Stromule extension along microtubules coordinated with actin-mediated anchoring guides perinuclear chloroplast movement during innate immunity

Abstract

Dynamic tubular extensions from chloroplasts called stromules have recently been shown to connect with nuclei and function during innate immunity. We demonstrate that stromules extend along microtubules (MTs) and MT organization directly affects stromule dynamics since stabilization of MTs chemically or genetically increases stromule numbers and length. Although actin filaments (AFs) are not required for stromule extension, they provide anchor points for stromules. Interestingly, there is a strong correlation between the direction of stromules from chloroplasts and the direction of chloroplast movement. Stromule-directed chloroplast movement was observed in steady-state conditions without immune induction, suggesting it is a general function of stromules in epidermal cells. Our results show that MTs and AFs may facilitate perinuclear clustering of chloroplasts during an innate immune response. We propose a model in which stromules extend along MTs and connect to AF anchor points surrounding nuclei, facilitating stromule-directed movement of chloroplasts to nuclei during innate immunity.

https://doi.org/10.7554/eLife.23625.001

eLife digest

Within a plant’s cells, compartments called chloroplasts harvest energy from sunlight. This process, termed photosynthesis, keeps the plant alive and growing. Yet this is not all that chloroplasts do. For example, if a harmful microbe infects the plant, its chloroplasts rapidly change shape and move toward the cell’s nucleus – the compartment of the cell that contains the bulk of the plant’s genetic material. The chloroplasts then send warning signals to the nucleus that boost the plant’s defenses.

In 2015, researchers showed that, during an infection, tiny tubes called stromules extend out from chloroplasts and make contact with the nucleus. Stromules are flexible structures that can extend and retract. However, it is not yet understood what exactly the stromules do, or how they establish connections with the nucleus.

Some scientists had suggested the internal skeleton of the cell – a complex network of protein filaments called actin and microtubules – might regulate the movements of the stromules. Now, Kumar, Park et al. – including several researchers from the 2015 study – have monitored the interaction between stromules and this internal “cytoskeleton” in leaf cells from a plant called Nicotiana benthamiana. Stromules, microtubules and actin were marked with fluorescent tags, which allowed them to be tracked under a microscope. This showed that the stromules actively extend along microtubules and anchor to a network of actin filaments.

Further work showed that chloroplasts move in the direction of the stromules. This suggests that chloroplast movement may actually be directed by these structures. This movement was seen in healthy plants growing under normal conditions, suggesting that it may be a more general occurrence in plants. Next, Kumar, Park et al. used a viral protein to provoke an immune response in the plants, and saw that a larger number of chloroplasts moved and clustered around the nucleus. This chloroplast clustering was also guided by stromules anchored to the actin filaments that surround the nucleus.

These findings shed new light on how chloroplasts communicate with the nucleus. Future work is needed to determine if stromules just guide chloroplast movement or if they provide a physical force that drives this process.

https://doi.org/10.7554/eLife.23625.002

Introduction

Stroma-filled tubular structures called stromules emanate from chloroplasts and have been observed in several genera in the plant kingdom, although they are most common in non-green plastids (Gray et al., 2001; Hanson and Sattarzadeh, 2008; Köhler and Hanson, 2000; Kumar et al., 2014; Natesan et al., 2005). Stromules are developmentally regulated and induced in response to biotic and abiotic stress, symbiotic association, and changes in plastid number and size (Brunkard et al., 2015; Caplan et al., 2015; Caplan et al., 2008; Erickson et al., 2014; Gray et al., 2012; Kumar et al., 2014; Schattat and Klösgen, 2011; Waters et al., 2004). The dynamic extension of stromules increases the surface area of chloroplasts, presumably facilitating transport of signals or macromolecules to the nucleus, cytosol, plasma membrane or other organelles (Gunning, 2005, 2004a; Kwok and Hanson, 2004c). We have recently shown that stromules are induced and function during innate immunity (Caplan et al., 2015). The induced stromules make connections with the nuclei to facilitate transport of chloroplast-localized defense protein NRIP1 (N receptor interacting protein 1) and the pro-defense molecule, hydrogen peroxide (H2O2), from chloroplasts into nuclei during an immune response (Caplan et al., 2015). Stromules may also facilitate certain number of chloroplasts to maintain contact with the moving nuclei (Erickson et al., 2017a). However, the mechanism(s) that facilitates chloroplast stromules connections to nuclei and eventual perinuclear clustering of chloroplasts is unknown.

Stromule length is variable as they extend, retract and branch, changing their shape and position (Gray et al., 2001; Gunning, 2005; Kwok and Hanson, 2004c; Waters et al., 2004). However, mechanisms that regulate the dynamic nature of stromule morphology and motility are poorly understood. Studies using inhibitors in non-green tissue have implicated cytoskeleton elements such as actin microfilaments (AFs) and microtubules (MTs) in regulating stromule frequency, length and motility (Gunning, 2005; Kwok and Hanson, 2003; Kwok and Hanson, 2004a). Treatment with AF inhibitors, Cytochalasin D (CTD) and Latrunculin B, resulted in the reduction of stromule frequency in tobacco hypocotyls (Kwok and Hanson, 2003). Stromules have been observed to extend parallel to AFs and the tips of stromules make contact with AFs in Arabidopsis hypocotyl epidermal cells (Kwok and Hanson, 2004a). Treatment with myosin ATPase inhibitor 2,3 butanedione 2-monoxime (BDM) affects stromule movement and length; furthermore, Myosin XI family motor proteins have been implicated in stromule movement and anchoring to the cytoskeleton in Nicotiana (Natesan et al., 2009; Sattarzadeh et al., 2009). These findings suggest that stromules move along AFs using myosin motors; however, direct evidence for movement along AFs is lacking. Treatment with MT inhibitor amiprophosmethyl (APM) reduced stromules, and co-treatment with AF and MT inhibitors decreased stromule frequency and length (Kwok and Hanson, 2003). In contrast, ‘chloroplast protrusions’ from mesophyll chloroplasts of the arctic plant Oxyria digyna remained unaffected by the MT inhibitor Oryzalin or the AF inhibitor LatB (Holzinger et al., 2007b). Therefore, the precise role of AFs and MTs during stromule dynamics in green tissue chloroplasts is not well understood.

Here, we analyzed the mechanism of stromule extension and movement in chloroplasts of green leaf tissue and perinuclear chloroplast clustering during innate immunity. Our results show that MTs are required for stromule extension and movement. MT depolymerization led to stromule retraction, and MT stabilization increased stromule frequency. Silencing the gene for γ-tubulin complex protein 4 (GCP4) caused enhanced bundling and disrupted dynamics of MTs, which resulted in longer stromules, but slower extension and retraction. Although stromule extension does not require AFs, they function as anchor points that stabilize stromules and anchor the body of chloroplasts. AFs play an important role in type of chloroplast movement that appears to be directed by stromules. This new type of stromule-directed movement is completely disrupted by AF inhibitors. However, stromule-directed chloroplast movement was still observed when AFs were partially disrupted, suggesting that chloroplast anchoring might restrict stromule directed movement. We hypothesize that a biological function of stromules is to direct the movement of chloroplasts. During an innate immune response, we propose a model where stromules extend along MTs towards nuclei and attach to the nuclei at actin anchor points; and, these perinuclear stromule attachments guide chloroplasts to the nucleus.

Results

Stromules interact and extend along microtubules

To examine the interactions of stromules with MTs, we expressed TagRFP fused to the N-terminal microtubule-associated protein domain of CKL6 (Ben-Nissan et al., 2008) (TagRFP-MAP-CKL6) in transgenic Nicotiana benthamiana plants expressing NRIP1 fused to Cerulean (NRIP1-Cerulean) that mark stromules (Caplan et al., 2015; Caplan et al., 2008). Marking both stromules and MTs revealed that these two structures overlapped in confocal microscopy images. These sites of overlap were designated as potential stromule-to-MT interactions. These observations were made in maximum intensity projections of z-stacks generated by confocal microscopy, and all observations in this study, were made in epidermal pavement cells of N. benthamiana plants. The varied morphology of stromules appeared to be correlated with MT interactions (Figure 1). Stromules often initiate as beak-like structures. The tips of beaks were seen interacting with MTs (Figure 1A; column 1). Beaks extend into longer stromules. Longer stromules were seen as just the tips of stromules interacting with MTs or the tip and the full length of the stromule aligned with MTs (Figure 1A; columns 2 and 3). More complex stromule structures, such as kinked or branched stromules, were found at the junction of two MTs (Figure 1A; columns 4 and 5). However, approximately 11% of stromules did not interact with MTs (Figure 1A, arrowhead), suggesting there is a MT-independent mechanism of stromule formation.

Figure 1 with 1 supplement see all
Chloroplast stromules extend along microtubules.

(A) MTs were marked by transiently expressing TagRFP-MAP-CKL6 (yellow) in N. benthamiana transgenic plants expressing NRIP1-Cerulean that marks stromules (cyan). Confocal micrographs of stromule-to-MT interactions in lower epidermal pavement cells are shown. Arrows indicate stromule beaking (1), tip contact (2), extension along (3), kinking (4) and branching (5) associations with MTs. Approximately 11% of stromules were not attached to MTs (2, arrowhead). Images are maximum intensity projection of confocal z-stacks. Scale bars equal 2 µm. Total of 103 stromules were observed in 11 biological replicates to generate this data. (B) Stromules were marked by expressing NRIP1(cTP)-TagRFP in transgenic GFP-TUA6 N. benthamiana plants. Time-lapse images were acquired and kymographs over 2 min were generated. Stromules were observed extending along MTs (left) and in both directions along MTs (right). Kymographs (bottom) were generated adjacent to the red lines in top images. (C) MTs were marked by expressing EB1-Citrine (yellow) in N. benthamiana transgenic lines expressing NRIP1-Cerulean that marks stromules (cyan). Stromules were observed extending with only the tip being pulled along MTs. Kymographs (bottom) were generated adjacent to the red lines in top image and show that a stromule tip translated along a MT at a constant rate and then rapidly changed direction (arrowhead). (D) The average velocity of stromules along MTs was calculated from manually tracked stromule tips moving along MTs marked with EB1-Citrine or GFP-TUA6. Data represented as the mean standard error of the mean (SEM), ****p<0.0001 by a Student’s t-test with Welch’s correction. (E) A stromule tip was tracked using a combination of fuzzy c-means and active contour algorithm, with shape analysis to calculate the length of the stromule, the tip velocity and the association with microtubules (Lu et al., 2017). Tip associations (green dots) with MTs (gray scale) were mapped over a time series. Tips not associating with MTs are depicted as red dots. Moving stromule tips were associated with MTs except when stromules were retracting (arrowhead).

https://doi.org/10.7554/eLife.23625.003
Figure 1—source data 1

Statistical analysis of the stromule velocity along MT.

A raw data for the average stromule velocity along MT presented in Figure 1D.

https://doi.org/10.7554/eLife.23625.005

A stromule-to-MT interaction was designated if these two structures were overlapping or not resolvable by confocal microscopy. However, since the resolution of confocal microscopy is relatively low, we verified the close interaction between stromules and MTs using transmission electron microscopy (TEM). Microtubules were originally detected and described in plants using TEM and can readily be observed as hollow, tubule-like structures that are 24 nm in diameter (Ledbetter and Porter, 1963). We were able to observe MTs by TEM and the close interactions of MTs with stromule tips and kink points (Figure 1—figure supplement 1). MTs were seen directly associated with the chloroplast outer envelope membrane at a kink point. Serial sections near the tip of a stromule that graze the chloroplast outer envelope membrane show MTs in line and in close proximity to the stromule.

Since our initial observations were from static images of stromules interacting with MTs, to look at the dynamics of stromules along MTs, we used an established transgenic N. benthamiana MT marker line expressing green fluorescent protein fused with the tubulin alpha 6 (GFP-TUA6) (Gillespie et al., 2002). In this transgenic line, we expressed NRIP1’s chloroplast transit peptide fused to TagRFP [NRIP1(cTP)-TagRFP] to mark stromules (Caplan et al., 2015). Time-lapse imaging of GFP-TUA6-labeled MTs revealed that stromules dynamically extended along MTs (Figure 1B, Video 1). Kymographs of the motion showed stromules extending and retracting in line with MTs in a single direction (Figure 1B, left kymograph) or moving bi-directionally in opposite directions (Figure 1B, bottom right kymograph). We also verified the stromule movement using another MT marker, the end binding one protein (Chan et al., 2003) fused to Citrine (EB1-Citrine). EB1-Citrine was initially chosen to examine the direction of movement because EB1 marks the positive end of MTs; however, the Agrobacterium-mediated expression often led to even staining of the MTs (Figure 1C). Kymographs of the motion of stromules showed clear movement of stromule tips along MTs (Figure 1C). The time lapse video show the dynamic interactions of stromules with MT, including branching, tip contact, transfer between microtubules, stromule initiation and bidirectional extension (Video 2). Our results from using three different MT markers via transgenic and transient expression indicate that stromules extend along MTs.

Video 1
Stromules extend along microtubules.

Stromules were marked by expressing NRIP1(cTP)-TagRFP (Cyan) in GFP-TUA6 (yellow) transgenic N. benthamiana plants. Time-lapse confocal microscopy was used to acquire images every 0.92 s that are displayed at 30 frames/s. The first half of the video (148 frames) shows a single stromule extending along microtubules and the second half (164 frames) of a stromule moving bidirectionally along microtubules. This video was used to generate the kymographs in Figure 1B.

https://doi.org/10.7554/eLife.23625.006
Video 2
Tracking of stromule tip interactions along MTs.

The first half of the video shows maximum intensity projections of time-lapse confocal microscopy of 120 images taken every 8.1 s and displayed at 8 frames/second. MTs were marked by expressing EB1-Citrine (yellow) in transgenic N. benthamiana plants expressing NRIP1-Cerulean that marks stromules (cyan). Chlorophyll autofluorescence is pseudo-colored in red. Stromules were observed extending with only the tip being pulled along MTs. The second half of the video shows the tracking of stromule tips using the method described in Figure 1—figure supplement 1. The contours of stromules were defined in blue. The tip associations (green dots) with microtubules (grayscale) and non-associating tips (red dots) were mapped over a time series. The video was used to generate Figure 1C–E.

https://doi.org/10.7554/eLife.23625.007

To quantify the motion, we manually tracked the velocity of stromules extending along MTs. All stromule extension was correlated with movement along MTs. The velocity of stromule extension was significantly lower when MTs were marked with EB1-Citrine (0.0565 µm/s) compared to GFP-TUA6 (0.146 µm/s) (Figure 1D). An automated algorithm for detecting stromule tips in maximum intensity projections was developed (Lu et al., 2017). The MTs were segmented and skeletonized (data not shown). Using the skeletonized images, the points of interaction (Figure 1E, green) and the points of no interaction (Figure 1E, red) were mapped over a time series (Video 2). Linear arrays of interaction points along MTs were clearly seen in time points T = 6, 12, and 18 min (Figure 1E). A retraction event had limited interaction with MTs (Figure 1E, arrowhead). The algorithm only accurately detected the slower moving motion when EB1-Citrine was used as a MT marker, and therefore, was not used in other experiments. The length, velocities, extension and retraction frequencies, and types of motion were quantified manually in all other experiments.

Stromule extensions through the ER is directed by microtubules

Stromules and the endoplasmic reticulum (ER) have correlated dynamics and three-dimensional arrangement; therefore, it was hypothesized that contact points along the ER direct their extension (Schattat et al., 2011). Since, our data suggested that the extension of stromules is directed by MTs, we examined stromules, ER and MTs simultaneously by co-expression of labels for the ER (SP-Citrine-HDEL) and MTs (TagRFP-MAP-CKL6) in NRIP1-Cerulean transgenic N. benthamiana plants that mark stromules. Similar to the previous report (Schattat et al., 2011), stromules were surrounded by ER, but here we show that MTs direct the movement through ER (Figure 2A; Video 3). Imaging using a high-resolution airyscan confocal microscope revealed that the ER forms channels around the stromules, and MTs were found at the stromule-to-ER interface (Figure 2B). Time lapse studies under similar imaging conditions showed that stromule extension occurred actively along the MTs, while the ER changed its direction and formed a channel around the extended stromule tip (Figure 2C, middle panel; Video 4). The stromule continued to extend along the MTs past the ER and no longer formed a channel around the extending stromule tip (Figure 2C, right panel). These time-lapse studies indicate that stromule extension is active along MTs and ER reorganization follows stromule extension along MTs.

Microtubules direct stromule movement along the ER.

(A) The endoplasmic reticulum (ER) were marked by transiently expressing SP-Citrine-HDEL (magenta) and MTs were marked by TagRFP-MAP-CKL6 (yellow) in N. benthamiana transgenic lines expressing NRIP1-Cerulean that marks stromules (cyan). Confocal microscopy time-lapse images of lower epidermis of leaves indicating changes in stromule extension along MTs and push through the ER network (bottom right, merged). Arrows indicate stromule extension within an ER channel (ER) along MT. (B) High resolution airyscan confocal micrograph showing the interaction of the stromule (S) with the MT within the ER channel. (C) High-resolution time-lapse images showing the extension of stromules (cyan) along the MTs (yellow) within and away from the ER channel (magenta). Arrows indicate active stromule extension, while the ER follows the course of the extending stromule (203 s) followed by the ER remodeling (219 s) independent of the stromule extension. Scale bars equal 5 µm.

https://doi.org/10.7554/eLife.23625.008
Video 3
Stromule move in channels of ER along MTs.

Video show stromules (cyan) first moving along MTs (yellow) to show dynamic association, then merged with the ER (magenta), and with stromules and ER only to show movement through ER channels. The video was used to generate Figure 2A.

https://doi.org/10.7554/eLife.23625.009
Video 4
Stromules remodeling ER.

Stromules (cyan) move along MTs (yellow) remodeling the ER (magenta). Video was created from three consecutive time series and was used to generate Figure 2B–C.

https://doi.org/10.7554/eLife.23625.010

Microtubules are required for stromule formation and extension

To further demonstrate that stromules extend along MTs, we expressed the MT marker TagRFP-MAP-CKL6 in NRIP1-Cerulean transgenic N. benthamiana plants and then disrupted MTs using 20 µM APM or 300 µM Oryzalin. Compared to the DMSO vehicle control (Figure 3A, top panel), depolymerization of MTs was noticeable 5 to 15 min after APM and Oryzalin treatment, leaving behind remnants of partially depolymerized MTs and an increase of the MAP-CKL6 MT marker in the cytosol (Figure 3A, middle and bottom panels). Although mock control with DMSO resulted in an increase in stromules compared to the infiltration media control, the APM or Oryzalin disruption of MTs for 15 min significantly inhibited this increase in stromule number (Figure 3B). MT depolymerizers APM and Oryzalin not only decreased stromule number but also restricted stromules to MT fragments causing changes in stromule movement (Video 5). Beak-like protrusions from chloroplasts that did not result in stromules were also observed in APM and Oryzalin treatment (Figure 3A, asterisk; Video 5), however, we did not determine if these increased with the treatments compared to the DMSO vehicle control. In the time-lapsed data set shown in Figure 3A, stromule length was gradually reduced during 15-min treatment with APM that was caused by stromule retraction and correlated with simultaneous depolymerization of the MTs and, eventually, complete retraction of stromules (Figure 3A, middle and bottom panels, arrowhead). Similarly, with Oryzalin treatment at 0 min, we observed a region of the stromule overlapped with a segment of the MT (Figure 3A, bottom panel, T = 0). As the time course progressed, the segment of MT became shorter and there was a corresponding reduction in the length of the stromule (Figure 3A, bottom panels). In that time-lapsed data set, at 15-min treatment with Oryzalin we observed that stromules completely retracted from and changed course from the MT (Figure 3A, bottom panel, T = 15; Video 5). These results indicate that the disappearance of stromules may be a combination of disruption of extended stromules and the prevention of induction of new stromules.

Microtubule disruption lead to stromule retraction and microtubule stabilization increased stromules.

(A) Dynamics of MT depolymerization and stromule retraction in the lower epidermis of NRIP1-Cerulean transgenic N. benthamiana plant leaves after infiltration of mock control (0.2% DMSO) or MT depolymerizing agents, APM (20 µM) or Oryzalin (300 µM). Images are maximum intensity projections represented at time points with a 5-min interval after infiltration. Arrows in mock (top row) indicate extended and branching stromules. MT depolymerization due to APM and Oryzalin causes simultaneous stromule retraction within 15 min (arrowhead; middle and bottom panels). A beaking was initiated but failed to progress to stromule (asterisk; middle and bottom panels). Scale bar equals 2 µm. (B) Stromules were increased by DMSO vehicle control treatment from 0 to 15 min compared to no significant increase from the infiltration media control. Compared to the DMSO vehicle control, stromules significantly decreased after treatment with APM (20 µM) and Oryzalin (300 µM) at 15 min and no other comparisons were significant. The experiment was repeated four times with three to five replicates per experiment. Error bars represent mean ± standard error of the mean (SEM) **p<0.05. (C) Treatment with microtubule stabilizing agent Paclitaxel (0.8 nM) produced multiple stromules from single chloroplast after 30 min (arrows, bottom left panel). The extended stromules overlapped with the MTs (arrows, merged panel). Images are maximum intensity projections. Scale bar equals 5 µm. (D) Paclitaxel treatment increased stromules per chloroplasts after 30-min treatment compared to mock treated leaves. The experiments were repeated four times with two replicates per experiment. Error bars indicate mean ± standard error of the mean (SEM) **p<0.05 by a Student’s t-test with Welch’s correction.

https://doi.org/10.7554/eLife.23625.011
Figure 3—source data 1

Statistics of the stromule frequency under MT disrupting drug treatments.

Raw datapoints of stromule frequency after APM, Oryzalin, or Paclitaxel, presented in Figure 3B and D. Statistical analyses and the original graph generated by Prism7 were also included.

https://doi.org/10.7554/eLife.23625.012
Video 5
Stromule dynamics are disrupted by APM or Oryzalin MT inhibitor treatment.

Disruption of MT (yellow) by APM (20 µM) resulted in a loss of stromule extensions (middle) compared to the mock treatment control (left). Disruption of MT (yellow) by Oryzalin (300 µM) resulted in a loss of stromule extension compared to the mock treatment control (left). Maximum intensity projections of time-lapse z-stacks were taken every 40 s for Mock and Oryzalin and every 29 s for APM. Videos are 143 fames displayed at 15 frames/s. The video was used to generate Figure 3A.

https://doi.org/10.7554/eLife.23625.013

Changes in microtubule organization increases stromule number, length and stability

Since our results from MT inhibitor studies indicated that stromule formation and extension require MTs, we tested the effect of stabilizing MTs using Taxol (Schiff and Horwitz, 1980). Infiltration of Paclitaxel-BODIPY conjugate into leaves of transgenic NRIP1-Cerulean N. benthamiana plants showed extensive MT stabilization after 30 min of treatment compared to the mock control (Figure 3C) and significantly induced stromules compared to mock control (Figure 3C–D). Interestingly, after Paclitaxel treatment, we observed long stromules and multiple stromules emanating from individual chloroplasts (Figure 3C, bottom panels). These results suggest that MT stabilization is sufficient to induce stromules.

To more specifically alter MT organization and dynamics, we knocked-down the expression of GCP4 in N. benthamiana plants using virus-induced gene silencing (VIGS) approach (Dinesh-Kumar et al., 2003). GCP4 is a subunit of the γ-tubulin complex and artificial miRNA (amiR)-mediated knockdown of Arabidopsis GCP4 resulted in hyper-parallel and bundled cortical MT in leaf epidermal cells (Kong et al., 2010). We silenced NbGCP4 in NRIP1-Cerulean and GFP-TUA6 transgenic N. benthamiana plants to visualize the effect on stromules and MTs, respectively. Since amiR-AtGCP4 in Arabidopsis plants resulted in a significant growth phenotype, we first determined how many days of NbGCP4 VIGS resulted in a MT alteration without a severe growth phenotype. Four days after silencing, NbGCP4-silenced plants phenotypically looked similar to that of VIGS vector control plants (Figure 4—figure supplement 1A). However, six days post-silencing, leaves of the NbGCP4-silenced plants developed a crinkled leaf phenotype (Figure 4—figure supplement 1B, right panel). In addition, at this time point, the NRIP1-Cerulean stromule marker begin to leak out of chloroplasts compared to the VIGS control plants (Figure 4—figure supplement 1C, right panel). Fourteen days post-silencing, NbGCP4-silenced plants showed severe growth arrest and morphological distortion (Figure 4—figure supplement 1D). Thus, we observed stromules in leaf epidermal cells of the plants after 4 days of NbGCP4 VIGS, to minimize potential physiological changes that might occur due to the alterations of MT organization and dynamics. Although at 4 days-post silencing, NbGCP4 mRNA levels are reduced by only 50% in the silenced plants compared to the VIGS control plants (Figure 4B), cortical MT organization was significantly altered in the leaves of NbGCP4-silenced plants compared to the control (Figure 4A) in a similar way to amiR-AtGCP4 in Arabidopsis (Kong et al., 2010). To quantify these changes, we used SOAX software that uses Stretching Open Active Contours (SOACs) to quantify filamentous networks (Xu et al., 2015). SOAX analysis showed that MTs were more parallel or aligned in NbGCP4-silenced plants compared to the vector control (Figure 4C), which was visible by displaying the MT direction by color-coding the azimuthal angles (Figure 4D). Quantitative SOAX analysis shows that silencing NbGCP4 decreases the curvature (Figure 4E) and increases the snake length fitted to MTs (Figure 4F). The snake length is not a direct measurement of MT length, since this approach cannot accurately distinguish between two MTs that are bundled together. Nonetheless, this measurement further suggests that silencing NbGCP4 alters MTs.

Figure 4 with 1 supplement see all
Alterations of MT is correlated with NbGCP4 silencing.

(A) In GFP-TUA6 transgenic N. benthamiana leaves that mark MTs (yellow), NbGCP4 silencing resulted in hyper-parallel and occasional bundling of MT (VIGS-NbGCP4, right) compared to the VIGS vector control (VIGS-EV, left). Images are maximum intensity projections of confocal z-stacks. Scale bar represents 40 µm. (B) qRT-PCR analysis of transcript levels of NbGCP4 in VIGS-EV control and VIGS-NbGCP4 plants showed a significant decrease in NbGCP4 transcript level 4 days after silencing by VIGS-NbGCP4 compared to the VIGS-EV control. Data represented as the mean +standard error of the mean (SEM), n = 12, ***p<0.001 (Student's t-test). (C) Azimuthal angles were analyzed by converting angles to mean resultant length (MRL) by converting the MT angles into individual vectors, adding the vectors together, and calculating the mean. MRL values are between 0 and 1, with 0 indicating that MT angles are random and one indicating all MT angles are the same and completely aligned. Data represented as the mean + standard error of the mean (SEM), *p<0.01 (student’s t-test). (D) SOAX analysis was conducted on the images in (A). MT filaments are color-coded based on the azimuthal angle so that parallel MTs are the same color. (E) Curvature analysis that measures the rate of change of tangent vectors shows MTs in VIGS-NbGCP4 have less curvature. Box covers from first to third quartiles while a bar in the middle of the box indicates median. Whiskers show from minimum to maximum. ****p<0.0001 by Mann-Whitney test. (F) Analysis of the snake length computed by SOAX analysis showed an increase in MTs length in VIGS-NbGCP4. Data represented as median and 95% confidence interval. ****p<0.0001 by Mann-Whitney test.

https://doi.org/10.7554/eLife.23625.014
Figure 4—source data 1

Statistics of the quantitative RT-PCR to validate GCP4 gene silencing.

Statistical analyses and the graph of the quantitative RT-PCR for the NbGCP4 gene silencing presented in Figure 4B were generated by Prism7.

https://doi.org/10.7554/eLife.23625.016
Figure 4—source data 2

Statistics of the Azimuthal angles of the MT in VIGS-NbGCP4 plants.

Statistical analyses and the graph of the Azimuthal angles of the MT in the NbGCP4 silenced plants presented in Figure 4C were generated by Prism7.

https://doi.org/10.7554/eLife.23625.017
Figure 4—source data 3

Statistics of the curvature of MTs in VIGS-NbGCP4 plants.

Statistical analyses and the graph of the curvature of the MT in the NbGCP4 silenced plants presented in Figure 4E were generated by Prism7.

https://doi.org/10.7554/eLife.23625.018
Figure 4—source data 4

Statistics of the SOAX analysis of MTs in VIGS-NbGCP4 plants.

Statistical analyses and the graph of the SOAX analysis of the MT in the NbGCP4 silenced plants presented in Figure 4F were generated by Prism7.

https://doi.org/10.7554/eLife.23625.019

The alteration in MT organization at four days-post silencing of NbGCP4 (Figure 4), resulted in more than twice the number of stromules in NbGCP4-silenced plants compared to the VIGS vector control (Figure 5A, top panels; 5B, compare bars in mock treatment). Stromules were on average significantly longer in NbGCP4-silenced plants compared to VIGS vector control (Figure 5A, top panels; Figure 5C, compare bars in mock treatment). Furthermore, a greater percentage of stromules were longer than 3 μm in NbGCP4-silenced plants (Figure 5—figure supplement 1A). We classified stromule movement into three types, smooth and constant movement, sudden and erratic movement, and side and tangential movement (Figure 5—figure supplement 1B) and found that, in NbGCP4-silenced plants, stromule movements were more constant than those in VIGS vector control (Figure 5—figure supplement 1C).

Figure 5 with 1 supplement see all
Microtubule stabilization induces stromules constitutively.

(A) Stromules were induced 24 hr after TMV-p50 effector infiltration as described in Caplan et al. (2015) (bottom left) compared to a mock control (top left). Stromules were induced both in mock (top right) and p50 infiltrated (bottom right) NbGCP4-silenced plants indicating that NbGCP4 silencing induces stromules constitutively. Stromules were often long and branched in mock-treated NbGCP4-silenced plants (top right, yellow arrowheads) compared to the VIGS-EV control (top left). Scale bar represents 40 µm. (B) Quantitative representation of stromules from A. Stromules were significantly induced in mock-treated NbGCP4-silenced plants compared to the mock-treated VIGS-EV control (compare left green bar and magenta bar). Stromules increased significantly in VIGS-EV control plants treated with p50 effector compared to the mock (compare green bars). No significant difference in stromule number was observed in NbGCP4-silenced mock-treated plants compared to the p50-treated plants (compare magenta bars). Four images per leaf were generated for quantification from total of 12 leaves for each condition. Data represented as the mean +standard error of the mean (SEM), ***p<0.001 (Student's t-test with Welch's correction). (C) Stromule length was significantly increased in mock-treated VIGS-NbGCP4 plants compared to the VIGS-EV control (left open bars). p50-effector-induced immune response increased stromule length in VIGS vector control plants compared to the mock-treated VIGS-EV control plants (compare green open bars). No significant change was observed between p50-treated and mock-treated VIGS-NbGCP4 plants (compare red open bars). Box and whisker plot was drawn with rank transformation. Box cover from first to third quartiles while a bar in the middle of box indicates median. Whiskers show from 5% to 95% of ranking. ***p<0.001, ⧫ comparison with VIGS-EV control, p<0.001 by Mann-Whitney test. Dots in the graph indicate outliers. (D) The velocity of stromule extension and retraction in VIGS-EV control and VIGS-NbGCP4 with or without TMV-p50-induced immune response. Data represented as the mean + standard error of the mean (SEM). Symbols at the top of bars indicate significant differences according to Dunnett’s multiple comparison test. Single symbol (*,⧫, ■), control set for each comparison; two-symbols (**,⧫⧫, ■■), p<0.05 and three-symbols (***,⧫⧫⧫, ■■■), p<0.001. Scale bars equal 40 µm. (E) TMV-p50-induced immune response resulted in hyper-parallel MTs (NN + p50, right) compared to the control (nn +p50, left) in transgenic N. benthamiana leaves that mark MTs (yellow). Images are maximum intensity projections of confocal z-stacks. Scale bar represents 20 µm. (F) Azimuthal angle differences of MT filaments were measured by the length of the arc. Data represented as the mean +standard error of the mean (SEM), p=0.0713 (Student’s t test with Welch’s correction). (G) Curvature analysis that measures the rate of change of tangent vectors shows MTs in NN + p50 have less curvature. Box covers from first to third quartiles while a bar in the middle of box indicates median. Whiskers show from minimum to maximum. *****p<0.00001 (Mann-Whitney test). (H) Analysis of the snake length computed by SOAX analysis showed an increase in MTs length in NN + p50. Data represented as median and 95% confidence interval. ***p<0.001 by Mann-Whitney test.

https://doi.org/10.7554/eLife.23625.020
Figure 5—source data 1

Statistics of the stromule dynamics in VIGS-NbGCP4 plants.

Statistical analyses and the graph of the stromule frequency, length, and velocity in the NbGCP4 silenced plants presented in Figure 5B-D by Prism7 are included.

https://doi.org/10.7554/eLife.23625.023
Figure 5—source data 2

Statistics of the MT alteration by TMV-p50 induced immune responses.

Statistical analyses and the graph of the MT distribution presented in Figure 5F-H were generated by Prism7.

https://doi.org/10.7554/eLife.23625.024

We recently reported that stromules are induced significantly during an immune response against bacterial and viral infections (Caplan et al., 2015). The nucleotide-binding domain leucine-rich repeat (NLR) immune receptor N recognizes p50 effector from Tobacco Mosaic Virus (TMV) and activate immune response to limit TMV to the infection site (Whitham et al., 1994). The stromules are significantly induced during N NLR-mediated immunity to TMV (Caplan et al., 2015). Therefore, we tested if N NLR-mediated activation of immune response could further increase stromule number and length in NbGCP4-silenced plants. For this, we silenced NbGCP4 in transgenic N. benthamiana expressing N NLR and NRIP1-Cerulean (stromule marker) for 3 days and then infiltrated with p50 and 24 hr later the observations were recorded. As shown before (Caplan et al., 2015), the number of stromules significantly increased in p50-treated VIGS vector control plants compared to mock-treatment (Figure 5A, compare left panels and 5B, compare green bars). The average length (Figure 5C, green bars) and percentage of stromules longer than 3 μm also increased during an immune response (Figure 5—figure supplement 1A). Interestingly, the increase in stromules in mock-treated NbGCP4-silenced plants and a p50-induced immune response in VIGS vector control were remarkably similar (Figure 5A, compare top right panel with bottom left panel; Figure 5B, compare mock-treated magenta bar with p50-treated green bar). There was no significant change in stromule number in p50-treated NbGCP4-silenced plants compared to the mock-treated NbGCP4-silenced plants (Figure 5A, right panels and 5B, compare magenta bars). Mock-treated NbGCP4-silenced plants also showed longer stromules compared to mock-treated VIGS vector control plants (Figure 5C). This increase was similar to that of in p50-treated VIGS vector control plants that showed significantly longer stromules compared to mock-treated plants (Figure 5C, compare green open bars). However, there was no significant difference in stromule length in p50- and mock-treated NbGCP4-silenced plants (Figure 5C, compare magenta open bars). Collectively, these results indicate that the activation of immune response does not further increase stromule number and length in NbGCP4-silenced plants that exhibit constitutive stromule induction.

The velocities of stromule extension and retraction were calculated as an indicator of stromule dynamicity and stability. The stromule extension and retraction velocities decreased in the NbGCP4-silenced plants compared to the VIGS vector control (Figure 5D), suggesting that stromules were less dynamic and more stable. These results indicate that specific alterations of MTs are correlated with change in stromule dynamics and further support a role for MTs in regulating stromules. Interestingly, p50-treated VIGS vector control compared to the mock treatment reduced the velocities of stromule extension and retraction (Figure 5D, compare green bars) suggesting that stromules are less dynamic and more stable during active immune response. To test if p50-induced immunity alters MT organization resulting in alteration in stromule dynamics, we observed MT dynamics upon TMV-p50 treatment. For this, MT marker TagRFP-MAP-CKL6 was infiltrated into transgenic N. benthamiana plants expressing N NLR and NRIP1-Cerulean (NN) or expressing only NRIP1-Celulean without N NLR (nn). 12 hr later, p50 was infiltrated into the same spot to induce an immune response. After 48 hr of TagRFP-MAP-CKL6 expression and 36 hr of p50 expression, the MT cytoskeleton was imaged and then analyzed by SOAX. Visible differences in MTs between immunity-induced plants (Figure 5E, NN + p50) and non-immunity-induced plants (Figure 5E, nn +p50) were difficult to observed in the images, but interestingly, SOAX analysis revealed that p50-induced immunity altered MT morphology (Figure 5F–H). Specifically, there were minor differences in orientation (Figure 5F), curvatures were significantly smaller (Figure 5G) and snake lengths were larger (Figure 5H) in NN + p50 compared to nn +p50. Collectively, these results indicate that changes in MT organization caused by NbGCP4-silencing plants or during p50-induced immunity are correlated with changes in stromule dynamics, indicating a possible direct or indirect role for MT organization in modulating stromule dynamics.

Actin filaments serve as anchor points but not as tracks for stromule extension

Since AFs were previously shown to regulate chloroplast movement and stromule morphology (Kwok and Hanson, 2003; Kwok and Hanson, 2004a), we tested if stromules extend along AFs. We expressed Lifeact-TagRFP that labels AF (Era et al., 2009; Riedl et al., 2008) in transgenic N. benthamiana plants expressing NRIP1-Cerulean that marks stromules (Caplan et al., 2015; Caplan et al., 2008). Out of 73 stromule tip extension events from 34 cells, the vast majority (93%) of stromule tip extensions were not observed along AFs. Stromules were occasionally observed to be aligned with AF (Figure 6A, asterisk), but high-resolution examination showed that they were not co-localizing (Figure 6—figure supplement 1). Instead, in many cases, stromules interacted at restricted foci (Figure 6A, arrowheads) that often corresponded with a kink in the stromule. We verified these interactions using TEM and found an AF bundle in close proximity to the apex of a stromule kink (Figure 6—figure supplement 2A–B). Stromule tips often reached actin filaments (Figure 6A, arrows); however, time-lapse videos showed stromules interacting with AFs, but not extending along AFs (Video 6).

Figure 6 with 3 supplements see all
Stromule are anchored to the actin microfilament network.

(A) AFs were marked by expressing Lifeact-TagRFP (magenta) in N. benthamiana transgenic lines expressing NRIP1-Cerulean that marks stromules (cyan). Stromules were seen interacting with AFs at kink points (arrow heads). Tips were commonly seen not associated with AFs (Arrows). Stromules were occasionally seen oriented along AFs but not overlapping with AFs (*). Scale bar equals 5 µm. (B) An extended stromule with a tip in close proximity to an AF, became kinked at a point overlapping with an AF near the midpoint of the stromule, and retracted back to the kink point interacting with the AF. (C) A kymograph was created along the stromule adjacent to the 9.7 µm green line in panel B over 8 min. The stromule tip that was in close proximity to an AF and then rapidly retracts to an actin anchor. It remained attached to the actin anchor point for an additional 4 min before retraction to the body of the chloroplast. (D) The percent of stromules pausing at AFs during retraction events was quantified. No pausing resulted in a full retraction back to the body of the chloroplast (grey bar). Stromule retractions that did not retract completely and paused for multiple time frames showed a correlation of the paused stromule tip with an AF (magenta bar) or no correlation with an AF (black bar). Data was collected from 22 biological replicates spanning eight different experimental replicates. Eighty-two retraction events were quantified from 30 different cells. Data represented as the mean +standard error of the mean (SEM), **p<0.001, ****p<0.00001 (Student’s t test with Welch’s correction). (E) AFs were marked by expressing Citrine-mTalin (magenta) in N. benthamiana transgenic lines expressing NRIP1-Cerulean that marks stromules (cyan). High-resolution airyscan confocal micrographs revealed thinning points of stromule or chloroplast interactions with AFs (arrowheads). Scale bar equals 2 µm.

https://doi.org/10.7554/eLife.23625.025
Figure 6—source data 1

Statistics of the quantification of stromules pausing at the AFs.

The raw data points to calculate the percentage of stromules pausing at the AFs during retraction events, presented in Figure 6D.

https://doi.org/10.7554/eLife.23625.030
Video 6
Stromules do not extend along AFs.

Stromules (cyan) interacted with AFs (magenta) at points along the stromule, but stromules did not directly extend along AFs. Sixty images were captured by time-lapse confocal microscopy every 18 s and displayed at 8 frames/s. The video was used to generate Figure 6A.

https://doi.org/10.7554/eLife.23625.031

To determine if actin plays another role in stromule dynamics, we performed time-lapse studies in epidermal cells expressing the actin marker Lifeact-TagRFP. Stromules appeared to interact statically, and not dynamically, with AFs, suggesting there are actin anchor points along stromules. Interactions were observed at the tips or at kink points (Figure 6B). Kymographs and time lapsed video show that retracting stromules paused for multiple, consecutive time frames or stopped completely at AFs (Figure 6C, Figure 6—figure supplement 1, Video 7). Due to the density of the AF network, stromules are often seen intersecting with AFs. To indirectly determine if those points of intersection are potential AF anchor points, we examined stromule retraction events. 19.4% of stromules retracted fully back to the body of the chloroplast without any pausing, often passing intersections with AFs. 77.1% of retracting stromule tips paused for multiple, consecutive frames and showed colocalization with an AF. 5.7% of retracting stromule tips paused for multiple, consecutive frames, but did not colocalize with AF (Figure 6D). The pausing of retracting stromules at AF cannot be explained by chance alone because the density of AFs and the colocalization of stromules with AFs observed appeared to be much less than 77.1% (Figure 6A; Figure 6—figure supplement 1; Video 7). Therefore, this data suggests that there are actin anchor points along stromules. We further examined the interaction of stromules and AFs by expressing mTalin-Citrine in NRIP1-Cerulean N. benthamiana plants and then examining by high-resolution airyscan confocal microscopy (Figure 6E). AF marker mTalin-Citrine has been used previously to detect chloroplast-associated actin (cp-actin) (Kadota et al., 2009). We could observe clear interactions of AF with chloroplast bodies (Figure 6E). Interestingly, we observed a clear thinning or constriction at the site of stromule-to-actin interaction points along the length and across the body of the chloroplast (Figure 6E; arrowheads). Three-dimensional modeling shows grooves across the body of the chloroplast that correlate with AFs (Figure 6E, Figure 6—figure supplement 2C–D). Although the mechanism of stromule thinning at actin interaction sites is unknown, collectively these results indicate that AFs provide anchor points for stromules but not tracks for stromule extension.

Video 7
Stromule retraction to actin anchor points.

Two stromules (cyan) retracted to multiple actin anchor points (arrows) along AFs (magenta). Thirty-seven images were captured by time-lapse confocal microscopy every 18 s and displayed at 12 frames/s. The video was used to generate Figure 6B–C.

https://doi.org/10.7554/eLife.23625.032

To determine the effect of AF disruption on stromule formation, we expressed mTalin-Citrine in NRIP1-Cerulean transgenic lines and applied 200 µM Cytochalasin D (CTD) to depolymerize AFs for 30 min (Figure 6—figure supplement 3). Since CTD treatment disrupted the actin network only in a fraction of the cells, only cells with a disrupted actin network were examined. Stromules were still present in cells where actin network was disrupted (Figure 6—figure supplement 3A) and the stromule number was similar between the CTD- and mock-treatments (Figure 6—figure supplement 3B).

Microtubules and actin filaments contribute to stromule dynamics

Several studies suggested that MT and AF networks might work cooperatively for maintaining cell structure and physiology in eukaryotic systems (reviewed in [Takeuchi et al., 2017]). Although stromule formation and extension is primarily associated with MTs, AFs might have a role in stromule dynamics. To examine the role of each cytoskeletal filament, we treated transgenic N. benthamiana plants expressing GFP-TUA6 that marks MTs and FABD2-GFP that marks AFs with longer treatments of low concentrations of cytoskeleton inhibitors that specifically disrupt one cytoskeleton component, but not other. These experiments are in contrast to shorter treatments of higher concentrations (Figure 3, Figure 6—figure supplement 3) that are not optimal for time lapsed acquisition of stromule dynamics. We found that treatment with 1 µM of oryzalin (ORY) treatment for 1 hr partially disrupted MTs and had no significant, visible effect on the AF network (Figure 7—figure supplement 1, middle panels). Next, we tested CTD treatment concentrations to disrupt AFs. 10 µM of CTD treatment for 1 hr fully disrupted actin filament AF network showing bright puncta of GFP-FABD2, but only had a mild effect on MT organization (Figure 7—figure supplement 1, right panels); therefore, we used 10 µM of CTD. NRIP1-Cerulean transgenic plants were treated with either 10 µM CTD or 1 µM ORY for 1 hr to disrupt AFs or MTs, respectively, and then stromule length and dynamics were analyzed (Figure 7A). Interestingly, although average stromule length in both drug treatments were not significantly different (Figure 7B), velocity of stromule extension was increased significantly in ORY treatment compared to DMSO control (Figure 7C). Furthermore, CTD treatment resulted in significant reduction in velocity of both stromule extension and retraction compared to the control (Figure 7C). Interestingly, CTD treatment increased constant and smooth movements of stromules and reduced sudden and erratic movements of stromules, suggesting that CTD treatment stabilizes stromule dynamics (Figure 7D). Together, these results indicate that both types of cytoskeletal filaments regulate stromule dynamics.

Figure 7 with 1 supplement see all
Disruption of cytoskeleton change stromule dynamics.

(A) Time lapsed images of stromules 1 hr post-treatment with 1 µM of MT inhibitor oryzalin (ORY) or with 1 µM of actin inhibitor cytochalasin D (CTD) on the leaf of the NRIP1-Cerulean transgenic N. benthamiana. At these concentrations, ORY disrupt MT organization slightly while no visible effect on actin cytoskeleton. On the other hand, CTD showed significant disruption of the actin filament while no significant effect on MT organization (see Figure 7—figure supplement 1). The experiments were repeated three times with six replicates per treatment. Scale bar equals 20 µm. (B)-(D) Quantification of stromule dynamics in A. (B) Stromules length did not change significantly upon inhibitor treatments. (C) ORY treatment increased stromule extension velocity (magenta bars compared to green bars), while CTD treatment reduced the velocity of both stromule extension and retraction (blue bars compare to green bars). Data represented as the mean +standard error of the mean (SEM), ****p<0.0001 (Dunn’s multiple comparison test). (D) The frequency of constant, smoothly extending stromules was increased (left panel) and the frequency of sudden extending stromules decreased (right panel) with CTD treatment. ORY treatment showed no significant difference. Data represented as the mean +standard error of the mean (SEM), *p<0.05, (Mann-Whitney test).

https://doi.org/10.7554/eLife.23625.033
Figure 7—source data 1

Statistics of the stromule dynamics.

All of the statistical analyses and the original graphs of the stromule dynamics presented in Figure 7B-D generated by Prism7 are shown.

https://doi.org/10.7554/eLife.23625.035

Stromules direct chloroplast movement

While analyzing time-lapsed images of stromule movements in N. benthamiana transgenic plants expressing NRIP1-Cerulean, we observed the movement of chloroplasts in the direction of stromules (Figure 8A; Video 8) or toward stromule kinks that are correlated with anchor points (Figure 6, Figure 8—figure supplement 1). This observation suggests that stromules might direct or guide chloroplast movement. To examine if this movement is correlated with the interactions with the cytoskeleton, we co-expressed Lifeact-TagRFP that marks AFs and NRIP1(cTP)-TagBFP that marks stromules in N. benthamiana transgenic plants expressing GFP-TUA6 that marks MTs (Figure 8B). Stromules were anchored to AF and connected to MT for extension at 0 min. Stromules extend along MT at 1 min and retracted to the actin anchor point at 3 min. Stromule reextend on MT at 8 min. Retraction of stromule at 9 min led to movement of chloroplast body toward the direction of the stromule movement (Figure 8B; Video 9).

Figure 8 with 2 supplements see all
Stromule directed chloroplast movement.

(A) Stromules and chloroplast movement events were imaged in NRIP1-Cerulean N. benthamiana transgenic plants. A chloroplast was tracked (red line) in time lapsed images. The direction of movement was correlated with the direction of the stromule. A connected chloroplast (asterisk) moved with the stromule directed chloroplast. Time is measured in seconds and scale bars equal 2 µm. (B) AFs and stromules were marked by transiently expressing Lifeact-TagRFP (magenta) and NRIP1(cTP)-TagBFP (cyan) in N. benthamiana transgenic plants expressing GFP-TUA6 that marks MTs (yellow). The top row shows the merged images and the bottom row is an illustration highlighting the MT and actin dynamic events. At 0 min, the stromule tip is bound to a MT and a branch point is bound to an AF. At 1 min, the stromule extended along the MT. At 3 min, the stromule retracted to an actin anchor point. At 8 min, the stromule re-extended along a MT. At 9 min, the stromule retracted and correlated with chloroplast movement. Scale bar equals 5 µm. (C) The direction of the stromule connected to the chloroplast body and the direction of chloroplast movement were measured in FIJI ImageJ. The difference in angle was calculated and plotted. Both Mock and Oryzalin (ORY), showed a high frequency of values close to 0. Randomly generated values were used as a control. (D) The percent of chloroplast movement that were stromule directed movements were quantified. Cytochalasin D (CTD) treatment resulted in a complete halt of movement. Oryzalin (ORY) treatment caused a decrease in stromule directed movement compared to the DMSO vehicle control. Data represented as the ±SEM, ****p<0.0001, **p<0.01 (one-way ANOVA). (E) Stromules retract more frequently (ORY; pink bar) and extend less frequently (ORY; magenta bar) with oryzalin treatment. Data represented as the mean +standard deviation (SD), *****p<0.00001 by Mann-Whitney test.

https://doi.org/10.7554/eLife.23625.036
Figure 8—source data 1

Statistical analyses of the stromule-directed movements.

Statistical analyses and the graph to support the stromule-directed movements presented in Figure 8D and E were generated by Prism7.

https://doi.org/10.7554/eLife.23625.040
Video 8
Stromule-directed movement.

Chloroplast bodies and stromules were visualized in a NRIP1-CFP transgenic N. benthamiana plants, in combination with chlorophyll autofluorescence. Time-lapse confocal microscopy was used to acquire images every 10 s for 600 s, displayed at 8 framess. In upper right corner, an instance of stromule-directed movement can be observed, which was used to generate Figure 8A. The chloroplast located in the bottom-center of the video, shows another instance of stromule-directed movement.

https://doi.org/10.7554/eLife.23625.041
Video 9
Dynamic control of stromules by extension along MTs and anchoring to AFs.

AFs and stromules were marked by transiently expressing Lifeact-TagRFP (magenta) and NRIP1(cTP)-TagBFP (cyan) in N. benthamiana transgenic lines expressing GFP-TUA6 that marks MTs (yellow). Sixteen maximum intensity projections of time-lapse z-stacks taken were taken every 24.5 s and displayed at 4 frames/s. The video was used to generate Figure 8B.

https://doi.org/10.7554/eLife.23625.042

Next, we investigated if the stromule angle and the angle of chloroplast movement are significantly correlated and changed by ORY or CTD treatment. Since CTD treatment resulted in a complete disruption of chloroplast movement (Video 10), it was not analyzed. Chloroplast movement was first identified as any movement larger than the radius of the chloroplast body and the direction of the movement was measured as the angle from the start and end points of each movement events. If a chloroplast changed direction, that was considered a separate movement event. Only chloroplasts containing one or more stromules were used for this analysis because it depends on comparing paired measurements of the angle of the stromule from the chloroplast body attachment point to the tip and the angle of chloroplast movement. We compared 33 pairs for DMSO control and 47 pairs for ORY to calculate a circular correlation coefficient, r(FL) (Fisher and Lee, 1983). An r(FL) value of 1.0 would indicate that the stromule angle and the angle of chloroplast movement are always identical, an r(FL) of −1.0 would mean the paired angles differ by 180 degrees, and if the angles are randomly matched the r(FL) will be close to zero. The r(FL) values for DMSO and ORY were 0.76 and 0.85, respectively. To test the statistical significance, each data set was randomly shuffled 10,000 times and the r(FL) calculated for each randomization; the observed f(FL) values were greater than all the randomized r(FL) values, so for both DMSO and ORY the stromule angle and the angle of chloroplast movement were significantly correlated (p<0.0001). Standard errors of the r(FL) values were calculated using the jackknife method (Sokal and Rohlf, 1995), and used in a two-sample t-test; the r(FL) values for DMSO and ORY were not significantly different from each other (p=0.52). We generated a scatter plot of chloroplast movement angles and stromule angles, which shows a fairly linear relationship compared to the randomized control (Figure 8—figure supplement 1). To further visualize these data, we calculated the difference between the two angles and plotted the frequency (Figure 8C). If the chloroplast movement angle and stromule angle are equal, then the difference will be zero. We observed a higher frequency around zero compared to the randomized control. An examination of only angle pairs with less than ±30 degree difference were highly correlated and had an r(FL) value of 0.95; we therefore defined a stromule-directed movement event as being within ±30 degrees. The circular correlation calculation requires paired stromule angles and chloroplast movement angles, and excludes chloroplasts that move but do not have stromules. Using the ±30 degree criteria for stromule directed movement, we were able to compare the percent of stromule driven movement compared to total movement events, which includes chloroplasts without stromules This analysis shows that ORY treatment decreased stromule-directed chloroplast movement and CTD disrupted nearly all chloroplast movement, including stromule-directed (Figure 8D). It is possible that stromule extension or retraction may provide the driving force for stromule-directed movement. Therefore, we quantified how many times stromule extension and retraction events occur in 10 mins after 1 hr of drug treatment. Interestingly, ORY treatment significantly increased retractions and reduced the number of extensions (Figure 8E); however, the remaining stromules extension showed a higher velocity (Figure 7C) suggesting the frequency rather than the velocity of stromule extension is with correlated chloroplast movement. Overall, these results suggest that ORY treatment caused the reduced stromule-directed chloroplast movement due to less extension events. Our data show that stromules may direct chloroplast movement in epidermal pavement cells; however, it remains unknown if stromules provide a driving force or only guide chloroplast movement.

Video 10
Chloroplasts and stromule dynamics following ORY and CTD treatments.

Chloroplast bodies and stromules were visualized in a NRIP1-CFP transgenic N. benthamiana plants, in combination with chlorophyll autofluorescence. Time-lapse confocal microscopy was used to acquire images every 10 s for 350 s, displayed at 16 frames/s. Displays chloroplast and stromule activity within plants subjected to either DMSO vehicle control (left), 1 µm of ORY (center), or 1 µm of CTD (right) drug treatments.

https://doi.org/10.7554/eLife.23625.043

The longer CTD treatment resulted in a complete disruption of AFs and nearly all chloroplast movement. Since chloroplasts are anchored to the AF network, we aimed to partially disrupt the AF network without fully abrogating all AF function. Treatment with CTD resulted in discontinuous AFs (Figure 8—figure supplement 2A, magenta) while the MTs were intact (Figure 8—figure supplement 2A, yellow). Examination of time lapsed maximum intensity projections of confocal micrographs showed that stromules were still present at 3 min and then briefly absent at approximately 8 min after CTD treatment (Figure 8—figure supplement 2A, Video 11). This brief disruption further supports that stromules are stabilized by AF anchors and disruption of AFs results in rapid retraction of stromules. However, despite the initial disruption, stromules re-extended along MTs and multiple stromules were observed after 20 min (Figure 8—figure supplement 2A). These observations explain why the disruption of stromules by CTD was missed during 30-min treatment (Figure 6—figure supplement 3). Tracking the stromule and chloroplast movement (Lu et al., 2017) showed that stromules can still direct chloroplast movement if AFs are only partially disrupted. One chloroplast (Cp1) had restricted movement and colocalized with an AF fragment (Figure 8—figure supplement 2A). Stromules were observed extending in opposite directions (Figure 8—figure supplement 2B). However, the second chloroplast (Cp2) did not co-localize with AF fragments (Figure 8—figure supplement 2A; Video 11). The stromule of this chloroplast not only extended, but its extension along the MTs facilitated a rapid pulling of the body of down the viewing plane (Figure 8—figure supplement 2C).

Video 11
Stromule directed movement in Cytochalasin-D-treated cell.

Stromules (cyan) from two chloroplasts are disrupted after Cytochalasin D treatment that fragmented AFs (magenta). Stromules re-extended along MTs (yellow) resulting in the movement of chloroplasts. 105 maximum intensity projections of time-lapse z-stacks were taken every 39 s and displayed at eight frames per second. The video was used to generate Figure 8—figure supplement 2.

https://doi.org/10.7554/eLife.23625.044

Actin microfilaments mediate perinuclear chloroplast clustering during plant immune response

Our previous findings indicate that N NLR immune receptor-triggered immunity to the TMV p50 effector resulted in stromule induction, stromule-to-nuclear connections and eventual perinuclear clustering of chloroplasts (Caplan et al., 2015). Electron microscopy results in our previous studies indicated that the chloroplast and nuclear membranes do not directly interact (Caplan et al., 2015), suggesting other cytoplasmic components are required for this interaction. To study the importance of cytoskeleton during the process of perinuclear chloroplast clustering, we expressed TMV-p50 to induce an immune response in N-containing NRIP1-Cerulean N. benthamiana transgenic plants (Caplan et al., 2015; Caplan et al., 2008). Since stromules extend along MTs, initially, we marked MTs and looked at stromules to nuclear connections, but we were unable to find significant connections of stromules to MTs around nuclei. Therefore, we next marked AFs with Lifeact-TagRFP and found connections between stromules and AFs surrounding nuclei (Figure 9). Time-lapse studies showed long stromules stably connecting to an AF attached to a nucleus for approximately 18 min (Figure 9A, arrowheads). After 18 min of continuous imaging, a long stromule retracted, bringing the chloroplast body close to the nucleus (Figure 9A, arrows; Video 12). We verified these results with another AF marker, mTalin-Citrine (Figure 9—figure supplement 1). We also observed that when the bodies of chloroplasts were in contact with nuclei, there were connections with AFs (Figure 9B–C, arrows).

Figure 9 with 1 supplement see all
Chloroplasts and stromules positioning during perinuclear chloroplast clustering during immune response.

(A) TMV-p50 effector and Lifeact-TagRFP (magenta; AFs) were expressed in transgenic NRIP1-Cerulean N. benthamiana plants that marks stromules (cyan). Time lapse images of stromule retraction toward the nucleus (N) after chloroplast positioning around the nucleus. A stromule tip (arrow head) remained stably associated with an AF associated with a nucleus and the body of the chloroplast was anchored away from the nucleus for 18 min. Stromule retraction from 18 to 30 min brought the chloroplast body (arrow) in close association to the nucleus. The body of the chloroplast was tracked (red line). Arrows indicate retracting stromule. Scale bar equals 5 µm. (B) TMV-p50 effector and Lifeact-TagRFP (magenta; AFs) were expressed in transgenic NRIP1-Cerulean N. benthamiana plants and then fixed as described previously (Caplan et al., 2015). Three interaction points of stromules with actin surrounding nuclei were detected (circles). The body of a chloroplast was also associated with perinuclear AFs (arrow). Image is a deconvolved maximum intensity projection of a confocal microscopy z-stack. (C) Enlargements of individual xy slices of the z-stack show connections of stromule tips (left) and a stromule kink point (top right) with AFs.

https://doi.org/10.7554/eLife.23625.045
Video 12
Stromule retraction resulted in the movement of a chloroplast to a nucleus.

A stromule (cyan) attached to an AF (magenta) pulls the chloroplast body to the nucleus by retracting. 100 maximum intensity projections of time-lapse z-stacks were taken every 38 s and displayed at 8 frames/s. The video was used to generate Figure 9A.

https://doi.org/10.7554/eLife.23625.047

Since p50-induced immunity leads to vigorous stromule induction (Caplan et al., 2015); Figure 5), we hypothesized that more chloroplasts might move toward nucleus by stromule-directed movement of chloroplast body. Therefore, we quantified the perinuclear chloroplast clustering during TMV-p50-induced immune response in N-containing NRIP1-Cerulean transgenic plants in a time course (Figure 10A–B). Although majority of nuclei had a low number of interacting chloroplasts in the control (Figure 10A, left panels), we observed a significantly higher number of chloroplasts around nuclei in TMV-p50-treated samples (Figure 10A, right panels). More than 80% of nuclei (85 out of 105) were surrounded by more than two chloroplasts in TMV-p50-treated samples compared to 50% of observed nuclei (56 out of 120) were surrounded by none or single chloroplast in the control (Figure 10A and Figure 10—figure supplement 1A). The ratio of nuclei-clustered with more than four chloroplasts was significantly higher in TMV-p50 treatment compared to the control (Figure 10B and Figure 10—figure supplement 1A). These results indicate significant induction of perinuclear chloroplast clustering during an immune response.

Figure 10 with 1 supplement see all
Perinuclear chloroplast clustering during immune response.

(A) Time course images of perinuclear chloroplasts during TMV-p50 effector induced immune responses compared to mock control. Chloroplasts and stromules marked by NRIP1-Cerulean were pseudocolored cyan while nuclei and cytoplasm were pseudocolored red. N, nucleus. Scale bar equals 20 µm. (B) Quantification of perinuclear chloroplasts during TMV-p50 effector induced immune response compared to mock control shown in A. Ratio of nuclei associated with more than four chloroplasts in TMV-p50 infiltrated cells (magenta bars) are compared to those in control cells (green bars). More than 92 nuclei were observed for each condition from 48 images. Experiments were repeated three times with four plants each. Data represented as the mean +standard deviation (SD), **p<0.001, *p<0.01 (Student’s t test with Welch’s correction). (C) Actin cytoskeleton disruption by treatment with cytochalasin D (CTD) lead to dissociation of chloroplasts near the nucleus while microtubule disruption by treatment with oryzalin (ORY) does not affect the chloroplast positioning. Perinuclear clustering of chloroplasts was induced by TMV-p50 effector. Images were acquired 36 hr after induction. Cytoskeleton Inhibitors were treated 35 hr after induction. Chloroplasts and stromules presented by NRIP1-Cerulean were pseudocolored cyan, while nuclei were pseudocolored red. N, nucleus. Scale bar equals 20 µm. (D) Quantification of cells associated with more than four chloroplasts in C. More than 60 nuclei were observed for each condition from 12 images. Experiments were repeated three times with two plants each (total 6 plants). Data represented as the mean +standard deviation (SD), **p<0.001 (Student’s t test with Welch’s correction). D, DMSO; O, oryzalin; C, cytochalasin D.

https://doi.org/10.7554/eLife.23625.048
Figure 10—source data 1

Statistics of the perinuclear clustering of the chloroplasts in TMV-p50 induced plant immune response and effect of cytoskeleton disrupting drugs.

Statistical analyses and the graph of the perinuclear clustering of the chloroplasts in TMV-p50 induced plant immune response and presented in Figure 10B and D were generated by Prism7.

https://doi.org/10.7554/eLife.23625.051

To determine, if AF anchoring plays a role in the immunity-induced perinuclear clustering of chloroplasts, we treated plants with CTD and ORY. Remarkably, CTD treatment significantly reduced the number of chloroplasts interacting with nuclei compared to the control and ORY treatment (Figure 10C–D and Figure 10—figure supplement 1B). These results support that anchoring of stromules to the AFs at the nucleus or more generally chloroplast movement is important for perinuclear clustering of chloroplasts during plant immune response. In conclusion, we propose a model in which perinuclear clustering of chloroplasts involves stromule anchoring to AFs surrounding nuclei and stromules guide chloroplasts toward nuclei during an immune response.

Discussion

Cytoskeletal elements in plant cells support several cellular functions, including cytoplasmic streaming, cell division, cell elongation, polar growth, vesicle trafficking, nuclear positioning and morphogenesis (Cai et al., 2015; Higa et al., 2014; Li et al., 2015). In this study, we show that dynamic stromules extend along MTs and AFs stabilize stromules and chloroplast-to-nuclear connections during innate immune response. Stromules have the ability to direct chloroplast movement, and AF anchoring of stromules may guide perinuclear chloroplast clustering during innate immunity.

Previous studies in non-green hypocotyls indicated a role for AFs and MTs during stromule formation (Kwok and Hanson, 2003, 2004a). The initial study (Kwok and Hanson, 2003) used cytoskeletal inhibitors to implicate AFs and MTs during stromule formation, suggesting that AFs promote while MTs restrict stromule and plastid movement. Stromules visualized by differential interference contrast were observed interacting directly with AFs labeled with GFP-hTalin and rearrangements of the AF network changed stromule morphology (Kwok and Hanson, 2004a). Movement along AFs was indirectly implicated by the identification of myosin XI cargo domain and a small tail domain that localize to chloroplasts (Natesan et al., 2009; Sattarzadeh et al., 2009). Knockdown of myosin XIs qualitatively disrupted stromules (Sattarzadeh et al., 2009) or quantitatively decreased the percent of plastids with stromules (Natesan et al., 2009). However, the dynamics of stromules moving along AFs were not examined in these studies (Natesan et al., 2009; Sattarzadeh et al., 2009). Furthermore, longer myosin XI tail domain (Reisen and Hanson, 2007) and full-length myosin XI (Avisar et al., 2008) do not localize to the chloroplasts. These studies prompted us to conduct a detailed time-lapse confocal microscopy of the dynamics of stromules and AFs. Surprisingly, our extensive investigations were unable to show extension of stromules along AFs. Instead, we discovered stromules were statically anchored to the AF network. Stromules were previously shown to actively move beyond AF attachment points via an unknown mechanism that was proposed to be either collisions with other components of the cytoplasm or interactions with very fine AFs (Kwok and Hanson, 2004a). Here, we have revealed that this unknown mechanism to be stromule extension along MTs by simultaneously monitoring MTs labeled with GFP-TUA6, AFs labeled with Lifeact-TagRFP and stromules labeled with NRIP1(cTP)-BFP via time-lapse confocal microscopy (Figure 8B). When our third revision version of the manuscript was under review, another report has proposed a model in which both stromule extension and slow anchoring occurs on MTs and rapid extension occurs on AFs (Erickson et al., 2017b). Our high-resolution imaging data clearly indicate that stromules do not extend along AF. Our data shows that static stromule anchoring is associated with the AF network and dynamic movement occurs along MTs. We used both Lifeact and mTalin, since they may label different pools of AFs. Lifeact results in even labeling of fine AF network and is accepted as one of the best markers for AF (Riedl et al., 2008). However, mTalin was used previously for labeling cp–actin interacting with chloroplasts and required for blue-light mediated movement (Kadota et al., 2009). We have found that each marker has its own advantages and disadvantages, and no single marker is perfect. The actin inhibitor CTD disrupted the actin network, briefly destabilizing stromules which then dynamically re-extended along MTs. It is possible that myosin XI silencing is causing a similar effect, since knockout of myosin XI in Arabidopsis resulted in inhibiting distribution and dynamics of actin network (Cai et al., 2014; Park and Nebenführ, 2013).

Early studies examining the role of MTs during stromule formation were conducted with MT inhibitors, APM or Oryzalin, leading to the conclusion that MTs have a limited role, because disruption resulted in either a 25% reduction in stromule length in hypocotyls treated with 5 µM of APM (Kwok and Hanson, 2003) or no alteration of stromules in Nicotiana leaves treated with 36 µM of oryzalin (Natesan et al., 2009). A recent study also shows that stromules remained extended after 100 µM Oryzalin treatment (Erickson et al., 2017b). We show that both 20 µM APM and 300 µM Oryzalin can disrupt stromules in Nicotiana leaves; and propose that the difference between these studies may be caused by differences in either cell type or inhibitor concentration. The plastids in epidermal pavement cells in Nicotiana are chloroplasts (Barton et al., 2017), compared to dark grown hypocotyls that lack chlorophyll-containing plastids (Kwok and Hanson, 2003). In general, the formation of stromules may vary based on differences in cell, plastid, or stimulus type. We found that 20 µM of APM or 300 µM of oryzalin MT inhibitor was required to observe a more complete disruption of MTs. By monitoring MTs and stromules with fluorescently tagged markers, we were able to directly observe the effect of inhibitors on MT formation and found that stromules maintained interactions with small fragments of MTs, but retracted after complete disruption of MTs. The study using 100 µM Oryzalin also observed that stromules would remain associated to small fragments of MT (Erickson et al., 2017b). Although they did not quantify changes in stromule frequency or dynamics like we describe here, they qualitatively observed more fast moving, short-lived stromules. This is consistent with the overall increased stromule extension velocity that we measured after 1 µM Oryzalin treatment. The role of MTs during stromule extension is also supported by time-lapse confocal microscopy that shows stromule tips interacted and dynamically extended along MTs. This is consistent with the recent study also showing stromule extension along MTs using mOrange2-MAP4 (Erickson et al., 2017b). The involvement of MTs was unexpected because of the previously implicated role of AFs; therefore, we repeated these experiments with three independent MT markers, GFP-TUA6, EB1-Citrine, and TagRFP-MAP-CKL6.

To rule out potentially indirect effects of MT drugs, we further examined the mechanistic function of MTs during stromule formation by stabilizing MTs either chemically or genetically. Taxol, which stabilize MTs (Schiff and Horwitz, 1980), doubled the average number of stromules per chloroplasts. Furthermore, we altered MTs genetically by silencing NbGCP4. The γ-tubulin forms a complex with γ-tubulin complex protein (GCP) such as GCP2-GCP4 to form γ-tubulin ring complex (γ-TuRC) that plays an important role in MT nucleation and organization (Moritz and Agard, 2001). GCP4-GCP6 subunits are not essential for γ-tubulin complex (Vinh et al., 2002), but these subunits are important for stabilizing the ring complex (Guillet et al., 2011). Knockdown of GCP4 in Arabidopsis leaf pavement cells resulted in hyper-parallel bundles of MT (Kong et al., 2010). Our results showed that NbGCP4 silencing in N. benthamiana leaves exhibited similar changes in MT organization via SOAX analysis. The change in MT structure induced by NbGCP4 silencing was sufficient to induce stromules constitutively. Increased stromule length in NbGCP4-silenced plants could be due to less dynamic stromules, since extension and retraction velocities decreased. It is possible that the decrease in stromule dynamicity is caused by a disrupted balance between MT branching and MT bundling in NbGCP4 silenced plants. These findings support that MT dynamics are a key regulator of stromule formation and dynamics.

It has been proposed that stromules may extend via an internal force, and not along MTs or AFs. Early studies have found filament-like structures in plastids and stromules, which potentially could provide an outward force (Bourett et al., 1999; Lawrence and Possingham, 1984). Recently, stromules were shown to form in vitro from isolated chloroplasts (Brunkard et al., 2015); but, clean chloroplast preparations resulted in only 1.1% of chloroplasts having short, spontaneous stromules and a 40-fold increase in stromules after the addition of cell extracts (Ho and Theg, 2016). We also have observed rapidly moving beak-like or small protrusions (Video 2, red dots) that do not interact with MTs. They also resemble ‘chloroplast protrusions’ observed in alpine plants that form independently of MTs (Buchner et al., 2007; Holzinger et al., 2007a; Moser et al., 2015). It was recently proposed that the small, fast moving stromules moved along actin because their rate of extension was similar to myosin motors (Erickson et al., 2017b). However, when AFs were marked with Lifeact-TagRFP, we did not observe a correlation of these stromules extending along AFs (N = 73). These studies combined suggest that there may be an alternative mechanism for stromule initiation that may depend an internal force. Another alternative to cytoskeleton driven stromule formation is force derived from membrane contact sites (MCS) with the ER (Schattat et al., 2011). Stromule and ER dynamics are correlated and it is possible that MCS stabilize stromules similar to actin anchors. They propose a model in which ER MCSs or the underlying cytoskeleton dictate stromule dynamics (Schattat et al., 2011). Our findings strongly supports a role for the cytoskeleton in which stromules require a combination of MT and AF interactions. Nonetheless, the function of the MCS between stromules and ER is intriguing and may assist in transfer of proteins, lipids or small molecules.

Blue-light-induced-chloroplast movement in plants is driven by chloroplast actin filaments (cp-actin) (Kadota et al., 2009). The chloroplast unusual positioning 1 (CHUP1) protein recruits actin to the leading edge of chloroplasts and is required for movement. Interestingly, the N-terminal coiled-coiled domain of CHUP1 is also required to anchor chloroplasts to the plasma membrane, revealing a complex, dual role of actin during chloroplast movement and anchoring (Oikawa et al., 2008). Chloroplasts are held by a cage of AFs (Kandasamy and Meagher, 1999) and additional recruitment of cp-actin via CHUP1 potentially may inhibit stromules by forming a physical constraint. Consistent with this hypothesis, CTD treatment disrupts actin around chloroplasts causing them to lose their ellipsoid shape to become round (Figure 8—figure supplement 2). Once released, chloroplasts moved in the direction of stromules extending along MTs. However, complete disruption of actin using longer treatments of CTD resulted in a complete disruption of all chloroplast movement, including stromule-directed movement. Stromule-guided movement was also seen without CTD treatment (Figure 8) and appears to be a novel type of organellar movement along MTs. Over 50% of all the chloroplast movement in steady-state epidermal pavement cells was stromule-directed, suggesting that this type of movement may significantly contribute to chloroplast movement and positioning. Interestingly, stromules in the green algae, Acetabularia, also have been implicated in chloroplast movement (Menzel, 1994), suggesting that both cp-actin and stromule-directed chloroplast movement are conserved between land plants and green algae (Suetsugu and Wada, 2016).

We have recently shown that stromules play an important role during innate immunity and programmed cell death (Caplan et al., 2015). During an immune response, chloroplasts move toward the nucleus and different types of chloroplast stromule-to-nuclear connections are established (Caplan et al., 2015). However, the mechanism behind perinuclear chloroplast clustering and chloroplast stromule-to-nuclear interactions is unknown. Our results described here using TMV-p50-induced, effector-triggered immunity indicate a role for AFs and MTs during perinuclear clustering. MTs promote stromule extensions, contributing to more stromule movement, while AFs provide anchors to position chloroplasts towards the nucleus. These results reinforce the role for AFs as anchor points for stromules that were also previously shown to exist in Arabidopsis hypocotyls (Kwok and Hanson, 2004b), and a recent study showing that stromules are involved in maintaining contact with nuclei (Erickson et al., 2017a). Overall, our results invoke a model in which, during effector-triggered immunity, MTs facilitate stromule extensions and stromules bind tightly to AFs around nuclei. The role of MTs during the formation of stromule-to-nuclei connections requires further studies. However, our data suggests that once those connections are formed, stromules may guide or pull chloroplasts toward the nucleus, which then results in perinuclear clustering of chloroplasts.

Results described here show that MT-mediated stromule extension and AF-mediated stromule anchoring are two complementary activities during stromule formation and movement. We provide mechanistic insights into how interactions with the cytoskeleton form and stabilize stromules. Furthermore, we describe a new type of organellar movement along MTs that is stromule-directed and reveal a mechanism for perinuclear clustering during innate immunity. In the future, it will be interesting to investigate the molecular components required for stromule dynamics and stromule-directed movement and importance of perinuclear chloroplast clustering during innate immunity.

Materials and methods

Key resources table
DesignationSource or referenceIdentifiersAdditional information
NRIP1PMID: 18267075; 26120031
GCP4This paper
p50PMID: 18267075; 26120031
GV2260PMID: 4022773
GV3101https://doi.org/10.1007/BF00331014
NRIP1-Cerulean transgenic plantPMID: 18267075; 26120031
GFP-TUA6 transgenic plantPMID:12084822
FABD2-GFP transgenic plantPMID:12084822
NRIP-Cerulean; N double transgenic plantPMID: 18267075; 26120031
TRV1PMID: 14501071
TRV2-EVPMID: 14501071
Nicotiana benthamianaTaxonomy ID: 4100
mTalin-CitrineThis paperask construct named ‘SPKD2681’
Lifeact-TagRFPThis paperask construct named ‘SPKD2209’
TagRFP-MAP-CKL6This paperask construct named ‘SPKD2386’
NRIP1(cTP)-TagBFPThis paperask construct named ‘SPKD3168’
TRV-NbGCP4This paperask construct named ‘SPKD3111’
Citrine-p50-U1This paperask construct named ‘SPKD1939’
CitrineThis paperask construct named ‘SPKD914’
p50-2xHAPMID: 18267075; 26120031
NLS-mCherryPMID: 28619883
Cytochalasin DSigma-Aldrich, St. Louis, MOC8273
Aminoprophos-methyl (APM)Sigma-Aldrich, St. Louis, MO03992
OryzalinSigma-Aldrich, St. Louis, MO36182
Paclitaxel-BODIPYThermofisher Scientific, Waltham, MAP7501
DMSOSigma-Aldrich, St. Louis, MOD8418
PRISM7GraphPad, La Jolla, CA
Stromule detection and tracking algorithmhttp://sigport.org/1807

Plasmids

Plasmids used in this study includes mTalin-Citrine (SPDK2681), Lifeact-TagRFP (SPDK2209), TagRFP-MAP-CKL6 (SPDK2386), NRIP1(cTP)-TagBFP (SPDK3168), TRV-NbGCP4 (SPDK3111), Citrine-p50-U1 (SPDK 1939), Citrine (SPDK914), and p50-2xHA (TBS44), NLS-mCherry. These were constructed by PCR and standard cloning methods. Details of constructions are available upon request.

Transgenic marker lines and transient expression by agroinfiltration

Request a detailed protocol

Transgenic N. benthamiana plant expressing the NRIP1-fused to Cerulean is described in (Caplan et al., 2015; Caplan et al., 2008). Transgenic N. benthamiana plants expressing GFP-TUA6 and FABD2-GFP were gifts from Drs. Manfred Heinlein and Karl Oparka and described in Gillespie et al. (2002). The plants were grown under continuous light at 20°C on growth carts for 4–5 weeks as described in Caplan et al. (2015), (2008). Cultures of GV2260 Agrobacterium containing the recombinant plasmids were grown on plates containing Streptomycin (50 mg/L), rifampicin (25 mg/L), and carbenicillin (50 mg/L) and spectinomycin (100 mg/L) antibiotics. Agrobacterium was resuspended in infiltration media containing 10 mM MgCl2, 10 mM 2-Morpholinoethanesulfonic acid (MES) and 200 μM acetosyringone and induced for at least 3 hr. Fully expanded leaves of 3- to 4-week-old N. benthamiana were used for agroinfiltration as described in Caplan et al. (2015, 2008).

Inhibitor treatment

Request a detailed protocol

Actin inhibitor Cytochalasin D (200 µM), microtubule inhibitors APM (20 µM) and Oryzalin (300 µM) and the microtubule stabilizing agent, Paclitaxel-BODIPY (0.8 nM) were prepared as 1M stocks in dimethyl sulfoxide (DMSO) and suspended at appropriate working concentrations in the infiltration medium prior to pressure infiltration for imaging. Working concentrations were determined after testing a range of concentrations of the respective inhibitors and agents. The concentrations that resulted in the microtubule depolymerization/stabilization without any lethal effect on the cells at the microscopic level were used further for experiments. Inhibitor treatments were performed by pressure infiltration. A small hole was made on the underside of the leaves with a razor blade. A 1-mL syringe was used to pressure infiltrate inhibitor solutions or a mock containing DMSO (≤0.2%) in infiltration media Leaf excisions approximately 4 mm2 were taken away from the infiltration point and mounted in a Nunc coverglass bottom chamber (Thermo Fisher Scientific). The center of the sample was imaged to minimize effects caused by excision-induced wounding. All time points started immediately following the pressure infiltration of the treatment. The 0–5 min time point after the respective treatments accounts for the time taken for sample preparation and mounting the samples after infiltration with the inhibitors and stabilization agents.

For 1 hr treatment, Cytochalasin D (10 µM) and Oryzalin (1 µM) as well as DMSO (0.1%) as a control were infiltrate in an area of about 3 cm diameter on the same leaf by needleless syringe infiltration. After 1 hr, around 4 mm2 leaf disc away from the infiltrated point were excised and mounted in a Nunc coverglass bottom chamber.

VIGS assay

Request a detailed protocol

NRIP1-Cerulean or GFP-TUA6 N. benthamiana transgenic plants were used for VIGS experiments as described in (Dinesh-Kumar et al., 2003). Agrobacterium culture containing pTRV1 was mixed with culture containing TRV2-EV, or TRV2-NbGCP4 in 1:1 ratio to adjust an OD600 to 0.5. Plants of 6 leaf stage were infiltrated and observed their stromules were observed in leaf epidermis 4 days after infiltration of VIGS vectors. In immune response experiments, Agrobacterium culture containing TMV-p50 effector was infiltrated on the third day after VIGS construct infiltration. A total of 48 images were taken from 12 plants by three independent experiments for each condition. Real-time RT-PCR was performed to determine the silencing efficiency. After imaging, RNA was extracted from leaves by plant RNeasy kit (Qiagen) and cDNA was generated by reverse transcription using Superscript III Reverse Transcriptase (Thermo-Fisher Scientific). Real time PCR was performed on a Bio-Rad CFX96 touchTM real-time PCR detection system (Bio-Rad) using iTaq Universal SYBR Green Supermix (Bio-Rad). GCP4-F-realtime 5’-GGATGGTTCATCTCATCAGC-3’ and GCP4-R-realtime 5’- AACAACAAGCTGCCACAGAT-3’ were used for NbGCP4 gene expression while EF1α-F-Realtime 5’-CTGGTGTCCTCAAGCCTGGTATGG-3’ and EF1α-R-Realtime 5’-TGGCTGGGTCATCCTTGGAGTTTG-3’ were used as for control PCR.

Perinuclear clustering of chloroplasts

Request a detailed protocol

To count chloroplast clustering under immune response, two leaves of N and NRIP1-cerulean transgenic N. benthamiana were infiltrated with agrobacterium containing citrine 48 hr prior to imaging. On the same leaf, non-recombinant cell or cells containing p50-HA were infiltrated 24, 30, or 36 hr before observation. 4 mm2 leaf tissues away from the infiltrated point were excised and imaged by a confocal microscope. To examine MT structure during immune response, transgenic N. benthamiana plants containing N and NRIP1-Cerulean or without N and NRIP1-Cerulean were infiltrated with Agrobacteria containing p50-HA and TagRFP-MAP-CKL6 36 hr prior to imaging. For the cytoskeleton inhibitor treatment after inducing immune response, transgenic N. benthamiana plants containing N and NRIP1-Cerulean were infiltrated with a mixture of Agrobacteria containing p50-HA and NLS-mCherry were infiltrated 35 hr before inhibitor treatment. Inhibitors were infiltrated one hour prior to the imaging.

Confocal microscopy

Request a detailed protocol

N. benthamiana leaf sections (4 mm2) away from the infiltrated point were excised, infiltrated with water and imaged on a Zeiss LSM 780 upright confocal microscope, LSM 710 inverted confocal microscope or LSM 880 inverted confocal microscope fitted with 40X C-Apochromat water immersion objective (NA = 1.2) (Carl Zeiss Inc, Thornwood, NY). The 405 nm, 458 nm, 488 nm, 514 nm, or 561 nm laser line was used for TagBFP, Cerulean, GFP, Citrine, or TagRFP, respectively. TagBFP and Cerulean were pseudo-colored cyan, Lifeact-TagRFP and mTalin-Citrine were pseudo-colored magenta, and GFP-TUA6, EB1-Citrine, and TagRFP-MAP-CKL6 were pseudo-colored yellow throughout the manuscript. In the perinuclear clustering experiment, Citrine for cytosol and nucleus diffusion was pseudo-colored blue and mCherry with nuclear localization signal was pseudo-colored blue for consistency of data presentation.

Image processing

Request a detailed protocol

Huygens Professional (Scientific Volume Imaging, Hilversum, Netherlands) was used on the majority of images to deconvolve using a Classical Maximum Likelihood Estimation (CLME) restoration method, to remove drift using the object stabilizer algorithm, to correct photobleaching across time lapsed images and to equalize brightness and contrast. Noise was removed from images that were not suited for deconvolution using a 3 × 3 median filter. Volocity (PerkinElmer, Waltham, MA) was used to generate images, kymographs and videos.

SOAX microtubule analysis

Request a detailed protocol

Bio-filament analyzing program SOAX (Xu et al., 2015), which utilizes multiple Stretching Open Active Contours (SOACs), was used in order to determine Curvature, Length and Azimuthal Angles for MT filaments within Maximum Intensity Projections (MIP) of epithelial leaf cells. High-resolution z-stacks were acquired on an LSM 880 confocal microscope or an LSM 710 confocal microscope and deconvolved in Huygens Professional batch conversion, with Regularization per channel decreased to a minimum of 2, and Quality Change Threshold changed to 0.05. Resulting images were then converted into MIPs using Fiji (ImageJ) and analyzed. Regions were selected from five maximum intensity projections for each treatment, toward the central region of epidermal pavement cells which poses clear MT network. Regions were uniform in a radius of 10 µm from the center point, and minor errors in regional snakes were corrected. Following this, each region was analyzed using curvature and snake length analysis. Points of high filament visibility and quality were analyzed within each cell. Curvature and Snake Length were then compiled, while Azimuthal Angles were converted to Mean Resultant Lengths for statistical analysis. All settings for SOAX analysis were kept at program defaults excluding Ridge Threshold, increased to a maximum of 0.04 with a minimum of 0.02, and Stretch Factor, increased to 1. Results were compiled and graphed with Prism 7 (GraphPad). Azimuthal angle color-coding was performed on SOAX analyzed images to display the orientation of MTs.

Quantification of stromules

Request a detailed protocol

Stromules were manually counted using ImageJ (National Institutes of Health, Bethesda, Maryland, USA) from the maximum intensity projections of the confocal images. Mean stromule ratios were determined by counting the total number of stromules and then dividing by the total number of chloroplasts. To quantify perinuclear clustering, Agrobacterium-containing p35S::Citrine T-DNA vector was infiltrated into N. benthamiana NRIP1-Cerulean transgenic plant leaves as described in Caplan et al. (2015, 2008). Twenty-four hours later, either Agrobacterium-containing TMV-p50 or empty vector control was infiltrated. Images in Z series were captured by confocal microscope as described in Caplan et al. (2015) at the indicated time points. Perinuclear chloroplasts were counted manually with the cell counter plugin in ImageJ. Experiments were repeated three times with similar results and graphed with Prism 7 (GraphPad).

Automated stromule tracking

Request a detailed protocol

Matlab code was written to perform Fuzzy c-means clustering (FCM), active contour framework, contour smoothing, unit normal feature analysis and branch analysis. In the FCM, we utilize both spectral energy and spatial energy functions for clustering. A 5 by five window around each pixel is used to compute the spatial component. In our experiments, we clustered the spectral domain into eight clusters and compute coefficients for each pixel. Pixels having 30% coefficients as background and 70% as the foreground were then used in the active snake formulation. Matlab code is also written to perform tracking stromule. Segmentation is first performed in 3–4 layers from total 8 layers of z stack. Results are then projected into one image to perform nearest neighbor based tracking.

Manual stromule tracking

Request a detailed protocol

Sixty frames of time-lapse z-stack images of stromule dynamics were acquired every 10 s in NRIP1-Cerulean transgenic plants silenced for NbGCP4 or vector control with and without the TMV-p50 effector. Maximum intensity projections of time-lapse z-stacks were generated in Zen software (Carl Zeiss) and motion types including, extension, retraction, constant smooth, sudden erratic, and side tangential were manually counted. The maximum and minimum stromule lengths were manually measured using the FIJI version of ImageJ (Schindelin et al., 2012). The extension and retraction velocities were calculated using the Cell Counter plugin in FIJI ImageJ, via frame-by-frame analysis. This allows quantification of movement of stromules between frames of a temporal stack, in 2D and 3D.

Transmission electron microscopy

Request a detailed protocol

Transmission electron microscopy was conducted as described previously in Caplan et al. (2015). Leaf excisions were fixed with 2% paraformaldehyde and 2% gluaraldehyde in PHEM (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, and 2 mM MgCl2, pH 6.9) buffer for 45 min) overnight at 4°C. Samples were washed with 0.1 M sodium cacodylate buffer (pH 7.4), postfixed with 1% osmium tetroxide in the same buffer for 2 hr, and then washed with buffer and water. Samples were dehydrated in an acetone series (25%, 50%, 75%, 95%, and twice in anhydrous 100% acetone; 30 min each step) and infiltrated with Quetol 651-NSA resin. Ultrathin serial sections were cut on a Reichert-Jung Ultracut E ultramicrotome and collected onto a film of 0.5% formvar using 2 × 1 single slot grids. Sections were post-stained with methanolic uranyl acetate and Reynolds’ lead citrate and examined with a Zeiss Libra 120 TEM operating at 120kV. Images were acquired with a Gatan Ultrascan 1000 2k × 2 k CCD.

Statistical analysis

Request a detailed protocol

Statistical analysis was performed using Microsoft Excel 2013 (Microsoft) and Prism 7 (GraphPad). Stromule counts were performed on 3–4 images obtained at the appropriate time points depending on the drug treatment. Experiments were repeated at least three times. For experiments involving the Paclitaxel-BODIPY treatment each image was considered a replicate and the experiment was repeated three times. Student’s t-test with Welch’s correction was performed to examine difference between treatments. For stromule frequency, the results passed the D’Agostino and Pearson’s normality test. Thus, t-test with Welch’s correction was used to evaluate the differences. For the stromule lengths, rank transformation was performed and Mann-Whitney test was used for comparison. Comparisons of velocities of stromule extension and retraction between all the conditions were done using Dunnett’s multiple comparison. For the perinuclear clustering, non-parametric Mann-Whitney t-tests were performed to evaluate the differences. All graphs were formed with Prism 7 (GraphPad). Statistical analyses and graph generations were performed using Prism 7.

References

    1. Köhler RH
    2. Hanson MR
    (2000)
    Plastid tubules of higher plants are tissue-specific and developmentally regulated
    Journal of Cell Science 113:81–89.
  1. Book
    1. Kumar A
    2. Dinesh-Kumar SP
    3. Caplan J
    (2014) Stromules
    In: Theg S, Wollman F.-A, editors. Plastid Biology. Springer. pp. 189–207.
    https://doi.org/10.1007/978-1-4939-1136-3_7
    1. Lawrence ME
    2. Possingham JV
    (1984)
    Observations of microtubule-like structures within spinach plastids
    Biology of the Cell 52:77–82.
  2. Book
    1. Sokal RR
    2. Rohlf FJ
    (1995)
    Biometry: The Principles and Practice of Statistics in Biological Research
    New York, United States: W H Freeman. ISBN 978-0-7167-2411-7.
  3. Book
    1. Takeuchi M
    2. Staehelin A
    3. Mineyuki Y
    (2017)
    Actin-Microtubule Interaction in Plants
    In: Jimenez-Lopez J, editors. Cytoskeleton - Structure, Dynamics, Function and Disease. InTechOpen. pp. 33–54.

Decision letter

  1. Jean T Greenberg
    Reviewing Editor; University of Chicago, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: this article was originally rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]

Thank you for submitting your work entitled "Stromule extension along microtubules coordinated with actin-mediated anchoring guides perinuclear chloroplast movement" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife in its present form (see details below).

The reviewers and reviewing editor discussed the manuscript and the reviews in detail. They concluded that, while there were some interesting observations, significantly more work would be needed to solidify the claims (see the reviews). A point that came up in the discussion is that some data are over-interpreted, such as the data to suggest that stromule extension requires MT's. The data presented indicates that application of MT depolymerizing drugs reduces the number of observed struggles per plastid. This observation indicates a possible role in stromule maintenance, not a requirement in their extension. Keeping with the eLife spirit of not inviting revision when the requested experiments will likely take longer than a couple of months, we are declining the work. However, we are interested in the topic. A thoroughly revised manuscript that addresses the reviewers' concerns would likely be reviewed again. While it will be treated as a new submission, we would try to solicit advice from the same reviewers.

Reviewer #1:

I have reviewed the manuscript by Kumar et al. on " Stromule extension along microtubules coordinated with actin-mediated anchoring guides peri-nuclear chloroplast movement." The researchers have used a number of probes to establish the role of microtubules and provided some excellent images. The portion dealing with the cytoskeleton is generally well done and I commend the researchers for it. However, I do have major concerns that must be addressed in order to make the work make a higher impact.

1) The researchers use 4mm2 portions of tobacco leaves (that have been previously pressure infiltrated with different solutions). I am concerned that wounding of the leaf tissue plays an important role in their conclusions. Although controls have been used and I am not questioning the observations as such, it would be a much more convincing observation if portions of intact leaves (or large wounding-free areas) were imaged.

2) The Materials and methods portion is very sketchy and does not provide sufficient detail to allow the experiments to be repeated and verified by other labs. The researchers/manuscript would greatly benefit from clear statements of how things were actually done in this study, rather than referring to earlier publications for methodology.

3) It is important to know the time at which leaves were infiltrated (how many hours after exposure to light/daylight)?

4) Results section

"Beaks extend into longer stromules. Longer stromules were seen associated as just the tips of stromules bound to MTs or the full-length of the stromule extended along MTs. More complex stromule structures were observed associated with MTs." What does the “or” in the statement mean? While the stromule-tip-association suggests a good link to stromule extension, what happens when a stromule is totally aligned with a MT. Is this a coincidence or is there more to it? What are the more complex stromule structures?

5) 11% of stromules were not associated with MTs. The authors explain this by resorting to complete speculation and "MT-independent mechanism of stromule formation ". What are these other mechanisms and where is the evidence for them and their link to the observations presented here? If all stromule formation and extension is not based on MTs then I do not see the big discovery here. The observations presented then just become a biased subset that is chosen to make a point while anything that does not fit in is neglected.

6) The authors wished to 'conclusively demonstrate that stromules extend along MTs' and therefore used two inhibitors to disrupt the MTs. Their evidence for MT depolymerization was that stromules disappeared after 15 minutes. How can a specific assay be used and the observations then used as evidence that the assay is working? I find the reasoning faulty and wonder if this not a case of circular reasoning?

The stromules disappeared 15 minutes after APM and Oryzalin treatment: is this 15 minutes after pressure infiltration? Is it 15 minutes after the 4mm2 leaf excision was taken for viewing? What does disappearance mean? Did they retract? Were they not formed? It must be made clear what disappearance means.

7) The concentrations of inhibitors used are, in general, extremely high and would completely disrupt cellular functions, especially after pressure infiltrations. The authors provide minimal information on this by stating "The concentrations that resulted in the microtubule depolymerization/stabilization without any lethal effect on the leaves were used further for experiments." It must be made clear in the Materials and methods section what was done to ascertain that the cells were actually surviving and functioning normally after the harsh treatments.

8) The VIGS assay creates a weak point in this study. The time frame for the experiments involving VIGS is well beyond the induction period for stromules. The researchers actually observed plants four days after VIGS and found numerous anomalies including significantly altered / bundled cortical MT organization in leaves. They also found twice the number of stromules compared to VIGS vector control. The conclusion of high stromules number after 4 days in NbGCP4-silenced plants is at best just a correlation; it is not a direct evidence suggesting a clear cause and effect relationship. Perhaps the greatly altered MTs changed a lot of other things that could then then also have affected stromule frequency. the plant was clearly no longer operating in a normal manner. The same reasoning applies to the further increase in stromule length and number upon expressing p50 subsequently. In this case the observations were taken 24 hours later. Could something more have happened to create a physiological change in the 24 hour period? What happened at 4, 8, 12 hrs? That is definitely sufficient time for stromule formation to have occurred (Brunkard et al., 2015).

9) There are some really obtuse statements:

“Stromules appeared to interact statically, and not dynamically, with AFs, suggesting there are actin anchor points along stromules.” What is meant by 'interact statically and not dynamically'? actions. Is the word 'interaction' itself not denoting a dynamic phenomenon?

Further, the stromules became thin at anchor points – “We observed a clear thinning or constriction at the site of stromule-to-actin interaction points along the length and across the body of the chloroplast". Did the actin anchor create this thinning? As I understand the authors are talking about an anchor point and not a band of F-actin. If not, then how is the thinning of a stromule explained ? What is meant by “across the body of the chloroplast”? Are we still talking about thinning?

10) In a similar manner the authors state "At 8 min, the stromule began to re-extend along the MT, between 8 min and 9 min, the body of the chloroplast was released and this resulted in stromule driven chloroplast movement." What are the authors trying to convey? The body of the chloroplast was released! From what (?) and how did this result in a conclusion of 'stromule driven chloroplast movement'? The authors did not actually establish 'stromule driven movement'. Did the stromule pull the body or push it? Did the stromule elongate further or did it retract in order to drive chloroplast movement?

11) It is well observed that stromules arise in many different areas of a chloroplast randomly. Does this always coincide with chloroplast movement? The authors address this issue: "To examine how disruption of stromule and chloroplast AF anchoring effects chloroplast movement, we followed two chloroplasts in an epidermal cell treated with CytD." As I understand from the manuscript only one chloroplast was actually observed to reach the conclusion since the 1st one had restricted movement. Interestingly this singular observation is supposed to have added to the evidence that stromules may direct chloroplast movement in the absence of intact AFs.

12) The weakest part of this manuscript is the proposed link to 'innate immunity'. This part of the study takes the manuscript into a highly speculative mode where the nice demonstrations of microtubule alignments and the presence of actin are undone by a lack of rigorous experiments. The authors point to a feature of innate immunity as 'eventual peri-nuclear clustering of chloroplasts'. It is notable that chloroplast clustering around nuclei happens routinely in leaves, even without bringing innate immunity into play. In the excised 4mm2 sections what is the baseline number of chloroplasts clumping together and how is it accounted for by the idea of innate immunity? Perhaps the wounding of tissue is responsible for innate immunity? This still does not explain the chloroplasts around nuclei in normal, unwounded (or in any untreated tissue). Clearly stromules have been observed by the authors under conditions that trigger the innate immunity but I fail to see an exclusive response where a cause and effect relationship is indicated with regard to stromules.

Some references are missing and should be added. Reference for the presence of chloroplasts in tobacco pavement cells is needed.

The Introduction as well as discussion of earlier work on actin and microtubules must be done more thoroughly rather than be given a cursory mention. The authors must clearly state how their results advance the field and are not a repetition of earlier inhibitor experiments with a rather speculative link to chloroplast movement and plant immunity.

The authors have only looked at chloroplasts in pavement cells. For this study to have a real impact the observations must take into account other kinds of plastids, including plastids in mesophyll tissue and leucoplasts, all of which are known to display stromules.

Reviewer #2:

In this study Amutha Sampath Kumar and colleagues investigated the dynamics of stromules generation, elongation and retraction. The authors’ major discoveries indicate that stromules extend along microtubules (MTs) but not actin filaments (AFs). Moreover, they show that stromules extension and number per chloroplast depend on MTs integrity and stability. They also show a possible role for AFs as anchor points for stromules and plastid bodies around nuclei. Interestingly, they also provide evidence that stromules extending along MTs drive the movement and clustering of plastids to the nuclei during defense responses. This stromules/plastid behavior was previously reported by the same research group as being important for resistance against viruses (Caplan et al., 2015).

This work provides new and a significant information about the formation/regulation and function of the understudied plastid stromules. Although one can argue that the work it is generally based on correlative results, I consider that this data will be a keystone for future studies in this topic. I think that the manuscript fits with eLife's scope. However, I have some concerns about how the results are presented, described and its interpretation. Also, I consider that authors should support with more data the proposed model. Particularly, authors should address two things:

- The authors use the terms associated, bound, touching, interacting, etc. as "synonyms" but without specifying what they exactly mean. They should define these terms in a "precise" stromules-MTs distance(s) manner. Indeed, I do not think authors demonstrated interaction in any case. Authors should do 3D co-localization analyzes. In addition, they should consider doing FRET analysis if this is possible.

- Authors should analyze the dynamics changes of MTs and AFs networks after p50 defense induction. Are MTs stabilized after p50 expression? Are perinuclear AFs increased after p50? These results could be of great support for the proposed model. I believe the authors could do both analysis using the data that they already have.

The authors should also address the following comments:

Abstract:

1) Misspelling: effects should be affect.

2) Authors should revise and change the statement suggesting that AFs have no role in stromule extension. I think that providing "…anchor points for stromules to prevent their retraction while extending on MTs…" sounds like an important role.

Results:

1) Section "Microtubules are required for stromule extension". Authors should consider that the drugs used inhibit the increase of stromules per plastid associated to mock treatment instead of reducing them.

2) Section "Microtubule stabilization increases stromule number, length and stability", first paragraph: What are authors considering as MTs stabilization in Figure 2C? The image showing cells treated with paclitaxel should be label indicating this MTs stabilization. Are stromules longer in cells treated with paclitaxel?

3) Section "Microtubule stabilization increases stromule number, length and stability". Figure 3F: Authors point out that the stromules number increase observed in NbGCP4-silenced and in p50-induced plants are "remarkably similar". They should analyze the MTs stability during p50 induction to support this statement.

4) Section "Microtubule stabilization increases stromule number, length and stability". Figure 3I: Authors should describe and discuss the extension/retraction results obtained after treatment with p50.

5) Section "Stromules direct chloroplast movement in the absence of actin filaments", second paragraph: Although authors mention that exist "opposing forces" and a "rapid pulling" provided by stromules, they do not have any measure of these traction forces. Authors should be cautious with results interpretation and re-phrase this paragraph.

6) Section "Actin microfilaments mediate perinuclear chloroplast clustering during plant immune response", second paragraph: authors should replace the sentence "…connections, but were unable to find…" by "…connections, but we were unable to find…"

Discussion:

1) Third paragraph: "…and propose the difference…" should be replace by "…and propose that the difference…"

2) Third paragraph: What do the authors mean with the statement "directly interacted"? I do not think they showed any interaction.

Figures:

1) Figure 1A: How authors define tip connection or extension along MTs? How close are stromules and MTs? Are authors using z-stacks projections to analyze this distance? Authors should clarify this point in every image/quantification where an interaction/connection/touching, etc. suggestion is made.

2) Figure 1B-E: Are these analyses using only z-stack projections or they consider 3D/4D dimensions? Figure 1D: Is the algorithm calculating the 3D distance between stromules and MTs?

3) Figure 2B: Authors should show if exist statistical differences between times/treatments.

4) Figure 2C: What do the authors consider as stabilization of MTs in this image? They should do an analysis similar to the Figure 3.

5) Figure 2C-D: Is there a positive correlation between the increase on stromules number and the stromules "associated" to MTs?

6) Figure 5E: Authors should add the separated MTs and AFs channels to this figure.

7) Figure 3—figure supplement 2: Authors should add stats to this quantification. They should also consider using this data as a main figure.

Reviewer #3:

Higher plant plastids are observed to produce projections called stromules. It has been speculated that stromules may serve in exchange of metabolites or molecular signals to other organelles, including the cell nucleus. In a previous publication, the authors presented evidence that stromules may in fact aid in transport of proteins and reactive oxygen species to the cell nucleus in the context of innate immunity. In this study, the authors address cellular mechanisms by which stromules are extended, how they may interact with other components of the cell, and how stromule formation and dynamics may guide plastid movements and accumulation around the cell nucleus during innate immune response.

Previous studies have implicated both the actin and microtubule cytoskeletons in modulating stromule extension and dynamics, with actin emerging as the dominant player in stromule extension and plastid interaction with the cell periphery. Here, the authors acquire beautiful dynamic images of stromule dynamics showing compelling evidence for the guidance of stromule extension by cortical microtubules, conclude that such interactions contribute to plastid movement, extend previous observations of actin-implicated anchoring of plastid membranes at the cell cortex, and examine these mechanisms in the context of plastid accumulation at cell nuclei during the innate immune response. They propose a model whereby stromule extension mediated by microtubules helps to bring stromules to the nuclear periphery, where they are captured by surface-localized actin which can serve as attachment points for stromule mediated movement of plastids to the nucleus.

The Introduction and Discussion of the paper are well written, touching on the key literature. The microscopy is beautiful and sophisticated analyses are performed on some of the image data sets using tools that other plant biologists will be interested in knowing about and perhaps applying. A strength of the study are the time series images showing evidence for guidance of stromule dynamics by cortical microtubules, this result was nicely controlled for by using multiple markers for microtubules, making the possibility of an aberrant interaction between stromules and an overexpressed microtubule marker unlikely. However, there are issues with data interpretation that need to be addressed and I felt that a number of major and secondary conclusions were not supported well by the data presented.

A) The authors test the role of microtubules in stromule maintenance by applying depolymerizing and stabilizing drugs and by manipulating microtubule organization by genetic knock down of a nucleation complex protein. Previous studies had indicated that microtubules play a relatively minor role in stromule abundance, but here microtubule depolymerization resulted in a significant reduction in stromules per plastid compared to the mock control. This was the strongest section of the paper but I have a few comments.

1) The following point is not a major criticism, but an odd thing with the drug experiments was that stromules significantly increased in abundance in the mock control over 15 minutes. The reduction in stromule numbers with drug treatment observed at the same time were only slightly lower than the numbers observed at time 0-5 minutes after treatment. If just stromule numbers at time zero and 15 minutes were compared, one would not infer a strong effect of microtubules. The reason for the increase in stromule number in the mock is neither explored nor explained. A couple of possibilities are that stromule formation may be stimulated by stress caused by placing the tissue in the microscope slide mount, or perhaps that light used for the imaging stimulates stromule formation. Since it is an effect on the increased stromule numbers that is observed with drug treatment, it would be nice to have bit more insight into why these number are elevated over the course of the experiment. The two possibilities mentioned could be tested by not imaging at time point 1, and by determining if media exchange during imaging can mitigate the induced increase in stromules over time in the slide mount.

2) It was a good idea to manipulate microtubules by a secondary means, other than drug treatment. However, here the authors conclude that altered microtubule dynamics were responsible for observed reduction in stromule extension and retraction rates. Since microtubule dynamics were not measured, nor were observations shown of possible interaction between stromule tips and microtubule ends, it is not clear why this conclusion was reached. What is evident is that microtubule organization is altered in the GCP4 knock down, with an increase in parallel organization of microtubules in leaf epidermal cells, a result that has been shown previously for knockdown of GCP4 and also for GCP-WD/NEDD1. While the results are striking, it is not clear why a change in polymer organization would drive a change in stromule dynamics. My suggestion would be to keep the interpretation open and not ascribe it to microtubule dynamics. It is possible that the effect is indirect. For example, by altering levels of free MAPs due to reduced total lattice in these cells.

3) The authors conclude that microtubules play a direct role regulating stromules. While agree that the data shown are consistent with that idea, it is possible that the relationship is less direct, as the authors touch on in their Discussion when considering the ER and membrane contact sites.

4) The tip tracking method is very nice. It is kind of a shame though that it is not used to more rigorously quantify stromule tracking of microtubules. Only one example of such tracking is shown, with no summary measurements made over many cells, something that could be done with such a method. A further point here is that it should be defined more clearly how "on tube" vs. "not on tube" is determined, since the underlying molecular relationships are sub-resolution.

5) The SOACs analysis is used to assess features of polymer orientation, curvature and length. While this analysis produced a nice color-coded image of the orientation of segmented features, these data were not used to quantitatively assess patterns of polymer orientation, which again, seems a missed opportunity to show robustness of observations over many cells and samples. A second point, and an important one, is that by inspection of the raw data and the processed data, the method is not an appropriate means to assess polymer length, and those data should be dropped. It is not possible with such segmentation to tell if extended structures are composed a few long polymers or many shorter polymers, nor can it tease apart what happens were polymers meet and overlap at angles. Determining the length distribution in these arrays is an important goal in the field, but it is not trivial.

6) It is stated that GCP4 knock down results in greater bundling. This is not determined in the present manuscript.

B) The authors conclude that actin filaments do not extend along actin filaments but that actin filaments do serve as static anchor points for stromules, and also prevent stromule retraction.

1) The data for stromule extension along actin cables are not quantified, but presented as example images and a video. These images show little evidence for stromule extension along actin bundles, but to be robust, a method should be used to census the relationship of stromules and actin structures so that data from multiple cells and tissue samples can be assessed. Even so, it should also be at least discussed that actin features may be present that did not label with the probes used.

2) In the CytD experiments, why didn't stromule number increase from 0 to 15 minutes in the mock as was observed for the microtubule drug experiments? If this effect does not occur in every experimental setup, the microtubule drug experiments should be repeated under conditions where stromule number does not go up in the mock.

3) The evidence of static anchor points at actin filaments is weakly presented. A single kymograph is shown for analysis of stalling during stromule retraction, and in this single analysis it is not clear that the stall actually occurs at the position of actin signal. The signal corresponding to the stromule tip is offset from the actin signal. A much more rigorous analysis of retraction and actin overlap sites is required.

4) Likewise, the image sequence in Figure 5E and corresponding video are not convincing of retraction back to an actin-defined anchor point – the stromule tip appears to overshoot the "kink" position at the actin branch site. This is also a single image sequence example.

5) The analysis in Figure 5C looks more quantitative, but it is a less direct measure of interaction. Rather, it is measuring a predicted consequence if the proposed interactions exist. It is also poorly explained and it took some time to determine what the graph represents. I think the second bar was meant to be labeled "not fully retracted", rather than "actin anchored', which is a conclusion of function rather than an observation.

C) The authors conclude that loss of actin anchoring allows stromules to direct plastid movements.

1) Once again, these data consist of example images, not more global measurements across cells and tissue samples. In fact, just one image series example. It is a beautiful video, but it cannot be determined from this one example how generalized the phenomena shown are, nor how strong is the correlation of plastid movement with the direction stromule extension. A suggested means of analysis would be to assess each plastid displacement of x distance or greater in a set of time series acquired from multiple cells and leaves. For each displacement, stromule position and orientation would be determined and an orientational resultant would be estimated. The relationship between the predicted resultant angle of "pull" and the actual direction of movement could be plotted and analyzed.

2) A partial loss of actin structure by the CytB treatment is somewhat unsatisfying. The authors chose not to use LatB, which is more effective, because MT organization is also affected. However, I think these data should still be shown to ask if the observed movements of plastids are still seen when actin is more completely disrupted. It will be evident if stromule interaction with MTs (MT-dependent interaction sites) is also affected.

D) The authors conclude that actin anchoring mediate plastid accumulation at cell nuclei during the innate immune response and suggest a model by which plastid accumulation is facilitated by plastid movement driven microtubule- and stromule-based plastid motility, followed by actin mediated capture at the nucleus.

Another case of data by limited example. While the example shown in Figure 7D is compelling, more than this is needed. I was surprised that the hypothesis was not tested by actin disruption, for example. Further, it should be asked if plastid motility, and nuclear accumulation are significantly reduced if microtubules are depolymerized.

E) In general, the authors should be careful about distinguishing conclusion and interpretation from observation. This issue occurs throughout the manuscript.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Stromule extension along microtubules coordinated with actin-mediated anchoring guides perinuclear chloroplast movement during innate immunity" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Christian Hardtke as the Senior Editor. The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As you can see from the detailed reviews below, the reviewers very much appreciate your efforts to improve the initial manuscript. Yet they feel some additional adjustments are required before the paper can be published. These are mostly of organizational/semantic nature, and do not require additional experimentation. Apart from answering point by point to the reviews below, please pay particular attention to the following:

1) The reviewers are of the opinion that the analysis of retraction pausing with sites of actin filaments needs to be strengthened quantitatively. Currently, the data consist of example image sequences and an indirect measure involving pausing frequency, which is not explained very clearly. E.g., it is not obvious that this experiment correlates pausing with sites of possible actin action.

2) The reviewers believe that it would be a good idea to quantify tracking of microtubules by stromule tips rather than to show one example of such tracking.

3) Clarification is needed about how the circular correlation coefficient data relate to the results with the 30 degree cutoff angle. These data are new to the revised manuscript and need to be explained more thoroughly.

Reviewer #2:

The authors have addressed to most of my concerns. Also, they've clarified some other aspects raised by the other reviewers.

They included in this version new experiments improving the manuscript. Particularly, they added supporting results for the proposed model where they link the cytoskeleton, immunity and plastid movement/clustering around nuclei. I consider that this is new and significant information for the research field. In addition, authors included interesting quantification data for a more robust correlation between plastid movement and the direction stromule extension.

Nevertheless, one caveat is that authors are not showing quantitative data on the stromules tip tracking. The authors argue that the algorithm only works for the overexpressing EB1 construct, which could be changing the general MTs dynamics. I think this could also be true for the other OE constructs. Thus, authors should consider adding tip tracking experiment quantifications, and show the reproducibility for this method, at least for EB1.

Altogether, I think that although mostly based on correlative data, this is a foundational work for future investigations in this topic and should be published in eLife.

Reviewer #3:

All in all, the authors have addressed many of the concerns and comments made in the first review. In addressing these comments, a number of substantive additions to the manuscript have been made, such as quantifying the relationship of the angle of stromule extension and the direction of subsequent plastid movement. However, some concerns remain and there are also some new questions raised by the new datasets presented. The main issues concern the quantitation of actin anchoring of stromules (Figure 6), some follow-up questions on the new data quantifying the relationship between stromule extension angle and plastid movement, and the entire section on perinuclear clustering (should it be included?). Specifics are delineated below:

Point-by-point responses to authors' replies to initial review.

Point A1: In my first read, I missed the fact that DMSO treatment was associated with an increase in stromule number over time. This is perhaps because the media control was not shown in the figure for comparison. The revised text makes it more clear that DMSO causes an increase in stromule number over the media control, and that drug treatment reduces this increase. However, I am now wondering why the media controls are not shown in the figure. It would be useful to know if drug treatment also causes a reduction of stromules in the media control, even if there is no increase in time otherwise. If not, does this mean that induced stromules are somehow different from existing stromules in regard to microtubule depolymerization?

Point A2: The revised text addresses the concern raised about ascribing the changes in stromule dynamics to changes specifically in microtubule dynamics. However, I still think it would be appropriate to make clear that, while the manipulation here is pretty specific, the observed changes in stromule dynamics may be either a direct or indirect consequence of changes in microtubule organization. For example, the statement – "Collectively, these results indicate that change in MT organization in NbGCP4-silenced plants or during p50 induced immunity control stromule dynamics." – is stronger than I think the data warrant and should be restated. "Collectively, these results indicate that changes in MT organization caused by NbGCP4-silencing plants or during p50 induced immunity are correlated with changes in stromule dynamics, indicating a possible role, direct or indirect, for MT organization in modulating stromule dynamics."

Point A3:.The revised text addresses the stated concern.

Point A4:.Thanks for the clarification. I agree that the figure is useful for portraying a sense of dynamics in a static format. The original text did not make it clear that the figure was produced manually. Unfortunately, this is also not clear in the revised text. The manual method is described in the Materials and methods, but the text and figure legend do not indicate how this figure panel was made. I think a simple edit to the figure legend would fix this.

Point A5:.Revised text is an improvement.

Point A6:.OK.

Point B1: I think it is important to give a sense of how much space was searched through to say "almost never". For example, you could state that "at least x stromules were examined in at least y cells, and only n stromules showed any evidence of extending along actin filaments. Even in these cases…"

Point B2: OK.

Point B3: See comment for 5 below.

Point B4: See comment for 5 below.

Point B5: While more examples of stromules retracting to locations proximate to actin filaments, Figure 6D provides the only quantitation of this behavior. As indicated in the first review, it took me a while to figure out what I thought this panel represented. In reviewing this experiment again, I confess that I still find it unclear what exactly was measured here. The bar graph shows the percentage of fully retracted vs. partially retracted stromules. The figure legend states: "The percent of stromules containing actin anchors was quantified by monitoring retraction events in the samples described in panel A. Retracting stromules without actin anchors resulted in full retraction." However, in the text it states "70.3% + 0.02% (SEM) of retracting stromules partially retracted to one or more AF." So, was just the percent of partial retraction measured, or were retraction pause events also correlated with actin signal? If the latter, I might expect to see an analysis of how often retraction pausing was observed at sites of actin signal, but these data are not shown. If all partial retractions are indeed caused by actin anchors, then the percent of partial retraction alone would serve as an indirect estimate of the prevalence of stromules with actin anchors. However, is this known? Can stromules pause for other reasons? If they can, this experiment does not really probe prevalence of actin anchors well, nor does it test for the role of actin in pausing.

Point C1: The new data measuring the correlation of stromule angle with movement angle are an important addition to the manuscript. The angular correlation plot in the supplementary data is actually stunning. I would put it in the main figure. However, I think it should be stated more clearly how plastids were selected for this analysis. Were just plastids with one stromule selected? If so, how were they chosen? Also, a significant question is raised by the oryzalin experiments. In oryzalin treated cells, the circular correlation coefficient of stromule angle with the direction of plastid movement actually appears to go up (.85 vs.76 for no oryzalin). Yet when a 30 degree difference in stromule and movement angles is established as a cutoff for positive stromule directed movement, it is concluded that oryzalin treatment reduces stromule directed movement. It is hard for me to see how these two results are reconciled. If the correlation coefficient alone is considered, one would not conclude that oryzalin treatment reduces stromule directed movement in fact, it may even increase it.

Point C2: OK.

Point D1: It seems that the new drug tests during perinuclear clustering were useful. I agree that it may be challenging to interpret the results with cytochalasin D as it now turns out that actin is required for plastid movement and not just anchoring in these cells. Interestingly, this duality of actin being required for both movement and anchoring was also observed by Wada and colleagues for light driven plastid movement (it would be good to point this out in the Discussion). I am, however, a bit confused by continuing to propose a model where plastids are directed or guided during the early stages of the clustering response by MT's, since no effect on clustering by MT disruption was observed. In other words, the new data do not support all the particulars of the model. There seemed to be general agreement in the first review that this last experimental section was not a strength of the manuscript. I believe it may be a good idea to leave these studies out for now and have them be part of a future manuscript addressing specifically the mechanism of the clustering response.

Point E1: The text is generally improved in regard to clearly distinguishing between observation and interpretation.

Other comments and questions on the revised text:

Data questions:

1) Section “Microtubules are required for stromule formation and extension”: In relating the results that support a role for MT's in stromule extension, it is stated that stromules in the drug treated cells completely retract after 15 min. of treatment. If this is the case, can the authors comment on the data shown in Figure 3B, which indicate only a modest drop in the number of observed stromules in the drug treated cells from time 0' to time 15'? Also, it is not clear that these differences are significant. The standard errors for APM look like they mutually overlap the means at 2 sigma (0' to 15'), It is not clear what is going on with the stats for the oryzalin data.

2) Section “Actin filaments serve as anchor points but not as tracks for stromule extension”, last paragraph: Can the authors comment on the observation that 70% of stromules were observed to only partially retract, a state that is attributed to actin interaction, whereas the number of observed stromules in cells treated with 200 µM CTD was the same as in mock treatments? If actin interaction is the primary means of preventing complete retraction of stromules, it seems that the number of observed stromules would be expected to be lower in cells where actin is disrupted.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Stromule extension along microtubules coordinated with actin-mediated anchoring guides perinuclear chloroplast movement during innate immunity" for further consideration at eLife. Your revised article has been favorably evaluated by Christian Hardtke (Senior Editor), a Reviewing Editor, and two reviewers.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance. These are outlined below under reviewer #3. Additionally, since the work was in review, another paper appeared in the Plant Journal with similar results to yours. You should refer to the other study in your revised manuscript and point out any similarities/differences in the work (see doi: 10.1111/tpj.13813).

Reviewer #3:

The authors have addressed my previous concerns save one. At the end of the Results, the authors conclude:

"In conclusion, we propose a model in which perinuclear clustering of chloroplasts involves stromule extension along the MTs, stabilization of these extensions by anchor points to AFs surrounding nuclei and stromules guide chloroplasts towards nuclei during an immune response."

However, the authors reported that no connections with microtubules by stromules were observed during P50 induction of perinuclear clustering, and treatment with oryzalin had no effect on clustering. Thus, the experimental results do not support a role for guidance of stromules by microtubules in perinuclear clustering, indeed, they appear to contradict this possibility. The statement about microtubule guidance needs to be omitted from the proposed model for perinuclear clustering.

The description of the experiments addressing pausing of stromule retraction at AFs is now much more clear. However, the statistics in Figure 6D compare the rates of stromule pausing with the rate of full retraction. This comparison is relevant for asking if pausing is significantly more frequent than full retraction, but the main question here is whether pausing occurs at AFs more frequently than might be explained by chance alone. The rate of pausing at AFs by chance alone is a function of AF density; the higher the AF density, the more often pauses would be observed in association with them, even if there is no functional connection. Thus, AF density needs to be taken into account in a statistical test for the observed rate of AF-associated pausing (this can be a bit tricky due to the dynamics of the AFs). However, it seems pretty clear from the figures and the movies that the density of labeled AF's is lower than could easily explain a 77% rate of pausing at labeled AFs by chance alone (this would mean that, on average, about 3/4 of the stromule length would overlap AF signal). Even though the association is challenging to test formally, a statement to this effect might make more clear why the observed rate of 77% is a reasonable suggestion of an association.

https://doi.org/10.7554/eLife.23625.054

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

I have reviewed the manuscript by Kumar et al. on " Stromule extension along microtubules coordinated with actin-mediated anchoring guides peri-nuclear chloroplast movement." The researchers have used a number of probes to establish the role of microtubules and provided some excellent images. The portion dealing with the cytoskeleton is generally well done and I commend the researchers for it. However, I do have major concerns that must be addressed in order to make the work make a higher impact.

1) The researchers use 4mm2 portions of tobacco leaves (that have been previously pressure infiltrated with different solutions). I am concerned that wounding of the leaf tissue plays an important role in their conclusions. Although controls have been used and I am not questioning the observations as such, it would be a much more convincing observation if portions of intact leaves (or large wounding-free areas) were imaged.

We thank the reviewer for bringing up this concern, since it is something we considered in our experiments. We have done other experiments looking at the edge close to the cut site, and observed wounding effects 3-4 cells into the tissue. Therefore, our treatments were done on intact leaves and a 4mm2 leaf section was excised away from the infiltrated point (no damage on the leaf section was observed), mounted in the observation chamber, and only the center was imaged immediately. This approach minimized variation due to wounding.

We clearly defined these conditions in the Materials and methods subsections “Inhibitor treatment”,” Perinuclear clustering of chloroplasts” and “Confocal Microscopy”.

2) The Materials and methods portion is very sketchy and does not provide sufficient detail to allow the experiments to be repeated and verified by other labs. The researchers/manuscript would greatly benefit from clear statements of how things were actually done in this study, rather than referring to earlier publications for methodology.

It is routine to refer to previous work unless there is a modification to the previously described method. Therefore, we either described how the experiments were performed or referred to our previous publications in the Materials and methods section. In many cases, we cited our previous publications that describe the method, and also, we reiterate the details of the experiment. We did cite some methods without reiterating details, and in these cases, we repeated the details from previous publications in this manuscript.

3) It is important to know the time at which leaves were infiltrated (how many hours after exposure to light/daylight)?

We have stated this in the Materials and methods section. All experiments were performed in similar continuous light conditions (subsection “Transgenic marker lines and transient expression by Agroinfiltration”).

4) Results section

"Beaks extend into longer stromules. Longer stromules were seen associated as just the tips of stromules bound to MTs or the full-length of the stromule extended along MTs. More complex stromule structures were observed associated with MTs." What does the “or” in the statement mean? While the stromule-tip-association suggests a good link to stromule extension, what happens when a stromule is totally aligned with a MT. Is this a coincidence or is there more to it? What are the more complex stromule structures?

We have adjusted this statement in the text to state “the full-length and the tip of the stromule”, since we have never seen the stromule extending along the MT without the tip also associated. We did not elaborate on these two observations, since we did not have any data suggesting they were entirely unique or a coincidence of the angle of the tip association to the body of the chloroplast. Nevertheless, there is a possibility that this is significant; we therefore feel it is still important to report here our observations. In response to the “more complex stromules”, these were described as kinked or branched stromules in the next sentence. We rewrote the text accordingly to be clear in the revised version in the first paragraph of the subsection “Stromules interact and extend along microtubules”.

5) 11% of stromules were not associated with MTs. The authors explain this by resorting to complete speculation and "MT-independent mechanism of stromule formation ". What are these other mechanisms and where is the evidence for them and their link to the observations presented here? If all stromule formation and extension is not based on MTs then I do not see the big discovery here. The observations presented then just become a biased subset that is chosen to make a point while anything that does not fit in is neglected.

Our intention was certainly not to provide a biased subset, but rather, to report the observations of stromules interacting with MT. Most induced stromules under observation (89%) are associated with MTs; but we do see that about 11% of the stromules are not. Our results are not completely speculative, since if there is no association with MT, it must be independent and that statement is merely factual. Furthermore, we have not ignored or neglected the other potential mechanisms for stromule extension and we hypothesized how our results fit in with observations reported by others in the discussion. In brief, we discussed the possibility that actin anchoring, associations with the endoplasmic reticulum or stromules originating from an internal force could control MT-independent mechanism. All of these may play various roles in stromule formation, but in Figure 1 we have focused on the MT interactions and describe other interactions with ER in Figure 2 and actin later in the manuscript. We would like to disagree that our findings are only significant if only 100% of stromules are associated with MTs. Often complex biological processes are not absolute, and we believe our study will significantly advance our understanding of the role of the cytoskeleton during stromule dynamics, formation and stabilization.

6) The authors wished to 'conclusively demonstrate that stromules extend along MTs' and therefore used two inhibitors to disrupt the MTs. Their evidence for MT depolymerization was that stromules disappeared after 15 minutes. How can a specific assay be used and the observations then used as evidence that the assay is working? I find the reasoning faulty and wonder if this not a case of circular reasoning?

We agree that using stromule disappearance, as evidence of MT depolymerization would be circular reasoning. However, we did not make that circular argument. Instead, we show that both APM and Oryzalin result in depolymerization of MTs after 15 min of treatment in Figure 3A and Video 5 and after 1 hour of 1µM of Oryzalin treatment in Figure 7—figure supplement 1. It must be noted that the prior studies (Kwok and Hanson, 2003; Natesan et al., 2009), used these compounds without examining the effects on MTs. Here, we have monitored the depolymerization using MT markers and time lapse microscopy. We restated our results to overcome this misunderstanding.

The stromules disappeared 15 minutes after APM and Oryzalin treatment: is this 15 minutes after pressure infiltration? Is it 15 minutes after the 4mm2 leaf excision was taken for viewing?

We clarified that it was 15 minutes after pressure infiltration of the treatments. The following text was added to the Materials and methods section, “All time points started immediately following the infiltration of the treatment.”

What does disappearance mean? Did they retract? Were they not formed? It must be made clear what disappearance means.

How stromules disappear during the treatment is an interesting question. We described our observations, but left the interpretation up to the reader. It appears that from our reported observations that it is a combination of both mechanisms and we stated this in the revised text. We do see a correlation between the depolymerization of a MT and the retraction of a stromule associated with it (Figure 3A, bottom row; Video 5). Also, we see beaking that does not appear to continue into an elongated stromule (best observed in Video 5). To address this more directly, we counted the number of extension and retraction events after Oryzalin treatment and show that extension decreases and retraction increases (Figure 8E), further supporting that both extension and retraction contribute to a disappearance of stromules.

7) The concentrations of inhibitors used are, in general, extremely high and would completely disrupt cellular functions, especially after pressure infiltrations. The authors provide minimal information on this by stating "The concentrations that resulted in the microtubule depolymerization/stabilization without any lethal effect on the leaves were used further for experiments." It must be made clear in the Materials and methods section what was done to ascertain that the cells were actually surviving and functioning normally after the harsh treatments.

We provided more details in the Materials and methods subsection “Inhibitor treatment”. Without any lethal effect means – leaf is not dying or showing any sign of microscopic abnormality. When a cell dies, we see increased autofluorescence in the tonoplast and nucleolus, and chloroplast swelling that eventually leads to rupture. We did not see these effects, but to make it more clear, we repeated Oryzalin inhibitor experiments using a lower concentration over a longer period of time. We also examined the actin microfilament network, which remained intact. Although MTs were not completely depolymerized under these conditions (Figure 7—figure supplement 1), these data were provided to address these concerns.

8) The VIGS assay creates a weak point in this study. The time frame for the experiments involving VIGS is well beyond the induction period for stromules. The researchers actually observed plants four days after VIGS and found numerous anomalies including significantly altered / bundled cortical MT organization in leaves. They also found twice the number of stromules compared to VIGS vector control. The conclusion of high stromules number after 4 days in NbGCP4-silenced plants is at best just a correlation; it is not a direct evidence suggesting a clear cause and effect relationship. Perhaps the greatly altered MTs changed a lot of other things that could then then also have affected stromule frequency. the plant was clearly no longer operating in a normal manner. The same reasoning applies to the further increase in stromule length and number upon expressing p50 subsequently. In this case the observations were taken 24 hours later. Could something more have happened to create a physiological change in the 24 hour period? What happened at 4, 8, 12 hrs? That is definitely sufficient time for stromule formation to have occurred (Brunkard et al., 2015).

We strongly feel that these experiments are a strong point of the study that provides genetic support for the role of MTs in stromule formation and extension; however, we are happy to rewrite this section to make our approach and results clear.

We altered MTs by silencing GCP4. This altered state of MTs induced stromules without any inducer, such as p50 treatment. We made the observations 4 days after silencing because VIGS takes about 3-10 days to knockdown the expression of a target gene. Since GCP4 is an essential gene, significant knockdown of this gene has effect on the plant phenotype. Hence, we choose to do our analyses after 4 days of silencing (~50% reduction in GCP4 expression) because plants look similar to the VIGS control plants. At this time frame compared to the control, GCP4 silencing alone is sufficient to induce stromules because the MTs are bundled in GCP4 silenced plants. As we state in the manuscript, the stabilization of MTs is sufficient to induce stromules. This also correlates with our Taxol treatment data.

Our method of inducing a defense response is different than the Brunkard et al., 2015 paper, and their time points do not make sense in the context of the experiments that we have described here. We cannot make observations at 4, 8, 12 hours because we need to give enough time for p50 to express after infiltration. Therefore, we did our observation at earliest time point possible after infiltration of p50; i. e. 24 hours. Brunkard et al., 2015 did not use p50 as the inducer. Most of their studies are based on the isolated chloroplasts and not using intact leaves under a biological treatment.

9) There are some really obtuse statements:

“Stromules appeared to interact statically, and not dynamically, with AFs, suggesting there are actin anchor points along stromules.” What is meant by 'interact statically and not dynamically'? Please explain the interactions. Is the word 'interaction' itself not denoting a dynamic phenomenon?

We apologize if our use of static and dynamic resulted in obtuse statements. Below we have defined our word usage, but if the reviewer believes different wording is better, we would be open and appreciative of constructive suggestions. However, we believe the description below will clarify our word choice:

The definition of static (adjective): “characterized by a lack of movement, animation, or progression”

The definition of dynamic (adjective): “marked by usually continuous and productive activity or change.”

The definition of interaction (noun): “a kind of action that occurs as two or more objects have an effect upon one another.”

Interaction does not denote a dynamic phenomenon on its own. We used interaction as a noun and static and dynamic as adjectives that qualify or modify the meaning of interaction. Similarly, we used interact as a verb, and statically and dynamically as adverbs. In this context, we use static to describe the interaction with AFs, since the interaction was stable along individual points along AFs and did not move. We use dynamic to describe the interaction with MTs, since the interaction point changed over time, resulting in movement or change in position.

Further, the stromules became thin at anchor points – “We observed a clear thinning or constriction at the site of stromule-to-actin interaction points along the length and across the body of the chloroplast". Did the actin anchor create this thinning? As I understand the authors are talking about an anchor point and not a band of F-actin. If not, then how is the thinning of a stromule explained?

These are interesting questions. We do not know if the actin anchor caused the thinning. The stromule thinning and the actin anchoring are correlated; stating that it is causal would be an over interpretation. Nevertheless, reporting this observation opens interesting questions into the mechanism, and we hope it incites future research. Our high-resolution imaging did not reveal a band or ring of F-actin and we hypothesize that instead a tension point from the interaction with AF is generated, causing the thinning. We currently added in that the mechanism for the thinning is unknown at the end of the second paragraph of the subsection “Actin filaments serve as anchor points but not as tracks for stromule extension”.

What is meant by “across the body of the chloroplast”? Are we still talking about thinning?

Here we see a thinning or constriction from one side to the opposite side of the chloroplast body, that forms a groove along the body of the chloroplast that correlates with actin microfilaments. We thank the reviewer for pointing out that this is unclear, and we provided a 3D model of grooves along the body of the chloroplast in Figure 6—figure supplement 2.

10) In a similar manner the authors state "At 8 min, the stromule began to re-extend along the MT, between 8 min and 9 min, the body of the chloroplast was released and this resulted in stromule driven chloroplast movement." What are the authors trying to convey? The body of the chloroplast was released! From what (?) and how did this result in a conclusion of 'stromule driven chloroplast movement'? The authors did not actually establish 'stromule driven movement'. Did the stromule pull the body or push it? Did the stromule elongate further or did it retract in order to drive chloroplast movement?

We have made this clear in the text and we have also added more data on stromule-driven movement (Figure 8A, C-E; Figure 8—figure supplement 1).

In the described data we were following an individual chloroplast and associated stromule as an example. It is clearly defined – at 1 minute during the observation, stromule is extending; at 3 min it is retracting; but then at 8 min, it begins to re-extend. We hypothesize, that the chloroplast body was released from actin, but since we do not have direct evidence, it was not explicitly stated. We did not want to over interpret the data, however, we included this in the sixth paragraph of the Discussion. Also, we changed our wording to “chloroplast movement is correlated with stromule movement”.

Related, we now have analyzed the correlation between the angle of stromules and the angle of chloroplast movement. These comprise new, exciting data in this revision. We analyzed the correlation based on the stromule angle in the first frame and the subsequent angle of movement of chloroplasts. Using a circular correlation coefficient, r(FL), we were able to determine if it was more likely a pulling or pushing force. A positive r(FL) value would suggest a pulling force and a negative r(FL) value would be consistent with a pushing force. The value was 0.76, suggesting that the mechanism is more likely a pulling force. However, in the text we did not directly state that it is a pulling force, but rather, proposed that stromules could potentially provide the force for movement, or alternatively, they may just guide chloroplast movement. Since our data does not clearly support either mechanism, we called it stromule-directed movement, rather than stromule driven or stromule guided movement.

11) It is well observed that stromules arise in many different areas of a chloroplast randomly. Does this always coincide with chloroplast movement? The authors address this issue: "To examine how disruption of stromule and chloroplast AF anchoring effects chloroplast movement, we followed two chloroplasts in an epidermal cell treated with CytD." As I understand from the manuscript only one chloroplast was actually observed to reach the conclusion since the 1st one had restricted movement. Interestingly this singular observation is supposed to have added to the evidence that stromules may direct chloroplast movement in the absence of intact AFs.

In the previous version of the manuscript we presented three different examples of chloroplast movement correlated with the movement or orientation of a stromule in Figures 5, 6 and 7. In the new manuscript, we analyze and quantified numerous stromule-directed chloroplast movement events and added in additional examples (Figures 69; Figure 8—figure supplement 1, 2). Over half of chloroplast movement is stromule-directed and this is partially disrupted by partial disruption of MTs.

12) The weakest part of this manuscript is the proposed link to 'innate immunity'. This part of the study takes the manuscript into a highly speculative mode where the nice demonstrations of microtubule alignments and the presence of actin are undone by a lack of rigorous experiments. The authors point to a feature of innate immunity as 'eventual peri-nuclear clustering of chloroplasts'. It is notable that chloroplast clustering around nuclei happens routinely in leaves, even without bringing innate immunity into play. In the excised 4mm2 sections what is the baseline number of chloroplasts clumping together and how is it accounted for by the idea of innate immunity? Perhaps the wounding of tissue is responsible for innate immunity? This still does not explain the chloroplasts around nuclei in normal, unwounded (or in any untreated tissue). Clearly stromules have been observed by the authors under conditions that trigger the innate immunity but I fail to see an exclusive response where a cause and effect relationship is indicated with regard to stromules.

The proposed link to innate immunity builds upon our prior work (Caplan et al., 2015; Caplan et al., 2008) and we do not believe that it is highly speculative. We improved our description of how this connects to our prior work and progresses our understanding of how and why chloroplasts cluster around nuclei during plant innate immunity. We agree that chloroplasts are seen around nuclei in untreated mock tissue and present this data in the new Figure 10A. Mock tissue has nuclei with chloroplasts clustered, but this increases during plant innate immunity. Similarly, chloroplasts often have stromules, and they increase during plant innate immunity as we have shown previously (Caplan et al., 2015). We have presented a working model in Caplan et al., (2015) that stromules and perinuclear clustering acts to amplify innate immunity signaling. Neither can induce a defense response on its own, but constitutive stromule induction can accelerate HR-PCD. In the revised version, we provided full microscopic images that we used to count the perinculear clustering; they clearly show that many chloroplasts cluster near nuclei in the presence of p50 (Figure 10A). We used appropriate controls to show that this is specifically induced during p50 response. We are not using the wounded tissue here; all our observations are with tissue away from the primary infiltration point. Furthermore, we have clearly shown in our published paper (Caplan et al., 2015), that stromules are specifically induced during p50 response in NN plants but not in nn plants.

Some references are missing and should be added. Reference for the presence of chloroplasts in tobacco pavement cells is needed.

Only one example has been provided, making it difficult to fully respond to the concern. To our knowledge, we have referred to all the relevant references except for the one provided by the reviewer, which we believe may not be required. A recent paper published by a group of leading plant cell biologists (Barton et al., 2016) clearly states that it generally well accepted that Nicotiana pavement cells have chloroplasts. Interestingly, that same manuscript reports that there are multiple lines of evidence that Arabidopsis also contains chloroplasts in epidermal pavement cells, even though a recent stromule paper (Brunkard et al., 2015) refers to these plastids as leucoplasts.

The Introduction as well as discussion of earlier work on actin and microtubules must be done more thoroughly rather than be given a cursory mention. The authors must clearly state how their results advance the field and are not a repetition of earlier inhibitor experiments with a rather speculative link to chloroplast movement and plant immunity.

The authors have only looked at chloroplasts in pavement cells. For this study to have a real impact the observations must take into account other kinds of plastids, including plastids in mesophyll tissue and leucoplasts, all of which are known to display stromules.

We have described and referred to the previous papers in the Discussion. We explain how our data compares to prior inhibitor studies (there are only two primary papers and others are reviews) and how our work advances our understanding of stromule formation and dynamics. We have already shown in our recent paper (Caplan et al., 2015) that stromules are induced in mesophyll plastids. We have focused on epidermal cells for consistency and because stromules and their dynamics can be more accurately measured in epidermal cells compared to mesophyll cells, which have more tightly packed chloroplasts. Examining stromules also in tissues containing leucoplasts, such as roots, would require an entire replicate of our experimentation, which we believe is beyond the scope of this manuscript. Overall, we anticipate that our initial study will open exciting new avenues of research.

Reviewer #2:

[…] This work provides new and a significant information about the formation/regulation and function of the understudied plastid stromules. Although one can argue that the work it is generally based on correlative results, I consider that this data will be a keystone for future studies in this topic. I think that the manuscript fits with eLife's scope. However, I have some concerns about how the results are presented, described and its interpretation. Also, I consider that authors should support with more data the proposed model. Particularly, authors should address two things:

- The authors use the terms associated, bound, touching, interacting, etc. as "synonyms" but without specifying what they exactly mean. They should define these terms in a "precise" stromules-MTs distance(s) manner. Indeed, I do not think authors demonstrated interaction in any case. Authors should do 3D co-localization analyzes. In addition, they should consider doing FRET analysis if this is possible.

We rewrote and restated the text to remove the terms associated and touching. For consistency we used the more general term interacting throughout the manuscript. We only used bound to describe static interactions with actin. Stromule tips do co-localize with MTs and AFs in 3D and the majority of our analysis is generated from 3D confocal microscopy z-stacks over time. We apologize if that was not clear in text. However, confocal microscopy cannot provide a “precise” stromule-MT distance, since it lacks the required resolution. We have included much higher resolution transmission electron micrographs of MT in close proximity (nanometers) to stromules (Figure 1—figure supplement 1).

FRET analysis is currently not possible. FRET requires a direct protein-protein interaction, and we do not know what proteins on the stromule surface interact with MT directly or MT associated proteins (MAPs or motors) indirectly. We thank the reviewer for the suggestion, but without the identity of these molecular components, we cannot conduct this experiment at this time, but we will consider them in the future.

- Authors should analyze the dynamics changes of MTs and AFs networks after p50 defense induction. Are MTs stabilized after p50 expression? Are perinuclear AFs increased after p50? These results could be of great support for the proposed model. I believe the authors could do both analysis using the data that they already have.

These are excellent questions and we included these data on dynamic changes of MTs and AF networks during a p50-induced defense in the revised manuscript (Figure 5E-H). We examined perinuclear clustering, but it was difficult to quantify. Instead, we disrupted AFs using CTD and observed a decrease in perinuclear clustering of chloroplasts following p50 defense induction (Figure 10 C-D).

The authors should also address the following comments:

Abstract:

1) Misspelling: effects should be affect.

Changed in the Abstract.

2) Authors should revise and change the statement suggesting that AFs have no role in stromule extension. I think that providing "…anchor points for stromules to prevent their retraction while extending on MTs…" sounds like an important role.

Modified in the Abstract: "Although actin filaments (AFs) are not required for stromule extension, they provide anchor points for stromules to prevent their retraction while extending on MTs."

Results:

1) Section "Microtubules are required for stromule extension". Authors should consider that the drugs used inhibit the increase of stromules per plastid associated to mock treatment instead of reducing them.

Changed accordingly.

2) Section "Microtubule stabilization increases stromule number, length and stability", first paragraph: What are authors considering as MTs stabilization in Figure 2C? The image showing cells treated with paclitaxel should be label indicating this MTs stabilization. Are stromules longer in cells treated with paclitaxel?

Paclitaxel has been used extensively for MT stabilization and cancer research (>30,000 publications on Pubmed), and therefore, we did not characterize it here. After Paclitaxel treatment, we observed longer stromules and multiple stromules emanating from individual chloroplasts. We clarified that description on (subsection “Changes in microtubule organization increases stromule number, length and stability”, first paragraph, see Figure 3). However, we chose to focus our extensive characterization of MTs and stromules in NbGCP4-silenced plants that have altered MTs. In those experiments, we examined both the effects on MTs and stromules, including stromule length (Figure 45). Since, it is difficult to directly ascribe the changes in stromules to MT stabilization, the section header was changed to: “Changes in microtubule organization increases stromule number, length and stability.”

3) Section "Microtubule stabilization increases stromule number, length and stability". Figure 3F: Authors point out that the stromules number increase observed in NbGCP4-silenced and in p50-induced plants are "remarkably similar". They should analyze the MTs stability during p50 induction to support this statement.

We have provided new data in Figure 5E-H (see subsection “Changes in microtubule organization increases stromule number, length and stability”, last paragraph). In summary, p50-induced immunity results in similar changes as NbGCP4-silenced plants.

4) Section "Microtubule stabilization increases stromule number, length and stability". Figure 3I: Authors should describe and discuss the extension/retraction results obtained after treatment with p50.

See new data in Figure 5D (see subsection “Changes in microtubule organization increases stromule number, length and stability”, last paragraph).

5) Section "Stromules direct chloroplast movement in the absence of actin filaments", second paragraph: Although authors mention that exist "opposing forces" and a "rapid pulling" provided by stromules, they do not have any measure of these traction forces. Authors should be cautious with results interpretation and re-phrase this paragraph.

We modified the text accordingly. We also added this statement: “Our data show that stromules may direct chloroplast movement in epidermal pavement cells; however, it remains unknown if stromules provide a driving force or only guide chloroplast movement.”

6) Section "Actin microfilaments mediate perinuclear chloroplast clustering during plant immune response", second paragraph: authors should replace the sentence "…connections, but were unable to find…" by "…connections, but we were unable to find…"

We have changed this in the first paragraph of the subsection “Actin microfilaments mediate perinuclear chloroplast clustering during plant immune response”.

Discussion:

1) Third paragraph: "…and propose the difference…" should be replace by "…and propose that the difference…"

We have changed this in the third paragraph of the Discussion.

2) Third paragraph: What do the authors mean with the statement "directly interacted"? I do not think they showed any interaction.

We removed the word “directly”.

Figures:

1) Figure 1A: How authors define tip connection or extension along MTs? How close are stromules and MTs? Are authors using z-stacks projections to analyze this distance? Authors should clarify this point in every image/quantification where an interaction/connection/touching, etc. suggestion is made.

The following sentences were modified: “These observations were made in maximum intensity projections of z-stacks generated by confocal microscopy….”, and: “A stromule-to-MT interaction was designated if these two structures were overlapping or not resolvable by confocal microscopy. However, since the resolution of confocal microscopy is relatively low, we verified the close interaction between stromules and MTs using transmission electron microscopy (TEM).”

2) Figure 1B-E: Are these analyses using only z-stack projections or they consider 3D/4D dimensions? Figure 1D: Is the algorithm calculating the 3D distance between stromules and MTs?

The algorithm does not detect 3D distances. In the last paragraph of the subsection “Stromules interact and extend along microtubules”, we added that the algorithm was used on maximum intensity projections.

3) Figure 2B: Authors should show if exist statistical differences between times/treatments.

Figure 2B in old version is now in Figure 3B. We have tested multiple comparisons across the time and treatments. There is no other statistically significant difference.

4) Figure 2C: What do the authors consider as stabilization of MTs in this image? They should do an analysis similar to the Figure 3.

Please see point 2 (Results) above.

5) Figure 2C-D: Is there a positive correlation between the increase on stromules number and the stromules "associated" to MTs?

We did not measure if there was a positive correlation, mainly due to the caveat that it is difficult to quantify if they are “associated” by confocal microscopy because of the resolution limit of the technique.

6) Figure 5E: Authors should add the separated MTs and AFs channels to this figure.

Due to space limitations, we chose to provide a diagrammatical representation rather than showing all channels separately. However, if the reviewer still believes that separated channels would provide additional information, we can provide it as a supplemental figure.

7) Figure 3—figure supplement 2: Authors should add stats to this quantification. They should also consider using this data as a main figure.

We added in the total number stromules measured and the statistical significance if it existed. We agree with the reviewer that there is enough important data presented to be a main figure. However, since this was from the same experimental data set and just additional quantifications of the same data, we chose to keep it as a supplement to this figure.

Reviewer #3:

[…] The Introduction and Discussion of the paper are well written, touching on the key literature. The microscopy is beautiful and sophisticated analyses are performed on some of the image data sets using tools that other plant biologists will be interested in knowing about and perhaps applying. A strength of the study are the time series images showing evidence for guidance of stromule dynamics by cortical microtubules, this result was nicely controlled for by using multiple markers for microtubules, making the possibility of an aberrant interaction between stromules and an overexpressed microtubule marker unlikely. However, there are issues with data interpretation that need to be addressed and I felt that a number of major and secondary conclusions were not supported well by the data presented.

A) The authors test the role of microtubules in stromule maintenance by applying depolymerizing and stabilizing drugs and by manipulating microtubule organization by genetic knock down of a nucleation complex protein. Previous studies had indicated that microtubules play a relatively minor role in stromule abundance, but here microtubule depolymerization resulted in a significant reduction in stromules per plastid compared to the mock control. This was the strongest section of the paper but I have a few comments.

1) The following point is not a major criticism, but an odd thing with the drug experiments was that stromules significantly increased in abundance in the mock control over 15 minutes. The reduction in stromule numbers with drug treatment observed at the same time were only slightly lower than the numbers observed at time 0-5 minutes after treatment. If just stromule numbers at time zero and 15 minutes were compared, one would not infer a strong effect of microtubules. The reason for the increase in stromule number in the mock is neither explored nor explained. A couple of possibilities are that stromule formation may be stimulated by stress caused by placing the tissue in the microscope slide mount, or perhaps that light used for the imaging stimulates stromule formation. Since it is an effect on the increased stromule numbers that is observed with drug treatment, it would be nice to have bit more insight into why these number are elevated over the course of the experiment. The two possibilities mentioned could be tested by not imaging at time point 1, and by determining if media exchange during imaging can mitigate the induced increase in stromules over time in the slide mount.

We apologize for not explaining the induction with the mock treatment. The only difference between the media control and the mock treatment is the 0.2% DMSO from the inhibitor stocks. The mock treatment (often referred to as the vehicle control in other fields) with DMSO is the proper control, but since we saw a difference with our additional media control, we felt scientifically obligated to report these results. Stromules can be induced by a wide variety of stimuli. DMSO induced stromules, but nonetheless, these stromules were disrupted by the drug treatments.

2) It was a good idea to manipulate microtubules by a secondary means, other than drug treatment. However, here the authors conclude that altered microtubule dynamics were responsible for observed reduction in stromule extension and retraction rates. Since microtubule dynamics were not measured, nor were observations shown of possible interaction between stromule tips and microtubule ends, it is not clear why this conclusion was reached. What is evident is that microtubule organization is altered in the GCP4 knock down, with an increase in parallel organization of microtubules in leaf epidermal cells, a result that has been shown previously for knockdown of GCP4 and also for GCP-WD/NEDD1. While the results are striking, it is not clear why a change in polymer organization would drive a change in stromule dynamics. My suggestion would be to keep the interpretation open and not ascribe it to microtubule dynamics. It is possible that the effect is indirect. For example, by altering levels of free MAPs due to reduced total lattice in these cells.

We agree with the reviewer and modified the text to keep the interpretation open and did not ascribe it directly to microtubule dynamics. We appreciate these thoughtful comments and changed the text accordingly.

3) The authors conclude that microtubules play a direct role regulating stromules. While agree that the data shown are consistent with that idea, it is possible that the relationship is less direct, as the authors touch on in their Discussion when considering the ER and membrane contact sites.

We adjusted our text to state that our data is consistent with microtubules playing a role regulating stromules. Thank you for the suggestion. We have also observed stromules pushing through the ER network while extending along MTs. We added Figure 2 and two videos (Video 34) showing these observations and described these data in the subsection “Stromule extensions through the ER is directed by microtubules”.

4) The tip tracking method is very nice. It is kind of a shame though that it is not used to more rigorously quantify stromule tracking of microtubules. Only one example of such tracking is shown, with no summary measurements made over many cells, something that could be done with such a method. A further point here is that it should be defined more clearly how "on tube" vs. "not on tube" is determined, since the underlying molecular relationships are sub-resolution.

Our algorithm worked well on MTs marked with EB1; however, our peers recommended that we do not use EB1 and instead conduct our studies using the GFP-TUA6 line. However, our algorithm does not work with GFP-TUA6 because of free tubulin masking clear detection of MTs. Related, quantitative data using EB1 might be misleading, since the overexpression could change the dynamics. The quantitative data on stromules and interactions of stromules with MTs was conducted manually. Despite these caveats, we believe that these data provide a nice illustration of chloroplast movement along MTs and more clearly defined that the “on tube” were points in which the tip colocalized with the skeletonized MTs.

5) The SOACs analysis is used to assess features of polymer orientation, curvature and length. While this analysis produced a nice color-coded image of the orientation of segmented features, these data were not used to quantitatively assess patterns of polymer orientation, which again, seems a missed opportunity to show robustness of observations over many cells and samples. A second point, and an important one, is that by inspection of the raw data and the processed data, the method is not an appropriate means to assess polymer length, and those data should be dropped. It is not possible with such segmentation to tell if extended structures are composed a few long polymers or many shorter polymers, nor can it tease apart what happens were polymers meet and overlap at angles. Determining the length distribution in these arrays is an important goal in the field, but it is not trivial.

Although these quantitative comparative measurements of orientation were not made in the original work describing, SOAX, and only a visual output was provided. However, we developed a method based on the mean resultant length of MTs to quantify these images. The mean resultant length was calculated by converting the MT angles into individual vectors, adding the vectors together, and calculating the mean. MRL values are between 0 and 1, with 0 indicating that MT angles are random and 1 indicating all MT angles are the same and completely aligned.

We agree with the reviewer that the SOAX analysis cannot accurately assess polymer length due to unknowns that are not resolvable by confocal microscopy. This is not a characteristic that is easily measured, although we feel that our quantitative measurement, albeit indirect, demonstrates a method for measuring changes in microtubules. If the reviewer still insists, we will remove these data, however we hope the reviewer agrees there is some value to these data if more carefully described as potentially being a few long polymers or many short polymers. We added in the following statement: “The snake length is not a direct measurement of MT length, since this approach cannot accurately distinguish between two MTs that are bundled together. Nonetheless, this measurement further suggests that silencing NbGCP4 alters MTs.”

6) It is stated that GCP4 knock down results in greater bundling. This is not determined in the present manuscript.

The reviewer is correct that we did not determine that GCP4 knock down results in greater bundling. Those conclusions were drawn from Kong et al. (2010) in which they state that amiR GCP4 results in bundling in Arabidopsis. We apologize if our text suggested that we showed this directly here, and we more carefully cited the prior research that knockdown of GCP4 results in bundling. We also removed the term “bundling” from the description of our work in the Results and left that speculation only in the Discussion.

B) The authors conclude that actin filaments do not extend along actin filaments but that actin filaments do serve as static anchor points for stromules, and also prevent stromule retraction.

1) The data for stromule extension along actin cables are not quantified, but presented as example images and a video. These images show little evidence for stromule extension along actin bundles, but to be robust, a method should be used to census the relationship of stromules and actin structures so that data from multiple cells and tissue samples can be assessed. Even so, it should also be at least discussed that actin features may be present that did not label with the probes used.

We agree with the reviewer that our data do not show stromules extending along actin bundles, and the best example when examined closely, did not appear to be “on actin” but simply close by (Figure 6—figure supplement 1). We made that clearer in the text. In general, we almost never saw stromules extending along actin microfilaments and hence did not see a need to quantify that negative data. We also agree that mTalin and LifeAct may not show all the actin and added that to the second paragraph of the Discussion. We chose to use two different markers, since they each have their caveats, but the reviewer is correct that neither perfectly labels all actin.

2) In the CytD experiments, why didn't stromule number increase from 0 to 15 minutes in the mock as was observed for the microtubule drug experiments? If this effect does not occur in every experimental setup, the microtubule drug experiments should be repeated under conditions where stromule number does not go up in the mock.

As described above, this is most likely due to the DMSO concentration. We repeated these experiments with much lower concentrations of DMSO and drug treatments. We also added in additional quantifications of length, extension and retraction events and velocity, stromule-directed movement and movement type (constant and sudden) following Oryzalin or Cytochalasin D treatments (Figure 7, 8D-E).

3) The evidence of static anchor points at actin filaments is weakly presented. A single kymograph is shown for analysis of stalling during stromule retraction, and in this single analysis it is not clear that the stall actually occurs at the position of actin signal. The signal corresponding to the stromule tip is offset from the actin signal. A much more rigorous analysis of retraction and actin overlap sites is required.

We regret that the image we chose does not show the retraction clearly. Since this is a new assay we developed, it may not have been completely understood. However, it is clear from the depth of this critique, that this reviewer is indeed an expert on cytoskeleton dynamics; so if our point was missed by the reviewer there is a good chance it will be missed by others. Previously, we showed just one image of a stromule partially retracted (hence, the tip was offset from the signal). In the revision, we show three time points in which it was fully extended, kinked/partially retracted, and retracted back to an actin microfilament (Figure 6B). We also show retraction back to an actin microfilament in Figure 8B. We rectified that misunderstanding by showing montage of these retraction events. To further show these data, we added in three additional representative examples in Figure 6—figure supplement 1.

4) Likewise, the image sequence in Figure 5E and corresponding video are not convincing of retraction back to an actin-defined anchor point – the stromule tip appears to overshoot the "kink" position at the actin branch site. This is also a single image sequence example.

We provided more examples of the retraction events to actin anchor points (Figure 6—figure supplement 1). Our retraction assay was our way to avoid the over interpretation that a simple intersection between a stromule and actin is an anchor point. Unlike, the interaction with MTs, which has a motion trajectory that matches the path of a MT, the interaction with actin is static and is more difficult to quantify. However, we did try to quantify this by looking at retraction events indirectly.

5) The analysis in Figure 5C looks more quantitative, but it is a less direct measure of interaction. Rather, it is measuring a predicted consequence if the proposed interactions exist. It is also poorly explained and it took some time to determine what the graph represents. I think the second bar was meant to be labeled "not fully retracted", rather than "actin anchored', which is a conclusion of function rather than an observation.

We changed the bar labels to “full retraction and “partial retraction” and we thank the reviewer for this suggestion. The reviewer is correct that this is an indirect interaction, but we currently do not have the tools to quantify it in a more direct way. It is indeed a consequence of what would occur if an actin anchor exists and is somewhat predictive. We described this assay more carefully. As I’m sure the reviewer is aware, it is difficult to quantify these interactions without knowing the proteins involved in the interactions.

C) The authors conclude that loss of actin anchoring allows stromules to direct plastid movements.

1) Once again, these data consist of example images, not more global measurements across cells and tissue samples. In fact, just one image series example. It is a beautiful video, but it cannot be determined from this one example how generalized the phenomena shown are, nor how strong is the correlation of plastid movement with the direction stromule extension. A suggested means of analysis would be to assess each plastid displacement of x distance or greater in a set of time series acquired from multiple cells and leaves. For each displacement, stromule position and orientation would be determined and an orientational resultant would be estimated. The relationship between the predicted resultant angle of "pull" and the actual direction of movement could be plotted and analyzed.

The data was a representative time lapse from multiple observations, and we thank the reviewer for clear guidance on how to quantify the relationship of the movement and we have now quantified these data. Excitingly, we found that stromule-directed movement is a more general phenomenon and does not require actin disruption. We hinted at this in our first submission in Figure 5E that showed dynamics with MTs and AFs, and here we quantified the movement. As suggested, we calculated the orientation angles for the initial stromule angle that would “pull” and the actual direction of chloroplast movement. We used a circular correlation coefficient to show that the angles are indeed correlated and not randomly similar. We plotted the relationship directly as an XY scatterplot as well as looking at the difference between the two angles. We also show that Oryzalin treatment partially disrupts stromule directed movement. We were hesitant to call it “pulling” since it is unclear from our data if the stromule provides a pulling force or just guides the movement.

2) A partial loss of actin structure by the CytB treatment is somewhat unsatisfying. The authors chose not to use LatB, which is more effective, because MT organization is also affected. However, I think these data should still be shown to ask if the observed movements of plastids are still seen when actin is more completely disrupted. It will be evident if stromule interaction with MTs (MT-dependent interaction sites) is also affected.

We repeated our CytD treatment experiment and quantify stromule movement as suggested. Using a lower concentration of Cytochalasin D over a longer period of time fully disrupted AFs, but did not disrupt MTs (Figure 7—figure supplement 1). Complete disruption of AFs resulted in a complete loss of chloroplast movement, including stromule-directed movement (Figure 8D). However, stromules still persisted and extension was more constant and less sudden with slower extension and retraction rates (Figure 7D).

D) The authors conclude that actin anchoring mediate plastid accumulation at cell nuclei during the innate immune response and suggest a model by which plastid accumulation is facilitated by plastid movement driven microtubule- and stromule-based plastid motility, followed by actin mediated capture at the nucleus.

Another case of data by limited example. While the example shown in Figure 7D is compelling, more than this is needed. I was surprised that the hypothesis was not tested by actin disruption, for example. Further, it should be asked if plastid motility, and nuclear accumulation are significantly reduced if microtubules are depolymerized.

We thank the reviewer for this suggestion and we disrupted MTs with Oryzalin and AFs with Cytochalasin D. Oryzalin treatment only partially disrupts stromules, and this level of disruption had no effect on perinuclear clustering. Interestingly, Cytochalasin D significantly reduced perinuclear clustering (Figure 10D). Although this supports our hypothesis, the data should be cautiously interpreted since our data suggests that actin plays multiple roles in this process, including interactions with the body of the chloroplast controlling positioning, along the stromule and with nuclei.

E) In general, the authors should be careful about distinguishing conclusion and interpretation from observation. This issue occurs throughout the manuscript.

We thank the reviewer for this suggestion and we adjusted the text to more accurately distinguish between conclusions and interpretation. As we hope the reviewer can gauge from our responses, we do have a clear understanding of the difference, but that was not conveyed properly in the text. We believe the changes to the text and the new data have corrected this in the revision.

[Editors’ note: the author responses to the re-review follow.]

As you can see from the detailed reviews below, the reviewers very much appreciate your efforts to improve the initial manuscript. Yet they feel some additional adjustments are required before the paper can be published. These are mostly of organizational/semantic nature, and do not require additional experimentation. Apart from answering point by point to the reviews below, please pay particular attention to the following:

1) The reviewers are of the opinion that the analysis of retraction pausing with sites of actin filaments needs to be strengthened quantitatively. Currently, the data consist of example image sequences and an indirect measure involving pausing frequency, which is not explained very clearly. E.g., it is not obvious that this experiment correlates pausing with sites of possible actin action.

The previous analysis was a quantification of retracting stromules specifically pausing at AFs. We apologize for not describing this clearly. We have rewritten the description of the assay and results in the text and believe that the new description is much clearer. We also added in additional quantitative data to show that the majority of stromules that paused during retraction had a retracting tip colocalized with an AF, suggesting that these are possible sites of actin anchoring. Only a small fraction of retracting stromules (5.7% of the total) paused and did not have a stromule tip colocalizing with AFs. We have presented the data in Figure 6D and description in the second paragraph of the subsection “Actin filaments serve as anchor points but not as tracks for stromule extension”.

2) The reviewers believe that it would be a good idea to quantify tracking of microtubules by stromule tips rather than to show one example of such tracking.

In response to this, we manually tracked and quantified the velocity of stromules tips extending along EB1-Citrine- and GFP-TUA6-marked microtubules. We are still hesitant to show more EB1 data, since this is not an ideal marker for microtubules and do not want to mislead others into quantifying stromule dynamics using EB1. Instead, we show here that the rate of stromule extension is significantly lower with EB1, compared to GFP-TUA6 or stromules extending with no MT marker. These quantitative data are presented in Figure 1D and corresponding text in the last paragraph of the subsection “Stromules interact and extend along microtubules”.

3) Clarification is needed about how the circular correlation coefficient data relate to the results with the 30 degree cutoff angle. These data are new to the revised manuscript and need to be explained more thoroughly.

The main difference between the circular correlation coefficient data and the analysis using a 30-degree cutoff angle is that the circular correlation coefficient requires paired stromule angles and chloroplast movement angles. Therefore, only chloroplasts with stromules can be analyzed. If we restrict the circular correlation analysis to only paired movement events with a 30-degree angle difference, the r(FL) value is 0.95. A r(FL) value of 0.95 suggests angles are highly correlated and this is why a 30-degree difference was chosen. Using this 30-degree cutoff, we were able to examine all movement events, including chloroplasts that have no stromules. We have included this in the second paragraph of the subsection “Stromules direct chloroplast movement”.

Reviewer #2:

[…] Nevertheless, one caveat is that authors are not showing quantitative data on the stromules tip tracking. The authors argue that the algorithm only works for the overexpressing EB1 construct, which could be changing the general MTs dynamics. I think this could also be true for the other OE constructs. Thus, authors should consider adding tip tracking experiment quantifications, and show the reproducibility for this method, at least for EB1.

We have now quantified the velocity of stromules tips extending along MTs. We did not use the algorithm for the tracking, but rather manually tracked the stromules in imageJ, and removed the single instance of tracking with the algorithm. We are still hesitant to add in more EB1-Citrine data in this manuscript, since it significantly decreased the speed of stromule extension (new data in Figure 1D). To make the limitations of the algorithm clear and to emphasize all of the data in the manuscript was quantified by manual tracking, we added in the following statement: “The algorithm only accurately detected the slower moving motion when EB1-Citrine was used as a MT marker, and therefore, was not used in other experiments. The length, velocities, extension and retraction frequencies, and types of motion were quantified manually in all other experiments.”

Furthermore, we agree with the reviewer that any overexpression construct for marking MTs may change MT dynamics. Consequently, we quantified changes in stromule dynamics in the MT inhibitor and NbGCP4-silencing experiments without MT overexpression markers.

Reviewer #3:

[…] Point-by-point responses to authors' replies to initial review.

Point A1: In my first read, I missed the fact that DMSO treatment was associated with an increase in stromule number over time. This is perhaps because the media control was not shown in the figure for comparison. The revised text makes it more clear that DMSO causes an increase in stromule number over the media control, and that drug treatment reduces this increase. However, I am now wondering why the media controls are not shown in the figure. It would be useful to know if drug treatment also causes a reduction of stromules in the media control, even if there is no increase in time otherwise.

We added in the infiltration media alone control to the graph. Most reports just show the vehicle control (DMSO in this case), but we agree with the reviewer that showing the media control would be useful for others in the field; we added this into Figure 3B. It is important to report that DMSO can induce stromules, and when conducting any drug treatment assays, the concentration of DMSO must be the same for all treatments. It is not the focus of this work, but we do believe it is important to show this effect. However, the DMSO treatment is just media with the same amount of DMSO required to do the drug treatment. The requested drug treatment in media control cannot be included, since DMSO is required to solubilize the drugs.

If not, does this mean that induced stromules are somehow different from existing stromules in regard to microtubule depolymerization?

This is an interesting question, and more generally, are there different types of stromules? Are stromules induced during plant innate immunity the same as stromules induced during other stimuli, such as DMSO? We thank the reviewer for the question and added in the following statement in the Discussion: “In general, the formation of stromules may vary based on differences in cell, plastid, or stimulus type.”

Point A2: The revised text addresses the concern raised about ascribing the changes in stromule dynamics to changes specifically in microtubule dynamics. However, I still think it would be appropriate to make clear that, while the manipulation here is pretty specific, the observed changes in stromule dynamics may be either a direct or indirect consequence of changes in microtubule organization. For example, the statement – "Collectively, these results indicate that change in MT organization in NbGCP4-silenced plants or during p50 induced immunity control stromule dynamics." – is stronger than I think the data warrant and should be restated. "Collectively, these results indicate that changes in MT organization caused by NbGCP4-silencing plants or during p50 induced immunity are correlated with changes in stromule dynamics, indicating a possible role, direct or indirect, for MT organization in modulating stromule dynamics."

In the previous revision, we made numerous changes to the text to emphasize that these events are correlated. We thank the reviewer for finding this statement that we missed. We changed it to “Collectively, these results indicate that changes in MT organization caused by NbGCP4-silencing plants or during p50 induced immunity are correlated with changes in stromule dynamics, indicating a possible role, direct or indirect, for MT organization in modulating stromule dynamics."

Point A4: Thanks for the clarification. I agree that the figure is useful for portraying a sense of dynamics in a static format. The original text did not make it clear that the figure was produced manually. Unfortunately, this is also not clear in the revised text. The manual method is described in the Materials and methods, but the text and figure legend do not indicate how this figure panel was made. I think a simple edit to the figure legend would fix this.

We edited the text and figure to make it clear that only the depiction of the motion in Figure 1E was conducted with the algorithm and all the quantitative data was collected using manual tracking. Please see our response to reviewer #2 for a detailed description.

Point B1: I think it is important to give a sense of how much space was searched through to say "almost never". For example, you could state that "at least x stromules were examined in at least y cells, and only n stromules showed any evidence of extending along actin filaments. Even in these cases…"

We thank the reviewer for this suggestion, and we added: “Out of 73 stromule tip extension events from 34 cells, the vast majority (93%) of stromule tip extensions were not observed along AFs.”

Point B3: See comment for 5 below.

Point B4: See comment for 5 below.

Point B5: While more examples of stromules retracting to locations proximate to actin filaments, Figure 6D provides the only quantitation of this behavior. As indicated in the first review, it took me a while to figure out what I thought this panel represented. In reviewing this experiment again, I confess that I still find it unclear what exactly was measured here. The bar graph shows the percentage of fully retracted vs. partially retracted stromules. The figure legend states: "The percent of stromules containing actin anchors was quantified by monitoring retraction events in the samples described in panel A. Retracting stromules without actin anchors resulted in full retraction." However, in the text it states "70.3% + 0.02% (SEM) of retracting stromules partially retracted to one or more AF." So, was just the percent of partial retraction measured, or were retraction pause events also correlated with actin signal? If the latter, I might expect to see an analysis of how often retraction pausing was observed at sites of actin signal, but these data are not shown. If all partial retractions are indeed caused by actin anchors, then the percent of partial retraction alone would serve as an indirect estimate of the prevalence of stromules with actin anchors. However, is this known? Can stromules pause for other reasons? If they can, this experiment does not really probe prevalence of actin anchors well, nor does it test for the role of actin in pausing.

We apologize that our description of this experiment was not clear. We repeated the analysis of these data and rewrote the description. The repeat of the analysis was spurred by the reviewer’s question, “can stromules pause for other reasons?” We originally, did not quantify this because it was a rare event, but it is an excellent question. Therefore, we quantified how often stromules tips paused during retraction without the tip colocalizing with an AF and found that 5.7% of stromule retractions show this behavior. It is unclear if these pause events are at AFs that are not labeled (common problem with AF markers) or if it is pausing at some other interaction point. Nonetheless, these data further emphasize that the majority of partial retraction events show a paused stromule tip colocalized with AF signal. We no longer describe it as a “partial retraction”, since this ambiguous descriptor may be the point of confusion. Instead, we describe it more accurately as “retracting stromule tips paused for multiple, consecutive frames and showed colocalization with an AF.” See the second paragraph of the subsection “Actin filaments serve as anchor points but not as tracks for stromule extension” and Figure 6D.

Point C1: The new data measuring the correlation of stromule angle with movement angle are an important addition to the manuscript. The angular correlation plot in the supplementary data is actually stunning. I would put it in the main figure. However, I think it should be stated more clearly how plastids were selected for this analysis. Were just plastids with one stromule selected? If so, how were they chosen?

We chose to show the histogram data in the main figure because this was used to define the 30 degree difference for stromule directed movement classification, and therefore, leads into the next graph of the that figure. Therefore, we believe it should be left in the main figure. However, we also thought the angular correlation plot was quite striking and visually showed how correlated the stromule angle and the chloroplast movement angle were in these data. To clarify how plastids were chosen, we added in the following statement: “Only chloroplasts containing one or more stromules were used for this analysis because it depends on comparing paired measurements of the angle of the stromule from the chloroplast body attachment point to the tip and the angle of chloroplast movement.”

Also, a significant question is raised by the oryzalin experiments. In oryzalin treated cells, the circular correlation coefficient of stromule angle with the direction of plastid movement actually appears to go up (.85 vs.76 for no oryzalin). Yet when a 30 degree difference in stromule and movement angles is established as a cutoff for positive stromule directed movement, it is concluded that oryzalin treatment reduces stromule directed movement. It is hard for me to see how these two results are reconciled. If the correlation coefficient alone is considered, one would not conclude that oryzalin treatment reduces stromule directed movement in fact, it may even increase it.

We examined the difference in r(FL) values for DMSO and ORY are found they are not significant. To examine this, we calculated the standard errors of the r(FL) values using a jackknife method. A two-sample t-test shows that the r(FL) values for DMSO and ORY were not significantly different from each other (P=0.52), and consequently, this data is not in conflict with the data using the 30 degree cutoff. We chose the 30 degree cutoff because when using that cutoff, the angles were highly correlated with an r(FL) value of 0.95. The circular correlation analysis requires that we only analyze chloroplasts with stromules. However, we can use the 30 degree cutoff to examine chloroplast movement events with and without stromules. When this analysis is conducted, chloroplasts still move when microtubules are disrupted with ORY, but there is less stromule directed movement. See the second paragraph of the subsection “Stromules direct chloroplast movement”.

Point D1: It seems that the new drug tests during perinuclear clustering were useful. I agree that it may be challenging to interpret the results with cytochalasin D as it now turns out that actin is required for plastid movement and not just anchoring in these cells. Interestingly, this duality of actin being required for both movement and anchoring was also observed by Wada and colleagues for light driven plastid movement (it would be good to point this out in the Discussion).

Thank you for this suggestion and we added in the following sentence: “Interestingly, the N-terminal coiled-coiled domain of CHUP1 is also required to anchor chloroplasts to the plasma membrane, revealing a complex, dual role of actin during chloroplast movement and anchoring (Oikawa et al., 2008)”.

I am, however, a bit confused by continuing to propose a model where plastids are directed or guided during the early stages of the clustering response by MT's, since no effect on clustering by MT disruption was observed. In other words, the new data do not support all the particulars of the model.

We speculate in the Discussion that since MTs are required for stromule extension and involved in chloroplast movement, they are likely also involved in stromule extension to nuclei. However, that is speculative and we changed the wording and added in that further studies need to be conducted: “The role of MTs during the formation of stromule-to-nuclei connections requires further studies. However, our data suggests that once those connections are formed, stromules may guide or pull chloroplasts towards the nucleus, which then results in perinuclear clustering of chloroplasts.”

There seemed to be general agreement in the first review that this last experimental section was not a strength of the manuscript. I believe it may be a good idea to leave these studies out for now and have them be part of a future manuscript addressing specifically the mechanism of the clustering response.

We agree that the other experimental sections of the manuscript are stronger, mainly because there are a lot of future experiments that could be done to elucidate the mechanisms involved in perinuclear clustering of chloroplasts. The need for future experiments can be viewed as a weakness, but also as a strength, since it will potentially spur discussions in our field and additional studies. Therefore, we thank the reviewer for suggesting making this a separate, future manuscript, but after much discussion, we believe presenting the data we have now will be more beneficial to other in this field of research.

Other comments and questions on the revised text:

Data questions:

1) Section “Microtubules are required for stromule formation and extension”: In relating the results that support a role for MT's in stromule extension, it is stated that stromules in the drug treated cells completely retract after 15 min. of treatment. If this is the case, can the authors comment on the data shown in Figure 3B, which indicate only a modest drop in the number of observed stromules in the drug treated cells from time 0' to time 15'? Also, it is not clear that these differences are significant. The standard errors for APM look like they mutually overlap the means at 2 sigma (0' to 15'), It is not clear what is going on with the stats for the oryzalin data.

We apologize for the confusion, but the statement in the subsection “Microtubules are required for stromule formation and extension”, was a description of a single data set shown in Figure 3A and Video 5; the quantification was shown in Figure 3A. It actually highlights how stromules will stay extended even on a small fragment of MT. We made it clear that the descriptions of stromule disruption are single examples shown in Figure 3A and Video 5 in the aforementioned subsection.

We are only showing significance between the DMSO vehicle control and APM or the DMSO vehicle control and oryzalin. Since there was an increase in stromules in the DMSO control, the significance was only examined at time 15’. There was no significance between time 0’ to time 15’, and consequently, we did not report significance. We now stated this explicitly in the Figure 1D legend as: “Compared to the DMSO vehicle control, stromules significantly decrease after treatment with APM (20 µM) and Oryzalin (300 µM) at 15 min; no other comparisons were significant.”

We would also like to mention that new extensive data on stromule dynamics after oryzalin treatment were add in Figure 7, 8, and 10 and complement these data in the last revision.

2) Section “Actin filaments serve as anchor points but not as tracks for stromule extension”, last paragraph: Can the authors comment on the observation that 70% of stromules were observed to only partially retract, a state that is attributed to actin interaction, whereas the number of observed stromules in cells treated with 200 µM CTD was the same as in mock treatments? If actin interaction is the primary means of preventing complete retraction of stromules, it seems that the number of observed stromules would be expected to be lower in cells where actin is disrupted.

The reviewer is correct that we expected to see a decrease with CTD, but did not observe one at 30 min after CTD treatment. In time-lapse data of this treatment, we saw that stromules initially retracted, but then re-extended on MTs (Figure 8—figure supplement 2, Video 11), and we speculate it’s this re-extension that results in no significant change after 30 min. It is also possible that actin anchoring just alters the dynamics of stromules and stromule driven movement as shown in Figures 7, 9 and 10, and has little role in maintaining the number of stromules. We hope that we or others will discover the proteins required for these interactions, so that their function can be studied more extensively in the future.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance. These are outlined below under reviewer #3. Additionally, since the work was in review, another paper appeared in the Plant Journal with similar results to yours. You should refer to the other study in your revised manuscript and point out any similarities/differences in the work (see doi: 10.1111/tpj.13813).

We also noticed the online version of Plant Journal paper while our third revised version of the manuscript was under review. As suggested, we have added similarities and differences in our Discussion section. The main similarity is that stromules extend along MT. Our study shows this more extensively and quantitatively, by using three different markers and reporting different stromule morphologies, the percent interaction with MTs, stromule length, the rate of extension and retraction under multiple conditions, including during an innate immune response and with various cytoskeleton inhibitors. Both studies with Oryzalin support the conclusion that the role of MTs during stromule formation was under appreciated because stromules can hold on to small fragments of MTs and complete disruption of MTs with Oryzalin is difficult.

There are multiple differences between our paper and the Plant Journal paper. One of the largest differences is their model proposes slow moving anchoring points on MT, while we propose anchoring points along AFs. Their study did not mark and monitor AF and stromules simultaneously, and consequently, they would not have observed AF anchoring. Our study marked MTs, AFs, and stromules simultaneously to observe their complex interactions. They also suggest that fast moving stromules move along AFs, but they did not mark AF to show the extension. We also saw fast moving stromules that did not associate with MTs. However, we did mark AF, but interestingly, we did not observe them along AFs. We propose that they extend by an unknown force, potentially an internal chloroplast force. Our paper also shows that perinuclear association of stromules with nuclei is mediated by AFs. Lastly, our work proposes a novel role for stromules during chloroplast movement. Overall, we are excited about this similar publication and anticipate that the similarities and differences of these papers published nearly back-to-back will spark a lot of discussion, enhancing the impact of both papers.

Reviewer #3:

The authors have addressed my previous concerns save one. At the end of the Results, the authors conclude:

"In conclusion, we propose a model in which perinuclear clustering of chloroplasts involves stromule extension along the MTs, stabilization of these extensions by anchor points to AFs surrounding nuclei and stromules guide chloroplasts towards nuclei during an immune response."

However, the authors reported that no connections with microtubules by stromules were observed during P50 induction of perinuclear clustering, and treatment with oryzalin had no effect on clustering. Thus, the experimental results do not support a role for guidance of stromules by microtubules in perinuclear clustering, indeed, they appear to contradict this possibility. The statement about microtubule guidance needs to be omitted from the proposed model for perinuclear clustering.

We have modified the statement: "In conclusion, we propose a model in which perinuclear clustering of chloroplasts involves stromule anchoring to AFs surrounding nuclei and stromules guide chloroplasts towards nuclei during an immune response.”

The description of the experiments addressing pausing of stromule retraction at AFs is now much more clear. However, the statistics in Figure 6D compare the rates of stromule pausing with the rate of full retraction. This comparison is relevant for asking if pausing is significantly more frequent than full retraction, but the main question here is whether pausing occurs at AFs more frequently than might be explained by chance alone. The rate of pausing at AFs by chance alone is a function of AF density; the higher the AF density, the more often pauses would be observed in association with them, even if there is no functional connection. Thus, AF density needs to be taken into account in a statistical test for the observed rate of AF-associated pausing (this can be a bit tricky due to the dynamics of the AFs). However, it seems pretty clear from the figures and the movies that the density of labeled AF's is lower than could easily explain a 77% rate of pausing at labeled AFs by chance alone (this would mean that, on average, about 3/4 of the stromule length would overlap AF signal). Even though the association is challenging to test formally, a statement to this effect might make more clear why the observed rate of 77% is a reasonable suggestion of an association.

The assay that we conducted to look at stromule pausing during retraction was new here, and we thank the reviewer for further assisting us in describing the method and results. We agree that pausing at AFs occurs more frequently than might be explained by chance alone and we have added this to the description. We added in the following statement to that effect: “The pausing of retracting stromules at AF cannot be explained by chance alone because the density of AFs and the colocalization of stromules with AFs observed appeared to be much less than 77.1% (Figure 6A; Figure 6—figure supplement 1; Video 7).”

https://doi.org/10.7554/eLife.23625.055

Article and author information

Author details

  1. Amutha Sampath Kumar

    Delaware Biotechnology Institute, University of Delaware, Newark, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Visualization, Methodology, Writing—original draft
    Contributed equally with
    Eunsook Park
    Competing interests
    No competing interests declared
  2. Eunsook Park

    1. Department of Plant Biology, College of Biological Sciences, University of California, Davis, Davis, United States
    2. The Genome Center, College of Biological Sciences, University of California, Davis, Davis, United States
    Present address
    Department of Plant Sciences, College of Agriculture and Life Sciences, Seoul National University, Seoul, Korea
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Amutha Sampath Kumar
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2984-3039
  3. Alexander Nedo

    1. Delaware Biotechnology Institute, University of Delaware, Newark, United States
    2. Department of Biological Sciences, College of Arts and Sciences, University of Delaware, Newark, United States
    Contribution
    Conceptualization, Data curation, Validation, Visualization, Methodology
    Competing interests
    No competing interests declared
  4. Ali Alqarni

    1. Delaware Biotechnology Institute, University of Delaware, Newark, United States
    2. Department of Biological Sciences, College of Arts and Sciences, University of Delaware, Newark, United States
    3. Department of Plant and Soil Sciences, College of Agriculture and Natural Resources, University of Delaware, Newark, United States
    Contribution
    Formal analysis, Validation, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Li Ren

    Department of Plant and Soil Sciences, College of Agriculture and Natural Resources, University of Delaware, Newark, United States
    Contribution
    Software, Methodology
    Competing interests
    No competing interests declared
  6. Kyle Hoban

    1. Delaware Biotechnology Institute, University of Delaware, Newark, United States
    2. Department of Biological Sciences, College of Arts and Sciences, University of Delaware, Newark, United States
    Contribution
    Formal analysis, Investigation, Visualization, Writing—review and editing
    Competing interests
    No competing interests declared
  7. Shannon Modla

    Delaware Biotechnology Institute, University of Delaware, Newark, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  8. John H McDonald

    Department of Biological Sciences, College of Arts and Sciences, University of Delaware, Newark, United States
    Contribution
    Formal analysis, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  9. Chandra Kambhamettu

    1. Department of Plant and Soil Sciences, College of Agriculture and Natural Resources, University of Delaware, Newark, United States
    2. Department of Computer and Information Sciences, College of Engineering, University of Delaware, Newark, United States
    Contribution
    Software, Formal analysis, Supervision, Methodology
    Competing interests
    No competing interests declared
  10. Savithramma P Dinesh-Kumar

    1. Department of Plant Biology, College of Biological Sciences, University of California, Davis, Davis, United States
    2. The Genome Center, College of Biological Sciences, University of California, Davis, Davis, United States
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    spdineshkumar@ucdavis.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5738-316X
  11. Jeffrey Lewis Caplan

    1. Delaware Biotechnology Institute, University of Delaware, Newark, United States
    2. Department of Biological Sciences, College of Arts and Sciences, University of Delaware, Newark, United States
    3. Department of Plant and Soil Sciences, College of Agriculture and Natural Resources, University of Delaware, Newark, United States
    Contribution
    Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    jcaplan@udel.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3991-0912

Funding

National Institutes of Health (R01 GM097587)

  • Savithramma P Dinesh-Kumar
  • Jeffrey Lewis Caplan

National Institutes of Health (P20 GM103446)

  • Jeffrey Lewis Caplan

National Institutes of Health (S10 OD016361)

  • Jeffrey Lewis Caplan

National Institutes of Health (S10 RR027273)

  • Jeffrey Lewis Caplan

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Drs. Jung-Youn Lee and Bo Liu for providing MAP-CKL6 and EB1 plasmids respectively. We thank Drs. Manfred Heinlein and Karl Oparka for providing GFP-TUA6 N. benthamiana transgenic seeds. The National Institute of Health R01 grant GM097587 to SPD-K and JLC, supported this work. Microscopy access was supported by grants from the NIH (P20 GM103446, S10 OD016361 and S10 RR027273).

Reviewing Editor

  1. Jean T Greenberg, University of Chicago, United States

Version history

  1. Received: November 24, 2016
  2. Accepted: January 16, 2018
  3. Accepted Manuscript published: January 17, 2018 (version 1)
  4. Version of Record published: February 16, 2018 (version 2)

Copyright

© 2018, Kumar et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 5,918
    Page views
  • 875
    Downloads
  • 55
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Amutha Sampath Kumar
  2. Eunsook Park
  3. Alexander Nedo
  4. Ali Alqarni
  5. Li Ren
  6. Kyle Hoban
  7. Shannon Modla
  8. John H McDonald
  9. Chandra Kambhamettu
  10. Savithramma P Dinesh-Kumar
  11. Jeffrey Lewis Caplan
(2018)
Stromule extension along microtubules coordinated with actin-mediated anchoring guides perinuclear chloroplast movement during innate immunity
eLife 7:e23625.
https://doi.org/10.7554/eLife.23625

Further reading

    1. Cell Biology
    2. Developmental Biology
    Simon Schneider, Andjela Kovacevic ... Hubert Schorle
    Research Article

    Cylicins are testis-specific proteins, which are exclusively expressed during spermiogenesis. In mice and humans, two Cylicins, the gonosomal X-linked Cylicin 1 (Cylc1/CYLC1) and the autosomal Cylicin 2 (Cylc2/CYLC2) genes, have been identified. Cylicins are cytoskeletal proteins with an overall positive charge due to lysine-rich repeats. While Cylicins have been localized in the acrosomal region of round spermatids, they resemble a major component of the calyx within the perinuclear theca at the posterior part of mature sperm nuclei. However, the role of Cylicins during spermiogenesis has not yet been investigated. Here, we applied CRISPR/Cas9-mediated gene editing in zygotes to establish Cylc1- and Cylc2-deficient mouse lines as a model to study the function of these proteins. Cylc1 deficiency resulted in male subfertility, whereas Cylc2-/-, Cylc1-/yCylc2+/-, and Cylc1-/yCylc2-/- males were infertile. Phenotypical characterization revealed that loss of Cylicins prevents proper calyx assembly during spermiogenesis. This results in decreased epididymal sperm counts, impaired shedding of excess cytoplasm, and severe structural malformations, ultimately resulting in impaired sperm motility. Furthermore, exome sequencing identified an infertile man with a hemizygous variant in CYLC1 and a heterozygous variant in CYLC2, displaying morphological abnormalities of the sperm including the absence of the acrosome. Thus, our study highlights the relevance and importance of Cylicins for spermiogenic remodeling and male fertility in human and mouse, and provides the basis for further studies on unraveling the complex molecular interactions between perinuclear theca proteins required during spermiogenesis.

    1. Cell Biology
    2. Structural Biology and Molecular Biophysics
    Bronwyn A Lucas, Benjamin A Himes, Nikolaus Grigorieff
    Research Advance

    Previously we showed that 2D template matching (2DTM) can be used to localize macromolecular complexes in images recorded by cryogenic electron microscopy (cryo-EM) with high precision, even in the presence of noise and cellular background (Lucas et al., 2021; Lucas et al., 2022). Here, we show that once localized, these particles may be averaged together to generate high-resolution 3D reconstructions. However, regions included in the template may suffer from template bias, leading to inflated resolution estimates and making the interpretation of high-resolution features unreliable. We evaluate conditions that minimize template bias while retaining the benefits of high-precision localization, and we show that molecular features not present in the template can be reconstructed at high resolution from targets found by 2DTM, extending prior work at low-resolution. Moreover, we present a quantitative metric for template bias to aid the interpretation of 3D reconstructions calculated with particles localized using high-resolution templates and fine angular sampling.