1. Developmental Biology and Stem Cells
  2. Plant Biology
Download icon

Cell type boundaries organize plant development

  1. Monica Pia Caggiano
  2. Xiulian Yu
  3. Neha Bhatia
  4. André Larsson
  5. Hasthi Ram
  6. Carolyn K Ohno
  7. Pia Sappl
  8. Elliot M Meyerowitz
  9. Henrik Jönsson
  10. Marcus G Heisler Is a corresponding author
  1. European Molecular Biology Laboratory, Germany
  2. University of Sydney, Australia
  3. Lund University, Sweden
  4. California Institute of Technology, Howard Hughes Medical Institute, United States
  5. University of Cambridge, United Kingdom
Research Article
Cited
1
Views
2,299
Comments
0
Cite as: eLife 2017;6:e27421 doi: 10.7554/eLife.27421

Abstract

In plants the dorsoventral boundary of leaves defines an axis of symmetry through the centre of the organ separating the top (dorsal) and bottom (ventral) tissues. Although the positioning of this boundary is critical for leaf morphogenesis, how the boundary is established and how it influences development remains unclear. Using live-imaging and perturbation experiments we show that leaf orientation, morphology and position are pre-patterned by HD-ZIPIII and KAN gene expression in the shoot, leading to a model in which dorsoventral genes coordinate to regulate plant development by localizing auxin response between their expression domains. However we also find that auxin levels feedback on dorsoventral patterning by spatially organizing HD-ZIPIII and KAN expression in the shoot periphery. By demonstrating that the regulation of these genes by auxin also governs their response to wounds, our results also provide a parsimonious explanation for the influence of wounds on leaf dorsoventrality.

https://doi.org/10.7554/eLife.27421.001

Introduction

Lateral organ development in plants and animals typically involves several processes occurring in a coordinated manner. These include organ positioning, the specification of different cell types and organ morphogenesis. Spatial cues specifying these processes are usually provided by a molecular pre-pattern present in precursor tissues, or from inductive signals emanating from neighboring regions. Unlike animals however, plant organs such as leaves arise continuously in regular patterns around the shoot apical meristem (SAM). Nevertheless, certain features of leaves are relatively constant including the restriction of their formation to the meristem periphery and their flattened, dorsoventral (top-bottom) orientation with respect to the shoot apex. How are these fundamental features specified?

Since the 1950s wounding experiments involving the isolation of leaf primordia from the meristem have suggested the presence of an inductive signal from the meristem that promotes dorsal identity within leaf primordia at the time of organ initiation (Reinhardt et al., 2005; Sussex, 1951). A variant on this theme is the proposal that transient auxin depletion in the adaxial (adjacent to the shoot axis) tissues of leaf primordia promotes dorsal identity (Qi et al., 2014). In contrast, other studies suggest that dorsoventral patterning is pre-established, being directly derived from central-peripheral patterning of the shoot (Hagemann and Gleissberg, 1996; Husbands et al., 2009; Kerstetter et al., 2001; Koch and Meinhardt, 1994). Supporting this proposal is the observation that transcription factors involved in both dorsal and ventral cell fate in the leaf are expressed in the SAM in a central and peripheral manner respectively (Emery et al., 2003; Kerstetter et al., 2001; McConnell et al., 2001; Yadav et al., 2013). Thus, the manner in which leaf dorsoventrality is first established in leaves remains unresolved (Kuhlemeier and Timmermans, 2016).

To explain how leaf dorsoventrality regulates morphogenesis, Waites and Hudson, 1995 took a cue from wing development in Drosophila (Diaz-Benjumea and Cohen, 1993) and proposed that tissues located along the boundary between dorsal and ventral leaf tissues, i.e. along the leaf margin, act as organizers by producing mobile signals that pattern lamina growth (Waites and Hudson, 1995). So far, several genes have been identified that are expressed along this boundary including KLUH (Anastasiou et al., 2007) and the WOX family transcription WOX1, WOX2 and PRS (Haecker et al., 2004; Matsumoto and Okada, 2001; Nakata et al., 2012). The WOX genes in particular are required for proper lamina growth but are expressed only at the boundary region suggesting they may promote long-range patterning (Nakata et al., 2012). However although their ectopic expression can cause filamentous outgrowths at the leaf base, overall lamina growth is fairly normal (Nakata et al., 2012) suggesting that additional factors must direct patterning. The WOX genes are also known to regulate the expression of both dorsal and ventrally expressed transcription factors and miRNAs (Nakata et al., 2012; Zhang et al., 2014; Zhang et al., 2017) indicating that the WOX genes promote lamina growth at least in part by maintaining the integrity of dorsal and ventral expression domains. Overall, this leaves the question of how dorsoventral boundaries actually regulate morphogenesis still unanswered.

Besides influencing leaf differentiation and shape, genes involved in leaf dorsoventrality influence leaf position. For instance, Arabidopsis plants mutant for the KANADI (KAN) genes develop leaves ectopically from the hypocotyl and leaf tissues (Izhaki and Bowman, 2007) while plants mutant for the Class III HD-ZIP (HD-ZIPIII) genes develop leaves from the center of the shoot (Emery et al., 2003). These observations indicate that the developmental mechanisms that specify leaf dorsoventrality may also be involved in organ positioning although how these processes relate is unclear.

In this study we investigate the origin of dorsoventral patterning in detail. We show that new organs are centered on a pre-patterned boundary region located between the expression domains of HD-ZIPIII and KAN genes and that both KAN and HD-ZIPIII genes act to repress organogenesis where they are expressed. This leads to a model in which dorsoventral genes control leaf morphogenesis and positioning by localizing auxin transcriptional response, which then polarizes cells non-cell autonomously (Bhatia et al., 2016). However, we also show that dynamic auxin levels play a central role in determining boundary position by spatially organizing HD-ZIPIII and KAN1 expression in the meristem periphery. By demonstrating that the regulation of these genes by auxin also governs their response to wounds, our results also provide a parsimonious explanation for the influence of wounds on leaf dorsoventrality.

Results

Genes involved in leaf dorsoventrality pre-pattern organ formation in the shoot

To better understand how leaf dorsoventrality is first established we used confocal microscopy to monitor the expression of genes involved in leaf dorsoventrality in the shoot apical meristem (SAM) in combination with the auxin efflux carrier PIN-FORMED1 (PIN1) using functional fluorescent protein reporters. In general, we found proteins involved in leaf dorsoventrality to be localized in non-overlapping concentric patterns with the dorsal Class III HD-ZIP protein REVOLUTA (REV) (Otsuga et al., 2001) being detected centrally, as marked by the expression of REVp::REV-2 × YPet (REV-2 × YPet), and the ventral protein KANADI1 (KAN1) (Kerstetter et al., 2001) expressed peripherally, as marked by the expression of KAN1p::KAN1−2 × GFP (KAN1−2 × GFP) (Figure 1, A to E). These concentric domains not only encircled the meristem but also extended contiguously around initiating leaves and floral bracts. Imaging PIN1p::PIN1-CFP (PIN1-CFP) (Gordon et al., 2007a) together with these reporters revealed regions of high PIN1-CFP expression, marking positions of organ inception (Heisler et al., 2005; Reinhardt et al., 2003), to be centered on a narrow region or ‘gap’ located between the dorsal and ventral domains (Figure 1D–K). As found previously, REV expression extended out into the PIN1-marked primordia over time (Heisler et al., 2005), reducing the ‘gap’ between the REV and KAN domains (Figure 1—figure supplement 1A–F,M–P). Nevertheless, at later stages a more substantial ‘gap’ reappeared between the REV and KAN1 domains (Figure 1C and D insets; Figure 1—figure supplement 1G–L and Q–R) indicating that the distance between the REV and KAN1 domains can change dynamically. Time-lapse imaging revealed that few cells alter their KAN1 expression during these early stages (Figure 2A–F) suggesting that the boundary of KAN1 expression in the SAM marks cells destined to form the future dorsoventral boundary of the organ.

Figure 1 with 1 supplement see all
Organ initiation is centered on a boundary between the expression domains of genes involved in leaf dorsoventrality.

(AD) Confocal projections showing REV-2 ×YPet (red), PIN1-CFP (blue) and KAN1−2 × GFP (green) expression in a vegetative shoot apical meristem at 3 days (A), 4 days (C) and 5 days (D) after stratification (DAS), respectively. Inset in (C) shows a longitudinal optical section across the dotted line in the right leaf. Inset in (D) shows a transverse optical section across the dotted line in the left leaf. White arrowheads in the insets in (C and D) mark a gap in between the REV-2 ×YPet and KAN1−2 × GFP expression domains where PIN1 expression is highest. (B) Longitudinal reconstructed section of seedling shown in (A). (E) Expression pattern of REV-2 ×YPet, KAN1−2 × GFP and PIN1-CFP in an inflorescence meristem. White arrow head marks region where KAN1−2 × GFP expression is being reestablished after floral bract emergence. (FK) Longitudinal optical sections across the dashed lines in (D) and (E) showing localized PIN1-CFP expression (magenta) marking organ inception at the REV-2 ×YPet/KAN1−2 × GFP boundary in both the vegetative meristem (F–H) and inflorescence meristem (I–K). White arrow heads mark cells in between the REV-2 ×YPet and KAN1−2 × GFP expression domains where PIN1-CFP expression is highest. Scale bars, 20 μm (A-E, inset in (D)) and 10 μm (F–K).

https://doi.org/10.7554/eLife.27421.002
The expression of KAN1−2 × GFP is relatively stable with respect to the underlying cells within initiating organs.

(A–B) Confocal projections showing an inflorescence meristem viewed from above expressing PIN1-CFP (blue) and KAN1−2 × GFP (green) at two time points (0 hr and 12 hr). Two incipient primordia are marked i1 and i2. (C–D) Close up views corresponding to primordium i1 from (A) and (B) with white arrows marking three cells at the edge of KAN1−2 × GFP expression that retain this expression over the time interval. Yellow arrow heads mark two cells in which KAN1−2 × GFP is absent at 12 hr. Similar colored dots mark the same cells in (C) and (D), tracked over 12 hr. (E) Close up view of primordium I2 in (A) with arrow marking adaxial edge of KAN1 expression. (F) Close up of primordium I2 in (B) showing the same cell marked in (E) remaining at the adaxial edge of KAN1 expression after 12 hr. Similar colored dots mark the same cells in (E) and (F), tracked over 12 hr. Scale bars, 20 μm in (A and B); 10 μm in (C–F).

https://doi.org/10.7554/eLife.27421.004

Ventrally expressed MIR165/166 (Kidner and Martienssen, 2004; Merelo et al., 2016; Nogueira et al., 2007; Yao et al., 2009) also appeared active in the SAM periphery as marked by a MIR166Ap::GFPER reporter and MIR165/166 biosensor (Figure 3A–I), consistent with previous studies (Miyashima et al., 2013). Both KAN1−2 × GFP and MIR166Ap::GFPER re-established their expression in the SAM periphery after organ outgrowth (Figure 1E arrowhead; Figure 1—figure supplement 1I,K and L and Figure 3C arrowhead). As members of the WOX family of transcription factors including WOX1 and PRS are expressed at dorsoventral boundaries (Nakata et al., 2012) we imaged functional translational fusions to both PRS (PRSp::PRS- 2 × GFP) and WOX1 (WOX1p:: 2 × GFP-WOX1) to examine their expression in the shoot relative to leaves. We found that although 2 × GFP-WOX1 expression was limited to the margins and middle domain of leaves (Figure 4A and B) and absent in the inflorescence meristem (Figure 4C), in the vegetative shoot PRS-2 ×GFP expression extended from the leaf middle domain to surround the SAM in a region between the KAN1 and REV expression domains (Figure 4D–H). In the inflorescence meristem however, PRS-2 × GFP expression was restricted to early cryptic bract primordia (Figure 4I and J). In contrast to all genes described so far, a FILp::dsRED-N7 marker was expressed in both the abaxial (away from the shoot axis) and adaxial regions of the primordium at inception, consistent with a previous study (Tameshige et al., 2013) (Figure 5). Overall these data reveal that in many respects although not all, dorsoventral patterning within young leaf primordia, including the middle region (Nakata et al., 2012), corresponds to and is contiguous with, the patterning of dorsoventral gene expression in the SAM.

Expression and activity of MIR165/166 is localized to the periphery of the shoot meristem.

(A) Expression of MIR166Ap::GFPER (green), PIN1-CFP (blue) and REV-2 × YPet (red) in the vegetative meristem (VM) at 3.5 DAS. (B) Longitudinal section of meristem shown in (A). (C) Expression of MIR166Ap::GFPER (green), PIN1-CFP (blue) and REV-2 × YPet (red) in the inflorescence meristem (IM). White arrow head marks the reestablishment of MIR166Ap::GFPER expression around the meristem after organ emergence. (D to F) Expression of REV-2 × YPet (red) alone (D), a MIR165/166 biosensor driven by the UBQ10 promoter (green) alone (E) and both combined in the same IM (F). (GI) Corresponding control for (D to F) where the MIR165/166 biosensor has been rendered insensitive to MIRNA activity. Bars represent 20 μm.

https://doi.org/10.7554/eLife.27421.005
Expression patterns of 2 × GFP-WOX1 and PRS-2 ×GFP.

Confocal projections showing PIN1-CFP (blue) and 2 × GFP-WOX1 (green) expression patterns in the vegetative meristem and leaves of seedlings at 3 DAS. (B) Longitudinal section of meristem shown in (A). (C) An inflorescence meristem image showing 2 × GFP-WOX1 is not expressed in the IM. (D) Confocal projection showing PIN1-CFP (blue), PRS-2 ×GFP (green) and REV-2 ×YPet (red) expression in the vegetative meristem and leaves at 3.5 DAS, where PRS-2 ×GFP is expressed surrounding the VM and along the leaf margins. (E and F) Cross sections of leaf on the right side in (D) showing the expression of PRS-2 × GFP in the middle domain of the leaf. (G and H) Confocal projection and cross section showing PRS-2 × GFP (green) and KAN1−2 × CFP (red) expression patterns in the vegetative meristem and leaves of seedlings at 3 DAS. (I and J) PRS-2 × GFP (green) is expressed in the young floral bracts in the IM, indicated with arrowhead in (J). Bar = 20 μm.

https://doi.org/10.7554/eLife.27421.006
FILp::dsREDN7 expression is broad during leaf initiation but is later excluded from dorsal tissues.

(A–F) Confocal projections and reconstructed sections of seedlings expressing FILp::dsREDN7 (green), REV-2 ×VENUS (red) and PIN1-CFP (blue). (A) Top view of seedling at 3DAS (BC) Longitudinal section of seedling shown in (A). Dashed line shows dorsoventral axis of first leaf and arrowheads mark dorsal cells expressing both REV-2 ×VENUS and FILp::dsREDN7. (D) Seedlings at 3.5 DAS with FILp::dsREDN7 expression more restricted to the developing ventral side of the leaf. (EF) Longitudinal sections of seedling shown in (D) showing a more complementary pattern of FILp::dsREDN7 relative to REV-2 ×VENUS compared to the earlier stage shown in (A) to (C). Scale bars represent 20 μm.

https://doi.org/10.7554/eLife.27421.007

The KAN and class III HD-ZIP genes repress organ initiation where they are expressed

The finding that sites of organ inception marked by PIN1 are located between KAN1 and REV expression domains in the SAM suggest that both KAN1 and REV may act to repress organ inception where they are expressed. Supporting this proposal, leaves develop ectopically in kan1 kan2 kan4 and rev phb phv mutants (Izhaki and Bowman, 2007). However it is possible that the HD-ZIPIII genes influence organ initiation only indirectly by promoting SAM formation during embryogenesis (Emery et al., 2003; Izhaki and Bowman, 2007). To distinguish between these possibilities we induced the expression of MIR165/166-resistant REVr-2 ×VENUS throughout the epidermis using a pOp6/LhGR two-component (Samalova et al., 2005) system with the ATML1 promoter (Sessions et al., 1999) driving LhGR and found that it caused an arrest of organ formation and repression of KAN1−2 × GFP in both the vegetative and inflorescence meristems (Figure 6A–E). Despite the lack of organs, stem growth for the inflorescence meristem continued without any obvious change to meristem size, indicating that the phenotype initially influenced organ initiation (inset in Figure 6E). Similar results were observed after induction of a short tandem target mimicry construct designed to repress MIR165/166 activity (Yan et al., 2012) (Figure 6F and G) or after epidermal induction of MIR165/166-resistant PHAVOLUTA (Figure 6H–J). Similarly, plants expressing KAN1 ectopically in the epidermis also stopped making new organs (Figure 6K and L) and the inflorescence meristem took on a dome shape before eventually arresting (Figure 6M). We conclude both the KAN and Class III HD-ZIP genes regulate organ positioning at least in part by repressing organ initiation where they are expressed.

Organ initiation depends on the restriction of Class III HD-ZIP and KANADI expression in the shoot.

(A) Confocal projection showing wild type control seedling at 7DAS viewed from above for comparison to (C) and (F). (B) Macroscopic view of control seedling at 7DAS for comparison to (D), (G) and (L). (C) Arrest of organogenesis after ectopic expression of REVr-2 × VENUS from the ATML1 promoter in the vegetative meristem (7 DAS) after germination on DEX, KAN1−2 × GFP (green) expression is down regulated and could only be detected in a few cells in the sub-epidermis. Although PIN1-CFP (blue) expression is patchy, no leaves developed. (D) Macroscopic view of plant in (C). (E) Arrest of organogenesis after ectopic expression of REVr-2 ×VENUS (red) from the ATML1 in the IM after 3 DEX treatments over 6 days. Note the absence of KAN1−2 × GFP signal. Inset shows longitudinal optical section of the meristem across the dashed line. (F) Seedlings at 7DAS showing similar phenotype to (C) after induction of a short tandem target mimic (STTM) designed to down regulate MIR165/166 activity. (G) Macroscopic view of plant in (F). (HJ) Ectopic expression of REV-2YPet (red) and arrest of organogenesis (PIN1-CFP in blue) in 4DAS seedling after induction of MIR165/166 resistant PHAVOLUTA. (H) Longitudinal view of un-induced control. Top view (I), and macroscopic view (J) of induced seedling showing arrest of organ development. (K and L) Confocal projection (K) and macroscopic view (L) of seedling at 7DAS after induction of KAN1-GFP (green) in the epidermis. No leaves have developed (autofluorescence shown in red). (M) Arrest of organogenesis after induction of KAN1-GFP (green) driven by the ATML1 promoter in the IM after 3 DEX treatments over 6 days; autofluorescence (red). Inset shows longitudinal optical section of the meristem across the dashed line. Scale bars represent 20 μm in (A, C, E, F, HI, K and M); 1 mm in (B, D, G, J and L).

https://doi.org/10.7554/eLife.27421.008

Expression patterns of REV and KAN1 in the shoot regulate leaf positioning and morphogenesis

To test whether boundaries between KAN1 and Class III HD-ZIP expression in the SAM can play an instructive role in positioning new organs and determining their subsequent dorsoventrality, we induced KAN1−2 × GFP expression ectopically at the center of the SAM using a pOp6/LhGR two-component (Samalova et al., 2005) system and the CLV3 promoter driving LhGR. After KAN1−2 × GFP induction, most seedlings initiated several new leaves before their growth stopped. Confocal imaging of seedlings five days after stratification (DAS) on dexamethasone (DEX) induction medium revealed that new organs, marked by high levels of PIN1-CFP expression, formed ectopically at the perimeter of an enlarged and irregular central domain of induced KAN1−2 × GFP expression, in which REV-2 × YPet expression had been repressed (Figure 7A and B). Although endogenous KAN1 was not monitored, ectopic KAN1−2 × GFP was only detected within or bordering organs during their initiation while REV-2 × YPet expression was often restricted during later developmental stages (Figure 7C–H). This indicates that patterns of KAN1 gene expression within the SAM can influence subsequent organ development. In particular, we noted that the distal margins of developing leaf primordia always correlated with boundaries of REV expression within the epidermis, even when REV expression was abnormally restricted (arrow heads in Figure 7C–H; Video 1). Several classes of phenotype, including leaves with an inverted orientation, could be distinguished at maturity (Figure 7I–T) that correlated with the patterns of REV-2 ×YPet expression observed during early development. These distinct morphological classes can be explained according to the configuration of the HD-ZIPIII-KAN boundaries at organ inception. Specifically, we infer that the number and orientation of boundaries within organ founder cells (as specified by high auxin levels) determines the configuration of later leaf marginal tissue (Figure 7P–S).

KANADI1 expression boundaries in the shoot specify organ position and orientation.

(A and B) Confocal projections showing organ initiation marked by REV-2 × YPet (red) and PIN1-CFP (blue) at border of KAN1−2 × GFP expression (green) in wild type (A) and after induction of KAN1−2 × GFP using the CLV3 promoter (B). Distance separating opposite organs was greater for induced (B) compared to control (A) (114.3 ± 3.3 μm, n = 19 vs 54.2 ± 1.0 μm n = 10 (mean ±SE, p<0.05, t-test)). (CH) Confocal projections (C, E and G) and longitudinal reconstructions corresponding to dashed lines (D, F and H respectively) showing restricted REV-2 ×YPet expression (red) after ectopic KAN1−2 × GFP induction (green). Regions in which neither REV-2 ×YPet nor KAN1−2 × GFP signal was detected may potentially express endogenous KAN1, which was not monitored. Four main configurations of REV expression and morphology were observed (labeled 1 to 4). Class 1 organs (E and F) correspond to the wild type, Class 2 (C and D) express REV-2 ×YPet centrally, Class 3 (E and F) express REV-2 ×YPet in a reversed orientation and Class 4 (G and H) express REV-2 ×YPet centrally and laterally only. Correspondence between REV-2 ×YPet expression boundaries and leaf margins indicated by arrowheads in D, F and H; m indicates meristem. Gamma value changed to highlight PIN1-CFP expression (blue) in (C) to (H). (IO) Examples of mature leaves corresponding to Classes 1 to 4, including the WT (I), cup-shaped (J), lotus-shaped (a variation of cup-shaped) (K), needle-like (a further decrease in extent of dorsal tissue compare to cup-shaped) (L), inverted (M and N) and four bladed (O). ‘D and V’ represent ‘dorsal’ and ‘ventral’ respectively in (O). (P to S) Diagrams summarizing proposed configurations of REV and KAN (green) gene expression in leaf founder cells (dashed circles) (upper diagram) leading to the observed phenotypic classes of leaf shape (numbered 1 to 4) (lower diagram) after induction of KAN1−2 × GFP using the CLV3 promoter. (P) represents the wild type Class 1 configuration, (Q) represents Class 2, (R) represents Class 3 and (S) represents Class 4. (T) Frequency of seedlings exhibiting different leaf morphologies after ectopic induction of KAN1−2 × GFP expression in the CLV3 domain Class of phenotype corresponds to those indicated in (I to O). Scale bars = 20 μm in A to H; 1 mm in I, J and M.

https://doi.org/10.7554/eLife.27421.009
Video 1
Movie shows confocal 3D projection of a single vegetative seedling corresponding to that shown in Figure 7G and H.

Several leaf-like organs have formed on the boundary of ectopic KAN1 expression driven by the CLV3 promoter (green). These organs express REV (red) in restricted patterns corresponding to the three classes described in Figure 7 (labeled Class 2, 3 and 4). Note developing leaf margins, marked by high PIN1 expression (blue) correlate with REV expression boundaries in the epidermis.

https://doi.org/10.7554/eLife.27421.010

Maximal auxin response is localized to HD-ZIPIII and KAN boundaries

The influence of dorsoventral gene expression on organogenesis suggests that dorsoventral boundaries may function generally to localize auxin response. In support of this proposal it has been reported that in the inflorescence meristem the auxin transcriptional marker DR5 is only responsive to auxin in the shoot periphery (de Reuille et al., 2006). We investigated DR5 expression (Liao et al., 2015) in the vegetative meristem by examining its expression both in the wild type and after 1-N-Naphthylphthalamic acid (NPA) treatment at 3 DAS. In mock-treated seedlings we found DR5 to be expressed at the locations of incipient primordia and at the distal tip of existing leaf primordia (Figure 8A and B). In contrast, NPA-treated seedlings expressed DR5 in a ring encircling the SAM prior to organ emergence and in the middle domain or dorsoventral boundary region of developing leaves (Figure 8C and D). Similar experiments with seedlings expressing the ratiometric R2D2 intracellular auxin reporter (Liao et al., 2015) revealed a generally broader distribution of auxin compared to the DR5 transcriptional response, especially after NPA treatment, where signal was not restricted to the meristem periphery or leaf tip (Figure 8E–H). These results indicate that like the inflorescence meristem, auxin transcriptional response outside the periphery of the vegetative meristem appears repressed. Since we also observed PRS expression localized to the meristem periphery (Figure 4D), we tested whether PRS as well as WOX1 are auxin inducible and found that both genes respond to auxin treatment within 12 and 15 hr respectively although their response was restricted to the boundary region (Figure 8I–P). Measuring transcript levels using qPCR indicated that the auxin response for both genes occurs at the transcriptional level (Figure 8Q).

Auxin promotes PRS and WOX expression.

(A–D) Response of pDR5V2−3 × VENUS-N7 (green) auxin transcriptional reporter to NPA in Arabidopsis seedlings treated at 3DAS (Days after Stratification). (A and B) Confocal projections 45 hr (A) and 72 hr (B) after treatment with mock solution. (C and D) Confocal projections 45 hr (C) and 72 hr (D) after treatment with 100 μM NPA solution (n = 5/5). White arrowhead in (C) marks absence of DR5 reporter in the center of the meristem. Inset in (D) shows transverse optical section through the ring-shaped organ showing most DR5 expression localized in the center of the organ. (E–H) Expression and response of R2D2 (magenta) to auxin along with PIN1-GFP expression (green) in Arabidopsis seedlings treated at 3DAS. (E and F) Confocal projections 48 hr (E) and 72 hr (F) after treatment with mock solution. (G and H) Confocal projections 48 hr (G) and 72 hr (H) after treatment with 100 μM NPA solution (n = 4/4). White arrowhead in (G) marks the presence of auxin in the meristem center (compare with (C) which shows absence of auxin signaling in the meristem center). (I–L) Expression and response of pPRS:: PRS-2 ×GFP to auxin in Arabidopsis seedlings. Confocal projections and transverse optical slices of seedlings 4DAS showing of pPRS:: PRS-2 ×GFP expression (green) 12 hr after treatment with mock solution (I and J) and 5 mM NAA (K and L) (n = 5/5). (M–P) Expression and response of 2 × GFP WOX to auxin in Arabidopsis seedlings. (M and N) Confocal projections (M and O) and corresponding optical slices (N and P) of seedlings 4DAS showing pWOX1::2 × GFP-WOX1 expression (green) 12 hr after treatment with mock solution (M and N) and 5 mM NAA (O and P). Note WOX expression increases but does not expand beyond its regular expression domain upon auxin addition (n = 5/5). (Q) Q-PCR analysis of PRS, WOX1 and positive control IAA20 transcripts after 5 mM NAA or mock treatment on 4 days old wild-type (Ler) seedlings. *=p < 0.05, **=p < 0.001. Scale bars 20 μm (A–I, K); 15 μm (J and L); 30 μm (M–P).

https://doi.org/10.7554/eLife.27421.011

These results not only reveal that general auxin response is maximized at the dorsoventral boundary but also, that genes already known to be expressed at the boundary are auxin responsive. Considering our results altogether and the finding that KAN1 represses PRS and WOX1 in ventral tissues (Nakata et al., 2012), we suggest a general scenario in which KAN and HD-ZIPIII genes repress auxin response where they are expressed, leaving a narrow domain of auxin responsive cells in between their expression domains.

We tested this proposal in silico by implementing a previous model for phyllotaxis (Jönsson et al., 2006; Smith et al., 2006), now supported experimentally (Bayer et al., 2009; Bhatia et al., 2016; Heisler et al., 2010), that incorporates the polarization of PIN1 towards cells with high intracellular auxin concentrations. By assuming that both KAN1 and REV repress auxin-induced transcription and by including a narrow region of cells located between the KAN1 and REV expression domains the model was able to self-organize the periodic formation of auxin maxima along the boundary as predicted, compared to a broader distribution of maxima when such boundaries are not included (Figure 9A–D; compare 9B to 9D).

Computational model illustrating how dorsoventral gene expression boundaries may restrict phyllotactic patterning to the SAM peripheral zone.

(A) Illustration of model interactions. Auxin is transported passively and actively via PIN1 between cells. PIN1 is polarized towards cells with high auxin, via a signaling pathway represented by X (previously suggested to be realized by increased stresses in the neighboring cells due to changes in mechanical wall properties (Heisler et al., 2010). (B) As shown previously (Heisler et al., 2010; Jönsson et al., 2006; Smith et al., 2006), peaks of auxin are formed spontaneously. (C) A pattern of KANADI (green) and REVOLUTA (red) is added to the template with a boundary domain in between in which REV expression is low or absent and KAN1 expression is absent. (D) If KANADI and REVOLUTA decrease the signal X in cells where they are expressed (dashed interactions in A), the formation of auxin peaks is restricted to the boundary.

https://doi.org/10.7554/eLife.27421.013

Auxin organises HD-ZIPIII and KAN expression in the shoot periphery

So far, our results indicate that both HD-ZIPIII and KAN1 suppress auxin-induced gene expression where they are expressed. However this does not exclude the possibility that auxin may play a role in patterning HD-ZIPIII and KAN expression. Indeed, auxin promotes the expression of several HD-ZIPIII genes in rice (Itoh et al., 2008) and in Arabidopsis, REV expression typically extends towards PIN1 polarity convergence patterns in the meristem periphery and disappears in axil regions where auxin is depleted (Heisler et al., 2005). To further investigate any potential regulation of REV and KAN expression by auxin we treated inflorescence meristems with synthetic auxin NAA(1-Naphthylacetic acid)as well as auxin transport inhibitor NPA (N-1-Naphthylthalamic acid) to prevent rapid auxin redistribution and examined the response of both REV and KAN1. We found that the region in between the REV and KAN1 domains narrowed due to a significant expansion of REV expression within 15 hr of treatment (Figure 10A and B). REV also became expressed in cells between the primordia and meristem where REV had presumably reduced its expression previously (Figure 10C and D). This was confirmed by following cells over time where we found that although KAN expression appeared static, more cells expressed REV (Figure 10C–H). A more subtle response was observed as early as 6 hr upon combined NAA and NPA application (Figure 10—figure supplement 1). Local application of IAA to pin1-4 meristems also resulted in a local response by REV within 16 hr of auxin application (Figure 10—figure supplement 2). In contrast, exogenous auxin had little effect on a pCLV3:GFP-ER marker for central zone identity (Figure 10—figure supplement 3). Apart from changes in REV we noticed that over the course of 48 hr, KAN1 expression did not appear as expected in between the meristem and floral bracts of stages i1 and i2 (but not P1) at the time of treatment (Figure 10I and J), compared to mock treated plants (Figure 10K and H), where KAN expression appeared adjacent to the primordia within 24 hr. In vegetative meristems this was more obvious where contiguous REV expression often appeared between the meristem and developing leaves where KAN1 expression would normally be expressed (Figure 11A–H; Video 2). These results indicate exogenous auxin promotes HD-ZIPIII expression locally and that exogenous auxin can prevent the establishment of KAN1 expression if applied at stages i1 and i2 of organ development.

Figure 10 with 3 supplements see all
Effect of auxin on the expression patterns of REV and KAN1 in the inflorescence meristem.

(A and B) Confocal projections of the IMs showing expression pattern of REV-2 × YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) before (A) and 15 hr after the combined application of 5 mM NAA and 100 μM NPA (B). Primordium (P) and incipient primordium (i) stages are numbered from i3-P2 based on convention described in (Heisler et al., 2005). Note up regulation and expansion of REV expression 15 hr after the combined application of NAA and NPA (n = 6). (C and D) Longitudinal optical sections along the dashed white lines in (A) and (B) respectively. Note the presence of REV expression in the epidermal cells marked by arrowheads in (D) and a corresponding absence or weak level of expression in (C). Similar colored arrowheads mark the same cells tracked over 15 hr. (E and F) Magnified views of the surface of i1, outlined by dotted rectangles in (A and B) showing expression pattern of REV-2 ×YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) before (E) and 15 hr after the combined application of 5 mM NAA and 100 μM NPA (F). Note the presence of REV in the cells marked by arrowheads in (F) and their absence in (E). Similar colored arrowheads mark the same cells tracked over 15 hr in (E and F). (G and H) Same as (E and F) but showing REV and KAN1 expression only. Note the presence of a gap between REV and KAN1 expression in (G) but its absence in (H), marked by white arrowheads in (G and H). (I–L) Confocal projections of IMs showing expression of REV-2 ×YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) before (I) and 48 hr after the combined application of 5 mM NAA and 100 μM NPA (J), and before (K) and 24 hr after treatment with mock solution (L). Initially for both control and treated meristems, KAN1 expression is absent between the meristem center and P1 (blue arrowheads), i1 (white arrowheads) and i2 (yellow arrowheads) (I and K). Under mock treatment, KAN1 expression appears in all three corresponding regions (marked by the same arrowheads) 24 hr later (L). However for the meristem treated with NAA and NPA, KAN1 expression is absent in regions previously corresponding to i1 (white arrowheads) and i2 (yellow arrowheads) but not P1 (blue arrowheads), even after 48 hr (J). Scale bars 20 μm (A–D, I–L) and 10 μm (E–H).

https://doi.org/10.7554/eLife.27421.014
High auxin levels prevent new KAN1 expression in the leaf axils.

(A–E) Confocal projection of the VM showing expression pattern of REV-2 × YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) 23 hr after the treatment with 5 mM NAA (A), 21 hr after combined treatment with NAA and NPA (B and C); panel (C) showing close-up view of the meristem in (B) with REV (red) and KAN (green) expression only, 70 hr after treatment with NPA alone (D) and 70 hr after treatment with mock (E). White arrowheads mark the presence of KAN1 expression in the cells adjacent to the grown out leaves in (A) and (E) and absence in (B–D). Note that REV expression (red) expanded towards leaves axils in (B–D). However also note that REV expression appears faint in the leaf axils in (B and C) compared to (D). This is due to the reason that the combined application of NAA and NPA resulted in the leaves growing at an acute angle to the meristem, which caused shading and made it difficult to reach leaf axils while imaging. (F) A bar graph showing percentages of leaves lacking KAN1 expression in their vicinity upon treatment with NAA, NPA and NAA + NPA. Scale bars 20 μm (A and B), 10 μm in (C) and 30 μm (D and E).

https://doi.org/10.7554/eLife.27421.018
Video 2
Confocal projection of an Arabidopsis seedling 70 hr after treatment with NPA showing view of the vegetative meristem from above with channels for PIN1-CFP (blue), REV-2x YPet (red) and KAN1-2xGFP (green) alternating (also shown as a snapshot in Figure 11C).

Note the absence of contiguous KAN1 expression in between the meristem and older leaves potentially due to auxin build up in the absence of its regular transport.

https://doi.org/10.7554/eLife.27421.019

To test whether REV or KAN1 expression depend on endogenous auxin for their patterns of expression we treated triple marker plants with the TIR1 antagonist auxinole (Hayashi et al., 2012) and found that after 18 hr of treatment KAN expression had expanded slightly toward the meristem center (Figure 12A and B; Figure 12—figure supplement 1A–D). As auxinole competes with endogenous auxin for the TIR1 binding pocket, we attempted to reduce endogenous auxin signaling further by simultaneously treating the meristems with yucasin (Nishimura et al., 2014) and kyn (He et al., 2011) to block auxin synthesis. This combination of treatments led to an arrest of organogenesis and ‘pin’ phenotype as well as a much more significant expansion and restriction of KAN1 and REV expression centrally, respectively (Figure 12C and D; Figure 12—figure supplement 1E–H Figure 12—figure supplement 2). KAN expression not only expanded into those cells originally located between the REV and KAN1 domains but also cells that had previously been expressing REV at high levels. However KAN1 expression remained excluded from the central most cells of the shoot and established floral bracts and leaf primordia, where REV expression still remained (Figure 12D; Figure 12—figure supplement 2. Extending the drug treatment time did not lead to a further change in expression. Monitoring the ratiometric auxin sensor R2D2 in response to a similar combined inhibitor treatment confirmed a strong decrease in overall intracellular auxin levels, including in the central meristem region and maturing organs (Figure 12E and F), indicating that auxin independent mechanisms regulate the expression patterns of these genes in the central zone and in organs after their initiation. In contrast to REV and KAN1, auxin depletion did not influence size of the central zone, as marked by the expression of pCLV3:GFP-ER (Figure 12—figure supplement 3).

Figure 12 with 3 supplements see all
Auxin depletion alters boundary position.

(A and B) Confocal projections of the IMs showing expression pattern of REV-2 × YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) before (A) and 18 hr (B) after the application of 100 μM auxinole. Inset shows close-up of the primordium outlined with the dotted rectangle. Similar colored dots mark the same cells at 0 hr and 18 hr time-points. Note the presence of KAN1 expression in the proximity of the cells marked with colored dots in the inset in (A) and its absence in the inset in (B). Similar color arrowheads in (A and B) mark the same cells that showed REV expression at 0 hr but KAN1 expression at 18 hr after treatment with auxinole. (C and D) Confocal projections of the IMs showing expression pattern of REV-2 × YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) before (C) and 18 hr (D) after the combined application of 100 μM auxinole, 100 μM KYN and 100 μM Yucasin (auxin depleting drugs). Note KAN-2 ×GFP expression has expanded centrally at the expense of REV-2 × YPet expression (compare the cells marked by arrowheads in (C) with (D), similar colored arrowheads mark the same cells tracked over 18 hr) (n = 6/6). (E and F) Confocal projections of the IMs indicating the predicted auxin distribution (magenta) based on R2D2 expression along with PIN1-GFP expression (green) before (E) and 17 hr after the combined application of 100 μM auxinole, 100 μM kyn and 100 μM yucasin (auxin depleting drugs) (F). Note lack of detectable auxin based on R2D2 expression in (F) compared to (E) after the combined drug application (n = 3/4). Scale bars 20 μm (A–D), 30 μm (E and F).

https://doi.org/10.7554/eLife.27421.020

Overall these data indicate that dynamic changes in auxin levels specifically within the meristem periphery play a central role in regulating dorsal and ventral identities. While high levels of auxin act to both trigger and maintain HD-ZIPIII expression, this only occurs in cells not already expressing KAN1. Such high levels of auxin also prevent the acropetal expansion of KAN1 expression, not only in cells within primordia but also in regions between the primordia and meristem. As our auxin treatments only prevented KAN1 expression from establishing around primordia at stages, i1 and i2 and not around P1, which corresponds to a stage when auxin depletion has already occurred (Heisler et al., 2005), our data also suggest that once cells are exposed to low auxin levels, their fate becomes determined, i.e subsequent auxin exposure cannot reverse their trajectory towards ventral or peripheral cell identity. Outside the meristem periphery, both in the meristem central zone and within more established organs, REV and KAN expression appears comparatively auxin insensitive.

Wounding induces KAN and represses REV expression in an auxin depletion-dependent manner

Since auxin is required to maintain REV at the expense of KAN expression in the meristem periphery and wounding causes repolarization of PIN1 away from wound sites (Heisler et al., 2010) resulting in auxin depletion (Landrein et al., 2015), we investigated whether wounding results in ectopic KAN1 and reductions in REV expression. Firstly, using a pulsed IR laser to ablate cells adjacent to young organ primordia we confirmed that auxin levels decrease in the vicinity of wounds by monitoring the expression of the R2D2 ratiometric marker (Figure 13A and B). Next, we monitored REV, KAN and PIN1 expression in response to such wounds. We found that KAN1 became expressed in cells adjacent to the wound on either side, regardless of wound orientation with respect to the SAM (Figure 13C and D; Figure 13—figure supplement 1A–H). Such a response argues against the possibility that ectopic KAN is the result of interruption of a signal emanating from the meristem that promotes dorsal and represses ventral identity (Sussex, 1955). Instead, it supports the proposal that KAN1 expression is promoted in the vicinity of wounds in general, possibly due to low auxin levels. To test this hypothesis, we repeated these experiments while treating the wounded meristems with combinations of NAA and NPA over a 48 hr period and found that the induction of KAN1 expression around wounds could be completely eliminated if NAA and NPA were combined (Figure 13E–G; Figure 13—figure supplement 1I–L). Although a similar response to wounding was found to occur in vegetative meristems, the wound response typically involved a more substantial reorganization of meristem structure, possibly due to the small size of the vegetative meristem relative to the wounds (Figure 13—figure supplement 1M–O). When new leaves subsequently formed, they were properly oriented with respect to the new meristem organization.

Figure 13 with 1 supplement see all
Wounding induces KANADI1 expression in response to low auxin.

(A and B) Confocal projections of IMs showing predicted auxin distribution (magenta) based on R2D2 expression 12 hr (A) and 36 hr (B) after wounding. Note low predicted auxin levels in the cells surrounding the ablated cells. (C and D) Confocal projections of the IMs showing expression pattern of REV-2 × YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) immediately after ablation (ablated cells indicated a white arrowhead) (C) and 48 hr after (D). Note KAN1 expression (green) has completely surrounded the wounded cells 48 hr after the ablation (D) compared to (C). (E and F) Confocal projections of the IMs showing expression pattern of REV-2 × YPet (red), KAN1−2 × GFP (green) and PIN1-CFP (blue) immediately after ablation (ablated cells are marked by white arrowhead) (E) and 48 hr after ablation and combined NAA and NPA application (F). Note absence of KAN1 expression (green) surrounding the wound when wounding is accompanied by the exogenous addition of auxin and NPA (compare to (D)). (G) Quantification of wounding induced ectopic KAN1 expression upon mock treatment (n = 6/6)), NAA application (n = 2/6)) and NAA + NPA combination (n = 0/6) application on the Arabidopsis IMs expressing REV-2 ×YPet, KAN1−2 × GFP and PIN1-CFP. Scale bars 30 μm (A–F).

https://doi.org/10.7554/eLife.27421.024

Discussion

In this study we shed new light on a long-standing question regarding leaf dorsoventrality in plants – when and how is it established? The early work of Sussex, based on histological analysis and wounding experiments, suggested that initiating leaf primordia require an inductive signal from meristem tissues to specify dorsal cell fate (Sussex, 1955). This proposal has been further supported by a more recent study in tomato (Reinhardt et al., 2005). In contrast, other workers in the field have claimed that dorsoventrality arises directly from radial patterning of the shoot (Husbands et al., 2009; Kerstetter et al., 2001), with some authors also linking the organogenic capacity of the SAM peripheral zone to the meristematic nature of leaf margins (Hagemann and Gleissberg, 1996) and dorsoventral boundaries (Koch and Meinhardt, 1994). Our results support elements of these latter proposals by revealing that organs are pre-patterned by domains of KAN and HD-ZIPIII expression not only in terms of their differentiation but also their position and morphology. The effect of ectopic KAN1 expression at the centre of the shoot on subsequent leaf positioning and growth is particularly striking and indicates that the spatial arrangement of HD-ZIPIII and KAN expression present within organ founder cells is incorporated into organs as they initiate and directs patterns of morphogenesis (Figure 7P–S; Figure 14). However, the exact configuration of this pre-pattern, including its propagation into developing organs, is dynamic and depends on changing auxin levels, although this regulation is unlikely to be direct. While high auxin levels promote HD-ZIPIII and represses KAN1 this is only true for cells not already expressing KAN1. Hence, at sites of incipient organ formation, where auxin is concentrated in between the two expression domains, HD-ZIPIII expression is promoted and can expand peripherally, but only in cells adaxial to where KAN is expressed. At the same time, this high level of auxin prevents KAN expression from expanding acropetally into HD-ZIPIII expressing cells, causing the boundary to remain relatively static with respect to the underlying cells (Figure 14A). However if auxin levels drop, for instance when PIN1 reverses polarity away from cells between the meristem and primordium (Heisler et al., 2005; Qi et al., 2014), HD-ZIPIII expression decreases and KAN expression takes its place (Figure 14B and C). Hence high levels of auxin ‘lock-in’ the spatial arrangement of HD-ZIPIII and KAN expression within organ founder cells at organ inception until a later stage when the pattern becomes auxin independent. In between primordia, where auxin levels are lower, cells continuously transition from expressing REV in the central zone (where HD-ZIPIII expression appears auxin insensitive) to expressing KAN1 as they are displaced peripherally, as also indicated by the expression patterns of these genes in pin1 meristems (Figure 10—figure supplement 2A,E and I).

Conceptual Model.

(A) During organ inception, PIN1 polarities (white arrows) and the alignment of microtubule arrays converge to create an auxin maximum and promote growth oriented towards the epidermal boundary between HD-ZIPIII and KANADI (KAN) expression domains. As auxin accumulates, it promotes the expression of HD-ZIPIII thereby resulting in its extension towards the PIN1 convergence site. At the same time, auxin also prevents the acropetal expansion of KAN. Thus the boundary becomes fixed to the underlying cells. (B) As the primordium grows, PIN1 polarity in cells adaxial to the primordium reverse towards the meristem center and adjacent incipient primordia, thereby creating an auxin depletion zone leading to a reduction in HD-ZIPIII expression. (C) The reduction in auxin results in the re-establishment of KAN expression between the meristem and organ. (D) Auxin in the vicinity of the boundary in adjacent tissues leads to a localized transcriptional response that orients the polarity of surrounding cells into a convergence pattern, most likely via mechanical signals (Bhatia et al., 2016).

https://doi.org/10.7554/eLife.27421.026

Overall these results imply that the dorsoventral patterning of organs results from a four-step process: (i) signals during embryonic development establish concentric patterns of Class III HD-ZIP and KAN gene expression; (ii) the boundary between these domains helps to define sites for auxin-dependent organogenesis i.e. the meristem peripheral zone; (iii) As organs form, dynamic auxin levels modulate the patterns of HD-ZIPIII and KAN expression, thereby helping to position new organs and establish their dorsoventrality; (iv) patterns of HD-ZIPIII and KAN gene expression within organs are stabilized and dictate future patterns of morphogenesis.

The regulation of HD-ZIPIII and KAN expression by auxin is not only relevant to understanding wild type organ development but also for understanding the reorganization of tissue types in response to wounds. Wounds in the meristem outer cell layer specifically alter cell polarities such that PIN1 becomes polarized away from wounds in adjacent cells (Heisler et al., 2010), leading to auxin depletion (Landrein et al., 2015). Consistent with this, we observe KAN1 to be upregulated and REV to be repressed around wounds. Furthermore, such changes can be completely blocked by auxin application. Altogether then, our results strongly indicate that wounds re-pattern peripheral cell identities due to the depletion of auxin in their vicinity. Specifically, we propose that auxin depletion triggers a local commitment to peripheral cell fate (which includes expressing KAN1), thereby providing a parsimonious explanation for the wound-associated disruptions to leaf polarity observed by Sussex and others (Sussex, 1955). However further experiments using the vegetative meristems of plants with larger meristems will be needed to confirm this proposal since we were not successful in altering the dorsoventrality of Arabidopsis leaves by wounding and therefore could not test whether auxin treatment could prevent such alterations.

Our conclusion that auxin promotes dorsal cell fate while inhibiting ventral cell fate contrasts with those of another study reporting that transient high levels of auxin inhibit establishment or maintenance of dorsal fate and/or promote establishment or maintenance of ventral cell fate (Qi et al., 2014). These findings were based on auxin application experiments in tomato that resulted in the ventralization of leaves as well as the observation that Arabidopsis pin1 and pid rev double mutants produce trumpet or rod-shaped shaped leaves. Although our results cannot easily explain the tomato data, since we find that high auxin levels are required to maintain REV expression during organ establishment, we would expect that in pin and pid mutants, lower auxin levels may well result in lower expression levels of REV and possibly other Class III HD-ZIPs, leading to leaf ventralization as reported in these mutant backgrounds (Qi et al., 2014). Further auxin application experiments in tomato that include an analysis of gene expression and auxin level changes may clarify the relationship between auxin and dorsoventral patterning in tomato compared to Arabidopsis.

How do DV boundaries control organ position and shape? Both our data as well as data from previous studies suggest this is through the regulation of auxin perception. For instance, organ initiation requires high auxin levels but auxin can only trigger organogenesis in the peripheral zone (Reinhardt et al., 2000) where the HD-ZIPIII/KAN boundary occurs. A similar relationship holds for leaves where auxin application results in growth but only from the leaf margins, again corresponding to a DV boundary (Koenig et al., 2009). We further show that the localization of auxin response to DV boundaries applies generally, i.e. DR5 expression appears higher at such boundaries compared to the broader predicted auxin distribution, as marked by R2D2. How is this restriction achieved? Imaging data reveals that the locations of PIN1 polarity convergences correspond to a region of cells in between the expression domains of HD-ZIPIII and KAN where their expression is low or absent. Given our results as well as genetic and molecular data indicating that both the HD-ZIPIII and KAN genes repress auxin activity (Huang et al., 2014; Merelo et al., 2013; Müller et al., 2016; Zhang et al., 2017), we propose that this localized absence of expression results in a local de-repression of auxin-induced transcription. Such localized auxin activity has recently been shown to orient cell polarity, including microtubule orientations, in a non-cell autonomous manner (Bhatia et al., 2016). Hence, we suggest that the proposed ability of dorsoventral boundaries to act as organizers analogous to those of the Drosophila wing (Diaz-Benjumea and Cohen, 1993; Waites and Hudson, 1995) rests in-part on their ability to orient cell polarity non-cell autonomously by localizing auxin activity (Figure 14D). The resulting growth may occur either in a periodic fashion along the boundary typical of phyllotaxis and complex leaves, or in a more continuous manner typical of simple leaves, depending on the strength of auxin transport or other modifications to the cell polarity feedback system (Bilsborough et al., 2011; Koenig et al., 2009). Investigating the dynamic consequences of directly juxtaposing REV and KAN expression should enable the testing of these hypotheses as well as lead to a better understanding of how auxin-responsive boundary domains are first established.

Materials and methods

Plant material

Plants were grown on soil at 22°C in continuous light-conditions and cultivated either on soil or on GM medium (1% sucrose, 1 × Murashige and Skoog basal salt mixture, 0.05% MES 2-(MN-morpholino)-ethane sulfonic acid, 0.8% Bacto Agar, 1 % MS vitamins, pH 5.7 with 1 M potassium hydroxide solution).

Construction of transgenes

Multiply transgenic lines were generated by Agrobacterium-mediated transformation into stable transgenic lines or by genetic crossing. The FILp::dsREDN7 and PIN1p::PIN1-GFP transgenes have been described elsewhere (Heisler et al., 2005). pREV::REV-2×VENUS in the T-DNA vector pMLBART (Gleave, 1992) is a modification of pREV::REV-VENUS (Heisler et al., 2005) that contains a translational fusion to two tandem copies of the fluorescent protein VENUS (Nagai et al., 2002). REVp::REV-2×Ypet containing a C-terminal fusion to the 2 × Ypet (Nguyen and Daugherty, 2005) in pMOA36 T-DNA (Barrell and Conner, 2006) was transformed into a PINp::PIN1-CFP line (Gordon et al., 2007b). The KAN1p::KAN1−2 × GFP transgene in pMOA34 T-DNA was created by amplifying 8.7 kb of KAN1 (At5g16560) genomic sequences with primers KAN1g F and KAN1g R (Supplementary file 1A) as a translational fusion to a 9 Ala linker and 2 × GFP followed by OCS terminator sequences. When transformed into kan1-2 kan2-1 segregating plants, this construct complements the mutant phenotype. The triple marker line was generated by transforming KAN1p::KAN1−2 × GFP into a REVp::REV-2 × Ypet; PIN1p::PIN1-CFP transgenic line. KAN1p::KAN1−2 × CFP or KAN1p::KAN1−2 × Ypet containing a fusion to 2 copies of CFP or Ypet, respectively, were constructed similarly. KAN1p::KAN1−2 × Ypet and PIN1p::PIN1-GFP were combined in T-DNA vector BGW (Karimi et al., 2002) by Gateway technology (Invitrogen) for generation of a double marker transgenic line.

The KAN1 cDNA was amplified by PCR with primers K1 cDNA F and K1 cDNA R to generate C-terminal translational fusion to a 9 Ala linker followed by single GFP or 2 × GFP followed by pea rbcS E9 terminator sequence (Zuo et al., 2001) and cloned into the pOp6/LhGR two-component system (Craft et al., 2005) for dexamethasone-inducible misexpression.

An ATML1p::LhGR driver containing 3.4 kb of the L1-specific ATML1 gene (At4g21750) fused to the chimeric LhGR transcription factor and a 6Op::KAN1-GFP expression construct in a pSULT sulfadiazine-resistant T-DNA vector (ATML1 >>KAN1 GFP) was generated. The pSULT T-DNA vector was derived from pMLBART by replacing the NOSp::BAR gene with 1’−2’p::SUL (Rosso et al., 2003), a plant selectable marker that confers resistance to sulfadiazine herbicide to create pSULT. A CLV3p::LhGR driver containing 1.49 kb of upstream regulatory sequences was PCR amplified with primers CLV3p F and CLV3p R along with 1.35 kb of downstream regulatory sequences with primers CLV3utr F and CLV3utr R was combined with 6Op::KAN1−2 × GFP in pSULT T-DNA vector (CLV3 >>KAN1-2×GFP).

ATML1 >>REVr-2×VENUS is a sulfadiazine-resistant T-DNA vector to misexpress microRNA resistant REV-2 ×VENUS fusion, where 6Op::REVr-2×VENUS was constructed by cloning a 1148 bp BamHI-XcmI microRNA resistant REV cDNA (a gift from J. Bowman) harbouring two previously characterized silent mutations that disrupt the binding of MIRNA 165/166 to the coding sequence of REV as previously described (Emery et al., 2003) downstream of the 6Op and in frame with the wild type REV-2 ×VENUS coding sequences. miR166Ap::GFPER T-DNA construct was kindly provided by K. Nakajima (Miyashima et al., 2011). The MIR165/166 biosensor was created based on the design presented by Smith Z. R. et al. (Smith and Long, 2010) in the AlcR/AlcA expression system (Roslan et al., 2001) for ethanol-inducible expression. The sequences conferring MIR165/166 sensitivity from the REV coding sequence (REV) and the sequences conferring MIR165/166 insensitivity (REVr) were fused to mCherry (Shaner et al., 2004) with endoplasmic reticulum localization sequences mCherryER, which was synthesized de novo (Genscript). The MIR165/166-mCherryER biosensors (both biosensor and control) were cloned as HindIII-BamHI fragments downstream of the AlcA regulatory sequences in the UBQ10p:AlcR_BJ36 plasmid vector. UBQ10p:AlcR was constructed by cloning the UBQ10 promoter 2 kb fragment upstream of AlcR and the OCS terminator. Both UBQ10p::AlcR and AlcA::REV-mCherryER or AlcA::REVr-mCherryER components were combined in the T-DNA vector pMOA34.

The WOX1p::2 × GFP-WOX1 construct in pMLBART T-DNA vector was generated as follows: 2.2 kb of WOX1 (At3g18010) upstream promoter sequence was amplified with primers WOX1p F and WOX1p R and cloned using restriction enzymes KpnI and BamHI. 3.6 kb of WOX1 coding sequence plus 1.65 kb 3’-regulatory sequences was amplified from wild-type Col-O genomic DNA with the primers WOX1g F and WOX1g R and cloned using restriction enzymes BglII and SpeI. 2 copies of GFP were inserted in frame at the start of the WOX1 coding sequence at the BamHI and BglII sites. A double marker was generated by transforming the WOX1p::2 × GFP-WOX1 into a PIN1p::PIN1-CFP transgenic line.

The PRSp::PRS-2×GFP construct in pMOA34 T-DNA vector was made by amplification of 3.9 kb PRS (At2g28610) genomic sequence (similar to [Shimizu et al., 2009]) with primers PRSg F and PRSg R to create a C-terminal fusion to 2 × GFP followed by OCS 3’regulatory sequences. Marker combinations were generated by transforming the PRSp::PRS-2×GFP into either a PIN1p::PIN1-CFP transgenic line or into REVp::REV-2×YPET PIN1p::PIN1-CFP line. (PRSp::PRS-2×GFP) and (KAN1p::KAN1−2 × CFP) were combined in T-DNA vector BGW (Karimi et al., 2002) by Gateway technology (Invitrogen) for generation of a double marker transgenic lines.

A short tandem target mimic (STTM) construct to target MIR165/166 (Yan et al., 2012) was generated in the pOp6/LhGR two-component system for dexamethasone-inducible expression with a UBQ10p::GRLh driver. STTM MIR165/166-88 sequence (Yan et al., 2012) was synthesized de novo (Genscript) and cloned downstream of 6 × Op to create 6 × Op::STTM 165/166. Both components were combined in a sulfadiazine T-DNA pSULT (UBQ10 >>STTM 165/166).

ATML1 >>PHVr is a sulfadiazine-resistant T-DNA vector containing a mutated version of PHV cDNA (a gift from J. Bowman) with a Gly to Asp amino acid change that disrupts the miRNA165/166 binding site in the PHV gene (McConnell et al., 2001). 6 × Op::PHVr was constructed by cloning a 2.6 kb XhoI-BamHI PHVr cDNA downstream of 6 × Op and upstream of pea3A terminator sequences. ATML1 >>PHVr was transformed into a REVp::REV-2×YPet; PIN1p::PIN1-CFP transgenic line.

DR5−3 × VENUS v2 reporter gene (Liao et al., 2015) was a generous gift from Prof Dolf Weijers (Wageningen University). The line R2D2 PIN1p::PIN1-GFP was described previously (Bhatia et al., 2016). The seeds for pin1-4 mutant were a kind gift from Prof. David Smyth (Monash University) and the heterozygote plants were crossed wild type plants carrying pREV::REV-2xVENUS and pKAN::KAN-2xGFP. pCLV3::GFP-ER has been described previously (Gordon et al., 2009).

Dexamethasone induction

For inducible gene perturbations in the vegetative SAM, seeds were germinated directly on GM medium containing 10 μM Dexamethasone (Sigma, stock solution was prepared in Ethanol). Seedlings were then dissected for imaging at 4 DAS, 5 DAS or 7 DAS depending on the experiment. For DEX induction in the IM, 10 μM DEX solution containing 0.015% Silwet L-77 was applied to the IM every second day three times. Inflorescences were then dissected and imaged. The number of T2 inducible transgenic lines that exhibit the presented phenotypes and the frequencies of phenotypes amongst imaged plants is shown in Supplementary file 1B and associated caption.

Confocal microscopy

Plants were dissected for imaging as previously described (Bhatia et al., 2016; Heisler and Ohno, 2014) and imaged with a Leica SP5 Upright confocal microscope using an Argon laser. The objective used was a water-immersion HCX IRAPO L25x/0.95 W (Leica). Excitation for CFP is 458 nm, GFP is 488 nm, YFP (YPet and VENUS) is 514 nm and tdTomato is 561 nm. Emission signal were collected at 460–480 nm for CFP, 490–512 nm for GFP, 516–550 nm for YFP (YPet and VENUS), and 570–620 nm for tdTomato. The resulting z-stacks were processed using the Dye Separation (Channel mode or automatic mode) function available in the LAS AF program in order to separate the GFP channel from the YFP (YPet or VENUS) channel. Three software packages: LAS AF from Leica, Imaris 8.0.2 by Bitplane and FIJI (https://fiji.sc) were used for data analysis. Ratios for R2D2 were calculated as described previously (Bhatia et al., 2016).

Measurement of distance between organs

For distance measurements between oppositely positioned leaves on plants transgenic for inducible KAN1 (CLV3 >>KAN1-2×GFP – see above) the measurement tool from Imaris 8.0.2 (Bitplane) was used. For comparisons to control, untreated seedlings grown on GM were compared to seedlings grown on DEX for 5 days. t-Test was performed using Excel.

Chemical treatments for auxin depletion in the inflorescence meristems

500 mM stock solutions of auxinole, yucasin and L-Kynurenine were prepared separately in DMSO. The stocks were diluted in 1 mL 0.1M phosphate buffer in sterile water to make a working solution containing all the three drugs (0.2 μL of each stock) to a final concentration of 100 mM each. The final concentration of DMSO in the working solution containing all the three drugs was 0.06%.

Treatments were carried out on the inflorescence meristems of whole plants transplanted from soil to boxes containing GM medium supplemented with vitamins. The older flowers were removed as described (Heisler and Ohno, 2014). The plants were chosen such that the stem of the meristem was a few millimeters above the rosette to prevent the drug solution from dispersing into the surrounding medium. After imaging, the meristems were carefully dried using a thin strip of sterile filter paper to remove excess water. Approximately 50 μL of the drug solution was added directly to the meristem, drop-wise. The meristems were treated only once in a time course of 12–18 hr.

NPA treatment on seedlings carrying R2D2 and DR5 markers

Seedling aged 3 DAS were dissected to expose the meristem and the first leaf pair as described (Bhatia et al., 2016). After imaging, seedlings were transferred to new GM medium containing plates and blotted dry with thin strips of sterile paper. 5–10 μL of 100 μM NPA in sterile water (100 mM stock in DMSO) was added directly to the dissected seedlings every twenty-four hours for three days in total.

Auxin treatment on seedlings carrying PRS and WOX1 markers

Seedlings aged 4DAS (days after stratification) were dissected to expose the meristem and the first leaf pair as described (Bhatia et al., 2016). 5 mM NAA (1M stock in 1M KOH) solution was prepared in liquid GM medium. Seedlings were then immersed in 100 μL of NAA containing medium in individual wells in 96 well plate and grown under continuous light without shaking for 12 hr.

NAA, NPA and combined NAA and NPA treatments

For treatments on the vegetative meristems, 3DAS old seedlings expressing REV, KAN and PIN1 markers were dissected and imaged first. Samples were then treated with a combined solution of 100 μM NPA and 5 mM NAA (both diluted in water) or 100 μM NPA alone or 5 mM NAA alone. Samples were imaged and treated every 24 hr for 72 hr in total. For treatments on the infloresecnce meristems, flowering plants expressing REV, KAN1 and PIN1 markers and CLV3 GFP-ER marker were transferred from soil to boxes containing GM vit medium as described above. The infloresecnce meristems were dissected to remove older flower buds covering the meristem. The meristems were imaged, dried and treated with 10 μL of combined solution of 100 μM NPA and 5 mM NAA. Plants were imaged at the required timepoints (6 hr, 15 hr and 48 hr). Plants were treated with NAA and NPA after every timepoint imaging.

Local application on pin1-4 mutants

pin1-4 mutant plants expressing REV and KAN1 markers with visibly smooth pin meristems were transferred from soil to boxes containing GM-vitamin medium. Leaves covering the meristems were dissected away. Control paste was applied locally to follow the meristem orientation over time. The meristems were imaged, dried and lanolin paste carrying 5 mM IAA (3-Indoleacetic acid) was administered locally at the meristem periphery. Plants were then imaged 15–17 hr later to follow REV response at the site of auxin paste application. For this, first a z-stack representing an overview of the meristem was recorded. The paste was then scrapped off using a microfiber of the blotting paper carefully without damaging the meristem. A z stack was recorded again to follow REV and KAN expression at the site of paste application. A separate set of treated plants were imaged 24 hr later to follow REV and KAN expression in the grown out primordia at the site of auxin paste application.

Pulsed laser ablations

Laser ablations on the inflorescence meristems were carried out using the Mai Tai multi-photon laser from Spectra Physics, which was controlled with LEICA SP5 confocal software. Z-stacks were acquired prior to ablation. Single cells were targeted one after the other using bleach point mode. Ablations were carried out at 800 nm with an output power of ~3W. Each pulse was shot for 1–15 milliseconds. Usually ablations were accomplished within 1–3 bursts of the laser. Ablated cells could be visually identified as their nuclei exploded resulting in unusual auto fluorescence. Z stacks were acquired immediately after the ablations.

Auxin and auxin plus NPA combined treatments on ablated inflorescence meristems

After ablations, the meristems were carefully blotted dry using thin strips of sterile filter paper. 20 μL 5 mM NAA in sterile water (0.5M stock in 1M KOH) or 20 μL of a solution containing 5 mM NAA and 100 μM NPA in sterile water (100 mM NPA stock in DMSO) or mock solution were added directly to the meristems every 24 hr for 48 hr in total.

Model for auxin and PIN1 dynamics

We developed a computational model to understand how interplay between the HD-ZIPIII and KAN pattern and PIN1 dynamics can influence auxin transport and primordia initiation. The model introduces a dependence on KANADI and REVOLUTA into previous models describing PIN1 and auxin dynamics (Bhatia et al., 2016; Heisler et al., 2010; Heisler and Jönsson, 2006; Jönsson et al., 2006; Sahlin et al., 2009). In the model, auxin resides in cell compartments and is able to move between cells either via active transport mediated by PIN1 proteins or passively via a diffusion-like process. PIN1 proteins cycle between cytosol and membrane compartments and a quasi-equilibrium model is used for determining its membrane localization at any time point. Auxin generates a signal able to polarize PIN1 in neighboring cells, i.e. a high auxin concentration increases the amount of PIN1 proteins in the neighboring cell membrane facing that cell. The molecule X in our model acts as a mediator of the signalling between auxin and PIN1, and the signal has previously been interpreted as a molecular (Jönsson et al., 2006), or mechanical stress signal (Heisler et al., 2010). In the model, the signal X is activated by auxin, and repressed by KAN and REV. The equations governing the dynamics of the molecules are

dAidt=cAdAAi+1Vi[Dj{Ni}aij(AjAi)+Tj{Ni}aij(PjiAjPijAi)],
dPitotdt=cP-dPPitot,
dXidt=VXAinXA(KXAnXA+AinXA)KXRnXR(KXRnXR+RinXR)KXKnXK(KXKnXK+KinXK)dXXi,

where Ai is the auxin concentration and Xi is the level of the signalling molecule in cell i.Ni is the set of cells neighboring cell i,Vi is the cell volume of cell i, and aij=aji is the cell wall area for the wall section between cells i and j.Pi[tot] is the total PIN1concentration in the cytosol and membrane compartments of cell i. Membrane-bound PIN1 appears in the equation as Pij, which is the PIN1 concentration in the membrane compartment of cell i that faces cell j. A simple linear feedback between the signal Xj and Pij is used and a quasi-stable assumption, introduced in Jönsson et al. (2006), leads to

Pij=Pi[tot][1-kp+kpXj]fp+k{Ni}[1-kp+kpXk],

where fp=kn/kx is the ratio of endocytosis and exocytosis rates and kp sets the relation between symmetric and polarized exocytosis (dPij/dt=kx[(1kp)+kpXj]PiknPij, where Pi is the PIN1 in the cytosol compartment). The feedback between auxin and PIN1 is identical to previous models if the KAN and REV factors in the dXi/dt equation are 1 (e.g. by setting KAN and REV to zero in all cells), while the polarising signal becomes reduced in regions where KAN and/or REV are expressed. This effect tunes the interaction between the dorsoventral patterning and PIN1/auxin dynamics (cf. Figure 8B and D).

The model was simulated using the in-house developed software tissue (available at the gitlab repository https://gitlab.com/SLCU/TeamHJ/tissue; please see README.txt file in Supplementary file 2 for details). A copy is archived at https://github.com/elifesciences-publications/TeamHJ-tissue. Files defining the model with parameter values (Supplementary file 1C) and initial configuration of (static) cell geometries and KAN and REV expression domains are provided in Supplementary file 2. The simulations use a 5th order Runge-Kutta solver with adaptive step size (Press et al., 2007), and initial auxin, PIN1 and X concentrations are set to zero in all compartments.

Generation of geometrical template

The model defined above was run on a template containing a predefined KAN/REV pattern (provided as Supplementary file 1C, Figure 9C). The geometry of the template was generated by a combination of cell/wall growth and mechanical interactions together with a shortest path division rule (Sahlin and Jönsson, 2010). A KAN/REV pattern was generated by the equations

dKidt=VKrinKKKnK+rinK-dKKi,
dRidt=VRrinRKRnR+rinR-dRRi.

This system was run to equilibrium on the above-mentioned template. In the above equations ri is the distance of cell i from the center of the template. The parameters were set to VK = VR = dK=dR = 1, KK = 30, KR = 25, nK = nR = 20. These parameters are set such that two distinct domains are created, with a small overlap of low KAN and REV concentrations in the boundary between these regions. To make the transition of KAN concentrations between domains sharper, KAN concentrations were set to 0 (1) if Ki ≤0.5 (>0.5).

Real time PCR

4 day-old wild-type Ler seedlings were immersed into 5 mM NAA solution in liquid GM medium, and grown under continuous light without shaking for 15 hr. Cotyledons, hypocotyl, and roots were removed under dissecting scope, and only shoot meristem and the first pair of leaves were collected and immediately frozen in liquid nitrogen. Each biological replicate, represents tissue from 10 to 15 individual seedlings. Five biological replicates were collected for both mock and NAA treatment. RNeasy Mini kit (Qiagen) was used according to manufacturer’s instruction for RNA extraction. 1 microgram of RNA was used for cDNA preparation using Super script III reverse transcriptase for Q-PCR analysis. Q-PCRs were performed in a StepOne Plus Real Time PCR system thermo cycler (The applied bio systems) using 20 μl of PCR reaction containing 10 μL of SYBR Green mix (Roche), 1 μl of primer mix (10 μm), 2 μl of 1:10 diluted cDNA and 7 μl of water. Transcript levels were normalized to ACTIN2 transcript levels. Data was analyzed using the 2−ΔΔCT method. A freely available online tool was used for analysis using an unpaired t-test of the RT-PCR results: http://graphpad.com/data-analysis-resource-center/. For p-value calculation, data entry format with mean, SD and N was used. Measurements and calculations for all replicates are provided in Figure 8—source data 1.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
    Flower Development
    1. MG Heisler
    2. C Ohno
    (2014)
    431–440, Live-imaging of the Arabidopsis inflorescence meristem, Flower Development, Springer.
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
    Numerical Recipes: The Art of Scientific Computing (3rd edn)
    1. WH Press
    2. SA Teukolsky
    3. WT Vetterling
    4. BP Flannery
    (2007)
    New York, NY, USA: Cambridge University Press.
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
    Morphogenesis in Solanum tuberosum I: experimental investigation of leafdorsiventrality and orientation in the juvenile shoot
    1. IM Sussex
    (1955)
    Phytomorphology 5:286–300.
  65. 65
  66. 66
    Phantastica - a Gene Required for Dorsoventrality of Leaves in Antirrhinum-Majus
    1. R Waites
    2. A Hudson
    (1995)
    Development 121:2143–2154.
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72

Decision letter

  1. Dominique C Bergmann
    Reviewing Editor; Stanford University/HHMI, 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 "Cell type boundaries organize plant development" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors and the evaluation has been overseen a Senior Editor. The reviewers have opted to remain anonymous.

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.

The reviewers and editors all agreed that this work focused on a long–standing issue of organ formation in plants and that the manuscript, using beautiful imaging and side–specific markers, presented convincing and high quality evidence for the presence of the boundary within the SAM and the influence of the location and form of this boundary on the position of leaf initiation, leaf polarity and the resulting leaf form. While this was appreciated, there was also a strong feeling from several reviewers that much of this, beautiful as it is, was largely confirmatory. In addition, there were concerns that some manipulations had technical issues (notably the long treatment with cocktails of auxin drugs) or use of multiple types of meristems that made the results somewhat ambiguous.

Reviewer #1:

Caggiano et al., address a long–standing issue of organ formation in plants. How do leaves acquire their characteristic arrangement around the shoot meristem and their flattened shape with different ad/abaxial domains? Is there a signal from the meristem, or are organs pre–patterned based on information in the meristem? Using detailed live–cell imaging and a number of genetic and experimental manipulations, the authors argue for pre–patterning in the meristem of ad/ab genes (primarily HD–ZIPs and Kanadi's) and for in instructive role of auxin in defining cell identity, potentially by influencing the expression domains of REV and KAN.

Overall, this work was interesting, the imaging of gene expression patterns was impressive and the overall view of pattern formation seems reasonable. I particularly like the extension of models from their previous SAM work to accommodate the new data provided here. Also, the phenotypes and diagrams explaining their potential origin in Figure 6 are quite cool! The authors point out alternative conclusions about D/V pattern that come from recently published work in tomato. I am afraid I don't have the background to evaluate the difference between these studies.

Reviewer #2:

Organ polarity is influenced by the organ placement relative to the shoot apical meristem (SAM). Several hypotheses have been proposed with respect to the nature of the positional information that directs organ polarity specification. One model suggests that pre–patterning in the SAM peripheral zone specifies the dorsal and ventral (abaxial and adaxial) sides of the leaf. In the current manuscript, using beautiful imaging and side–specific markers, the authors clearly show the existence of polarity within the peripheral zone of the SAM. They further show that the site of organ initiation is determined by the location of this dorsal–ventral boundary within the SAM. Alteration of this boundary by manipulation of the expression domain of KANADI, which promotes ventral identity, influences the initiation site. They further analyze the effect of auxin treatment and wounding on the placement of the boundary.

The manuscripts presents convincing and high quality evidence for the presence of the boundary within the SAM and the influence of the location and form of this boundary on the position of leaf initiation, leaf polarity and the resulting leaf form. The question is whether these findings are mainly confirmatory or bring new biological insights. The dependence of organ initiation and blade outgrowth on dorsal–ventral polarity has been shown before. It also has been shown that there is preexisting polarity within the peripheral zone of the SAM, but this manuscript shows it more clearly, including the pre–existing polar expression of ventral and dorsal determinants. The causal effect of auxin on this boundary is less convincing, as it is hard to tell how a combined NPA and auxin treatment or a treatment of a combination of 3 auxin biosynthesis and response inhibitors affects the different aspects of the process. The effect of these treatments is shown after a relatively long time, and these effects on the localization of the examined markers could result from an effect on the domain patterning within the SAM rather than an effect on REV expression.

Reviewer #3:

Caggiano et al., address the question, how the dorso–ventral (DV) pattern of leaves is established. Using fluorescent protein reporters they show that DV patterning of leaves correlates with central–peripheral patterning of the shoot apical meristem (SAM). The experimental data related to this paragraph are shown in 5 figures (Figures 15), of which many details are not addressed in the text. The novelty of this part is limited, because similar data have been published earlier (e.g. McConnell et al., 2001; reviewed by Barton et al., 2009).

Subsequently, the authors provide evidence that HD–ZIPIII and KAN genes repress leaf primordium initiation in the vegetative SAM. However, the question whether this is a direct influence on organ initiation or an indirect influence on SAM maintenance was not answered, because SAM organization in the arrested plants was not analyzed. Similar results were obtained by Kerstetter et al., 2001.

In the next paragraph, the consequences of ectopic KAN expression in the central domain of the SAM are studied. Most seedlings initiate several abnormal organs before their growth is arrested. The distinct organ phenotypes observed can be explained by the orientation of boundaries within organ founder cells.

Furthermore, the manuscript focuses on the interplay between auxin, HD–ZIPIII–expression and KAN–expression. Experiments using either NAA and NPA or a combination of auxinole, yuccasin and kyn led to the conclusion that high levels of auxin promote REV expression, whereas depletion of auxin promotes KAN expression. In this context, the treatment with a combination of three different drugs, which is necessary to see the described extension of KAN expression, may also have a different effect on gene expression than the one mediated through the observed depletion of auxin. The interplay of auxin, HD–ZIPIII expression and KAN–expression has also been addressed in previous studies (e.g. Izhaki and Bowman, 2007; Ilegems et al., 2010) with similar results.

In the last paragraph, the authors show that wounding in inflorescence meristems (IMs) leads to a depletion of auxin around the wound followed by an accumulation of KAN, which could be eliminated by an NAA + NPA treatment. Again, part of these findings were already known (Heisler et al., 2010; Landrein et al., 2015). In addition, IMs may react different from vegetative meristems.

Taken together, the manuscript presents a nice set of experiments supporting the view that DV organization of leaves is pre–patterned by the expression domains of REV and KAN in the SAM. However, previous studies revealed many similar results. Therefore, the study by Caggiano et al., presents only an incremental advance in our understanding.

Aiming for an explanation of the origin of DV patterning of leaves, it seems problematic to mix results from experiments using vegetative meristems and using inflorescence meristems, because a flower primordium is not a structure showing a DV pattern like a leaf primordium.

The wounding experiments offer an alternative explanation for the observations made by I. Sussex, but we think they do not rule out the existence of a signal from the main meristem to the developing leaf primordium.

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

Thank you for resubmitting your work entitled "Cell type boundaries organize plant development" for further consideration at eLife. Your revised article has been favorably evaluated by Christian Hardtke (Senior editor) and three reviewers (mix of new and previous), one of whom is a member of our Board of Reviewing Editors.

There was, as in the first version, considerable appreciation of the imaging quality and evidence for a prepattern mechanism. There was also concern that the role of auxin in the process was not strongly supported by the provided data. Specifically, there was continued concern about the length of auxin treatments and the potential for KAN and REV responses to be secondary. Additionally, because the highest levels of auxin–signaling activity are found in the boundary region, but REV is precisely not expressed there, it was unclear what REV induction there by NPA+NAA treatment means. Moreover, how would this observation relate to a potential auxin–mediated maintenance of REV expression in the central region of the meristem, where there appears to be very little auxin–signaling activity?

After considerable deliberation, we have decided to consider a revised manuscript, IF it can address this critical concern about the proposed mechanism of auxin–mediated pre–patterning via REV activation and KAN repression. The misexpression of boundary genes (reviewer 1) could be one way to add more evidence, but alternative experiments would also be acceptable.

A revised version should also take care to tone down the discussion related to the "Sussex signal" because those experiments were done in floral meristems where the D/V polarity is limited. Ideally the issue would be addressed experimentally (e.g. using tomato apices), but the authors could also consider simply rewriting this section.

Reviewer #1:

In this revision, the authors have addressed my concerns about data presentation and have clarified many of the confusing passages of the previous version.

In this current version, I see that they have generated a very high resolution description of organ boundary gene expression and have a plausible model for auxin (though like most auxin models, there are enough feedbacks that it’s hard to find a way to experimentally disprove it). To my mind a manipulation in which they misexpress boundary genes in such a way that polarity becomes locally reoriented (something like making fly wing clones that create artificial boundaries) would be the real test of their models. It seems to me that reagents to do these experiments should be available, and I would be very supportive of this manuscript if such data could be added.

Reviewer #2:

In the revised manuscript, Caggiano et al., address some of the concerns raised by the reviewers and explain and clarify the new findings presented in this paper. As before, the imaging is impressive and convincing with respect to the pre–patterning of in the polarity in the SAM and the causative role of the boundary in organ initiation.

The authors better clarify the reason for the use of multiple drugs to assess the role of auxin in the patterning. However, I still think that the presented data on the effect of auxin does not support the interpretation on the role of auxin in specifying the spatial patterning in the SAM periphery. The time from application to effect is still long, and the effect on KAN and REV may be secondary to other effects. Combined with the relatively broad application, the results in my opinion do not support the proposed mechanism of auxin–mediated pre–patterning via REV activation and KAN repression.

Reviewer #4:

This manuscript addresses a long–standing question in plant biology, the acquisition of dorso–ventral (DV; or adaxial/abaxial) polarity in plant leaves. The authors use fluorescence live imaging combined with transgenic manipulation of gene–expression domains and pharmacological treatments to characterize the role of the boundary region between dorsal and ventral gene–expression domains in the meristem in leaf–polarity acquisition. They confirm and provide further detail on the pre–pattern in the shoot meristem involving centrally expressed HD–ZIPIII and peripherally expressed KANADI1 genes, and show that the boundary between the two domains, where neither gene is expressed strongly, is the site of auxin convergence driven by PIN1 and the highest auxin–signaling activity. Overexpressing either REV or KAN1 throughout the meristem epidermis abolishes organ initiation, consistent with their documented role in suppressing auxin activity. Together this suggests that the auxin maxima can only form in the boundary domain to trigger organ outgrowth there. Forcing KAN1 expression in the meristem centre and thus creating a new DV boundary there can lead to leaf radialization or even polarity reversal, indicating that the pre–pattern in the meristem can have an instructive function for leaf DV polarity. Having demonstrated that the pattern of REV and KAN1 expression determines the pattern of auxin responsiveness in the meristem, the authors then ask whether auxin in turn influences the expression pattern of these genes. Treating meristems with NAA and NPA causes an increase in the REV expression domain into the boundary region and can prevent the re–establishment of KAN1 expression in the periphery separating an outgrowing primordium from the meristem; conversely, inhibiting auxin signaling leads to an encroachment of KAN1 expression towards the meristem centre. Lastly, the authors show that the previously demonstrated auxin depletion around wound sites leads to an upregulation of KAN1 expression around the wound. This latter observation is offered as an explanation for the leaf radialization seen after separating a leaf primordium from the meristem in Sussex' classical experiments.

How DV polarity in leaves is determined is a question of fundamental importance and broad interest in plant development. As the ongoing interest in the quest for the proposed Sussex signal indicates, the work presented is clearly highly timely. The manuscript provides a mix of rather confirmatory and novel results. Important novel aspects are the demonstration that disturbing the DV expression pre–pattern in the meristem can lead to leaf–polarity defects; the finding that auxin in turn can modulate this DV expression pre–pattern; and the observation that peripheral/abaxial identity is induced surrounding wounds in the meristem.

I have the following comments:

1) The finding of a likely causal role of the pre–pattern on leaf polarity is an important advance. Also, the observation that auxin can modulate the DV expression pattern is interesting, yet its biological significance appears somewhat unclear. As the authors show in Figure 1 and in their model in Figure 9, the highest levels of auxin–signaling activity are found in the boundary region. However, it is precisely there that REV is not expressed. Thus, what does it mean that REV can be induced there by NPA+NAA treatment? And how does this observation relate to a potential auxin–mediated maintenance of REV expression in the central region of the meristem, where there appears to be very little auxin–signaling activity? Have the authors attempted to incorporate the auxin effect on REV expression into their model? Does this provide any information about the possible significance of this interaction?

2) The alternative explanation offered for the effects of wounds separating a primordium from the meristem is novel, interesting and potentially of broad significance. However, its impact appears limited, as the KAN1 upregulation around the entire wound is currently only shown clearly for Arabidopsis inflorescence meristems (while the seedling meristems undergo more dramatic re–organization, making their behavior more difficult to interpret), where the action of a Sussex signal is hard to demonstrate due to the limited DV polarity of floral meristems. This means that the alternative explanation is not backed up by functional experiments. Ideally, the authors would directly test in a different system, such as the tomato apex, whether NAA+NPA treatment after appropriate wounding could prevent organ abaxialization.

https://doi.org/10.7554/eLife.27421.030

Author response

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

Reviewer #2:

Organ polarity is influenced by the organ placement relative to the shoot apical meristem (SAM). Several hypotheses have been proposed with respect to the nature of the positional information that directs organ polarity specification. One model suggests that pre–patterning in the SAM peripheral zone specifies the dorsal and ventral (abaxial and adaxial) sides of the leaf. In the current manuscript, using beautiful imaging and side–specific markers, the authors clearly show the existence of polarity within the peripheral zone of the SAM. They further show that the site of organ initiation is determined by the location of this dorsal–ventral boundary within the SAM. Alteration of this boundary by manipulation of the expression domain of KANADI, which promotes ventral identity, influences the initiation site. They further analyze the effect of auxin treatment and wounding on the placement of the boundary.

The manuscripts presents convincing and high quality evidence for the presence of the boundary within the SAM and the influence of the location and form of this boundary on the position of leaf initiation, leaf polarity and the resulting leaf form.

We thank the reviewer for these positive remarks.

The question is whether these findings are mainly confirmatory or bring new biological insights.

Although we agree that, in some respects, the paper presents confirmation of one theory over others regarding how polarity is established and the role of a “Sussex signal”, we feel that this is definitely not the whole story and in any case, that it is quite a big deal.

Firstly, the debate regarding a pre–pattern or inductive signal has been going on for 70 years and recent papers have highlighted the fact that the issue is still not resolved. For instance, an extensive review published just recently on this exact topic by Cris Kuhlemeier and Marja Timmermans in Development pushes the Sussex signal idea. In this review they “postulate possible candidates for the ‘Sussex signal’ – the elusive meristem–derived factor that first ignited interest in this important developmental problem.” (Kuhlemeier and Timmermans, 2016).

Regarding other novel findings, our results on the role of auxin on REV and KAN expression are entirely new and provide great clarity for understanding exactly how dorsoventrality is established. We show that although there is a pre–pattern of dorsoventral gene activity in the SAM, this pattern is highly dynamic, being dependent on changing auxin levels and only becomes anchored to the cells when high auxin levels are present at organ inception. This finding reveals that auxin not only acts to trigger organ growth but also, that it plays a key role in establishing patterns of organ differentiation and the continuance of the dorsoventral boundary itself. We now provide extra data to back these conclusions (see below) and highlight it more clearly in the text and with our new Figure 14. Our unique high resolution data also enables us to propose the first explicit model based on known molecular players to explain how these boundaries actually work. We have added text to the intro and discussion as well as a diagram (Figure 14) to highlight this point.

These points are reiterated in more detail below.

The dependence of organ initiation and blade outgrowth on dorsal–ventral polarity has been shown before. It also has been shown that there is preexisting polarity within the peripheral zone of the SAM, but this manuscript shows it more clearly, including the pre–existing polar expression of ventral and dorsal determinants.

Regarding the previously published work on organ initiation / blade outgrowth and DV polarity, the very fact that the pre–pattern issue has not been settled confirms to us that the clues gained from these studies haven’t provided the clarity we need in order to know what’s really going on. For instance, one of the main figures from the recent Kuhlemeier and Timmermans review (Figure 6) shows the rough expression domains of HD–ZIPIII and KAN in the SAM but the rest is simply a model they propose (Kuhlemeier and Timmermans, 2016). There are several errors including the proposal that KAN1 is excluded from the primordium initially. Many other reviews indicate no pre–pattern at all (e.g. (Kalve et al., 2014; Liu et al., 2012; Nakata et al., 2012). Even when PIN1 and REV were examined at high resolution together in our previous paper (Heisler et al., 2005), we could only suggest that the DV boundary may be positioned by PIN1. Only now, by looking at KAN1 and testing the role of the boundary directly are we able to demonstrate the opposite relationship, i.e. the boundary influences where PIN1 localizes auxin.

In terms of functional data, Izhaki et al., (2007) shows through genetics that both KAN and HD–ZIPIII activities are required for repressing ectopic organogenesis (Izhaki and Bowman, 2007). However, the explanation given does not mention dorsoventral boundaries. Rather, the authors propose that the KAN genes control PIN1 expression and therefore auxin flow and that the HD–ZIPIIIs are seen as necessary for SAM formation. Ectopic organs therefore form because of aberrant auxin transport patterns and a missing SAM. This paper was published after the polar expression domains of KAN1 and REV in embryos were reported and yet no link was made between the positioning of organs and DV boundaries because the exact spatial relationship between these factors and PIN1 was not examined.

In contrast, our paper not only reports how all relevant factors relate spatially, but also, that the boundaries of the SAM are contiguous with that of the leaf and that, just as ectopic expression of REV and KAN blocks leaf blade growth, these factors also block leaf initiation. This has not been reported previously. We think this in itself is a major finding because now we can start to think of these seemingly distinct development processes as being not so different and controlled by one master patterning mechanism.

Besides concretely linking the activity of DV boundaries to organ formation, the high–resolution data we provide also forms the basis of a new hypothesis for how these boundaries actually work. By being able to see that PIN1 initiates organs specifically in a small gap between the REV and KAN domains and linking this with our other novel finding that HD–ZIPIII activity blocks organ initiation, we were able to present the first explicit model based on known molecular players to account for DV boundary activity. We feel this is a major step forward since we now have a testable hypothesis to help understand boundary function.

Additional major findings not reported previously include:

1) A new relationship between auxin and cell differentiation. In our revised version, we have further highlighted this discovery by adding additional data showing that the changes to REVKAN patterning in the inflorescence meristem caused by high auxin levels are even more dramatic in the vegetative meristem, where the reestablishment of KAN1 expression between the meristem and leaf can be completely blocked (Figure 11A–E).

2) Understanding how meristems reorganize themselves in response to wounds is a major question in plant development. So far, we know that factors such as WUS re–pattern themselves in response to wounds and we have models for how this occurs. However more recently it’s been discovered that wounding alters cell polarity and as a consequence, auxin becomes depleted around wounds. What has been missing then is an understanding of the functional consequence of this depletion. Our paper now reveals that such changes in auxin play a key role in enabling the meristem to re–pattern DV gene expression. This work also enables us to provide a novel, straightforward and parsimonious explanation for the observations that originally prompted Sussex’s inductive signaling hypothesis.

In summary, we feel strongly that the imaging data we provide does more than simply add detail to what is already published, it also reveals key features that lead to major new insights and discoveries.

The causal effect of auxin on this boundary is less convincing, as it is hard to tell how a combined NPA and auxin treatment or a treatment of a combination of 3 auxin biosynthesis and response inhibitors affects the different aspects of the process. The effect of these treatments is shown after a relatively long time, and these effects on the localization of the examined markers could result from an effect on the domain patterning within the SAM rather than an effect on REV expression.

Firstly, we note that there was an error in the legend of Figure 10 that stated that the combined drug treatment on the inflorescence meristem was over 48hrs. Instead, the period was 18 hours as we now make clearer in the text.

Secondly, we want to clarify that we do not try to make the case that auxin acts directly on REV. Rather, our aim is to show that auxin directly or indirectly promotes HD–ZIPIII expression and prevents KAN expression in cells not already expressing KAN. To build on this proposal we have now added additional data (Figure 10—figure supplement 1) showing that in the inflorescence meristem the response of REV and KAN1 to 2,4–D is identical to combined NAA and NPA treatment but over a shorter period of 24hrs. We have also tracked cells showing that while auxin promotes REV expression it does this only in cells not already expressing KAN (Figure 10C–H). This leads to a firmer conclusion as to how polarity is established in organs and how this contrasts with what happens in between organs where auxin levels are low (see Discussion section).

Regarding the influence of the three drugs, we have added additional data to our paper showing that auxinole treatment alone causes a very mild version of the phenotype we see when adding all three drugs simultaneously (Figure 12A and B). Importantly, as auxinole competes for the auxin binding pocket in TIR1, we think it makes perfect sense to add the auxin synthesis inhibitors as well to obtain a more extreme loss of auxin activity. We also note that the phenotype we obtain is exactly what you would might expect, i.e. low auxin levels and a “pin” apex.

Reviewer #3:

Caggiano et al., address the question, how the dorso–ventral (DV) pattern of leaves is established. Using fluorescent protein reporters they show that DV patterning of leaves correlates with central–peripheral patterning of the shoot apical meristem (SAM). The experimental data related to this paragraph are shown in 5 figures (Figures 15), of which many details are not addressed in the text. The novelty of this part is limited, because similar data have been published earlier (e.g. McConnell et al., 2001; reviewed by Barton et al., 2009).

Although other studies have shown that the HD–ZIPIIIs are expressed centrally and that KAN1 and miR165/166 reporters are expressed peripherally, our aim here was to examine the precise spatio–temporal relationship between the patterning of these genes and organ initiation. In other words, where and when does organ initiation occur with respect to these patterns? None of the studies mentioned address this.

We find that organ initiation is centered precisely between the REV and KAN domains, which is a novel finding and suggests a pre–pattern that may function to control one aspect of organ positioning.

A similar effort to address the question of whether there is a pre–pattern or not for leaf dorsoventrality is the study by Tameshige et al., 2013 which addressed whether FIL is initially polarized with respect to leaf primordia or not. They concluded that FIL is initially not polar leading to the proposal that leaf primordia are not polarized to begin with (Tameshige et al., 2013). Our findings confirm their result for FIL but also show that FIL is the odd one out in this respect. We mention this just as an example to help illustrate that our findings with respect to leaf pre–patterning are novel. Further evidence is provided in or response to reviewer 2 above.

Subsequently, the authors provide evidence that HD–ZIPIII and KAN genes repress leaf primordium initiation in the vegetative SAM. However, the question whether this is a direct influence on organ initiation or an indirect influence on SAM maintenance was not answered, because SAM organization in the arrested plants was not analyzed. Similar results were obtained by Kerstetter et al., 2001.

Although we did not examine the expression of SAM markers, the central zone expresses HD–ZIPIII genes normally during its development. It is difficult to explain then how peripheral REV in the epidermis might shut down meristem growth. In fact, inflorescence apices ectopically expressing REV continue to grow vertically without a change in size as evidenced by the elongated “pin–like” phenotype. This also indicates that the SAM remains intact and functioning.

As for ectopic KAN1, we note in the paper that meristem arrest does occur. However, since we see the formation of a “pin” meristem when we activate KAN1 post–embryonically, this implies that organ formation stops prior to arrest of the meristem since stem tissue must have been produced while organs were not. We disagree with the reviewer that our results are similar to Kerstetter et al., (2001). Kerstetter et al., 2001 found that ectopic KAN1 gave rise to seedlings completely lacking a SAM (Kerstetter et al., 2001). This contrasts with our “pin” phenotype. However, this is not surprising given the contrasting methods to induce KAN1, i.e. our study does it post–embryonically and only in the epidermis.

In the next paragraph, the consequences of ectopic KAN expression in the central domain of the SAM are studied. Most seedlings initiate several abnormal organs before their growth is arrested. The distinct organ phenotypes observed can be explained by the orientation of boundaries within organ founder cells.

We are unsure how to respond to this as this comment corresponds to our stated conclusion.

Furthermore, the manuscript focuses on the interplay between auxin, HD–ZIPIII–expression and KAN–expression. Experiments using either NAA and NPA or a combination of auxinole, yuccasin and kyn led to the conclusion that high levels of auxin promote REV expression, whereas depletion of auxin promotes KAN expression. In this context, the treatment with a combination of three different drugs, which is necessary to see the described extension of KAN expression, may also have a different effect on gene expression than the one mediated through the observed depletion of auxin.

Please see our response to reviewer 2 with regards to the auxin and drug treatments.

The interplay of auxin, HD–ZIPIII expression and KAN–expression has also been addressed in previous studies (e.g. Izhaki and Bowman, 2007; Ilegems et al., 2010) with similar results.

Neither of these papers examined the response of REV and KAN1 to auxin. Nor did they examine the response of auxin to ectopic REV. Rather, they examined the ectopic production of organs in HD–ZIPIII and KAN mutants (Izhaki and Bowman, 2007) as well as the response of vascular tissues to loss and gain of KAN expression (Ilegems et al., 2010). Regarding the Izhaki paper, please see our response to reviewer 2 above where we discuss this paper in detail. For the second paper, there is no overlap in experiments or major conclusions since this study focused on vascular formation.

In the last paragraph, the authors show that wounding in inflorescence meristems (IMs) leads to a depletion of auxin around the wound followed by an accumulation of KAN, which could be eliminated by an NAA + NPA treatment. Again, part of these findings were already known (Heisler et al., 2010; Landrein et al., 2015). In addition, IMs may react different from vegetative meristems.

The major point of our ablations is not to demonstrate auxin depletion. It is to demonstrate a link between wound–induced auxin depletion and changes in REV and KAN1 expression. We start by citing those exact papers (Heisler et al., 2010; Landrein et al., 2015) as support for our hypothesis. We then repeat the experiment to monitor auxin depletion simply as a control to make sure the wounds are having the effect we expect in our particular context (i.e. without prior NPA treatment). Our major findings on the influence of wounds on REV and KAN expression and the role of auxin in this are completely novel.

Taken together, the manuscript presents a nice set of experiments supporting the view that DV organization of leaves is pre–patterned by the expression domains of REV and KAN in the SAM. However, previous studies revealed many similar results. Therefore, the study by Caggiano et al., presents only an incremental advance in our understanding.

Please see our response to reviewer 2. The fact that the issue of a Sussex signal is still a major controversy after 70 years demonstrates that our study is highly relevant. We argue that our results are also conclusive in terms of the pre–pattern and further provide an explanation for the original observed wounding responses. Our paper also contains many other major new findings as described in more detail above and now better highlighted in the new version of our manuscript.

Aiming for an explanation of the origin of DV patterning of leaves, it seems problematic to mix results from experiments using vegetative meristems and using inflorescence meristems, because a flower primordium is not a structure showing a DV pattern like a leaf primordium.

We apologize for the confusion. We should have used the term “floral bract” rather than primordium and we have now done so. All our results so far indicate that the establishment of polarity for the cryptic bract is almost identical to that of leaves consistent with published data indicating that the initial structure is indeed leaf–like rather than a floral meristem. The floral meristem develops some time later, after polarity for the bract has already been established and therefore we do not think it relevant. In any case, most of our results are presented for both the inflorescence and vegetative meristem contexts.

The wounding experiments offer an alternative explanation for the observations made by I. Sussex, but we think they do not rule out the existence of a signal from the main meristem to the developing leaf primordium.

At least for wounds, since KAN1 expression surrounds the wound regardless of its orientation with respect to the meristem, we maintain this is strong evidence against a meristem–oriented signal, as originally proposed by Sussex. The added evidence demonstrating the involvement of auxin provides a different and more parsimonious interpretation of Sussex’s experimental results. However, we agree that there are still important questions to be answered and write in our discussion that further experiments are required in tomato and other species.

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

Reviewer #1:

In this revision, the authors have addressed my concerns about data presentation and have clarified many of the confusing passages of the previous version.

In this current version, I see that they have generated a very high resolution description of organ boundary gene expression and have a plausible model for auxin (though like most auxin models, there are enough feedbacks that it’s hard to find a way to experimentally disprove it). To my mind a manipulation in which they misexpress boundary genes in such a way that polarity becomes locally reoriented (something like making fly wing clones that create artificial boundaries) would be the real test of their models. It seems to me that reagents to do these experiments should be available, and I would be very supportive of this manuscript if such data could be added.

We thank reviewer 1 for their positive comments. With regards to testing the model we propose using mosaics, we feel this is a major technical undertaking and would deserve to be presented in its own separate study. It is something we are working on for the future.

Reviewer #2:

In the revised manuscript, Caggiano et al., address some of the concerns raised by the reviewers and explain and clarify the new findings presented in this paper. As before, the imaging is impressive and convincing with respect to the pre–patterning of in the polarity in the SAM and the causative role of the boundary in organ initiation.

The authors better clarify the reason for the use of multiple drugs to assess the role of auxin in the patterning. However, I still think that the presented data on the effect of auxin does not support the interpretation on the role of auxin in specifying the spatial patterning in the SAM periphery. The time from application to effect is still long, and the effect on KAN and REV may be secondary to other effects. Combined with the relatively broad application, the results in my opinion do not support the proposed mechanism of auxin–mediated pre–patterning via REV activation and KAN repression.

We have now directly addressed these concerns with additional experiments. Firstly, we found that when auxin was applied to meristems on intact plants rather than plants in which the apices had been detached, the response of REV to exogenous auxin was much faster, starting within 6hrs (Figure 10—figure supplement 1) but becoming more obvious in less than 18hrs (Figure 10). Importantly, this exactly matches the wild type timing of events during regular development since REV expression visibly extends from the central zone into developing primordia a plastochron later than when PIN1 and DR5 mark these primordia (Heisler et al., 2005), i.e. around a 12hr delay.

As with the previous version of Figure 10, in the new Figure 10 we also show that auxin treatment inhibits the reestablishment of KAN1 in between primordia and the meristem (yellow and white arrows in Figure 10, I and J compared to control K and L).

Regarding the timing for KAN1, during regular development, KAN1 expression is well established by stage P2 (Figure 1—figure supplement 1), around 1 plastochron (approx. 12hrs) after auxin is depleted in boundary regions at stage P1 (the primordia are staged according to Heisler et al., 2005 based on the time of PIN1 polarity reversal). Hence it is to be expected that any induction of KAN1 in response to low auxin levels will take a similar length of time, which is entirely consistent with the results of our auxin depletion experiments (i.e. 18hrs shown in Figure 12).

Our data also suggest that once auxin depletion has occurred, cells are committed to expressing KAN1 even if auxin levels subsequently increase. This is indicated by the finding that exogenous auxin treatment cannot prevent the onset of KAN1 expression if auxin is applied at the P1 stage (when auxin has already been depleted in the surrounding cells – blue arrowheads in Figure 10I and J). This is in contrast to applying auxin at the i1 and i2 stages when PIN1 reversal and auxin depletion have not occurred yet (white and yellow arrowheads in Figure 10 I and J compared to controls K and L).

To further show that auxin acts locally to influence the expression domains of both REV and KAN1 in the periphery we applied auxin locally to pin1 mutant meristems and found that the response of REV and KAN1 is also local (Figure 10—figure supplement 2).

Lastly, we now also present data showing that neither auxin addition nor depletion obviously influence meristem patterning overall as evidenced by the central zone marker CLV3, indicating a more direct influence on primordium patterning (Figure 10—figure supplement 3 and Figure 12—figure supplement 3).

Again, we want to emphasize that the conclusions of our paper do not depend on REV and KAN being direct auxin targets and therefore that they must respond extremely rapidly (e.g. 30mins) to auxin. Accordingly, given the timing, we have now explicitly stated in our discussion that the regulation of these genes by auxin is likely to be indirect (Discussion section). At the same time however, our data do show that auxin modulates their expression patterns locally in the periphery and reasonably rapidly, and that this regulation is relevant to wound response.

Reviewer #4:

This manuscript addresses a long–standing question in plant biology, the acquisition of dorso–ventral (DV; or adaxial/abaxial) polarity in plant leaves. The authors use fluorescence live imaging combined with transgenic manipulation of gene–expression domains and pharmacological treatments to characterize the role of the boundary region between dorsal and ventral gene–expression domains in the meristem in leaf–polarity acquisition. They confirm and provide further detail on the pre–pattern in the shoot meristem involving centrally expressed HD–ZIPIII and peripherally expressed KANADI1 genes, and show that the boundary between the two domains, where neither gene is expressed strongly, is the site of auxin convergence driven by PIN1 and the highest auxin–signaling activity. Overexpressing either REV or KAN1 throughout the meristem epidermis abolishes organ initiation, consistent with their documented role in suppressing auxin activity. Together this suggests that the auxin maxima can only form in the boundary domain to trigger organ outgrowth there. Forcing KAN1 expression in the meristem centre and thus creating a new DV boundary there can lead to leaf radialization or even polarity reversal, indicating that the pre–pattern in the meristem can have an instructive function for leaf DV polarity. Having demonstrated that the pattern of REV and KAN1 expression determines the pattern of auxin responsiveness in the meristem, the authors then ask whether auxin in turn influences the expression pattern of these genes. Treating meristems with NAA and NPA causes an increase in the REV expression domain into the boundary region and can prevent the re–establishment of KAN1 expression in the periphery separating an outgrowing primordium from the meristem; conversely, inhibiting auxin signaling leads to an encroachment of KAN1 expression towards the meristem centre. Lastly, the authors show that the previously demonstrated auxin depletion around wound sites leads to an upregulation of KAN1 expression around the wound. This latter observation is offered as an explanation for the leaf radialization seen after separating a leaf primordium from the meristem in Sussex' classical experiments.

How DV polarity in leaves is determined is a question of fundamental importance and broad interest in plant development. As the ongoing interest in the quest for the proposed Sussex signal indicates, the work presented is clearly highly timely. The manuscript provides a mix of rather confirmatory and novel results. Important novel aspects are the demonstration that disturbing the DV expression pre–pattern in the meristem can lead to leaf–polarity defects; the finding that auxin in turn can modulate this DV expression pre–pattern; and the observation that peripheral/abaxial identity is induced surrounding wounds in the meristem.

I have the following comments:

1) The finding of a likely causal role of the pre–pattern on leaf polarity is an important advance. Also, the observation that auxin can modulate the DV expression pattern is interesting, yet its biological significance appears somewhat unclear. As the authors show in Figure 1 and in their model in Figure 9, the highest levels of auxin–signaling activity are found in the boundary region. However, it is precisely there that REV is not expressed. Thus, what does it mean that REV can be induced there by NPA+NAA treatment? And how does this observation relate to a potential auxin–mediated maintenance of REV expression in the central region of the meristem, where there appears to be very little auxin–signaling activity? Have the authors attempted to incorporate the auxin effect on REV expression into their model? Does this provide any information about the possible significance of this interaction?

We thank reviewer 4 for their positive comments.

Regarding the location of auxin signalling and REV expression, we note that during normal development REV expression extends into the boundary region that normally separates REV and KAN1 a plastochron after PIN1 upregulation. We apologize as this was not properly described in association with Figure 1 previously and so we have now added more text and a sup figure to fully detail this observation (Figure 1—figure supplement. 1). As can be seen from the new supplementary figure, the size of the gap shrinks initially as REV expression increases and extends but then becomes more apparent at later stages. Hence our treatments with NPA+NAA recapitulate the initial extension and increase in REV expression observed during the early stages of primordium development.

What does this regulation by auxin mean? The first point of significance is that we show that auxin acts to establish as well as maintain proper DV cell type patterning in new primordia. This is a new role for auxin that is distinct to its well–known function in organ initiation. This role is difficult to uncover because if you take auxin away you don’t get a primordium. Only by depleting auxin after organs have initiated and observing with time–lapse are we able to come to this conclusion. The second point of significance is in relation to the next section where we link the regulation of REV/KAN by auxin to wounding.

Regarding the central zone, an important point is that while REV/KAN expression depends on auxin in the meristem periphery, our experiments show that in the central zone, this is not the case. REV is maintained in the central zone no matter how much we deplete auxin. Thus, we conclude that the regulation of REV/KAN in the meristem central zone occurs via a distinct, auxin independent pathway. Hence only once cells move out of the central zone into the periphery, do high auxin levels become important for maintaining REV and repressing KAN. We have tried to make this conclusion clearer in the relevant Results section and Discussion section.

Regarding incorporating auxin regulation of REV/KAN in the modelling, this has begun. However, the modelling is still inconclusive at this point since it lacks certain features we feel are necessary to fully test our ideas, such as localized cell proliferation.

2) The alternative explanation offered for the effects of wounds separating a primordium from the meristem is novel, interesting and potentially of broad significance. However, its impact appears limited, as the KAN1 upregulation around the entire wound is currently only shown clearly for Arabidopsis inflorescence meristems (while the seedling meristems undergo more dramatic re–organization, making their behavior more difficult to interpret), where the action of a Sussex signal is hard to demonstrate due to the limited DV polarity of floral meristems. This means that the alternative explanation is not backed up by functional experiments. Ideally, the authors would directly test in a different system, such as the tomato apex, whether NAA+NPA treatment after appropriate wounding could prevent organ abaxialization.

While we agree that further experiments in other species would increase the impact of the current work, we also feel that this would represent a major long–term project in itself. Therefore, at this stage, we have rewritten our discussion of these results to further highlight these concerns (Discussion section).

https://doi.org/10.7554/eLife.27421.031

Article and author information

Author details

  1. Monica Pia Caggiano

    European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Xiulian Yu and Neha Bhatia
    Competing interests
    No competing interests declared
  2. Xiulian Yu

    European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Monica Pia Caggiano and Neha Bhatia
    Competing interests
    No competing interests declared
  3. Neha Bhatia

    1. European Molecular Biology Laboratory, Heidelberg, Germany
    2. School of Life and Environmental Sciences, University of Sydney, Sydney, Australia
    Contribution
    Conceptualization, Data curation, Formal analysis, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Monica Pia Caggiano and Xiulian Yu
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-2165-5183
  4. André Larsson

    Computational Biology and Biological Physics, Department of Astronomy and Theoretical Physics, Lund University, Lund, Sweden
    Contribution
    Software, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft
    Competing interests
    No competing interests declared
  5. Hasthi Ram

    European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  6. Carolyn K Ohno

    1. European Molecular Biology Laboratory, Heidelberg, Germany
    2. School of Life and Environmental Sciences, University of Sydney, Sydney, Australia
    Contribution
    Resources, Investigation, Methodology, Writing—original draft, Project administration
    Competing interests
    No competing interests declared
  7. Pia Sappl

    European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  8. Elliot M Meyerowitz

    Division of Biology and Biological Engineering, California Institute of Technology, Howard Hughes Medical Institute, Pasadena, United States
    Contribution
    Supervision, Funding acquisition, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon 0000-0003-4798-5153
  9. Henrik Jönsson

    1. Computational Biology and Biological Physics, Department of Astronomy and Theoretical Physics, Lund University, Lund, Sweden
    2. Sainsbury Laboratory, University of Cambridge, Cambridge, United Kingdom
    3. Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Conceptualization, Software, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon 0000-0003-2340-588X
  10. Marcus G Heisler

    1. European Molecular Biology Laboratory, Heidelberg, Germany
    2. School of Life and Environmental Sciences, University of Sydney, Sydney, Australia
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    marcus.heisler@sydney.edu.au
    Competing interests
    No competing interests declared
    ORCID icon 0000-0001-5644-8398

Funding

H2020 European Research Council (261081)

  • Marcus G Heisler

Marie Curie Actions (255089)

  • Pia Sappl

Gordon and Betty Moore Foundation (GBMF3406)

  • Elliot M Meyerowitz

Gatsby Charitable Foundation (GAT3395/PR4)

  • Henrik Jönsson

Swedish Research Council (VR2013-4632)

  • Henrik Jönsson

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

Acknowledgements

We thank M E Byrne for helpful feedback and ideas on the manuscript. We thank J Bowman for helpful discussions as well as for plasmids containing REVr and PHVr cDNAs. We thank D Weijers and C Y Liao for a plasmid containing pDR5v2::ntdTomato; K Nakajima for the T-DNA construct of miR166Ap::GFPER, Dr Atsushi Miyawaki from RIKEN Brain Science Institute for the VENUS fluorescent protein which was obtained through an MTA. We thank Professor Ken-ichiro Hayashi, Okayama University of Science for providing Auxinole. The EMM laboratory is supported by funds from the Howard Hughes Medical Institute and the Gordon and Betty Moore Foundation (through grant GBMF3406). The research leading to these results received funding from the Australian Research Council (MGH) and European Research Council under the European Union’s Seventh Framework Programme (FP/2007–2013)/ERC Grant Agreement n. 261081 (MGH), as well as the People Programme (Marie Curie Actions) under REA grant agreement n. 255089 (PS). The work was also supported by: the European Molecular Biology Laboratory (XY, MPC, CO, PS, NB, HR and MGH); the EMBL International PhD Programme (XY, NB and MPC); Gatsby Charitable Foundation (GAT3395/PR4) (HJ) and Swedish Research Council (VR2013-4632) (HJ). The authors declare no competing financial interests. Supplement contains additional data and movie. We declare there are no competing interests.

Reviewing Editor

  1. Dominique C Bergmann, Reviewing Editor, Stanford University/HHMI, United States

Publication history

  1. Received: April 3, 2017
  2. Accepted: September 11, 2017
  3. Accepted Manuscript published: September 12, 2017 (version 1)
  4. Accepted Manuscript updated: September 25, 2017 (version 2)
  5. Version of Record published: September 27, 2017 (version 3)

Copyright

© 2017, Caggiano 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

  • 2,299
    Page views
  • 455
    Downloads
  • 1
    Citations

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

Comments

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)

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

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

Further reading

    1. Cancer Biology
    2. Neuroscience
    Peter M LoCoco et al.
    Research Article Updated