1. Neuroscience
Download icon

Imaging neuropeptide release at synapses with a genetically engineered reporter

  1. Keke Ding
  2. Yifu Han
  3. Taylor W Seid
  4. Christopher Buser
  5. Tomomi Karigo
  6. Shishuo Zhang
  7. Dion K Dickman
  8. David J Anderson  Is a corresponding author
  1. California Institute of Technology, United States
  2. University of Southern California, United States
  3. Oak Crest Institute of Science, United States
  4. Howard Hughes Medical Institute, California Institute of Technology, United States
  5. Tianqiao and Chrissy Chen Institute for Neuroscience, California Institute of Technology, United States
Tools and Resources
  • Cited 0
  • Views 3,805
  • Annotations
Cite this article as: eLife 2019;8:e46421 doi: 10.7554/eLife.46421

Abstract

Research on neuropeptide function has advanced rapidly, yet there is still no spatio-temporally resolved method to measure the release of neuropeptides in vivo. Here we introduce Neuropeptide Release Reporters (NPRRs): novel genetically-encoded sensors with high temporal resolution and genetic specificity. Using the Drosophila larval neuromuscular junction (NMJ) as a model, we provide evidence that NPRRs recapitulate the trafficking and packaging of native neuropeptides, and report stimulation-evoked neuropeptide release events as real-time changes in fluorescence intensity, with sub-second temporal resolution.

https://doi.org/10.7554/eLife.46421.001

Introduction

Neuropeptides (NPs) exert an important but complex influence on neural function and behavior (Hökfelt et al., 2000; Insel and Young, 2000; Nässel and Winther, 2010; Bargmann and Marder, 2013). A major lacuna in the study of NPs is the lack of a method for imaging NP release in vivo, with subcellular spatial resolution and subsecond temporal resolution. Available techniques for measuring NP release include microdialysis (Kendrick, 1990), antibody-coated microprobes (Schaible et al., 1990) and GFP-tagged propeptides visualized either by standard fluorescence microscopy (van den Pol, 2012), or by TIRF imaging of cultured neurons (Xia et al., 2009). In Drosophila, a fusion between rat Atrial Natriuretic Peptide/Factor (ANP/F) and GFP was used to investigate neuropeptide trafficking at the fly neuromuscular junction (NMJ) (Rao et al., 2001). Release was measured indirectly, as a decrease in ANP-GFP fluorescence intensity at nerve terminals reporting residual unreleased peptide, on a time-scale of seconds (Wong et al., 2015). None of these methods combines NP specificity, genetically addressable cell type-specificity, high temporal resolution and applicability to in vivo preparations (Supplementary file 1). A major challenge is to develop a tool that encompasses all these features for direct, robust measurement of NP release in vivo.

Results

Neuropeptides are synthesized as precursors, sorted into dense core vesicles (DCVs), post-translationally modified and cleaved into active forms prior to release (Taghert and Veenstra, 2003). We reasoned that an optimal in vivo real-time NP release reporter should include (1) a reporter domain that reflects the physico-chemical contrast between the intravesicular milieu and the extracellular space (Figure 1—figure supplement 1A); and (2) a sorting domain that ensures its selective trafficking into DCVs (Figure 1—figure supplement 1b). The NP precursor may function as the sorting domain, suggested by studies of DCV fusion using pIAPP-EGFP (Barg et al., 2002) and NPY-pHluorin (Zhu et al., 2007) in cultured neurons, or ANP-GFP in Drosophila (Rao et al., 2001). We therefore developed a pipeline to screen various transgenes comprising NP precursors fused at different sites to fluorescent reporters, in adult flies (Figure 1—figure supplement 1B–C). A total of 54 constructs were tested. We found that optimal trafficking was achieved by substituting the reporter for the NP precursor C-terminal domain that follows the final peptide (Figure 1—figure supplement 1B). In order to maintain covalent linkage with the reporter domain, we removed the dibasic cleavage site C-terminal to the final peptide.

The DCV lumen has lower pH and free calcium (pH = 5.5–6.75, [Ca2+]~30 µM) compared to the extracellular space (pH = 7.3, [Ca2+]~2 mM) (Mitchell et al., 2001; Sturman et al., 2006). These differences prompted us to test validated sorting domains in a functional ex vivo screen using either pH-sensitive fluorescent proteins (Miesenböck et al., 1998) or genetically-encoded calcium indicators (GECIs) (Tian et al., 2012; Lin and Schnitzer, 2016) (Figure 1—figure supplement 1A–D). Reporters based on pHluorins (Miesenböck et al., 1998) did not perform well in our hands, therefore we focused on GCaMP6s (Chen et al., 2013). The calcium sensitivity threshold of GCaMP6s is below the calcium concentration in both DCVs and the extracellular space. However, GCaMP6s fluorescence is quenched in the acidic DCV lumen (Barykina et al., 2016), enabling it to function as a dual calcium/pH indicator (Figure 1A). These key properties should boost the contrast between GCaMP6s fluorescence in unreleased vs. released DCVs, potentially allowing us to trace NP release at the cellular level in vivo.

Figure 1 with 4 supplements see all
Design and Synaptic Localization of an NPRR.

(A) Schematic illustrating the principle of NPRRs (Neuropeptide Release Reporters). NPRR molecules in the DCV lumen (low pH/low calcium, left) exhibit increased fluorescence when released by fusion into the extracellular space (neutral pH/high calcium, right). NPRR fluorescent signal is expected to decay following diffusion into the synaptic cleft. New NPRR-containing DCVs are produced by synthesis and transport from the soma, not by recycling. NP: Neuropeptide. DCV: Dense Core Vesicle. SV: Synaptic Vesicle. (B) Distinct motor neuron subtypes at the Drosophila NMJ (muscle 12/13) have different proportions of DCVs vs. SVs. The GAL4 driver R57C10-Gal4 (nsyb-GAL4) labels all subtypes, while R20C11-GAL4 selectively labels only Type III neurons, which lack SVs (‘Type III-GAL4’). Light gray circles, black lines and dark gray shading represent boutons, inter-bouton intervals and subsynaptic reticulum respectively. The studies in this paper focus on Type Ib neurons and Type III neurons (in red rectangles). (C) Triple immunolabeling for GFP (green), Bursicon (blue) and vGluT (red), in flies containing nsyb-GAL4 driving UAS-GCaMP6s (upper), or NPRRANP (lower). Type Ib and Type III boutons are indicated. Scale bar, 5 µm. Inset image (NPRRANP, a-GFP channel) shows details of puncta distribution of NPRRANP in Type Ib neuron. Scale bar, 2 µm. (D) TEM images of boutons immunolabeled with anti-GFP (5 nm gold particle-conjugated) to detect nsyb>NPRRANP-GFP, which has an identical structure to NPRRANP, but is a GFP rather than GCaMP6s fusion to improve antigenicity (see Figure 1—figure supplement 4). Note strong labeling in DCVs (arrows) and the neuronal plasma membrane (arrowheads). Scale bar, 200 nm. Lower panel shows representative images of labeled DCVs. Scale bar,100 nm. (E) Quantification for TEM images in (D).

https://doi.org/10.7554/eLife.46421.002

We sought to test several NP precursor-GCaMP6s fusion proteins, called NPRRs (NeuroPeptide Release Reporters; unless otherwise indicated all NPRRs refer to fusions with GCaMP6s), in an intact preparation using electrical stimulation to evoke release. Initially for proof-of-principle experiments, we used the Drosophila larval NMJ to test NPRRANP, a GCaMP6s fusion with rat ANP (Burke et al., 1997). NMJ terminals are large, individually identifiable, and easy to image and record. In particular, boutons on muscle 12/13 are diverse -- Type Ib and Type Is boutons contain mostly synaptic vesicles and few DCVs, while Type III boutons contain an abundance of DCVs but no synaptic vesicles (Menon et al., 2013); moreover, Type III-specific GAL4 drivers are available (Koon and Budnik, 2012) (Figure 1B).

Expression of NPRRANP pan-neuronally (under the control of nsyb-GAL4) followed by double immuno-staining for ANP and GCaMP (anti-GFP) indicated that the sorting domain and the reporter domains showed a similar localization in Type III neurons (Figure 1—figure supplement 2). Moreover, the distribution of NPRRANP overlapped that of Bursicon (Figure 1—figure supplement 3D), an NP that is endogenously expressed in Type III neurons (Loveall and Deitcher, 2010). Both GCaMP and Bursicon immunoreactivity were strongest within boutons, consistent with the known subcellular localization of DCVs (Gorczyca and Budnik, 2006).

Glutamate is the only known canonical neurotransmitter used at the larval NMJ (Menon et al., 2013). This allowed visualization of the subcellular localization of small synaptic vesicles (SV) by immuno-staining for vGluT, a vesicular glutamate transporter (Fremeau et al., 2001; Kempf et al., 2013). In Type Ib neurons (which contain relatively few DCVs relative to SVs [Menon et al., 2013]), vGluT staining was observed as patches with a dim center, which may reflect clustered SVs, while NPRRANP immunoreactivity was seen in dispersed, non-overlapping punctae (Figure 1C, α-GFP, inset). In Type III neurons, NPRRs were strongly expressed but no vGluT immunoreactivity was detected (Figure 1C). The subcellular distribution of this NPRR in larval NMJ neurons, therefore, is similar to that of other DCV-targeted markers previously used in this system (Rao et al., 2001; Shakiryanova et al., 2006), and appears to reflect exclusion from SVs.

The diffraction limit of light microscopy precluded definitive co-localization of NPRRs in DCVs. Therefore, we employed Immuno-Electron microscopy (Immuno-EM) to investigate the subcellular localization of NPRRs at the nanometer scale. To maximize antigenicity for Immuno-EM, we generated constructs that replaced GCaMP6s with GFP (NPRRANP-GFP;). NPRRANP-GFP showed dense labeling in association with DCVs (Figure 1D, arrows), where the average number of gold particles/µm2 was substantially and significantly higher than in neighboring bouton cytoplasm (DCV/Bouton ~ 14.26) (Figure 1E, Supplementary file 2). Taken together, these data indicate that NPRRANP-GFP is localized to DCVs. By extension, they suggest that NPRRANP-GCaMP6s (which has an identical structure to NPRRANP-GFP except for the modifications that confer calcium sensitivity) is similarly packaged in DCVs. While these two reporters show indistinguishable distributions by immunofluorescence (Figure 1—figure supplement 4), we cannot formally exclude that the substitution of GCaMP for GFP may subtly alter subcellular localization of the NPRR in a manner undetectable by light microscopy.

To measure the release of NPRRs from DCVs, we next expressed NPRRANP in Type III neurons using a specific GAL4 driver for these cells (Koon and Budnik, 2012) (Figure 2E and Figure 1—figure supplement 3D). We delivered 4 trials of 70 Hz electrical stimulation to the nerve bundle, a frequency reported to trigger NP release as measured by ANF-GFP fluorescence decrease (Rao et al., 2001; Shakiryanova et al., 2006), and used an extracellular calcium concentration that promotes full fusion mode (Alés et al., 1999). This stimulation paradigm produced a relative increase in NPRRANP fluorescence intensity (ΔF/F), whose peak magnitude increased across successive trials (Figure 2A, red bars and 2D; Video 1; Figure 2—figure supplement 1, A1 vs. A7). Responses in each trial showed a tri-phasic temporal pattern: (1) In the ‘rising’ phase, NPRRANP ∆F/F peaked 0.5–5 secs after stimulation onset, in contrast to the virtually instantaneous peak seen in positive control specimens expressing conventional GCaMP6s in Type III neurons (Figure 2A–B). The NPRRANP latency to peak was similar to the reported DCV fusion latency following depolarization in hippocampal neurons (Xia et al., 2009). This delay is thought to reflect the kinetic difference between calcium influx and DCV exocytosis due to the loose association between DCVs and calcium channels (Xia et al., 2009). (2) In the ‘falling’ phase, NPRRANP ∆F/F began to decline 1–5 s before the termination of each stimulation trial, presumably reflecting depletion of the available pool of releasable vesicles. In contrast, GCaMP6s fluorescence did not return to baseline until after stimulation offset (Figure 2A–B). (3) Finally, unlike GCaMP6s, NPRRANP exhibited an ‘undershoot’ (∆F/F below baseline) during the post-stimulation intervals, followed by a ‘recovering’ phase (Figure 2A; Figures 2C,I14). This undershoot may reflect dilution of released fluorescent NPRR molecules by diffusion into the synaptic cleft (van den Pol, 2012), while recovery may reflect DCV replenishment in the boutons from vesicles proximal to the imaged release site.

Figure 2 with 3 supplements see all
NPRR specifically reports neuropeptide release.

(A) Trace from a representative experiment showing changes in NPRRANP fluorescence intensity (∆F/F) in Type III motor neurons at the larval NMJ evoked by electrical stimulation. BG: background. S1-S4: Stimulation trials 1–4. I1-I4: Inter-stimulation Intervals (ISIs) 1–4. Green line: ∆F/F averaged across all boutons in the field of view. Gray shading: s.e.m envelope. Red bar: electrical stimulation trials (70 Hz). The three typical phases of the response are indicated in S4. The peak height of the response on the first trial is characteristically lower (see also (D)), and may reflect competition with unlabeled DCVs in the readily releasable pool. (B) ΔF/F traces in control flies expressing cytoplasmic GCaMP6s in Type III neurons. (C) Integrated NPRRANP ΔF/F values during trials S1-4 and intervals I1-4. A.U.: arbitrary units. n = 8. ***, p<0.001. (D) Average NPRRANP ΔF/F peak heights for trials S1-4. n = 8. *, p<0.05. Plotted values in (C–D) are mean ± s.e.m. (E1–E2) Representative selection of ROIs (yellow). Details see Materials and methods. Scale bar, 5 µm. (F) NPRRANP ΔF/F response are abolished in Type III GAL4>UAS NPRRANP flies bearing UAS-TNT (F1) but not UAS-TNTimp (F2). (G) Average peak heights of NPRRANP ΔF/F in combined stimulation trials (S1-4) from (F). ****, p<0.0001. (H) Average ‘undershoot’, defined as the integrated ΔF/F during ISIs I1-4 (see (C)). In (C–D) and (G–H).

https://doi.org/10.7554/eLife.46421.009
Video 1
Activation of NPRRANPin situ.

Representative video recording of NPRRANP-expressing Type III neurons at the larval NMJ. In the preparation, the Type III neuron was stimulated as described in Figure 2A.

https://doi.org/10.7554/eLife.46421.017

Because NPRRANP fluorescence was preferentially accumulated within boutons, we asked whether these regions contributed to ∆F/F peaks more significantly than the inter-bouton intervals (IBIs). To do this, we partitioned the processes into boutons and IBI fields (Figure 2—figure supplement 2A), and compared the ∆F/F in these regions during stimulation trials. The time-averaged ratio of bouton/IBI ΔF/F (see Materials and Methods) was significantly higher for NPRRANP than for GCaMP6s, particularly during later stimulation trials (Figure 2—figure supplement 2B, green bars, S2-4). This contrast indicates that NPRRANP signals are preferentially observed in boutons, where DCVs are located, and do not reflect differences in cytoplasmic free Ca2+ levels between these regions as detected by GCaMP6s.

To test definitively if NPRRANP ∆F/F signals are dependent upon NP release, we blocked vesicle fusion at terminals of Type III neurons using expression of tetanus toxin light chain (TNT) (Sweeney et al., 1995), a protease that cleaves n-synaptobrevin, a v-snare required for DCV fusion (Figure 2—figure supplement 3) (Xu et al., 1998). As a control, we used impotent TNT (TNTimp), a reduced activity variant (Sweeney et al., 1995). TNT expression completely abolished stimulation-induced ∆F/F increases from NPRRANP, while TNTimp did not (Figure 2F). Further analysis revealed that both the ∆F/F peaks and inter-stimulation undershoots were diminished by TNT (Figure 2G–H). In contrast, neither TNT nor TNTimp affected the kinetics of GCaMP6s signals in Type III neurons (Figure 2—figure supplement 2C), which report cytosolic Ca2+ influx. Taken together, these data support the idea that NPRRANP signals specifically reflect DCV release.

ANP is a rat NP that lacks a Drosophila homolog (Rao et al., 2001). To determine whether our method could be applied to detect the release of a specific, endogenous fly NP, we tested NPRRdTK, one of 6 different reporter variants we initially generated from the Drosophila neuropeptide precursor, DTK (Figure 1—figure supplement 1B). In contrast to ANP which encodes a single peptide, DTK yields multiple NP derivatives (Winther et al., 2003). Light microscopy (Figure 3A) and Immuno-EM (Figure 3B, arrows) confirmed that NPRRdTK, like NPRRANP, was localized to DCVs (DCV/bouton ~ 22.19, Figure 3C). Using the Type III-specific GAL4 driver to express NPRRdTK and the same stimulation protocol as used for NPRRANP, the basic tri-phasic response profile was also observed (Figure 3D). However, peak heights and baseline fluorescence fell progressively with successive stimulation trials (Figure 3E), in contrast to NPRRANP where the first peak and undershoot were lower (Figure 2C–D). The reason for this difference is currently unclear.

Application of the NPRR approach to a Drosophila neuropeptide.

(A) Triple immunolabeling for GFP (green), Bursicon (blue) and vGluT (red) in Type III-GAL4> UAS NPRRdTK flies. Scale bar, 5 µm. (B) TEM images of boutons immunolabeled against GFP (5 nm gold) in nsyb-GAL4>UAS NPRRdTK-GFP flies. Note strong labeling in DCVs (arrows) and bouton plasma membrane (arrowheads). Scale bar, 200 nm. Lower panel shows representative images of labeled DCVs. Scale bar,100 nm. (C) Quantification of TEM images in (B). (D) NPRRdTK ΔF/F curve; stimulation conditions as in Figure 2A. (E) Average NPRRdTK ΔF/F peak height above pre-stimulation baseline (corrected; see Materials and methods) for stimulation trials S1-4. n = 6. **, p<0.01.

https://doi.org/10.7554/eLife.46421.013

We next investigated the relationship between NPRR signal and stimulation intensity, by delivering to the Type III neurons a series of low to high frequency electrical stimuli (1–70 Hz; Levitan et al., 2007) while imaging the nerve terminals. For direct comparison of NPRR responses across different preparations, we applied a posteriori normalization of fluorescent peaks in each trial to the highest response obtained among all trials. For both NPRRANP and NPRRdTK (Figure 4A–B), the peak responses showed a positive correlation with stimulation frequency, analogous to that observed using cytosolic GCaMP6s (Figure 4C). In Type III neurons, the responses of both NPRRs to stimulation frequencies < 30 Hz (1,5,10,20 Hz) were not statistically significant from zero. NPRRANP showed a higher sensitivity to high stimulation frequencies (30 Hz: 18.14%, 50 Hz: 82.40% Normalized peak ∆F/F), while NPRRdTK showed a higher stimulation threshold and lower sensitivity (30 Hz: 3.57%, 50 Hz: 24.67% Normalized peak ∆F/F).

Figure 4 with 1 supplement see all
NPRR reveals distinct cell-type specific peptide release properties.

For each preparation, a series of stimulation trials were delivered at frequencies from 1 Hz to 70 Hz, as indicated. In-stimulation response peaks were normalized to 70 Hz. The normalized peaks of NPRRs or calcium responses (measured with cytosolic GCaMP6s) were pooled and plotted for both Type III (A-C) and Type Ib (D-F) neurons. Responses were compared to zero. n = 6–12. n.s., not significant. *, p<0.05. **, p<0.01. ***, p<0.001. ****, p<0.0001.

https://doi.org/10.7554/eLife.46421.015

We next investigated whether the relatively high stimulation frequency required to observe significant responses with NPRRs was a function of the reporters, or rather of the cell class in which they were tested. To do this, we expressed both NPRRs in Type Ib neurons, a class of motor neurons that contains both SVs and DCVs (Figure 1B, Figure 4D–F), and performed stimulation frequency titration experiments. Strikingly, in Type Ib neurons, significant increases in ∆F/F could be observed at frequencies as low as 10 Hz (Figure 4D,E; NPRRANP @ 20 Hz: 12.50%, NPRRdTK @ 20 Hz: 17.67% normalized peak ∆F/F). The reason for the difference in NPRR threshold between Type III and Type Ib neurons is unknown, but parallels their difference in GCaMP6s response to electrical stimulation (Figure 4C vs. Figure 4F).

Notably, although NPRRANP and NPRRdTK presented distinct response profiles in Type III neurons, their performance in Type Ib neurons was more similar (Figure 4A vs. Figure 4B; cf. Figure 4D vs. Figure 4E). In summary, the differences in performance we observed between the two NPRRs appeared to be specific to Type III neurons, and were minor in comparison to the differences in performance of both reporters between the two cell classes. The reason for the differences between NPRRANP and NPRRdTK sensitivity and kinetics in Type III neurons is unknown but may reflect differences in how well these reporters compete with the high levels of endogenous neuropeptide (Bursicon) for packaging, transport or release.

Discussion

Here we present proof-of-principle for a method to detect the release of different neuropeptides in intact neural tissue, with subcellular spatial and sub-second temporal resolution. By exploiting the fluorescent change of GCaMP in response to a shift in pH and [Ca2+], we visualized the release of neuropeptides by capturing the difference between the intravesicular and extracellular microenvironment. NPRR responses exhibited triphasic kinetics, including rising, falling and recovering phases. In the falling phase, a post-stimulus ‘undershoot’, was observed in which the fluorescent intensity fell below pre-stimulation baseline. This undershoot presumably reflect the slow kinetics of DCV replenishment relative to release.

The molecular mechanisms of NP release are incompletely understood (Xu and Xu, 2008). It is possible that individual DCVs only unload part of their cargo during stimulation, in which case many DCVs that underwent fusion may still contain unreleased NPRR molecules following a stimulus pulse. Although we are convinced that NPRR signals do indeed reflect NP release, due to the presence of the recovering phase, we cannot formally exclude that unreleased NPRRs may contribute to the signal change due to their experience of intravesicular [Ca2+]/pH changes that occur during stimulation. To resolve this issue in the future, an ideal experiment would be to co-express an NPRR together with a [Ca2+]/pH-invariant NP-reporter fusion. Multiple attempts to generate such fusions with RFP were unsuccessful, due to cryptic proteolytic cleavage sites in the protein which presumably result in degradation by DCV proteases during packaging.

To test if NPRRANP ∆F/F signals are dependent on NP release, we expressed the light chain of tetanus toxin (TNT), a reagent shown to effectively block NP release in many (McNabb and Truman, 2008; Hentze et al., 2015; Zandawala et al., 2018), if not all (Umezaki et al., 2011), systems. We observed a striking difference in NPRR kinetics in flies co-expressing TNT vs. its proteolytically inactive ‘impotent’ control form TNTimp (Figure 2F). The strong reduction of NPRR signals by TNT-mediated n-syb cleavage is consistent with the idea that these signals reflect the release of NPRRs from DCVs.

We have tested the generalizability of the principles used to generate NPRRs by (1) constructing a surrogate NP reporter NPRRANP as well as a multi-peptide-producing endogenous Drosophila NP reporter NPRRdTK (Figures 23); (2) characterized NPRR signals in response to varying intensities of electrical stimulation; and (3) recorded NPRR signals in two different classes of NMJ motor neurons containing DCVs with or without SVs, respectively (Figure 4). These experiments revealed, to our surprise, that NPRR responses exhibit cell-type specific characteristics (Figure 4). As NPRRs are applied to other neuropeptides and cell types, a systematic characterization of neuropeptide release properties in different peptidergic neurons should become possible, furthering our understanding of neuropeptide biology.

The method described here can, in principle, be extended to an in vivo setting. This would open the possibility of addressing several important unresolved issues in the study of NP function in vivo. These include the ‘which’ problem (which neuron(s) release(s) NPs under particular behavioral conditions?); the ‘when’ problem (when do these neurons release NPs relative to a particular behavior or physiological event?); the ‘where’ problem (are NPs released from axons, dendrites or both?); and the ‘how’ problem (how is NP release regulated?). The application of NPRRs to measuring NP release dynamics in awake, freely behaving animals may yield answers to these important long-standing questions.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
information
Genetic reagent (D. melanogaster)UAS-NPRRANP (attp2)this paperSee Materials and methods, subsection Construction of transgenic animals.
Genetic reagent (D. melanogaster)UAS-NPRRdTK (attp2)this paperSame as above.
Genetic reagent (D. melanogaster)UAS-TNTimpBloomington Drosophila Stock CenterBDSC:28840; FLYB:FBti0038575; RRID:BDSC_28840Flybase symbol: w[*]; P{w[+mC]=UAS TeTxLC.(-)V}A2
Genetic reagent (D. melanogaster)UAS-TNTBloomington Drosophila Stock CenterBDSC:28838; FLYB:FBti0038527; RRID:BDSC_28838Flybase symbol: w[*]; P{w[+mC]=UAS TeTxLC.tnt}G2
Genetic reagent (D. melanogaster)w; +; UAS-GCaMP6s (su(Hw)attp1)Hoopfer et al., 2015
Antibodyanti-GFP (chicken polyclonal)AveslabAveslab: GFP-1020; RRID:AB_2307313(1:250:Immuno-EM, 1:1000: IHC)
Antibodyanti-ANP (rabbit polyclonal)abcamabcam #14348(1:500)

Fly strains

All experimental flies were reared on a 12/12 hr day-night cycle at 25°C. Standard chromosomal balancers and genetic strategies were used for all crosses and for maintaining mutant lines. Detailed genotypes used are summarized in Supplementary file 3. The following strains were obtained from Bloomington Stock Center (Indiana University): R20C11-Gal4 (#48887), R57C10-Gal4 (#39171), UAS-mCD8::GFP (#32185), UAS-TNT (#28838), UAS-TNTimp (#28840). UAS-opGCaMP6s was made by Barret Pfeiffer (Gerald Rubin’s lab, Janelia Farm) (Hoopfer et al., 2015).

Construction of transgenic animals

All PCR reactions were performed using PrimeSTAR HS DNA polymerase (Takara #R045Q). All constructs were verified via DNA sequencing (Laragen).

To construct UAS-NPRRANP, Drosophila codon-optimized ANP and GCaMP6s were synthesized using gBlocks service (Integrated DNA Technologies), and subcloned into pJFRC7 vector (from Addgene #26220) (Pfeiffer et al., 2010) using Gibson cloning. UAS-dTK-NPRR is built in a similar way except the dTK fragment was cloned from the Drosophila brain cDNA. NPRRdTK-GFP and NPRRANP-GFP were built similarly except Drosophila codon-optimized GFP was used for the subcloning. All the vectors were injected and integrated into attP2 or attp40 sites (Bestgene Inc; see Supplementary file 3 for attP sites used for each genotype employed).

Expression screening of NPRR candidates

Adult fly brains were dissected in chilled PBS and fixed in 4% formaldehyde for 55 min at room temperature. After three 10 min rinses with PBS, the brains were cleared with Vectashield (#1000, Vectorlabs), mounted, and used for native fluorescence measurements. We trace the NPF neuron somata and arborization as ROIs. We selected regions next to NPF neurons and measured its fluorescent intensity as a reference, which represents background autofluorescence. Candidates whose fluorescence reached at least 2-fold higher than reference were selected for functional screening.

Functional screening of NPRR candidates

For the baseline fluorescence measurement, we crossed NPF-Gal4 to the candidate lines and generated NPF-Gal4>NPRRx (x = candidate label) flies for tests. The dissected adult fly brains were mounted on a petri dish and immersed in Drosophila imaging saline (108 mM NaCl, 5 mM KCl, 2 mM CaCl2, 8.2 mM MgCl2, 4 mM NaHCO3, 1 mM NaH2PO4, 5 mM trehalose, 10 mM sucrose, 5 mM HEPES, pH 7.5). To deliver high potassium challenge, High-K imaging saline was perfused (43 mM NaCl, 70 mM KCl, 2 mM CaCl2, 8.2 mM MgCl2, 4 mM NaHCO3, 1 mM NaH2PO4, 5 mM trehalose, 10 mM sucrose, 5 mM HEPES, pH 7.5). Live imaging series were acquired using a Fluoview FV3000 Confocal laser scanning biological microscope (Olympus) with a 40×, 0.8 N.A. (Numerical Aperture) water immersion objective (Olympus). Candidates whose post-stimulation fluorescence reached at least 2-fold of baseline fluorescence (measured as averaged pre-stimulation fluorescence) were selected for in vivo tests at NMJ. For each candidate line, at least three brains were tested and fold-change of each was averaged.

Immunohistochemistry

Larval dissection was performed in chilled HL3 solution (70 mM NaCl, 5 mM KCl, 20 mM MgCl2, 10 mM NaHCO3, 115 mM sucrose, 5 mM trehalose, 5 mM HEPES and 1.5 mM CaCl2, pH 7.2). Dissected tissues were fixed in 4% formaldehyde or Bouin’s solution for 30 min at room temperature. After three 15 min rinses with PBS, tissues were incubated with primary antibodies overnight at 4°C. Following three 15 min rinses with PBS, tissues were incubated with secondary antibody for 2 hr at room temperature. Following three 15 min rinses, tissues were cleared with Vectashield (#1000, Vectorlabs) and mounted. Confocal serial optical sections were acquired using a Fluoview FV3000 Confocal laser scanning biological microscope (Olympus) with a 60×, 1.30 N.A. silicone oil objective (Olympus). All image processing and analyses were done using ImageJ (National Institute of Health).

The following primary antibodies were used: Chicken anti-GFP (1:250-1:1000, Aveslab #1020), Rabbit anti-ANP (1:500, abcam #14348), Guinea pig anti-vGluT (Goel and Dickman, 2018) (1:1500), Rabbit anti-syt1 (Littleton et al., 1993) (1:500) and Rabbit anti-Bursicon (1:2000, a gift from Dr. Benjamin White).

The following secondary antibodies were used: Alexa Fluor 488 Goat anti-Chicken IgY (#A11039, Invitrogen), Alexa Fluor 488 Goat anti-Rabbit IgG (#A11008, Invitrogen), Alexa Fluor 568 Goat anti-Rabbit IgG(H + L) (#A11011, Invitrogen), Alexa Fluor 633 Goat anti-Rabbit IgG(H + L) (#A21070, Invitrogen), Alexa Fluor 488 Goat anti-Guinea Pig IgG(H + L) (#A11073, Invitrogen), Alexa Fluor 568 Goat anti-Guinea Pig IgG(H + L) (#A11075, Invitrogen), Alexa Fluor 568 Goat anti-Mouse IgG(H + L) (#A11004, Invitrogen) and Alexa Fluor 633 Goat anti-Mouse IgG(H + L) (#A21050, Invitrogen).

Electron microscopy

Drosophila tissues were fixed in 4% formaldehyde in PBS and stored at 4°C until preparation by high-pressure freezing (HPF) and freeze-substitution (FS) (Buser and Walther, 2008; Buser and Drubin, 2013). Tissues were cryoprotected in 2.3 M sucrose for 45 min, transferred to 200 µm deep planchettes and high-pressure frozen in an EMPact2 with RTS (Leica, Vienna, Austria). FS was carried out in an AFS2 (Leica, Vienna, Austria) in methanol containing 5% water, 0.05% glutaraldehyde and 0.1% uranyl acetate (−90°C, 3 hr; −90 to −80°C, 10 hr; −80°C, 4 hr; −80°C to 4°C, 24 hr). Samples were washed once in methanol containing 5% water, infiltrated with hard grade LR White (Electron Microscopy Sciences, Hatfield, PA, USA) at 4°C ([LR White]: [methanol containing 5% water] 1:1, 24 hr; 100% LR White, 3 × 24 hr) and polymerized in a fresh change of LR White using a Pelco BioWave (Ted Pella, Inc, Redding, CA, USA) set to 750 W, 95°C for 45 min.

60 nm thin sections (UCT ultramicrotome, Leica, Vienna, Austria) were picked up on formvar-coated 50 mesh copper grids. The sections were blocked for 3 min in blocking buffer (PBS with 0.5% bovine serum albumin, which was used for all antibody dilutions), incubated in anti-GFP antibody (1:500, Aveslab #1020) for 5 min, washed 3 times in blocking buffer, incubated in rabbit anti chicken antibody (1:50, MP Biomedicals #55302) for 5 min, washed 3 times on blocking buffer, incubated on protein A - 5 nm gold (1:50, Utrecht, Netherlands), and washed 3 times in PBS and 3 times in distilled water. The sections were stained in uranyl acetate or uranyl acetate and Reynolds lead citrate depending on the desired contrast and imaged at 80 kV in a Zeiss EM10C (Zeiss, Oberkochen, Germany) using a CCD camera (Gatan, Pleasanton, CA, USA).

Labeling density was estimated using stereological methods (Griffiths and Hoppeler, 1986). Cross-sections through boutons were recorded and the following parameters were measured: total image area, total number of gold particles, number of visible dense core vesicles (DCV), number of gold particles within a 50 nm radius of the DCV center, bouton area (grid intersection estimate), gold within the bouton cytoplasm, gold within 20 nm of the bouton plasma membrane, gold outside of the bouton (mainly sER). Background labeling was estimated using internal controls (labeling on blank resin and on muscle fibers) and a biological control (non-GFP expressing genotype). Occasional obvious, large gold aggregates were disregarded. Background was consistently below 0.6 gold/µm2 in independently repeated labeling experiments.

Electrical stimulation

The dissection of third-instar larvae was performed in zero-calcium HL3 saline. The CNS was removed to avoid spontaneous motor neuron activity. To minimize muscle contraction induced by electrical stimulation of motor neurons, the larval body walls were slightly stretched and incubated in HL3 saline supplemented with 10 mM glutamate for 5 mins after dissection to desensitize postsynaptic glutamate receptors. Samples were then shifted to HL3 saline containing 1 mM glutamate and 1.5 mM Ca2+. Motor nerves were sucked into a glass micropipette with a stimulation electrode. In Figure 2 and Figure 3, to induce maximum dense core vesicle release at type III motor neuron terminals, four repetitive bursts (70 Hz stimulation for 18–20 s with pulse width of 1 ms) with intervals of 40–42 s were programmed and triggered with a Master-9 stimulator (A.M.P.I., Israel) connected to an iso-flex pulse stimulator (A.M.P.I., Israel). The stimulation intensity was tested and set to double the intensity required to induce muscle contraction by a single pulse stimulation.

In Figure 4, stimulation trials were delivered with the same duration, but with a series of frequencies spanning 1 Hz to 70 Hz.

Calcium imaging

A Nikon A1R confocal microscope with resonant scanner and NIS Element software were used to acquire live Ca2+ imaging on third instar larvae, bathed with 1 mM glutamate added in 1.5 mM Ca2+ HL3 saline. Type III motor neuron terminals in abdominal segments from A2 to A5 were imaged using a 60x APO 1.4 N.A. water immersion objective with 488 nm excitation laser. A 5 min period was used for time-lapse imaging at a resonance frequency of 1 fps (512 × 512 pixels or 1024 × 1024 pixels), with z-stacks (step length varying from 1 to 1.5 μm) covering the depth of entire type III motor neuron terminals. The repetitive electrical stimulation of 70 Hz was delivered during the imaging session. Samples with severe muscle contractions were abandoned due to imaging difficulties. Maximum intensity projection (MIP) and image registration were conducted using Image J. Plugins including Image Stabilizer (K. Li, CMU) and Template Matching (Q. Tseng) were used for compensating drifting and correcting movement induced by electrical stimulations. ROIs were manually selected by tracing the outer edge of each neuron based on the baseline fluorescence. If the fluorescence was too weak to trace, we established a reference stack by empirically adjusting the contrast on a duplicate of the raw image stack. We used the reference stack for ROI selection and projected the selected ROIs back onto to the raw image stack for measurement. For frames in which the sample movement could not be automatically corrected, we manually outlined the ROIs used for measurements. Preparations with severe movement or deformation artifacts were abandoned to avoid unreliable measurements. Each ROI represent a traceable neuronal branch except Figure 2—figure supplement 2B, in which the ROIs were further manually partitioned into boutons and IBIs (Inter-Bouton Intervals) based on morphology. Fluorescence change were normalized to the pre-stimulation background except for Figure 3E, for which the data in each trial was normalized to the average ∆F/F during a 5 s period just before stimulation was initiated. No sample size is predetermined based on statistics. Ca2+ imaging data were acquired from at least six independent NMJs from at least five animals.

Statistical analysis

Data are presented as mean ± s.e.m. All data analysis was performed with Graphpad Prism 6, Microsoft Excel and custom Matlab codes (Source code 1). Mann-Whitney U test was used for comparison except in Figure 4, where One-sample T test was used for comparison with a specified value (0).

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
  22. 22
    Expression of synaptotagmin in Drosophila reveals transport and localization of synaptic vesicles to the synapse
    1. JT Littleton
    2. HJ Bellen
    3. MS Perin
    (1993)
    Development 118:1077–1088.
  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
    Differential Regulation of Small Clear Vesicles and Large Dense-Core Vesicles
    1. T Xu
    2. P Xu
    (2008)
    In: Z. W Wang, editors. Molecular Mechanisms of Neurotransmitter Release. Totowa, NJ: Humana Press. pp. 327–399.
  44. 44
  45. 45

Decision letter

  1. Leslie C Griffith
    Reviewing Editor; Brandeis University, United States
  2. Eve Marder
    Senior Editor; Brandeis University, United States
  3. Leslie C Griffith
    Reviewer; Brandeis University, United States
  4. Paul Taghert
    Reviewer; Washington University Medical School, 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.

Thank you for submitting your article "Imaging neuropeptide release at synapses with a genetically engineered reporter" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Leslie C Griffith as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Eve Marder as the Senior Editor. The following individual involved in review of your submission has also agreed to reveal their identity: Paul Taghert (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This paper develops a new set of tools for looking at release of peptides in real time. The idea is a good one and there is a need in the field for such tools. The authors describe a genetically encoded reporter system for dense core vesicle (DCV) fusion in Drosophila. They generate a host of fusions of GCaMP6 to ANF and DTK neuropeptides and identify a specific fusion event that localizes well to DCVs by both immunocytochemistry and immuno-EM. The authors show the sensor can respond to strong stimulation and compare the response to cytosolic GCaMP6 as a read-out of intra-terminal calcium changes. Overall, it's a potentially useful tool for the field, but the study could add a bit more quantification to demonstrate how well the sensor is actually working.

Essential revisions:

1) The major issue the reviewers had with the feasibility of this approach for dissecting DCV release is the conditions needed to trigger these changes – 70 Hz stimulation for 18 secs. These motor neurons would never experience such extreme stimulation in vivo. Peptidergic neurons typically fire in the 1-5 Hz range and while motor neurons fire at higher rates, natural stimulation is not of the duration used here. The signal-to-noise ratio (δ F changes) of this DCV sensor (compared to GCaMP alone which appears to be at least 5-10x more sensitive) is also rather poor. As such, it would be informative to have more information on the dynamics of their new reporter.

a) Provide a dose response curve to demonstrate both the required stimulation frequency (only 70 Hz is reported), the required length of stimulation (only 18 seconds is reported) and the calcium response curve. What does 1 Hz look like? Or 5 Hz?

b) The authors only report changes in the Type 3 terminals. What did the authors see in the Type 1 terminals that have fewer DCVs? Was DCV release detectable in these neurons, as has been reported in the field for the classical ANF-GFP DCV sensor? These data may be more useful than Type III data for investigators contemplating using this tool in neurons that do not have huge numbers of DCVs.

c) It would be helpful to use a benchmark to compare the methodology described here to that currently used in the field – the loss of ANF-GFP signal following stimulation. Is the current system actually an upgrade over that? There is unique value here in the potential to endogenously target specific neuropeptides for this technique, but if the sensitivity is far below that of ANF-GFP, the older system might still be the one of choice for dissecting the biology of DCV fusion overall. Without these data it is hard to claim an advance.

2) It would be helpful for the authors to further clarify in discussion what they actually think the sensor is reporting over time. It was not clear to reviewers exactly what the rise/falling/undershoot/recovery phases actually represent. The recovery phase is especially mysterious since the soma is gone so it cannot be replenishment via transport. DCVs are known to only release part of their cargo during stimulation, so presumably many of these DCVs still contain GCaMP-tether neuropeptide within them. Do the DCVs that have fused become less acidic and take up come extracellular calcium and thus activate the remaining GCaMP within them? Or do the authors think that the DCVs release all their content, or not take up calcium or lose their acidity during partial fusion events? This seems a potential confound – the authors are measuring a transient rise in calcium influx and pH change within the lumen of the DCV, not the actual release of the neuropeptide-GCaMP.

While the reviewers did not feel that it was required that this point be addressed with new experiments, it would be incredibly informative to assay their sensor in animals concurrently expressing an ANF-mCherry, ANF-mOrange (or other red-shifted tagged neuropeptide) where loss of the neuropeptide is directly reported in addition to the neuropeptide-GCaMP signal. That would really help clarify what is likely to be happening with their sensor during these phases of DCV release or partial fusion.

https://doi.org/10.7554/eLife.46421.024

Author response

Essential revisions:

1) The major issue the reviewers had with the feasibility of this approach for dissecting DCV release is the conditions needed to trigger these changes – 70 Hz stimulation for 18 secs. These motor neurons would never experience such extreme stimulation in vivo. Peptidergic neurons typically fire in the 1-5 Hz range and while motor neurons fire at higher rates, natural stimulation is not of the duration used here. The signal-to-noise ratio (δ F changes) of this DCV sensor (compared to GCaMP alone which appears to be at least 5-10x more sensitive) is also rather poor. As such, it would be informative to have more information on the dynamics of their new reporter.

a) Provide a dose response curve to demonstrate both the required stimulation frequency (only 70 Hz is reported), the required length of stimulation (only 18 seconds is reported) and the calcium response curve. What does 1 Hz look like? Or 5 Hz?

b) The authors only report changes in the Type 3 terminals. What did the authors see in the Type 1 terminals that have fewer DCVs? Was DCV release detectable in these neurons, as has been reported in the field for the classical ANF-GFP DCV sensor? These data may be more useful than Type III data for investigators contemplating using this tool in neurons that do not have huge numbers of DCVs.

We share the reviewers’ concern. In response to points 1a and 1b, we have added a new Figure 4, where we present stimulation frequency titrations of NPRRANP, NPRRdTK and GCaMP6 responses in both Type III neurons and Type Ib neurons. Because 70 mM potassium or 70 Hz electric stimulation were reported necessary for inducing maximum dense core vesicle release events in Drosophila type III motor neurons 1, we covered stimulation frequencies from 1 Hz to 70 Hz as the reviewer requested (1a). To compare and contrast the results between preparations and genotypes, for each preparation we normalized the ∆F/F peaks to responses obtained at 70Hz. We did not change the duration of stimulation, as it adds an additional dimension of complexity to the experiments.

Type III motor neurons contain primarily DCVs and exhibit a distinct morphology compared to type Ib motor neurons. As reviewer requested (1b), we have applied the same electric stimulation series to the glutamatergic type Ib motor neurons, which contain both SVs and DCVs 2.

To answer the reviewers’ questions in reverse order:

1b) NPRR responses can be observed in Type Ib as well as in Type III neurons (new Figure 4).

1a) In Type Ib neurons, no statistically significant NPRR responses could be observed at 1 Hz (NPRRANP: P=0.2817; NPRRdTK: P=0.0723) or 5 Hz (NPRRANP: P=0.0727; NPRRdTK: P=0.3467). However, responses at 10 Hz were statistically significant (NPRRANP: P=0.0108; NPRRdTK: P=0.0124). GCaMP responses were statistically significant at 5 Hz (P=0.0152) but not at 1 Hz (P=0.1194). Importantly, unlike the case in Type III neurons, we did not observe significant differences in the performance of the NPRRANP and NPRRdTK reporters in Type Ib neurons (Figure 4—figure supplement 1). In Type III neurons, no statistically significant responses were observed at 1 Hz (NPRRANP: P=0.3304; NPRRdTK: P=0.0719), 5 Hz (NPRRANP: P=0.3699; NPRRdTK: P=0.3467) and 10 Hz (NPRRANP: P=0.3084; NPRRdTK: P=0.5660). We observed a statistically significant increase in ∆F/F at 30 Hz for NPRRANP but not NPRRdTK; at 50 Hz and above we observed significant responses with both reporters. (NPRRANP: P=0.0021; NPRRdTK: P=0.034). The difference between both NPRRs at 30 Hz or 50 Hz was also statistically significant ((Figure 4—figure supplement 1).

(*Statistical methods: For in-sample comparison, one-sample T-test was used to compare response at a given stimulation frequency to 0. For between-sample comparison at the same stimulation frequency, Mann-Whitney U test was used.)

These results reveal two important new findings: (1) the difference between the performance of the two NPRRs we tested in Type III neurons is not a general property of these reporters, but a peculiarity of Type III neurons; (2) the high threshold for detection of NPRR responses in Type III neurons (30-50 Hz) is also not a general property of these reporters, since significant responses can be observed at frequencies as low as 10 Hz in Type 1b neurons. 10 Hz is well within the range of physiological rates for many neurons. These observations should allay concerns that NPRRs will have limited applicability.

Indeed, these new results reveal a novel application for NPRRs: they can be used to compare the relationship between stimulation and neuropeptide release between different types of neurons. In the present case, the difference in NPRR release detection thresholds between Type Ib and Type III neurons appears to be correlated with differences in the overall responses of these two cell types to electrical stimulation, as revealed by conventional GcaMP imaging (Figure 4C, F). However further studies will be necessary to substantiate this. To our knowledge, such a comparison of NP release thresholds between different neuronal cell types has not been performed previously. These data therefore illustrate a new area of investigation that is opened up by this technology.

In response to the reviewer’s comment that the signal-to-noise ratio of this sensor is rather poor, in comparison to that of GCaMP6, we would respectfully point out that the signal-to-noise ratio of the first-generation GCaMPs were also rather poor (max ∆F/F ≪100%). It took a managed team of protein engineers at Janelia 5 years and $10 million to optimize GCaMP from version 1.3, to version 6 (the first version with single-spike sensitivity). We would respectfully submit that the expectation that a first-generation reporter of a new class would perform as well as the 6th generation GcaMP is unrealistic. We anticipate that there will be multiple iterations of NPRRs to optimize their performance, but such optimization is beyond the scope of this manuscript.

c) It would be helpful to use a benchmark to compare the methodology described here to that currently used in the field – the loss of ANF-GFP signal following stimulation. Is the current system actually an upgrade over that? There is unique value here in the potential to endogenously target specific neuropeptides for this technique, but if the sensitivity is far below that of ANF-GFP, the older system might still be the one of choice for dissecting the biology of DCV fusion overall. Without these data it is hard to claim an advance.

We took the reviewers’ point and attempted to benchmark our method to ANP-GFP. While both methods showed comparable decreases in fluorescence in response to 70 mM KCl (see below), in response to electrical stimulation, we failed to observe any decrease in ANP-GFP signals in Type Ib or Type III neurons (Author response image 1A1,2). In contrast, under the same stimulation conditions NPRRANP clearly exhibited an increase in dF/F (Author response image 1A3,4). In other words, in our hands the ANPGFP method yielded no signal (decreased fluorescence) in response to electrical stimulation, whereas the NPRR clearly did so. Therefore rather than it being the case that the sensitivity of NPRRs is “far below that of ANF-GFP,” it appears that NPRRs are far more sensitive than ANF-GFP, in the context of electrical stimulation.

Author response image 1
Benchmarking NPRR with ANP-GFP.

Trace from a representative experiment showing changes in fluorescence intensity (∆F/F) of ANP-GFP and NPRRANP evoked by electrical stimulation (Red bar, frequency shown in A1). No significant ANP-GFP fluorescent change was observed within stimulations, in both Type Ib (A1) and Type III neurons (A2), while NPRRANP displays distinct characteristics in Type Ib (A3) and Type III(A4). Green line: mean ∆F/F; Grey shading: s.e.m envelope.

Details: We measured the native fluorescent intensity of ANP-GFP or of NPRRANP directly in fixed specimens following stimulation with 70 mM KCl. In this case, we observed a significant decrease in ANP-GFP fluorescence (36.04% decrease, N=7) after 10 mins, consistent with published work from the Levitan laboratory 1,3. In parallel, NPRRANP underwent a comparable level of fluorescent loss (33.57% decrease, N=7), suggesting a similar sensitivity of the two reporters under these conditions of bath depolarization.

We cannot explain our failure to observe any fluorescent loss of ANP-GFP in response to tenable electrical stimulation, in contrast to published results from the Levitan laboratory 1,3. Whatever the explanation, our results demonstrate the superiority of the NPRR methodology. In addition to being able to visualize specific neuropeptides, the ultimate objective of generating NPRRs is to be able to measure NP release in behaving animals over time. For this purpose, a positive signal would be more optimal than a negative signal (ANP-GFP), as fluorescent loss could be confounded by other technical issues, such as photobleaching or movement.

2) It would be helpful for the authors to further clarify in discussion what they actually think the sensor is reporting over time. It was not clear to reviewers exactly what the rise/falling/undershoot/recovery phases actually represent. The recovery phase is especially mysterious since the soma is gone so it cannot be replenishment via transport. DCVs are known to only release part of their cargo during stimulation, so presumably many of these DCVs still contain GCaMP-tether neuropeptide within them. Do the DCVs that have fused become less acidic and take up come extracellular calcium and thus activate the remaining GCaMP within them? Or do the authors think that the DCVs release all their content, or not take up calcium or lose their acidity during partial fusion events? This seems a potential confound – the authors are measuring a transient rise in calcium influx and pH change within the lumen of the DCV, not the actual release of the neuropeptide-GCaMP.

We appreciate the reviewers’ request to clarify the temporal kinetics we observed. As we stated in the original submission, we interpret the rising phase as reflecting epochs of NP release. The falling phase where an undershoot was observed suggests that the overall decrease of GCaMP fluorescence represents a reduction of total DCV NPRR cargo in the bouton area following vesicle fusion.

The reviewers state that the recovery phase is mysterious, because replenishment via transport cannot occur in the absence of the soma, which is removed in these preparations. The reviewer’s statement is partially incorrect: recent studies have demonstrated that axonal transport indeed occurs during short-term presynaptic protein remodelling with4 or without5 the ventral verve cord (i.e., somata) removed, over a time course ranging from 10 – 30 min 6,7, a timeframe similar to that in our experimental design. Thus, replenishment of DCVs via continuous transport from unstimulated axon segments contained within our preparation seems the most parsimonious explanation for the recovery phase.

DCV release occurs in several modes. At this point, we are unable to determine whether an individual DCV unloads all or a portion of its content in our experiments, although as mentioned above (response to point 1c) we detected an ~36% decrease in overall GCaMP fluorescence following high potassium stimulation. Furthermore, the ∆F/F undershoot observed in the recovery phase supports the idea that at least some, if not all, NPRR molecules are released to the extracellular space, since a purely intravesicular change in calcium/pH would not cause such an undershoot. Nevertheless, we cannot fully exclude the possibility that some fraction of NPRR increased ∆F/F signal is contributed by unreleased NPRR molecules, and now acknowledge this in the Discussion (second paragraph).

While the reviewers did not feel that it was required that this point be addressed with new experiments, it would be incredibly informative to assay their sensor in animals concurrently expressing an ANF-mCherry, ANF-mOrange (or other red-shifted tagged neuropeptide) where loss of the neuropeptide is directly reported in addition to the neuropeptide-GCaMP signal. That would really help clarify what is likely to be happening with their sensor during these phases of DCV release or partial fusion.

We agree with the reviewers’ suggestion to have an additional channel to report neuropeptide loss directly. In fact, we initially generated sixteen different neuropeptide precursor-Red Fluorescent Protein fusions (Figure 1—figure supplement 1). Unfortunately, all 16 red NPRRs exhibited 1) nearly invisible baseline fluorescence; and 2) no fluorescence change following high potassium challenge (unlike GFP fusions). We tried to investigate the reason for this failure using Neuropred (http://stagbeetle.animal.uiuc.edu/cgi-bin/neuropred.py), an online algorithm that predicts potential peptidergic cleavage sites within a given protein. This analysis revealed that RFP and tdTomato have 10 and 25 predicted cleavage sites, respectively, as opposed to 3 sites in GCaMP6s, mainly reflecting the high abundance of positively-charged residue repeats (Lys-Lys, Lys-Arg) in RFPs. If these cleavage sites are accessible to DCV proteases that processes packaged NP precursors, it would explain why no red fluorescence was observed with these reporters. We have now mentioned this in the text. Apparently, generating such RFP-based reporters would require mutagenizing all such sites to avoid cleavage, while retaining red fluorescence. Such protein engineering is desirable, but outside the scope of the present manuscript.

We appreciate the suggestion to use orange fluorescent proteins such as mOrange or mKO as a fusion alternative. Although it was not included in our first round of design, we would be glad to try it in the future.

References:

1) Levitan, E. S., Lanni, F. and Shakiryanova, D. In vivo imaging of vesicle motion and release at the Drosophila neuromuscular junction. Nat Protoc 2, 1117-1125, doi:10.1038/nprot.2007.142 (2007).

2) Rao, S., Lang, C., Levitan, E. S. and Deitcher, D. L. Visualization of neuropeptide expression, transport, and exocytosis in Drosophila melanogaster. J Neurobiol 49, 159-172 (2001).

3) Shakiryanova, D., Tully, A. and Levitan, E. S. Activity-dependent synaptic capture of transiting peptidergic vesicles. Nat Neurosci 9, 896-900, doi:10.1038/nn1719 (2006).

4) Goel, P. et al. Homeostatic scaling of active zone scaffolds maintains global synaptic strength. J Cell Biol 218, 1706-1724, doi:10.1083/jcb.201807165 (2019).

5) Bohme, M. A. et al. Rapid active zone remodeling consolidates presynaptic potentiation. Nat Commun 10, 1085, doi:10.1038/s41467-019-08977-6 (2019).

https://doi.org/10.7554/eLife.46421.025

Article and author information

Author details

  1. Keke Ding

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Yifu Han
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5261-4843
  2. Yifu Han

    1. Department of Neurobiology, University of Southern California, Los Angeles, United States
    2. Neuroscience Graduate Program, University of Southern California, Los Angeles, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology, Writing—review and editing
    Contributed equally with
    Keke Ding
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1201-654X
  3. Taylor W Seid

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Validation, Investigation, Methodology
    Competing interests
    No competing interests declared
  4. Christopher Buser

    Oak Crest Institute of Science, Monrovia, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4379-3878
  5. Tomomi Karigo

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Shishuo Zhang

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Validation, Investigation
    Competing interests
    No competing interests declared
  7. Dion K Dickman

    Department of Neurobiology, University of Southern California, Los Angeles, United States
    Contribution
    Resources, Formal analysis, Supervision, Funding acquisition, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1884-284X
  8. David J Anderson

    1. Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    2. Howard Hughes Medical Institute, California Institute of Technology, Pasadena, United States
    3. Tianqiao and Chrissy Chen Institute for Neuroscience, California Institute of Technology, Pasadena, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    wuwei@caltech.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6175-3872

Funding

National Institutes of Health (R21EY026432)

  • David J Anderson

National Institutes of Health (R01DA031389)

  • David J Anderson

National Institutes of Health (NS091546)

  • Dion K Dickman

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

Acknowledgements

We thank T Südhof for advice and encouragement at the inception of this project, T Littleton for suggesting GCaMP as a potential reporter for DCV release, B Pfeiffer for advice on construct design, members of the Anderson laboratory for critical feedback, K Zinn for helpful comments on the manuscript, P Goel for the information on antibodies, B Kiragasi for Master-9 programming, K Chen and D Ma for advice on imaging analysis and programming, G Mancuso and C Chiu for administrative assistance and lab management, and A Sanchez for maintenance of fly stocks. This work was supported by NIH BRAIN Initiative grant R21EY026432 and R01 DA031389 (to DJA) and NS091546 (to DD). DJA is an Investigator of the Howard Hughes Medical Institute.

Senior Editor

  1. Eve Marder, Brandeis University, United States

Reviewing Editor

  1. Leslie C Griffith, Brandeis University, United States

Reviewers

  1. Leslie C Griffith, Brandeis University, United States
  2. Paul Taghert, Washington University Medical School, United States

Publication history

  1. Received: February 27, 2019
  2. Accepted: June 25, 2019
  3. Accepted Manuscript published: June 26, 2019 (version 1)
  4. Version of Record published: July 4, 2019 (version 2)

Copyright

© 2019, Ding 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

  • 3,805
    Page views
  • 781
    Downloads
  • 0
    Citations

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

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. Developmental Biology
    2. Neuroscience
    Jesús Rodriguez Curt et al.
    Research Article Updated
    1. Neuroscience
    Xiaoxing Zhang et al.
    Research Article Updated