Innervation modulates the functional connectivity between pancreatic endocrine cells

  1. Yu Hsuan Carol Yang  Is a corresponding author
  2. Linford JB Briant
  3. Christopher A Raab
  4. Sri Teja Mullapudi
  5. Hans-Martin Maischein
  6. Koichi Kawakami
  7. Didier YR Stainier  Is a corresponding author
  1. Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Germany
  2. Exeter Centre of Excellence for Diabetes Research, Institute of Biomedical and Clinical Science, University of Exeter Medical School, United Kingdom
  3. Oxford Centre for Diabetes, Endocrinology and Metabolism, Radcliffe Department of Medicine, University of Oxford, United Kingdom
  4. Division of Molecular and Developmental Biology, National Institute of Genetics, Department of Genetics, SOKENDAI, Japan

Abstract

The importance of pancreatic endocrine cell activity modulation by autonomic innervation has been debated. To investigate this question, we established an in vivo imaging model that also allows chronic and acute neuromodulation with genetic and optogenetic tools. Using the GCaMP6s biosensor together with endocrine cell fluorescent reporters, we imaged calcium dynamics simultaneously in multiple pancreatic islet cell types in live animals in control states and upon changes in innervation. We find that by 4 days post fertilization in zebrafish, a stage when islet architecture is reminiscent of that in adult rodents, prominent activity coupling between beta cells is present in basal glucose conditions. Furthermore, we show that both chronic and acute loss of nerve activity result in diminished beta–beta and alpha–beta activity coupling. Pancreatic nerves are in contact with all islet cell types, but predominantly with beta and delta cells. Surprisingly, a subset of delta cells with detectable peri-islet neural activity coupling had significantly higher homotypic coupling with other delta cells suggesting that some delta cells receive innervation that coordinates their output. Overall, these data show that innervation plays a vital role in the maintenance of homotypic and heterotypic cellular connectivity in pancreatic islets, a process critical for islet function.

Editor's evaluation

The role of islet innervation on endocrine cell activity is currently not well defined. Previously, Yang et al. described embryonic islet innervation dynamics in zebrafish and the role of neural activity in proper glucose homeostasis and δ-cell formation. In their new manuscript, the authors introduced a novel transgenic model for whole islet in vivo calcium imaging. By applying this tool in a series of innovative and technically challenging approaches, the authors provide novel insights into how neuronal inputs influence pancreatic endocrine cell connectivity and function. Overall, this is an important study that adds to our understanding for the role of neuronal interactions with the islet.

https://doi.org/10.7554/eLife.64526.sa0

Introduction

Tight regulation of hormone release from pancreatic islets is critical for glucose homeostasis and its disruption can lead to diabetes mellitus (Noguchi and Huising, 2019). Pancreatic islets are composed of different cell types, including the hormone producing alpha, beta, and delta cells, peripheral nerves, and vascular endothelial and smooth muscle cells. Studies have implicated signaling from the vascular scaffold (Almaça et al., 2014; Mullapudi et al., 2019) and nerve networks (Rodriguez-Diaz et al., 2012; Taborsky et al., 1998; Borden et al., 2013; Yang et al., 2018; Makhmutova et al., 2019; Tarussio et al., 2014) during the development and function of pancreatic islet cells. However, it remains difficult to investigate the immediate effects of acute nerve modulation on islet cell function. Given the alterations in islet innervation architecture in some models of diabetes (Mundinger and Taborsky, 2016; Mundinger et al., 2016; Tang et al., 2018), it is imperative to understand whether disruption of nervous control can contribute to diabetes etiology.

Different methods of assessing islet cell function have provided important clues into the role of autocrine and paracrine signaling in this process. Electrophysiological recordings have provided fundamental insights into isolated islet cell function (Vergari et al., 2020; Camunas-Soler et al., 2020; Hastoy et al., 2018), including functional connectivity studies that identified homotypic as well as heterotypic coupling between endocrine cells (Moreno et al., 2005; Briant et al., 2018). However, assessing islet function in live animals with undisrupted vascular and nerve networks remains challenging. Calcium dynamics is a good readout of the function of all islet cell types because its influx is critical for hormone release. However, no studies to date have been able to record simultaneously the activity of all islet endocrine cell types in the intact organ of a living animal, which is required to understand how the different endocrine cell types respond to physiological perturbations individually and interdependently. To this end, we established an in vivo imaging platform to visualize the activity of all the islet cell types by combining calcium imaging with cell type reporters. We investigated the functional connectivity between homotypic and heterotypic cell pairs by analyzing the correlation patterns in their intracellular calcium changes. Chronic and acute inhibition of nerve activity captured its dynamic control of the functional connectivity between islet endocrine cells.

Results and discussion

The activity of all pancreatic endocrine cell types can be studied simultaneously in vivo

The zebrafish primary islet becomes highly innervated (Yang et al., 2018) and vascularized (Mullapudi et al., 2019; Hen et al., 2015; Toselli et al., 2019) early in development (Figure 1A). Fluorescent reporters for different pancreatic endocrine cell types, including beta, alpha, and delta cells, were used to study the establishment of islet cytoarchitecture (Figure 1B). By 100 hours post fertilization (hpf), a beta cell core and alpha cell mantle layout are observed (Figure 1B, C), in agreement with previous studies (Biemar et al., 2001), and reminiscent of adult rodent islets (Brereton et al., 2015) and small human islets (Bonner-Weir et al., 2015). Simultaneous functional assessment of all islet cell types in vivo required reporters for cell activity and cell identity. We used the Et(1121A:GAL4FF) enhancer trap line with the Tg(UAS:GCaMP6s) line for calcium imaging of all islet cells and a subset of peri-islet neurons (Figure 1D, E, Figure 1—video 1), as well as the Tg(ins:mCardinal) and Tg(sst2:RFP) lines to assign beta and delta cell identities, respectively; alpha cells were identified by their mantle localization and/or by immunostaining (Figure 1E). Thus, for the first time, we were able to assess the activity of all islet cells along with some peri-islet neurons in their native environment in an intact living animal (Figure 1F, Figure 1—video 2).

Figure 1 with 2 supplements see all
Pancreatic islet cell activity is visualized in vivo with preserved vascular and neural networks.

(A) Wholemount immunostaining of wild-type zebrafish at 100 hours post fertilization (hpf) for acetylated Tubulin (nerves), Tg(kdrl:GFP) expression (vessels), and Insulin (beta cells), and counterstaining with DAPI (diamidino-2-phenylindole, DNA). (B) 100 hpf Tg(ins:mCardinal); Tg(sst2:RFP); Tg(gcga:GFP) zebrafish stained with DAPI (DNA). (C) Mean distance of pancreatic islet cells to islet core reveals a beta cell core and alpha cell mantle; mean ± SEM, n = 11 animals, p values from one-way analysis of variance (ANOVA) with Holm–Sidak’s multiple comparisons test; see Figure 1—source data 1. (D) 100 hpf Et(1121A:GAL4FF); Tg(UAS:GCaMP6s); Tg(ins:mCardinal); Tg(sst2:RFP) zebrafish stained with DAPI (DNA). (E) 100 hpf Et(1121A:GAL4FF); Tg(UAS:GCaMP6s); Tg(ins:mCardinal) zebrafish stained for Glucagon (alpha cells) and DNA. (F) Schematic of documented interactions between beta, delta, and alpha cells and of intracellular calcium recordings in each of these cell types. Maximum intensity projections or single planes are presented; A, anterior; D, dorsal; V, vagus nerve; P, pancreas.

Figure 1—source data 1

Figure 1C.

Mean distance of pancreatic islet cells to islet core reveals a beta cell core and alpha cell mantle.

https://cdn.elifesciences.org/articles/64526/elife-64526-fig1-data1-v2.xlsx

A subset of islet cells display activity coupling with peri-islet neurons

Pancreatic islet innervation is in contact with all islet cell types (Figure 2A, B). However, both the beta and delta cells have a higher density of nerves in contact with their cell surfaces (Figure 2C). Prior to 120 hpf, this innervation is only from the vagus nerve (Yang et al., 2018). To investigate whether peri-islet neurons could actively modulate intra-islet coordination of activity, we first used the Et(1121A:GAL4FF); Tg(UAS:GCaMP6s) line which also labels a subset of peri-islet neurons (Figure 1—video 1) to simultaneously image the calcium dynamics in neurons and islet cells. While a higher number of neural activity connection to beta cells were observed, normalization to the number of cells imaged revealed no significant differences in the percentage of beta, delta, and alpha cells that display activity coupling with peri-islet neurons (Figure 2D, E). From the normalized single-cell calcium traces, correlation matrices, and average correlation coefficients (Ravg), we observed that homotypic coupling between beta cells is more prominent than those between delta and alpha cells (Figure 2F–H). Notably, we found a significant increase in homotypic coupling for the delta cells that display neural activity connection (Figure 2H). Future studies will determine whether direct neural activity connection is critical for the regulation of this delta cell subset.

Pancreatic nerves display differential interactions with islet cell types.

(A) 100 hpf Tg(ins:mCardinal), Tg(sst2:RFP), and Tg(gcga:GFP) zebrafish immuno-stained for acetylated Tubulin (nerves). (B) Segmentation and classification of islet nerves that are in contact with beta, delta, or alpha cell surfaces (magenta). All remaining islet nerves are colored in cyan. (C) Percentage of islet nerve density in contact with the specified pancreatic islet cell type; a majority of the nerves are in contact with beta and delta cells; mean ± standard error of the mean (SEM), n = 7–18 animals, p values from one-way analysis of variance (ANOVA) with Holm–Sidak’s multiple comparisons test; see Figure 2—source data 1. (D) Following exposure to elevated glucose and calcium imaging of pancreatic islet cells and peri-islet neurons at 105–110 hpf, correlation analysis was used to identify beta, delta, and alpha cells with neural activity connection. The percentage of cells of a given type with neural activity connection shows no significant difference between the different cell types; mean ± SEM, n = 6 animals, p values from one-way ANOVA with Holm–Sidak’s multiple comparisons test; see Figure 2—source data 2. (E) Correlation maps of a peri-islet neuron and beta, delta, and alpha cells. Individual cells are plotted with their coordinates. The strength of the correlation (Ravg) between cell pairs is drawn with a grayscale line and the number of connections for each cell is represented by the circle size. (F) Normalized calcium traces of pancreatic islet cells (including delta, beta, alpha, and unidentified cells) and a neuron (n). Individual islet cells were assigned to a cell type and given a cell id. (G) Correlation matrix of cell activity. Individual cells were assigned to a cell type (including delta, beta, alpha, and unidentified cells, and neurons) and average correlation coefficients for given cell pairs (matrix row-column intersects) were calculated. Areas displaying homotypic interactions are highlighted. (H) Average correlation coefficients from individual animals between homotypic cell pairs were divided into two groups (cells with or without neural connection); n = 6 animals, p values from paired t-tests.

Figure 2—source data 1

Figure 2C.

Percentage of islet nerve density in contact with the specified pancreatic islet cell type.

https://cdn.elifesciences.org/articles/64526/elife-64526-fig2-data1-v2.xlsx
Figure 2—source data 2

Figure 2D.

The percentage of cells of a given type with neural activity connection.

https://cdn.elifesciences.org/articles/64526/elife-64526-fig2-data2-v2.xlsx

Homotypic and heterotypic coupling between endocrine cells requires pancreatic innervation

Activity coupling between pancreatic endocrine cells can be mediated by autocrine and paracrine signaling, gap junctions, and other means. To investigate whether pancreatic innervation is critical for intra-islet coordination of activity, we used different approaches to chronically or acutely inhibit neural signaling. We used endoderm transplantation to generate chimeric zebrafish that express two GAL4/UAS systems in different germ layer-derived tissues and investigated the role of chronic neural inhibition on islet function (Figure 3A). Pan-neural expression of botulinum toxin (BoTx) chronically inhibits neurotransmitter release (Yang et al., 2018; Sternberg et al., 2016) and leads to elevated glucose levels at 100 hpf (Yang et al., 2018; Figure 3B). While primary islet volume was consistently greater in BoTx+ larvae at 100 hpf (Figure 3C; as we reported for earlier stages Yang et al., 2018), we did not observe changes in the architectural arrangement of the different islet cell types (Figure 3D). From the normalized single-cell calcium traces, correlation matrices, and average correlation coefficients (Ravg), we observed that the calcium dynamics in BoTx+ larvae were significantly disrupted, with impairment in beta cell coupling under both basal glucose and glucose stimulated conditions (Figure 3E–G, Figure 3—figure supplement 1). Our measure of Ravg over increasing intercellular distance revealed the expected decline in coupling over distance, and the significant difference in the elevation of the linear regression further confirmed the altered synchronicity between beta cells (Figure 3H).

Figure 3 with 3 supplements see all
Chronic inhibition of synaptic transmission disrupts islet cell activity.

(A) Schematic of transplants to generate chimeras with endodermal organs derived entirely from donor embryos. (B) Whole larva-free glucose-level measurements at 100 hpf; mean ± standard error of the mean (SEM), n = 24–32 batches of 5 larvae per replicate, p value from t-test; see Figure 3—source data 1. (C) Quantification of primary islet mass; n = 21–29, p value from t-test; see Figure 3—source data 2. (D) Mean distance of pancreatic islet cells to islet core; mean ± SEM, n = 21–29 animals, p values from paired t-tests; see Figure 3—source data 3. (E) Normalized calcium traces of pancreatic islet cells (including delta, beta, alpha, and unidentified cells). Individual islet cells were assigned to a cell type and given a cell id; LG, basal condition; HG, glucose treated condition. (F) Correlation matrices of islet cell activity. Individual islet cells were assigned to a cell type and given a cell id, and average correlation coefficients for given cell pairs (matrix row-column intersects) were calculated for LG (basal condition; bottom left triangle) and HG (glucose treated condition; top right triangle). (G) Average beta cell correlation coefficients in individual larvae; n = 5–8 animals, p values from two-way analysis of variance (ANOVA) with Holm–Sidak’s multiple comparisons test; *, corresponding calcium traces and correlation matrices shown in panels E and F. (H) Average homotypic (beta–beta) and heterotypic (delta–beta and alpha–beta) cell correlation coefficients with cell distance distribution from 1 (close) to 10 (far), mean and linear regression (solid lines) with 95% confidence intervals; n = 5–8 animals, p values of slope and intercept from simple linear regression. (I–K) Fraction time analysis of heterotypic delta–beta, alpha–delta, and alpha–beta cell pairs for times when both are active (I), both are silent (J), and one is active and one is silent (K); mean ± SEM, n = 32–160 cell pairs, p values from two-way ANOVA with Holm–Sidak’s multiple comparisons test; magenta circle, active state; cyan circle, silent state.

To determine how perturbations in neural signaling influenced communication between different endocrine cell types, we conducted correlation analysis over increasing distance as well as fraction time analysis of heterotypic cell pairs that were in nearest proximity to each other (Figure 3H–K, Figure 3—figure supplement 2). Significant changes in the intercepts suggest impairments in delta–beta and alpha–beta heterotypic coupling (Figure 3H). Nearest neighbors have a greater likelihood of displaying coupling, as seen in our Ravg over increasing distance analysis. We analyzed single-cell calcium traces and determined the fraction of time a given nearest cell pair resides in a state when (1) both cells are active, (2) both cells are silent, and (3) one cell is silent, the other active. This analysis of activity patterns between nearest heterotypic cell pairs further supported the observed heterotypic coupling defects upon chronic neural inhibition (Figure 3I–K). Aside from the quiet phase, when both cell types are silent, we found changes in activity patterns between alpha–beta, delta–beta, and alpha–delta cell pairs (Figure 3I–K): upon chronic neural inhibition, the delta-silent/beta-active state was decreased while the delta-active/beta-silent state and the delta-active/alpha-silent state were both increased (Figure 3K). Although we cannot exclude potential defects in endocrine cell development, as reported in our previous study (Yang et al., 2018), these changes were not accompanied by alterations in delta cell calcium oscillation frequency nor peak height or duration (Figure 3—figure supplement 3). We also found a significant decrease in the alpha-active/beta-active state upon chronic neural inhibition (Figure 3I), suggesting that neural signaling is an important regulator of alpha–beta connectivity.

Given the potential role of pancreatic innervation on islet cell maturation, we next investigated the effects of acutely blocking neural activity using two different approaches. By lineage tracing, we found that the neural crest-derived peri-islet neurons were also labeled by the Et(1121A:GAL4FF) enhancer trap (Figure 4A, B), thereby allowing us to investigate the effects of photo-ablating a subset of peri-islet neurons on islet cell activity (Figure 4C). The photo-ablation of peri-islet neurons reduced islet nerve density (Figure 4—figure supplement 1A, B). While the oscillatory pattern of calcium dynamics was maintained in beta cells upon this ablation (Figure 4D), the coupling between beta cells was significantly decreased (Figure 4E–G). Notably, we did not observe further impairment in beta cell coupling over increasing distance (Figure 4G), suggesting that upon ablation of peri-islet neurons, the signal that initiates beta cell coupling was blunted while beta cells maintained their propensity for coupling across the islet. Whether this beta cell coupling is due to gap junctional (Benninger et al., 2008) and/or soluble factors warrants further studies. Delta–beta coupling was also reduced (Figure 4—figure supplement 1C). Unlike in the chronic neural inhibition scenario, in the fraction time analysis, delta–beta and alpha–delta coupling was not affected upon ablating peri-islet neurons (Figure 4H–J). However, we observed a decrease in alpha–beta coupling and in the alpha-active/beta-active state (Figure 4G–H). The observed coupling defects are unlikely due to changes in individual calcium spike characteristics, as no significant differences were observed in calcium oscillation frequency, peak height, or peak duration (Figure 4—figure supplement 2). Overall, the ablation of peri-islet neurons significantly disrupted beta–beta and alpha–beta connectivity, while conclusions regarding other heterotypic interactions will require further investigation into the various neural subsets that were targeted. It is likely that our targeting of peri-islet neurons affected at least those that guide the activity coupling between alpha and beta cells.

Figure 4 with 2 supplements see all
Targeted ablation studies reveal the crucial role of peri-islet neurons for islet cell activity.

(A) Schematic of lineage tracing of neural crest cells in Tg(sox10:CreERT2, myl7:GFP); Tg(ubb:loxP-CFP-loxP-nuc-mCherry); Tg(ins:mCardinal); Et(1121A:GAL4FF); Tg(UAS:GCaMP6s) zebrafish following 5 µM tamoxifen (TM) treatment from 16 to 24 hpf and staining at 108 hpf. (B) Wholemount immunostaining at 108 hpf for mCherry expression (neural crest-derived cells) and counterstaining with DAPI (DNA). Yellow arrowheads point to neural crest-derived cells positive for GCaMP6s expression. (C) Schematic of two-photon ablation experiment. (D) Normalized calcium traces of pancreatic islet cells (including delta, beta, alpha, and unidentified cells). Individual islet cells were assigned to a cell type and given a cell id; LG, basal condition; HG, glucose treated condition; HG a, ablation or mock ablation condition. (E) Correlation matrices of islet cell activity; LG, basal condition; HG, glucose treated condition; HG a, ablation or mock ablation condition. (F) Average beta cell correlation coefficients in individual larvae; n = 11–12 animals, p values from two-way analysis of variance (ANOVA) with Holm–Sidak’s multiple comparisons test. (G) Average homotypic (beta–beta) and heterotypic (alpha–beta) cell correlation coefficients with cell distance distribution from 1 (close) to 10 (far), mean and linear regression (solid lines) with 95% confidence intervals; n = 11–12 animals, p values of slope and intercept from simple linear regression. Fraction time analysis of heterotypic delta–beta, alpha–delta, and alpha–beta cell pairs for times when both are active (H), both are silent (I), and one is active and one is silent (J); mean ± standard error of the mean (SEM), n = 178–215 cell pairs, p values from two-way ANOVA with Holm–Sidak’s multiple comparisons test; magenta circle, active state; cyan circle, silent state.

Next, we took an optogenetic approach by generating a transgenic line that allows one to acutely photo-inhibit the release of neurotransmitters with a single pulse of blue light. This method has previously been used in Drosophila (Makhijani et al., 2017) and C. elegans (Lin et al., 2013; Qi et al., 2012) for targeted photo-ablation and photo-inhibition. Pan-neural expression of a singlet oxygen generator, miniSOG2, tethered to synaptic granules resulted in a blue light inducible loss of swimming activity in 110 hpf larvae (Figure 5—figure supplement 1A–C). Following this confirmation of the effectiveness of the tool, we studied neural control of islet cell activity upon acute photo-inhibition (Figure 5—figure supplement 1D). Surprisingly, photo-inhibition decreased glucose levels compared with transgene-negative zebrafish exposed to the same light condition (Figure 5A). Similar to peri-islet neural ablation, pan-neural photo-inhibition decreased beta cell connectivity (Figure 5B–E). Changes in delta–beta and alpha–beta heterotypic interactions were also observed upon acute neural inhibition (Figure 5E–H). A significant decrease in delta-silent/beta-active state reflects what we observed upon chronic neural inhibition (Figure 5H). Like with the photo-ablation of peri-islet neurons, no changes in nearest alpha–delta interactions were observed (Figure 5F–H). Notably, following acute neural inhibition, alpha–beta coupling (Figure 5E) and alpha-active/beta-active states (Figure 5F) were significantly decreased. While we cannot exclude a role for soluble factors from other peripheral organs, these changes in alpha–beta interactions were consistently observed upon both acute pan-neural and peri-islet inhibition, possibly reflecting a role for neurons in the maintenance of alpha–beta coupling.

Figure 5 with 2 supplements see all
Acute optogenetic inhibition of neurotransmitter release disrupts islet cell activity.

(A) Whole larva-free glucose-level measurements of Tg(elavl3:sypb-miniSOG2-P2A-mScarlet) zebrafish at 110 hpf; mean ± standard error of the mean (SEM), n = 22–31 batches of 5 larvae per replicate, p values from two-way analysis of variance (ANOVA) with Holm–Sidak’s multiple comparisons test; see Figure 5—source data 1. (B) Normalized calcium traces of pancreatic islet cells (including delta, beta, alpha, and unidentified cells). Individual islet cells were assigned to a cell type and given a cell id; LG, basal condition; HG, glucose treated condition. (C) Correlation matrices of islet cell activity; LG, basal condition; HG, glucose treated condition; HG λ, post-blue light treatment. (D) Average beta cell correlation coefficients in individual larvae; n = 10 animals, p values from two-way ANOVA with Holm–Sidak’s multiple comparisons test. (E) Average homotypic (beta–beta) and heterotypic (alpha–beta) cell correlation coefficients with cell distance distribution from 1 (close) to 10 (far), mean and linear regression (solid lines) with 95% confidence intervals; n = 10 animals, p values of slope and intercept from simple linear regression. Fraction time analysis of heterotypic delta–beta, alpha–delta, and alpha–beta cell pairs for times when both are active (F), both are silent (G), and one is active and one is silent (H); mean ± SEM, n = 141–220 cell pairs, p values from two-way ANOVA with Holm–Sidak’s multiple comparisons test; magenta circle, active state; cyan circle, silent state.

Figure 5—source data 1

Figure 5A.

Whole larva-free glucose-level measurements of Tg(elavl3:sypb-miniSOG2-P2A-mScarlet) zebrafish.

https://cdn.elifesciences.org/articles/64526/elife-64526-fig5-data1-v2.xlsx

Dissecting the complex interplay of local autocrine, paracrine, and gap junctional communication between different endocrine cells, in addition to vascular and nerve interactions, is often hindered by our inability to simultaneously study them in an intact organ within its innate environment. Imaging calcium dynamics with genetically encoded biosensors or calcium sensitive fluorescent indicators in individual islet cell types has been conducted in vitro with dispersed cells (Yang et al., 2013; Albrecht et al., 2015, whole islets Benninger et al., 2008; Johnston et al., 2016), and perfused pancreas slices (Stožer et al., 2013; Panzer et al., 2020), and in vivo with islets transplanted into the anterior chamber of the eye (Rodriguez-Diaz et al., 2012; Salem et al., 2019), as well as intravital imaging of the mouse pancreas (Adams et al., 2021). We report a noninvasive in vivo imaging strategy to study all the different pancreatic endocrine cell types within the same animal. Our three approaches to inhibit neural control, ranging in temporal and spatial specificity, provided useful insights into the role of neurons in regulating pancreatic islet function (Figure 6). Whether activity coupling between beta cells is in part mediated by gap junctions warrants further studies, but our data suggest that neural regulation is critical for the establishment and maintenance of beta cell connectivity as we consistently found decreased beta cell coupling upon chronic and acute neural inhibition. Given that our targeted neural ablation approach also led to this decline in beta cell coupling, it is likely that autonomic neural control is required for beta cell connectivity independently of possible indirect effects resulting from neural regulation of other organs. Whether this neural regulation is directly on beta cells or a bystander effect resulting from the regulation of other cell types, including delta and endothelial cells, requires further studies. It has been proposed that glucose sensing neurons regulate early postnatal beta cell proliferation and maintenance of beta cell function (Tarussio et al., 2014), and our data support the peri-islet localization of such neurons.

Graphical abstract summarizing the research findings.

Neuromodulation, with varying temporal and spatial specificity, provides useful insights into the role of neurons in regulating homotypic and heterotypic activity coupling between pancreatic islet cells. Green check marks indicate when significant defects were observed in correlation-based analysis across the whole islet. Blue check marks indicate when significant defects were observed in fraction time analysis between nearest neighbors.

We have focused our studies on the pancreatic beta, alpha, and delta cells; however, it is important to note that there are other endocrine cell types (gamma and epsilon cells) that remain undefined, but do display glucose induced activity coupling with beta cells (Figure 3F). Correlation analysis allowed us to study heterotypic coupling and fraction time analysis further allowed us to study activity patterns of heterotypic cell pairs that are near each other, upon loss of neural signaling. We found changes in delta–beta activity coupling, supporting a role for gap junctions in mediating electrical coupling between delta and beta cells (Briant et al., 2018). Both chronic and acute neural inhibition also blunted the functional connectivity between alpha and beta cells suggesting that neurons may have an important role in the paracrine potentiation of beta cell activity by glucagon released from alpha cells (Svendsen et al., 2018). We propose that alpha–beta interactions are specifically regulated by pancreatic innervation since peri-islet neural ablation led to a similar decrease in alpha–beta coupling and the alpha-active/beta-active state. However, given that pan-neural inhibition is required to induce changes in nearest delta–beta interactions, it is possible that central nervous control of other organs could in part be driving these changes. Importantly, our data further support a role for neurons in modulating delta cell activity, since delta cells that display neural activity connection exhibit a significant increase in their coupling to other delta cells. Future studies will determine whether the observed changes in delta cell activity directly reflect alterations in somatostatin release.

Selective targeting of subsets of neurons will advance our understanding of the pancreatic islet-neural interplay in health and disease, including diabetes pathophysiology. Through our in vivo studies of homotypic and heterotypic activity coupling, we have illustrated how studying functional connectivity can be achieved for the endocrine pancreas and discovered a critical role for neurons in mediating these connections. Given the cellular heterogeneity of organ composition, simultaneously evaluating the function of the different cell types that make up an organ provides insights that can be missed by simply investigating one cell type at a time. Interrogating the neural regulation of other organs, such as the liver, intestine, and kidney, can be achieved through the extended application of these tools. In combination with the ability to monitor organ development, function, and regeneration in vivo, this approach will allow one to address complex questions pertaining to the autonomic nervous system and its role in organ maintenance and dysfunction.

Materials and methods

Zebrafish transgenic lines and husbandry

Request a detailed protocol

All zebrafish husbandry was performed under standard conditions in accordance with institutional (MPG) and national ethical and animal welfare guidelines (Proposal numbers: B2/1041, B2/Anz. 1007, B2/1218). All procedures conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. Adult zebrafish were fed a combination of fry food (Special Diet Services) and brine shrimp five times daily and maintained under a light cycle of 14 hr light:10 hr dark at 28.5°C. Zebrafish embryos and larvae were grown in egg water at 28.5°C. Transgenic and mutant lines used were on the mitfaw2/w2 background and as described in Table 1.

Table 1
List of zebrafish transgenic lines.
NameSpecificity/purposeReference
Tissue specific promoter lines
Et(1121A:GAL4FF)nkgsaizgffm1121APancreatic islet cells, neuronsCurrent
Tg(ins:mCardinal)bns162Pancreatic beta cellsCurrent
Tg(elavl3:sypb-miniSOG2-P2A-mScarlet)bns529Inhibit postmitotic neuronsCurrent
Tg(elavl3:GAL4-VP16)zf357Postmitotic neuronsStevenson et al., 2012
Tg(sst2:RFP)gz19Pancreatic delta cellsLi et al., 2009
Tg(gcga:GFP)ia1Pancreatic alpha cellsZecchin et al., 2007
Tg(sox10:CreERT2, myl7:GFP)t007Neural crest cells, heart markerMongera et al., 2013
Tg(ubb:loxP-CFP-loxP-nuc-mCherry)jh63UbiquitousWang et al., 2015
Tg(kdrl:EGFP)s843Endothelial cellsJin et al., 2005
UAS lines
Tg(UAS:GCaMP6s)UAShspzGCaMP6sVisualize intracellular calciumMuto et al., 2017
Tg(UAS:BoTxBLC-GFP)icm21Inhibit neurotransmitter releaseYang et al., 2018; Sternberg et al., 2016

Generation of new transgenic lines

Request a detailed protocol

The Tg(elavl3:sypb-miniSOG2-P2A-mScarlet) line was generated by Tol2 transgenesis of a 8.7 kb elavl3 promoter (Addgene: 59531, AgeI restriction enzyme digested) driving expression of sypb (Addgene: 74316), miniSOG2 (Addgene: 87410), P2A-mScarlet (gift from A. Beisaw) cloned with Cold Fusion (System Biosciences). The Tg(ins:mCardinal) line was generated by Tol2 transgenesis of a 1.1 kb ins promoter (in-house plasmid, MluI restriction enzyme digested) driving expression of mCardinal (Addgene: 51311) cloned with Cold Fusion.

Generation of chimeric zebrafish

Request a detailed protocol

Et(1121A:GAL4FF); Tg(UAS:GCaMP6s); Tg(sst2:RFP); Tg(ins:mCardinal) donor embryos were injected with 200 pg sox32 mRNA, generated with mMESSAGE mMACHINE SP6 transcription kit (Thermo), at the one-cell stage to enhance endoderm formation. Tg(elavl3:GAL4-VP16); Tg(UAS:BoTxBLC-GFP) host embryos were injected with 0.3 ng sox32 morpholino (Gene Tools LLC) at the one-cell stage to inhibit endoderm formation. At the 1000-cell stage, single cells were removed from donor embryos with a glass capillary needle, and ~40 cells were transplanted into host embryos in the region fated to give rise to endoderm-derived tissues. Only larvae that displayed complete endoderm transplant, without obvious defects, and carried the relevant transgenes were used for downstream imaging experiments. Control chimeras were negative for BoTxBLC-GFP expression.

In vivo confocal microscopy

Request a detailed protocol

Live zebrafish between 100 and 110 hpf were anesthetized with 0.015% Tricaine and mounted in 0.8% low melting agarose in egg water containing 0.005% Tricaine for confocal imaging. A Zeiss LSM880 upright laser scanning confocal microscope equipped with a Plan-Apochromat ×20/NA1.0 dipping lens was used for imaging. Time-lapse calcium imaging was conducted in 25°C conditions and z-stacks were taken at 5-s intervals for 1–2 hr. Larvae were exposed to 75 mM glucose containing egg water at the indicated time points during the time-lapse imaging to increase whole larvae glucose levels within a physiological range (Figure 3—figure supplement 1). For the photo-inhibition experiments, the GCaMP6s signal was visualized with a Chameleon Vision II Ti:Sapphire Laser (Coherent) laser at 920 nm. For all other experiments, the GCaMP6s signal was visualized with an Argon laser at 488 nm.

Two-photon laser ablation

Request a detailed protocol

A Chameleon Vision II Ti:Sapphire Laser (Coherent) mounted on a Zeiss LSM880 microscope was used for two-photon single-cell laser ablation of peri-islet neurons labeled by the Et(1121A:GAL4FF) enhancer trap. The tunable laser was set at 800 nm to scan an ablation area of 4 µm2 at a scan speed of 1 with 10 iterations. Following ablation, embryos were rested for 5 min before continuation of calcium imaging. Controls were mock ablation of cells within 20 µm from the peri-islet neurons. Animals from the same clutch were randomly assigned to ablation and control groups.

Photo-inhibition of nerve activity

Request a detailed protocol

All experiments with Tg(elavl3:sypb-miniSOG2-P2A-mScarlet) animals were done on F1 to F3 larvae displaying strong and uniform mScarlet signal. Controls were transgene-negative siblings. To assess swimming behavior, 110 hpf larvae were individually isolated in 9-mm diameter circular wells filled with 200 µl egg water. A Nikon SMZ25 stereomicroscope with a P2-SHR PlanApo ×1/NA0.15 objective was used for time-lapse imaging of larvae at 1-s intervals; the same animals were imaged prelight and postexposure to a 5-min pulse of blue light (3.8 mW LED, 466/40 filter, Lumencor Sola Light Engine). To assess changes in glucose levels, pools of 40 larvae swimming in 9 cm diameter Petri dishes filled with 35 ml egg water were exposed to 0.3 mW blue LED light for 30 min prior to sample collection; control animals were not exposed to blue light and were randomly selected siblings from the same clutch. Notably, acute blue light exposure alone mildly reduced swimming behavior and increased glucose levels (Figure 5—figure supplement 1C, Figure 5A). Induction of a stress response likely led to this increase in glucose levels in wild-type larvae, as chronic 14 hr exposure to blue LED light for 3 consecutive days kills zebrafish larvae (Ustundag et al., 2019). To assess calcium changes, larvae were exposed to a 3-min pulse of blue light (2.1 mW 470 nm LED, Colibri) and the GCaMP6s signal was measured pre- and post-blue light exposure. While blinding was not possible, we randomized the order in which animals on a given day were imaged.

Wholemount immunostaining

Request a detailed protocol

Zebrafish were euthanized with Tricaine overdose prior to overnight fixation in 4% paraformaldehyde dissolved in phosphate-buffered saline (PBS) containing 120 µM CaCl2 and 4% sucrose, pH 7.4. The skin was manually removed with forceps, without disturbing the internal organs and the zebrafish were permeabilized with 1% Triton X-100, 1% dimethyl sulfoxide (DMSO) containing PBS for 3 hr at room temperature. Following blocking with 5% donkey serum (Jackson Immunoresearch) in blocking buffer (Dako), samples were incubated in primary antibodies overnight at 4°C, washed 4× with 0.025% Triton X-100 containing PBS, incubated in secondary antibodies overnight at 4°C, washed 4×, incubated in an increasing glycerol gradient of 25%, 50%, and 75%, and mounted in VectorShield mounting medium. The following antibodies and dilutions were used: mouse anti-glucagon (1:200, Sigma G2654), chicken anti-GFP (1:200, Aves GFP-1020), mouse anti-acetylated Tubulin (1:200, Sigma T7451), and guinea pig anti-insulin (1:300, DAKO A0564). Secondary antibodies used in this study include donkey anti-guinea pig AlexaFluor647 (1:200, Jackson 706-605-148), donkey anti-mouse AlexaFluor488 (1:300, Jackson 715-545-150) and AlexaFluor647 (1:200, Jackson 715-605-150), and donkey anti-chicken AlexaFluor488 (1:300, Jackson 703-545-155). Nuclei were stained with 25 µg/ml diamidino-2-phenylindole (DAPI, Sigma). Images were taken on a Zeiss LSM880 or LSM800 laser scanning confocal microscope equipped with a ×25/NA0.8 objective.

Data analysis

Request a detailed protocol

Image data were analyzed using Imaris (Bitplane) and Fiji (ImageJ). Correlation and fraction time analyses were performed using Matlab (code available upon request). Statistical analysis was performed using Prism 8 (GraphPad) and the type of test used is indicated in the figure legends. For comparison between two groups, a two-tailed Student’s t-test was used to determine the p values. For comparison between more than two groups, an ordinary one-way analysis of variance ANOVA with Holm–Sidak’s multiple comparisons test or a two-way ANOVA with Holm–Sidak’s multiple comparisons test was used to determine the p values. The number of animals or cells analyzed, and the p values are reported in the figure and figure legends. All experiments were repeated on different days using at least three different clutches of animals.

Data availability

All data generated or analysed during this study are included in the manuscript, figures, and figure legends. Source data files have been provided for Figures 1, 2, 3, and 5.

References

Decision letter

  1. Lori Sussel
    Reviewing Editor; University of Colorado Anschutz Medical Campus, United States
  2. Marianne E Bronner
    Senior Editor; California Institute of Technology, United States
  3. Sarah M Knox
    Reviewer; University of California, San Francisco, United States
  4. Dirk Meyer
    Reviewer; University of Innsbruck, Innsbruck, Austria

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for submitting your article "Innervation modulates the functional connectivity between pancreatic endocrine cells" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Marianne Bronner as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Sarah M Knox (Reviewer #3); Dirk Meyer (Reviewer #4).

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

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

The role of islet innervation on endocrine cell activity is currently not well defined. Previously, Yang et al. provided a detailed description of embryonic islet innervation dynamics in zebrafish and they established that innervation and/or neural activity is required for proper glucose homeostasis and δ-cell formation. In their new manuscript, they introduced a novel transgenic model for whole islet in vivo calcium imaging. By applying this tool in a series of innovative and technically challenging approaches, the authors employ 3 distinct methods to modulate neuronal activity using a range of reporter lines, mutant lines and intravital imaging to examine these islet cell interactions. In addition they demonstrate that larval β-cells express the gap-junction protein Cx43 and that β-cell coupling is disturbed in cx43 mutants. On the whole these methods are robustly applied, in addition to the analytical methods used to interpret 'connectivity' from the ca2+ dynamics. Overall, this is an important study that adds to our currently weak understanding for the role of neuronal interactions with the islet.

Summary:

Overall, the reviewers are supportive of publication and study provides important novel insights into our understanding of the underlying mechanisms mediating coordinated pancreatic endocrine cell responses. However, they indicate the need for a number of important clarifications (particularly related to interpreting the connectivity analysis) and places where data presentation is not always clear. In addition, all 3 reviewers indicated that throughout the manuscript the authors' made overly speculative statements and there is some concern that the study does not provide evidence of direct influence of neuronal activity. Finally, some of the experiments require additional and/or alternative statistical analyses. While most of these revisions will not require additional experimentation, a small number of experiments are suggested that would help support the authors' conclusions.

Essential revisions:

1. An important caveat in this study is that connectivity is used to refer to cells showing similar ca2+ elevations, whether calculated via correlation coefficients or fraction of time two neighboring cells are 'active'/'inactive'. This does not per se indicate if the two cells truly are directly interacting, or whether they are instead each interacting with some other factor. For instance, an increased correlation coefficient between β cells upon innervation could be due to enhanced gap junction coupling or paracrine communication between β cells or it could be due to neuronal activity driving the activity of the β cells. Some literature (e.g. Zhang et al., Biophys.J. 2008) would suggest the latter, not the former, and thus conflicts with the authors conclusions.

2. Related to this above point, in figures 2-4 the authors use one type of analysis for assessing homotypic interactions between β cells, but a separate analysis for assessing heterotypic interactions. Why do the authors not use a correlation based analysis between cell neighbors for heterotypic interactions? Would they observe changes if such correlation analysis is employed? (I see why they cannot use it across all cells).

Also, if the 'connectivity' between two cell types changes (e.g. β-δ), then the apparent connectivity with the other cell type (α) would very likely change irrespective of whether any actual connection is present. For instance, it seems to me that changes in β cell correlation would change the apparent β-δ and β-α connectivity, irrespective of the actual link between these cell types. Δ-α 'connectivity' doesn't change under any of the experimental conditions which would support this alternative conclusion. How do the authors exclude this alternative conclusion?

Finally, are α, β and δ cell activity patterns themselves altered under each neuronal modulation and Cx43 deletion? E.g. the duration and amplitude of ca2+ transients. If so could these explain differences in apparent 'connectivity'?

3. There are many overly speculative statements that need either toning down or including data that better supports the conclusion: e.g.

– Abstract: "δ cells receive innervation which coordinates its (the islet) output." Such an interpretation would rely on modulating neuronal-δ cell interactions specifically.

– Lines 123-124: hyper-activation of δ cells is a strong statement. Instead, it seems that δ cells are activated more than their neighboring β and α cells. But the change in δ cell activity is not especially strong. Is δ cell activity itself (% time elevated or spike height) different?

– Line 124: similarly, the authors did not measure somatostatin release so cannot say this is directly the result of δ cell activity.

– Line 192-194: the authors should combine their Cx43 with neural modulation to test if pancreatic innervation is indeed mediating the remaining intra-islet coordination.

4. In performing neuronal modulation, pan-neural chronic botox treatment (Figure 2) significantly disrupted both homotypic and heterotypic interactions; acute photo-ablation of peri-islet neurons (Figure 3) only disrupted homotypic interactions (although note my comment 1); and pan-neural acute inhibition of neurotransmitter release (Figure 4) significantly disrupted both homotypic and heterotypic interactions. How can the authors exclude the pan-neural disruptions affecting islet function independent of pancreas innervation, e.g. innervation affecting another organ and changing some circulating factor? Especially given both pan neural methods generate different findings than the peri-islet disruption.

5. Line 140-141: you cannot say there is a trend towards a decrease in α-active β active state based on a p of 0.06 within a set of data that contains 12 related comparisons. This leads me to ask if the authors have corrected for multiple comparisons in their statistical analysis? (for 2J-L, for 3I-K, for4G-I). Throughout, the authors should state in the figure captions what statistical analysis was performed for which comparison indicated.

6. Figure 2 G: the change in Ravg (correlation) is marginally significant. The results refer to these changes as 'significantly disrupted', 'severe disruption', drastically altered' which needs toning down, with 'marginal' being a more apt term. Furthermore, in F the 'representative' correlation matrices are not congruent with the data in G. At the very least in G the data points corresponding to the example in F should be indicated, and better would be to include more representative matrices.

7. In figures 2-4 I do not follow the Ravg vs distance measurement. This is interpreted to be indicating β cell wave propagation. I do not see how the decline in relative correlation with distance would refer to wave propagation? This is really a decorrelation which is entirely expected for a system that is not perfectly coupled. Other phenomena than wave propagation could explain such decorrelation. I suggest the authors either analyze the time-index of Ca across the islet to specifically show wave propagation, or refer to this data appropriately.

8. While the islet cell type specific reporters coupled with calcium imaging represents a novel and exciting new technology for understanding zebrafish endocrine functionality, the authors utilize it along with three methods to address the same question; namely, do nerves regulate islet cell functional connectivity. Of the three, one method (photo-ablation; Figure 3), shows contradictory/weak results when compared to the other two and is thus not warranted as a main figure and could be moved to supplement.

9. In figure 3 (photo ablation of peri-islet nerves) the authors show no evidence that photo ablation does indeed lead to non-functional/disrupted nerve architecture. To address this, the authors could fix and stain for nerves directly after photo ablation to see if neuronal axons are still present.

10. While the findings of cx43 KO leading to impaired β cell functionality are novel, they have no direct tie to the previous 4 figures of the paper. The relationship to the nerves is supported by solid data, indeed, there is very, very little data to indicate nerve input plays a role here.

11. The authors state that δ cell activity is increased in the chronic and acute models of neuronal inhibition (BoTx and optogenetics), suggestion that nerves play an inhibitory role in δ cell function. This, however, cannot be proven directly by the current experiments performed by the authors. Furthermore, the current experiments cannot definitively say that the nerves play a direct role in regulating neuronal activity vs. bystander effect of modulating other surrounding cell types (i.e. vasoconstriction of vascularization within the islet).

12. The authors cannot attribute which branch of the peripheral nervous system (parasympathetic vs. sympathetic) are leading to functional disruptions upon inhibition. While botox is predominantly thought of as an inhibitor of cholinergic neurons, there is evidence that administration affects pre-ganglionic ganglia that activate sympathetic neurons. Furthermore, optogenetic inhibition of pancreatic nerves will potentially inhibit both sympathetic and parasympathetic nerves. Thus, determining which peripheral nerve subtype is important for functional signaling in pancreatic islet cell would be novel and beneficial to the field.

13. Does the BoTx affect the health of the zebrafish, which might affect the function of the pancreatic islet cells? This needs to be addressed.

14. While the authors also shed new light on the previously unclear role of gap-junctions in the zebrafish islet and β-cell coupling, it remains unclear how these results connect to the innervation phenotypes. Rather, it might have been more interesting to use Et(1121A:LAL4FF) driven GCaMP expression in peri-islet neurons to directly correlate neuronal with endocrine activity – in particular with respect to the suggested role of neural input in modulating δ cell function.

15. 109, Figure 2G: The data show no significant difference between LG – HG, suggesting that there is no glucose-induced response, same for Figure 3F and 4D (such a difference is only seen for the control in 5E ?). In addition: paired sample t-test rather than student's t-test should be used for comparing LG and HG (3F: repeated measures ANOVA?)

16. 109-111 "[…] indicating drastically altered synchronicity and wave propagation (Figure 2H-I)." – data in H show no difference in slopes between both groups, purpose/meaning of AOC is not explained and relevance appears unclear as there is no significant change.

17. 123-125: "[…] resulted in the hyperactivation of δ cells, […]" – Figure 2J-L: These conclusions need more detailed statistics, in particular as it is not clear which statistical tests were used – information about sample size, standard deviation and or confidence intervals, degrees of freedom, t-value are missing. (same for corresponding data in Figure 3-5).

18. 183-184: "From normalized single cell calcium traces, we observed blunted

glucose-induced coupling of β cell activity glucose-induced coupling" – see above: it seems this is the only data set supporting a glucose response in control animals?

19. 520: "[…]; p-values from ANOVA." P-values are from post-hoc test? Holm Sidak's multiple comparisons should be applied instead.

https://doi.org/10.7554/eLife.64526.sa1

Author response

Summary:

Overall, the reviewers are supportive of publication and study provides important novel insights into our understanding of the underlying mechanisms mediating coordinated pancreatic endocrine cell responses. However, they indicate the need for a number of important clarifications (particularly related to interpreting the connectivity analysis) and places where data presentation is not always clear. In addition, all 3 reviewers indicated that throughout the manuscript the authors' made overly speculative statements and there is some concern that the study does not provide evidence of direct influence of neuronal activity. Finally, some of the experiments require additional and/or alternative statistical analyses. While most of these revisions will not require additional experimentation, a small number of experiments are suggested that would help support the authors' conclusions.

We thank the reviewers for their supportive and constructive comments. We have conducted additional experiments and analyses and have made the recommended changes in toning down our language. The revised Figure 2 provides further support for direct neural regulation of a subset of pancreatic islet cells. We have addressed all of the reviewers’ comments as detailed below.

Essential revisions:

1. An important caveat in this study is that connectivity is used to refer to cells showing similar ca2+ elevations, whether calculated via correlation coefficients or fraction of time two neighboring cells are 'active'/'inactive'. This does not per se indicate if the two cells truly are directly interacting, or whether they are instead each interacting with some other factor. For instance, an increased correlation coefficient between β cells upon innervation could be due to enhanced gap junction coupling or paracrine communication between β cells or it could be due to neuronal activity driving the activity of the β cells. Some literature (e.g. Zhang et al., Biophys.J. 2008) would suggest the latter, not the former, and thus conflicts with the authors conclusions.

We thank the reviewers for this comment and have rephrased the sentences that implicated direct interactions. We have now also cited the suggested paper.

2. Related to this above point, in figures 2-4 the authors use one type of analysis for assessing homotypic interactions between β cells, but a separate analysis for assessing heterotypic interactions. Why do the authors not use a correlation based analysis between cell neighbors for heterotypic interactions? Would they observe changes if such correlation analysis is employed? (I see why they cannot use it across all cells).

We have now also included a correlation-based analysis for heterotypic interactions and presented the results alongside the fraction time analysis.

Also, if the 'connectivity' between two cell types changes (e.g. β-δ), then the apparent connectivity with the other cell type (α) would very likely change irrespective of whether any actual connection is present. For instance, it seems to me that changes in β cell correlation would change the apparent β-δ and β-α connectivity, irrespective of the actual link between these cell types. Δ-α 'connectivity' doesn't change under any of the experimental conditions which would support this alternative conclusion. How do the authors exclude this alternative conclusion?

While we cannot exclude the alternative conclusion completely, we think that the differential effects observed across the different conditions that all display β cell correlation changes support the presence of connectivity between heterotypic cell pairs. For instance, there is some evidence of δ-α connectivity changes in both chronic and acute pan-neural inhibition condition, but not in the acute, peri-islet ablation setting. In these ablation experiments, only α-β changes were concurrently observed for both the correlation and fraction time analysis.

Finally, are α, β and δ cell activity patterns themselves altered under each neuronal modulation and Cx43 deletion? E.g. the duration and amplitude of ca2+ transients. If so could these explain differences in apparent 'connectivity'?

We have now included the analysis of duration and amplitude of the calcium transients for the ablation condition. There doesn’t appear to be any significant differences in frequency, duration, or amplitude of calcium transients for β, δ, or α cells across the different conditions and upon ablation.

3. There are many overly speculative statements that need either toning down or including data that better supports the conclusion: e.g.

– Abstract: "δ cells receive innervation which coordinates its (the islet) output." Such an interpretation would rely on modulating neuronal-δ cell interactions specifically.

– Lines 123-124: hyper-activation of δ cells is a strong statement. Instead, it seems that δ cells are activated more than their neighboring β and α cells. But the change in δ cell activity is not especially strong. Is δ cell activity itself (% time elevated or spike height) different?

– Line 124: similarly, the authors did not measure somatostatin release so cannot say this is directly the result of δ cell activity.

– Line 192-194: the authors should combine their Cx43 with neural modulation to test if pancreatic innervation is indeed mediating the remaining intra-islet coordination.

We have toned down the language.

As noted by the reviewers, further analysis of δ cell activity did not show any significant difference in calcium peak frequency, height, or duration (Figure 3 – Figure Suppl. 3). We have therefore removed the claims of δ cell hyper-activation.

We thank the reviewers for this suggestion. While the proposed cx43 experiments would certainly be very interesting to carry out, it was not feasible given the COVID19 related and lab startup delays. We have therefore decided to remove the cx43 data.

4. In performing neuronal modulation, pan-neural chronic botox treatment (Figure 2) significantly disrupted both homotypic and heterotypic interactions; acute photo-ablation of peri-islet neurons (Figure 3) only disrupted homotypic interactions (although note my comment 1); and pan-neural acute inhibition of neurotransmitter release (Figure 4) significantly disrupted both homotypic and heterotypic interactions. How can the authors exclude the pan-neural disruptions affecting islet function independent of pancreas innervation, e.g. innervation affecting another organ and changing some circulating factor? Especially given both pan neural methods generate different findings than the peri-islet disruption.

We cannot exclude the possibility that the pan-neural disruption affects the pancreas indirectly, and have now added a sentence on this point.

5. Line 140-141: you cannot say there is a trend towards a decrease in α-active β active state based on a p of 0.06 within a set of data that contains 12 related comparisons. This leads me to ask if the authors have corrected for multiple comparisons in their statistical analysis? (for 2J-L, for 3I-K, for4G-I). Throughout, the authors should state in the figure captions what statistical analysis was performed for which comparison indicated.

Our study with the pan-neural ablation was under powered, and we have now increased the number of biological replicates. The change in α-β heterotypic interaction is statistically significant, both with the correlation analysis and fraction time analysis, suggesting that innervation does play a role in α-β interactions.

Information about the statistical analysis performed is now included in all the figure legends.

6. Figure 2 G: the change in Ravg (correlation) is marginally significant. The results refer to these changes as 'significantly disrupted', 'severe disruption', drastically altered' which needs toning down, with 'marginal' being a more apt term. Furthermore, in F the 'representative' correlation matrices are not congruent with the data in G. At the very least in G the data points corresponding to the example in F should be indicated, and better would be to include more representative matrices.

The language has been toned down.

The correlation matrices in panel F match the calcium traces in panel E. For panel G, we have now indicated the data points corresponding to the examples in panels E & F.

7. In figures 2-4 I do not follow the Ravg vs distance measurement. This is interpreted to be indicating β cell wave propagation. I do not see how the decline in relative correlation with distance would refer to wave propagation? This is really a decorrelation which is entirely expected for a system that is not perfectly coupled. Other phenomena than wave propagation could explain such decorrelation. I suggest the authors either analyze the time-index of Ca across the islet to specifically show wave propagation, or refer to this data appropriately.

We have rephrased our interpretation as recommended and removed all reference to wave propagation.

8. While the islet cell type specific reporters coupled with calcium imaging represents a novel and exciting new technology for understanding zebrafish endocrine functionality, the authors utilize it along with three methods to address the same question; namely, do nerves regulate islet cell functional connectivity. Of the three, one method (photo-ablation; Figure 3), shows contradictory/weak results when compared to the other two and is thus not warranted as a main figure and could be moved to supplement.

The photo-ablation experiments add a level of specificity to targeting pancreatic innervation. Although, we don’t yet know the exact identity of these neurons, more animals have since been analyzed to increase the power of our analysis.

9. In figure 3 (photo ablation of peri-islet nerves) the authors show no evidence that photo ablation does indeed lead to non-functional/disrupted nerve architecture. To address this, the authors could fix and stain for nerves directly after photo ablation to see if neuronal axons are still present.

To support the claim of disrupted islet nerve architecture upon photo-ablation of peri-islet neurons (now Figure 4), we have included quantification of islet nerve density following peri-islet neuron ablation in Figure 4 —figure supplement 1A-B.

10. While the findings of cx43 KO leading to impaired β cell functionality are novel, they have no direct tie to the previous 4 figures of the paper. The relationship to the nerves is supported by solid data, indeed, there is very, very little data to indicate nerve input plays a role here.

We thank the reviewer for this comment and have decided to remove the cx43 data from this manuscript.

11. The authors state that δ cell activity is increased in the chronic and acute models of neuronal inhibition (BoTx and optogenetics), suggestion that nerves play an inhibitory role in δ cell function. This, however, cannot be proven directly by the current experiments performed by the authors. Furthermore, the current experiments cannot definitively say that the nerves play a direct role in regulating neuronal activity vs. bystander effect of modulating other surrounding cell types (i.e. vasoconstriction of vascularization within the islet).

We have toned down the language. Additionally, and as suggested by the reviewers, we have now analyzed the activity of peri-islet neurons in parallel with islet cells (new Figure 2) and the data do suggest that a subset of δ cells may be directly modulated by nerve activity. We cannot exclude bystander effects and have included a statement to that effect in the discussion.

12. The authors cannot attribute which branch of the peripheral nervous system (parasympathetic vs. sympathetic) are leading to functional disruptions upon inhibition. While botox is predominantly thought of as an inhibitor of cholinergic neurons, there is evidence that administration affects pre-ganglionic ganglia that activate sympathetic neurons. Furthermore, optogenetic inhibition of pancreatic nerves will potentially inhibit both sympathetic and parasympathetic nerves. Thus, determining which peripheral nerve subtype is important for functional signaling in pancreatic islet cell would be novel and beneficial to the field.

We have previously demonstrated that at the stages when these experiments were carried out, innervation is only from the vagus nerve and have since found that sympathetic innervation does not enter the islet until after 120 hpf. We have now included this point in the results and plan to analyze effects of sympathetic nerves with appropriate transgenic lines.

13. Does the BoTx affect the health of the zebrafish, which might affect the function of the pancreatic islet cells? This needs to be addressed.

Pan-neural expression of BoTx does have other effects that may influence pancreatic islet function. As we have previously described (PMID: 29916364), these fish may also have defects in the development of δ cells. This point has now been mentioned in the results.

14. While the authors also shed new light on the previously unclear role of gap-junctions in the zebrafish islet and β-cell coupling, it remains unclear how these results connect to the innervation phenotypes. Rather, it might have been more interesting to use Et(1121A:LAL4FF) driven GCaMP expression in peri-islet neurons to directly correlate neuronal with endocrine activity – in particular with respect to the suggested role of neural input in modulating δ cell function.

We thank the reviewers for this comment. We have removed the cx43 data and added examples of correlation of neural activity with endocrine activity (new Figure 2). Notably, we found that δ cells displaying significant activity coupling to a peri-islet neuron had significantly higher activity coupling to other δ cells within the same islet in comparison to those that were not coupled to the neuron.

15. 109, Figure 2G: The data show no significant difference between LG – HG, suggesting that there is no glucose-induced response, same for Figure 3F and 4D (such a difference is only seen for the control in 5E ?). In addition: paired sample t-test rather than student's t-test should be used for comparing LG and HG (3F: repeated measures ANOVA?)

The statistics have now been updated and described in the figure legends. There is a significant difference between LG – HG in the control of the BoTx experiment (Figure 3G). However as noted by the reviewers, this glucose induced response is not observed in Figure 4F and 5D. The photo-inhibition/photo-ablation experiments had to be carried out 5 hours later than the BoTx experiment, due to the inability to photo-ablate the neurons at earlier stages. The ~20% rise in glucose level in the earlier stages (Figure 3- Figure Suppl. 1) had a significant effect on β cell correlation. The increased basal glucose level in the later stages (PMID: 23201900) likely led to less prominent induction of glucose, and thus a lesser effect on β cell correlation.

16. 109-111 "[…] indicating drastically altered synchronicity and wave propagation (Figure 2H-I)." – data in H show no difference in slopes between both groups, purpose/meaning of AOC is not explained and relevance appears unclear as there is no significant change.

This statement has now been removed since, as the reviewers pointed out, our analyses do not allow one to infer any changes in wave propagation.

17. 123-125: "[…] resulted in the hyperactivation of δ cells, […]" – Figure 2J-L: These conclusions need more detailed statistics, in particular as it is not clear which statistical tests were used – information about sample size, standard deviation and or confidence intervals, degrees of freedom, t-value are missing. (same for corresponding data in Figure 3-5).

We have removed this statement as further analysis of δ cell activity did not reveal any significant difference in calcium peak frequency, height, or duration (Figure 3 – Figure Suppl. 3). Figure legends have been updated.

18. 183-184: "From normalized single cell calcium traces, we observed blunted

glucose-induced coupling of β cell activity glucose-induced coupling" – see above: it seems this is the only data set supporting a glucose response in control animals?

The section on cx43 has been removed.

19. 520: "[…]; p-values from ANOVA." P-values are from post-hoc test? Holm Sidak's multiple comparisons should be applied instead.

Appropriate post-hoc tests were performed and described in the figure legends and methods.

https://doi.org/10.7554/eLife.64526.sa2

Article and author information

Author details

  1. Yu Hsuan Carol Yang

    1. Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    2. Exeter Centre of Excellence for Diabetes Research, Institute of Biomedical and Clinical Science, University of Exeter Medical School, Exeter, United Kingdom
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review and editing
    For correspondence
    Y.C.Yang@exeter.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6663-0302
  2. Linford JB Briant

    Oxford Centre for Diabetes, Endocrinology and Metabolism, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom
    Contribution
    Formal analysis, Software, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3619-3177
  3. Christopher A Raab

    Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  4. Sri Teja Mullapudi

    Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    Contribution
    Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3916-8148
  5. Hans-Martin Maischein

    Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  6. Koichi Kawakami

    Division of Molecular and Developmental Biology, National Institute of Genetics, Department of Genetics, SOKENDAI, Mishima, Japan
    Contribution
    Resources, Writing – review and editing
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9993-1435
  7. Didier YR Stainier

    Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    Contribution
    Funding acquisition, Resources, Writing – review and editing
    For correspondence
    Didier.Stainier@mpi-bn.mpg.de
    Competing interests
    Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0382-0026

Funding

Max Planck Society

  • Didier YR Stainier

Canadian Institutes of Health Research

  • Yu Hsuan Carol Yang

Human Frontier Science Program (LT000159/2015)

  • Yu Hsuan Carol Yang

EMBO (ALTF 773-2014)

  • Yu Hsuan Carol Yang

NIG-JOINT

  • Yu Hsuan Carol Yang

AMED

  • Koichi Kawakami

Wellcome Trust

  • Linford JB Briant

UKRI Expanding Excellence in England

  • Yu Hsuan Carol Yang

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

Acknowledgements

These studies were supported by funds from the Max Planck Society to DYRS. YHCY was supported by CIHR Postdoctoral Fellowships, an HFSP Long-Term Fellowship (LT000159/2015), an EMBO Long-Term Fellowship (ALTF 773-2014), NIG-JOINT funding, and UKRI Expanding Excellence in England. LJBB was supported by a Sir Henry Wellcome Postdoctoral Fellowship. KK was supported by an NBRP grant from AMED. We thank Arica Beisaw for the plasmid with P2A-mScarlet and James Johnson for critical reading of the manuscript.

Ethics

All zebrafish husbandry was performed under standard conditions in accordance with institutional (MPG) and national ethical and animal welfare guidelines (Proposal numbers: B2/1041, B2/Anz. 1007, B2/1218). All procedures conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.

Senior Editor

  1. Marianne E Bronner, California Institute of Technology, United States

Reviewing Editor

  1. Lori Sussel, University of Colorado Anschutz Medical Campus, United States

Reviewers

  1. Sarah M Knox, University of California, San Francisco, United States
  2. Dirk Meyer, University of Innsbruck, Innsbruck, Austria

Publication history

  1. Received: November 1, 2020
  2. Preprint posted: November 5, 2020 (view preprint)
  3. Accepted: April 3, 2022
  4. Accepted Manuscript published: April 4, 2022 (version 1)
  5. Version of Record published: April 13, 2022 (version 2)

Copyright

© 2022, Yang 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

  • 1,378
    Page views
  • 227
    Downloads
  • 2
    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)

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

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

  1. Yu Hsuan Carol Yang
  2. Linford JB Briant
  3. Christopher A Raab
  4. Sri Teja Mullapudi
  5. Hans-Martin Maischein
  6. Koichi Kawakami
  7. Didier YR Stainier
(2022)
Innervation modulates the functional connectivity between pancreatic endocrine cells
eLife 11:e64526.
https://doi.org/10.7554/eLife.64526

Further reading

    1. Cell Biology
    Swapneeta S Date et al.
    Research Article Updated

    Deciphering mechanisms controlling SNARE localization within the Golgi complex is crucial to understanding protein trafficking patterns within the secretory pathway. SNAREs are also thought to prime coatomer protein I (COPI) assembly to ensure incorporation of these essential cargoes into vesicles, but the regulation of these events is poorly understood. Here, we report roles for ubiquitin recognition by COPI in SNARE trafficking and in stabilizing interactions between Arf, COPI, and Golgi SNAREs in Saccharomyces cerevisiae. The ability of COPI to bind ubiquitin, but not the dilysine motif, through its N-terminal WD repeat domain of β′-COP or through an unrelated ubiquitin-binding domain is essential for the proper localization of Golgi SNAREs Bet1 and Gos1. We find that COPI, the ArfGAP Glo3, and multiple Golgi SNAREs are ubiquitinated. Notably, the binding of Arf and COPI to Gos1 is markedly enhanced by ubiquitination of these components. Glo3 is proposed to prime COPI–SNARE interactions; however, Glo3 is not enriched in the ubiquitin-stabilized SNARE–Arf–COPI complex but is instead enriched with COPI complexes that lack SNAREs. These results support a new model for how posttranslational modifications drive COPI priming events crucial for Golgi SNARE localization.

    1. Cell Biology
    Eva Kaulich et al.
    Research Article Updated

    Biological clocks are fundamental to an organism’s health, controlling periodicity of behaviour and metabolism. Here, we identify two acid-sensing ion channels, with very different proton sensing properties, and describe their role in an ultradian clock, the defecation motor program (DMP) of the nematode Caenorhabditis elegans. An ACD-5-containing channel, on the apical membrane of the intestinal epithelium, is essential for maintenance of luminal acidity, and thus the rhythmic oscillations in lumen pH. In contrast, the second channel, composed of FLR-1, ACD-3 and/or DEL-5, located on the basolateral membrane, controls the intracellular Ca2+ wave and forms a core component of the master oscillator that controls the timing and rhythmicity of the DMP. flr-1 and acd-3/del-5 mutants show severe developmental and metabolic defects. We thus directly link the proton-sensing properties of these channels to their physiological roles in pH regulation and Ca2+ signalling, the generation of an ultradian oscillator, and its metabolic consequences.