Intravital imaging of islet Ca2+ dynamics reveals enhanced β cell connectivity after bariatric surgery in mice

Bariatric surgery improves both insulin sensitivity and secretion in type 2 diabetes. However, these changes are difficult to monitor directly and independently. In particular, the degree and the time course over which surgery impacts β cell function, versus mass, have been difficult to establish. In this study, we investigated the effect of bariatric surgery on β cell function in vivo by imaging Ca2+ dynamics prospectively and at the single cell level in islets engrafted into the anterior eye chamber. Islets expressing GCaMP6f selectively in the β cell were transplanted into obese male hyperglycaemic mice that were then subjected to either vertical sleeve gastrectomy (VSG) or sham surgery. Imaged in vivo in the eye, VSG improved coordinated Ca2+ activity, with 90% of islets observed exhibiting enhanced Ca2+ wave activity ten weeks post-surgery, while islet wave activity in sham animals fell to zero discernible coordinated islet Ca2+ activity at the same time point. Correspondingly, VSG mice displayed significantly improved glucose tolerance and insulin secretion. Circulating fasting levels of GLP-1 were also increased after surgery, potentially contributing to improved β cell performance. We thus demonstrate that bariatric surgery leads to time-dependent increases in individual β cell function and intra-islet connectivity, together driving increased insulin secretion and diabetes remission, in a weight-loss independent fashion. Significance Statement Used widely to treat obesity, bariatric surgery also relieves the symptoms of type 2 diabetes. The mechanisms involved in diabetes remission are still contested, with increased insulin sensitivity and elevated insulin secretion from pancreatic β cells both implicated. Whilst the speed of remission – usually within a few days – argues for improvements in β cell function rather than increases in mass, a direct demonstration of changes at the level of individual β cells or islets has been elusive. Here, we combine vertical sleeve gastrectomy with intravital imaging of islets engrafted into the mouse anterior eye chamber to reveal that surgery causes a time-dependent improvement in glucose-induced Ca2+ dynamics and β cell - β cell connectivity, both of which likely underlie increased insulin release.

Bariatric surgery improves both insulin sensitivity and secretion in type 2 diabetes. However, 2 these changes are difficult to monitor directly and independently. In particular, the degree and 3 the time course over which surgery impacts β cell function, versus mass, have been difficult 4 to establish. In this study, we investigated the effect of bariatric surgery on β cell function in 5 vivo by imaging Ca 2+ dynamics prospectively and at the single cell level in islets engrafted into 6 the anterior eye chamber. Islets expressing GCaMP6f selectively in the β cell were 7 transplanted into obese male hyperglycaemic mice that were then subjected to either vertical 8 sleeve gastrectomy (VSG) or sham surgery. Imaged in vivo in the eye, VSG improved 9 coordinated Ca 2+ activity, with 90% of islets observed exhibiting enhanced Ca 2+ wave activity 10 ten weeks post-surgery, while islet wave activity in sham animals fell to zero discernible 11 coordinated islet Ca 2+ activity at the same time point. Correspondingly, VSG mice displayed 12 significantly improved glucose tolerance and insulin secretion. Circulating fasting levels of 13 GLP-1 were also increased after surgery, potentially contributing to improved β cell 14 performance. We thus demonstrate that bariatric surgery leads to time-dependent increases 15 in individual β cell function and intra-islet connectivity, together driving increased insulin 16 secretion and diabetes remission, in a weight-loss independent fashion. 17

19
Significance Statement: 20 Used widely to treat obesity, bariatric surgery also relieves the symptoms of type 2 diabetes. 21 The mechanisms involved in diabetes remission are still contested, with increased insulin 22 sensitivity and elevated insulin secretion from pancreatic β cells both implicated. Whilst the 23 speed of remissionusually within a few daysargues for improvements in β cell function 24 rather than increases in mass, a direct demonstration of changes at the level of individual β 25 cells or islets has been elusive. Here, we combine vertical sleeve gastrectomy with intravital 26 imaging of islets engrafted into the mouse anterior eye chamber to reveal that surgery causes 27 a time-dependent improvement in glucose-induced Ca 2+ dynamics and β cellβ cell 28 connectivity, both of which likely underlie increased insulin release. 29

Introduction 1
An estimated 30 million individuals in the US (9.4 % of the population) have diabetes (1), with 2 ~90% of cases thought to be Type 2 Diabetes (T2D), while in the United Kingdom it is 3 predicted that by 2025 more than five million people will be diagnosed with the disease (2). In 4 response to this epidemic, an abundance of pharmacological, dietary, exercise and 5 behavioural interventions have been deployed but often focus on T2D management rather 6 than long-term disease resolution (3, 4). Several clinical trials have now reported that bariatric 7 surgery, a group of gastrointestinal procedures originally developed to aid weight loss, 8 improves long-term glycaemia more effectively than caloric restriction or medical intervention 9 (5-7). 10 Numerous studies (8-12) have attempted to unravel the mechanisms through which blood 11 glucose control is improved post-operatively. One hypothesis to explain post-bariatric T2D 12 remission is that it results from the increased release of incretins from the gut, such as the 13 gastrointestinal insulin-stimulating hormone Glucagon-like Peptide 1 (GLP-1), as upregulated 14 postprandial levels have been reported following bariatric surgery (13-15). Preclinical and 15 clinical data have shown that bariatric surgery improves both hepatic and peripheral insulin 16 sensitivity, as well as increases in insulin secretion (16)(17)(18)(19). However, the exact mechanisms 17 through which surgery impacts the β cell, including the identity of all the extra-pancreatic 18 signals involved, and the relative importance of changes in β cell function and mass, have 19 remained elusive. Nonetheless, the rapid (hours-days) reversal of diabetes in human subjects 20 treated with bariatric surgery (20, 21) has provided powerful evidence that an improvement of 21 β cell function plays an important, and possibly the dominant, role in increasing pancreatic 22 insulin output. 23 A critical limitation in investigating β cell function in living humans or preclinical models is that, 24 in the absence of robust in vivo imaging technologies (22), function must rely mainly on indirect 25 measurements of circulating insulin or C-peptide. These approaches preclude any quantitation 26 of changes over time, a detailed examination of function at the level of single β cells, or the 27 connections between them. The latter has become an important issue since we (23) and 28 others (24) have reported that weaker intercellular connections, and the loss of highly 29 connected cells, that can often initiate Ca 2+ waves (sometimes referred to as "hubs"), underlie 30 the loss of insulin secretion observed in response to challenges associated with diabetes 31 gluco(lipo)toxicity, low inflammation level, etc. (23, 25, 26). However, untangling these 32 functional changes from alterations in β cell mass in vivo is challenging, since the latter can 33 only reliably be determined post-mortem via pancreatic biopsies, and thus at a single time 34 point. 35 In an effort to overcome these limitations, the present study aimed to investigate the effect of 36 Vertical Sleeve Gastrectomy (VSG) on pancreatic β cell function in mice, by transplanting 37 "reporter" islets in the anterior chamber of the eye. This approach was established by Berggren 38 and colleagues (27) and has recently been developed by ourselves (26) to assess coordinated 39 islet behaviour in vivo. Importantly, this technique has allowed us to image Ca 2+ dynamics 40 recursively, in the same islet, over time and with near single cell resolution, following surgery.

41
We show that VSG increases β cell Ca 2+ dynamics within eight weeks post-surgery when 42 compared to pre-operative baseline and a sham operated group. Moreover, we demonstrate 43 that VSG increases the number and strength of β to β cell connections at ten weeks after 44 surgery. These changes were associated with increased fasting levels of GLP-1, suggesting 45 that enhanced incretin production may contribute to postoperative improvements in β cell 1 performance. 2 3 Results 4 5 Vertical Sleeve Gastrectomy improves glucose tolerance 6 Our experimental protocol is summarized in Figure 1A. In brief, mice were placed on a high 7 fat high sucrose diet (HFHSD), at eight weeks of age, eight weeks before sham or vertical 8 sleeve gastrectomy (VSG) surgery (week 0). This protocol led to fasting hyperglycaemia, 9 indicative of β cell decompensation and defective insulin secretion, as expected (28). 10 Ins1Cre:GCaMPf fl/fl islets were isolated from donor mice and transplanted at week (-4). 11 Baseline islet Ca 2+ dynamics were imaged at week (-1 (OGTT) at post-operative week 8 (Fig. 1C) and intraperitoneal glucose tolerance test (IPGTT) 17 four and ten weeks post operatively (p<0.01 at min. 30, 60, 90 min, Fig. 1D, 1E respectively). 18 Strikingly, in all tolerance tests performed on VSG-treated mice, glucose peaked at 15 min. 19 post glucose injection (3g/kg) and dropped to baseline levels within 60 min. by week eight, 20 and near baseline levels at week ten. In contrast, in sham-operated mice, glucose peaked at 21 30 min. and did not fully recover within the first 2 h of measurement. 22

23
Vertical Sleeve Gastrectomy improves insulin secretion and sensitivity but does not increase 24 β cell mass 25 In order to understand the marked increase in the rate of glucose clearance in mice that had 26 undergone VSG, we measured insulin secretion in vivo as a response to an IP glucose load 27 (3g/kg). Insulin secretion was increased significantly in VSG versus sham mice as early as 28 four weeks post operatively (Fig. 1F, with the observed peak at 15 min. almost three-fold 29 higher compared to sham mice (p<0.05). VSG mice were also significantly more insulin 30 sensitive when compared to sham mice, as assessed by intraperitoneal insulin tolerance test 31 (ITT, 1.5U/kg) (p<0.01, Fig. 1 G). However, pancreatic β cell mass was not increased in the 32 VSG group relative to sham controls (Supp. Fig. 1A, 2). Notably, the ratio of α to β cell mass 33 was significantly higher in the VSG group, yet α cell mass was not significantly increased 34 (Supp Fig. 1B, C, 2). 35

36
Vertical Sleeve Gastrectomy enhances GLP-1 secretion 37 To assess whether enhanced incretin release may contribute to the euglycemic effect of VSG 38 we observed during IPGTT and OGTT, we measured plasma GLP-1 levels during fasting and 39 15 min. following an orally administered glucose load (3g/kg) (Fig. 1C). Fasting GLP-1 was 1 significantly higher in the VSG group, when compared to sham (Fig. 1H). Moreover, whilst 2 glucose failed to increase GLP-1 levels significantly in the sham group, a highly significant 3 increase in response to glucose gavage was observed in VSG-treated animals (Fig. 1H). 4 Significantly lower glucose levels were apparent in VSG-treated versus sham-treated mice, 5 both fasting and following glucose gavage (Fig. 1I). 6 7 β cell Ca 2+ dynamics are enhanced following Vertical Sleeve Gastrectomy 8 In order to explore changes in β cell function after surgery, we monitored intracellular Ca 2+ 9 changes prospectively and in the same islets by confocal imaging of the anterior eye chamber 10 (26, 27). Ca 2+ increases, measured at ambient blood glucose concentrations in the range 11 12.5±0.7 mmol/L for both VSG-treated and sham mice, which occurred at a single or multiple 12 site across the islet but did not advance across the islet, were defined as "oscillations". 13 Increases that had a defined site of origin but did not spread across the full width of the imaged 14 plane, were defined as "partial" waves ( Fig. 2Bi, Supp. Mov. 1D). Those increases spreading 15 across the whole islet were termed "waves" (Fig. 2Ai, 2Bii Supp. Mov. 1A, E). If the latter wave 16 type was recurrent, we defined the behaviour as a "super wave" (Fig 2Biii, Supp. Mov. 1F).

17
As illustrated in Fig. 2A, when imaged 0, 4 and 10 weeks after surgery, islets in sham-operated 18 animals displayed a progressive loss of Ca 2+ dynamics, as defined by the frequency and type 19 of waves. Thus, when imaged at 0 weeks ( Fig. 2Ai), wave behaviour (beginning at the bottom 20 right; red area) area moved rapidly across the areas identified in yellow and blue. Comparable 21 behaviour was seen at 4 weeks, with a similar site of origin of the wave (Fig. 2Aii) but was lost 22 at 10 weeks post sham surgery, even though there was no significant weight difference 23 between the two groups ( Fig. 2Aiii, Supp. Mov. 1C). 24 In contrast, islets implanted into mice subject to VSG displayed sustained or gradually 25 improving Ca 2+ dynamics following surgery. Thus, the islet shown in Fig. 2Bi initially showed 26 partial wave activity but progressed to full wave activity by week 4 (Fig. 2Bii, Supp. Mov. 1D) 27 and to super wave by week 10 (Fig. 2Biii, Supp. Mov. 1F). A similar progression was seen for 28 eight islets in three separate mice subjected to VSG (Fig. 3A, Supp. Fig. 3), whilst in six islets 29 in three sham-operated mice a decline in behaviour was apparent after surgery ( Fig. 3A).

30
Remarkably, almost all islets transplanted into VSG animals displayed either wave or 31 superwave behaviour by week eight, even if VSG-treated animals did not display further 32 weight loss. This is significantly higher when compared to sham mice at the same timepoint 33 (p=0.02) (Fig. 3A). By week 10, the activity of all sham-transplanted islets dropped to almost 34 zero (p=0.004) (Fig. 3A). Mean wave front velocity, a measure of the speed of the wave 35 calculated by distance (μm) divided by time (sec), across the islet was not different between 36 groups at any time point explored. Similarly, no differences were apparent between wave 37 velocities for the different wave types in either VSG or sham operated (Fig. 3B). Coordinated activity of β cells is a feature of the healthy islet, and is likely to be important for 41 the regulation of pulsatile insulin secretion (29). As shown in Fig. 4A and B, Pearson 42 correlation analysis revealed no differences in apparent connectivity at week 0 (prior to 43 surgery), whereas a progressive decline in connectivity was observed in the sham group. The 1 number of connected cells (Fig. 4B), or the mean connectivity strength (R) (Fig. 4C, D) 2 remained relatively constant in the VSG group, such that by week 10 these islets displayed 3 significantly greater connectivity than the sham group (Fig. 4B). In summary, glucose-related 4 Ca 2+ signalling in VSG mice was characterized by higher magnitude and higher sensitivity to 5 glucose when compared with sham mice, suggesting changes in glucose metabolism in the 6 islets following VSG. 7 8 Discussion 9 Using an intravital imaging approach developed in recent years to monitor islet function in vivo 10 (26, 27), we provide here evidence that VSG causes a dramatic improvement in β cell Ca 2+ 11 dynamics, a useful assay of normal cellular function and proxy for insulin secretion (22, 30, 12 31). The use of such an approach addresses the challenges in dissecting the relative 13 importance of the actions of bariatric surgery in changes observed in: (a) pancreatic insulin 14 output versus peripheral insulin sensitivity, (b) β cell function versus mass, and (c) the time 15 courses of changes, post-surgery. 16 Critically, we demonstrate that at similar, stimulatory glucose concentrations, islet Ca 2+ 17 dynamics and connectivity are dramatically increased in VSG versus sham-operated animals. 18 Our data provide the first evidence we are aware of that alterations in β cell function occur 19 both at the level of individual cells and across the islet ensemble after surgery, and are thus 20 likely to play a pivotal role in improving insulin output. Changes in both β cell identity, reflecting 21 altered gene expression (32-34), and in coordinated β cell activity across the islet, are 22 important features of T2D (25, 35). The normalization of either thus presents an attractive 23 therapeutic route towards improving insulin secretion in this disease. Importantly, whilst 24 several studies have demonstrated changes in islet gene expression in rodent models related 25 to hyperglycaemia and diabetes progression, such as obese diabetic (ZDF) rats and HFHSD 26 mice (36, 37), few have examined the potential for reversing these changes as a therapeutic 27 strategy (37, 38). 28 Central to the present study has been the use of VSG in obese mice as a model of human 29 bariatric surgery (39, 40). Roux-en-Y Gastric Bypass (RYGB) and VSG are routinely deployed 30 as an approach to treat human obesity, and both cause similar rates of T2D remission within 31 the first post-operative year in man (41). In mice, VSG leads to initial rapid weight loss followed 32 by a weight regain, unlike RYGB, but sustains improved glucose tolerance while offering a 33 more tractable approach, with lower mortality (42, 43). Importantly, our study had a ten-week 34 post-operative follow up and, by week eight, there was no significant weight difference 35 between sham and VSG group. This allowed us to separate marked improvements in insulin 36 secretion from significant weight loss without the need to pair-feed the sham group (43). 37 Moreover, it corresponds with our previous findings in lean VSG-treated mice that 38 demonstrated no weight difference when compared to sham mice at four weeks post-39 operatively, yet displayed improved glucose tolerance and corresponding insulin secretion 40 curves during an IPGTT (44). Insulin tolerance tests at week eight demonstrated that VSG 41 mice had improved insulin sensitivity, an effect previously attributed in humans to rapid and 42 significant enhancement of post-operative hepatic insulin clearance (45). It has recently been 43 suggested that, in animal models of surgery, hepatic insulin clearance is related to peripheral, 1 rather than hepatic, insulin sensitivity (46). 2 Accelerated glucose clearance in VSG-treated animals was accompanied by increased insulin 3 secretion in response to glucose at 15 and 30 min., consistent with previous studies using 4 VSG models (39, 47, 48). The fact that the insulin response to IPGTT was equally robust 5 suggests that this effect is not solely due to an increased spike in blood glucose associated 6 with elevated gastric emptying rates and upregulated glucose absorption, as has been 7 previously postulated (9, 49). Increased insulin secretion in the face of lower plasma glucose 8 demonstrates enhanced β cell glucose sensitivity, consistent with cell-autonomous changes 9 in islet function, alterations in circulating levels of other regulators of secretion, or an increase 10 in β cell number. Analysis of the endogenous pancreatic β cell mass showed no increase in 11 VSG versus sham-operated mice, pointing to a functional change rather than a change in 12 endogenous β cell mass, as underlying increased insulin output. Moreover, and though this 13 could not be quantitated accurately due to the lack of focal distances stacking data, we saw 14 no evidence for a change in the β cell mass of islets engrafted into the eye. This is in line with 15 previous findings demonstrating that there is no islet hyperplasia or increased β cell turnover 16 following bariatric surgery in humans or rats with obesity (50, 51). These findings contrast 17 other studies that have reported -over similar times scales-increasing (52-54) or decreasing 18 (48, 55) β cell mass differences, which may reflect pre-operative metabolic state or other 19 factors. 20 Given many reports of increased GLP-1 release after bariatric surgery in both humans (56) 21 and rodents (42, 57), here we explored levels of both circulating fasting and post-glucose 22 gavage GLP-1 levels. Importantly, the peak in GLP-1 following oral gavage did not differ 23 between sham and VSG mice, indicating that an enhanced insulinotropic effect of the incretin 24 is unlikely to explain dramatic increase in insulin secretion observed. Furthermore, enhanced 25 insulin secretion was seen in mice treated with VSG even during IPGTT, where the stimulation 26 of GLP-1 secretion is negligible. 27 Interestingly, VSG-treated mice did display significantly higher circulating GLP-1 levels under 28 basal (fasting) conditions. Apart from increasing glucose-stimulated insulin secretion and 29 enhancing insulin gene transcription (58, 59), GLP-1 also inhibits β cell apoptosis in animal 30 models of diabetes (60, 61). Thus, although the underlying mechanisms remain unclear, 31 increased basal GLP-1 levels might provide a partial explanation for the enhanced responses 32 to glucose and the euglycemic effects observed after surgery. functional connectivity between β cells before and after VSG or sham surgery. The percentage 2 of significantly-connected cell pairs and correlation coefficient decreased substantially in the 3 sham group at week ten, while in the VSG group these parameters remained stable for the 4 duration of the study. We would note that hub/follower behaviour (i.e. the existence of a "power 5 law" in the degree of connectedness) (23) was not readily apparent in the present study. More 6 rapid acquisition rates are likely to be needed to reveal such a hierarchy. Furthermore, we 7 note that wave-like behaviour is more often apparent in the islet in vivo using GCaMP6f as the 8 Ca 2+ sensor (26) than in some of our own and others' earlier studies (23, 25) using entrapped, 9 synthetic Ca 2+ probes. Nonetheless, clusters of apparent "leader" β cells, corresponding to 10 the point at which a rise in Ca 2+ was first observed at the beginning of a wave, were easily 11 identified in many cases (e.g. Fig. 2A), and these have previously been reported (23) to 12 correspond to the hub cell population. Interestingly, the origin of the waves was similar within 13 a given islet assayed several weeks apart, indicating that the cells which initiate them (leaders) 14 represent a stable population, at least over the time frame (< 10 weeks) studied here. 15 Taken together, our results indicate that bariatric surgery improves glycaemic control at least 16 partially by maintaining: (a) functional β cell identity and (b) coordinated activity across the 17 islet. A possible explanation for our data may be that the improvement in islet function follows 18 the improvement in glycaemia. However, since insulin sensitivity was barely altered by VSG, 19 it is unclear whether extrapancreatic events could be the drivers for improved islet function 20 and insulin output. postmortem and showed that the percentage of islets displaying Ca 2+ oscillations in response 28 to glucose was enhanced 2.2-fold in the VSG group, indicating increased islet glucose 29 sensitivity after surgery. In addition, VSG altered the islet transcriptome, affecting genes 30 involved in insulin secretion and Ca 2+ signaling (43). However, these earlier studies were 31 cross-sectional in nature, and as such did not explore the apparent reactivation in vivo of 32 individual islets and β cells in the living animal, as described here. 33 In conclusion, our findings provide further evidence for the protective effect of bariatric surgery 34 in T2D, irrespective of weight loss, and demonstrate direct effects on β-cell function and 35 coordination in the living animal. Future challenges are to understand more fully the 36 mechanisms through which these changes are affected at the paracrine, endocrine and 37 cellular levels. New Brunswick, NJ) ad libitum to induce obesity and diabetes. Four weeks after the start of 7 this diet, the animals underwent islet transplantation into the anterior chamber of the eye of 8 genetically modified islets expressing GCaMP6f to allow for intravital measurements of 9 cytosolic Ca 2+ . Four weeks after islet transplantation, the animals underwent either a vertical 10 sleeve gastrectomy or a sham surgery as described below. 11 Ins1Cre:GCaMPf fl/fl mice, used as donors for islet transplantation were generated by crossing 12 crossed Ins1Cre mice (provided by J Ferrer, this Department) to mice that express GCaMP6f 13 downstream of a LoxP-flanked STOP cassette (The Jackson Laboratory, stock no. 028865). 14 Islets donated from either sex were used for transplantation. 15 Islet transplantation into the anterior chamber of the mouse eye (ACE) -Pancreatic islets were 16 isolated and cultured as described previously (75). For transplantation, 10-20 islets were 17 aspirated with a 27-gauge blunt eye cannula (BeaverVisitec, UK) connected to a 100ul 18 Hamilton syringe (Hamilton) via 0.4-mm polyethylene tubing (Portex Limited). Prior to surgery, 19 mice were anaesthetised with 2-4% isoflurane (Zoetis) and placed in a stereotactic frame to 20 stabilise the head. The cornea was incised near the junction with the sclera, being careful not 21 to damage the iris. Then, the blunt cannula, pre-loaded with islets, was inserted into the ACE 22 and islets were expelled (average injection volume 20 µl for 10 islets). Carprofen (Bayer, UK) 23 and eye ointment were administered post-surgery. 24 Vertical Sleeve Gastrectomy -Three days before bariatric or sham surgery, animals were 25 exposed to liquid diet (20% dextrose) and remained on this diet for up to four days post 26 operatively. Following this, mice were returned to high fat/high sucrose diet until euthanasia 27 and tissues harvested ten weeks post bariatric surgery. Anaesthesia was induced and 28 maintained with isoflurane (1.5-2%). A laparotomy incision was made, and the stomach was 29 isolated outside the abdominal cavity. A simple continuous pattern of suture extending through 30 the gastric wall and along both gastric walls was placed to ensure the main blood vessels were 31 contained. Approximately 60% of the stomach was removed, leaving a tubular remnant. The 32 edges of the stomach were inverted and closed by placing two serosae only sutures, using 33 Lembert pattern. The initial full thickness suture was subsequently removed. Sham surgeries 34 were performed by isolating the stomach and performing a 1 mm gastrotomy on the gastric 35 wall of the fundus. All animals received a five-day course of SC antibiotic injections 36 (Ciprofloxacin 0.1mg/kg). 37 In vivo Ca 2+ imaging of Ins1Cre:GCaMPf fl/fl islets in the ACE -A minimum of four weeks was 38 allowed for full implantation of islets before imaging. Imaging sessions were performed as 39 previously described (26) with the mouse held in a stereotactic frame and the eye gently 40 retracted, with the animal maintained under 2-4% isoflurane anaesthesia. All imaging 41 experiments were conducted using a spinning disk confocal microscope (Nikon Eclipse Ti, 42 Crest spinning disk, 20x water dipping 1.0 NA objective). The signal from GCaMP6f 43 fluorophore (ex. 488 nm, em. 525±25 nm) was monitored in time-series experiments for up to 44 20 min. at a rate of 3 frames/ sec. Ca 2+ traces were recorded for three min, with a mean blood 1 glucose reading (across six islets in three separate animals per group) of 12.5±0.7. mmol/L. 2 Islets were continuously monitored, and the focus was manually adjusted to counteract 3 movement. Animals were imaged 3 days prior to Vertical Sleeve Gastrectomy (baseline) and 4 then at four, eight and ten weeks post-operatively. 5 Glucose Tolerance Tests -Mice were fasted overnight (total 16 h) and given free access to 6 water. At 0900, glucose (3 g/kg body weight) was administered via intraperitoneal injection or 7 oral gavage. Blood was sampled from the tail vein at 0, 5, 15, 30, 60 and 90 min. after glucose 8 administration. Blood glucose was measured with an automatic glucometer (Accuchek; 9 Roche, Burgess Hill, UK). antibody. Samples were mounted on glass slides using VectashieldTM (Vector Laboratories, 28 USA) containing DAPI. Images were captured on a Zeiss Axio Observer.Z1 motorised inverted 29 widefield microscope fitted with a Hamamatsu Flash 4.0 Camera using a Plan-Apochromat 30 206/0.8 M27 air objective with Colibri.2 LED illumination. Data acquisition was controlled with 31 Zeiss Zen Blue 2012 Software. Fluorescence quantification was achieved using Image J 32 (https://imagej.nih.gov/ij/). Whole pancreas was used to quantitate cell mass. 33 Statistical Analysis -Data were analysed using GraphPad PRISM 7.0 software. Significance 34 was tested using unpaired Student's two-tailed t-tests with Bonferroni post-tests for multiple 35 comparisons, or two-way ANOVA as indicated. P<0.05 was considered significant and errors 36 signify ± SEM. 37      ii.

Pearson (R)-based connectivity and correlation analyses -Correlation analyses between the
iii. i. ii. iii.