In vitro pancreatic islet cluster expansion facilitated by hormones and chemicals

Tissue regeneration, such as pancreatic islet tissue propagation in vitro, could serve as a promising strategy for diabetes therapy and personalized drug testing. However, such a protocol has not been realized yet. Propagation could be divided by two steps, which are: (1) expansion in vitro and (2) repeat passaging. Even the in vitro expansion of the islet has not been achieved to date. Here, we describe a method to enable the expansion of islet clusters isolated from pregnant mice or wild-type rats by employing a combination of specific regeneration factors and chemical compounds in vitro. The expanded islet clusters expressed insulin, glucagon and somatostatin, which are markers corresponding to pancreatic β cells, α cells and δ cells, respectively. These different types of cells grouped together, were spatially organized and functioned similarly to primary islets. Further mechanistic analysis revealed that forskolin in our recipe contributed to renewal and regeneration, whereas exendin4 was essential for preserving islet cell identity. Our results provide a novel method for the in vitro expansion of islet clusters, which is an important step forward in developing future protocols and medium used for islet tissue propagation in vitro. Such method is important for future regenerative diabetes therapies and personalized medicine using large amounts of pancreatic islets derived from the same person.

compared to the pregnant mouse islets (Fig. 3g). In response to high glucose stimulation, both 140 other labs and our lab have shown that UCN3 serves an endogenous paracrine factor secreted 141 by pancreatic β cells to stimulate endogenous somatostatin secretion from pancreatic δ 142 cells 7,10,24 . We therefore stimulated the expanded islet clusters with both glucose and UCN3 to 143 minimize the amount required for islet cluster usage. The in vitro expanded islet clusters 144 displayed significantly more somatostatin release in response to glucose and UCN3 145 stimulation (Fig. 3h). Moreover, the expanded islet clusters secreted more glucagon in 146 response to 25mM Arginine stimulation, hallmarking the normal function of pancreatic islet α 147 cells, which is similar to the pregnant mouse islets (Fig. 3i). These results confirmed the 148 functional integrity of the in vitro expanded islet clusters isolated from pregnant mice. 149 Although the expanded pancreatic islet clusters showed similar SST and Glucagon 150 secretion compared to the primary isolated islets in response to specific physiological 151 stimulations, it is worth to note that their insulin secretion in response to the combined 152 stimulation of glucose and GLP-1 is significantly weakened. Some extent degeneration during 153 the islet expansion may occur during current in vitro expansion condition, which awaits for 154 further investigation and methods optimization in our future work. 155 156

Expansion of rat islet clusters in vitro 157
We then examined the expansion of the dispersed rat islet single cells and clusters in PIEM. 158 Similar to those from pregnant mice, approximately 5~15% of the rat pancreatic islet clusters 159 were able to expand from a surface area of 5,000-8,000 µm 2 (200~350 cells) to a surface area 160 of 20,000-25,000 µm 2 (1,000~1,900 cells), whereas dispersed rat islet single cells showed no 161 such expansion ability (Fig. 4a, b). Furthermore, qRT-PCR revealed that the expanded rat islet 162 clusters maintained the same expression of the α cell markers Gcg and Mafb as the primary 163 rat islets but showed significantly higher expression of the β cell markers Ins-1,  Mafa and the δ cell marker Sst (Fig. 4c). 165 166 Increased gene expression related to dedifferentiation, pluripotency and proliferation in 167 expanded islet clusters 168 We next examined whether the expanded islet clusters increased the expression level of 169 genes functionally associated with cell proliferation, renewal and regeneration, which are 170 important factors for in vitro regeneration. Importantly, significantly higher expression of 171 Ki67, Ccnd1 and Pcna was found in expanded islet clusters than in islets derived from 172 pregnant mice or wild-type rats, highlighting the increased proliferation ability of the 173 expanded islet clusters (Fig. 5a, d). Moreover, significantly higher Nanog and Sox9 174 expression were found in expanded islet clusters derived from both pregnant mice and rats 175 (Fig. 5b, e). These results provide putative explanations accounting for the better proliferation 176 and renewal abilities of isolated islet clusters than integral islets. 177 We suspected that these islets gained proliferation and pluripotency due to dedifferentiation. 178 Thus, we examined dedifferentiation markers in expanded islet clusters and compared them 179 with those in isolated pancreatic islets. Interestingly, the expanded islet clusters showed 180 significantly higher expression of Ngn3 and Hlxb9, markers that characterize pancreatic islet 181 progenitor cells (Fig. 5c, f). Although the expanded islet clusters derived from pregnant mice 182 showed higher expression of Gata6, they showed decreased expression of Gata4, both of 183 which are pancreatic progenitor markers (Fig. 5c, f). The protein expression of KI67 and 184 SOX9 in expanded islet clusters were confirmed by immunofluorescence, higher than in 185 primary islets ( Supplementary Fig. S2a-d). Similarly, the expanded islet clusters derived from 186 rats did not display significant pancreatic progenitor characteristics, as they showed decreased 187 expression of Gata6 (Fig. 5c, f). The data suggested that these expanded islet clusters 188 underwent one-step dedifferentiation towards islet progenitor cells, but they did not ultimately 189 reach the pancreatic progenitor cell stage. Taken together, this one-step dedifferentiation of 190 isolated islet clusters cultured in PIEM towards pancreatic islet progenitor cells contributed to 191 the gain of cell proliferation, renewal and regeneration functions of expanded islet clusters. 192

Essential role of FSK and exendin4 in PIEM 194
During our formula component screening and recipe formation for PIEM, we identified that 195 FSK and exendin4 are both required for robust islet cluster expansion in vitro. It is known that 196 exendin4 induced cAMP accumulation after it activates GLP-1R and downstream Gs 197 proteins 25 . FSK is known as a cAMP agonist through direct binding to adenyl cyclase 26 . 198 Moreover, recent reports indicate that FSK is required for proliferation of liver organoids in 199 addition to A83-01 27 . Interestingly, removing FSK or exendin4 from PIEM has different 200 effects on the gene expression profiles of islet identity, proliferation, renewal and regeneration 201 markers. Removing FSK from PIEM had partial effects on islet identity and cell proliferation 202 markers, including decreased expression of Mafa, Sst and Ccnb1 (Fig. 6a, b). In particular, 203 FSK was required for the expression of cell renewal and regeneration markers, including 204 Nanog and Sox17 (Fig. 6c). In contrast, exendin4 was essential for islet identity and 205 proliferation because it maintained the expression of the pancreatic β cell markers Ins-1, Pdx1 206 and Mafa and the proliferation markers Ki67, Ccnb1 and Cdk4 (Fig. 6a, b). In particular, 207 exendin4 was required for the expression of Ngn3 and Hxlb9, two islet progenitor markers 208 (Fig. 6d). These results indicated that FSK in PIEM renders cell renewal and regeneration 209 ability to isolated islet clusters, whereas exendin4 is essential for maintaining islet β cell 210 identity, proliferation and dedifferentiation. The in vitro expansion of primary pancreatic β cells or pancreatic islet tissues has not been 215 reported before. Here, we developed a medium recipe, which we named PIEM, that enabled 216 the proliferation of dispersed islet clusters isolated from pregnant mice or wild-type rats in 217 vitro (Fig. 7). Isolated islet clusters normally grow from an initial surface area of 5,000-9,000 218 µm 2 to a surface area of 20,000-30,000 µm 2 , with an estimated 5~8-fold increase in cell 219 number. Compared to primary pancreatic islets, expanded cell clusters have similar or higher 220 mRNA expression levels of insulin, glucagon and somatostatin, which are markers of 221 pancreatic β cells, α cells and δ cells, respectively. Expanded islet clusters demonstrated 222 normal insulin and somatostatin secretion in response to physiological stimulation and normal 223 glucagon secretion in response to amino acids, suggesting that the clusters recapitulate 224 specific functions of pancreatic islets. Immunostaining further confirmed that the cells 225 expressing insulin, glucagon and somatostatin were grouped together and spatially organized 226 in expanded islet clusters similar to primary islets. These data indicated that the expanded 227 islet clusters behave similar to an in vitro organoid, thus serving as an initial step for future 228 islet propagation methods developing in vitro. The ultimate goal of developing efficient 229 methods for propagation islets in vitro is to fulfil the demand for regenerative diabetes 230 therapy and personalized drug tests. 231 It is worth noting that we observed the expansion of islet clusters isolated from only 232 pregnant mice and wild-type rats in PIEM, not those isolated from wild-type mice. Pregnant 233 mice are well known for their increased pancreatic β cell proliferation 28-31 , which provides a 234 useful starting point for testing medium recipes. Although we did not observe the 235 proliferation of islet clusters isolated from wild-type mice, we did observe the expansion of 236 islet clusters isolated from wild-type rats, which share more properties identical to human 237 islets than mouse islets. An urgent need is to test whether PIEM enables the expansion of 238 primary islet clusters isolated from human patients. 239 Another notable observation is that only dispersed islet clusters were able to proliferate in 240 vitro, whereas dispersed single primary islet cells could not. Two important factors may 241 contribute to this discrepancy. First, the islet clusters contain multiple cell types, which are 242 not only the origins of the different cell types in the finally expanded islet clusters but also 243 may form certain cell circuits and gradient hormone concentrations to support in vitro 244 expansion. Second, primary single cells must undergo harsher digestion than islet clusters, 245 which means that the extracellular parts of the membrane or matrix proteins, including 246 receptors, ion channels and transporters, may be digested and their function impaired. These 247 membrane or matrix proteins may be key for in vitro tissue expansion. Therefore, future 248 analysis of the different expression patterns and functions of membrane proteins and 249 signalling circuit differences between isolated islet clusters and single cells may provide more 250 clues for unravelling the secrets of the in vitro expansion of islets and provide guidance for a 251 better strategy for in vitro islet cluster expansion. 252 We noticed that expanded islet clusters have higher expression of Ngn3 and Hlxb9, which 253 are markers of pancreatic islet progenitor cells. Although the Nanog in islet clusters increased 254 approximately 10 fold compared to normal islets, however, this level is more mimic the whole 255 embryo tissue but much lower (more than 1,000 fold lower) than the embryo stem cells 256 ( Supplementary Fig. S2e). Whether this amount Nanog is required for islet expansion and 257 propagation in vitro, and whether it has cancer potential require further investigation. This 258 observation indicated that in vitro expanded islets underwent certain dedifferentiation. 259 However, we doubt that the islet clusters gained proliferation ability through this 260 dedifferentiation. Further experiments with withdrawal of key chemicals in the medium will 261 test this hypothesis. For clinical usage, the in vitro expansion of 100 patient islets to 1 million 262 islets without significant gene mutations, while preserving pancreatic islet function, and 263 transferring them back to the patient to recover glucose homeostasis is ideal. However, the 264 current approach of PIEM allows only 5~8-fold growth of approximately 5%-15% of isolated 265 islet clusters. The expansion of the islet clusters was only observed in the first passage.

Pancreatic islet isolation and digestion into single cells or cell clusters. 297
We isolated pancreatic islets from C57BL/6 mice, pregnant C57BL/6 mice and Wistar rat.

Production of Rspo1-conditioned medium 319
The RSPO1 conditioned medium is home-made and will be described in other papers.  Supplementary Table S1.

Immunofluorescence staining 351
Expanded islet clusters were harvested using cell recovery solution (Corning, 354253) and 352 fixed in 4% paraformaldehyde for 30 min at room temperature, followed with blocking and 353 permeabilizing in PBS with 0.5% Triton X-100 (Solarbio, T8200) and 5% donkey serum 354 (Solarbio, SL050) for 30min at room temperature. Then, samples were incubated with 355 primary antibody at 4℃ overnight, followed by incubation with secondary antibody for 2h at 356 room temperature. DAPI (Beyotime, C1002) was used to stain the nucleus and find islets. The

Insulin secretion measurement 365
The insulin secretions were preformed similar to previously described in our group 7,9,24 . 366 Fifty day 7 expanded islet clusters (cell number is comparable to 1/10 of primarily isolated 367 islets, 1,000~2,000 cells) or ten islets for each group were starved in 2.8 mM glucose 368

Somatostatin measurement 376
The somatostatin secretions were preformed similar to previously described in our 377 group 7,9,24 . Fifty day 7 expanded islet clusters or ten primary islets for each group were 378 starved in 1 mM glucose MKRBB buffer for 1 hour at 37℃, then treated with 20 mM glucose 379 and 100 nM UCN3, whereas the control groups were treated with 1 mM glucose for 1 hour. 380 The supernatant fractions were collected for somatostatin measurement using the Phoenix 381 Pharmaceuticals ELISA kit (EK-060-03), as what indicated by the manufacturer's 382

Glucagon measurement 385
The glucagon secretions were preformed similar to previously described in our group 7,9,24 . 386 Fifty day 7 expanded islet clusters or ten primary islets for each group were starved in 12 mM 387 glucose MKRBB buffer for 1 hour at 37℃, and then treated with 1 mM glucose and 25 mM 388 Arginine, whereas the control groups were treated with 5 mM Arginine for 1 hour. The

Statistical analysis 394
All the qRT-PCR data were performed independently for at least three times. Statistical 395 analyses were performed using GraphPad Prism 7 software. Experimental data were 396 performed using unpaired two-tailed Student's t-test. Results are presented as mean ± SEM; P 397 < 0.05 was considered statistically significant.  g Insulin secretion of primary islets or expanded islet clusters treated with 20 mM glucose and 540 100 nM glucagon peptide 1 (GLP-1) were measured for 30 minutes. The control groups were 541 treated with 2.8 mM glucose. ***, p < 0.001, stimulation groups were compared with control 542 groups. ###, p < 0.001, expanded islet clusters were compared with primary islets. 543 h Somatostatin secretion of primary islets and expanded islet clusters treated with 20 mM 544 glucose and 100 nM UCN3 were measured for 1 hour. The control groups were treated with 1 545 mM glucose. **, p < 0.01, stimulation groups were compared with control groups. NS, no 546 significance, expanded islet clusters were compared with primary islets. 547 i Glucagon secretion of primary islets and expanded islet clusters treated with 25 mM 548 Arginine were measured for 1 hour. The control groups were treated with 5mM Arginine. ***, 549 p < 0.001, stimulation groups were compared with control groups. NS, no significance, 550 expanded islet clusters were compared with primary islets. 551 The data are shown as the mean ± SEM of at least three independent experiments. The data 552 statistics were analysed using an unpaired two-tailed Student's t-test. compared with primary islets. *, p <0.05; **, p < 0.01; clusters were compared with primary 562 rat islets. The data are shown as the mean ± SEM of at least three independent experiments. 563 The data statistics were analysed using an unpaired two-tailed Student's t-test. f Comparison of the expression of islet and pancreatic progenitor markers in rat expanded 578 clusters vs. isolated primary islets. 579 *, p < 0.05; **, p < 0.01; ***, p < 0.001; clusters were compared with primary islets. The data 580 are shown as the mean ± SEM of at least three independent experiments. The data statistics 581 were analysed using an unpaired two-tailed Student's t-test. Schematic depicting the isolation, seeding and the expansion of primary islet cell clusters. 595 After the islets were isolated from pregnant mice or wild-type rats, the islets were digested 596 into appropriate cell clusters by incubation with the dispase II. The digestion time were 597 optimized and the mechanical blow force was used for clusters generation. These clusters 598 were cultured 3D in PIEM to achieve the in vitro expansion of the islet cell clusters. 599