Multipotent mesenchymal stromal cells derived from porcine exocrine pancreas improve insulin secretion from juvenile porcine islet cell clusters

Neonatal and juvenile porcine islet cell clusters (ICC) present an unlimited source for islet xenotransplantation to treat type 1 diabetes patients. We isolated ICC from pancreata of 14 days old juvenile piglets and characterized their maturation by immunofluorescence and insulin secretion assays. Multipotent mesenchymal stromal cells derived from exocrine tissue of same pancreata (pMSC) were characterized for their differentiation potential and ability to sustain ICC insulin secretion in vitro and in vivo. Isolation of ICC resulted in 142 ± 50 × 103 IEQ per pancreas. Immunofluorescence staining revealed increasing presence of insulin‐positive beta cells between day 9 and 21 in culture and insulin content per 500IEC of ICC increased progressively over time from 1178.4 ± 450 µg/L to 4479.7 ± 1954.2 µg/L from day 7 to 14, P < .001. Highest glucose‐induced insulin secretion by ICC was obtained at day 7 of culture and reached a fold increase of 2.9 ± 0.4 compared to basal. Expansion of adherent cells from the pig exocrine tissue resulted in a homogenous CD90+, CD34−, and CD45− fibroblast‐like cell population and differentiation into adipocytes and chondrocytes demonstrated their multipotency. Insulin release from ICC was increased in the presence of pMSC and dependent on cell‐cell contact (glucose‐induced fold increase: ICC alone: 1.6 ± 0.2; ICC + pMSC + contact: 3.2 ± 0.5, P = .0057; ICC + pMSC no‐contact: 1.9 ± 0.3; theophylline stimulation: alone: 5.4 ± 0.7; pMSC + contact: 8.4 ± 0.9, P = .013; pMSC no‐contact: 5.2 ± 0.7). After transplantation of encapsulated ICC using Ca2+‐alginate (alg) microcapsules into streptozotocin‐induced diabetic and immunocompetent mice, transient normalization of glycemia was obtained up to day 7 post‐transplant, whereas ICC co‐encapsulated with pMSC did not improve glycemia and showed increased pericapsular fibrosis. We conclude that pMSC derived from juvenile porcine exocrine pancreas improves insulin secretion of ICC by direct cell‐cell contact. For transplantation purposes, the use of pMSC to support beta‐cell function will depend on the development of new anti‐fibrotic polymers and/or on genetically modified pigs with lower immunogenicity.

Isolation of ICC resulted in 142 ± 50 × 10 3 IEQ per pancreas. Immunofluorescence staining revealed increasing presence of insulin-positive beta cells between day 9 and 21 in culture and insulin content per 500IEC of ICC increased progressively over time from 1178.4 ± 450 µg/L to 4479.7 ± 1954.2 µg/L from day 7 to 14, P < .001.
Highest glucose-induced insulin secretion by ICC was obtained at day 7 of culture and reached a fold increase of 2.9 ± 0.4 compared to basal. Expansion of adherent cells from the pig exocrine tissue resulted in a homogenous CD90 + , CD34 − , and CD45 − fibroblast-like cell population and differentiation into adipocytes and chondrocytes demonstrated their multipotency. Insulin release from ICC was increased in the presence of pMSC and dependent on cell-cell contact (glucose-induced fold increase: ICC alone: 1.6 ± 0.2; ICC + pMSC + contact: 3.2 ± 0.5, P = .0057; ICC + pMSC no-contact: 1.9 ± 0.3; theophylline stimulation: alone: 5.4 ± 0.7; pMSC + contact: 8.4 ± 0.9, P = .013; pMSC no-contact: 5.2 ± 0.7). After transplantation of encapsulated ICC using Ca 2+ -alginate (alg) microcapsules into streptozotocin-induced diabetic and immunocompetent mice, transient normalization of glycemia was obtained up to day 7 post-transplant, whereas ICC co-encapsulated with pMSC did not improve glycemia and showed increased pericapsular fibrosis. We conclude that pMSC derived from juvenile porcine exocrine pancreas improves insulin secretion of ICC by direct cell-cell contact. For transplantation purposes, the use of pMSC to support beta-cell function will depend on the development of new anti-fibrotic polymers and/or on genetically modified pigs with lower immunogenicity.

| INTRODUC TI ON
It is currently recognized that type 1 diabetes (T1D) patients presenting long-term complications can be treated by allotransplantation of human islets. Most transplanted patients reach improved glycemic control, but this treatment rarely leads to insulin independence. 1 An effective treatment requires multiple human pancreas donors and lifelong immunosuppression. The limited availability of human organ donors is a major hurdle for this clinical application and has driven investigation to find other sources for islets or beta cells, including xenogenic islets isolated from porcine fetal, neonatal, or adult pancreases. 2 The isolation of islets from adult pigs is highly inefficient due to their fragility and in contrast, protocols for the digestion of fetal, neonatal, or juvenile pancreas allow to obtain high amounts of pancreatic cell clusters or islet-like clusters. 3 Cell clusters derived from the neonatal porcine are called either, neonatal porcine islets (NPI), porcine neonatal pancreatic cell clusters (NPCCs) or also neonatal porcine islet-like cluster (NPICC). [4][5][6][7] We will use the term of islet cell cluster (ICC). ICC display no defined structure and contains only low amounts of insulin-positive beta cells (5%) and a majority of duct or acinar cells. Several protocols exist, to increase postnatal beta-cell mass in theses cell clusters by beta-cell maturation in vitro. [8][9][10][11] ICC also mature after transplantation into immuno-deficient diabetic mice, where they respond poorly to glucose for several weeks before restoring normoglycaemia after 4-10 weeks depending on the amount of ICC transplanted. [12][13][14] In addition, first clinical trials of transplantation of encapsulated neonatal porcine islets in humans showed a reduction in unaware hypoglycemia events suggesting a clinical benefit in patients with unstable type 1 diabetes. 15,16 Currently, multipotent mesenchymal stromal cells (MSC) are investigated for regenerative cell therapies and islet transplantation purposes. MSC are potentially effective in supporting function and viability of isolated islets after transplantation and also dampen inflammatory immune reactions. 17 MSC are isolated and expanded from many tissues, such as bone marrow, adipose tissue, umbilical cord, and pancreas. [18][19][20][21] MSC are considered as "biomolecule dispensers" since they release bioactive molecules which improve tissue regeneration by increasing vascularization, 22-25 angiogenesis, 26 and reducing apoptosis. 27 It remains open whether MSC from different tissues have same capacities. [28][29][30] The transplantation of rodent or human islets with syngeneic, 23,27,31,32 autologous, 33 allogeneic, 34 and xenogeneic 35 MSC resulted in improved islet graft survival and function in rodent models. Similarly, immuno-deficient diabetic mice co-transplanted with porcine neonatal ICC and human MSC reached normoglycemia significantly earlier than mice transplanted with ICC alone. 36 The aim of this study was to investigate whether porcine MSC derived from the pancreatic exocrine tissue interacts with juvenile porcine ICC and is able to modulate insulin secretion.
We also performed co-encapsulation and transplantation of ICC with pMSC into immunocompetent mice to analyze their effect on graft function.

| Isolation and maturation of ICC
Fourteen (±1) days old piglets (Appel, Switzerland) were used to isolate ICC as described previously with some modifications. 37  tissue, containing cell clusters of diameters ranging between 50 and 550 µm, were placed in beta-cell maturation media composed of HAM's-F10 culture medium containing 0.5% BSA (fraction V; Sigma-Aldrich), 50 µmol/L IBMX (3-isobutyl-1-methylxanthine), 10 mmol/L glucose, 2 mmol/L glutamine, 10 mmol/L nicotinamide, 1.6 mmol/L sodium bicarbonate, and 1% streptomycin/penicillin, at 37°C and 5% CO 2 . 13 Tissue was placed in non-adherent flasks. From first day until day 5 of culture, 4/5 volume of media was changed every day, then media was replaced every second day. A non-negligible amount of ICC (1/3) was adherent and had to be detached from the bottom of flasks through manual pipetting every day. At day 5, flasks were changed and ICC cultured at density of ~20-25 000 IEQ/T175 flask.
For pMSC isolation and expansion, media containing exocrine tissue recovered after first 24-hour culture was used. ICC were taken after 3, 7, 14, 21, and 28 days to analyze maturation and functionality. At day 7, the final number of cell clusters was determined and expressed as IEQ.

| Immunofluorescence staining on juvenile pancreatic tissue and isolated ICC
Immunofluorescence staining was performed on sections of formalin-fixed and paraffin-embedded juvenile pancreases. ICC were collected at various time points fixed and collected using Histogel (Thermo Scientific), following the manufacturer's recommendations.
Briefly, 4-μm sections were treated with 0.01 mol/L citrate for 15 minutes in a microwave, to unmask epitopes. To avoid nonspecific binding, slides were incubated with 0.5% BSA for 30 minutes at  Microscope images were acquired using a fluorescence microscope (Mirax Midi, Zeiss).

| Isolation and expansion of MSC from juvenile porcine exocrine pancreas
Maturation media from the first 24-hour culture of porcine pancreatic cell cluster culture (from 3 pancreata) was centrifuged for 5 minutes at 320g to obtain a pellet of exocrine tissue. This pellet was

| Real-time cell analysis
Doubling time was assessed by real-time cell analyzer system (RTCA). Cell index (CI) was recorded by measuring impedance.
Cells at passage between 4 and 6 were seeded in duplicate at a density of 5000, 10 000, and 20 000 cells/well and their growth assessed for 120 hours using the instrument xCELLigence RTCA; (ACEA Biosciences Inc). Doubling time was calculated in the exponential growth phase with the following formula: (t2-t1) × Log(2)/ (Log(CI(t2)) -Log(CI(t1))) where t is time.

| Differentiation into adipocytes, chondrocytes, and osteoblasts
The expanded mesenchymal cells and human bone marrow-derived MSC (positive control) were differentiated into adipocytes, chondrocytes, and osteoblasts as previously described and with minor modification. 39 Briefly, for adipocyte differentiation, cells were seeded at high density (20 000-25 000 cells/cm 2 ) and cultured at 37°C in presence of 5% CO 2 for 3 weeks in adipogenic dif-

| Co-culture conditions and Insulin secretion assay
Five hundred IEQ (considered as 500 000 islet cells) were cultured either alone or with pMSC (50 000) (cell ratio 10:1). ICC and pMSC were co-cultured in cell-cell contact or in transwell condition without cell-cell contact. Experiments were performed in triplicates using complete ICC maturation media in 24-well plates for 3 days before performing insulin secretion assay. We used an allogenic experimental setting, in which pancreas-derived pMSC were always used with ICC obtained from different pancreas donors (different piglets). Insulin secretion was measured by performing a static incubation assay. Briefly, ICC were rinsed twice with oxy-

| Diabetes induction and transplantation of microencapsulated ICC in mice
MS transplantation was performed following protocols approved by the Geneva cantonal veterinary authorities (license GE/151/16).

Diabetes was induced by injection of streptozotocin (SZ) (Sigma)
at 220 mg/kg into the peritoneum of C57BL/6 male mice (Janvier, France). Mice were considered diabetic at blood glucose level >20 mmol/L. Three days after SZ injection, the diabetic mice were anesthetized with isoflurane, and each mouse received 1 mL of Caalg MS containing 30 000 IEQ of ICC alone or with 3.10 6 pMSC (cell ratio 10:1) into the peritoneum through a small incision. Blood samples were obtained from the tail vein for glucose measurement and graft failure was stated when glucose level was >20 mmol/L for 3 consecutive measurements. Body weight of transplanted and nontransplanted diabetic mice was measured until 14 or 21 days after transplantation.

| Statistical analysis
Data are expressed as mean ± SEM of numbers (n) of independent observations. The statistical significance of differences was calculated with the Student t test or the 1-way ANOVA.

| In situ hormone-positive cell aggregates in juvenile porcine pancreata
To evaluate the presence and distribution of endocrine cell aggregates in the pancreas of 14 days old juvenile piglets, immunostainings of alpha and beta cells were performed on random pancreas sections. Hormone-positive aggregates were varying in size, from few-cell aggregates to well-defined islets ( Figure 1). In few-cell aggregates, the presence of insulin-positive beta cells was predominant whereas in small well-defined islets, the glucagon-positive alpha cells were mostly present. The well-defined small islets showed that beta and α-cells appeared to be intermingled similarly as in human islets. 40 As the scope of the study was to isolate MSCs from the exocrine pancreas, pancreas sections were stained for vimentin, a marker for mesenchymal cells. Vimentin-positive mesenchymal cells were present throughout the exocrine tissue in the juvenile pig pancreata and often in close vicinity to small hormone-positive cell aggregates.

| Juvenile porcine pancreatic islet cell cluster isolation and maturation
For each of the 11 ICC isolations, 2-3 pancreata from piglets of the same litter were digested. After digestion, the pancreatic tissue including the exocrine component was cultured in neonatal pig islet differentiation media that was changed every day. ICC were counted at day 7 in culture, when exocrine tissue had completely disappeared. Table 1 Figure 2B). Further, to evaluate glucose responsiveness, insulin secretion assays were performed. ICC showed strongest insulin release to high glucose stimulation (5-fold increase) and maximal stimulation with theophylline (15-fold increase) after 7 days in culture, compared to basal release ( Figure 2C). Therefore, ICC after 10 days in culture were selected for further experiments.

| Expansion and characterization of MSC isolated from juvenile porcine exocrine pancreas
To investigate whether the porcine exocrine pancreas contains MSC, pancreatic exocrine tissue was recovered from media of ICC cultures after 24 hours of culture. Exocrine tissues from 3 pancreases were collected and placed in specific media for the expansion of human bone marrow-derived MSC. At passage 0 of culture, we observed morphologically heterogeneous cells, which had evolved into a homogenous fibroblast-like cell population at passage 3 ( Figure 3A).
Flow cytometry analysis showed that cells were CD90 + , CD45 − , and CD34 − , indicating a mesenchymal phenotype ( Figure 3B). To demonstrate multipotency, we differentiated these cells into adipocytes and chondrocytes. Figure 3C shows the adipocyte differentiation of

| Co-culturing ICC in direct contact with porcine exocrine pancreas-derived MSC supports insulin secretion
Earlier studies showed that human bone marrow-derived MSC in direct cellular contact with human islets showed a beneficial effect on insulin secretion in vitro and on graft survival and function after transplantation into diabetic mice. 34 Therefore, we investigated the effect of human and porcine MSC on insulin secretion by ICC with or without cell-cell contact using the transwell system. ICC and pMSC at a proportion of 10 islet cells for 1 pMSC were co-cultured for 3 days as depicted in Figure 4A.
As shown in Figure 4B, insulin secretion by ICC was significantly

| Transplantation of co-encapsulated ICC and pMSC is not beneficial for graft function compared to ICC alone and worsens pericapsular fibrosis
The effect of pMSC on graft function of ICC was investigated after

| D ISCUSS I ON
Human islets transplantation, as a treatment for type 1 diabetes patients, is limited by the shortage of human organ donors. The neonatal and juvenile porcine pancreas represent an unlimited source for the preparation of immature islets, also called ICC. These ICC contain few hormone-positive cells and mainly ductal and ductal precursor cells.
Maturation media was subsequently developed to increase the proportion of endocrine cells. 9,10 In view of future clinical applications, the production of neonatal and juvenile porcine ICC has the advantage to be more cost-effective compared to adult porcine islets which is related to the strict hygiene conditions necessary for the breeding and housing conditions of animal donors. Further, the isolation technique of juvenile porcine islets is more efficient and cultured immature ICC showed increased viability compared to adult islets. 42 In this study, juvenile porcine ICC were isolated and obtained with an average yield of IEQ at 142 ± 50 × 10 3 per pancreas at day 7.
In the literature, the average yield of ICC for one neonatal and young porcine pancreas amounts to 30.4 ± 1.2 × 10 3 3 and 33.3 ± 6.4 × 10 3 IEQ, 43 respectively. In the study of Lamb and coworkers, the mean age of the young piglets was 20 days (range 4-30 days). They described an almost 2-fold increase of the average weight of pancreas from piglets aged 11-18 days (5.25 ± 1.6 g) compared to piglets aged 4-10 days (2.75 ± 1.3 g). They also described a progressive decrease  of the amount of IEQ/g of pancreas isolated from piglets aged between 11 and 18 days and those of 20 and 30 days. Therefore, the increased number of ICC obtained in this study, might be resulting from the combination of both parameters, the increased size of the pancreas of 14-day-old piglets (compared to neonatal) and the still efficient digestion of the pancreas in this range of age.
Immunofluorescence staining on pancreas sections for alpha and beta cells from same litters showed a size-disparity of the hormone-positive cell aggregates. We also observed differences in in- It is known, however, that mesenchymal cells regulate pancreatic growth at early and late development stages, probably through expansion of the pool of epithelial pancreatic precursor cells, since the mesenchymal ablation at a late development stage in mice (e13. 5) leads to reduced beta-and acinar-cell mass. 45 Tracing the fate of the pancreatic mesenchyme in mice also revealed that nearly all mesenchymal cells acquire a pericyte fate at embryonic stage e13.5.
Spatial distribution and abundance of pancreatic PDGFRβ+ pericytes further indicate that these cells represent the primary pancreatic mesenchymal cell population and are the predominant fate of the embryonic pancreatic mesenchyme. 46 The identification of the exact cell type from which the isolated and expanded pancreatic MSC originate and their relation to pericytes or other types of fibroblasts needs further investigation. 47,48 A recent study demonstrated that in vivo, pericytes of multiple organs do not behave as MSC. This study challenges the current view that pericytes act as multipotent tissue-resident progenitors in vivo and suggest that the in vitro plasticity may arise from cell manipulations ex vivo. 49 Here we show that insulin secretion by ICC was significantly increased when co-cultured in direct cellular contact with exocrine tissue-derived pMSC for three days. High glucose and maximal stimulation with theophylline significantly increased insulin secretion compared to ICC cultured alone, suggesting that cell-cell interactions are important to support islet functionality.
This confirmed our previous studies and by other groups demonstrating that cell-cell contact between islets and MSC are necessary to improve islets function in vitro and after transplantation. 32,34 The interaction of the adhesion molecule N-cadherin expressed on islets and MSC were shown to be necessary to achieve the increased insulin secretion. Further, increased islet function was also obtained in engineered cell sheets composed of human islets and supporting human bone marrow, or adipose tissue-derived MSC or fibroblasts, 50 suggesting that cell-cell contact improves the micro-environment of isolated islets. Increased functionality and survival of isolated human and murine islets was observed by Gamble et al, when islets were cultured in suspension with MSC allowing direct cellular contact. 51 However, other studies showed that trophic factors alone are sufficient to induce beneficial effects. Park and coworkers showed that preconditioned murine islets through co-culture with human MSC without contact, increased survival, and function of transplanted murine islets in a syngeneic model. Increased amounts of trophic factors such as IL-6, IL-8, VEGF-A, HGF, and TGF-β measured in MSC-conditioned media have been suggested to be responsible for the activation of signaling pathways involved in angiogenesis and anti-apoptotic signaling of cultured islets. 52 Our results do not exclude that soluble factors derived from pMSC play a role, but this effect was not sufficient to increase ICC function. Therefore, we concluded that pMSC isolated from exocrine pancreas significantly supported insulin secretion from juvenile porcine ICC. A significant beneficial effect was not observed with human MSC, suggesting that species-specific cellular interactions are implicated.
Immunoregulatory effects have been shown for allogenic MSC when transplanted in non-human primates. In cynomolgus monkeys, intraportal co-infusion of allogenic MSC and islets increased islets engraftment and function, shown by a reduced number of islets necessary to reach normoglycemia. 22,53 However, in our model, after transplantation of Ca-alg MS containing ICC with pMSC into immunocompetent and streptozotocin-induced diabetic mice, a beneficial modulation of the immune and/or inflammation responses was not observed. Although encapsulated ICC alone decreased hyperglycemia up to 5-6 days, the survival and function of ICC co-transplanted with pMSC was not prolonged. The recovered MS containing ICC and pMSC showed increased pericapsular fibrosis compared to MS with ICC alone, suggesting that Ca-alg MS released increased amounts of xenoantigens.
It is known that porcine xenoantigens provoke a strong adaptive immune and inflammatory response, and until today transplantation of encapsulated ICC into mice without immunosuppression has been unsuccessful. However, by using a combination therapy of anti-CD154 and anti-LFA-1 monoclonal antibodies prolonged graft survival of encapsulated neonatal pig islets in immunocompetent mice has been obtained. Under such condition, transplanted and treated mice achieved normoglycemia within 10-35 days and around 50% of treated mice remained normoglycemic for more than 100 days. 54 We concluded that in immunocompetent mice, pMSC encapsulated together with ICC in Ca-alg MS are not able to alleviate the immune response against encapsulated porcine ICC. This is also in line with a recent study indicating that encapsulated MSC do not mitigate the foreign body responses against alg MS. 55 Semi-permeable Ca-alg MS that are developed to protect alloor xenografts present instability over time due to swelling and Ca 2+ loss, which causes increased pore size and ultimately capsule breakage. 56 Therefore, the future use of microcapsules for longterm transplantation will depend on the ability to further develop hydrogels that are more resistant to leakage and mitigate the foreign body response, as well as the genetic engineering of pigs with lower immunogenicity.
Herein, we showed that cellular contact between pancreatic-derived MSC and ICC supports insulin secretion. Our results suggest that pMSC improves the micro-environment for isolated porcine ICC and might be beneficial for porcine ICC or islet transplantation studies. The role of pancreatic MSC in the juvenile pancreas needs further investigations. Therefore, ICC from all development stages are a potential source of islets for the treatment of type 1 diabetes in humans and also represent a complementary experimental system for studying islet development and maturation. 57

ACK N OWLED G M ENT
We thank Jean-Pierre Giliberto for his excellent work in porcine anesthesia and for surgical assistance and Marlene Sanchez for her excellent technical help.

AUTH O R CO NTR I B UTI O N S
EM contributed to the study design, collected data, and performed data analysis and interpretation. LS produced polymers and contributed to the study design, collection of data, and data analysis.
AB and JM performed surgery to collect porcine pancreases and contributed to research data collection. NPM, JP, and M-NG contributed to study design, research data collection, and interpretation. SG contributed to the study concept, design of polymers, critical revision of the manuscript, and securing of funding. LB and BE contributed to study concept, design, and critical revision of the manuscript and securing of funding. CGG contributed to study concept and design, wrote the manuscript, collected data, and performed data analysis and interpretation and takes responsibility of the integrity of the study. All authors approved the final version of the manuscript.