The Impact of Different Implantation Sites and Sex on the Differentiation of Human Pancreatic Endoderm Cells Into Insulin-Secreting Cells In Vivo

Human embryonic stem cells (hESCs) can be efficiently differentiated into pancreatic endoderm cells (PECs) in vitro that further differentiate into pancreatic endocrine cells following implant (1–5). The signals directing developing β-cells to become fully functional in vivo remain unknown, but it is evident that the cell environment influences development; for example, hESC-derived PECs implanted in chronic hypothyroid mice displayed impaired β-cell development (6) and developed glucose-stimulated human C-peptide secretion earlier when implanted in female mice compared with males (7).

Previous groups have studied the differentiation of hESC-derived PECs in different implantation sites (8–10), including Kroon et al. (1), who observed that male mice with PECs implanted in the gonadal fat pad (FP) developed higher levels of fasted and stimulated circulating C-peptide than those with cells implanted under the kidney capsule (KC). Additionally, this group noted that mice with PECs implanted in a subcutaneous site took longer to develop similar levels of glucose-stimulated human C-peptide than mice receiving cells in the gonadal FP (1). We performed a direct comparison of PECs implanted under the KC, in the gonadal FP, and subcutaneously within macroencapsulation devices (TheraCyte [TC], San Clemente, CA) in both male and female mice. This study demonstrates that both the site of implantation and the sex of the recipient can impact the identity and function of hESC-derived PECs into endocrine cells.

CyT49 hESCs were differentiated into PECs using a 12-day, four-stage protocol, as previously described (1,11,12). PECs were cryopreserved (11,12) until they were thawed and shipped overnight to the University of British Columbia (UBC) by ViaCyte (San Diego, CA). Upon receipt, the PECs were cultured up to 3 days using 100 mL Vertical-Wheel bioreactors (IA-UNI-B-501; PBS Biotech, Camarillo, CA) mixing at 45 rpm in the incubator (20% O₂, 8% CO₂, 37°C). Experiments were approved by the Canadian Stem Cell Oversight Committee and the UBC Clinical Research Ethics Board.

Age-matched male and female Fox Chase SCID-Beige (CB17.Cg-Prkdc^scid Lyst^bg-J/Crl) mice were received from Charles River Laboratories (Wilmington, MA). Mice were on a 12-h light/dark cycle and fed ad libitum the standard irradiated Teklad Diet no. 2918 (Harlan Laboratories, Madison, WI). Mice were housed at a density of one to five mice per cage; some mice required single housing due to fighting. Experiments were approved by the UBC Animal Care Committee and conducted in accordance with the Canadian Council on Animal Care guidelines.

At ∼8 weeks of age, mice were anesthetized with inhalable isoflurane and implanted with 5 × 10⁶ PECs. Mice were randomly assigned to receive cells under the KC, in the gonadal FP using a Gelfoam scaffold and Matrigel overlay, or subcutaneously within 20 µL TC macroencapsulation devices. Blood glucose was measured 2 days before operation and on the operation date. Operations were performed over 3 days on 48 mice, using procedures outlined in the Supplementary Material.

Metabolic tests were conducted in conscious, restrained mice. Blood was collected at the indicated time points via the saphenous vein. Body weight and blood glucose were measured after a 4-h fast. Blood glucose levels were measured using a OneTouch Verio handheld glucometer (Lifescan, Burnaby, British Columbia, Canada). Plasma human C-peptide levels were assessed following an overnight (16-h) fast and an intraperitoneal (i.p.) injection of glucose (3 g/kg body wt, 30% solution; Vétoquinol, Lavaltrie, Quebec, Canada). The insulin tolerance and arginine tolerance tests were performed following a 4-h fast, with an i.p. injection of human synthetic insulin (0.7 IU/kg body wt; Novolin ge Toronto, Novo Nordisk, Mississauga, Ontario, Canada) or l-arginine (2 g/kg body wt, 40% solution; Sigma-Aldrich, Oakville, Ontario, Canada), respectively. Further details of the assays performed are described in the Supplementary Material.

Samples were fixed in 4% paraformaldehyde overnight and stored in 70% ethanol before paraffin embedding. For immunofluorescent staining, paraffin-embedded 5-μm-thick sections underwent heat-induced epitope retrieval for 10–15 min at 95°C using 10 mmol/L citrate buffer at pH 6.0. Sections were incubated with primary antibodies (see Supplementary Table 1) overnight and scanned using an ImageXpress Micro Imaging System (Molecular Devices Corp, Sunnyvale, CA). Hematoxylin and eosin-stained slides were scanned by Wax-It Histology Services (Vancouver, British Columbia, Canada).

Statistical analyses for metabolic tests in the mice were performed in R version 4.1.0 (13). A Bayesian approach with multilevel regression modeling was used due to the high number of correlated experiments with repeated measurements (14), as detailed in the Supplementary Material.

Code and data for analyses are available at https://github.com/caraee/TJKV06.

Male mice that received hESC-derived PECs under the KC weighed less than mice receiving cells subcutaneously within TC devices from weeks 1 to 5 and in the FP from weeks 4 to 20, while there were no differences between the groups of female mice (Supplementary Fig. 1A).

Mice that received PECs had lower fasted blood glucose compared with preimplant from 4 weeks postimplant onwards, with a final decrease of 2.0–3.9 mmol/L (point estimate; credible interval 1.0, 5.1) in males (Supplementary Fig. 1B, bottom) and from 4 (KC, except at 13 weeks), 9 (TC), or 7 weeks (FP) postimplant until the end of the study, with a final decrease of 2.3–3.9 mmol/L (1.1, 5.0) in females (Supplementary Fig. 1B, top). Mice had lower fasted blood glucose with PECs in TC devices than in the FP from 11 weeks postimplant (0.9–3.1 mmol/L [0.1, 4.5]) in males and from 9 to 26 weeks (1.7 mmol/L [0.3, 3.1]) postimplant in females. Male mice that received PECs under the KC had lower fasted blood glucose compared with in the FP only at 4 (1.4 mmol/L [0.7, 2.2]) and 20 (2.3 mmol/L [1.2, 3.5]) weeks postimplant, whereas female mice were consistently lower from 14 to 27 weeks postimplant (2.1 mmol/L [0.6, 3.6]). These results suggest that in both male and female mice, PECs under the KC or in TC devices lowered fasted blood glucose more consistently and robustly compared with the FP.

Mice with cells in TC devices had improved glucose tolerance compared with under the KC or in the FP from 10 (female) (Fig. 1A, top) and 12 (male) weeks (Fig. 1A, bottom) to 21 weeks (see Supplementary Fig. 2 for individual data). These improvements in glucose tolerance correspond with increasing human C-peptide levels; female mice with cells in the TC device had glucose-stimulated C-peptide secretion as early as 8 weeks postimplant (increase of 0.4 nmol/L [credible interval 0.0, 0.7] at 60 min) (Fig. 1B, top). Female mice with cells under the KC had increased C-peptide in response to glucose during week 12 postimplant (0.2 nmol/L [0.0, 0.3] at 60 min), and female mice with cells in the FP had modestly increased C-peptide in response to glucose during week 16 (0.2 nmol/L [0.0, 0.4] at 60 min). A similar pattern of C-peptide was seen in male mice (Fig. 1B, bottom), but glucose-stimulated C-peptide secretion was not observed until 10 weeks in the TC device group (0.7 nmol/L [0.3, 1.2] at 60 min).

The improved glucose tolerance in female mice compared with males with the TC device may reflect increased insulin sensitivity in females (15). However, in alignment with our previous results (7), females implanted with PECs under the KC had higher levels of C-peptide and were able to improve glucose tolerance earlier than in males. This suggests that cells mature sufficiently to respond to a glucose challenge in a manner similar to human islets (2) earlier in female mice under the KC compared with male mice, earlier in the TC device compared with other sites in both sexes, and less effectively in the FP in both sexes.

All groups displayed a reduction in blood glucose and human C-peptide levels following i.p. injection of insulin at 19 weeks postimplant (Fig. 2A and B; see Supplementary Fig. 3 for individual data). Female mice with PECs in the TC device (1.6 mmol/L [credible interval 0.7, 2.6]) and under the KC (1.3 mmol/L [0.4, 2.3]) had lower fasted blood glucose compared with in the FP (Fig. 2A, top), as well as higher C-peptide preinjection (0.8 nmol/L [0.5, 1.2]) (Fig. 2B, top). Males with PECs in the TC device had lower fasted blood glucose than in the FP (2.9 mmol/L [1.9, 3.9]) or under the KC (1.9 mmol/L [0.9, 2.8]) (Fig. 2A, bottom) as well as higher fasted C-peptide than in the FP (1.0 nmol/L [0.6, 1.3]) or under the KC (0.7 nmol/L [0.3, 1.0]) (Fig. 2B, bottom).

All groups displayed similar changes in blood glucose levels relative to baseline following an i.p. injection of arginine at 24 weeks postimplant (Fig. 3A). Male mice with cells in the TC devices had higher insulin at all time points compared with cells in the FP and at 7 and 15 min compared with the KC group and compared with female mice with cells in the TC device (1.0 nmol/L) (Fig. 3B). Female mice with cells in the TC device (all time points) or under the KC (7 min) had moderately higher arginine-stimulated insulin compared with the FP group (0.3 nmol/L), and the TC group had higher insulin compared with the KC group (0.2 nmol/L [credible interval 0, 0.4]).

Some mice had fasted plasma glucagon and glucagon-like peptide 1 (GLP-1) levels below the limit of detection (Fig. 3C and D), but all mice displayed an increase in glucagon and GLP-1 levels in response to arginine. Female mice with cells under the KC had higher GLP-1 at all time points compared with the FP group (15 min: 18.2 pmol/L [credible interval 10.5, 25.5]) and TC group (15 min: 15.4 pmol/L [6.9, 23.5]) (Fig. 3D), whereas male mice with cells in the TC device had higher GLP-1 compared with in the FP at 15 (12.1 pmol/L [6.0, 18.0]) and 60 min. Female mice had higher GLP-1 compared with male mice with cells under the KC (15 min: 14.5 pmol/L [5.2, 23.3]). Similar results were seen in plasma glucagon levels (Fig. 3C).

Prior to implantation, PECs contained cells positive for the endocrine cell marker synaptophysin, insulin, glucagon, and somatostatin (Supplementary Fig. 4A). Costaining of insulin with glucagon or somatostatin was also observed. Additionally, PECs displayed immunoreactivity for the nuclear transcription factors PDX1 and NKX6.1 but not MAFA. Immunofluorescent staining of adult human pancreas sections was used as a positive control (Supplementary Fig. 5). Expression of these proteins were also determined at the RNA level relative to human islet levels using quantitative PCR analysis (Supplementary Fig. 4B).

The images in Fig. 4 are representative of all TC (n = 15), KC (n = 16), and FP (n = 15) grafts. Since there were no differences between male and female grafts, only male grafts are shown. TC device grafts had more immunoreactivity for synaptophysin rather than the ductal cell marker cytokeratin 19 (CK19) whereas KC and FP grafts displayed a mix of cells immunoreactive for synaptophysin and CK19 at 28 weeks postimplant (Fig. 4). Immunoreactivity for the exocrine cell marker trypsin was not observed in any grafts. All grafts displayed insulin-, glucagon-, and somatostatin-positive cells. The nuclear markers PDX1, NKX6.1, and MAFA were also observed in insulin-positive cells in all grafts. While cysts were present in the KC and FP grafts, none were observed in the TC grafts. Despite the absence of cysts, a portion of cells in the TC grafts were immunoreactive for SOX9, a nuclear marker for ductal cells (Supplementary Fig. 6). As expected, due to the abundance of cysts and ductal structures, KC and FP grafts contained many cells immunoreactive for SOX9. SOX9-positive cells were also present in PECs before implantation.

Overall, this study demonstrates how the site of implantation and sex of the recipient can influence the in vivo differentiation of hESC-derived PECs into endocrine cells. PECs differentiated into glucose-responsive insulin-secreting cells earliest when implanted within TC devices regardless of sex. TC grafts also consisted mostly of synaptophysin-positive cells and, unlike KC and FP grafts, were devoid of large cysts. Consistent with our previous study (7), we observed earlier glucose-stimulated C-peptide secretion with PECs implanted under the KC in female compared with male mice. We observed higher levels of glucagon and GLP-1 in female mice with PECs under the KC compared with males, suggesting that the recipient’s sex plays a role in determining the identity of implanted PECS, by mechanisms that have yet to be elucidated.

While we cannot control for the effect of aging in these animals, the lowering of 4 h fasted blood glucose as C-peptide increases suggests that the implants are contributing to the glycemic set point. Human C-peptide levels of >0.3 nmol/L (1 ng/mL) are associated with substantial blood glucose lowering in mice (2,3). Mice with PECs in the TC device achieved this C-peptide level by week 10 postimplant compared with week 16 in the FP or under the KC. We observed that grafts from TC devices did not develop cyst-like structures and consisted mostly of endocrine cells compared with a mixed population of endocrine and ductal cells in the FP and under the KC.

In contrast to the high levels of C-peptide Kroon et al. (1) observed at 8 weeks postimplant in male mice that received PECs in the gonadal FP, we found PECs took the longest to develop glucose-stimulated C-peptide secretion when implanted in this site. Pepper et al. (9) also found that PECs in the FP of diabetic mice had not developed significant glucose-stimulated C-peptide secretion at 20 weeks postimplant. These differences highlight the role that the environment plays in PEC differentiation and maturation in vivo; differences in mouse strain, housing conditions, or diet could be contributing to the disparity between studies.

Vascularization is important for the survival, maturation, and function of implanted PECs. Pepper et al. (9) found that PECs implanted in a prevascularized subcutaneous site developed glucose-stimulated C-peptide secretion at 8 weeks postimplant, whereas cells implanted in subcutaneous sites that were not prevascularized had not developed significant function even at 20 weeks postimplant. PECs implanted within TC devices did not directly contact the blood vessels, whereas cells in the FP did, yet had inferior function. Vascularization likely plays an important role in cell identity and function, but our results suggest direct vascularization is not required for the development of glucose-responsive human β-cells, at least in mice.

The cells in TC devices are restricted in a confined space, whereas the KC and gonadal FP sites allow for more cell spreading. Mamidi et al. (16) found that cell confinement directed PECs toward an endocrine cell fate, whereas cell spreading promoted a ductal cell fate. Cells that were confined maintained high PDX1 and NKX6.1 expression, whereas spread cells downregulated PDX1 and NKX6.1 expression (16). Increased cell-to-cell contact within the TC devices therefore may have promoted the differentiation of these cells toward an endocrine cell rather than a ductal cell fate. KC and FP grafts displayed more immunoreactivity for the ductal marker SOX9 than TC grafts. The confined space of the TC devices may have also prevented the formation of cysts due to less off-target differentiation or due to the physical restraint provided from the devices making it difficult for cysts to form. Motté et al. (4) found that nonencapsulated hESC-derived PECs displayed a lower percentage of endocrine cells and large, cystic cavities compared with encapsulated cells implanted in the subcutaneous site.

We found that PECs implanted within TC devices were more consistently glucose-responsive in female mice compared with males. Additionally, this is now the third cohort of mice where we found that PECs, derived from either CyT49 or H1 cell lines, developed glucose-stimulated C-peptide secretion faster under the KC in female mice compared with male mice (7). PECs implanted under the KC in female mice secreted more glucagon and GLP-1 during an arginine tolerance test than in male mice. We did not observe any sex differences in PECs implanted in the gonadal FP, but these cells took longer than previously reported (1) to develop glucose-stimulated C-peptide secretion, and we might have seen differences had we followed the FP grafts longer.

Our findings support the use of implanting hESC-derived PECs in macroencapsulation devices as a means of containing off-target growth. How the site of implant may affect maturation and function of cells implanted at a more differentiated state remains to be determined. This study also highlights the importance of including both male and female subjects when investigating the maturation and differentiation of hESC-derived PECs.

Acknowledgments. The authors thank Shannon O’Dwyer (UBC) and Travis Webber (UBC) for their technical assistance and Evert Kroon (ViaCyte) for providing PEC and the protocol for FP implants.

Funding. This work was funded by the Stem Cell Network, Canadian Institutes of Health Research (CIHR), and JDRF. N.S. received a graduate scholarship from the National Sciences and Engineering Research Council of Canada.

Duality of Interest. This work was funded by STEMCELL Technologies. A.R. is an employee of CRISPR Therapeutics, and T.J.K. was an employee of ViaCyte. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. N.S. and C.E.E. wrote the manuscript. N.S., C.E.E., D.G.I., R.K.B., A.R., and T.J.K. were involved in the acquisition, analysis, and interpretation of the data, contributed to manuscript revisions, and approved the final version of the manuscript. N.S., C.E.E., A.R., and T.J.K. contributed to the conception and design of the experiments. T.J.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.