Xenocell therapy from neonate or adult pig pancreatic islets is one of the most promising alternatives to allograft in type 1 diabetes for addressing organ shortage. In humans, however, natural and elicited antibodies specific for pig xenoantigens, α-(1,3)-galactose (GAL) and N-glycolylneuraminic acid (Neu5Gc), are likely to significantly contribute to xenoislet rejection. We obtained double-knockout (DKO) pigs lacking GAL and Neu5Gc. Because Neu5Gc−/− mice exhibit glycemic dysregulations and pancreatic β-cell dysfunctions, we evaluated islet function and glucose metabolism regulation in DKO pigs. Isolation of islets from neonate piglets yielded identical islet equivalent quantities to quantities obtained from control wild-type pigs. In contrast to wild-type islets, DKO islets did not induce anti-Neu5Gc antibody when grafted in cytidine monophosphate-N-acetylneuraminic acid hydroxylase KO mice and exhibited in vitro normal insulin secretion stimulated by glucose and theophylline. Adult DKO pancreata showed no histological abnormalities, and immunostaining of insulin and glucagon was similar to that from wild-type pancreata. Blood glucose, insulin, C-peptide, the insulin-to-glucagon ratio, and HOMA-insulin resistance in fasted adult DKO pigs and blood glucose and C-peptide changes after intravenous glucose or insulin administration were similar to wild-type pigs. This first evaluation of glucose homeostasis in DKO pigs for two major xenoantigens paves the way to their use in (pre)clinical studies.
Pancreatic islet allotransplantation is a realistic alternative or complement to insulin therapy in type 1 diabetes (T1D) to prevent serious long-term complications but is limited by the lack of pancreas. Pig pancreas remains a promising complementary islet source. Reproducible evidence exists on the long-term therapeutic benefit of islet xenotransplantation in pig to nonhuman primate preclinical models, using encapsulation for islet immunoprotection (1,2) or using immunosuppressant treatments (3–6). Methods for pig islet purification have significantly improved, in particular for neonatal pig islet-like cell clusters (NPCCs), which are easy to isolate and functional in vivo in the long-term (4,6,7). However, strong humoral response to pig antigens, especially against the α-(1,3)-galactose (GAL) sugar, was detected in nonhuman primates after grafting of neonate (2) and adult (1) pig islets, even encapsulated in alginate hydrogel, and may even affect encapsulated islet survival (1). Humans have lost expression of α-1,3 galactosyl transferase (GGTA1) enzyme generating the GAL epitope (8) and produce natural and elicited antibodies to GAL (9). Among non-GAL, the N-glycolylneuraminic acid (Neu5Gc) sugar is also a major pig xenoantigen (10). Humans have also lost the ability to synthesize Neu5Gc from the N-acetylneuraminic form, after mutation of cytidine monophosphate-N-acetylneuraminic acid hydroxylase (CMAH) (11). Pig islets express Neu5Gc, which reacts with human natural antibodies (12,13), and Neu5Gc knockout (KO) mice reject pancreatic islets from wild-type (WT) counterparts (14).
Using transcription activator-like effector nucleases (TALEN), we obtained a double-knockout (DKO) pig line for the CMAH and GGTA1 enzymes. However, CMAH−/− mice exhibited higher fasting blood glucose and insulin than WTs, associated with mild glucose intolerance and significant β-cell dysfunctions, including higher insulin secretion by isolated islets in response to glucose (15). As a prerequisite for xenocell therapy for T1D and to study the metabolic effects of CMAH KO in a species more closely related to humans than mice, we analyzed glucose homeostasis in our DKO adult pigs and insulin secretion of isolated DKO NPCCs.
Research Design and Methods
Animals and Ethical Statements
DKO pigs were obtained by targeting the swine CMAH gene using TALEN on the GGTA1−/− background (obtained from Dr. D.H. Sachs, Massachusetts General Hospital, Boston, MA) and by somatic cell nuclear transfer (SCNT) (Supplementary Data). WT pigs were Landrace × Large-White (EARL Pont Romain, Surzur, and ANSES, Ploufragan, France). Adult and neonate pigs were males, respectively, 5–6 months old weighing 37.90 ± 3.47 kg and 3–10 days old weighing 1–3 kg. CMAH−/− (B6.129 × 1-Cmahtm1Avrk/J; The Jackson Laboratory, Bar Harbor, ME) and control mice (C57BL/6; Janvier Laboratories, Le Genest-Saint-Isle, France) were 8–12 weeks old. Experiments were approved by the Ethics Committees and performed in accordance with relevant Italian (DGL116/92) and French regulations (2001-464 and 2013-118, Approval 01074.01/02). All efforts were made to minimize animal suffering and to restrict the number of experimental animals.
Insulin and Glucagon Immunostaining and Content in Pancreas
Formalin-fixed paraffin-embedded pancreas cross-sections were incubated overnight with rabbit anti-insulin (C27C9; Cell Signaling Technology, Beverly, MA) and mouse anti-glucagon (G2654; Sigma-Aldrich, St. Louis, MO) IgGs and then 1 h with secondary Alexa-Fluor 488 donkey anti-rabbit IgG and Alexa-Fluor 555 donkey anti-mouse IgG antibodies (Thermo Fisher Scientific, Rockford, IL). Pancreatic insulin and glucagon contents were evaluated by ELISA (Mercodia, Uppsala, Sweden) after acid-ethanol extraction of three 1 cm3 frozen blocks per pancreas (head, body, and tail) according to the Animal Models of Diabetic Complications Consortium (http://www.diacomp.org/shared/document.aspx?id=73&docType=Protocol - Ed. Leiter) protocol.
Metabolic Investigations in Adult Pigs
Glycemia (Stat-Strip-Glucometer; Nova Biomedical, Waltham, MA), serum or plasma insulin, glucagon, and C-peptide (ELISA; Mercodia) were assessed in fasted pigs and after an intravenous (I.V.) glucose tolerance test (IVGTT) and insulin tolerance test (ITT). ELISA samples were run in duplicates with variation coefficient <10%. HOMA of insulin resistance (HOMA-IR) was (glucose [mmol/L]*insulin [mIU/L])/22.5).
In IVGTT, the glucose disappearance rate was (Ln [Glucose T5min] – Ln [Glucose T30min])*100/25 min. Acute insulin response to glucose (AIRglu) was the mean of serum insulin levels 3, 5, 10, and 15 min after glucose injection, minus the prechallenge (t = 0) level. Areas under the curve (AUC) were calculated above t = 0 values using the trapezoidal rule.
In the ITT, KITT measurement of glucose disappearance was 0.693*100/t1/2, where t1/2 is the time to serum glucose level half the level at 2 min.
Isolated Neonatal Pig Islets
Isolation from pancreas, culture, and capacity of NPCCs to secrete insulin in response to acute glucose, with or without theophylline (potentiating insulin secretion), were performed as described previously by Korbutt et al. (7) (Supplementary Data). The insulin stimulatory index was the quantity of insulin (ELISA) in the supernatant of NPCCs cultured with stimulus (20 mmol/L glucose ± 10 mmol/L theophylline), divided by the insulin level in the basal medium (RPMI, 2.8 mmol/L glucose, 2 mmol/L l-glutamine, 0.5% BSA). In the graft experiments, NPCCs were encapsulated into 200 μL 2% silanized-hydroxypropyl methylcellulose (Si-HPMC) hydrogel (16) and injected subcutaneously in 2% isoflurane anesthetized mice.
Data are expressed as mean ± SEM. P < 0.05 was considered statistically different (Mann- Whitney test; Prism, GraphPad Software, Inc., La Jolla, CA).
DKO Pig Clone Generation
GAL−/− pig primary fibroblasts were cotransfected with plasmids encoding a TALEN pair targeting the HindIII site located in pig CMAH exon 4 fragment (Supplementary Fig. 1), and the targeting donor DNA construct containing a 1.65-kb fragment of the CMAH exon 4, and a phosphoglycerate kinase promoter-PuroR-pA expression cassette in the HindIII site in the CMAH exon 4 fragment. PCR genotyping of the 27 puromycin-resistant fibroblast clones identified 3 clones subsequently used for SCNT (A6, D1, E8) (Supplementary Fig. 1). Sequence analysis showed deletions ranging from 7 to 40 nucleotides around the targeted HindIII site (Supplementary Fig. 2). Eight living piglets from two farrowings were tested to confirm CMAH−/− genotype (four piglets originated from clone D1, two from A6, and two from E8). We confirmed the absence of Neu5Gc (Fig. 1A) and CMAH mRNA (Fig. 1B) in pig fibroblasts.
Necropsy and Histological Characterization of Adult DKO Pigs
DKO and WT adult pigs showed no obvious developmental abnormalities at necropsy (Supplementary Data). The DKO pancreas showed no evident macroscopic abnormalities. Pancreatic islets from four DKO and four WT pigs underwent histological analyses (Fig. 2A). DKO islets seemed slightly larger and numerous than in WT pigs but without a significant difference (P = 0.1143) (Fig. 2C–E). Mean islet volume density (Supplementary Data) did not differ (3.16 ± 0.13% for DKO pancreas and 2.69 ± 0.12% for WT pancreas, P = 0.1143). Immunodetected (Fig. 2B) insulin and glucagon areas on pancreas sections and percentage of glucagon area/insulin area exhibited no significant differences between DKO and WT pigs (P = 0.4857 [Fig. 2F], P = 0.8286 [Fig. 2G], and P = 0.8286 [Fig. 2H], respectively). Insulin (Fig. 2I) and glucagon (Fig. 2J) pancreatic contents did not differ between pig groups (P = 0.8286).
Glycemic Homeostasis and Pancreatic Endocrine Functions in Adult DKO Pigs
Fasting blood glucose was slightly lower in adult DKO pigs (3.77 ± 0.17 mmol/L, n = 4) than in age-matched WT control pigs (4.73 ± 0.03, n = 2) (Supplementary Table 1). Fasting serum insulin was similar (1.74 ± 0.24 and 1.73 ± 0.21 pmol/L, respectively) (Supplementary Table 1). The serum insulin-to-glucagon molar ratio (0.53 ± 0.17 for DKOs vs. 0.55 ± 0.24 for WTs) (Supplementary Table 1) and HOMA-IRs (2.04 ± 0.33 vs. 2.52 ± 0.32) after an overnight fast appeared almost identical in pig groups.
In IVGTT, blood glucose curves were very similar between DKO and WT pigs. Blood glucose peaked 3 min after glucose administration and then dropped to reach the fasting value after 60–90 min (Fig. 3A). The rate of glucose disappearance did not differ (2.42 ± 0.63 mmol/L/min for DKOs and 2.47 ± 0.78 for WTs) (Supplementary Table 1). AUCglu were also almost identical (421.8 ± 46.58 mmol*120 min/L and 435.5 ± 19 for DKOs and WTs, respectively). Serum insulin increased 8–17-fold in the first minutes after glucose injection (Fig. 3B). C-peptide also showed rapid elevation (Fig. 3C). The DKO pig number 3 (DKO3) had a delayed insulin (15 min after glucose administration) and C-peptide (10 min) peak. Consistently, DKO3 presented a moderately elevated AUCglu associated with a slightly lower AIRglu and AUCins than WT and other DKO pigs, but not objectified by a different AUCC-pep (Supplementary Table 1). Glucagon suppression was more rapid and intense in WTs and DKO3 than in the other DKOs (Fig. 3D).
During ITT, blood glucose in DKO pigs dropped rapidly and was similar to controls (Fig. 3E). Glucose disappearance did not differ between pig groups (KITT, 11.36 ± 1.50% for DKOs vs. 10.73 ± 0.82 for WTs) (Supplementary Table 1). C-peptide curves were also similar between pigs, except for the DKO pig number 2 (DKO2) exhibiting a chaotic waning of C-peptide after insulin injection (Fig. 3F).
Pancreatic Islets From Neonate DKO Pigs
The number of NPCC islet equivalent quantities (IEQs) obtained after isolation was comparable in DKOs (16.44 ± 1 IEQs/g, n = 5) versus WTs (14.637 ± 3.234, P = 0.5921, n = 10). Viability of DKO NPCCs was not affected compared with WTs (data not shown). NPCCs exhibited a normal insulin secretion after glucose/theophylline stimulation (Fig. 4A). The average insulin secretion induced by glucose alone appeared higher for DKO (n = 5) NPCCs than WT (n = 6), without reaching significance (P = 0.1255) (Fig. 4A). As expected, theophylline increased insulin secretion by WT (P = 0.026) (Fig. 4A) and DKO islets (however, not significantly, probably because of the positive effect of glucose alone on insulin secretion by DKO islets). The proportion of cellular insulin secreted in response to glucose plus theophylline stimulation by DKO islets increased between 24 h and 4–5 days’ in vitro culture (36.64 ± 5.35% at 24 h vs. 71.00 ± 9.76% at 4–5 days, P = 0.032) (Supplementary Table 2), suggesting the beginning of NPCC maturation was even more rapid than for WT islets (34.73 ± 5.58% at 24 h vs. 36.69 ± 10.25 at 4–5 days).
Finally, we checked induction of anti-Neu5Gc antibodies in CMAH−/− mice after a subcutaneous graft of pig NPCCs encapsulated in Si-HPMC to prolong their survival (unpublished data). We confirmed that WT and GAL−/− NPCCs induced anti-Neu5Gc antibody response in CMAH−/− mice unlike DKO NPCCs (Fig. 4B).
Anti-GAL antibodies have been considered for years as the first obstacle to successful xenotransplantation. Natural anti-Neu5Gc antibodies are likely to be important non-GAL antibodies because they are present in most normal human sera (12,17) and are elicited by challenges with xenogenic tissues (18). Preexisting antibodies present in human serum bind to GAL−/− pig islets (19). Moreover, WT syngeneic pancreatic islets are rapidly rejected by CMAH−/− mice and not by GAL−/− mice (14).
The methodology we described here to disable CMAH gene expression in pig GAL−/− fibroblasts, using TALEN and SCNT, produces viable DKO pigs efficiently and rapidly (within 8 months from TALEN design). Others have also generated GAL and Neu5Gc DKO pigs using zinc finger nucleases (20), TALEN (21), or CRISPR-Cas9 (22). Only single GAL KO pigs have been investigated previously for glucose homeostasis and isolated islet functions (23).
Unlike the CMAH−/− mouse (15), fasting blood glucose and insulin and glucagon secretions appeared normal in adult DKO pigs. Furthermore, DKO pigs seemed to exhibit neither glucose intolerance nor insulin resistance as indicated by changes in blood glucose levels after IVGTT and ITT, respectively (confirmed by HOMA-IR). Although the number of adult control pigs tested in our study was limited, the results with control WT pigs were similar to those with DKOs, and values remained within the physiological range for fasting and IVGTT in pigs (24). After glucose administration, one DKO pig had delayed insulin secretion without this clearly affecting glycemia regulation, probably as glucagon rapidly decreased in this pig.
DKO pigs also displayed preserved islet architecture (islet number, area, volume density, and insulin and glucagon immunostaining). Further studies with a high-energy diet would be helpful to investigate the effect of the CMAH KO on β-cell and islet compensation in response to severe obesity-induced insulin resistance (15).
Furthermore, isolation of NPCCs yielded an IEQs-to-g pancreas ratio similar to that resulting in control pigs. DKO NPCCs were functional ex vivo and exhibited insulin secretion after stimulation with glucose and theophylline. Pig islet Neu5Gc and sialic antigens clearly contribute to pig islet antigenicity (12,13). Also, our results clearly confirmed that anti-Neu5Gc antibodies are induced in CMAH−/− mice by WT encapsulated NPCCs, probably mimicking what is expected in the human context. CMAH editing in our pigs prevented this humoral-specific response, consistent with earlier observations of reduced binding of human antibodies to peripheral blood mononuclear cells from GGTA1/CMAH KO pigs (25). CMAH editing will most likely be necessary when hydrogel-encapsulated pig islets are used for T1D xenotherapy because encapsulation in hydrogel does not fully prevent antisugar humoral responses (1,2).
In conclusion, our study opens the way for the use of DKO pig pancreatic islets in (pre)clinical studies on T1D cell therapy.
Acknowledgments. The authors are very grateful to David H. Sachs (Massachusetts General Hospital, Boston, MA) for providing the GAL KO pig. The authors thank Odile Duvaux (Xenothera, Nantes, France) for the scientific discussions.
Funding. This work was supported by Pays de la Loire Region (France) (Xenothera academic program and to A.S.), the Société d’Accélération du Transfert de Technologies Ouest Valorisation (to A.S.), the European Center for Transplantation and Immunotherapy Sciences (ECTIS IHU, Nantes, France), the National Research Agency “Investment Into The Future” programs (ANR-10-IBHU-005, to X.L., and ANR-II-INSB-0014), the European Commission’s Xenome Sixth Framework Programme (LSHB-CT-2006-037377), and the European Seventh Framework Programme “Translink” research program (grant agreement 603049 and to L.L.B.).
Duality of Interest. A.S. is currently an employee of the start-up Xenothera. J.-P.S. and J.-M.B. are cofounders of the start-up Xenothera. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. A.S., M.M., X.L., A.P., J-.P.J., C.D., S.P., A.M., D.J., N.G., J.A., O.G., S.L.B.-B., L.L.B., J.H., D.M., S.Bo., I.L., R.D., G.L., S.C., and J.-M.B. performed most of the experiments. A.S., M.M., X.L., C.D., S.P., A.M., D.J., S.Bo., S.C., C.G., J.-P.S., and J.-M.B. analyzed the data. A.S., M.M., X.L., A.P., A.M., D.R., S.Br., E.C., S.C., C.G., J.-P.S., and J.-M.B. wrote or reviewed the paper. A.P., J.-P.C., S.C., and C.G. generated the DKO pigs. D.R., S.Br., G.B., J.-P.S., and J.-M.B. helped acquire study funding. P.W. and J.G. provided Si-HPMC and helped with islet encapsulation design. C.G., J.-P.S., and J.-M.B. devised the scientific strategy and experimental design. J.-M.B. 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.