Trefoil Factor 2 Expressed by the Murine Pancreatic Acinar Cells Is Required for the Development of Islets and for β-Cell Function During Aging

The pancreas contains exocrine and endocrine compartments, which secrete digestive enzymes and hormones that regulate blood glucose homeostasis, respectively. Dysfunction of the endocrine β-cells can result in type 1 or type 2 diabetes (T1D/T2D), which remains a global health concern (1). Increasing evidence indicates abnormalities in the exocrine pancreas impact β-cell function (2), such as cystic fibrosis (3,4), pancreatitis (5,6), and pancreatic cancer (7,8), potentially through secreted factors. However, the exocrine cell-derived signaling factors that regulate β-cell function remain largely unknown.

The trefoil family of secretory peptides consists of three proteins: Tff1, Tff2, and Tff3. In adult murine pancreas, Tff2 (9) and Tff3 (10) but not Tff1 (11) are detected, and Tff2 is expressed at higher levels than Tff3 (12). Tff2 is conserved from birds to humans (13), and is highly resistant to denaturing conditions and enzymatic digestion due to the seven intramolecular disulfide bonds in its tertiary structure and compact quaternary structure (14). Tff2 is abundantly expressed in stomach (15,16) and immune cells (17), where it binds with mucins to form protective gastric mucus barriers (18) and regulates immune response (19,20). Additionally, Tff2 is expressed in the central nervous system (21,22), anterior pituitary (21), hypothalamus (23), and pancreas (12,24). When added exogenously in vitro, Tff2 promotes the proliferation of adult murine pancreatic β-cells (25) and rescues embryonic β-cell apoptosis (9). However, the physiological role of Tff2 in the pancreas has not been discerned. Furthermore, immunohistochemistry studies disagree on whether Tff2 is expressed by endocrine (25) or exocrine pancreas in mice (9). Thus, there is a critical need to clarify which compartment expresses Tff2.

Here, we generated a conditional Tff2 knockout mouse model to test the physiological requirement of Tff2 in the pancreas, and to clarify the expression pattern of Tff2. We find that Tff2 protein is preferentially expressed by adult murine pancreatic acinar cells, and that Tff2 deficiency results in reduced β-cell mass in the embryonic pancreas and impaired glucose tolerance (IGT) in aged mice. Our findings demonstrate previously unknown roles of Tff2 in the endogenous pancreas and implicate Tff2 as a key signaling factor mediating exocrine-to-endocrine crosstalk.

Experiments on mice were approved by the City of Hope Institutional Animal Care and Use Committee, Duarte, CA. All mice had a C57BL/6J genetic background, which harbors nicotinamide nucleotide transhydrogenase mutation (26) (Supplementary Fig. 1). The generation of Pdx1:Cre;Tff2^fl/fl mice was achieved by intercrossing Pdx1:Cre mice (Jax #014647) with the Tff2 floxed (Tff2<tm1.1Htk> MGI:7620295; Tff2^fl/fl) mice (see below).

The mice harboring conditional knockout allele for Tff2 were designed and produced by Ozgene Pty Ltd. (Perth, Australia) (Supplementary Fig. 2).

PCR on genomic DNA with allele-specific primers (Supplementary Table 1) was used to determine genotypes.

Total RNA was isolated, converted into cDNA, and assayed using Taqman probes (Supplementary Table 2). β-Actin was used as the internal control.

Tile images from frozen and paraffin slides were collected, and then processed and quantified using QuPath v0.2.3 software.

Static glucose-stimulated insulin secretion (GSIS) was performed as previously described (27). The levels of insulin release were adjusted based on total protein content and number of islets. Islet perifusion was conducted as previously described (28).

Statistical significance was determined by unpaired, two-tailed Student t test when comparing two groups; Welch correction was used when comparing unequally sized samples. One-way ANOVA followed by Tukey post hoc analysis was used for comparing more than three groups. Data were analyzed with the GraphPad Prism 9 software and presented as mean ± SD or mean ± SEM.

Detailed methods are described in the Supplementary Material.

All data are available in the article or Supplementary Material. Data sets and resources are available from the corresponding author upon reasonable request.

Quantitative RT-PCR analyses revealed that Tff2 was most expressed followed by Tff3 and Tff1 at embryonic day 16.5 (E16.5), postnatal day 0 (P0), P7, and 6 weeks (Fig. 1A). Consistent with a previous report (9), Tff2 increased over time in the pancreas (Fig. 1B). Publicly available single-cell RNA-sequencing (scRNA-seq) databases indicate Tff2 is expressed in trunk, acinar, endocrine progenitor, and endocrine cells at E15.5 and/or E18.5 (29) (Supplementary Fig. 3), and in adult acinar and ductal cells (12) (Supplementary Fig. 4). Because of this promiscuous gene expression pattern, we deleted Tff2 in the whole pancreas.

We generated Tff2-floxed mice containing two loxP sites that flank exons 2 and 3 of the Tff2 gene, which code for the trefoil domain necessary for Tff2 function (30), as well as a Neo cassette flanked by two Frt sites inserted to the 5′ end of exon 3 (Fig. 1C). Subsequently, the Neo cassette was removed by crossing with Pgk1-Flp recombinase transgenic mice to generate the Tff2^fl/fl mice, which were mated to transgenic Pdx1-Cre mice to delete Tff2 in pancreatic cells (Pdx1:Cre;Tff2^fl/fl).

Brightfield analysis of E16.5 embryos showed no observable differences in pancreas size and architecture between mutants and controls (Fig. 1D and,Supplementary Fig. 5). Co-immunofluorescence (co-IF) staining of Tff2 and E-cadherin (E-cad, Cdh1) confirmed reduction of Tff2 protein in the pancreatic epithelium of mutants (Fig. 1E). Additionally, Tff2 transcripts were reduced in the pancreas (Fig. 1F) but not in the stomach (Fig. 1G) of E16.5 mutants, confirming specificity. Furthermore, Tff2 protein in E16.5 stomach was maintained (Supplementary Fig. 6), but reduced in duodenum in mutants (Supplementary Fig. 6), as Pdx1 is expressed in both the duodenum and pancreas (31). Together, these results demonstrate that Tff2 is specifically reduced in the pancreas and duodenum.

To clarify Tff2 protein expression patterns, co-IF analyses revealed that Tff2 was absent in β-cells (Fig. 2A) and endocrine cells (Supplementary Fig. 7A). Instead, Tff2 colocalized with acinar cells (Fig. 2B and C) but not ductal cells (Fig. 2D). Colocalization of an epithelial cell marker (EpCAM) with Tff2 in P7 pancreas, duodenum, and stomach validated the specificity of the anti-Tff2 antibody (Fig. 2E and Supplementary Fig. 7B and C). Together, these results demonstrate that Tff2 protein is expressed preferentially in the murine acinar cells compared with β-cells or ductal cells.

To gain insights into the consequence of Tff2 knockout, genome-wide gene expression analysis was performed using bulk mRNA-sequencing (RNA-seq) on E16.5 pancreata. Multidimensional scaling analysis separated mutants from controls, suggesting distinct gene expression patterns (Supplementary Fig. 8A). A total of 1,143 differentially expressed genes (DEGs) were identified, with 695 upregulated and 448 downregulated (Fig. 3A, Supplementary Fig. 8B, and Supplementary Data Set 1). Gene set enrichment analysis (GSEA) of DEGs using Hallmark molecular signature database identified downregulated biological pathways: pancreatic β-cells, oxidative phosphorylation, and DNA repair (Fig. 3B). In contrast, upregulated pathways included epithelial mesenchymal transition, mitotic spindle, and G2M checkpoint (Fig. 3B). Another computational tool, Database for Annotation, Visualization and Integrated Discovery, identified downregulated biological processes in mutants, including peptide transport, hormone secretion, endocrine pancreas development, maturity onset diabetes of the young, and type II diabetes mellitus (T2D). Upregulated pathways in mutants included nervous system development, cell adhesion molecules, and phagosome (Supplementary Fig. 8C and D). These analyses suggest that early knockout of Tff2 impacts a wide range of biological processes in the embryonic pancreas.

Because pancreatic β-cells was the most significant downregulated pathway in the mutant embryos (Fig. 3B), we examined endocrine marker expression. Tff2 was the most downregulated gene (Fig. 3C), confirming the knockout. Many downregulated genes were known endocrine lineage markers, such as Ins1, Ins2, Gcg, Ppy, and Sst (Fig. 3C). Additional β-cell–related genes were also decreased in mutant pancreas samples, such as Rfx6, Pax4, Pax6, Slc2a2, Iapp, Nkx6-1, Abcc8, Insm1, and Syt13 (Fig. 3D). qRT-PCR analysis confirmed the reduction of Ins1, Ins2, Gcg, Ppy, Sst, Chga, Iapp, Abcc8, and G6pc2 in mutant pancreata (Fig. 3E). The co-IF staining revealed that the ratios of total endocrine cell (Chga⁺) (Fig. 3F and H) and β-cell (Insulin⁺) area to total pancreas area (Fig. 3G and I) were reduced in mutants. The pancreas area (E-cad⁺) was not different between mutant and control embryos (Fig. 3J) nor was the percent sampling size (Supplementary Fig. 9). The ratio of ductal area (DBA⁺) to total pancreas area was also not different (Fig. 3K). Together, these results demonstrate a loss of the endocrine compartment at E16.5 in response to Tff2 knockout.

Given the reduction in the embryonic β-cells at E16.5 (Fig. 3), we anticipated hyperglycemia in mutant mice after birth. Surprisingly, P7 mutant mice exhibited normal nonfasting blood glucose level, body weight, and pancreas weight (Fig. 4A–C). Mice were born in a Mendelian inheritance ratio with no observable changes in gross anatomy of the P7 mutant pancreas (Fig. 4D). Similar to E16.5, Amylase⁺ acinar, DBA⁺ ductal cells (Fig. 4E–G and Supplementary Fig. 10A), and pancreas area (Supplementary Fig. 10B) were not different in P7 mutants. However, unlike E16.5, the endocrine cells in P7 pancreas were no longer reduced in mutants (Fig. 4H and I). Also, no difference in Ki67⁺ proliferating cells among mutant and control endocrine cells was observed (Fig. 4J).

Gene compensation is known to occur when mutations are deleterious (32). Given the similarities between the trefoil family proteins (14) and the known compensation of Tff3 in the stomach of a Tff2 global knockout mouse model (33), potential compensation by Tff1 and/or Tff3 in the mutant pancreas was examined. Tff3, but not Tff1, was upregulated in P7 mutant pancreas, and Tff2 remained reduced (Fig. 4K). To further clarify, P7 pancreatic islets and exocrine tissues were separately examined: the only changes found were decreased Tff2 and increased Tff3 in the mutant exocrine compartment (Fig. 4L). IF staining confirmed the absence of Tff2 (Fig. 4M) and the presence of Tff3 protein (Fig. 4N) in the exocrine pancreas of P7 mutant mice. Antibody staining for Tff2 and Tff3 in positive control tissues confirmed specificity (Fig. 4O and Supplementary Fig. 10C).

Further examination of P0 pancreas revealed no difference in body weight, pancreas weight, gross anatomy, Tff3 expression, endocrine pancreas area, proliferation of endocrine cells, or islet cell mass (Supplementary Fig. 10D–N). Together, these results indicate that early loss of Tff2 in the embryonic pancreas leads to recovery of endocrine cell mass by P0 and Tff3 compensation in acinar cells by P7.

Despite Tff3 compensation in the pancreas of P7 mutant mice, we hypothesized that the loss of Tff2 may impact endocrine function because 1) Tff2 has a much higher (2,722 ± 797-fold; n = 3 each group; range 1,826–4,311) gene expression level compared with Tff3 in 6-week pancreata (Fig. 1A), and 2) Tff2 transcript in acinar cells is found to be highest among 20 organs from adult mice in scRNA-seq analyses (12) (Supplementary Fig. 11). Therefore, homozygous (Pdx1:Cre;Tff2^fl/fl) and heterozygous (Pdx1:Cre;Tff2^fl/+) mutants and controls (Tff2^fl/fl) were monitored from 10 to 60 weeks of age. Body weight increased at 60 weeks for both male (Fig. 5A and Supplementary Fig. 12A) and female (Fig. 5B and Supplementary Fig. 12B) homozygous mutants compared with heterozygous and control mice. Fasting blood glucose levels were elevated >200 mg/dL at 40 weeks in male (Fig. 5C), but not female (Fig. 5D), homozygous mutant mice compared with heterozygous and control mice, suggesting females were protected. However, by 60 weeks, both male and female homozygous mutants had elevated fasting blood glucose compared with heterozygous mutants and controls (Fig. 5C and D and Supplementary Fig. 12C and D). Glucose tolerance, assayed by intraperitoneal glucose tolerance test (IP-GTT), was not different at 20 weeks of age (Fig. 5E and F). However, at 40 weeks, male (Fig. 5G) but not female (Fig. 5H) homozygous mutants became glucose intolerant, again indicating various degrees of impact by sex. By 60 weeks of age, male heterozygous mutants also became glucose intolerant (Fig. 5I), demonstrating Tff2 haplodeficiency impacts male mice. Female homozygous mutants also became glucose intolerant by 60 weeks (Fig. 5J).

To determine whether the glucose intolerance of mutant mice was due to insulin resistance, an intraperitoneal insulin tolerance test was performed. No difference was observed for both males and females (Supplementary Fig. 12E and F), indicating that insulin sensitivity in the peripheral tissues is conserved in homozygous and heterozygous mutant mice. Together, these results demonstrate the requirement of pancreatic Tff2 in maintaining glucose homeostasis during aging. Based on these results, we focused our attention on homozygous mutant male mice at ∼1 year of age in subsequent studies.

Decreased glucose clearance (Fig. 5I and J) with unchanged insulin sensitivity (Supplementary Fig. 12E and F) suggested a defect in β-cells in 1-year-old mutants. Tff2 knockout was first confirmed in acinar cells of homozygous mutants (Supplementary Fig. 13A and B). Surprisingly, pancreas weight and pancreas-to-body weight ratio were reduced for both mutant males (Fig. 6A and B) and females (Supplementary Fig. 13C and D).

To determine changes in islets, morphometric analysis was performed with a 5% sampling rate (Supplementary Fig. 13E). Hematoxylin and eosin staining (Fig. 6C) first confirmed that total pancreas area (Fig. 6D) and exocrine cell mass (Fig. 6E and Supplementary Fig. 13F) were reduced in male homozygous mutants. However, the islet-to-pancreas ratio (Supplementary Fig. 13G) and islet cell mass (Fig. 6F) were not different. Next, insulin colorimetric staining and morphometric analysis (Fig. 6G) revealed lowered total pancreas area (Fig. 6H) without changes in insulin-to-pancreas area ratio and β-cell mass (Fig. 6I and J) in mutant mice. Together, these results indicate that early loss of Tff2 perturbs pancreas size without altering islet and β-cell mass during aging.

To determine changes in β-cell function, we isolated islets from aged male mice and performed an in vitro GSIS assay; islets were sequentially treated with low (2.8 mmol/L), high (20 mmol/L), and low concentrations of d-glucose, followed by 30 mmol/L KCl to stimulate maximal release of insulin (Fig. 6K and L). As expected, islets isolated from control mice secreted more insulin in response to high glucose or KCl compared with low glucose (Fig. 6L, left). Islets from homozygous mutant mice also responded to high glucose or KCl, but the levels of insulin secreted were reduced (Fig. 6L, right). The protein content was not different between the mutant and control islets (Fig. 6M), indicating equivalent islet mass was used. These results indicate that static insulin secretion is impaired in the homozygous mutant islets.

To discern differences in insulin secretion dynamics, in vitro perifusion GSIS was performed. Insulin secretion was recorded every minute during glucose concentration changes, allowing quantification of the first and second phase of insulin secretion (Fig. 6N). As expected (28), control islets reached a peak in insulin secretion within 5 min of high-glucose stimulation, followed by a reduced yet sustained second phase that persisted in the presence of glucose (Fig. 6N). The areas under the curve (AUCs) for the first phase (14–20 min) trended lower but did not reach significance in the homozygous mutants (Fig. 6O, 1st phase). However, AUC of the second phase (22–45 min) was ∼45% lower in islets from homozygous mutants (Fig. 6O, 2nd phase), suggesting the recruitment of insulin granules from the internal storage pool is affected in β-cells. Tff2 was reduced in the exocrine fraction post–islet isolation for the perifusion experiment (Fig. 6P, left), whereas Tff3 expression remained higher in homozygous mutant exocrine fractions (Fig. 6P, right). Overall, these results suggest that loss of exocrine Tff2 impacts the second phase of insulin secretion in aged mouse β-cells.

To further characterize global gene expression changes in islets from homozygous mutants (n = 5 male mice, each group), bulk RNA-seq was performed. Surprisingly, only 22 DEGs were identified, with 15 genes upregulated and 7 genes downregulated (Fig. 7A and B  and Supplementary Data Set 2). The exocrine fraction from homozygous mutant pancreata was confirmed to express lower levels of Tff2 and higher levels of Tff3 (Supplementary Fig. 14A and B). GSEA using Hallmark and Gene Ontology, Biological Processes molecular signature databases revealed downregulated pathways including protein secretion, Golgi vesicle transport, vesicle organization, establishment of protein localization to organelle, and vesicle localization (Fig. 7C and D). In contrast, upregulated pathways included TNFα signaling via NFκB, oxidative phosphorylation, glycolysis, cytoplasmic translation, inflammatory response, and leukocyte chemotaxis. This analysis suggests that reduced vesicle transport to the plasma membrane and increased metabolism and inflammatory response affects islets from homozygous mutant mice.

Melanophilin (Mlph) was identified as the most downregulated gene (Fig. 7B). Mlph in β-cells plays key roles in the second phase of insulin secretion (34). Co-IF analysis revealed that Mlph was coexpressed with insulin in aged control islets (Fig. 7E, top), as expected. However, aged mutant islets exhibited reduced Mlph expression among β-cells (Fig. 7E, bottom, and 7F). Together, these results demonstrate that the absence of pancreatic Tff2 correlates with reduced Mlph in aged β-cells.

In this study, we generated a conditional Tff2 knockout mouse model and demonstrated the critical physiological role of Tff2 in the pancreas. Because of Tff2’s expression in multiple organs, investigating the pancreas in the global Tff2 knockout mice would have introduced confounding factors, especially from stomach, pituitary, and hypothalamus, which are known to participate in the metabolism of the whole body (35). We found that Tff2 loss in the pancreas resulted in a reduction of the endocrine compartment including the β-cells in E16.5 pancreas (Fig. 3). Upon reaching 1 year of age, homozygous mutant mice (both male and female) eventually developed glucose intolerance (Fig. 5) and showed defects in the second phase of GSIS (Fig. 6). Mutant males exhibited IGT much earlier than female mutants (Fig. 5), suggesting some impact by sex. This may be explained by the estrogen hormone produced by females that is known to protect against the development of metabolic syndrome, obesity, T2D, and insulin resistance in the C57BL/6 strain of mice (36).

Interestingly, endocrine mass recovered as early as P0 (Supplementary Fig. 10N), while Tff3 compensation in acinar was observed starting in P7 (Fig. 4L) and continued in 1-year-old homozygous mutant pancreas (Fig. 6P). These results suggest that Tff3 upregulation is not directly responsible for endocrine recovery prior to birth. However, the upregulated Tff3 in acinar cells (Supplementary Fig. 10N and Fig. 4L) paired with the Tff3 in islet cells (Fig. 4L and N) may be sufficient to maintain glucose metabolism in mutant mice from P7 until 1 year of age because of Tff3′s effects on β-cell proliferation (10) and insulin sensitivity (37).

Although endocrine cell mass was preserved, pancreas size was reduced in the aged homozygous mutant mice (Fig. 6), suggesting requirement of Tff2 in maintaining exocrine cell mass. This observation is echoed in a prior study by Kodama et al. where Pdx1 was specifically knocked down in acinar cells (38); these mice developed smaller exocrine pancreas and β-cell dysfunction. Interestingly, a follow-up study from the same group found that the smaller exocrine tissue was associated with reduced Tff2 (9). Our study provides causal evidence of Tff2 deficiency in exocrine size reduction. Pancreas weight and area was not changed at P0 (Supplementary Fig. 10E and K) or P7 (Fig. 4C and Supplementary Fig. 10B) in our mutant mice. Thus, pancreas size reduction occurs between P8 and 1 year, which can be influenced through proliferation in the exocrine pancreas during postnatal growth (P7–P28) (39), or a secondary effect of the lowered insulin trophic factor for acinar cells (40) from T1D or T2D islets (41,42).

We caution that Tff2 deficiency in the duodenum (Supplementary Fig. 6B) may contribute to body weight increase (Fig. 5) and pancreas size decrease (Fig. 6) in 1-year-old mutants. For example, duodenal ulcers can lead to weight gain due to postprandial pain relief (43), and duodenal obstruction may contribute to pancreatitis, affecting pancreas organ size (44). However, we did not observe morphological characteristics indicative of pancreatitis in aged mutants. Interestingly, our findings on weight gain differ from those reported by De Giorgio et al., who demonstrated that Tff2 null mice were protected from weight gain, because of increased energy expenditure and locomotor activity (23). This discrepancy may be attributed to confounding effects between global versus conditional Tff2 knockout mice.

Using co-IF staining, we clarified that Tff2 protein expression in the pancreas, between E16.5 and 6 weeks of age, is restricted to acinar cells (Fig. 2). Orime et al. demonstrated strong Tff2 protein expression in the islets of adult (10–14 weeks) murine pancreas (25). Using a different antibody, Hirata et al. showed Tff2 protein expression in the exocrine (both acinar and ductal cells) pancreas at E16.5, E18.5, and P1 (9); however, co-IF was not done to confirm lineage. We used the same commercially available antibody as Hirata et al., but with a higher dilution, and in combination with lineage markers. Hirata et al. further investigated the gene localization of Tff2 using in situ hybridization, and found it expressed only in the acinar cells in 8-week-old pancreas (9). scRNA-seq analysis of 20 adult (10–15 weeks old) murine organs showed stronger expression of Tff2 in pancreatic acinar compared with ductal cells (Supplementary Fig. 11C), and an absence of Tff2 in islet cells (Supplementary Fig. 4B) (12). We found that Tff2 was only expressed in the exocrine tissue (Fig. 4L and M). More recently, constitutive Tff2-Cre mice were generated that show embryonic lineage tracing to adult acinar but not endocrine or ductal cells (45). Notably, the aforementioned studies also utilized mice with C57BL/6 background, which indicates that the differences we observed in expression patterns were not due to strain differences. Taken together, we conclude that Tff2 is preferentially expressed by the exocrine acinar cells in the murine pancreas.

Transcriptomic analysis of aged islets from homozygous mutants versus control led to the identification of melanophilin (Mlph), a leading-edge gene in the vesicle localization pathway that is downregulated in mutants (Fig. 7B and D). Mlph functions in membrane trafficking, and, in skin, melanocytes form a complex with a GTPase, Rab27a, and myosin-Va on cortical actin filaments to direct the transfer of melanosomes (organelles that carry pigment) to the plasma membrane for secretion to neighboring keratinocytes (46). Mutations in Mlph cause problems with melanosome transportation that result in leaden (light-colored) mice compared with control mice with dark color (47). Additionally, leaden mice exhibit IGT due to reduced Mlph-associated insulin granule transportation to the plasma membrane during the second phase of insulin secretion (34,48). Our Tff2-deficient aged mice demonstrate the same reduction in the second phase of insulin secretion (Fig. 6N and O), which may be explained by the downregulation of Mlph (Fig. 7B and F). Future mechanistic studies are underway to clarify how the absence of Tff2 in the mouse pancreas leads to a downregulation of Mlph in aged β-cells. Interestingly, Mlph was also downregulated in the E16.5 mutants compared with controls (Fig. 3C), potentially through a different mechanism because E16.5 β-cells are not yet functional in GSIS (49).

Our results add to a growing number of studies that demonstrate that signals from the exocrine pancreas can affect the endocrine pancreas. For example, the risk of developing diabetes is increased by exocrine diseases, such as cystic fibrosis (3,4), pancreatitis (5,6), and pancreatic cancer (7,8). Also, patients with a mutation in an acinar-specific gene, carboxyl ester lipase, known as MODY8 gene, develop diabetes through secretion of the toxic mutant carboxyl ester lipase that is uptaken by the β-cells, causing dysfunction (50). Our study now identifies Tff2 as a beneficial protein, secreted from exocrine tissue, required for maintaining β-cell function.

In summary, we demonstrate that Tff2 is an exocrine regulator required for maintaining β-cell number during embryo development and sustaining β-cell secretory function during aging. Our finding has implications for diabetes disease progression and therapy.

Acknowledgments. The authors appreciate services provided by the following research cores at the City of Hope: Pathology, Light Microscopy Digital Imaging, Animal Resource Center, Integrative Genomics, and Comprehensive Metabolic Phenotyping. Some illustrations and the graphical abstract were created using BioRender.com.

Funding. Pathology, Light Microscopy Digital Imaging, Animal Resource Center, Integrative Genomics, and Comprehensive Metabolic Phenotyping are supported in part by the City of Hope Research Facilities grant P30CA033572 from the National Institutes of Health. J.A.O. received predoctoral fellowships from the Ford Foundation and from the Helen & Morgan Chu Fellowship at the Irell and Manella Graduate School of Biological Sciences. This project was supported in part by a City of Hope Arthur Riggs Diabetes & Metabolism Research Institute pilot grant and a grant from the National Institutes of Health, R01DK134652, to H.T.K., P.T.F., and D.C.T. Support from anonymous donor S.S. to H.T.K. is also gratefully acknowledged.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Author Contributions. J.A.O. and H.T.K. conceived and planned the study. J.R., F.K., P.T.F., and D.C.T. provided key reagents. J.A.O., N.G., K.L., J.R., E.M.M., E.O., J.M.I., K.J., J.M., M.-H.C., X.W., H.N.Z., J.C.Q., N.E., and A.L. performed the experiments. J.A.O., J.M.I., J.M., M.-H.C., X.W., H.N.Z., J.C.Q., N.E., and H.T.K. performed the analyses. J.A.O., P.T.F., D.C.T., and H.T.K. interpreted data. J.A.O. and H.T.K. wrote the manuscript. J.A.O., J.M., H.N.Z., J.C.Q., P.T.F., D.C.T., and H.T.K. edited and revised draft manuscript. P.T.F., D.C.T., and H.T.K. acquired funding. H.T.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.

Prior Presentation. Part of this work was presented at the Society of Developmental Biology National Conference, virtual meeting, 14 July 2021, and the National Institutes of Health–National Institute of Diabetes and Digestive and Kidney Diseases Integrated Physiology of the Exocrine and Endocrine Compartments in Pancreatic Diseases Workshop, Bethesda, MD, 29–30 June 2022.