Translating the blueprint of cell fate: eIF5A-mediated translation regulates cellular identity in the pancreas

Abstract

Cellular identity is fundamentally determined by the precise regulation of protein synthesis, which governs growth, differentiation, and function. In the pancreas, the balance between exocrine and endocrine cell types is critical for organ function, and the disruption of protein synthesis in these cells can lead to diseases such as exocrine insufficiency and diabetes. The specialized mRNA translation factor eukaryotic initiation factor 5A (eIF5A) has emerged as an essential regulator of on-demand protein synthesis in professional secretory cells. Here, we investigate the role of eIF5A-mediated mRNA translation in lineage specification during pancreas development. Using genetic mouse models, our studies reveal that loss of eIF5A results in a marked reduction of exocrine volume and a paradoxical expansion of the insulin-producing beta cell population. We reveal that these cellular changes are driven by impaired on-demand protein synthesis during the critical stage of pancreatic cell differentiation. Mechanistically, we show that eIF5A deficiency disrupts the synthesis of proteins critical for proper pathway signaling—most notably Notch—that instruct cell fate decisions. As a result, we observe impaired ductal branching and tip formation as well increased Ngn3+ endocrine progenitors within the ducts. These changes in lineage allocations directly contribute to decreased acinar cell and increased beta cell mass. Remarkably, eIF5A-deficient mice maintain elevated beta cell mass and exhibit preserved glucose tolerance despite severe exocrine deficiency. Collectively, our findings establish that eIF5A-mediated mRNA translation regulates critical developmental signaling pathways and reinforces the finding that disruptions in protein synthesis can reprogram cellular identity and drive disease pathogenesis.

Highlights

• •We show that eIF5A, which mediates “on-demand” translation, regulates lineage decisions during development by driving synthesis of specific proteins.
• •Eif5a deletion in the developing pancreas results in fewer exocrine cells but paradoxically an expanded beta cell population.
• •Without eIF5A-mediated translation, signaling pathways including Notch are disrupted, which upsets endocrine-exocrine progenitor cell balance.
• •Mutant mice maintain normal blood sugar despite severe digestive enzyme deficiency—a pattern observed in certain human pancreatic disorders.
• •Our study reveals that translational control, not just gene regulation, determines cellular identity during development.

1. Introduction

Cellular identity is fundamentally determined by the differential expression of specific proteins that govern growth, differentiation, and functional capacity. In experimental and clinical contexts, cell types are typically defined by the presence or absence of proteins that are important for development and function or that characterize their healthy state. Resultantly, a change in the expression of these critical proteins frequently underlies cellular dysfunction and disease. Pancreatic pathologies can exemplify this relationship: diabetes is characterized by the loss of beta cells and consequent reduction in serum insulin levels, while exocrine pancreatic insufficiency results from diminished acinar cell function and reduced trypsinogen [[1], [2], [3], [4], [5], [6]]. These examples underscore the central role of protein synthesis in maintaining cellular homeostasis and highlight how the dysregulation of protein expression directly contributes to pathology. In short, cellular identity and protein synthesis are intimately linked in both health and disease.

Unquestionably mRNA translation is universally essential; however, each cell type may have a specific timing on need. Stem cell studies provide evidence of a flexibility in the programs that control quiescence, proliferation, self-renewal, and differentiation, which are clearly dictated by changes in protein expression. In line with this, mouse embryonic stem cells (mESCs) are reported to show changing rates of translation as the mESCs transition from activated stem cells to progenitor cells that initiate differentiation to terminally differentiated cells [7]. Furthermore, genetic disease research has provided data that interrupting mRNA translation during development can have tissue-specific effects. For example, Schwachman-Diamond syndrome (SDS) is a rare autosomal recessive disorder caused by mutations in the SBDS or EFL1 genes, which are involved in translation activation, resulting in a disease that is characterized most prominently by exocrine pancreatic insufficiency [8,9]. Other diseases that impair protein synthesis show a similar cell type-specific impact [10], highlighting that certain cells may have a greater sensitivity to altered translation, which could cause a loss of cell identity and/or onset of disease in both the developmental and postnatal setting.

Professional secretory cells such as pancreatic beta cells and acinar cells are specific examples of cells that must maintain substantial protein production capacity and therefore would be particularly susceptible to even subtle disruptions in translation. Our previous work discovered that the specialized mRNA translation factor eukaryotic initiation factor 5a (eIF5A) regulates on-demand protein synthesis in the pancreatic beta cell [11]. eIF5A is unique because it must be post-translationally modified to become active and perform its function in mRNA translation. This post-translational modification is catalyzed by the rate-limiting enzyme deoxyhypusine synthase (DHPS), which uses the polyamine spermidine as a cofactor to modify the lysine at position 50 of eIF5A in a process known as hypusination [12]. The result of hypusination of eIF5A is the activation of its mRNA translation function [12,13]. In mouse models generated by our lab, eIF5A-mediated translation has been linked with regulating the translation of specific proteins in beta cells [11], acinar cells [14], and neurons [15]; all of these cell types are characterized by a need to produce high levels of specific proteins at specific times in order to perform their functions. Given that cellular identity is fundamentally determined by the differential expression of specific proteins at specific times, we speculated that eIF5A-mediated translation may also drive lineage decisions during development.

The lineage decisions that give rise to exocrine and endocrine cells during pancreas development involve a tightly regulated progression, where epithelial cells transition from pancreatic progenitors to lineage-specific progenitors and then to their differentiated states. While numerous transcription factors and signaling pathways have been implicated in orchestrating these fate choices [16], the spatial organization of progenitors within the pancreatic epithelium also plays an important role [[17], [18], [19]]. For example, at the onset of the secondary transition around embryonic day 12.5 (E12.5) in the mouse, the epithelium begins to segregate into peripheral tips and centralized trunks, which contributes to further differentiation. Cells in the tips, which upregulate Ptf1a and Cpa1, will differentiate into acinar cells, whereas certain cells within the Sox9-expressing trunk epithelium will upregulate Ngn3 to give rise to endocrine progenitors and subsequently all islet cell lineages [20,21]. In addition to tip-trunk segregation influencing lineage differentiation, the influence of signaling pathways such as Notch [22] and the establishment of apical-basal polarity have been shown to promote the endocrine fate as well as the formation of tip cells [23,24]. These studies highlight the contribution of spatial organization within the epithelium to lineage specification.

In this study, we tested the hypothesis that on-demand protein synthesis regulates the signals that drive cell fate decisions in the developing pancreas. We generated a genetic mouse model of Eif5a deletion in the developing pancreas and observed exocrine insufficiency accompanied by a paradoxical expansion of the beta cell population. Our experiments further revealed that impaired eIF5A-mediated mRNA translation during pancreatic differentiation reduced the synthesis of specific proteins and pathways critical for directing the differentiation of the endocrine versus exocrine cell lineages. We show that impaired ductal tip and branch formation is accompanied by an epithelium containing an increase in endocrine progenitors. The timing of these alterations during a critical window of differentiation in the developing pancreas resulted in changes in pancreatic cell type quantity (increased beta cells and reduced acinar cells) and produced postnatal exocrine insufficiency without endocrine dysfunction. Taken together, our studies have deciphered how pancreatic lineage differentiation is regulated by eIF5A-mediated translation and that impairment of this protein synthesis regulatory mechanism precipitates disease.

2. Materials and methods

2.1. Animal husbandry

Animal studies were approved by the Indiana University School of Science Institutional Animal Care and Use Committee. Mice containing the Eif5a^LoxP allele (B6.Cg-Eif5a^tm1.1Tlm/J) [14] were mated with mice carrying the Ptf1a-cre driver (Ptf1a^{tm1.1(cre)Cvw}) [25] and R26R^Tomato reporter allele (B6.Cg-Gt(ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J) [26] and subsequent interbreeding resulted in the generation of mutant mice (Eif5a^LoxP/LoxP;Ptf1a-cre;R26R^Tomato; hereon denoted eIF5A^ΔPANC). Control mice (Ptf1a-cre;R26R^Tomato) from these matings were used for comparison. Mice containing either the Eif5a^LoxP allele (B6.Cg-Eif5a^tm1.1Tlm/J) [14] or the R26R^DTA allele (Gt(ROSA)^{26Sortm1(DTA)Lky}) [27] were mated with mice carrying the Ptf1a-cre^ERTM driver (Ptf1a^{tm2(cre/ESR1)Cvw}/J) [28] and R26R^Tomato reporter allele (B6.Cg-Gt(ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J) [26] to generate pregnant dams for analysis. Subsequent breeding resulted in the generation of mutant mice (Eif5a^LoxP/LoxP;Ptf1a-creER;R26R^Tomato; hereon denoted eIF5A^ΔEXO and Ptf1a-cre^ERTM;R26R^DTA;R26R^Tomato; hereon denoted ΔEXO) as well as control mice (Ptf1a-cre^ERTM;R26R^Tomato) for comparison. All mice were maintained on a mixed background, ad libitum fed chow diet (5001, LabDiet), and kept on a 12-h light/dark cycle. At 3 weeks-of-age mice were weaned, ear-tagged, and tail-biopsied for genotyping. For postnatal experiments, mutant and control mice were allowed to reach experimental age and then were dissected for tissue. Body weight and blood glucose were measured on the morning of dissection using a digital scale (Fisher) and AlphaTrak2 or AlphaTrak3 glucose monitor and test strips (ADW Diabetes). For embryonic dissections, timed matings were established, and females were checked for seminal plugs before 9am the following day. If a plug was observed, noon that day was considered E0.5.

2.2. Genotyping

DNA was isolated from tail snips. Tail snips were digested in 400 μL of tail lysis buffer (0.5 mL 1M Tris pH 8.0; 50 μL 0.5M EDTA; 100 μL 10% SDS; 200 μL 5M NaCl; 4.1 mL ddH2O) and 40 μL proteinase K (10 mg/ml) (BP1700-500, Fisher) at 55 °C overnight. Subsequently, digested tails were centrifuged at 15,000 rpm for 8 min, the supernatant transferred to a new tube containing 400 μL isopropanol (A4154, Fisher) for DNA precipitation; purified DNA was resuspended in 400 μL of MilliQ water. Genotyping was performed as previously described using published primer sets [14,[26], [27], [28], [29], [30]].

2.3. Tissue collection, preservation, and cryosectioning

For embryonic tissue analysis, pregnant mice were euthanized via cervical dislocation and the embryos were removed and fixed in 4% paraformaldehyde (PFA) (41678-5000, Acros Organics) overnight at 4 °C. Following fixation, embryos were washed twice in cold PBS (BP3994, Fisher) for 15 min. After washing, embryos were cryopreserved in 30% sucrose (BP2201, Fisher) at 4 °C overnight. The next day, tissue was rocked first for 15 min in a 50:50 solution of optimal cutting temperature solution (OCT) (Fisher Scientific) and 30% sucrose, then for 15 min in OCT only. Embryos were embedded in OCT and stored at −80 °C. Embryos were cryosectioned (Leica CM1860) with consecutive tissue sections from across the entire pancreas collected onto superfrost plus microscope slides (Fisher Scientific) and then stored at −80 °C.

For postnatal tissue analyses, animals were euthanized via cervical dislocation and pancreata were dissected together with the spleen. Dissected tissues were fixed and processed using the same method as described above. Cryosections of pancreas tissue were collected at 8 μm thickness onto superfrost plus microscope slides. To sample the entire pancreas, twenty 8 μm sections were collected from each level across the entire pancreas, with 150 μm trim cuts between each level, resulting in approximately 10 levels per pancreas. Slides were stored at −80 °C until use.

2.4. Immunofluorescence and morphometric analysis

Slides with embryonic or postnatal pancreas sections were equilibrated to room temperature and washed in 0.3% hydrogen peroxide followed by two washes in PBS with Triton (PBST). A hydrophobic pen was used to create a barrier around the tissue, which was then incubated in 200 μL of blocking solution (20 μL PBS:1 μL donkey serum; D9663, Sigma) at room temperature for 1 h. Supplemental Table 1 (Supporting Information) contains antibodies and staining reagents. Primary antibodies were diluted in blocking solution, applied to slides, and incubated overnight at 4 °C. The following day, the slides were washed in PBST three times for 5 min each. Fluorophore-conjugated secondary antibodies (Supplemental Table 1) were then diluted in blocking solution, applied to slides, and incubated at room temperature for 2 h. Following incubation, secondary antibodies were removed, and a nuclear stain was applied and incubated at room temperature for 30 min. Slides were washed in PBS and coverslipped using fluorogel mounting media (Electron Microscopy Sciences). Each pancreas section was imaged using a tile scanning function on the Nikon C2 confocal microscope at 10X, 20X, or 40X. For tissue from E11.5 embryos, every 3rd section was evaluated across the entire pancreas, which resulted in 6 sections analyzed per embryo. For tissue from E14.5 or E18.5 embryos, every 10th section was evaluated across the entire pancreas, which resulted in 7 sections analyzed per E14.5 embryo and 10 sections analyzed per E18.5 embryo. For tissue from 2- and 4- week-old animals, one section from every level was evaluated across the entire pancreas, which resulted in 10 sections analyzed per pancreas. Animal weight was used to normalize data given this was consistent between genotypes.

For analysis of the multipotent progenitor cells and microlumen area, tissue sections from E11.5 Ptf1a-cre and eIF5A^ΔPANC embryos were stained using primary antibodies against Pdx1, Mucin1, Hes1, or E-cadherin. Pdx1+ cell counting, and ductal lumen area measurements were performed using QuPath software (https://qupath.github.io/). Hes1+ cell counting was performed using NIS Elements Software V5.42 (Nikon Corporation). For cellular analyses at E14.5, tissue sections from Ptf1a-cre and eIF5A^ΔPANC embryos were stained with combinations of primary antibodies including Carboxypeptidase A (CPA), Sox9, Neurogenin3 (Ngn3), Ptf1a, Nkx6.1, Hes1, and E-cadherin. The analyses of E-cadherin-expressing area, tip cell organization, or Ngn3-expressing cell abundance were performed using NIS Elements Software V5.42 (Nikon Corporation). Cellular measurements at E14.5 were normalized to pancreas area. For cellular analyses at E18.5, tissue sections from Ptf1a-cre, Ptf1a-cre^ERTM, eIF5A^ΔPANC, eIF5A^ΔEXO, and ΔEXO embryos were stained with combinations of primary antibodies including insulin, glucagon, somatostatin, amylase, and carboxypeptidase A (CPA). Insulin+, glucagon+, somatostatin+, amylase+, and CPA + absolute area measurements were analyzed using NIS Elements Software V5.42 (Nikon Corporation). Cellular measurements at E18.5 were presented as absolute area given that normalizing to pancreas area when there was a 50% loss of acinar cells would skew the data. For morphometric analysis of pancreas tissue from 2- and 4-week-old Ptf1a-cre and eIF5A^ΔPANC animals, primary antibodies against insulin, glucagon, somtatostatin, and amylase were used. Insulin+, glucagon+, somatostatin+ and amylase+ area was measured using NIS Elements Software V5.42 (Nikon Corporation) and normalized to animal weight. Statistical significance was determined using a student's t test (GraphPad Prism 10, GraphPad Software).

2.5. Tamoxifen administration to mice

Tamoxifen (T5648, Sigma) was dissolved in peanut oil (P2144, Sigma) to generate a stock solution with a concentration of 25 mg/mL. Tamoxifen was administered to pregnant dams on gestational day E14.5 at a dosage of 10 mg/40 g body weight using a U-100 BD Insulin Syringe (14-826-79, Fisher).

2.6. Metabolic tests

Fed blood glucose and body weight were measured from the tail vein of each animal weekly from 2 weeks to 2 years-of-age using an AlphaTrak2 glucose monitor. Glucose tolerance tests (GTT) were performed following a 5 h fast. Fasted blood glucose was measured at timepoint zero of the GTT. Following intraperitoneal administration of 1.5 g/kg glucose to males or 2 g/kg glucose to females, blood glucose measurements were taken at 15, 30, 45, 60, 90 and 120 min. Data were analyzed using a two-way analysis over variance (ANOVA) test. The blood from timepoint zero of the GTT was collected in an EDTA coated tube, processed for serum, and stored in the −80 °C until use. Serum insulin and glucagon were measured by ELISA. Unpaired t-test was performed for comparison between Ptf1a-cre and eIF5A^ΔPANC animals. All statistical analysis was carried out using Prism 10 (GraphPad).

2.7. Quantitative mass spectrometry and pathway analysis

Whole E18.5 pancreata were isolated from embryos and subjected to proteomic analysis by mass spectrometry. Pancreata were collected from eIF5A^ΔPANC, Ptf1a-cre, eIF5A^ΔEXO, and Ptf1a-cre^ERTM embryos. Pancreata from two individual embryos were pooled together to generate one sample. Sample preparation and mass spectrometry were performed in collaboration with the Indiana University Center for Proteome Analysis at the Indiana University School of Medicine, as previously published [14].

To analyze the proteomic enrichment and to determine significantly altered pathways, differentially expressed proteins (DEPs) were classified as significant with a P < 0.05. The following groups were compared using the DAVID bioinformatics tool (https://davidbioinformatics.nih.gov/tools.jsp): eIF5A^ΔEXO vs. Ptf1a-cre^ERTM, and eIF5A^ΔPANC vs. eIF5A^ΔEXO, and altered KEGG pathways were determined. Data from the eIF5A^ΔEXO mutants and Ptf1a-cre^ERTM controls was deposited in the MassIVE repository with accession MSV00010074. Data from the eIF5A^ΔPANC mutants and Ptf1a-cre controls was previously published ([14]; MassIVE accession MSV000099697).

2.8. Quantitative real time (RT) PCR analysis

Real-time PCR was performed as previously described [11,31]. Briefly, pancreata were isolated from E14.5 embryos and RNA was extracted using the Rneasy Micro Kit (QIAGEN). Isolated RNA was first subjected to the NanoDrop to assess RNA quantity and subsequently submitted for analysis using an Agilent 2100 Bioanalyzer at the Center for Medical Genomics at Indiana University School of Medicine to assess RNA quality (fragmentation and degradation). Total RNA was reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (Fisher Scientific). Real-time PCR was performed using TaqMan Gene Expression Master Mix (Applied Biosystems) and TaqMan probes for Notch1 (catalog #: Mm00627185_m1), Hes1 (catalog #: Mm00468601_m1), Ngn3 (catalog #: Mm00437606_s1), Eif5a (catalog #: Mm00839121_gH), Ins1 (catalog # Mm01259683_g1), Gcg (catalog # Mm00801714_m1), Sst (catalog # Mm00436671_m1), and Luciferase (catalog #: Mr03987587_mr). Gene expression was normalized to the spiked-in Luciferase (1 μg/μL, L4561, Promega). The mRNA abundance was graphed as a relative expression. Statistical significance was determined using a Student t test (GraphPad Prism 10, GraphPad Software).

3. Results

3.1. Deletion of Eif5a in the developing pancreas results in decreased exocrine mass concomitant with increased beta cell mass

We previously showed that genetic deletion of Eif5a at the start of pancreatic organogenesis at E9.5 (Eif5a^loxP/loxP;Ptf1a-cre; denoted eIF5A^ΔPANC) resulted most significantly in a 50% reduction in exocrine mass by the end of development at E18.5 [14]. Moreover, we identified reduced synthesis of many proteins in the pancreas including a significant number of exocrine-related proteins, thereby implicating eIF5A in the translational regulation of exocrine pancreas growth and function [14]. Whereas islets were observed in the postnatal pancreas of eIF5A^ΔPANC animals [14], it remained undetermined if loss of eIF5A also influenced growth and function of the endocrine cells. Therefore, we evaluated E18.5 pancreata from eIF5A^ΔPANC mutants and Ptf1a-cre controls to determine if there were changes to the insulin-expressing beta cells, glucagon-expressing alpha cells, and somatostatin-expressing delta cells (Figure 1A–D). Unexpectedly, morphometric analysis revealed that eIF5A^ΔPANC mutants exhibited a significant increase in insulin-expressing cells compared with Ptf1a-cre controls (Figure 1E). In contrast, no significant difference was observed in the abundance of glucagon-expressing cells or somatostatin-expressing cells (Figure 1F–G). Real time PCR analysis confirmed increased insulin (Ins1) gene expression and no change in glucagon (Gcg) or somatostatin (Sst) gene expression at E18.5 in eIF5A^ΔPANC mutants compared with Ptf1a-cre controls (Figure 1H–J). Together these results suggest that loss of eIF5A-mediated translation may specifically influence beta cells.

Figure 1.

Loss of eIF5A in the embryonic pancreas results in an increased beta cell area at E18.5. Representative images of E18.5 (A) Ptf1a-cre control and (B) eIF5A^ΔPANC mutant pancreata stained for insulin (green) and amylase (red). Representative images of E18.5 (C) Ptf1a-cre control and (D) eIF5A^ΔPANC mutant pancreata stained for glucagon (green), somatostatin (red), and nuclei (DAPI; blue). (C′ and D’) Increased magnification to visualize areas with glucagon- and somatostatin-expressing cells. Quantification of absolute (E) beta cell area, (F) alpha cell area and (G) delta cell area in eIF5A^ΔPANC mutants and Ptf1a-cre controls. Real time PCR analysis of (H) Ins1, (I) Gcg, and (J) Sst gene expression in eIF5A^ΔPANC mutants and Ptf1a-cre controls. Data presented as mean +/− SEM; n = 3–5/group; ∗, p < 0.05; ∗∗, p < 0.01. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.2. The absence of eIF5A during the secondary transition, not the reduction in exocrine mass, drives beta cell growth

The increase in beta cell mass was observed later in development and concomitant with reduced exocrine volume; therefore, we questioned whether it was the loss of acinar cells or the absence of eIF5A that influenced beta cell growth. Moreover, given our previous finding that beta and alpha cells were present at E14.5 in similar quantity to controls [14], we rationalized that it might be changes during the secondary transition that influenced later development. Therefore, we generated two additional mouse models: a model of pancreatic exocrine ablation during the secondary transition (Ptf1a-cre^ERTM;R26R^DTA;R26R^Tomato, denoted ΔEXO) (Figure 2A) and a model of induced genetic deletion of Eif5a during the secondary transition (Eif5a^flox/flox;Ptf1a-cre^ERTM; R26R^Tomato, denoted eIF5A^ΔEXO) (Figure 2B). To produce the ΔEXO mouse model, we used tamoxifen to induce expression of diphtheria toxin alpha (DTA) in the Ptf1a-expressing cells at E14.5, which stimulated apoptosis of these cells. Similar to previous analyses, we used amylase to visualize the acinar cells and insulin to identify the beta cells; R26R^Tomato reporter expression in the control tissue confirmed cre activity (Figure 2C–D). Morphometric analysis identified a significant decrease in exocrine area in the ΔEXO pancreas compared with Ptf1a-cre^ERTM controls (Figure 2E); however, no significant difference in beta cell area was observed (Figure 2F). This suggests that a loss of exocrine volume during the secondary transition does not directly influence beta cell growth.

Figure 2.

eIF5A deletion during the stage of differentiation increases beta cell area. (A) Schematic of exocrine ablation model generation. (B) Schematic of model generation of eIF5A deletion during differentiation. For both models, pregnant dams were injected with tamoxifen (TAM) at E14.5 and collected for analysis at E18.5. Representative images of E18.5 (C) Ptf1a-cre^ER control pancreas and (D) ΔEXO mutant pancreas stained for amylase (green), insulin (red), tdTomato (blue), and nuclei (DAPI, white). E18.5 ΔEXO mutants show (E) a significant decrease in exocrine area and (F) no change in beta cell area. Representative images of E18.5 (G) Ptf1a-cre^ER control pancreas and (H) eIF5A^ΔEXO mutant pancreas stained for amylase (green), insulin (red), tdTomato (blue), and nuclei (DAPI, white). E18.5 eIF5A^ΔEXO mutants exhibit (I) no change in exocrine area and (J) a significant increase in beta cell area. Data are presented as mean+/−SEM. n = 6/group; ∗p < 0.05,∗∗∗∗p < 0.0001.

In parallel, we generated the eIF5A^ΔEXO mutant using tamoxifen to induce deletion of Eif5a in the Ptf1a-expressing cells at E14.5. Morphometric analysis quantified exocrine (amylase) and beta cell (insulin) area in the pancreas of Ptf1a-cre^ERTM controls and eIF5A^ΔEXO mutants (Figure 2G–H). Interestingly, we observed no significant difference in the exocrine area of the eIF5A^ΔEXO mutant pancreas compared with controls (Figure 2I); however, there was a significant increase in insulin-expressing beta cell area in the eIF5A^ΔEXO mutants (Figure 2J). These data indicate that the absence of eIF5A during the secondary transition drives an increase in beta cell differentiation at this stage; however, the loss of exocrine mass may be patterned earlier.

3.3. Comparative analysis of eIF5A^ΔPANC and eIF5A^ΔEXO mutants revealed altered signaling pathways at E18.5

The similar exocrine volume between the eIF5A^ΔEXO mutants and controls suggests that the influence of eIF5A on exocrine patterning occurs before E14.5. The phenotype of the eIF5A^ΔEXO mutants can also confirm if eIF5A-mediated translation regulates the synthesis of exocrine-derived proteins later in pancreas development. To address this, we performed quantitative mass spectrometry on pancreas tissue from E18.5 eIF5A^ΔEXO mutants and Ptf1a-cre^ERTM controls and identified the differentially expressed proteins. Differentially expressed proteins (DEPs) were classified as significant with a p < 0.05 (Figure 3A; Supplemental Table 2) and pathway analysis performed using the bioinformatics tool DAVID identified pathways including pancreatic secretion, protein digestion and absorption, focal adhesion, cytoskeleton regulation, tight junctions, and metabolism were altered in the eIF5A^ΔEXO mutant pancreas compared with Ptf1a-cre^ERTM controls (Figure 3B). Furthermore, we compared this eIF5A^ΔEXO dataset with our previously published proteomic analysis of the eIF5A^ΔPANC mutants at E18.5 [14](Supplemental Table 3) and observed similar trends in the abundance of many acinar-related proteins including digestive enzymes, which were categorized in the pathways “protein digestion and absorption” and “pancreatic secretion”. Importantly, of the 28 acinar function proteins identified in our previous study as differentially expressed within the eIF5A^ΔPANC mutants [14], 26 show similar differential protein expression in the eIF5A^ΔEXO mutants (Figure 3C; Supplemental Table 4). This data confirms that eIF5A-mediated translation regulates the synthesis of proteins needed for acinar cell function.

Figure 3.

Quantitative proteomic analysis of the eIF5A^ΔEXO mutant pancreas. (A) Volcano plot visualizing the differentially expressed proteins in the eIF5A^ΔEXO compared with Ptf1a-cre^ERTM. (B) KEGG pathways altered in the eIF5A^ΔEXO compared with Ptf1a-cre^ERTM pancreas at E18.5 (C) Comparison of the eIF5A^ΔPANC and eIF5A^ΔEXO mutants for expression of proteins required for exocrine function. All proteins were previously reported as significantly downregulated in the eIF5A^ΔPANC mutant pancreas; those in bold font were also significantly downregulated in eIF5A^ΔEXO mutant pancreata whereas those in regular font show a trend toward downregulation. (D) Proteins identified as differentially expressed at E18.5 in eIF5A^ΔEXO mutants that are components of pathways contributing endocrine versus exocrine cells differentiation. n = 6/group.

In addition to the pathways “protein digestion and absorption” and “pancreatic secretion”, which contain numerous digestive enzymes needed for acinar cell function (Figure 3C; [14]), some of the altered pathways previously identified in the eIF5A^ΔPANC mutant pancreas included “focal adhesion”, “PI3K-Akt signaling”, “Hippo signaling” and “tight junctions” [14]. Given that these pathways contribute to endocrine and exocrine differentiation [[32], [33], [34], [35]] and have been shown to directly influence Notch signaling, which also plays a role in endocrine versus exocrine lineage decisions [23,[36], [37], [38], [39], [40], [41], [42], [43]], we asked if any DEPs in the eIF5A^ΔEXO were components of these pathways. We hypothesized that some of the outcomes from dysregulated eIF5A-mediated translation during the stage of pancreas specification should be similar to changes observed during the stage of differentiation. Excitingly, we identified differentially expressed proteins in PI3K-Akt signaling, Hippo signaling, focal adhesion, and tight junctions (Figure 3D; Supplemental Table 5). It should be noted that these changes represent a downstream effect of the loss of eIF5A-mediated translation earlier in pancreas development and therefore the synthesis of proteins that act earlier in development and influence these pathways may be the direct targets of eIF5A-mediated translation.

3.4. Pancreas progenitors and ductal organization are unaltered in eIF5A^ΔPANC mutants during the primary transition

The data from the eIF5A^ΔPANC and eIF5A^ΔEXO mutants suggests that eIF5A-mediated translation influences beta cell differentiation during the secondary transition and exocrine differentiation earlier than E14.5 (Figure 2, Figure 3). Resultantly, it is possible that eIF5A-mediated translation impacts progenitor cell organization and abundance during the primary transition, which would subsequently influence the differentiation of beta and acinar cells. Therefore, we analyzed E11.5 eIF5A^ΔPANC mutant pancreata and Ptf1a-cre controls, which is a time point during the primary transition and the stage of development defined by pancreatic progenitor cell expansion and formation, and organization of the ductal epithelium. We stained this tissue for Pdx1 to mark all pancreatic progenitors, and mucin1 to visualize ductal microlumens (Figure 4A–B). Morphometric analysis revealed no significant change in the number of Pdx1-expressing pancreatic progenitor cells between genotypes (Figure 4C). Moreover, we quantified mucin1-expressing area as well as the average ductal microlumen size, which directly influences subsequent ductal structure, and observed no significant differences (Figure 4D–E). These data suggest that neither reduced pancreatic progenitor cell number nor defective lumen formation in the early stage of pancreas development result from loss of eIF5A.

Figure 4.

eIF5A^ΔPANC mutants exhibit no changes in pancreatic progenitor cells or ductal lumen formation during the primary transition. Representative images of E11.5 (A) Ptf1a-cre control and (B) eIF5A^ΔPANC mutant pancreas stained for Mucin1 (green), Pdx1 (blue), and nuclei (DAPI, white). (C) Quantification of the total number of Pdx1+ nuclei per pancreas showed no significant difference between Ptf1a-cre control and eIF5A^ΔPANC mutant pancreata. (D) Quantification of the total microlumen area measured by Mucin expression showed no significant difference between Ptf1a-cre control and eIF5A^ΔPANC mutant pancreata. (E) Quantification of the average microlumen size showed no significant difference between Ptf1a-cre control and eIF5A^ΔPANC mutant pancreata. n = 4–6/group. Data are presented as mean+/−SEM.

3.5. eIF5A^ΔPANC mutants have an imbalance in progenitor cells during the secondary transition

The absence of altered duct formation and pancreatic progenitors during the primary transition indicates that eIF5A-mediated translation acts after E11.5 to influence the beta and acinar cell populations. Therefore, we examined the eIF5A^ΔPANC mutant pancreas at E14.5 for organization of the ductal epithelium, which houses the endocrine and exocrine progenitor cells. Using Nkx6.1 to mark the ductal trunk (which contains endocrine progenitors) and Ptf1a to mark the ductal tips (which contain exocrine progenitors), we observed no change in Nkx6.1-expressing cells but a significant reduction in the number of Ptf1a-expressing ductal tips (Figure 5A–C; S1A Figure). The reduction in the number of ductal tips remained significant even when normalized to pancreas area (Figure 5D). We next evaluated the organization of the ductal epithelium using Sox9 to visualize the ducts and carboxypeptidase A (CPA) to mark the developing acinar cells at the ductal tips (Figure 5E–F). Compared with Ptf1a-cre controls, the eIF5A^ΔPANC mutant pancreas showed no change in Sox9-expressing cells but reduced acinar cell containing ductal terminations (Figure 5G; S1B Figure) and significantly decreased number of ductal branch points (Figure 5H), which suggests that there are fewer terminal duct endings where acini can form. Finally, given the increased number of beta cells observed in the eIF5A^ΔPANC mutant pancreas at E18.5, we examined the quantity of Ngn3-expressing endocrine progenitor cells within the Sox9-expressing epithelium at E14.5 (Figure 5I–J). The eIF5A^ΔPANC mutant pancreas showed a significantly increased number of Ngn3-expressing cells within each segment of ductal trunk (Figure 5K), equating to a significantly greater number of endocrine progenitors compared with the Ptf1a-cre control pancreata (Figure 5L). Together, these data demonstrate that the absence of eIF5A in the developing pancreas results in a disorganized ductal epithelium at E14.5 with fewer ductal terminations and greater endocrine progenitor cells, which precedes the reduced exocrine and increased beta cell mass observed in the mutants at E18.5.

Figure 5.

eIF5A^ΔPANC mutants exhibit impaired tip cell formation concomitant with increased endocrine progenitor cells at E14.5. Representative images of E14.5 (A) Ptf1a-cre control and (B) eIF5A^ΔPANC pancreas stained for Ptf1a (green), Nkx6.1 (red), E-cadherin (white), and nuclei (DAPI, blue). (C) Quantification of Ptf1a-expressing tips showed a significant decrease in the eIF5A^ΔPANC mutants. (D) Quantification of the E-cadherin-expressing pancreas area showed no significant difference between the eIF5A^ΔPANC mutants and Ptf1a-cre controls. Representative images of E14.5 (E) Ptf1a-cre and (F) eIF5A^ΔPANC pancreas stained for CPA (green) and Sox9 (red). (G) Quantification of the average number of ductal terminations showed a significant decrease in the eIF5A^ΔPANC mutants. (H) Quantification of the average number of ductal branching points showed a significant decrease in eIF5A^ΔPANC mutants. Representative images of E14.5 (I) Ptf1a-cre and (J) eIF5A^ΔPANC pancreas stained for Ngn3 (green), Sox9 (red), and E-cadherin (white). (K) Quantification of the number of Ngn3-expressing cells per Sox9-expressing ductal segment showed a significant increase in the eIF5A^ΔPANC. (L) Quantification of Ngn3-expressing cells in the pancreas showed a significant increase in the eIF5A^ΔPANC mutants. Data are presented as mean +/− SEM; n = 3–4/group; ∗, p < 0.05; ∗∗, p < 0.01.

3.6. Deletion of Eif5a alters Notch signaling during the secondary transition

Our data showed an increase in beta cells at E18.5 (Figure 1), preceded by altered organization of the ductal epithelium including an increase in Ngn3-expressing cells at E14.5 (Figure 5). Next, we asked what could drive the increase in Ngn3 and how might the stimulate only an increase in beta cell mass? Previous work has shown that the fate of the pancreatic progenitors is driven by Notch signaling acting during the secondary transition to directly influence the expression of Ngn3 [44,45]. Specifically, Notch signaling functions in pancreatic lineage specification by repressing Ngn3 expression, thereby preventing the premature differentiation of progenitor cells into endocrine cells. To do this, the transcriptional repressor Hes1 is activated by Notch signaling and specifically represses Ngn3 transcription in adjacent cells and thus the differentiation into endocrine cells. Given that we observed an increase in Ngn3 at E14.5, we hypothesized that a decrease in Hes1 expression was driving the increased expression in Ngn3. First, to ensure this change in cellular distribution was not due an imbalance in Hes1 expression during specification, we examined Hes1 expression in E11.5 Ptf1a-cre controls and eIF5A^ΔPANC mutants (Figure 6A–B). We observed no significant changes to the number of Hes1-expressing cells within the pancreas (Figure 6C), which is in line with our data that showed no morphometric changes during the primary transition (Figure 4). Next, we stained pancreas from E14.5 Ptf1a-cre and eIF5A^ΔPANC animals with E-cadherin and Hes1 to examine if the increase in Ngn3-expressing cells at this age coincided with a decrease in Hes1-expressing cells (Figure 6D–E). The eIF5A^ΔPANC mutants displayed a significant reduction in the number of Hes1-expressing cells compared with Ptf1a-cre controls (Figure 6F), suggesting a disruption in Notch signaling.

Figure 6.

Notch signaling is impaired in the eIF5A^ΔPANC mutant pancreas. Representative images of E11.5 (A) Ptf1a-cre and (B) eIF5A^ΔPANC pancreata stained for E-cadherin (green), Hes1 (red) and nuclei (DAPI, blue). These tissue sections also show expression of the tdTomato reporter, which is visualized in the green channel. (C) Quantification of the Hes1-expressing cells within the pancreas showed no significant difference between Ptf1a-cre controls and eIF5A^ΔPANC mutants. Representative images of E14.5 (D) Ptf1a-cre and (E) eIF5A^ΔPANC pancreata stained for E-cadherin (green), Hes1 (red) and nuclei (DAPI, blue). (F) Quantification of the Hes1-expressing cells within the pancreas showed a significant decrease in eIF5A^ΔPANC mutants. Gene expression analysis of (G) Hes1, (H) Notch1, (I) Ngn3, and (J) Eif5a in E14.5 pancreata from Ptf1a-cre controls and eIF5A^ΔPANC mutants. Data are presented as mean +/− SEM; n = 4–5/group; ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001.

To determine if these protein expression changes were driven by changes at the level of mRNA translation, we performed real time PCR on RNA extracted from E14.5 Ptf1a-cre and eIF5A^ΔPANC pancreas and measured the abundance of Hes1, Notch1, and Ngn3. Gene expression analysis revealed no significant difference in Hes1 or Notch1 expression (Figure 6G–H), indicating that loss of Hes1 protein expression was not due to a reduction in mRNA transcript. Interestingly, our analysis also revealed an increase in the amount of Ngn3 transcript (Figure 6I), which further supports the increase in the number of Ngn3-expressing cells (Figure 5). Moreover, to confirm genetic deletion of Eif5a in the embryonic pancreas, we measured the abundance of Eif5a, which showed the expected significant decrease (Figure 6J). Altogether, our data indicate that eIF5A-mediated translation regulates the synthesis of Notch pathway components during endocrine differentiation and that phenotypic changes observed at this stage result, at least in part, from the imbalance in progenitor cell populations in the ductal epithelium.

3.7. Increased beta cell mass and loss of exocrine volume is maintained postnatally in eIF5A^ΔPANC mice

Given the observation of decreased exocrine mass and increased beta cell mass by the end of embryonic development at E18.5, we evaluated postnatal viability and the pancreas phenotype. Based on genotyping of litters at 2 weeks-of-age, mutant animals were viable, and we observed eIF5A^ΔPANC and controls in Mendelian ratio (χ² = 6.481; p = 0.262). Subsequent morphometric analysis of the pancreas from these animals, with amylase marking the acinar cells and insulin marking the beta cells (Figure 7A–B), showed a significant reduction of exocrine volume in the eIF5A^ΔPANC mutant pancreata compared with controls (Figure 7C) and a continued significant increase in beta cell area (Figure 7D). We performed the identical analysis of pancreas tissue from animals after weaning, at 4 weeks-of-age. We were interested in the weaning stage, given that this time period involves a switch from a lipid-rich (maternal milk) diet to a carbohydrate-rich (chow) diet and therefore an increased stress on pancreatic function [46,47]. Morphometric analysis of pancreas tissue from 4-week-old Ptf1a-cre and eIF5A^ΔPANC animals (Figure 7E–F) revealed the same phenotype as observed at 2 weeks-of-age – a significant decrease in exocrine area and significant increase in beta cell area in the eIF5A^ΔPANC mutants compared with controls (Figure 7G–H; S2A-B Figure). Furthermore, in line with our observations at E18.5, morphometric analysis showed no difference in glucagon-expressing alpha cells or somatostatin-expressing delta cells between eIF5A^ΔPANC mutants and controls (Figure 7I – L; S2C-D Figure). Altogether, these data suggest that islets with an expanded beta cell mass can be maintained postnatally in a pancreas environment devoid of exocrine.

Figure 7.

Morphometric analysis of exocrine and beta cell area in pancreas from 2- and 4-week-old animals. Representative images of pancreas from 2-week-old (A) Ptf1a-cre and (B) eIF5A^ΔPANC mice stained for amylase (green), insulin (red), and nuclei (DAPI, white). (C) Quantification of amylase-expressing exocrine area showed a significant decrease in the eIF5A^ΔPANC mutants. (D) Quantification of insulin-expressing beta cell area showed a significant increase in the eIF5A^ΔPANC mutants. Representative images of pancreas from 4-week-old (E) Ptf1a-cre and (F) eIF5A^ΔPANC mice stained for amylase (green), insulin (red), and nuclei (DAPI, white). (G) Quantification of male amylase-expressing exocrine area showed a significant decrease in the eIF5A^ΔPANC. (H) Quantification of male insulin-expressing beta cell area showed a significant increase in eIF5A^ΔPANC. Representative images of pancreas from 4-week-old (I) Ptf1a-cre and (J) eIF5A^ΔPANC mice stained for somatostatin (green), glucagon (red), and nuclei (DAPI, white). Quantification of male (K) glucagon-expressing alpha cell area and (L) somatostatin-expressing delta cell area showed no significant differences between the eIF5A^ΔPANC mutants and Ptf1a-cre controls. Data are presented as mean +/− SEM; n = 3–8/group. ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗∗, p < 0.0001.

3.8. In the setting of exocrine insufficiency, the expanded beta cells in the eIF5A^ΔPANC mice are functional

Given our finding of islets with increased beta cell mass in the setting of exocrine insufficiency, we evaluated the metabolic health of eIF5A^ΔPANC mutants by measuring body weight and ad libidum fed blood glucose from the perinatal to adult stage of life (2–8 weeks-of-age). Analysis of body weight indicated that in cohorts of either males or females, the eIF5A^ΔPANC mutants and Ptf1a-cre controls begin at the same body weight but the mutants gained weight at a slower rate compared with controls (Figure 8A–B). Moreover, whereas eIF5A^ΔPANC mutants (both male and female) showed lower relative blood glucose before weaning, mutant eIF5A^ΔPANC animals displayed blood glucose levels similar to Ptf1a-cre controls from 4 to 8 weeks-of-age (Figure 8C–D). Additionally, glucose tolerance tests were performed on animals at 7 weeks-of-age and revealed normal glucose tolerance in male eIF5A^ΔPANC mutants and improved glucose tolerance in female eIF5A^ΔPANC mutants compared with controls (Figure 8E–H). Furthermore, plasma insulin and glucagon levels in 7-week-old male and female eIF5A^ΔPANC mutants were not significantly altered compared with Ptf1a-cre controls (Figure 8I-L). Altogether, these data indicate that eIF5A^ΔPANC mutants have normal to improved glucose tolerance despite a dramatic loss of the exocrine mass. In summary, our eIF5A^ΔPANC mutant represents a unique model of islet function in the setting of exocrine insufficiency.

Figure 8.

Metabolic measurements in Ptf1a-cre controls and eIF5A^ΔPANC mutants from 2 weeks-to 8 weeks-of-age. From 2 through 8-weeks-of-age, body weight was measured in (A) male Ptf1a-cre (n = 6) and eIF5A^ΔPANC (n = 10) animals, and (B) female Ptf1a-cre (n = 6) and eIF5A^ΔPANC (n = 8) animals. In parallel, blood glucose was measured during the same time frame in (C) male Ptf1a-cre (n = 6) and eIF5A^ΔPANC (n = 10) animals, and (D) female Ptf1a-cre (n = 6) and eIF5A^ΔPANC (n = 8) animals. (E) Glucose tolerance tests were performed on male Ptf1a-cre (n = 6) and eIF5A^ΔPANC (n = 7) animals at 7 weeks-of-age. (F) Area under the curve analysis showed no significant difference in glucose tolerance between male control and mutant mice. (G) Glucose tolerance tests were performed on female Ptf1a-cre (n = 4) and eIF5A^ΔPANC (n = 7) animals at 7 weeks-of-age. (H) Area under the curve showed improved glucose tolerance in female mutant mice compared with controls. Plasma (I) insulin and (J) glucagon levels were measured for male Ptf1a-cre controls and eIF5A^ΔPANC mutants at 7 weeks-of-age. Plasma (K) insulin and (L) glucagon levels were measured for female Ptf1a-cre controls and eIF5A^ΔPANC mutants at 7 weeks-of-age. n = 3/group. Data presented as mean +/− SEM. ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001, ∗∗∗∗, p < 0.0001.

4. Discussion

In this study, we have discovered that eIF5A-mediated translation regulates the synthesis of proteins needed for signaling pathways that instruct progenitor cell allocation in the pancreatic ductal epithelium of mice. Impairing this regulatory mechanism causes altered endocrine and exocrine lineage differentiation and a resultant pancreas devoid of acinar cells, concomitant with increased functional beta cell mass.

We identified that eIF5A-mediated translation specifically functions during the secondary transition, which is the stage of significant cellular differentiation during pancreas development. Whereas the loss of exocrine pancreas volume was expected given the phenotype of the DHPS^ΔPANC mutant mice we previous published [14], the concomitant increase in beta cell mass was unexpected. Moreover, we expected that our inducible model of eIF5A loss during the secondary transition (eIF5A^ΔEXO) would phenocopy our eIF5A^ΔPANC model; however, we observed a significant increase in the beta cell area with no significant changes to the acinar cell abundance. We speculate that inducing impaired eIF5A-mediated translation in the middle of the secondary transition, which impairs Notch signaling at this timepoint, is past the developmental stage that influences exocrine progenitor development. More specifically, our data shows that ductal organization and the distribution of exocrine progenitor cells within the ducts must be specified earlier than E14.5 but after E11.5. Together these data refine the timing of influence of eIF5A-mediated translation such that the synthesis of proteins that regulate ductal branching and exocrine progenitor cell differentiation occurs before the induction of pathways that coordinate beta cell differentiation.

Our model of induced deletion of Eif5a during the secondary transition (eIF5A^ΔEXO) also clarifies the importance and timing of translational regulation of digestive enzymes in acinar cells. Acinar cells are known to have the highest rate of protein synthesis for a cell in any mammalian organ, and approximately 90% of all transcripts encode for digestive enzymes [48]. With this in mind, it would follow that this secretory cell, which needs to make and secrete these proteins on demand, would have a mechanism that regulates protein synthesis to control timing and quantity. Therefore the eIF5A^ΔEXO mutant revealed that the functional characteristics of acinar cells are imparted as early as E14.5. Drawing from the studies by Connors∗, Villaca∗ et al., [11] and Levasseur et al., [49], where beta cell growth was only impacted in beta cell-specific DHPS mutants once there was a need for on-demand protein synthesis, we speculate that, similar to the eIF5A^ΔEXO, a postnatal deletion of DHPS/eIF5A in adult acinar cells would not display any detrimental effects on exocrine growth but would impair the abundance of digestive enzymes when secretory function is induced. Generating this mutant mouse model would be an interesting future study.

In addition to clarifying the timing of translational regulation of acinar cell function, our mutant mice revealed important observations about the regulation of beta cell differentiation. Unexpectedly, we observed that alterations to Notch signaling upon Eif5a deletion predominantly impacted the beta cell population without affecting alpha cells and delta cells. Specifically, our eIF5A^ΔPANC mutant pancreas showed reduced Hes1-expressing cells and increased Ngn3-expressing endocrine progenitors in the ductal epithelium at E14.5, with a subsequent significant increase in beta cells but not alpha or delta cells at E18.5. This data suggests that loss of eIF5A in the early developing pancreas results in decreased Notch signaling leading to a reduced number of Hes1-expressing cells in epithelium at E14.5. As Hes1 is a known repressor of Ngn3 [50], its reduction increases Ngn3-expressing endocrine progenitors at E14.5, which results in beta cell differentiation. The predominant increase in beta cells after a specific increase in Ngn3-expressing cells at E14.5 is in line with work from Henrik Semb's group [23], which demonstrates that the establishment of Ngn3 high-expressing endocrine progenitors is influenced by changes in ductal organization and Notch signaling, which subsequently drives beta cell specific differentiation [23]. Moreover, Johansson and colleagues [51] show that the induction of Ngn3 expression at E14.5 led to predominantly beta cell formation by E18.5. Thus, our data supports a mechanism wherein impaired eIF5A-mediated translation during the primary transition dysregulates Notch signaling that in turn upsets the balance of Hes1-and Ngn3-expressing cells at E14.5, which drives differentiation of beta cells by E18.5.

The source of Notch signaling is an interesting question. Published work shows that Notch signaling functions in pancreatic lineage specification by repressing Ngn3 expression and preventing the premature differentiation of progenitor cells into endocrine cells. In particular Notch/Hes1 was shown to control Ngn3 protein stability [44] and Hes1 oscillations were found to be associated with the bipotent versus pro-acinar fate choice [45]. Given that we observed the increase in Ngn3 and decreased Hes1 at E14.5, this supports the mechanism that decreased Hes1 was driving the increased expression of Ngn3 and subsequently influencing the balance of endocrine and exocrine progenitors. Whereas exocrine cells have been shown to upregulate Notch signaling [52], the most likely scenario is that the source of the decreased Notch signaling is a cell population that is present at or before E14.5. Pinpointing the exact age in early pancreas development when altered Notch signaling influences lineage decisions is a question that requires further investigation.

Our study highlights that eIF5A-mediated translation has cell type-specific and time-dependent targets, such that the loss of on-demand protein synthesis in one cell type versus another or in prenatal versus postnatal growth will impact different proteins and thus produce different outcomes. As demonstrated in the eIF5A^ΔPANC model compared with the eIF5A^ΔEXO model, loss of eIF5A-mediated translation at the beginning of pancreas development impacted both the endocrine and exocrine compartment while loss at E14.5 only impacted the endocrine. Furthermore, the target genes translationally regulated by eIF5A varied depending on the cell types and the age at which translation was impaired; the proteins needed for lineage decisions and cellular functions are specific to that tissue. This finding is in line with a study from the Mirmira lab, which showed that DHPS/eIF5A-mediated translation was needed for the compensatory response of beta cell proliferation following high fat diet feeding [49]. Whereas Levasseur and colleagues focused on the postnatal beta cell in the setting of diabetogenic injury, our work in the developing beta cell revealed that impaired eIF5A-mediated on-demand protein synthesis altered the synthesis of proteins needed for beta cell maturation including Insulin, Glut2, Ucn3 and ChgA [11]. Together these studies underscore how this regulatory mechanism functions in a time and cell type-specific manner.

Additional work in the field supports the idea that eIF5A-mediated translation is tissue/cell specific. In muscle, eIF5A-mediated translation is necessary for muscle cell identity and activation, as eIF5A is required for the translation of MyoD, which is a transcription factor that defines the muscle cell lineage [53,54]. Moreover, eIF5A-mediated translation has a regulatory role in immune cells including the modulation of mitochondrial respiration and polarization to a pro-inflammatory phenotype [[55], [56], [57]]. Also, eIF5A has been shown to be necessary for the translation of the autophagy protein ATG3, which functions in the formation of the autophagosome and protein lipidation [58]. These studies further support the finding that eIF5A-mediated translation functions in a context-dependent manner.

Whereas many studies, including our own, have focused on DHPS/eIF5A^HYP and the function of this specialized translation factor, a study by Anderson and colleagues highlighted that we also must consider the role of the unhypusinated form of eIF5A, and suggested that this unmodified form may also act in the beta cell. The authors directly targeted eIF5A (rather than DHPS) in the postnatal beta cell in the setting of high fat diet and revealed that compensatory beta cell proliferation was not altered [59], and speculated that the differing phenotype from their mouse model of DHPS loss in the adult beta cell in the setting of high fat diet was due to a unique role of unhypusinated eIF5A in regulating functional responses in the pancreas [49]. Our eIF5A^ΔPANC mouse model also specifically targets eIF5A and shows that the regulation of synthesis of specific proteins needed for lineage decisions during development is impaired when Eif5a is deleted in the pancreatic progenitor cells, which is in line with our published mouse model of Dhps deletion in the developing pancreas [14]. Comparing these mouse models, one could speculate that when cell lineage is not fixed (e.g. progenitor cells in the developing pancreas) the function of the hypusinated form of eIF5A is most prominent so that the correct rate of translation and abundance of proteins can be maintained and resultantly permit proper cell formation and differentiation. However, it is possible that once the lineage and functional capacity of a cell is fixed (e.g. 8-week-old mature adult beta cell) there is a role for the unhypusinated form of eIF5A. Clearly, further study is needed to understand the contribution of unhypusinated eIF5A in different cellular contexts.

Our studies have refined our understanding of this regulatory mechanism in the setting of pancreas development; however, our mouse model has also provided insight into disease states. Often diseases of the exocrine pancreas negatively impact the islets, contributing to diabetes onset. However, clinical data has also revealed that 20–65% of individuals with exocrine diseases including pancreatitis, exocrine insufficiency and cystic fibrosis, do not develop diabetes [[60], [61], [62], [63], [64]]. How beta cells grow and maintain function despite the presence of catastrophic exocrine disease is not understood. Our eIF5A^ΔPANC mouse model provides an opportunity to further study this clinical phenomenon, given the phenotype of exocrine insufficiency without endocrine dysfunction. Moreover, our findings suggest that individuals with exocrine disease without endocrine dysfunction may have a greater functional beta cell mass.

Our findings also suggest that this mechanism may be common to and critical for the development, growth, and function of many different professional secretory cells. As mentioned above, secretory cells in the body make large amounts of specific proteins and release these proteins to facilitate cellular functions. Specifically, pancreatic beta cells make and secrete insulin into the blood to maintain metabolic homeostasis, pancreatic acinar cells make and secrete enzymes that travel to the intestine and aid in digestion, and neurons make and secrete neurotransmitters in the brain that act as chemical messengers, relaying signals from one cell to the next thereby triggering a response such as an electrical signal. In each case, these secretory cells make and secrete critical proteins in response to physiologic demand. Our work implies that impaired on-demand protein synthesis in the beta cell results in diabetes [11], in the acinar cells it results in exocrine insufficiency [14], and in the brain it results in epilepsy [15]. It would be exciting to propose that restoring on-demand protein synthesis in each of these settings could reverse cellular dysfunction and possibly disease.

CRediT authorship contribution statement

Danielle L. Overton: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Investigation, Formal analysis. Catharina BP. Villaca: Writing – review & editing, Validation, Methodology, Investigation, Formal analysis. Dorian J. Dale: Writing – review & editing, Validation, Investigation, Formal analysis. Caleb D. Rutan: Writing – review & editing, Validation, Investigation, Formal analysis. Morgan A. Robertson: Writing – review & editing, Validation, Investigation, Formal analysis. Craig T. Connors: Writing – review & editing, Validation, Investigation, Formal analysis. Emily K. Anderson-Baucum: Writing – review & editing, Validation, Investigation, Formal analysis. Teresa L. Mastracci: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization.

Funding

This work was supported by funding from the National Institutes of Health (NIH) to DLO (F31DK137482) and TLM (R01DK121987). The mass spectrometry work was performed by the Indiana University School of Medicine Center for Proteome Analysis and acquisition of the Indiana University School of Medicine Proteomics instrumentation was provided by the Indiana University Precision Health Initiative. The proteomics work was supported, in part, by the Indiana Clinical and Translational Sciences Institute funded in part by funds (UL1TR002529) from the NIH, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award, and the Cancer Center Support Grant for the IU Simon Comprehensive Cancer Center (P30CA082709) from the National Cancer Institute.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.