Pancreatic injury induces β–cell regeneration in axolotl

Abstract

Background:

Diabetes is a condition characterized by a loss of pancreatic β–cell function which results in the dysregulation of insulin homeostasis. Using a partial pancreatectomy model in axolotl, we aimed to observe the pancreatic response to injury.

Results:

Here we show a comprehensive histological characterization of pancreatic islets in axolotl. Following pancreatic injury, no apparent blastema-like structure was observed. We found a significant, organ-wide increase in cellular proliferation post-resection in the pancreas compared to sham-operated controls. This proliferative response was most robust at the site of injury. Further, an increase in nuclear density was observed, suggesting compensatory congestion as a mechanism of regeneration. We found that β–cells actively contributed to the increased rates of proliferation upon injury. β–cell proliferation manifested in increased β–cell mass in injured tissue at two weeks post injury. At four weeks post injury, we found organ-wide proliferation to be extinguished while proliferation at the injury site persisted, corresponding to pancreatic tissue recovery. Similarly, total β–cell mass was comparable to sham after four weeks.

Conclusions:

Our findings suggest a non-blastema-mediated regeneration process takes place in the pancreas, by which pancreatic resection induces whole-organ β–cell proliferation without the formation of a blastemal structure. This process is analogous to other models of compensatory congestion in axolotl.

Background

Diabetes is a condition characterized by a loss of pancreatic β–cell function which results in compromised insulin homeostasis. Treatment for diabetes can be, in most cases, lifelong, requiring administration of exogenous insulin or glycemic control agents, and glucose monitoring. Diabetes can cause complications such as heart failure, kidney disease, and tissue necrosis requiring limb amputation.^1,2,3 Several treatment strategies, such as cadaveric β–cell transplantation, polymer encapsulated stem cell-derived β–cell transplantation, transplantation of chemically induced pluripotent stem-cell-derived islets, and personalized endoderm stem cell-derived islet tissue, aimed at functional restoration, have been implemented to varying degrees of success.^4,5,6,7 These strategies, however, are invasive procedures that do not address the root cause of initial β–cell loss of function. A thorough understanding of genetic pathways that can promote pancreatic regeneration is important for developing therapeutic strategies for diabetes based on stimulating endogenous β-cell replacement.

Pancreatic regeneration has been reported in several animal models, such as mice and zebrafish, with varying degrees of regenerative organ recovery.^8,9 Pancreatic regeneration studies typically focus on the endocrine pancreas i.e. β–cells due to their relevance to diabetes. β–cells are the principal insulin producing cells in the body; characterizing their development, function, and fate has been a topic of research for decades. Partial pancreatectomy studies in rats and mice have reported proliferation of existing islet and exocrine cells, demonstrating regenerative ability in these organisms.^8,10 In a landmark study, Dor et al. reported that mice, which have been subject to a partial pancreatectomy, demonstrate a 2.6-fold increase in β–cell proliferation as a result of pre-existing β–cell differentiation, indicating β-cell regeneration by self-duplication.¹⁰ However, although a modest regenerative capacity is observed in mice following pancreatic injury, this capacity sharply declines in adulthood and is absent in adult humans.⁴ Despite several decades of characterization of the pancreas in mammals, endogenous β–cell regeneration has not been accomplished in human patients.

The axolotl (Ambystoma mexicanum) is a highly regenerative species of salamander that can provide insights into how molecular mechanisms of regeneration might be harnessed therapeutically in humans. The axolotl has been shown to regenerate several organs and appendages, such as limbs, lungs, liver, heart, and tail, among others.^11,12,13 The molecular mechanisms governing these regeneration events are still being elucidated, but they all feature cellular proliferation and regulation of cell-to-cell signaling processes. Despite the investigation of several internal organs, few studies have investigated pancreatic regenerative ability in axolotls. Conservation of pancreatic regeneration in many other organisms leads us to reason the axolotl may display a similar proficiency for pancreatic response to injury.

Although the pancreas is well studied in humans and mice, only one study has reported on pancreatic regeneration in axolotls.¹⁴ This study, which established a diabetic axolotl model via chemical ablation of β–cells using streptozotocin, demonstrated a return to normal glycemic function after 84 days. These results indicated a restoration of β–cell function – possibly due to regeneration. However, streptozotocin was reported to cause side effects such as edema, lowered red blood cell count, and increased mortality.¹⁴ These findings introduced the axolotl as a promising animal model to study diabetes and illustrated the need to develop an injury model that includes all pancreatic cell types to show whether axolotls exhibit robust pancreatic regeneration. We sought to expand on the limited literature currently available in regard to axolotl pancreas regeneration by studying the regenerative response to partial pancreatectomy in axolotls.

Here, we characterize pancreatic tissue morphology and the pancreatic response to injury in axolotls. We established a novel pancreatic resection surgery model to investigate how animals respond to this injury over the course of 28 days. We identified key genes such as Pdx1, Ins, Tgf-β1, Ctrb2, Marco, Kazald2, and Cirbp to be differentially regulated in response to pancreatic injury.

Results

Axolotl pancreatic tissue morphology

The axolotl pancreas is a thin, triangular tissue connected directly to the base of the duodenum. It has been reported that the axolotl pancreas exhibits protein level insulin expression similar to the mammalian pancreas.¹⁵ Recently, Ma et al. characterized the morphology of the axolotl pancreas by reporting the presence of pancreatic ducts, acinar cells, and islets.¹⁶ We utilized hybridized chain reaction-fluorescent in situ hybridization (HCR-FISH) to show spatial transcript level expression of several pancreatic markers, corroborating previous findings in both axolotl and newts.^{16, 17, 18} Our results show the presence of Ins- (insulin) and Nkx6–1- expressing β–cells, Gcg- (glucagon) expressing α-cells, Sst- (somatostatin) expressing δ-cells, (Figure 1A–B). Pdx1, a critical gene in pancreatic development and function, was found to be expressed in both endocrine and exocrine cells (Figure 1B).¹⁹

Figure 1. Transcript level pancreatic islet morphology in axolotl.

A. HCR-FISH of Gcg, Ins, and Sst transcripts within pancreatic islets. B. HCR-FISH of Ins, Pdx1, and Nkx6–1 transcripts. Scale bars are 50 μm.

A novel pancreatic resection model in the axolotl

We established a pancreatic resection surgery model with a sham-operated counterpart in order to investigate the injury response after a loss of pancreatic tissue mass (Figure 2A–B). Pancreatic resection was performed by surgical incision at a site adjacent to the base of the duodenum, resulting in removal of approximately twenty percent of the length of the organ. Animals with larger portions of their pancreas resected generally did not survive. Injury response was investigated at 14 and 28 days post injury (dpi). Ninety percent of animals survived the surgery, and death was non-specific to the sham or resection group. Mortality could have resulted from blood loss, infection, or other unknown causes. By 14 dpi, the abdominal incision wound had partially healed, while the pancreas appeared to have fully healed. The abdominal incision had fully healed by 28 dpi. No visible increase in tissue mass or blastema-like structure were observed at the site of pancreatic resection. Pancreatic resection studies in mice suggest that regeneration in pancreatic tissue can materialize in several forms, leading us to hypothesize that non-blastemal regeneration mechanisms such as transdifferentiation, endocrine cell-specific proliferation, or compensatory congestion are possible and could have been utilized by the axolotl in response to injury.²⁰

Figure 2. Pancreatic resection surgery model.

A. Cartoon depicting sham and pancreatic resection surgery. B. Morphology of sham and pancreatic resection surgery at 0, 14 and 28 dpi. The pancreas is outlined in blue while black arrows point to the duodenum. 0 dpi image of the resection condition was taken immediately following 20% removal of pancreas by length. The pancreas extends into the abdominal cavity where it connects to the liver (not visible in Figure B). In each image, the orientation from left to right follows a rostral to caudal direction. Black scale bars in the bottom right corner of each image are 2 mm.

Pancreatic resection leads to significant increases in cellular proliferation

To assess cellular proliferation, we used 5-ethynyl-2´-deoxyuridine (EdU) staining on coronal sections of pancreas (whole organ) from the sham and resection animal groups at both 14 and 28 dpi (Figure 3A & 3F). We used a single pulse of EdU, followed by a short chase period (18 hours), to identify all cells which have recently gone through S-phase of the cell cycle. Organ-wide proliferation was analyzed and compared to proliferation within a standardized area local to the site of injury. We found a significant increase in organ-wide cellular proliferation at 14 dpi in resected pancreas samples compared to the sham samples (Figure 3B). Additionally, proliferation local to the site of injury was significantly increased in comparison to an analogous area in sham samples (Figure 3C). This proliferative response was significantly increased local to the site of injury in comparison to organ-wide proliferation within resection samples (Figure 3D). Our findings suggest that at 14 dpi, the axolotl pancreas undergoes both an organ-wide proliferative response and a proliferative response local to the site of injury. Further, we found that nuclear density is significantly increased in resection samples local to the site of injury when compared to an analogous area in sham samples, indicating compensatory congestion (Figure 3E).

Figure 3. Pancreatic resection initially provokes substantial proliferative responses throughout the entire organ, but cell proliferation becomes restricted local to the site of injury later in the process.

A. EdU and insulin (IHC) stain of coronal sections of pancreas samples at 14 dpi (sham n=5, resection n=6). B. Organ-wide proliferation is increased in resection samples at 14 dpi (p=0.0162). C. Proliferation local to the site of injury is increased at 14 dpi (p=0.0012). D. Proliferation within resection samples is higher local to the site of injury at 14 dpi (p=0.0032).E. Nuclear density is increased locally in resection samples. F. EdU and insulin stain of coronal sections of pancreas samples at 28 dpi (sham n=5, resection n=5). G. Organ-wide proliferation in resection samples is non-significant in comparison to sham at 28 dpi (p=0.8459). H. Proliferation local to the site of injury is increased at 28 dpi (p=0.0079).I. At 28 dpi, proliferation local to the site of injury is higher than the organ-wide proliferation within resection samples (p=0.0085). J. Nuclear density is not significantly different between sham and resection at 28 dpi. K. Organ-wide proliferation is decreased between resection samples from 14 to 28 dpi (p=0.0009). L. Local proliferation is decreased between resection samples from 14 to 28 dpi (p=0.0005). M. Cartoon representation of pancreatic resection, color coded based on condition described in statistical analysis. Scale bars are 100 μm.

At 28 dpi, we found that the organ-wide proliferative response was comparable to the sham-operated controls (Figure 3G). However, the proliferative response local to the site of injury was still significantly increased in comparison to an analogous area in sham samples (Figure 3H). Likewise, the proliferative response local to the site of injury remained significantly increased in comparison to the organ-wide response within resection samples (Figure 3I). These results suggest that the whole-organ proliferative response had subsided at 28 dpi while the local proliferative response persisted. Further, we found that nuclear density is slightly increased in resection samples local to the site of injury when compared to an analogous area in sham samples, although not significantly; indicating nuclei may have spread throughout the pancreas after temporary congestion (Figure 3J).

We also compared the injury response between 14 and 28 dpi pancreas samples. Compared to 14 dpi, there was a significant decrease in proliferation both throughout the organ and local to the site of injury at 28 dpi (Figure 3K–L). We included a cartoon representing sham and resection conditions with color coding specifying global and local areas; corresponding color coding can be found throughout all plots in Figure 3.

Pancreatic resection induces β–cell regeneration

The identity of proliferating cells is important for understanding the regenerative mechanisms at play, which will ultimately determine whether an organ will be capable of proper function after regeneration occurs. We identified and quantified proliferating β–cells as EdU positive cells expressing insulin/ proinsulin protein in order to understand how this specific subset of endocrine cells responds to injury (Figure 4A–B). We found that β–cells proliferated at a significantly higher rate in resected samples when compared to the sham. In previous studies in mice, Dor et al. describe a 2.3-fold increase in β-cell proliferation by self-duplication as β-cell regeneration.¹⁰ On average, we observed a 6.2-fold increase in β-cell proliferation in resected samples, which is indicative of β–cell regeneration in response to injury (Figure 4C).¹⁰ Additionally, at 14 dpi, there was a significant increase in the number of β–cells in the resection condition; we refer to this as increased β–cell mass (Figure 4D). The observed increase in β-cell mass was almost double that of uninjured samples (Figure 4D). By 28 dpi, there was a non-significant difference in the amount of β–cells between the sham and resected samples (Figure 4E–F).

Figure 4. β–cell proliferation is observed at 14 days post injury but subsides by 28 days.

A. Representative image of a coronal section of regenerating pancreatic tissue at 14 dpi. White arrows point to proliferating β–cells. B. High resolution image of a proliferating β-cell (marked by white arrow), 90x magnification. C. A significant increase in β–cell proliferation was observed at 14 dpi (p = 0.0463) (sham n=5, resection n=6). D. A significant increase in β–cell population was observed at 14 dpi (p = 0.005) (sham n=5, resection n=6). E. No significant difference in β–cell proliferation was observed at 28 dpi (p = 0.0936) (sham n=5, resection n=5). F. No significant difference in β–cell population was observed at 28 dpi (p = 0.3181) (sham n=5, resection n=5). Scale bar is 200 μm in panel A, 20 μm in panel B.

Axolotl pancreatic resection induces transcriptional changes in well-studied injury response pathways

To identify changes in gene expression associated with pancreatic resection, we generated cDNA libraries (sham n=4, resection n=4) with polyadenylated RNA extracted from whole-organ pancreas tissue at 14 dpi. cDNA was sequenced using the Oxford Nanopore Technologies (ONT) PromethION sequencing platform. Sequencing produced 160.34M reads, of which 138.6M reads received a qscore greater than 8, as calculated by Dorado. After filtering by quality, 97.79% of reads were successfully demultiplexed by their respective barcode (Figure 5A). Classified reads were aligned to the axolotl transcriptome with a median read alignment identity of 98.41% and alignment N50 of 526.48 bp (Figure 5B).²¹ Subsequent analyses of the sequencing data showed 1618 differentially expressed genes between the sham and resection groups (Figure 5C–D).

Figure 5. Nanopore sequencing data reveals transcriptomic changes in response to injury.

A. Bar plot detailing reads per sample after demultiplexing. Unclassified reads and reads with a qscore less than 8 were filtered out. B. Histogram detailing aligned read identity vs total read counts. C. Heatmap of a subset of differentially expressed genes at 14 dpi D. Volcano plot of differentially expressed genes at 14 dpi.

Many of the significantly modulated genes detected have been implicated in axolotl limb regeneration such as Kazald2, which was found to be differentially upregulated in resection samples.²² We also found that genes Cirbp and Marco were differentially upregulated, which we validated via HCR-FISH (Figure 6A–B). We found several pancreatic development and maintenance genes to be differentially expressed such as Ctrb2, Ins, and Pdx1. Ctrb2 is a gene encoding chymotrypsin, an essential molecule produced by the exocrine pancreas to aid in digestion. We use Ctrb2 to visualize the distinction between the exocrine and endocrine pancreas (Figure 6A).

Figure 6: Genes previously implicated in regeneration exhibit upregulation in axolotl pancreatic injury model.

A. HCR-FISH showing differential expression of Ins and Marco transcripts. Ctrb2 was not found to be differentially expressed; however, it provides structural context for exocrine tissue. Marco+ cells shown with arrows. B. HCR-FISH showing differential expression of Ins and Cirbp transcripts. Scale bars are 50 μm.

Discussion

Axolotl pancreas morphology

Here, we show transcript-level expression of evolutionarily conserved pancreatic hormones such as insulin, glucagon and somatostatin in the axolotl via HCR-FISH. Additionally, key developmental markers such as Nkx6–1 and Pdx1 are shown to be conserved in this context. These transcriptional characteristics provide evidence that the axolotl may use similar gene regulatory networks as those which operate in mice and humans during pancreatic development and function.⁸

Pancreatic regenerative responses are observed after injury

Axolotl limb regeneration is well documented as a blastema-mediated regeneration process. Our observations from the pancreatic resection surgery do not show any blastema-like structure—the plane of resection remains visible, and no tissue growth was visually observed at the site of injury (Figure 2B). However, non-blastema mediated regeneration has been reported in the axolotl from studies on liver and ovary regeneration via compensatory congestion.^13,23 Liver regeneration in the axolotl materializes in whole-organ cell proliferation, which significantly increases cell density, and partly restores the mass of liver removed by partial hepatectomy (in contrast to mammalian compensatory growth which restores the full mass of the organ).¹³ At all times throughout the liver regeneration process, there is no blastemal structure, and the plane of resection remains visible.¹³ Likewise, ovaries utilize compensatory mechanisms by increasing proliferation and differentiation as a response to tissue loss.²³

In the context of axolotl pancreas regeneration, our data suggests a similar, non-blastema mediated regeneration process. We observed a whole-organ proliferative response over the course of 14 days which showed an overall reduction by 28 days (Figure 3J–K). Further, we measured a significant increase in nuclear density as a result of increased proliferation, suggesting compensatory congestion as a mechanism of pancreatic regeneration (Figure 3E). A portion of these proliferating cells were identified as β–cells and a significant increase in β–cell mass was observed at 14 dpi (Figure 4C–D). We measured a 6-fold increase in β-cell proliferation and a 1.8-fold increase in β-cell mass, indicating β-cell regeneration in response to injury. Interestingly, our sequencing results indicated that transcription of insulin was downregulated at the 14 dpi timepoint. We can evoke several different possible explanations for pancreatic cells downregulating insulin transcription post-injury. However, developing these hypotheses further and testing them will require additional experimentation outside the scope of this work.

Transcriptional shifts in pancreatic regeneration

Several important genes were found to be differentially expressed in injured samples at 14 days post pancreatic resection such as Kazald2, Cirbp, and Marco. Interestingly, Kazald2 and Cirbp are regeneration markers in other contexts such as axolotl limb regeneration.²² Marco is a macrophage marker, suggesting an immune response is provoked by pancreatic resection, which parallels the macrophage-mediated immune response necessary in axolotl limb regeneration and in a variety of other organ/appendage regeneration contexts in other species.^24,25,26 Confirmation of these hypotheses could lead to a mechanistic understanding of the pancreatic response to injury and whether it shares characteristics with other regenerative organs in the axolotl, future work investigating these pathways is crucial.

Additionally, we found Tgf-β1 to be differentially upregulated. Tgf-β1 is a key regulator of wound epithelium formation, blood clot formation, and inflammation in mammals, all of which are essential for regeneration to take place in axolotls.²⁷ Further, Levesque et al. functionally demonstrated Tgf-β1 to be essential in the initiation and control of the axolotl limb regeneration process.²⁷ Parallels in axolotl limb and pancreas regeneration may reveal conserved regeneration pathways that are present in all axolotl injury responses. To evaluate a possible functional role of Tgf-β1 in pancreatic regeneration, experimentally inhibiting Tgf-β1 in combination with pancreatic resection will be necessary.

Interestingly, both Pdx1 and Ins were found to be downregulated at 14 dpi. Ins is the transcript responsible for insulin production by β–cells. Pdx1 is a master regulator of pancreatic development and function.¹⁹ Pdx1 is expressed throughout the gastrointestinal tract and central nervous system during development; however, it is primarily expressed in β–cells in early development and persists into maturity.¹⁹ Investigating the downregulation of these transcripts is an active area of study and could lead to crucial elements of the mechanisms at play in pancreatic regeneration. Developing these hypotheses further and testing them will require additional experimentation outside the scope of this work.

Future studies should aim to investigate the functional role of notable genes in promoting or preventing pancreatic cell function and proliferation in the context of injury. The animals used for this study were too small to tolerate repeated blood draws of the quantity needed for ELISA assays. As such, our findings would benefit from future studies that identify whether the injury-induced increase in β–cell mass in axolotl allows for the maintenance of functional insulin homeostasis. Future work should also characterize the origin of new β–cells. There is significant debate in the field as to whether pancreatic stem cells exist in mammals.¹⁰ Axolotl salamanders have been reported to use several stem-cell-mediated mechanisms for regeneration.^26,28 Lineage tracing should be performed in axolotl pancreatic injury models to ascertain whether new β–cells originate from existing β–cells or from pancreatic stem cells. Further, it is worth noting that a large portion of EdU+ cells were negative for endocrine markers, which indicates a large part of the regenerative response is from the exocrine pancreas. Future work should aim to characterize the exocrine pancreatic response to injury.

We conclude that axolotl pancreatic islets closely resemble those in mice and humans, demonstrated by the conservation of several key pancreatic gene markers (Figure 1A–B). Further, a significant increase in β–cell proliferation is a direct parallel between mouse and axolotl pancreatic regeneration (Figure 4C).¹⁰ Although mouse models are very well characterized in the context of pancreatic regeneration, translating these insights into natural β–cell regeneration has not yet manifested in humans. We assert that the axolotl is an important animal model to study because its highly regenerative nature could reveal additional factors required to improve regenerative outcomes in humans.

Experimental Procedures

Pancreas resection surgery

All animal experimentation was approved by and conducted in accordance with Harvard University’s Institutional Animal Care and Use Committee (Protocol # 19–02-346). Adult wild type axolotl salamanders, 16 cm in length, were anesthetized in 0.1% tricaine solution and subjected to either a pancreatic resection surgery or a sham control surgery. The sham surgery consisted of a 2 inch lateral incision of the abdomen followed by gentle perturbation of the pancreas. The abdominal incision was closed with two horizontal mattress stitches using 6–0 nylon sutures. The resection surgery consisted of a similar 2 inch lateral incision of the abdomen followed by an approximate 20% resection of the pancreas starting from the base of the duodenum. Substantial bleeding was observed. Sterile wooden cotton tipped applicators were used to successfully clot the bleeding in all cases. The abdominal incision was closed with two horizontal mattress stitches using 6–0 nylon sutures. The animals were allowed to respond to this injury for 14 days, or 28 days post-incision, after which the whole pancreas was collected and used for histology/ sequencing.

Tissue collection and cryosectioning

Upon collection of the whole pancreas at 14 days, or 28 days post incision, animals were perfused with diethyl pyrocarbonate phosphate buffered saline (DEPC PBS) via aortic injection in order to clear tissue of blood and digestive enzymes while preserving RNA integrity. Tissue samples were subsequently preserved using various methods based on the intended analysis of that tissue:

Pancreas tissue samples intended for use in histology were transferred to a 4% paraformaldehyde in DEPC PBS solution upon collection. Samples were rocked overnight at 4°C. The following morning, samples were washed with DEPC PBS and transferred to a 30% sucrose in DEPC PBS solution. Samples were rocked overnight at 4°C. The following morning, samples were transferred to an OCT −30% sucrose mixture for 1 hour at room temperature. Subsequently, samples were embedded in 100% OCT on dry ice. Samples were stored at −80°C until further use. Tissue samples embedded in OCT were cryosectioned on the Leica cm 1950 cryostat at −22°C into 10 μm sections and mounted on Fisher Brand Superfrost Plus microscope slides. Samples were stored at −80°C.

Pancreas tissue samples intended for bulk RNA-seq were subject to a modified protocol based on Jun et al. which required samples to be injected with RNAlater upon collection and minced into small pieces to maximize tissue surface area contact with RNAlater²⁹. Samples were subsequently transferred into 500μL of RNAlater and snap frozen on dry ice. Samples were stored at −80°C until further use. RNA extraction was completed using a phenol/chloroform phase separation with subsequent clean up using the Zymo Clean & Concentrator 5 kit.

EdU incorporation

A stock 5-ethynyl-2´-deoxyuridine (EdU) solution was made by dissolving EdU in dimethylsulfoxide (DMSO) at a concentration of 5 μg/μL. This stock solution was diluted to 0.1 μg/μL in 0.7× PBS for injection. Animals were anesthetized in 0.1% tricaine solution for 30 minutes prior to injection. EdU solution was injected intraperitoneally with care not to puncture internal organs. Each animal received 20 μL per gram body weight of the injection solution. Animals were returned to normal housing conditions to incorporate with EdU for 18 hours post injection. Tissue collection immediately followed.

EdU and immunohistochemistry

Slides were rehydrated in 1× PBS and permeabilized with 0.5% Triton-X PBS (TX-PBS) for 30 minutes. EdU reaction was completely using copper sulfate (CuSO₄), ascorbic acid (C₆H₈O₆) and sulfo-Cy3 azide. Slides were blocked with 2% BSA in 0.1% TX-PBS for 30 minutes followed by addition of primary antibody (1:200 in blocking solution) against insulin/ proinsulin overnight (D3E7 (5B6/6), ThermoFisher Scientific, cat no. MA1–83256). Secondary antibody was added for 2 hours (AlexaFluor Donkey Anti-Mouse 647, ThermoFisher Scientific, cat no. A-31571).

HCR FISH

Oligonucleotides were designed using a probe generator model developed by the Monaghan Lab at Northeastern University with transcripts identified in our sequencing data through alignment to the Nowoshilow transcriptome²¹. HCR stain was completed as described by Lovely et al.¹⁸.

Imaging and quantification

Histological imaging was performed using the Nikon Spinning Disk Microscope, Zeiss Axioscan Microscope, and Zeiss LSM 900 Microscope. EdU & insulin/ proinsulin stains were quantified using Cell Profiler and Qupath. Technical replicates were quantified individually and then averaged into a single biological replicate. For EdU+ nuclei, all EdU+ nuclei in a section were counted and normalized to the total number of nuclei in that section. This was expressed as %EdU+ nuclei, calculated as (EdU+/total nuclei) * 100. Qupath used a spatial categorization method to count insulin/ proinsulin & EdU+ nuclei. Similarly, the proportion of insulin/proinsulin and EdU+ nuclei was determined by counting these nuclei in a section and normalizing them to the total number of nuclei in the section, expressed as %insulin & EdU+ nuclei = (insulin & EdU+/total nuclei) × 100. Generally, tissue sections were composed of 5–10 thousand total nuclei. Biological replicates at 14 dpi: Sham n=5; Resection n=6. Biological replicates at 28 dpi: Sham n=5; Resection n=5.

For selection of area local to the site of resection, a 350 × 500 μm control area was selected randomly along the plane of resection in histology images. Nuclei within the control area were analyzed as described above. A comparable area in sham samples was selected by utilizing a 350 × 500 μm control area placed along the edge of the tissue section. Nuclei in the control area were analyzed as described above. Randomized repetitions of control area placement yielded comparable results.

RNA extraction

Whole pancreas tissue from the sham and resection groups (n=4) was transferred to 500mL of Trizol and homogenized for 30-seconds with the Bio-Gen PRO200 Hand-Held homogenizer. A phenol chloroform phase separation was performed, and the aqueous layer was transferred to Zymo spin columns. The RNA was cleaned up following the protocol for the Zymo Clean & Concentrator 5 kit. RNA quality was determined by Nanodrop spectrophotometer, Tapestation, and Qubit.

Nanopore cDNA Sequencing

High quality RNA from the previous step was library prepped for sequencing according to protocol for the ONT PCR cDNA kit with barcode capability (SQK-PCB111.24). Each sample was barcoded, the libraries were combined and sequenced simultaneously on two PromethION flow cells.

Sequencing data was analyzed using a custom in-house Nextflow pipeline. This pipeline utilized Dorado to basecall, Minimap2 to align reads to the axolotl transcriptome, and featureCounts to quantify the data.²¹ Differential gene expression was calculated using DESeq2.