Ductal or Ngn3⁺ cells do not contribute to adult pancreatic islet beta-cell neogenesis in homeostasis

Abstract

The adult pancreatic ducts have long been proposed to contain rare progenitors, some of which expressing Ngn3, that generate new beta cells in endocrine-islet homeostasis. Due to their postulated rarity and the lack of definitive markers, the existence or absence of ductal endocrine progenitors remains unsettled despite many studies. Genetic lineage tracing of ductal cells or Ngn3⁺ cells with currently available CreER drivers has been complicated by off-target labeling of pre-existing beta cells. Here, using dual-recombinase-mediated intersectional genetic strategy and newly-derived Ngn3-2A-CreER and Hnf1b-2A-CreER knock-in drivers, we succeeded in specifically labeling Ngn3-positive cells and Hnf1b-positive ductal cells without marking pre-existing beta cells. These data revealed no evidence of de novo generation of insulin-producing beta cells from ductal cells or endogenous Ngn3-positive cells in the adult pancreas during homeostasis.

Synopsis

Adult pancreatic ducts have long been proposed to contain rare progenitors that can contribute to endocrine islets. Using intersectional genetic tracing approaches, this study finds no evidence that pancreatic ductal cells or endogenous Ngn3-positive non-beta cells in the adult mouse pancreas generate beta cells during homeostasis.

• New knock-in mouse lines Ngn3-2A-CreER and Hnf1b-2A-CreER allow efficient lineage tracing in adult pancreas.

• Dual-recombinase-mediated genetic strategies achieve improved precision of pancreatic-lineage labeling.

• Pancreatic ductal cells or ductal Ngn3⁺ progenitors do not contribute to new beta cells during adult homeostasis.

• After extreme beta-cell loss, Ngn3⁺ non-beta cells can generate new beta cells, while the number of ductal Ngn3⁺ cells does not increase.

Dual-recombinase-mediated genetic tracing with newly developed CreER drivers shows no evidence of de novo generation of insulin-producing beta cells from ductal progenitors.

Introduction

A major cause of type 1 diabetes and a subset of type 2 diabetes is the deficiency of functional beta cells in the pancreas (Aguayo-Mazzucato and Bonner-Weir, 2018; Zhou and Melton, 2018). Understanding potential sources of new beta cells in the adult pancreas will facilitate both basic and clinical research and inform the treatment of insulin-dependent diabetes. Numerous studies have reported on two major replenishing sources of the beta cell mass: self-replication of pre-existing beta cells and beta cell neogenesis (generation of new beta cells from progenitors or non-beta cells without genetic manipulation) (Zhou and Melton, 2018). The self-replication model was first proved by Dor et al (Dor et al, 2004), with other research groups subsequently coming to the same conclusion using genetic tools (Blaine et al, 2010; Cox et al, 2016; Nir et al, 2007; Perez-Frances et al, 2022; Rankin et al, 2013; Xiao et al, 2013a; Zhao et al, 2021a) or chemical labeling (Teta et al, 2007). The neogenesis model often invokes the existence of progenitor cells expressing Neurogenin 3 (Neurog3, also known as Ngn3), a master regulator of endocrine specification, in extra-islet locations such as the pancreatic ducts (Gradwohl et al, 2000; Gribben et al, 2021; Gribben et al, 2023; Gu et al, 2002; Magenheim et al, 2023; Seymour et al, 2008; Solar et al, 2009; Xu et al, 2008; Yu et al, 2016; Zhou et al, 2007). In 2008, Xu et al utilized a transgenic Nng3-nLacZ mouse line and reported for the first time that adult pancreatic Ngn3⁺ progenitors could contribute to the generation of beta cells after partial duct ligation (PDL) injury (Xu et al, 2008). Other studies, however, showed that Ngn3 mRNA and protein could be detected in pre-existing beta cells (Wang et al, 2009), and although PDL injury could activate Ngn3 expression in pancreatic ducts, it does not lead to beta cell neogenesis from duct cells (Kopp et al, 2011; Xiao et al, 2013b).

In addition to Ngn3⁺ cells, genetic lineage tracing was used extensively to detect endocrine differentiation from pancreatic ducts. Evidence for and against ductal endocrine progenitors was put forward by different groups (Furuyama et al, 2011; Inada et al, 2008; Kopinke et al, 2011; Kopinke and Murtaugh, 2010; Kopp et al, 2011; Solar et al, 2009), with the most recent being Gribben et al reporting a population of ductal Ngn3⁺ progenitors making a sizable contribution to beta cell mass in homeostasis (Gribben et al, 2021). This report sparked a vigorous debate (Bonner-Weir, 2021; Gribben et al, 2023; Magenheim et al, 2023; Zhao et al, 2021b). In a Matters Arising, Magenheim et al provided new data to challenge this conclusion (Magenheim et al, 2023).

In our view, much of the uncertainty about ductal- or exocrine-resident endocrine progenitors can be traced to deficiencies in the genetic lineage tracing tools presently available. In particular, Ngn3-CreER and Hnf1b-CreER (used in tracing pancreatic ductal cells) have “off-target” labeling of pre-existing beta cells. In addition, the modest labeling efficiency in some of the ductal tracing experiments was criticized as potentially missing rare ductal cell populations (Gribben et al, 2023). To resolve these issues, we need genetic tools to mark ducts at exceptionally high efficiency without off-target labeling. Here, we generated new knock-in mouse lines Ngn3-2A-CreER and Hnf1b-2A-CreER, which do not affect endogenous gene expression and are highly efficient in the pancreas. Using a dual-recombinase-mediated intersectional genetic approach, we systematically investigated whether Ngn3-expressing progenitors and pancreatic ductal epithelial cells contribute to beta cell neogenesis during homeostasis in adults. Our results unambiguously support the contention that Nng3-expressing progenitors and pancreatic ductal cells do not generate beta cells de novo in adult pancreas homeostasis.

Results

Ngn3-CreER and Ngn3-2A-CreER label ductal cells and also various islet cells

The Ngn3-CreER transgenic line used in Gribben et al (Gribben et al, 2021) appears to label a population of pre-existing beta cells in addition to rare ductal cells. Magenheim and colleagues (Magenheim et al, 2023) further analyzed a Ngn3-CreER knockin line that labels islet but not ductal cells. The question arises as to exactly what cell lineages Ngn3-CreER should be labeling that faithfully recapitulate its endogenous expression.

We first evaluated the transgenic mouse line Ngn3-CreER (Gu et al, 2002) as Gribben et al reported (Gribben et al, 2021), and crossed it with R26-tdT (R26-loxP-Stop-loxP-tdTomato) (Madisen et al, 2010) mice to generate Ngn3-CreER;R26-tdT for pulse-and-chase experiments (Fig. 1A). We treated Ngn3-CreER;R26-tdT with Tam (tamoxifen) at the adult stage and collected the pancreases for analysis at 2 weeks and 12 weeks afterwards (Fig. 1B). Wholemount fluorescent images highlighted localized tdT (tdTomato) signals in pancreatic tissues collected at both 2 weeks and 12 weeks after Tam induction (Fig. 1C). To clarify the identities of the tdT⁺ cells, we performed immunostaining for tdT and the pancreatic ductal cell marker CK19 on tissue sections. We found few tdT⁺ cells lining the ductal epithelium, similar to the findings of Gribben et al (Gribben et al, 2021) (Fig. 1D). We also found that the ductal tdT⁺ cells expressed insulin (Ins) and the delta cell marker somatostatin (Sst) (Fig. 1D). Consistent with the findings of Magenheim et al (Magenheim et al, 2023), a substantial proportion of pancreatic endocrine cells (including beta cells, delta cells, alpha cells, and PP cells) were labeled at both the “pulse” time point (2 weeks post-Tam) and the “chase” time point (12 weeks of post-Tam) in Ngn3-CreER;R26-tdT mice (Fig. 1E). Quantitively, 16.86  ± 10.77% of the beta cells were tdT⁺ at the “pulse” time point, and the percentage of pulse-labeled beta cells did not increase significantly after chase period (23.47 ± 5.04% at 12 weeks post-Tam) (Fig. 1F).

Figure 1. Genetic lineage tracing of Ngn3⁺ cells in the pancreas.

(A) Schematic diagram of the lineage tracing strategy by Ngn3-CreER;R26-tdT mice. (B) Schematic showing the experimental strategy. (C) Wholemount fluorescent images of pancreas collected from Ngn3-CreER;R26-tdT at the indicated time points after Tam treatment. Arrowheads, tdT⁺ pancreatic islets. (D) Immunostaining for tdT and CK19 or Ins or somatostatin (Sst) on pancreatic sections of Ngn3-CreER;R26-tdT. (E) Immunostaining for tdT and Ins or Sst or glucagon (Gcg) or pancreatic polypeptide (Ppy) on pancreatic sections of Ngn3-CreER;R26-tdT mice. Arrowheads, tdT⁺lineage⁺ pancreatic endocrine cells. (F) Quantification of the percentage of tdT⁺ cells among pancreatic endocrine cell lineages in the Ngn3-CreER;R26-tdT mice. Data are mean ± SD; +2 weeks, n = 5 biological replicates; +12 weeks, n = 4 biological replicates; Ins: P = 0.30; Sst: P = 0.72; Gcg: P = 0.97; PP: P = 0.62; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t-tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. (G) Schematic of the Ngn3-2A-CreER knock-in strategy. (H) Schematic of lineage tracing strategy for Ngn3-2A-CreER;R26-tdT. (I) Schematic showing the experimental strategy. (J) Wholemount fluorescent images of the pancreas from Ngn3-2A-CreER;R26-tdT mice at the indicated time-points after Tam treatment. Arrowheads, tdT⁺ pancreatic islets. (K, L) Immunostaining for tdT and E-cad (K) or CK19 or Ins or Sst (L) on pancreatic sections of Ngn3-2A-CreER;R26-tdT mice. Arrowheads in (K), tdT⁺ islets. (M) Immunostaining for tdT and Ins or Sst or Gcg or Ppy on pancreatic sections of Ngn3-2A-CreER;R26-tdT Mice. Arrowheads, tdT⁺lineage⁺ pancreatic endocrine cells. (N) Quantification of the percentage of tdT⁺ cells in pancreatic endocrine cell lineages of Ngn3-2A-CreER;R26-tdT mice. Data are mean ± SD; +2 weeks, n = 5 biological replicates; +12 weeks, n = 5 biological replicates; Ins: P = 0.76; Sst: P = 0.26; Gcg: P = 0.42; PP: P = 0.67; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. (O) Cartoon image showing that Ngn3-CreER and Ngn3-2A-CreER label a subset of ductal cells and endocrine cells in the adult pancreas. Scale bars, 1 mm (yellow) and 100 μm (white). Each image is representative of 4–5 individual mouse samples. See also Fig. EV1. Source data are available online for this figure.

The above mouse line, Ngn3-CreER, was generated based on random transgene integration (Gu et al, 2002). The Ngn3-CreER knockin line used in the Magenheim study inactivated one Ngn3 allele, which may perturb endogenous Ngn3 transcription (Magenheim et al, 2023). We therefore decided to create a new mouse line, Ngn3-2A-CreER by inserting the CreER into the 3’UTR (3’ untranslated region) of the endogenous Ngn3 gene, preserving endogenous Ngn3 transcription (Fig. 1G). We crossed Ngn3-2A-CreER with the R26-tdT reporter mouse line to obtain Ngn3-2A-CreER;R26-tdT mice (Fig. 1H) and analyzed the mice at 2 weeks and 12 weeks post-Tam (Fig. 1I). Wholemount imaging of Ngn3-2A-CreER;R26-tdT pancreases revealed tdT⁺ signal patterns similar to that of Ngn3-CreER;R26-tdT (Fig. 1J). Ngn3-2A-CreER labeled scattered ductal cells, the majority of Sst⁺ delta cells in islets (88.08 ± 12.89%), and a sizable population of Ins⁺ beta-cells (32.18 ± 5.34%) at 2 weeks post-Tam (Fig. 1K–M). At 12 weeks post-Tam, the proportion of labeled endocrine cells did not increase significantly over time (Fig. 1N). Without Tam, very few tdT⁺ cells were detected in the islets from Ngn3-CreER;R26-tdT (0.04 ± 0.03%), and Ngn3-2A-CreER;R26-tdT (0.02 ± 0.02%) (Fig. EV1A–J).

Figure EV1. Characterization of Ngn3-CreER;R26-tdT and Ngn3-2A-CreER;R26-tdT without tamoxifen treatment, related to Fig. 1.

(A) Schematic showing the experimental strategy of Ngn3-CreER;R26-tdT with oil treatment. (B) Whole-mount fluorescent images of pancreas collected from Ngn3-CreER;R26-tdT after oil treatment. (C, D) Immunostaining for tdT and CK19 (C) or Insulin (Ins, D) on pancreatic sections of Ngn3-CreER;R26-tdT after oil treatment. (E) Quantification of the percentage of tdT⁺ cells in Ins⁺ beta cells of Ngn3-CreER;R26-tdT after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (F) Schematic showing the experimental strategy of Ngn3-2A-CreER;R26-tdT with oil treatment. (G) Whole-mount fluorescent images of pancreas collected from Ngn3-2A-CreER;R26-tdT after oil treatment. (H, I) Immunostaining for tdT and CK19 (H) or Ins (I) on pancreatic sections of Ngn3-2A-CreER;R26-tdT after oil treatment. (J) Quantification of the percentage of tdT⁺ cells in Ins⁺ beta cells of Ngn3-2A-CreER;R26-tdT after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. Scale bars, yellow, 1 mm; white, 100 μm. Each image is representative of 5 individual samples.

These data demonstrate that the transgenic Ngn3-CreER and the knock-in Ngn3-2A-CreER both labeled hormone-positive endocrine cells in the ductal epithelium and pancreatic islets (Fig. 1O). The similar labeling percentage of endocrine cells from 2 weeks to 12 weeks post-Tam suggests a lack of substantial de novo proliferation or transdifferentiation by the labeled cells during the pulse-and-chase time window. These data also highlighted difficulties in interpreting the lineage tracing data due to “ectopic” labeling of pre-existing endocrine cells.

Simultaneous tracing of Ngn3⁺ non-beta cells, Ngn3⁺ beta cells, and Ngn3^– beta cells by a traffic light reporter

A dual-recombinase-based lineage tracing strategy that combines the Cre/loxP and Dre/rox systems could allow more specific labeling of particular cell lineages (He et al, 2017). We employed the R26-TLR (Rosa26-traffic light reporter) mouse line, which has three distinct fluorescent reporters that can trace Cre⁺Dre⁺, Cre^–Dre⁺, Cre⁺Dre^– cells simultaneously in one mouse (Liu et al, 2020). We also utilized the highly efficient insulin-specific inducible mouse line Ins2-DreER, which labels over 99% of beta cells (Zhao et al, 2021a). We generated Ins2-DreER;Ngn3-CreER;R26-TLR mice (Fig. 2A) in which Ins⁺ beta cells were marked by ZsGreen, Ngn3⁺Ins^– cells by tdT, and Ngn3⁺Ins⁺ cells by both ZsGreen and tdT (Fig. 2A). In this design, if Ngn3⁺Ins^– non-beta cells, which contain putative progenitors, give rise to new beta cells, then a subset of beta cells would be marked by tdT⁺ only after the chase period.

Figure 2. Simultaneous tracing of Ngn3⁺Ins^– non-beta cells, Ngn3⁺Ins⁺ beta cells, and Ngn3^–Ins⁺ beta cells by a triple light reporter.

(A) Schematic illustrating the labeling strategy by Ins2-DreER;Ngn3-CreER;R26-TLR mice. (B) Schematic showing the experimental strategy. (C) Wholemount fluorescent images of pancreas collected from Ins2-DreER;Ngn3-CreER;R26-TLR mice after 2 weeks of Tam treatment. Arrowheads, pancreatic islets. (D) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Ngn3-CreER;R26-TLR mice. (E) Immunostaining for ZsGreen, tdT, and Sst on pancreatic sections of Ins2-DreER;Ngn3-CreER;R26-TLR mice. (F, G) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Ngn3-CreER;R26-TLR mice after 2 weeks (F) and 12 weeks (G) of Tam treatment. Arrowheads, ZsGreen⁺tdT⁺Ins⁺ pancreatic beta cells. (H) Quantification of the percentage of ZsGreen^–tdT⁺ and ZsGreen⁺tdT⁺ cells and ZsGreen⁺ cells in Ins⁺ beta cells of Ins2-DreER;Ngn3-CreER;R26-TLR mice. Data are mean ± SD; +2 weeks, n = 5 biological replicates; +12 weeks, n = 6 biological replicates; ZsGreen^–tdT⁺: P = 0.13; ZsGreen⁺tdT⁺: P = 0.71; ZsGreen⁺: P = 0.51; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. (I) Schematic diagram showing the experimental strategy. (J) Wholemount fluorescence images of the pancreas from Ins2-DreER;Ngn3-2A-CreER;R26-TLR mice after 2 weeks of Tam treatment. Arrowheads, pancreatic islets. (K) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;R26-TLR mice after 2 weeks (top) and 12 weeks (bottom) of Tam treatment. (L, M) Immunostaining for ZsGreen, tdT, and Sst (L) or Ins (M) on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;R26-TLR mice after 2 weeks (top) and 12 weeks (bottom) of Tam treatment. Arrowheads, ZsGreen⁺tdT⁺Ins⁺ pancreatic beta cells. (N) Quantification of the percentages of ZsGreen^–tdT⁺ and ZsGreen⁺tdT⁺ cells and ZsGreen⁺ cells in Ins⁺ beta cells. Data are mean ± SD; +2 weeks, n = 4 biological replicates; +12 weeks, n = 6 biological replicates; ZsGreen^–tdT⁺: P = 0.73; ZsGreen⁺tdT⁺: P = 0.55; ZsGreen⁺: P = 0.36; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. (O) Cartoon image showing simultaneous tracing of Ngn3⁺Ins^– non-beta cells, Ngn3⁺Ins⁺ beta cells, and Ngn3^–Ins⁺ beta cells in the pancreas using Ins2-DreER;Ngn3-CreER;R26-TLR and Ins2-DreER;Ngn3-2A-CreER;R26-TLR mice. Ngn3⁺Ins^– non-beta cells do not contribute to beta cell neogenesis during homeostasis. Scale bars, 1 mm (yellow) and 100 μm (white). Each image is representative of 4–6 individual mouse samples. See also Fig. EV2. Source data are available online for this figure.

We first validated that nearly all Ins⁺ beta cells (99.92 ± 0.07%) could be labeled with ZsGreen in Ins2-DreER;R26-TLR after Tam injection with high specificity (Fig. EV2A–F). Next, analyzing samples of Ins2-DreER;Ngn3-CreER;R26-TLR pancreases 2 weeks after Tam, we readily detected ZsGreen and tdT fluorescence in wholemount preparations (Fig. 2B,C). Combinatorial staining of pancreatic sections with antibodies against ZsGreen, tdT, CK19, Sst, and Ins revealed three distinct cell populations in ducts and islets: tdT⁺ZsGreen^–Ins^– non-beta cells, tdT^–ZsGreen⁺Ins⁺ beta cells, and tdT⁺ZsGreen⁺ beta cells (i.e. Ngn3⁺Ins⁺ beta cells) (Fig. 2D–F). The vast majority of tdT⁺ZsGreen^– non-beta cells were located in the pancreatic islets, most of which were pancreatic Sst⁺ delta cells, and approximately single-digit numbers of tdT⁺ZsGreen^– non-beta cells could be found located in the pancreatic ductal epithelium per tissue section, in agreement with the findings of Gribben et al (Gribben et al, 2021) (Fig. 2D–F). 12 weeks post-Tam, we did not detect any Ins⁺ cells expressing tdT single fluorescence in ducts or in islets (Fig. 2G). Quantitative data showed that none of Ins⁺ beta cells was labeled with tdT at 2 or 12 weeks after Tam treatment (Fig. 2H, 2 weeks: 0.02 ± 0.01%; 12 weeks: 0.06 ± 0.06%). Moreover, there was no significant difference in the proportion of ZsGreen⁺tdT⁺ beta cells at 2 weeks (18.43 ± 5.54%) or at 12 weeks (17.42 ± 2.82%) (Fig. 2H), suggesting a lack of substantial proliferative advantage for the Ngn3⁺Ins⁺ beta cell subpopulation during homeostasis. In addition, we found no detectable dilution of ZsGreen⁺ beta cells labeled with Ins2-DreER from 2 weeks to 12 weeks (99.68 ± 0.31% in 2 weeks and 99.79 ± 0.22% in 12 weeks, Fig. 2H). These data together indicate that Ngn3⁺ non-beta cells do not generate any detectable new beta cells during homeostasis, consistent with the finding of previous studies (Dor et al, 2004; Perez-Frances et al, 2022; Xiao et al, 2013a; Zhao et al, 2021a).

Figure EV2. Characterization of Ins2-DreER;R26-TLR, and characterization of Ins2-DreER;Ngn3-CreER;R26-TLR and Ins2-DreER;Ngn3-2A-CreER;R26-TLR without tamoxifen treatment, related to Fig. 2.

(A) Schematic showing the labeling strategy of Ins2-DreER;R26-TLR. (B) Schematic showing the experimental strategy of Ins2-DreER;R26-TLR with tamoxifen (Tam) treatment. (C) Whole-mount fluorescent images of pancreas from Ins2-DreER;R26-TLR. Arrowheads, zsGreen⁺ islets. (D) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;R26-TLR with Tam treatment. Arrowheads, zsGreen⁺ beta cells. (E) Immunostaining for ZsGreen and Ins on serial sections of one pancreatic islet from Ins2-DreER;R26-TLR with Tam treatment. (F) Quantification of the percentages of ZsGreen⁺tdT^– or tdT⁺ZsGreen^– or ZsGreen⁺tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;R26-TLR. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (G) Schematic showing the experimental strategy of Ins2-DreER;Ngn3-CreER;R26-TLR with oil treatment. (H) Whole-mount fluorescent images of pancreas from Ins2-DreER;Ngn3-CreER;R26-TLR after oil treatment. (I) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;Ngn3-CreER;R26-TLR after oil treatment. (J) Quantification of the percentage of ZsGreen⁺tdT^– or tdT⁺ZsGreen^– or ZsGreen⁺tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Ngn3-CreER;R26-TLR after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (K) Schematic showing the experimental strategy of Ins2-DreER;Ngn3-2A-CreER;R26-TLR with oil treatment. (L) Whole-mount fluorescent images of pancreas from Ins2-DreER;Ngn3-2A-CreER;R26-TLR after oil treatment. (M) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;R26-TLR after oil treatment. (N) Quantification of the percentage of ZsGreen⁺ or tdT⁺ or ZsGreen⁺tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Ngn3-2A-CreER;R26-TLR after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. Scale bars, yellow, 1 mm; white, 100 μm. Each image is representative of 5 individual samples.

To validate the above finding, we crossed Ins2-DreER;R26-TLR with the knockin Ngn3-2A-CreER (Fig. 2I). Wholemount fluorescence imaging of Tam-treated pancreases revealed a patchy distribution of ZsGreen⁺ islets (Fig. 2J). Combinatorial immunostaining revealed no Ins⁺ beta cells expressing tdT single fluorescence at either 2 weeks or 12 weeks (Fig. 2J–N, 2 weeks: 0.04 ± 0.03%; 12 weeks: 0.03 ± 0.03%), similar to the results of the Ins2-DreER;Ngn3-CreER;R26-TLR system. As controls, both Ins2-DreER;Ngn3-CreER;R26-TLR and Ins2-DreER;Ngn3-2A-CreER;R26-TLR mice expressed virtually no fluorescent proteins in the absence of Tam (Fig. EV2G–N). The above data demonstrated that Ngn3⁺Ins^– non-beta cells (containing Ngn3⁺ progenitors) do not give rise to new Ins⁺ beta cells under homeostatic conditions in adults (Fig. 2O).

Ngn3⁺ non-beta cells did not contribute to beta cells during homeostasis in adults

To independently address whether Ngn3⁺ non-beta cells contribute to beta cells in adults, we adopted another nested reporter, NR1 (He et al, 2017), in which Cre-based lineage tracing is controlled by Dre/rox recombination. Briefly, this strategy using Ins2-DreER;Ngn3-CreER;NR1 mice labels Ngn3⁺ non-beta cells with ZsGreen, whereas all beta cells including Ngn3⁺ beta cells, are labeled with tdT, as any Cre-activated reporter genes are excised by Dre/rox in beta cells (Fig. 3A). In this design, if Ngn3⁺ non-beta cells could differentiate into beta cells during the chase period, we would detect ZsGreen⁺tdT^– beta cells at the chase time point but not before (Fig. 3A).

Figure 3. Nested strategy shows Ngn3⁺Ins^– non-beta cells do not contribute to Ins⁺ beta cells during adult homeostasis.

(A) Schematic of the labeling strategy by Ins2-DreER;Ngn3-CreER;NR1 mice. (B) Schematic showing the experimental strategy. (C) Wholemount fluorescence image of the pancreas from Ins2-DreER;Ngn3-CreER;NR1 mice after 2 weeks of Tam treatment. Arrowheads, pancreatic islets. (D) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Ngn3-CreER;NR1 mice after 2 weeks of Tam treatment. (E) Immunostaining for ZsGreen, tdT, and Sst on pancreatic sections of Ins2-DreER;Ngn3-CreER;NR1 mice. Arrowheads, ZsGreen⁺Sst⁺ pancreatic delta cells. (F, G) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Ngn3-CreER;NR1 mice after 2 weeks (F) and 12 weeks (G) of Tam treatment. Arrowheads, tdT⁺Ins⁺ pancreatic beta cells. (H) Quantification of the percentages of ZsGreen⁺ or tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Ngn3-CreER;NR1 mice. Data are mean ± SD; +2 weeks, n = 5 biological replicates; +12 weeks, n = 5 biological replicates; tdT⁺: P = 0.14; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. (I) Schematic showing the experimental strategy. (J) Wholemount fluorescence images of the pancreas from Ins2-DreER;Ngn3-2A-CreER;NR1 mice after 2 weeks (left) and 12 weeks (right) of Tam treatment. (K) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;NR1 mice after 2 weeks (top) and 12 weeks (bottom) of Tam treatment. (L, M) Immunostaining for ZsGreen, tdT, and Sst (L) or Ins (M) on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;NR1 mice after 2 weeks (top) and 12 weeks (bottom) of Tam treatment. Arrowheads, ZsGreen⁺Sst⁺ pancreatic delta cells (L) and tdT⁺Ins⁺ pancreatic beta cells (M). (N) Quantification of the percentages of ZsGreen⁺ or tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Ngn3-2A-CreER;NR1 mice. Data are mean ± SD; +2 weeks, n = 4 biological replicates; +12 weeks, n = 4 biological replicates; tdT⁺: P = 0.09; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t tests were used for statistical comparisons and p < 0.05 was accepted as statistically significant. (O) Cartoon image showing tracing of Ngn3⁺Ins^– non-beta cells and Ins⁺ beta cells in the pancreas with distinct reporters using Ins2-DreER;Ngn3-CreER;NR1 and Ins2-DreER;Ngn3-2A-CreER;R26-NR1 mice. Ngn3⁺Ins^– non-beta cells do not convert to Ins⁺ beta cells in homeostatic condition. Scale bars, 1 mm (yellow) and 100 μm (white). Each image is representative of 4–5 individual mouse samples. See also Fig. EV3. Source data are available online for this figure.

We first examined the labeling efficiency of beta cells using Ins2-DreER;NR1, and found that virtually all Ins⁺ beta cells (99.87 ± 0.21%) could be labeled by Dre/rox recombination (Fig. EV3A–F). Next, we treated Ins2-DreER;Ngn3-CreER;NR1 with Tam to simultaneously trace Ngn3⁺Ins^– non-beta cells (ZsGreen⁺) and beta cells (tdT⁺) in the pancreas (Fig. 3A,B). Wholemount imaging revealed sporadic tdT⁺ islet signals in the pancreas collected 2 weeks after Tam treatment (Fig. 3C). Immunostaining for ZsGreen, tdT, CK19, Ins, and Sst showed that the Ins⁺ cells in the ductal epithelium were labeled with tdT but not ZsGreen, whereas some ductal Sst⁺ cells were labeled with ZsGreen (Fig. 3D). In islets, many Sst⁺ delta cells expressed the ZsGreen reporter but Ins⁺ beta cells did not express ZsGreen (0.00 ± 0.00%) (Fig. 3E,F). In pancreatic sections collected 12 weeks after Tam, 99.28 ± 0.53% of Ins⁺ beta cells expressed tdT. Importantly no ZsGreen⁺tdT^–Ins⁺ beta cells (0.00 ± 0.00%) were observed in ducts or in islets (Fig. 3G,H).

Figure EV3. Characterization of Ins2-DreER;NR1, and characterization of Ins2-DreER;Ngn3-CreER;NR1 and Ins2-DreER;Ngn3-2A-CreER;NR1 without tamoxifen treatment, related to Fig. 3.

(A) Schematic showing the labeling strategy of Ins2-DreER;NR1. (B) Schematic showing the experimental strategy of Ins2-DreER;NR1 with Tam treatment. (C) Whole-mount fluorescent images of pancreas from Ins2-DreER;NR1 with Tam treatment. Arrowheads, tdT⁺ islets. (D) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;NR1 with Tam treatment. Arrowheads, tdT⁺ beta cells. (E) Immunostaining for tdT and Ins on serial sections of one pancreatic islet from Ins2-DreER;NR1 with Tam treatment. (F) Quantification of the percentage of ZsGreen⁺ or tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;NR1. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (G) Schematic showing the experimental strategy of Ins2-DreER;Ngn3-CreER;NR1 with oil treatment. (H) Whole-mount fluorescent images of pancreas from Ins2-DreER;Ngn3-CreER;NR1 after oil treatment. (I) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;Ngn3-CreER;NR1 after oil treatment. (J) Quantification of the percentage of ZsGreen⁺ or tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Ngn3-CreER;NR1 after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (K) Schematic showing the experimental strategy of Ins2-DreER;Ngn3-2A-CreER;NR1 with oil treatment. (L) Whole-mount fluorescent images of pancreas from Ins2-DreER;Ngn3-2A-CreER;NR1 after oil treatment. (M) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;NR1 after oil treatment. (N) Quantification of the percentage of ZsGreen⁺ or tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Ngn3-2A-CreER;NR1 after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. Scale bars, yellow, 1 mm; white, 100 μm. Each image is representative of 5 individual samples.

Next, we used Ngn3-2A-CreER mice and generated Ins2-DreER;Ngn3-2A-CreER;NR1 triple knock-in mice (Fig. 3I). Analysis of pancreases collected at 2 weeks and 12 weeks post-Tam revealed that no ZsGreen⁺ tdT^–Ins⁺ beta cells were generated during the chase period (Fig. 3J–N, 2 weeks: 0.00 ± 0.00%; 12 weeks: 0.00 ± 0.00%). As controls, we did not detect any fluorescent signal in the islets of both Ins2-DreER;Ngn3-CreER;NR1 and Ins2-DreER;Ngn3-2A-CreER;NR1 mice in the absence of Tam (Fig. EV3G–N). In summary, these data indicate that during homeostasis, both Ngn3⁺ ductal cells and Ngn3⁺ non-beta islet cells (most being Sst⁺ cells) did not undergo lineage (trans) differentiation toward beta cells (Fig. 3O).

Ngn3⁺ non-beta cells gave rise to beta cells after extreme beta-cell loss

Previous studies have reported that non-beta islet cells, including alpha and delta cells, can generate new beta cells after extreme beta cell loss (Chera et al, 2014; Perez-Frances et al, 2021; Thorel et al, 2010; Zhao et al, 2021a). To test whether Ngn3⁺ non-beta cells could convert to beta cells after extreme beta-cell loss and to provide a positive technical control for our dual-genetic lineage tracing system, we utilized the R26-R-tdT-DTR mouse line (Wang et al, 2022) for genetic ablation of beta cells. Upon crossbreeding with Ins2-DreER, Dre/rox-mediated expression of diphtheria toxin receptor (DTR) in beta cells can lead to depletion of pre-existing beta cells after diphtheria toxin (DT) injection (Fig. 4A). We first examined the recombination efficiency of Ins2-DreER;R26-R-tdT-DTR by collecting the pancreases 2 weeks after Tam, with multiple DT injections before tissue harvest (Fig. 4B). Wholemount fluorescence revealed tdT signals in PBS-treated but not DT-treated pancreases (Fig. 4C). Consistently, all beta cells in the control group were labeled with tdT whereas virtually all beta cells in the DT-treated group were ablated (Fig. 4D,E), demonstrating high ablation efficiency in this model. Concurrently, there was a notable increase in the population of non-beta endocrine cells, including alpha, delta, and PP cells (Fig. EV4A,B).

Figure 4. Ngn3⁺Ins^– non-beta cells contribute to new Ins⁺ beta cells after extreme loss of beta cells.

(A) Schematic illustrating the labeling strategy by Ins2-DreER;R26-R-tdT-DTR mice. (B) Schematic showing the experimental strategy. (C) Wholemount fluorescence images of the pancreas from Ins2-DreER;R26-R-tdT-DTR mice after PBS (left) or DT (right) treatment. Arrowheads, pancreatic islets. (D) Immunostaining for tdT and Ins on pancreatic sections of Ins2-DreER;R26-R-tdT-DTR mice after PBS (left) or DT (right) treatment. Arrowheads, tdT⁺Ins⁺ pancreatic beta cells. (E) Immunostaining for tdT and Ins on serial sections of the pancreatic islets from Ins2-DreER;R26-R-tdT-DTR mice after PBS treatment. (F) Cartoon image showing the labeling strategy by Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice. (G) Schematic showing the experimental strategy. (H) Wholemount fluorescence images of the pancreas from Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice after PBS (top) or DT (bottom) treatment. Arrowheads, pancreatic islets. (I) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR after PBS (top) or DT (bottom) treatment. Arrowheads, tdT⁺Ins⁺ pancreatic beta cells. (J) Quantification of the beta cell numbers per islet sections of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice after PBS or DT treatment. Data are mean ± SD; PBS, n = 4 biological replicates; DT, n = 4 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (K) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice with PBS (left) or DT (right) treatment 4 weeks after Tam treatment. (L) Quantification of the percentages of ZsGreen⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice with PBS or DT treatment. Data are mean ± SD; PBS, n = 4 biological replicates; DT, n = 4 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (M) Immunostaining for ZsGreen, tdT, and Pdx1 and Nkx6.1 on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice with DT treatment 4 weeks after Tam treatment. (N) Cartoon image showing that Ngn3⁺Ins^– non-beta cells contribute to new Ins⁺ beta cells after genetic ablation of beta cells. Scale bars, 1 mm (yellow) and 100 μm (white). Each image is representative of 4–5 individual mouse samples. See also Fig. EV4. Source data are available online for this figure.

Figure EV4. Characterization of Ins2-DreER;Ngn3-CreER;NR1 and Ins2-DreER;Ngn3-2A-CreER;NR1 with PDL injury, related to Fig. 4.

(A) Immunostaining for tdT, Ins, Gcg, Sst, and PP on pancreatic sections of Ins2-DreER;R26-R-tdT-DTR with PBS or DT treatment. (B) Immunostaining for tdT, Gcg, Sst, PP and E-cad on pancreatic sections of Ins2-DreER;R26-R-tdT-DTR with PBS or DT treatment. (C) Schematic showing the experimental strategy of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR. (D) Immunostaining for tdT, zsGreen and CK19 on pancreatic sections of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice with PBS or DT treatment. (E) Quantification of zsGreen⁺ ductal cells per section of Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice with PBS or DT treatment. Data are mean ± SD; PBS, n = 4 biological replicates; DT, n = 4 biological replicates; P = 0.68; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t-tests were used for statistical comparisons and p < 0.05 was accepted as statistically significant. (F) Schematic showing the experimental strategy. (G) H.E. image of unligated head and ligated tail of the pancreas from Ins2-DreER;Ngn3-CreER;NR1 mice 2 weeks after pancreatic ductal ligation (PDL) injury. (H) Immunostaining for tdT, zsGreen and Insulin on sections of unligated head of the pancreas from Ins2-DreER;Ngn3-CreER;NR1 mice. (I) Immunostaining for tdT, zsGreen and Insulin on sections of ligated tail of the pancreas from Ins2-DreER;Ngn3-CreER;NR1 mice. (J) Quantification of the percentages of ZsGreen⁺ cells in Ins⁺ pancreatic beta cells of unligated head and ligated tail of the pancreas from Ins2-DreER;Ngn3-CreER;NR1 mice. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (K) Schematic showing the experimental strategy. (L) H.E. image of unligated head and ligated tail of the pancreas from Ins2-DreER;Ngn3-2A-CreER;NR1 mice 2 weeks after PDL injury. (M) Immunostaining for tdT, zsGreen and Insulin on sections of unligated head of the pancreas from Ins2-DreER;Ngn3-2A-CreER;NR1 mice. (N) Immunostaining for tdT, zsGreen and Insulin on sections of ligated tail of the pancreas from Ins2-DreER;Ngn3-2A-CreER;NR1 mice. (O) Quantification of the percentages of ZsGreen⁺ cells in Ins⁺ pancreatic beta cells of unligated head and ligated tail of the pancreas from Ins2-DreER;Ngn3-2A-CreER;NR1 mice. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. Scale bars, 100 μm. Each image is representative of 4–5 individual mouse samples.

Next, we crossed R26-R-tdT-DTR with Ins2-DreER;Ngn3-2A-CreER;NR1 (Fig. 4F). By design, pre-existing beta cells will be ablated after DT treatment, and if Ngn3⁺ non-beta cells (ZsGreen⁺) could convert to beta cells, the newly generated beta cells would be ZsGreen⁺. After Tam treatment and a washout period, we induced DT injection and collected pancreases from Ins2-DreER;Ngn3-2A-CreER;NR1;R26-R-tdT-DTR mice 2 weeks or 4 weeks post-Tam (Fig. 4G). At 2 weeks post-Tam, wholemount imaging showed tdT⁺ islets in the PBS-control group but not in the DT-treated group (Fig. 4H), which was confirmed by sectional analysis (Fig. 4I,J). At 4 weeks post-Tam, several Ins⁺ beta cells reappeared and 53.92 ± 9.41% of these newly generated beta cells were ZsGreen⁺ (Fig. 4K,L). Furthermore, these ZsGreen⁺ cells also expressed Pdx1 and Nkx6.1 (Fig. 4M). Meanwhile, we could hardly find Ins⁺ cells in the pancreatic duct at this time point. These data suggested that Ngn3⁺ non-beta cells could convert to beta cells after extreme beta-cell loss, providing a positive technical control for detecting beta-cell neogenesis by our dual-genetic lineage tracing system (Fig. 4M,N).

These Ngn3⁺ non-beta cells could likely be alpha cells, delta cells or PP cells that convert to beta cells under special conditions (Chera et al, 2014; Perez-Frances et al, 2021; Thorel et al, 2010; Zhao et al, 2021a). Our results showed that with DT treatment, there were 3.20 ± 1.41 zsGreen⁺ ductal cells per section, compared to 3.60 ± 1.23 zsGreen⁺ ductal cells per section in the PBS-treated control group (Fig. EV4C–E). Notably, we did not observe a significant increase in the number of Ngn3⁺ non-beta cells within the ducts following DT treatment (Fig. EV4E). These results strongly suggest that rare Ngn3⁺ non-beta cells within the ducts are highly unlikely to migrate and convert to islet beta cells, even under conditions of severe beta cell loss.

Previous studies have shown conflicting results regarding whether Ngn3⁺ progenitors contribute to the generation of new beta cells in the pancreatic ductal ligation (PDL) injury model (Van de Casteele et al, 2013; Xiao et al, 2013a; Xiao et al, 2013b; Xu et al, 2008). Compared to the extreme beta cell ablation injury, the PDL injury causes relatively gentler damage to the islets. To address this issue, we used two mouse models, Ins2-DreER;Ngn3-CreER;NR1 and Ins2-DreER;Ngn3-2A-CreER;NR1 (Fig. EV4F,K). We subjected these mice to PDL injury and collected samples for analysis 2 weeks later. Hematoxylin and eosin staining revealed that the pancreatic structure in the unligated head region remained largely normal, while there was a significant accumulation of ductal cells in the ligated tail region. Despite these morphological changes, we did not detect any ZsGreen⁺tdT^–Ins⁺ beta cells in either the pancreatic head or the tail under these conditions (Fig. EV4F–O). These findings suggest that in the context of PDL injury, Ngn3⁺ non-beta cells do not contribute to the generation of new beta cells, providing new insights into the mechanisms of beta cell neogenesis in response to different types of pancreatic injuries.

Ductal epithelial cells do not contribute to beta cell neogenesis during adult homeostasis

To improve lineage tracing of pancreatic ducts, we generated a new mouse knockin line Hnf1b-2A-CreER that could label ductal epithelial cells with exceptionally high efficiency. We crossed Hnf1b-2A-CreER with Ins2-DreER;R26-TLR mice (Fig. 5A). In this dual-genetic lineage tracing system, we can simultaneously label and trace Hnf1b⁺Ins^– non-beta cells (ZsGreen^–tdT⁺), Hnf1b⁺Ins⁺ beta cells (ZsGreen⁺tdT⁺), and Hnf1b^–Ins⁺ beta cells (ZsGreen⁺tdT^–) with distinct fluorescent markers (Fig. 5A).

Figure 5. Simultaneous tracing of Hnf1b⁺Ins^– non-beta cells, Hnf1b⁺Ins⁺ beta cells, and Hnf1b^–Ins⁺ beta cells by the triple light reporter.

(A) Schematic illustrating the labeling strategy by Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice. (B) Schematic showing the experimental strategy. (C) Wholemount fluorescence images of the pancreas from Ins2-DreER;Hnf1b-2A-CreER;R26-TLR 2 weeks (left) and 12 weeks (right) after Tam treatment. Arrowheads, pancreatic islets. (D) Immunostaining for ZsGreen, tdT, and CK19 on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice 2 weeks after Tam treatment. (E) Quantification of the percentage of tdT⁺CK19⁺ cells among CK19⁺ pancreatic ductal cells of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (F) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice 2 weeks after Tam treatment (G) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice 2 weeks after Tam treatment. Arrowheads, ZsGreen⁺tdT⁺Ins⁺ pancreatic beta cells. (H) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice 12 weeks after Tam treatment. (I) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice 12 weeks after Tam treatment. Arrowheads, ZsGreen⁺tdT⁺Ins⁺ pancreatic beta cells. (J) Quantification of the percentages of ZsGreen^–tdT⁺ and ZsGreen⁺tdT⁺ or ZsGreen⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR mice. Data are mean ± SD; +2 weeks, n = 5 biological replicates; +12 weeks, n = 5 biological replicates; ZsGreen⁺tdT⁺: P = 0.27; ZsGreen⁺: P = 0.57; n.s., non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. (K) Cartoon image showing that Ins2-DreER;Hnf1b-2A-CreER;R26-TLR label Hnf1b⁺Ins^– non-beta cells, Hnf1b⁺Ins⁺ beta cells, and Hnf1b^–Ins⁺ beta cells in pancreas simultaneously. Hnf1b⁺Ins^– non-beta cells do not contribute to beta cells was detected during homeostasis. Scale bars, 1 mm (yellow) and 100 μm (white). Each image is representative of five individual mouse samples. See also Fig. EV5. Source data are available online for this figure.

We treated Ins2-DreER;Hnf1b-2A-CreER;R26-TLR adult mice with Tam and analyzed them after 2 and 12 weeks (Fig. 5B). We performed wholemount fluorescence imaging and found that ZsGreen⁺ islets expressed strong tdT after Tam induction (Fig. 5C). Quantitively, 99.76 ± 0.26% of CK19⁺ ductal cells were positive for tdT (Fig. 5D,E). We also found that almost all Ins⁺ cells lining the ductal epithelium were ZsGreen⁺tdT⁺ whereas Sst⁺ cells in the duct were ZsGreen^–tdT⁺ (Fig. 5F). In the pancreatic islets, while vast majority of beta cells (99.94 ± 0.11%) were labeled with ZsGreen, a portion of the beta cells were also labeled with tdT, but we never detected tdT⁺ZsGreen^– beta cells (0.00 ± 0.00%) (Fig. 5G). After the chase period, all beta cells in ducts and islets were either ZsGreen⁺tdT^– or ZsGreen⁺tdT⁺, and there was no tdT⁺ZsGreen^– beta cells (0.00 ± 0.00%) (Fig. 5H–J). Moreover, the proportion of ZsGreen⁺tdT⁺ beta cells was not significantly different before and after the chase period (8.32 ± 0.73% for 2 weeks and 9.09 ± 1.25% for 12 weeks, Fig. 5J), indicating that the Hnf1b⁺Ins⁺ beta cell subpopulation does not harbor a proliferative advantage over Hnf1b^–Ins⁺ beta cells. As a technical control, we rarely detect ZsGreen⁺ or tdT⁺ cells without Tam treatment (Fig. EV5A–F). The above data demonstrated that pancreatic ductal cells did not adopt beta cell fate in the 10-week chase period in the adult pancreas (Fig. 5K).

Figure EV5. Characterization of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR and Ins2-DreER;Hnf1b-2A-CreER;NR1 without tamoxifen treatment, related to Figs. 5 and 6.

(A) Schematic showing the experimental strategy of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR with oil treatment. (B) Whole-mount fluorescent images of pancreas from Ins2-DreER;Hnf1b-2A-CreER;R26-TLR after oil treatment. (C) Immunostaining for ZsGreen, tdT and CK19 on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR after oil treatment. (D) Quantification of the percentage of tdT⁺ cells in CK19⁺ pancreatic ductal cells of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (E) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR after oil treatment. (F) Quantification of the percentage of ZsGreen⁺tdT^– or tdT⁺ZsGreen^– or ZsGreen⁺tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Hnf1b-2A-CreER;R26-TLR after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (G) Schematic showing the experimental strategy of Ins2-DreER;Hnf1b-2A-CreER;NR1 with oil treatment. (H) Whole-mount fluorescent images of pancreas from Ins2-DreER;Hnf1b-2A-CreER;NR1 after oil treatment. (I) Immunostaining for ZsGreen, tdT and CK19 on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-NR1 after oil treatment. (J) Quantification of the percentage of ZsGreen⁺ cells in CK19⁺ pancreatic ductal cells of Ins2-DreER;Hnf1b-2A-CreER;R26-NR1 after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. (K) Immunostaining for ZsGreen, tdT and Ins on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;R26-NR1 after oil treatment. (L) Quantification of the percentage of ZsGreen⁺ or tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Hnf1b-2A-CreER;R26-NR1 after oil treatment. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from 10 pancreas sections were quantified. Scale bars, yellow, 1 mm; white, 100 μm. Each image is representative of 5 individual samples.

We next crossed Hnf1b-2A-CreER with the Ins2-DreER;NR1 mouse line in which all beta cells were labeled with tdT whereas Hnf1b⁺ non-beta cells, including ductal epithelial cells, were labeled with ZsGreen (Fig. 6A). We treated Ins2-DreER;Hnf1b-2A-CreER;NR1 with Tam and collected the pancreases 2 weeks and 12 weeks later (Fig. 6B). Wholemount imaging revealed tdT⁺ islets (Fig. 6C). Virtually all ductal cells (99.61 ± 0.35%) were labeled with the ZsGreen reporter (Fig. 6D,E). Moreover, we detected a few tdT⁺ZsGreen^– cells in the ductal epithelium that were stained positive for insulin (Fig. 6F). The Sst⁺ ductal cells were labeled with ZsGreen (Fig. 6F). Examination of pancreatic sections from Ins2-DreER;Hnf1b-2A-CreER;NR1 mice revealed that none of the ZsGreen⁺tdT^– cells expressed Ins in the islets (0.00 ± 0.00%) (Fig. 6G). After 12 weeks of Tam treatment, we analyzed the pancreases and found no ZsGreen⁺tdT^– beta cells in either ducts or islets (0.00 ± 0.00%) (Fig. 6H–J). As a control, very few fluorescent cells were detected in the Ins2-DreER;Hnf1b-2A-CreER;NR1 pancreas without Tam treatment (Fig. EV5G–L), suggesting negligible leakage in this system. Taken together, these results indicated that pancreatic ductal cells did not contribute to new beta cells in adult homeostasis (Fig. 6K).

Figure 6. Nested strategy shows Hnf1b⁺Ins^– non-beta cells do not contribute to Ins⁺ beta cells during adult homeostasis.

(A) Schematic of the labeling strategy by Ins2-DreER;Hnf1b-2A-CreER;NR1 mice. (B) Schematic showing the experimental strategy. (C) Wholemount fluorescence images of the pancreas from Ins2-DreER;Hnf1b-2A-CreER;NR1 2 weeks and 12 weeks after Tam treatment. Arrowheads, pancreatic islets. (D) Immunostaining for ZsGreen, tdT, and CK19 on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;NR1 mice 2 weeks after Tam treatment. (E) Quantification of the percentage of ZsGreen⁺CK19⁺ cells in CK19⁺ pancreatic ductal cells of Ins2-DreER;Hnf1b-2A-CreER;NR1 mice. Data are mean ± SD; n = 5 biological replicates. In each sample, islets from ten pancreas sections were quantified. (F) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;NR1 mice 2 weeks after Tam treatment. (G) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;NR1 mice 2 weeks after Tam treatment. Arrowheads, tdT⁺Ins⁺ pancreatic beta cells. (H) Immunostaining for ZsGreen, tdT, and CK19 or Ins or Sst on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;NR1 mice 12 weeks after Tam treatment. (I) Immunostaining for ZsGreen, tdT, and Ins on pancreatic sections of Ins2-DreER;Hnf1b-2A-CreER;NR1 mice 12 weeks after Tam treatment. Arrowheads, tdT⁺Ins⁺ pancreatic beta cells. (J) Quantification of the percentage of ZsGreen⁺ or tdT⁺ cells in Ins⁺ pancreatic beta cells of Ins2-DreER;Hnf1b-2A-CreER;NR1 mice. Data are mean ± SD; +2 weeks, n = 5 biological replicates; +12 weeks, n = 5 biological replicates; tdT⁺: P = 0.21; n.s. non-significant. In each sample, islets from 10 pancreas sections were quantified. Two-tailed unpaired Student’s t tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. (K) Cartoon image showing that Ins2-DreER;Hnf1b-2A-CreER;NR1 label Hnf1b⁺Ins^– non-beta cells and Ins⁺ beta cells in pancreas simultaneously. Hnf1b⁺Ins^– non-beta cells do not generate new beta cells in adult homeostasis. Scale bars, 1 mm (yellow) and 100 μm (white). Each image is representative of five individual mouse samples. See also Fig. EV5. Source data are available online for this figure.

Discussion

Over the last two decades, the contribution of Ngn3⁺ progenitors and pancreatic ductal cells to new beta cells in adults has remained highly controversial and requires further investigation to clarify this issue. In this study, we employed a newly developed dual-genetic tracing strategy to address the question of whether Ngn3⁺ progenitors or other potential progenitors existing in the pancreatic ductal epithelium could contribute to beta cell neogenesis during murine adult homeostasis. Using independent lineage tracing strategies, we provided direct in vivo genetic evidence that Ngn3⁺ non-beta cells or Hnf1b⁺ non-beta cells (including virtually all ductal cells) do not contribute to de novo beta cells during adult homeostasis. Our study also demonstrated that Ngn3⁺ progenitors could generate new beta cells under the condition of extreme ablation of pre-existing beta cells. This transformation serves as a positive technical control for the detection of beta cell neogenesis by our dual-genetic lineage tracing system.

Precise genetic lineage tracing study is essential for elucidating specific cell behaviors and fate conversions (He et al, 2017; Kretzschmar and Watt, 2012; Weng et al, 2022). Previous lineage tracing studies have been performed using conventional Cre/loxP recombination system, with many controversies for either supporting or opposing the adult stem cells in tissue repair and regeneration (He et al, 2017; Weng et al, 2022). However, the Cre recombinase expression in non-specific cells may lead to misinterpretation of lineage tracing data. For example, supportive evidence for cKit⁺ cardiac stem cells should be based on the fact that Kit is not expressed in pre-existing cardiomyocytes, but subsequent studies have shown that Kit is expressed in cardiomyocytes, and unintentional labeling of Kit-CreER confounds this interpretation (He et al, 2019, 2017). Pancreatic beta cells are one of the most highly differentiated cells in adults and one of the most important features of the putative beta-cell progenitors is that they do not express insulin themselves. Despite differences in mouse lines, the expression of Ngn3 and Hnf1b in pre-existing beta cells is widely recognized (Magenheim et al, 2023; Solar et al, 2009; Wang et al, 2009), and single-cell-RNA sequencing analyses have also revealed the presence of Ngn3⁺Ins⁺ beta cell populations in human beta cells (Li et al, 2020). This is likely to be one of the factors that have given rise to the debate as to whether Ngn3⁺ progenitors (insulin negative) and ductal cells contribute to beta-cell neogenesis.

Here, we generated new knock-in mouse lines Ngn3-2A-CreER and Hnf1b-2A-CreER, which are efficient for labeling and do not affect endogenous gene expression compared to previous lines. We found that Ngn3-2A-CreER and Hnf1b-2A-CreER also label a fraction of beta cells at the pulse time point. Taking advantage of dual-genetic tracing technology, we could distinguish the “ectopic” labeling introduced by Cre recombinase expression in pre-existing beta cells. Using intersectional and nested reporter designs, respectively, we demonstrated that Dre/rox-mediated recombination effectively blocked the “unwanted” Cre reporter expression in beta cells, thereby facilitating the clear interpretation of lineage tracing data. We addressed the long-standing debate on the contribution of Ngn3⁺ progenitors and pancreatic ductal cells to the beta cells in the adult pancreas. Our dual-genetic lineage tracing data demonstrated that Ngn3⁺ non-beta cells and pancreatic ductal cells do not generate new beta cells de novo during homeostasis in murine adults.

The dual-genetic lineage tracing strategy not only controls Cre/loxP recombination in “unwanted” cells more precisely but also sensitively captures lineage conversions in vivo. By using the genetic ablation approach, we could also detect the in vivo conversion of Ngn3⁺ non-beta cells to beta cells under certain conditions. However, a limitation of our approach is that it is unable to distinguish between Ngn3⁺ islet cells and Ngn3⁺ ductal cells as the exact source for newly generated insulin-expressing cells. This is because Ngn3-2A-CreER labels subpopulations of alpha, delta, and PP cells in the islets, as well as Ngn3-expressing cells located in the ducts, all of which are labeled by zsGreen. Based on previous studies, we postulated that these Ngn3⁺ non-beta cells responsible for the generation of Ins⁺ cells are predominantly subpopulations of endocrine cells within the pancreatic islets (Chera et al, 2014; Perez-Frances et al, 2021; Thorel et al, 2010). Our model provides a rigorous technical standard and an unprecedented strategy for revisiting the contribution of putative pancreatic stem cells or progenitors in beta cell renewal and regeneration. Our study does not examine the potential of Ngn3⁺ ductal progenitor cells for beta-cell neogenesis under various injury models or some extreme conditions. However, our study highlights the urgent need for further research in this debated area. Our findings convincingly demonstrate that during homeostasis, Ngn3⁺ non-beta cells, which include the potential Ngn3⁺ ductal or islet progenitors, do not contribute to beta-cell neogenesis in the adult pancreas. Currently, it remains unclear which type of injury or extreme condition may unveil the regenerative capacity of rare Ngn3⁺ duct cells to contribute to beta-cell regeneration. In the case of the Hnf1b-2A-CreER line, although it labeled ductal cells with high-efficiency, it also labeled a subset of islet endocrine cells. Therefore, using the dual-genetic tracing strategy based on the intersection of Ngn3 and Hnf1b (for example, Ngn3-2A-CreER;Hnf1b-2A-DreER;R26-RSR-LSL-tdT) cannot specifically target the potential Ngn3⁺ ductal progenitors. There is no available genetic tool that can definitively resolve whether rare Ngn3⁺ duct cells contribute to beta cell neogenesis after severe beta cell loss. However, we did not observe a significant increase in the number of Ngn3⁺ non-beta cells within the ducts following DT treatment. These results strongly suggest that these rare Ngn3⁺ non-beta cells within the ducts are highly unlikely to migrate and convert to islet beta cells, even under conditions of severe beta cell loss. This finding provides quantitative support that these Ngn3⁺ non-beta cells within the ducts are unlikely to play a meaningful biological role in islet beta cell neogenesis. Future studies should also focus on the as-yet-unrevealed mechanism driving beta cell regeneration in the adult pancreas, including the promotion of replication of pre-existing beta cells or forced expression of exogenous reprogramming factors for beta cell regeneration.

Methods

Reagents and Tools Table

Reagent/resource	Reference or source	Identifier or catalog number
Experimental models
Ngn3-CreER	Gu et al, 2002	N/A
R26-tdT	Madisen et al, 2010	N/A
Ins2-DreER	Zhao et al, 2021a	N/A
R26-TLR	Liu et al, 2020	N/A
NR1	He et al, 2017	N/A
R26-R-tdT-DTR	Wang et al, 2022	N/A
Ngn3-2A-CreER	This study	N/A
Hnf1b-2A-CreER	This study	N/A
Recombinant DNA
N/A
Antibodies
Goat anti-tdTomato	Rockland	Cat# 200-101-379; RRID: AB_2744552
Rabbit anti-tdTomato	Rockland	Cat# 600-401-379; RRID: AB_2209751
Rat anti-tdTomato	Proteintech (ChromoTek)	Cat# 5F8; RRID: AB_2336064
Rabbit anti-ZsGreen	Clontech	Cat# 632474; RRID: AB_2491179
Guinea pig-Insulin	Dako Agilent	Cat# A0564; RRID: AB_10013624
Rabbit anti-Insulin	Abcam	Cat# ab63820; RRID: AB_1925116
Rat anti-CK19	Developmental Studies Hybridoma Bank	Cat# TROMA-III; RRID: AB_2133570
Goat anti-E-cad	R & D systems	Cat# AF748; RRID: AB_355568
Rabbit anti-Glucagon	Sigma-Aldrich	Cat# SAB4501137; RRID: AB_10761583
Rat anti-Somatostatin	Santa Cruz	Cat# sc-47706; RRID: AB_628268
Rabbit anti-Somatostatin	Abcam	Cat# ab111912; RRID: AB_10903864
Goat anti-Pancreatic Polypeptide	Abcam	Cat# ab77192; RRID: AB_1524152
Donkey anti-rabbit 488	Invitrogen	Cat# A21206; RRID: AB_2535792
Donkey anti-rabbit 555	Invitrogen	Cat# A31572; RRID: AB_162543
Donkey anti-rabbit 647	Invitrogen	Cat# A31573; RRID: AB_2536183
Donkey anti-rat 594	Jackson ImmunoResearch Inc	Cat# 712-585-153; RRID: AB_2340689
Donkey anti-rat 647	Abcam	Cat# ab150155; RRID: AB_2813835
Donkey anti-goat 555	Invitrogen	Cat# A21432; RRID: AB_2535853
Donkey anti-goat 647	Invitrogen	Cat# A21447; RRID: AB_141844
JIR Cy5 Donkey a-Guinea Pig IgG(H + L)	Jackson ImmunoResearch Inc	Cat# 706-175-148; RRID: AB_2340462
PCR primers - Ngn3-2A-CreER genotyping - mut	This study	5’-TACTCCCCAGTCTCCCAAGC-3’; 5’-CGCGCGCCTGAAGATATAGA-3’
PCR primers - Ngn3-2A-CreER genotyping - wt	This study	5’-ACAAAGATCGAGACCCTGCG-3’; 5’-CCTGTGAGAGGGCTAGGGAT-3’
PCR primers - Hnf1b-2A-CreER genotyping - mut	This study	5’-CCTGGCCATTCCATCCAAGT-3’; 5’-GTTGCATCGACCGGTAATGC-3’
PCR primers - Hnf1b-2A-CreER genotyping - wt	This study	5’-AAGTGCCCACCGTATTCCAG-3’; 5’-ACTTGGCATCTTGGGAGAGC-3’
Chemicals, enzymes and other reagents
Tamoxifen	Sigma-Aldrich	Cat# T5648
Diphtheria toxin	Sigma-Aldrich	Cat# D0564
Paraformaldehyde (PFA)	Sigma-Aldrich	Cat# P6148
O.C.T.	Sakura	Cat# 4583
Normal donkey serum	Jackson ImmunoResearch Inc	Cat# 017-000-121
Triton X-100	Sigma-Aldrich	Cat# X100
PBS	Invitrogen	Cat# C10010500BT
DAPI	Invitrogen	Cat# D21490
UitraPure Distilled Water	Invitrogen	Cat# 10977023
Software
Fiji	https://ImageJ.net/software/fiji/	N/A
GraphPad Prism 8 software	https://www.graphpad.com	N/A
PhotoLine	https://www.pl32.com/	N/A
Other
N/A

Mice

All animal experiments were performed under the guidelines of the Institutional Animal Care and Use Committee of the State Key Laboratory of Cell Biology, Center for Excellence in Molecular Cell Science, University of Chinese Academy of Science, Chinese Academy of Science. All mice used in this study were maintained on a C57BL6/ICR mixed background. Both male and female mice were included in the study, and the starting time point (Day0) involved adult mice aged 8-10 weeks. The Ngn3-CreER (Gu et al, 2002), R26-tdT (Madisen et al, 2010), Ins2-DreER (Zhao et al, 2021a), R26-TLR (Liu et al, 2020), NR1 (He et al, 2017), and R26-R-tdT-DTR (Wang et al, 2022) mouse lines have been reported previously. The Ngn3-2A-CreER mouse line was generated by inserting the 2A-CreER cassette after the endogenous Ngn3 gene via homologous recombination. The Hnf1b-2A-CreER mouse line was generated by inserting the 2A-CreER cassette after the endogenous Hnf1b gene via homologous recombination. Both Ngn3-2A-CreER and Hnf1b-2A-CreER mice were generated by the Shanghai Model Organisms Center, Inc. Tam (Sigma, T5648) was dissolved in corn oil and was introduced by gavage (0.2 mg/g) at the indicated time points. DT (Sigma, D0564) was injected intraperitoneally (10 ng/g) at the indicated time points.

Genomic PCR

Mouse tails or tissues were lysed in lysis buffer consisting of 100 mM Tris-HCl (adjusted to pH 7.8), 0.2% SDS, 5 mM EDTA, 200 mM NaCl, and 100 μg/ml Proteinase K at 55 °C for more than 6 h. After lysis, the mixture was centrifuged at maximum speed for 8 min, and the supernatant was separated and collected. The supernatant was then precipitated with isopropanol and centrifuged again at maximum speed for 5 min to obtain genomic DNA, which was then washed with 70% ethanol and dissolved in deionized water. To determine the genotype of all mice, specific primers adapted to the knock-in sequence were used to distinguish from the wild-type allele.

Wholemount fluorescence imaging

Wholemount fluorescence imaging was performed as previously described (He et al, 2017). Pancreatic tissues were washed three times in phosphate-buffered saline (PBS) and then fixed in 4% paraformaldehyde (PFA) at 4 °C for 1 h and washed three times with PBS. Then the tissues were placed in PBS, and wholemount brightfield and fluorescence imaging were taken by a Zeiss stereo microscope (Zeiss Axio Zoom. V16).

Immunostaining

The immunostaining protocol was performed as described previously (He et al, 2017). Briefly, pancreatic tissues were fixed in 4% PFA at 4 °C for 1 h and then washed with PBS. Then the tissues were dehydrated in 30% sucrose at 4 °C overnight and embedded in the optimum cutting temperature (O.C.T., Sakura). 10 μm cryosections of each block were collected on slides. For immunostaining, slides were air-dried at room temperature and washed with PBS. Next, the slides were blocked with 5% PBSST (0.1% Triton X-100 and 5% normal donkey serum in PBS) blocking buffer at room temperature for 30 min and then incubated with primary antibodies at 4 °C overnight. The primary antibodies used were as follows: tdT (Rockland, 200-101-379, 1:1000), tdT (Rockland, 600-401-379, 1:1000), tdT (ChromoTek, 5F8, 1:200), ZsGreen (Clontech, 632474, 1:1000), Insulin (Dako, A0564, 1:500), Insulin (Abcam, ab63820, 1:500), CK19 (Developmental Studies Hybridoma Bank, TROMA-III, 1:500), E-cad (R&D, AF748, 1:500), Glucagon (Sigma, SAB4501137, 1:500), Somatostatin (Santa Cruz, sc-47706, 1:500), Somatostatin (Abcam, ab111912, 1:500), Pancreatic Polypeptide (Abcam, ab77192, 1:200), Pdx1 (Abcam, ab47267, 1:500), and Nkx6.1 (Abcam, ab221549, 1:500). The slides were washed three times with PBS to remove the primary antibodies in the next day, and incubated with Fluor-conjugated secondary antibodies at room temperature for 30 min. After three washes with PBS, the slides were mounted with a mounting medium containing the nuclear stain DAPI. Immunostaining images were acquired with Olympus confocal microscopes (FV1200, FV3000 and FV4000) and a Zeiss confocal microscope (Zeiss LSM880), and analyzed with ImageJ (NIH).

Pancreatic ductal ligation

Pancreatic ductal ligation (PDL) injury was performed as previously described (Zhao et al, 2021a). In brief, adult mice were first anesthetized in a sealed chamber using 2% isoflurane gas. Once anesthetized, the mice were immediately transferred to a pre-warmed heat pad to ensure the maintenance of their body temperature. Anesthesia was continuously sustained through the inhalation of isoflurane throughout the procedure. Subsequently, the abdominal fur of the mice was carefully removed. A midline incision was then made through the abdominal skin and underlying muscles. The pancreatic head and the duodenum were gently lifted from the retroperitoneum to expose the relevant anatomical structures. The pancreatic duct, located to the left side of the portal vein, was carefully ligated, which demarcates the splenic and gastro-duodenal lobes of the pancreas. After surgery, each mouse was placed in an individual cage and provided with a standard laboratory chow diet and tap water for postoperative recovery.

H.E. staining

H.E. staining was performed according to a previously described method (Zhao et al, 2021a). First, 10 μm thick cryosections were prepared. These sections were then immersed in PBS for 15 min to wash them. After that, the sections were transferred to hematoxylin A solution and incubated for 3 min. Following the hematoxylin staining, the sections were rinsed three times under running tap water. Subsequently, they were immersed in a solution of 1% concentrated hydrochloric acid diluted in 70% ethanol for 1 min and then washed three times with water. Then the sections were incubated in 1% ammonia water for 1 min and washed three times with water. Next, the sections were stained with Eosin-Y solution for 8–10 sections. After eosin staining, the sections were dehydrated by passing them through a series of ethanol and xylene solutions. Finally, the stained sections were mounted with a resinous medium. All images of the stained sections were captured using an Olympus microscope (Olympus, DP72).

Quantification

All data were presented as the mean values ± standard deviations (SD). All mice were randomly assigned to different groups. The “n” in the figure legends represented the number of biological replicates used in each experiment. The manual counting was used for cell quantification and the raw cell count numbers per section and per individual mouse used in the quantifications were presented in the Appendix Table (Appendix Tables S1–11). Two-tailed unpaired Student’s t tests were used for statistical comparisons and P < 0.05 was accepted as statistically significant. No statistical methods were used to estimate sample size. No blinding was done. No sample were excluded from the analysis.

Author contributions

Xiuzhen Huang: Data curation; Validation. Huan Zhao: Data curation; Supervision; Funding acquisition; Validation; Writing—original draft. Hui Chen: Data curation; Validation; Methodology. Zixin Liu: Data curation; Formal analysis. Kuo Liu: Data curation. Zan Lv: Data curation. Xiuxiu Liu: Data curation. Ximeng Han: Data curation. Maoying Han: Data curation. Jie Lu: Writing—review and editing. Qiao Zhou: Writing—review and editing. Bin Zhou: Conceptualization; Resources; Supervision; Funding acquisition; Writing—review and editing.

Source data underlying figure panels in this paper may have individual authorship assigned. Where available, figure panel/source data authorship is listed in the following database record: biostudies:S-SCDT-10_1038-S44318-025-00434-z.

Data availability

This study includes no data deposited in external repositories.

The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44318-025-00434-z.

Disclosure and competing interests statement

The authors declare no competing interests.

Supplementary information

Expanded view data, supplementary information, appendices are available for this paper at 10.1038/s44318-025-00434-z.