Genetic recording of pancreatic beta cell proliferation

Significance

Pancreatic beta cell proliferation is essential for maintaining metabolic homeostasis, and its dysregulation is a key factor in the development of diabetes. Current methods cannot trace beta cell proliferation seamlessly over extended periods in vivo. Here, we developed beta cell ProTracer, a dual recombinase-based genetic system that enables the continuous, long-term recording of proliferated beta cells. This provides an important tool for evaluating beta cell proliferation during homeostasis, injury, and regeneration and will facilitate the validation of therapeutic strategies aimed at increasing beta cell mass.

Abstract

Pancreatic beta cell generation and proliferation primarily rely on the self renewal of preexisting beta cells, making precise quantification of beta cell proliferation essential for understanding pancreatic homeostasis and pathogenesis. However, previous methods had limitations and were unable to trace beta cell proliferation in vivo over extended periods. Here, we developed beta cell ProTracer, a genetic system that uses dual recombinase technology to enable continuous and cell type-specific recording of beta cell proliferation in mice. The beta cell ProTracer system allows for the quantification of the proliferation rate of adult pancreatic beta cells during homeostasis and in response to injury or drug treatment. Clonal analysis of the proliferated beta cell population reveals a uniform proliferative capacity among these cells. Our findings enhance the understanding of beta cell proliferation dynamics in homeostasis, repair, and regeneration.

Diabetes, characterized by hyperglycemia, is a complex metabolic disorder primarily caused by dysfunction of insulin-producing pancreatic beta cells, including type 1 and type 2 diabetes (1). Diabetes can lead to a variety of complications, including blindness, kidney failure, stroke, and coronary vascular diseases, which not only significantly reduces the quality of life of patients but also imposes a substantial economic and social burden (1). Identifying the sources of new beta cells is key to promoting pancreatic beta cell regeneration and diabetes treatment strategies. Previous studies, using lineage tracing techniques and isotope incorporation assays, have demonstrated that pancreatic beta cells mainly derived from self replication rather than stem cell differentiation in adults (2–8). Despite these significant findings, an important question remains: How can we enhance the proliferative ability of pancreatic beta cells to achieve in situ functional beta cell regeneration (9)?

Over the past few decades, numerous studies have investigated the potential factors that can promote pancreatic beta cell proliferation (10–23). Growth factors, such as platelet-derived growth factor (PDGF), have been demonstrated to stimulate beta cell division (16). Through high-throughput compound screening, researchers found that harmine, a chemical inhibitor of the dual-specificity tyrosine-regulated kinase 1a (DYRK1A), could effectively promote pancreatic beta cell proliferation in murine models and even in human beta cells cultured in vitro (17), which provides a promising direction for the development of drugs targeting pancreatic beta cell proliferation. Subsequent studies showed that the combined inhibition strategy could have a synergistic effect on pancreatic beta cell regeneration. Specifically, the coinhibition of DYRK1A and glycogen synthase kinase 3b (GSK3B) could induce the proliferation of pancreatic beta cells (24). Additionally, simultaneously inhibiting DYRK1A, the SMAD pathway, and the trithorax pathways could work in synergy to promote the replication of pancreatic beta cells (14). Furthermore, the combined use of a DYRK1A inhibitor and a glucagon-like peptide 1 (GLP-1) receptor agonist has a synergistic impact on promoting the regeneration of pancreatic beta cells (25). The knockout of the insulin inhibitory receptors (Inceptor) in pancreatic beta cells or blocking them with a monoclonal antibody can increase the number of functional beta cells (26). Overall, these studies have provided new insights into the regulation of pancreatic beta cell proliferation and identified potential therapeutic targets for diabetes treatment.

However, achieving precise and long-term monitoring of beta cell proliferation in the adult pancreas still poses technical challenges. The relatively low proliferation rate of beta cells in the adult pancreas makes it difficult to study their proliferation dynamics over time. Conventional techniques for detecting cell proliferation, such as immunostaining using cell proliferation markers like Ki67 or pHH3, have inherent limitations (27, 28). Ki67 immunostaining or Ki67-CreER genetic tracing in mice merely provides a static snapshot of cell proliferation at a single time point, being incapable of capturing the continuous process of pancreatic beta cell proliferation over an extended period (27–30). Furthermore, the question of whether there exists a subpopulation of pancreatic beta cells with distinct proliferative capabilities remains unresolved. Some studies suggested that pancreatic beta cells are homogeneous in terms of their proliferation potential (4, 31), whereas others indicated the presence of heterogeneity in beta cell generation (32–34). Unraveling the beta cell proliferation dynamics is fundamental to understanding the basic mechanisms underlying pancreatic beta cell proliferation and to the development of targeted therapies for diabetes.

In this context, the development of a reliable and sensitive method to trace pancreatic beta cell proliferation over time holds paramount significance. Such a tool could provide detailed quantitative information about the proliferation rate of pancreatic beta cells during normal physiological conditions and in response to injury or drug treatments. In this study, we developed a genetic system named beta cell ProTracer to provide such information and offer insights into the dynamics of pancreatic beta cell generation and proliferation.

Results

ProTracer System Enables Monitoring of Pancreatic Endocrine Cell Proliferation.

Ki67, a well-established cell proliferation marker (29), has been widely used in previous studies. The Ki67-CreER mouse line, in particular, has been employed to label proliferating cells(29, 30). However, the traditional Ki67-CreER system has limitations. Tamoxifen (Tam), which is used to induce the system, has a halflife of approximately 12 to 24 h in mice (35). Prolonged Tam treatment over several wk or mo is not only technically challenging but also potentially detrimental, as it may cause adverse effects on the cells and the overall physiological state of the animals (8, 36–38). To overcome these issues, we have generated Ki67-CrexER mouse line and developed a genetic recording system named ProTracer (R26-DreER;Ki67-CrexER;R26-RSR-LSL-tdT) (27, 28, 39). Specifically, the Ki67-CrexER mouse line was generated by knocking the CrexER (Cre-rox-ER-rox) cassette into the endogenous Ki67 locus, precisely inserted after the last exon (removal of translational stop codon) and before the 3’untranslated region (3’UTR). The working principle of the ProTracer system is based on the dual recombinase-mediated genetic lineage tracing technology using Cre+Dre systems. After Tam induction, DreER enters the nucleus and excises the rox-flanked estrogen receptor (ER) in the CrexER allele. This excision event triggers a transformation at the DNA level, converting the inducible CrexER into its constitutively active form, Cre. Subsequently, Dre and Cre work on the R26-RSR-LSL-tdT reporter allele, thereby activating the ProTracer system during cell proliferation. Theoretically, a single pulse of Tam is sufficient to enable this system to continuously and seamlessly capture and label all proliferating cells in vivo throughout the animal’s lifespan, regardless of the time window of observation (Fig. 1 A and B).

Fig. 1.

Genetic tracing of pancreatic endocrine cell proliferation during homeostasis. (A) Schematic of the Ki67 expression pattern and the fate mapping strategy. (B) Schematic of the genetic tracing strategy for seamless proliferation recording using R26-DreER;Ki67-CrexER;R26-RSR-LSL-tdT (ProTracer). (C) Experimental timeline for ProTracer-mediated lineage tracing. (D) Immunostaining for tdTomato (tdT) and Insulin (Ins) on pancreatic sections from ProTracer mice. (E–G) Immunostaining for tdT and glucagon (Gcg, E), somatostatin (Sst, F), or pancreatic polypeptide (Ppy, G) on pancreatic sections from ProTracer mice. (H) Quantification of the percentage of tdT⁺ cells within each indicated endocrine cell population. Data are mean ± SD, n = 5. (I) Cartoon image showing the ProTracer-based labeling of proliferating pancreatic endocrine cells. (Scale bar, 100 μm.) Each image is representative of five individual samples.

To evaluate the performance of the ProTracer system in the pancreas, we treated adult ProTracer mice (aged 7 to 10 wk) with Tam and collected pancreatic tissues at 4, 8, and 12 wk afterward (Fig. 1C). By immunostaining the pancreatic tissue sections with the beta cell marker insulin (Ins), we observed a time-dependent increase in the number of tdT⁺Ins⁺ beta cells (Fig. 1D), while rare tdT⁺ cells were detected in the corn oil-treated group (SI Appendix, Fig. S1 A and B). Quantitatively, the percentage of Ins⁺ beta cells expressing tdT was 11.45 ± 1.56%, 20.92 ± 2.77%, and 28.61 ± 1.99% at 4, 8, and 12 wk after Tam induction, respectively (Fig. 1H). We further extended our analysis to other endocrine cell lineages. By immunostaining for markers specific to alpha (Gcg⁺), delta (Sst⁺), and PP (Ppy⁺) cells, we measured their self renewal rates. The percentage of Gcg⁺ alpha cells expressing tdT was 7.29 ± 1.06%, 13.82 ± 1.52%, and 23.76 ± 3.47% at 4, 8, and 12 wk after Tam induction (Fig. 1 E and H). Similarly, the percentage of Sst⁺ delta cells expressing tdT was 6.99 ± 1.33%, 14.32 ± 2.38%, and 25.70 ± 2.73% at these time points (Fig. 1 F and H). For Ppy⁺ PP cells, the percentage of tdT expressing cells was 7.52 ± 0.95%, 13.81 ± 2.24%, and 25.95 ± 1.58% at 4, 8, and 12 wk after Tam induction (Fig. 1 G and H). To ensure the specificity and reliability of the ProTracer system, we included some technical controls. We treated R26-DreER;R26-RSR-LSL-tdT and Ki67-CrexER;R26-RSR-LSL-tdT mice with Tam and could hardly find any tdT⁺ cells (SI Appendix, Fig. S1 C–F).

We performed two sets of experiments to quantify the excision efficiency of the R26-DreER driver on the Ki67-CrexER and R26-RSR-tdT alleles, respectively. First, genomic DNA from hand-picked islets of tamoxifen-treated R26-DreER;Ki67-CrexER mice was analyzed by qPCR using primers spanning the CrexER cassette. The fraction of the unexcised cassette was 51.61 ± 11.65% relative to vehicle-treated controls (SI Appendix, Fig. S1 G–K), corresponding to a 48.39% Dre-mediated excision efficiency at the Ki67 locus. Second, immunofluorescence staining of pancreatic sections from tamoxifen-treated R26-DreER;R26-RSR-tdT mice showed that 92.57 ± 3.97% of beta cells were tdT⁺ (SI Appendix, Fig. S1 L–O), indicating high-efficiency recombination at the Rosa26 reporter locus. Therefore, the combined effective labeling efficiency of the universal ProTracer system was therefore estimated to be approximately 45% (the product of the two sequential recombination efficiencies) in the pancreatic islets.

Beta Cell ProTracer System Monitors Pancreatic Beta Cell Proliferation.

The ProTracer system can capture in vivo cell proliferation over a long period, but it fails to achieve cell type specificity (Fig. 1I). Therefore, to specifically and precisely monitor beta cell proliferation in a cell type–specific manner, we developed the beta cell ProTracer system. This system is composed of three mouse lines: Ins2-DreER, Ki67-CrexER, and R26-RSR-LSL-tdT (Fig. 2A). The Ins2-DreER line is engineered to efficiently recombine within beta cells following Tam treatment (6). Our hypothesis was that DreER-rox recombination would specifically prime the Ki67-Cre genotype in insulin-expressing beta cells. This would result in the constitutive expression of tdT exclusively in Ins⁺Ki67⁺ beta cells, thereby enabling the specific tracing of beta cell proliferation (Fig. 2A).

Fig. 2.

Genetic tracing of beta cell proliferation during homeostasis. (A) Schematic of the Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT system for seamless recording of beta cell proliferation. (B) Experimental timeline for Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice. (C) Immunostaining for tdT and Ins on pancreatic sections from Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice. (D) Quantification of the percentage of tdT⁺ beta cells in Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice. Data are mean ± SD, n = 5. (E) Experimental timeline for the low-dose tamoxifen (Tam) labeling strategy in Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice. (F) Immunostaining for tdT on 100 μm-thick pancreatic sections from Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice after low-dose Tam treatment. (G) Distribution of tdT⁺ cell cluster sizes (number of cells per cluster) in Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice following low-dose Tam treatment. (H) Schematic of the Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT system for seamless recording of beta cell proliferation. (I) Schematic of the experimental timeline for Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT mice. (J) Immunostaining for tdT and Ins on pancreatic sections from Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT mice. (K) Quantification of the percentage of tdT⁺ beta cells in Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT mice. Data are mean ± SD, n = 5. (L) Cartoon image showing the specific labeling of pancreatic beta cell proliferation using the beta cell ProTracer system. (Scale bar, 100 μm.) Each image is representative of five individual samples.

To test the functionality of the beta cell ProTracer system, we treated adult beta cell ProTracer mice (aged 7 to 10 wk) with Tam and then euthanized the mice at 4, 8, and 12 wk afterward (Fig. 2B), according to the Institutional Animal Care and Use Committee (IACUC) guidelines at Center for Excellence in Molecular Cell Sciences (SIBCB-S374-1702-001-C1). Immunostaining for tdT and Ins on pancreatic sections from these mice revealed a stepwise increase in the number of tdT⁺ beta cells at 4, 8, and 12 wk after Tam treatment (Fig. 2C). Quantitatively, the proportion of Ins⁺ beta cells expressing tdT was 22.69 ± 2.94%, 38.96 ± 2.52%, and 59.24 ± 2.84% at 4, 8, and 12 wk post Tam induction, respectively (Fig. 2D). Notably, nearly all tdT⁺ pancreatic cells of beta cell ProTracer collected at 12 wk after Tam induction were confirmed to be Ins⁺ beta cells by immunofluorescent staining (SI Appendix, Fig. S2 A and B). This exceptional specificity originates from the beta cell–restricted Ins2 promoter driving DreER expression. We further validated this using a hard-to-recombine reporter line (Ins2-DreER;NR1), in which Ins2-DreER must excise a large (2.63 kb) DNA segment to activate tdTomato expression (SI Appendix, Fig. S2 C–E). The results showed that approximately 99.76% of beta cells were labeled with tdT, and 99.89% of tdT⁺ cells were Ins⁺ beta cells (SI Appendix, Fig. S2 F–G), confirming that Ins2-DreER has the strong recombinase activity and subsequent ProTracer activation are virtually exclusive to beta cells. This finding strongly supports the notion that the beta cell ProTracer system is highly specific for recording beta cell proliferation and does not label other cell lineages.

To further validate the system, we included some technical controls. In the pancreas collected from Tam-treated Ins2-DreER;R26-RSR-LSL-tdT and corn oil-treated Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice, hardly any tdT⁺ cells were detected in tissue sections (SI Appendix, Fig. S3 A–D). To confirm whether Ki67-Cre traced beta cells had indeed undergone cell cycle, we collected the pancreas from Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice after a Two-wk EdU water treatment (SI Appendix, Fig. S3E). A remarkable similarity was observed between tdT expression and EdU labeling (SI Appendix, Fig. S3F). Approximately 83.18% of tdT⁺ beta cells were also labeled by EdU (SI Appendix, Fig. S3G). Additionally, to determine whether most Ki67⁺ beta cells labeled by our system had indeed undergone cell division, we used a low-dose Tam to sparsely label Ki67⁺ beta cells in Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice (Fig. 2E). By analyzing thick (100 μm) pancreatic sections, we found that the majority of tdT⁺ clones consisted of a pair of neighboring cells, and approximately 82.50% of tdT beta cells were in clusters of two or more cells (Fig. 2 F and G). These results confirm that the beta cell ProTracer system primarily labels proliferating beta cells that have completed the cell cycle. The percentage of tdT⁺ beta cells increases over time because our system acts as an irreversible, cumulative recorder. Once activated by tamoxifen, it permanently marks every beta cell that subsequently begins to proliferate. This is not a snapshot, but rather an integral of all proliferation events over the observation period. Consistent with this, long-term tracing (48 wk) under homeostasis showed that approximately 83% of beta cells became tdT⁺ (SI Appendix, Fig. S3 H–J), which supports this idea. This plateau likely represents the high efficiency of the beta cell ProTracer system within its target population, which may be attributed to two nonexclusive factors: 1) the intrinsic recombination efficiency of the genetic tool may have reached its upper limit, and 2) a small subset of beta cells may have remained truly quiescent throughout the extended observation window.

We also generated an alternative beta cell ProTracer system based on another cell cycle gene Ccna2 (Fig. 2H) (27, 28). Similar to the Ki67-based strategy, we treated Tam to Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT mice and collected the pancreas for analysis 4, 8, and 12 wk afterward (Fig. 2I). The quantification data obtained from this alternative system closely resembled those of the Ki67-based ProTracer strategy. Specifically, the percentage of Ins⁺ beta cells expressing tdT was 19.95 ± 2.78%, 37.55 ± 2.68%, and 57.79 ± 3.59% at 4, 8, and 12 wk after Tam induction, respectively (Fig. 2 J and K). Moreover, in the pancreas samples collected from Tam-treated Ccna2-CrexER;R26-RSR-LSL-tdT and corn oil-treated Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT and mice, hardly any tdT⁺ cells were detected (SI Appendix, Fig. S3 K–N). Collectively, these experiments precisely and specifically quantify the proliferation rate of pancreatic beta cells in adult mice under normal physiological conditions (Fig. 2L).

Assessment of Beta Cell Proliferation Heterogeneity.

The potential heterogeneity in beta cell proliferation has significant implications for understanding the maintenance and regeneration of the beta cell population. If such heterogeneity exists, it is likely that specific subpopulations of beta cells play a more critical role in replenishing the new beta cell pool. To molecularly characterize proliferated beta cells and to explore potential heterogeneity, we manually isolated pancreatic islets from tamoxifen-treated Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice and performed single-cell RNA sequencing (scRNA-seq) (Fig. 3A). After quality control and dimensionality reduction, we identified the major islet cell types (beta, alpha, delta, and PP cells) along with other populations with smaller number, such as endothelial and immune cells (Fig. 3B). Focusing on beta cells, unsupervised clustering revealed four transcriptionally defined subpopulations, all of which robustly expressed canonical beta cell markers (e.g., Ins1, Ins2) (Fig. 3C). To investigate whether there are proliferative beta cell subsets with distinct transcripts, we detected the expression of tdT mRNA. We found that approximately 15.46% of the beta cells were tdT⁺, which was consistent with the histological quantifications (Fig. 3D). Importantly, tdT transcripts were uniformly distributed across all four beta cell subsets (Fig. 3 D and E), indicating that proliferated beta cells do not constitute a unique transcriptional state. Cell cycle analysis showed no enrichment of cycling-state cells in any specific subcluster of beta cells (Fig. 3F).

Fig. 3.

Single-cell RNA sequencing and clonal analysis of pancreatic beta cell proliferation. (A) Experimental strategy for isolating pancreatic islets to perform scRNA-seq from tamoxifen-treated Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice. (B) Left: t-SNE (t-distributed Stochastic Neighbor Embedding) visualization of the annotated cell types in pancreatic islets from tamoxifen-treated Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice. Cells are colored by their corresponding cell annotation according to the legend. Right: Heatmap showing the expression levels of marker genes used for cell-type annotation, with colors indicating the relative average expression across cell types. (C) Left: t-SNE visualization of pancreatic beta cell subpopulations from Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice. Cells are colored by their corresponding cell annotation according to the legend. Right: Expression levels of the canonical beta cell markers Ins1 and Ins2 across these subpopulations. Cells are colored by their corresponding gene expression levels. (D) Left: Expression levels of the tdTomato (tdT) across pancreatic beta cell subpopulations. Cells are colored by their corresponding gene expression levels. Middle: t-SNE visualization of tdT⁺ (tdT expression > 0) and tdT^– pancreatic beta cells. Cells are colored according to their annotated identities. Right: pie chart showing the proportion of tdT⁺ beta cells among all beta cells. (E) The proportion of tdT⁺ beta cells within each beta cell subcluster. (F) Cell cycle state of different beta cell clusters. The red dashed line indicates the 50% cell fraction threshold. (G) Comparison of cell cycle states between tdT⁺ and tdT^– beta cells. (H) Gene Ontology (GO) enrichment analysis highlighting the functional annotations of genes upregulated in tdT⁺ beta cells compared with tdT^– beta cells. The dashed red line indicates an adjusted P value of 0.05. (I) Representative differentially expressed genes that are upregulated in tdT⁺ beta cells compared to tdT^– beta cells. Statistical analysis was performed using a t test. ****P < 0.0001. % Ribo reads: The percentage of reads that map to the ribosome genome. (J) Schematic showing the clonal labeling strategy using the Ins2-DreER;Ki67-CrexER;R26-Confetti system. (K) Experimental timeline for Ins2-DreER;Ki67-CrexER;R26-Confetti mice. (L) Representative fluorescent images of pancreatic sections from Ins2-DreER;Ki67-CrexER;R26-Confetti mice. (M) Quantification of cell numbers per clone in pancreatic sections from Ins2-DreER;Ki67-CrexER;R26-Confetti mice. (N) Cartoon image showing the clonal labeling of pancreatic beta cell proliferation using Ins2-DreER;Ki67-CrexER;R26-Confetti mice. (Scale bar, 100 μm.) Images are representative of five individual samples.

We next directly compared the transcriptional heterogeneity of tdT⁺ and tdT^– beta cells. Cell cycle scoring based on transcriptional signatures confirmed the expected trend, with tdT⁺ cells showing no enrichment in S and G2/M phases (tdT⁺: 32.14% S, 37.93% G2/M; tdT^–: 34.17% S, 38.19% G2/M; Fig. 3G). Gene ontology (GO) analysis of upregulated genes in tdT⁺ beta cells showed enrichment for terms related to ribosome biogenesis and translation (Fig. 3H). Differential expression analysis identified a limited set of significantly upregulated genes (~130 total) in tdT⁺ beta cells, including cell cycle regulators (e.g., Cdk8) and ribosomal components (Fig. 3I). The subtlety of these transcriptional differences is consistent with the design of our ProTracer system as a cumulative recorder, which permanently labels cells that have proliferated during the tracing window, rather than capturing only those actively cycling at the moment of analysis. These results suggest that most beta cells return to a transcriptional state similar to that of nonproliferating peers by the time of tissue collection.

In addition, we used the clonal analysis approach (Ins2-DreER;Ki67-CrexER;R26-Confetti) on proliferating beta cells. The R26-Confetti allele allows for the labeling of individual cells with different fluorescent proteins, facilitating the analysis of cell clones (Fig. 3J). In our experimental design, these mice were treated with low-dose Tam and analyzed after 12 wk (Fig. 3K). The majority of the cell clones we observed were composed of either single cells or pairs of cells, and there were almost no clones with large number of beta cells (Fig. 3 L and M). This observation indicates that among the subset of beta cells that entered the cell cycle during the tracing period, there was no significant heterogeneity in subsequent proliferative output (reflected by similar clone sizes). (Fig. 3N). These data indicate that proliferated beta cells exhibit a relatively homogeneous transcriptional profile within the broader beta cell population.

Beta Cell Proliferation in Response to Injury or Drug Treatment.

The adult pancreas exhibits a remarkable regenerative ability following partial pancreatectomy (PPX). This well-established phenomenon provided an ideal physiological context to evaluate the efficacy of our beta cell ProTracer system. To this end, we treated beta cell ProTracer mice with Tam and performed PPX. We collected the pancreas from beta cell ProTracer for analysis 2 wk afterward (Fig. 4A). We performed immunostaining on pancreatic sections, using antibodies against tdT and Ins. As expected from previous research, the tdT signal was significantly more prominent in the PPX group compared to the control mice (Fig. 4 B–D). Quantitatively, in control Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice, 10.39 ± 2.03% of beta cells were tdT⁺, whereas in the PPX-treated mice, 71.29 ± 6.15% of beta cells were tdT⁺ (Fig. 4B). And approximately 85.62% of tdT⁺ cells exhibited EdU incorporation (SI Appendix, Fig. S4 A–C). Similarly, in control Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT mice,9.71 ± 2.13% of beta cells were tdT⁺, and this percentage increased to 68.45 ± 6.73% in the PPX-treated mice (Fig. 4C). These results clearly demonstrate that our beta cell ProTracer system can effectively capture the enhanced beta cell proliferation that occurs in response to the stress of PPX (Fig. 4D), validating its utility in monitoring beta cell proliferation under physiological conditions that promote cell division.

Fig. 4.

Beta cell proliferation in response to injury or drug treatment. (A) Experimental timeline for the beta cell ProTracer system following partial pancreatectomy (PPX) injury. (B and C) Representative immunofluorescence images of pancreatic sections stained for tdT and Ins (Left), and quantification of the percentage of tdT⁺ beta cells (Right) following PPX. Data are mean ± SD, n = 5. (D) Cartoon image showing the genetic labeling of pancreatic beta cell proliferation in response to PPX using the beta cell ProTracer system. (E) Experimental timeline for the beta cell ProTracer system with harmine and GW788388 (Harmine+GW) treatment. (F and G) Immunostaining for tdT and Ins on pancreatic sections of the indicated beta cell ProTracer mice (Left), and quantification of the percentage of tdT⁺ beta cells (Right). Data are mean ± SD, n = 5. (H) Cartoon image showing the genetic labeling of pancreatic beta cell proliferation in response to Harmine+GW treatment using the beta cell ProTracer system. (I) Experimental timeline for the beta cell ProTracer system with pancreatic ductal ligation (PDL) injury. (J and K) Immunostaining for tdT and Ins on pancreatic sections of the indicated beta cell ProTracer mice (Left), and quantification of the percentage of tdT⁺ beta cells (Right). Data are mean ± SD, n = 5. (L) Cartoon image showing the genetic labeling of pancreatic beta cell proliferation in response to PDL using the beta cell ProTracer system. (Scale bar, 100 μm.) Each image is representative of five individual samples.

Previous studies have reported that the combination of harmine (DYRK1A inhibitor) and GW788388 (GW, TGFbSF inhibitor) can significantly stimulate beta cell proliferation in vivo (14, 17). To further validate our system, we first treated beta cell ProTracer mice with Tam. After a One-wk washout period, harmine and GW were administered via daily intraperitoneal injection for One wk (Fig. 4E). Following this treatment, we performed immunostaining for tdT and Ins on pancreatic sections. The results revealed a significant increase in beta cell proliferation in the Harmine+GW-treated mice compared to the control group (Fig. 4 F–H). Quantitatively, in control Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice, 9.95 ± 2.49% of beta cells were tdT⁺, while in the Harmine+GW-treated mice, 37.35 ± 13.17% of beta cells were tdT⁺ (Fig. 4F). In Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT mice, 10.21 ± 1.32% of beta cells were tdT⁺ in the control group, and this increased to 41.34 ± 13.22% in the Harmine+GW-treated mice (Fig. 4G). These findings demonstrated that our beta cell ProTracer system can accurately detect changes in beta cell proliferation induced by drug treatment (Fig. 4H), suggesting its potential as a valuable tool for drug screening and validation. To evaluate the metabolic effects of beta cell proliferation induced by Harmine+GW, we performed a metabolic analysis on wildtype male mice after one wk of treatment (SI Appendix, Fig. S4D). We quantified body weight and fasting blood glucose at the end of the treatment period. The results showed that there was no difference in weight between the two groups (SI Appendix, Fig. S4E). Meanwhile, the fasting blood glucose level was not significantly different between the treatment group and the control group (SI Appendix, Fig. S4F). These data confirm that one wk of Harmine+GW treatment did not induce hyperglycemia or diabetes, indicating that glucose homeostasis remained unchanged despite increased beta cell proliferation. Next, we performed the intraperitoneal glucose tolerance test (IPGTT) to evaluate the overall glucose clearance rate. The area under the curve (AUC) of blood glucose during the 120 min IPGTT was significantly lower than in controls (SI Appendix, Fig. S4G). These data demonstrate that short-term, pharmacologically induced beta cell proliferation does not compromise, and may even transiently enhance, systemic glucose homeostasis.

Whether pancreatic ductal ligation (PDL) injury can stimulate beta cell proliferation remains unresolved (40–42). To address this, we next performed PDL in beta cell ProTracer mice (Fig. 4I). We compared the number of tdT⁺ beta cells in the PDL-affected tail region with that in the PDL-head region, which served as an internal control. We did not detect a significant increase of tdT⁺ beta cells in PDL-tail compared to the PDL-head (Fig. 4 J–L). Quantitatively, in the PDL-head region of Ins2-DreER;Ki67-CrexER;R26-RSR-LSL-tdT mice, 10.20 ± 1.76% of beta cells were tdT⁺, and in the PDL-tail region, 10.30 ± 1.24% of beta cells were tdT⁺ (Fig. 4J). Similarly, in the PDL-head region of Ins2-DreER;Ccna2-CrexER;R26-RSR-LSL-tdT,12.86 ± 1.81% of beta cells were tdT⁺, and in the PDL-tail region, 13.66 ± 2.10% of beta cells were tdT⁺ (Fig. 4K). Taken together, these genetic tracing results extend the previous findings on the proliferative beta cells after injury or drug treatment. Recapitulating beta cell regeneration by continuous beta cell ProTracer recording shows the value for studying or exploring the multiple methods to stimulate beta cell proliferation.

Discussion

The development of techniques to trace in vivo pancreatic beta cell proliferation is important for the exploration of methods to promote pancreas regeneration and holds promise for advancing diabetes treatment strategies. In this study, we generated the beta cell ProTracer based on the dual recombinases system, which is designed to record beta cell proliferation events during both homeostasis and tissue regeneration. Using the beta cell ProTracer system, we found that beta cell proliferation increases over time under homeostatic conditions. Single-cell RNA sequencing and clonal analysis revealed homogeneous proliferative potential among proliferated beta cells. We also used PPX and pharmacological stimulation to monitor beta cell proliferation. Both interventions induced robust proliferative responses, validating the effectiveness of the system in quantifying beta cell proliferation. Additionally, we addressed the controversial role of pancreatic duct ligation in promoting beta cell proliferation. Despite the previous report suggesting pancreatic duct ligation promotes beta cell proliferation (40), our ProTracer-based analysis demonstrated no significant increase in beta cell proliferation following pancreatic ductal ligation.

A defining feature of the beta cell ProTracer system is its ability to permanently record cumulative beta cell proliferation events over a long period of time. It acts as a “proliferation recorder” rather than merely providing a transient “snapshot” of proliferation at a single time point, which differs from many traditional methods. Our system can trace cumulative proliferation events from wk to mo, capturing the total number of beta cells that have proliferated throughout the observation period. The ProTracer system can be potentially used to monitor beta cell proliferation in diabetic models, enabling in vivo screening and validation of drugs or small molecules that promote beta cell proliferation, and providing important insights into disease progression and therapeutic efficacy. In contrast, thymidine analogues only label cells in S phase during a brief administration period. They cannot accumulate signals from proliferation events occurring before or after dosing, and their signal is further diluted by subsequent cell divisions, which makes long-term cumulative quantification impossible. Similarly, many existing genetic proliferation models (e.g., Ki67-CreER) rely on transient CreER activation by a single tamoxifen pulse. This pulse only marks cells that are proliferating at that time. The conventional models lack the irreversible “recording” capacity to capture cumulative proliferation over time. The second advantage is the system’s specificity for beta cells, ensuring that quantified proliferation events are exclusively from beta cells, without interference from other pancreatic or extrapancreatic cell types. The specificity is achieved by using the beta cell–restricted Ins2 promoter to drive the DreER, which restricts the initial activation of the ProTracer cassette to beta cells. Thymidine analogues label proliferating cells, but they cannot distinguish beta cells from other cell types. They rely on coimmunostaining with beta cell markers to identify beta cell–specific proliferation, which may be prone to signal overlap in some conditions. In addition, the permanent tdT labeling offers greater flexibility for downstream analyses that complement proliferation quantification. For example, stable tdT expression allows us to subsequently isolate the cumulative pool of proliferated beta cells through fluorescence-activated cell sorting for functional assays or molecular profiling.

However, it is important to acknowledge some limitations of the beta cell ProTracer system. Although Ki67 is a well-established marker for cell proliferation, being expressed during DNA replication and nuclear division in the cell cycle, the use of this marker may result in the inclusion of beta cells that have initiated but not completed the cell cycle among the labeled cells. The observation that approximately 20% of tdT⁺ beta cells lack EdU incorporation or do not form doublets is consistent with this understanding and could be attributed to several nonexclusive possibilities. First, some of these cells likely represent cases in which Ki67 was transiently activated at the initiation of cell proliferation but did not complete the cell division cycle. This could occur if the cells arrest at a subsequent checkpoint or exit the cycle prematurely. This is a limitation of Ki67-based proliferation assays. Second, for the subset of cells without EdU incorporation, one possible explanation is insufficient EdU uptake. Our EdU administration strategy, which involves adding EdU to the drinking water, may not ensure 100% uptake by all proliferating cells. While this method is widely used for long-term in vivo labeling to avoid repeated injections and minimize stress, it depends on the mice to consistently and voluntarily drink the EdU-containing water. Third, the absence of doublets could be due to the following two possible conditions. After cell division, one daughter cell may undergo apoptosis, leaving only a single detectable tdT⁺ cell. Alternatively, while one cell is positioned in the tissue section, another paired cell may not be captured in the current section during tissue sectioning, thus failing to observe a detectable doublet. These minor limitations are inherent to most genetic tracing systems and do not diminish the primary utility of our model. The strong concordance between our Ki67- and Ccna2-based ProTracer systems provides robust, corroborative evidence for the proliferative events we describe.

The substantial difference in maximum labeling efficiency between the Ins2-DreER and R26-DreER drivers directly explains the discrepancy in reported percentages of proliferating beta cells between the universal ProTracer system and beta cell–specific ProTracer system. Therefore, for studies specifically focused on beta cell proliferation, we recommend the beta cell–specific ProTracer system due to its higher sensitivity within the lineage of interest. The universal ProTracer system is more suitable for research that requires the simultaneous assessment of the proliferation across multiple cell types in the pancreas.

Moreover, caution must be exercised when translating the findings from rodents to clinical applications. There are notable differences in pancreatic islets of humans and rodents, particularly in terms of structure and gene expression (1, 43). Meanwhile, there is an inverse relationship between cellular mature functionality and proliferation capacity of beta cells (32, 44–47), which highlights a new aspect of the drugs promoting beta cell proliferation should be used with caution. Despite the substantial beta cell expansion, the preservation of fasting normoglycemia can be explained by considering broader pharmacological effects. As a DYRK1A inhibitor, harmine can promote the proliferation of other islet cell types, including glucagon-producing alpha cells (14, 17). A dynamic balance may have been established between the expanded beta and alpha cell populations, which helps maintain the baseline glucose levels. Upon exogenous glucose stimulation (IPGTT), the increased beta cell mass may provide greater secretory capacity, leading to enhanced glucose clearance ability. These results refine the discussion on the inverse relationship between proliferation and maturity, that short-term pharmacological induction of proliferation might expand the functional beta cell pool. These results suggest that the “inverse relationship” is context-dependent and represents a dynamic balance that can be favorably regulated.

In light of these challenges, future research should focus on overcoming these key issues to safely and effectively promote the proliferation of pancreatic islet beta cells without causing adverse effects, especially tumor formation. Our study has provided valuable tools and insights into beta cell proliferation dynamics, but significant work remains to translate these findings into practical clinical applications.

Materials and Methods

All mouse experiments were performed in accordance with the guidelines of the IACUC at the Center for Excellence in Molecular Cell Science, Chinese Academy of Sciences. The methods included the following, mice information, genomic PCR (SI Appendix, Fig. S4H), pancreatic islet isolation and digestion, quantitative PCR, EdU incorporation assay, Harmine and GW788388 treatment, intraperitoneal glucose tolerance test, PPX, pancreatic ductal ligation, immunofluorescence staining, single-cell RNA sequencing, and statistical analysis. A detailed description of the methods used in this study is provided in the SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The scRNA-seq data generated in this study have been deposited in the Gene Expression Omnibus (GEO) database under accession number GSE313841 (48). All study data are included in the article and/or the SI Appendix.