β-Cell Dedifferentiation in HOMA-β^low and HOMA-β^high Subjects

Abstract

In recent years, mounting evidence has suggested that β-cell dedifferentiation may contribute to the loss of functional β-cell mass, thereby exacerbating the progression of diabetes (1-4). β-Cell dedifferentiation, evidenced by loss of the mature differentiated phenotype and regression to a precursor-like state, was significantly induced in diabetic islets in rodents (5-10) and humans (11-14). Our previous studies have demonstrated that the ratio of β-cell dedifferentiation could be enhanced 3 to 4 times by inflammation (12) or the cancer microenvironment (15) before hyperglycemia occurs, which might be a major cause of type 3c diabetes. Moreover, our team reported that defective unfolded protein response (UPR) linked β-cell dedifferentiation in elderly individuals without diabetes might be responsible for age-related β-cell failure (16). However, considering the relative low rates of β-cell dedifferentiation in diabetic islets (10-25%) (11-13), one might wonder the real contribution of this process to β-cell failure in vivo. Thus, it is important to establish a direct link between β-cell dedifferentiation in situ and insulin-releasing function in vivo. Moreover, the novel candidates for β-cell failure identified from rodents warrant further exploration in human.

In the current work, we acquired homeostasis model assessment of (β-cell function HOMA-β) values 1 week before pancreatectomy in patients with benign pancreatic cystic neoplasm (PCN) or intrapancreatic accessory spleen, and evaluated their rates of β-cell dedifferentiation in islets from paraneoplastic pancreatic tissue. We found that the rate of dedifferentiation was negatively correlated with their HOMA-β values. We further classified subjects according to their HOMA-β values, and found that the pancreatic islets from HOMA-β^low subjects were marked in situ as (1) increased β-cell dedifferentiation ratio, (2) induction of dedifferentiation maker Aldh1a3, and (3) lower expression of β-cell functional markers Glut2 and Ucn3. We also found that basic leucine zipper transcription factor 2 (Bach2), a key driver of type 2 diabetes (T2D) cell state transitions (17), was enriched in HOMA-β^low human islets and was positively related to the ratio of β-cell dedifferentiation, reinforcing its unique role in β-cell dedifferentiation during the development of T2D in humans.

Materials and Methods

Patients and Human Pancreas

A total of 4843 subjects with partial pancreatectomy for various reasons performed in the Department of Surgery in Ruijin Hospital from November 2017 to January 2023 were recruited. By reviewing the pathology diagnosis, subjects who had pancreatic malignant tumors were excluded. Cases were enrolled in the current study according to the following criteria: (1) with benign PCN or intrapancreatic accessory spleen; (2) no prior record of diabetes; (3) no history of chronic pancreatitis; (4) with a comprehensive glycemic test, including fasting plasma glucose, 2-hour postprandial plasma glucose (2hPG), fasting C-peptide, 2-hour postprandial plasma C-peptide (2hr C-peptide), and HbA1c% within 1 week before pancreatectomy.

All the clinical characteristics of the 13 subjects enrolled in the study are summarized in Table 1. Paraffin sections of pancreatic tissues far from the surgical margin of the pancreatectomy were acquired from the Department of Pathology at Ruijin Hospital for further examination. These tissue samples underwent reconfirmation by a pathologist to validate the accuracy of the final pathology diagnosis.

This study was approved by the Institutional Review Board of Ruijin Hospital affiliated to Shanghai Jiao-Tong University School of Medicine and was in accordance with the principles of the Declaration of Helsinki II. Informed consent was waived due to the retrospective and observational nature of the study. Only patients with telephone follow-up were provided with informed consent in this study.

Immunostaining

Human pancreas specimens were fixed and prepared for immunohistochemistry using a standard protocol, as previously described (18). All slides were treated with tissue antigen recovery (H-3300, Vector) and were incubated at 4 °C overnight with primary antibodies in Dako Antibody Diluent (Dako, Burlington Canada). The dilutions were as follows: guinea pig anti-insulin (1:10, Agilent Cat# IR002, RRID:AB_2800361), rabbit antiglucagon (1:200, Cell Signaling Technology, Cat# 2760S, RRID:AB_659831), rabbit antisomatostatin (1:200, proteintech, Cat#17512-1-AP, RRID:AB_2195910), rabbit antipancreatic polypeptide (1:200, Millipore, Cat#AB939, RRID:AB_92383), mouse antichromogranin A (1:100, Millipore, Cat#MAB5268, RRID:AB_11213294), rabbit anti-Aldh1a3 (1:200, Novus, Cat#NBP215339, RRID:AB_2665496), rabbit anti-Glut2 (1:200, Proteintech, Cat#20436-1-AP, RRID:AB_2750600), rabbit anti-UCN3 (1:200, Sigma-Aldrich, Cat#HPA038281, RRID:AB_10672408), rabbit anti-Ki67(1:100, Bethyl, CAT#IHC00375, RRID:AB_1547959), anti-Gck (1:100, Proteintech, Cat#19666-1-AP, RRID:AB_10863656), rabbit anti-PS6 (Ser240/244) (1:100, Cell Signaling Technology, Cat#5364S, RRID:AB_10694233), rabbit anti-Nrf2 (1:100, Abcam, Cat#ab62352, RRID:AB_944418), and rabbit anti-Cyb5r3 (1:100, Proteintech, Cat#10894-1-AP, RRID:AB_2292715). Apoptosis cells were determined using the One-step TUNEL Cell Apoptosis Detection Kit (red fluorescence) (Beyotime, Cat#C1090). Antibodies were detected using Alexa Fluor 488, 594, and 647 (1:500, Jackson ImmunoResearch Laboratories) as secondary antibodies. Nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI, Vector Laboratories; Burlingame, CA).

Image Acquisition and Analysis

Zeiss LSM 880 confocal microscopes were used to obtain immunofluorescence images. We employed the cell counter function of Image J in a blinded manner to analyze and count cells across the entire section, and analyzed 10 islets and at least 980 ± 110 cells per case. Initially, we detected the presence of chromogranin A (CGA)–positive cells, indicative of endocrine-derived cells, and subsequently identified CGA-positive cells lacking hormone expression as dedifferentiated cells. The fluorescence intensity was quantified in a blinded manner utilizing ImageJ software (National Institutes of Health, Bethesda, MD) with the EzColocalization plugin following the methodology outlined by Stauffer and colleagues (19).

Statistical Analysis

Data are presented as means ± SEM. Statistical analysis was performed using the unpaired Student t test with data from 2 groups. Correlation coefficients were calculated by simple and multiple regression analyses. P < .05 was considered to be statistically significant. All statistical analyses were conducted using SPSS version 24.0 statistical software (IBM, Armonk, NY).

Results

β-Cell Dedifferentiation Ratio and in Vivo β Cell Function (HOMA-β) in Humans

To investigate whether β-cell dedifferentiation is associated with β-cell function in humans, we calculated HOMA-β values and β-cell dedifferentiation ratios of 13 individuals who had no history of diabetes, chronic pancreatitis, and pancreatic cancer. Their fasting plasma glucose and 2hPG, fasting C-peptide, 2hr C-peptide, HbA1c%, and other clinical parameters are summarized in Table 1. Their HOMA-β values (ranged from 56.9% to 147.8%) were determined using the HOMA2 Calculator (Diabetes Trials Unit, University of Oxford).

We assessed the number of CGA-positive and islet hormone-negative cells (defined as dedifferentiated cells, arrows in Fig. 1A) in pancreas sections from all subjects as previously described (12, 15, 16). Interestingly, we found a robust inverse correlation between the ratio of β-cell dedifferentiation and their HOMA-β values (r = −0.79, P = .0012; Fig. 1B). Moreover, the ratio of β-cell dedifferentiation was inversely related to 2hr C-peptide as well (r = −0.55, P = .051; Fig. 1C). In contrast, there is no significant difference between the ratio of multihormone cells (coexpressing insulin and Gcg/Sst/Pp) and HOMA-β values (r = 0.18, P = .56; Fig. S1A and 1B (20)) in these subjects.

Figure 1.

Representative immunofluorescence sections showing dedifferentiated cells. (A) Pancreas from case # 309 (%diff/CGA: 3.50, %HOMA-β: 128.0), case # 682 (%diff/CGA: 7.68, %HOMA-β: 84.2), case # 306 (%diff/CGA: 8.71, %HOMA-β: 97.1) stained with CGA (green), insulin (white), endocrine cocktail (Gcg/Sst/Pp, red), and DAPI (blue) are shown. The specific area of the islet is highlighted within a white box, and enlarged to demonstrate the presence of dedifferentiated cells, indicated by yellow arrows. (B, C) The Pearson correlations between HOMA-β and 2hr C-peptide with dedifferentiated cells/CGA-positive cells. (D) Multivariate association of the clinical variable with the ratio of dedifferentiated cells in islets. Scale bars, 20 μm. The r of Pearson correlation and P values are shown in each panel. n = 13, *P < .05 was considered to be statistically significant.

Open in new tabDownload slide

We then conducted multiple regression analyses to examine the relationship between the percentage of dedifferentiated cells and their clinical parameters (ie, body mass index [BMI], HOMA-β, HbA1c%). The results showed that HOMA-β, rather than BMI and HbA1c%, was independently and negatively correlated with the percentage of dedifferentiated cells (β = .0659, P = .0060; Fig. 1D). These findings strongly support that the process of human β-cell dedifferentiation in situ is closely linked to their insulin-releasing function in vivo.

Aldehyde dehydrogenase 1 isoformA3 (Aldh1a3) is a marker of β-cell dedifferentiation (11, 12, 15, 21). We performed immunohistochemistry with Aldh1a3, insulin, and glucagon on pancreas section of all subjects and noticed that most Aldh1a3-positive cells in islets were Gcg⁺ cells, but there was still a portion of Ins⁺Gcg⁻ cells expressing Aldh1a3 protein (Fig. S2A (20)). We counted the percentage of Aldh1a3⁺Ins⁺ cells and found a significant and positive correlation between the rate of β-cell dedifferentiation and Aldh1a3⁺Ins⁺/Ins⁺ cells (r = 0.65, P = .017; Fig. 2B) or Aldh1a3⁺ cells per islet (r = 0.63, P = .022; Fig. S2B) (20). Moreover, we detected a negative correlation between the percentage of Aldh1a3-positive β cells and HOMA-β (r = −0.74, P = .006; Fig. 2C), providing additional evidence that β-cell dedifferentiation was directly linked to β-cell function in human.

Figure 2.

Correlation of Aldh1a3, Glut2 and Ucn3 expression with the ratio of dedifferentiated cells/CGA-positive cells. (A) Representative images co-labeled with Aldh1a3 (red) and insulin (green) in subjects. (B, C) Correlation between Aldh1a3⁺Ins⁺/Ins⁺ cells and the dedifferentiated cells/CGA-positive cells (B) or HOMA-β (C). (D) Representative images co-labeled with Glut2 (red) and insulin (green) in subjects. (E, F) Correlation between Glut2 expression and the dedifferentiated cells/CGA-positive cells (E) or HOMA-β (F). (G) Representative images co-labeled with Ucn3 (red) and insulin (green) in subjects. (H, I) Correlation between Ucn3 expression and the dedifferentiated cells/CGA-positive cells (H) or HOMA-β (I). Scale bars, 20 μm. The r of Pearson correlation and P values are displayed in each panel, n = 11–13.

Open in new tabDownload slide

Human β-cell Dedifferentiation and in Situ β Cell Function

To evaluate whether β-cell dedifferentiation is also closely related to β-cell function in situ, we carried out immunohistochemistry with Glut2 (encoded by SLC2A2) or urocortin3 (Ucn3), both of which are essential for glucose-stimulated insulin secretion (22-28), together with insulin. We noticed that Glut2 expression tended to be more abundant in islets with a lower β-cell dedifferentiation rate (Fig. 2D). Quantified Glut2 intensity was negatively correlated with the rate of β-cell dedifferentiation (r = −0.61, P = .045; Fig. 2E) and positively correlated with HOMA-β (r = 0.82, P = .0018; Fig. 2F). This was also the case for Ucn3, whose expression showed a negative relation with the β-cell dedifferentiation rate (r = −0.58, P = .060; Fig. 2G). These in situ findings further support the close link between β-cell dedifferentiation and β-cell function in human.

β-Cell Dedifferentiation and Bach2 Expression in Human Individuals With low and High HOMA-β Levels

It has been reported that β-cell function from normal young adults were tested with a recalibrated HOMA-β model to give a HOMA-β value of 100% (29). Therefore, we divided our subjects into HOMA-β^low (<100) and HOMA-β^high (>100) groups. Significant differences in HOMA-β values (80.7% ± 8.5% in HOMA-β^low vs 123.4% ± 4.2% in HOMA-β^high, P = .0003), fasting C-peptide (1.4 ± 0.1 in HOMA-β^low vs 2.5 ± 0.2 in HOMA-β^high, P = .007) and 2hr C-peptide (8.4 ± 0.9 in HOMA-β^low vs 14.7 ± 1.5 in HOMA-β^high, P = .024) were detected between the 2 groups; however, age, sex, BMI, fasting plasma glucose, 2hPG and HbA1c% were comparable (Fig. 3A).

Figure 3.

The rate of dedifferentiation in individuals with HOMA-β^low and HOMA-β^high levels. (A) Characteristic of the study subjects. (B, C) Quantitative analysis of dedifferentiated cells/CGA-positive cells (B) and Aldh1a3⁺Ins⁺/Ins⁺ cells (C) in HOMA-β^low and HOMA-β^high groups (n = 13). (D, E) Average fluorescent intensity of Glut2 (D) and Ucn3 (E) in HOMA-β^low and HOMA-β^high groups (n = 11). Data are presented as means ± SEM. The 2 groups were compared using an unpaired Student t test, *P < .05, **P < .01, ****P < .0001.

Open in new tabDownload slide

Importantly, there was an approximately 2-fold rise in the percentage of dedifferentiated cells (8.42% ± 0.26% in HOMA-β^low vs 4.09% ± 0.26% in HOMA-β^high, P < .0001; Fig. 3B) as well as in Aldh1a3⁺Ins⁺/Ins⁺ cells (14.33% ± 1.66% in HOMA-β^low vs 7.28% ± 1.17% in HOMA-β^high, P = .0061; Fig. 3C) in HOMA-β^low group. In contrast, Glut2 intensity was significantly lower (9.22 ± 2.92 in HOMA-β^low vs 18.53 ± 2.06 in HOMA-β^high, P = .026; Fig. 3D) in islets from HOMA-β^low group. Consistently, Ucn3 expression was slightly decreased (28.48 ± 3.00 in HOMA-β^low vs 36.63 ± 3.46 in HOMA-β^high, P = .15; Fig. 3E) as well in HOMA-β^low group. There was no difference in the rate of β-cell proliferation and apoptosis between HOMA-β^low and HOMA-β^high groups (Fig. S3A and 3B) (20). Thus, the pancreas tissues from HOMA-β^low and HOMA-β^high subjects in the present study can also be used for screening candidates related to β-cell dedifferentiation/dysfunction in human.

Several important targets have been reported to involve rodent β-cell dedifferentiation/function, including the mechanistic target of rapamycin complex 1 (mTORC1) (30-32), Nuclear factor erythroid 2–related factor (Nrf2) (33-35), β-Cell glucokinase (Gck) (36, 37), oxidoreductase cytochrome b5 reductase 3 (Cyb5r3) (38), and basic leucine zipper transcription factor 2 (Bach2) (17). We therefore stained for insulin and/or PS6 (mTORC1 target), Nrf2, Gck, Cyb5r3, and Bach2 in pancreas sections of these subjects. Bach2 was mainly expressed in alpha cells in the islets, but there was still a portion of Ins⁺Gcg⁻ cells that had Bach2 expression (Fig. S4 (20)). Interestingly, we found very few Bach2⁺Ins⁺ cells in islets with low β-cell dedifferentiation ratio, but a strong induction of Bach2⁺Ins⁺ cells in islets with high β-cell dedifferentiation ratio (Fig. 4A). The percentage of Bach2⁺ β-cells was positively correlated with the ratio of β-cell dedifferentiation (r = 0.63, P = .0498; Fig. 4B). We then counted the percentage of Bach2⁺ β-cells between HOMA-β^low and HOMA-β^high groups and found a nearly 2-fold increase in the percentage of Bach2⁺Ins⁺ cells in HOMA-β^low group (9.21% ± 1.54% in HOMA-β^low vs 5.55% ± 0.65% in HOMA-β^high, P = .029; Fig. 4C). In contrast, we did not find significant differences in the expression levels of PS6, Gck, Nrf2, and Cyb5r3 between the 2 groups (Fig. 4D-4G; Fig. S5A-5D) (20).

Figure 4.

The increase of Bach2 expression positively correlates with the dedifferentiation of pancreatic β-cells in human pancreatic islets. (A) Representative images co-labeled with Bach2 (red) and insulin (green) in subjects. (B) Correlation between Bach2 expression and the dedifferentiated cells/CGA-positive cells. (C) Quantitative analysis of dedifferentiated cells/CGA-positive cells in HOMA-β^low and HOMA-β^high groups (n = 10). (D-G) Average fluorescent intensity of PS6 (D), Gck (E) Nrf2 (F), and Cyb5r3 (G) in HOMA-β^low and HOMA-β^high groups (n = 11–13). Scale bars, 20 μm. Data present means ± SEM. The 2 groups were compared using an unpaired Student t test. The r of Pearson correlation and P values are displayed in each panel, *P < .05.

Open in new tabDownload slide

Discussion

The progressive nature of T2D is mainly due to continuous loss of β-cell function and mass (39). In 2003, Butler et al utilized pancreas autopsy samples to measure β-cell mass and found an approximately 50% reduction in relative β-cell volume in T2D compared with individuals without diabetes (40). The UKPDS study demonstrated an approximately 50% reduction of β-cell function, evaluated by the HOMA1 equation (%HOMA-β = 20 × FPI (fasting plasma insulin)/(FPG – 3.5)) in patients with T2D at the time of diagnosis (41). In recent years, multiple studies have provided evidence that the dedifferentiation of β-cells in rodent diabetic islets might play a pivotal role in the decline of β-cell function and mass, thereby exacerbating the advancement of diabetes in rodents (5-10). In humans, we and others have found that the ratios of dedifferentiated β-cells in diabetic islets, varying from 10% to 25% (11-13) due to differences in BMI, diabetes duration, utilization of antidiabetic drug, were always several times higher than the ratio found in islets from people without diabetes (1.2∼6.5%) (11-13, 16). However, the direct link between β-cell dedifferentiation in situ and β-cell function in vivo has not yet been well established.

In the present study, we provided evidence that β-cell dedifferentiation is closely related to β-cell function in human. First, both the frequency of dedifferentiated cells and the dedifferentiation marker (Aldh1a3) were inversely related to their insulin releasing function (HOMA-β) in vivo. Second, the frequency of β-cell dedifferentiation and Aldh1a3 was negatively related to the expression levels of β-cell functional markers Glut2 and/or Ucn3, which are closely related to insulin secretion of functional β-cells (22-25). Third, when subjects were subclassified according to their HOMA-β values, we found that the islets from HOMA-β^low individuals were manifested as significantly increased ratios of dedifferentiation cells and Aldh1a3 protein abundance, as well as reduced Glut2 and Ucn3 expression, compared with HOMA-β^high subjects. We noticed that the percentage of Aldh1a3⁺ cells (14.33% ± 1.17% in HOMA-β^low) was higher than the dedifferentiation ratio (8.42% ± 0.26% in HOMA-β^low), indicating that more β-cells may be on their way to dedifferentiate, albeit without reaching the stage of complete loss of insulin production. Considering the subjects enrolled in the present study were individuals without a history of diabetes and some of them were diagnosed as prediabetes/diabetes for the first time, it is rational to believe that the dedifferentiation ratio might rise substantially during disease progression (13).

The phenomenon of β-cell dedifferentiation was found to be reversible upon normalization of blood glucose levels (5, 42), and hyperglycemia and several pathways of glucotoxicity, including oxidative stress (31, 43-45), endoplasmic reticulum stress (16, 46-53), and hypoxia (15, 54-56), were proposed to participate in β-cell dedifferentiation. Our previous studies have demonstrated that adaptive UPR defects (16), inflammation (12), and/or cancer microenvironment (15) can induce human β-cell dedifferentiation. Aldh1a3 is a marker of β-cell dedifferentiation in mice (5-10) and humans (11, 12). To understand the mechanism underlying β-cell dedifferentiation/dysfunction, Accili’s group isolated ALDH⁺ islet cells from β-cell–specific Foxo knockouts and identified a narrow set of candidate genes, including Bach2 and Cyb5r3, that may affect the transition from a healthy to a dysfunctional β-cell (21). Rodent studies found that Cyb5r3 linked FoxO1-dependent mitochondrial dysfunction to β-cell failure (38), but did not involve β-cell dedifferentiation (57). In the present study, we also did not detect a direct relation between Cyb5r3 expression and β-cell dedifferentiation. However, a positive correlation between the expression of Bach2 and the ratio of β-cell dedifferentiation in human islets was established. We also found a robust increase in Bach2 levels in HOMA-β^low subjects. This finding is consistent with our previous observation in which Bach2 was enriched in diabetic human islets (17). More importantly, Son et al found that treatment with a Bach inhibitor could lower glycemia and restore insulin secretion in diabetic mice and human islets (17).

In summary, our findings demonstrated a close link between β-cell dedifferentiation in situ and β-cell function in vivo in humans. Furthermore, we underlined the importance of Bach2 in human β-cell dedifferentiation and function, reinforcing its unique role as a drug target for the treatment of diabetes.

Funding

This work was supported by the National Natural Sciences Foundation of China (82070795, 82270841, 82311530702).

Author Contributions

F.Y.K. performed the experiments, analyzed the data, and wrote the manuscript. Z.Z. revised the manuscript and contributed to the discussion. H.F. and. J.J.S. performed some experiments. J.Z. collected human samples and contributed to the discussion. Q.D.W. designed and supervised the project, wrote and revised the manuscript. All authors critically reviewed and approved the final version of the manuscript. Q.D.W. and J.Z. take responsibility for the integrity of the data and the accuracy of the data analysis.

Disclosures

The authors have nothing to disclose.

Data Availability

All data generated or analyzed during this study are included in the article. Further information about the data are available from the corresponding author upon request.

References

Abbreviations

 
• 2hPG

  2-hour postprandial plasma glucose

 
• 2hr C-peptide

  2-hour postprandial plasma C-peptide

 
• BMI

  body mass index

 
• CGA

  chromogranin A

 
• FPI

  fasting plasma insulin

 
• HOMA-β

  homeostasis model assessment of β-cell function

 
• PCN

  pancreatic cystic neoplasm

 
• T2D

  type 2 diabetes

 
• UPR

  unfolded protein response

Author notes