Abstract
Type 2 diabetes (T2DM) is associated with pancreatic islet dysfunction. Loss of β-cell identity has been implicated via dedifferentiation or conversion to other pancreatic endocrine cell types. How these transitions contribute to the onset and progression of T2DM in vivo is unknown. The aims of this study were to determine the degree of epithelial-to-mesenchymal transition occurring in α and β cells in vivo and to relate this to diabetes-associated (patho)physiological conditions. The proportion of islet cells expressing the mesenchymal marker vimentin was determined by immunohistochemistry and quantitative morphometry in specimens of pancreas from human donors with T2DM (n = 28) and without diabetes (ND, n = 38) and in non-human primates at different stages of the diabetic syndrome: normoglycaemic (ND, n = 4), obese, hyperinsulinaemic (HI, n = 4) and hyperglycaemic (DM, n = 8). Vimentin co-localised more frequently with glucagon (α-cells) than with insulin (β-cells) in the human ND group (1.43% total α-cells, 0.98% total β-cells, median; P < 0.05); these proportions were higher in T2DM than ND (median 4.53% α-, 2.53% β-cells; P < 0.05). Vimentin-positive β-cells were not apoptotic, had reduced expression of Nkx6.1 and Pdx1, and were not associated with islet amyloidosis or with bihormonal expression (insulin + glucagon). In non-human primates, vimentin-positive β-cell proportion was larger in the diabetic than the ND group (6.85 vs 0.50%, medians respectively, P < 0.05), but was similar in ND and HI groups. In conclusion, islet cell expression of vimentin indicates a degree of plasticity and dedifferentiation with potential loss of cellular identity in diabetes. This could contribute to α- and β-cell dysfunction in T2DM.
Introduction
Phenotypic plasticity of fully differentiated cells in adult pancreatic islets has been demonstrated in several rodent models of diabetes in vivo (Thorel et al. 2010, Talchai et al. 2012, Brereton et al. 2014, Piran et al. 2014, Wang et al. 2014, Kim-Muller et al. 2016), in human islet cells in vitro (Bramswig et al. 2013, Spijker et al. 2013) and in human pancreatic islets in vivo (Spijker et al. 2015, Butler et al. 2016, Cinti et al. 2016). These cellular transitions of one mature endocrine cell type to a progenitor-like state or another islet cell type demonstrate that fully differentiated islet cells in adult pancreas have the capacity for conversion. This is likely to affect hormone secretion of individual cells (Brereton et al. 2014), but it is as yet unknown whether it affects islet function to a clinically relevant degree in humans.
In rodent models, transcription factors that are important in islet cell development and embryonic differentiation (e.g. Pdx1, Ngn3, MafA, Foxo1, Arx, Pax4) are involved in the regulation of these processes and can be triggered by stimuli including hyperglycaemia and cellular stress (Jonas et al. 1999, Dhawan et al. 2011, Papizan et al. 2011, Yang et al. 2011, Szabat et al. 2012, Talchai et al. 2012, Al-Hasani et al. 2013, Courtney et al. 2013, Gao et al. 2014, Nishimura et al. 2015). Recently, our group reported reprogramming of β-cells in human pancreatic islets in vitro and in vivo (Spijker et al. 2013, 2015); both bihormonal cells and cells transitioning from β- to α-cells were identified to be associated with changes in relevant transcription factors. Similar transitioning phenotypes and endocrine cells expressing the mesenchymal protein vimentin have been described in Foxo1-knockout mice in combination with metabolic stress (Talchai et al. 2012); this was associated with the onset of hyperglycaemia, suggesting that β-cells can convert to a mesenchymal-like phenotype in diabetes.
Human β-cells can undergo an epithelial-to-mesenchymal transition (EMT) (including expression of vimentin) when cultured as a monolayer ex vivo (Gershengorn et al. 2004, Russ et al. 2009). Vimentin expression in a subset of islet cells has been described as a limited number of human post-mortem pancreas specimens (Rukstalis & Habener 2009, Guo et al. 2013, Gao et al. 2014). In addition, case reports of 3 subjects proposed that vimentin expression was linked to aetiology of type 2 diabetes (T2DM) (White et al. 2013). However, it is not known if this aberrant expression of a mesenchymal protein is a more widespread feature of T2DM or how vimentin expression is linked to the aetiology of T2DM and islet (patho)physiology in vivo (White et al. 2016).
The aim of this study was to determine the proportions of vimentin-positive endocrine cells in islets in vivo, using quantitative immunohistochemistry in post-mortem samples of human and non-human primate pancreas and to determine if this was associated with the onset and pathophysiology of T2DM.
Materials and methods
Human pancreas and islets
Pancreatic tissue was obtained from subjects at post-mortem examination, including human organ donors. Samples were taken from subjects with a documented history of type 2 diabetes (n = 28) and without diabetes (n = 38) (Supplementary Table 1, see section on supplementary data given at the end of this article). Pancreata from organ donors were procured through donor programmes in The Netherlands and in Oxford, UK. Post-mortem and multi-organ donor tissue was obtained according to National laws and National and Local ethics guidelines and permissions. Pancreatic tissue or islets from the donor programmes were only studied if they could not be used for clinical purposes (pancreas or islet transplantation), and if research consent had been obtained. Tissue samples from the pancreatic tail region were fixed overnight in 4% formaldehyde and embedded in wax. For quantitative light microscopy, analyses were performed on 4-μm thick sections of pancreas. For electron microscopy, human isolated islets were fixed in 2.5% glutaraldehyde and embedded in resin. Human islet isolations were performed in the Good Manufacturing Practice facility in Leiden and in the Diabetes Research & Wellness Foundation Human Islet Isolation Facility in Oxford.
Pancreatic tissue from non-human primates
Macaca mulatta and Macaca fascicularis, non-human primates, were selected from a larger colony that has been extensively characterised for features of diabetes and metabolism (Hansen & Bodkin 1986). The animals were maintained in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (USA).
Animals for this study were divided into three groups based on the metabolic profile at killing, being normoglycaemic and normoinsulinaemic, normoglycaemic and hyperinsulinaemic (fasting immunoreactive insulin (IRI) >600 pmol/L) or overtly diabetic (fasting plasma glucose (FPG) >7.8 mmol/L) (Supplementary Table 2). Samples of the pancreatic tail were taken during killing, fixed in 4% formaldehyde and processed for wax embedding; 4-μm sections were cut for analysis.
Immunofluorescent labelling and quantification
Labelling of tissue sections and cell quantification were performed independently in Leiden and Oxford laboratories on different materials using similar but not identical methodologies. Tissue sections were rehydrated, and antigen retrieval was performed by heating slides in 0.01 M citrate buffer (pH = 6.0). Primary antibodies against insulin, glucagon, vimentin, Nkx6.1, Pdx1, N-cadherin, smooth muscle actin and active caspase 3 were used. Fluorescent secondary antibodies were chosen to allow three antigens to be co-visualised (Supplementary Table 3 for more information on antibody sources and methods). The tyramide amplification system (Invitrogen, Thermo Fisher) was used as a secondary antibody system to improve antigen visualisation in Oxford. Sections were mounted in antifade reagent (VECTASHIELD, Vector Laboratories, Burlingame, CA, USA).
Sections were imaged using LSM 7 MP (Zeiss) or Biorad, Radiance 2100 (Biorad) confocal microscopes with pinhole set at 1 Airy unit. Images of at least 25 islets per post-mortem specimen were quantified; the labelled pancreatic section was scanned in sequential but not overlapping fields and images of all islets in the section were collected for analysis. The frequency of all double- (vimentin plus one other label) and triple-positive (vimentin plus two other labels) cells was quantified by semi-automatic image analysis systems (AxioVision, Zeiss) and Zen, Zeiss and ImageJ, and expressed as percentages of the total number of insulin- or glucagon-positive cells examined/subject. Cells that expressed a clear nuclear Pdx-1 or Nkx6.1 staining were considered as Pdx-1 or Nkx6.1 positive. To establish that two/three antigens were located in the same cell, each colour image was examined separately. Importantly, the investigators were blind to the presence of diabetes or other clinical features during quantification.
The association of vimentin positivity (vim+) with β- to α-cell conversion was assessed by the examination of cells expressing both insulin and glucagon (vim+ins+glu+). Vim+ cell proportion was examined in glu+ or ins+ cells, in relation to diabetic status, age, gender, BMI and in organ donors, the status of the donor and duration of cold ischaemic time of the organ donation.
To examine the relationship of vim+ cells with islet amyloid deposits in pancreatic tissue samples of patients with T2DM, sections were stained for amyloid with thioflavin s (Sigma Aldrich) before undertaking the immunolabelling. The proportion of vim+ cells was assessed in each islet noting the presence of thioflavin s–stained amyloid. In addition, the vim+/endocrine-positive cell proportion for each subject (and each animal) was examined in relation to the total pancreatic amyloid prevalence (percentage of islets containing amyloid throughout the pancreas).
Electron microscopy
Isolated islets from human donors with (n = 2) and without (n = 3) T2DM (Supplementary Table 4) were contrast enhanced with 2% uranyl acetate (Agar Scientific, Stanstead, UK), dehydrated and embedded in London Resin Gold (LRG) (Agar Scientific) without osmium post-fixation to preserve antigenicity. Sections were immunolabelled with antibodies to vimentin (Dako) or the lysosomal marker cathepsin D (Sigma) using biotin-conjugated secondary antibodies (Dako) and streptavidin gold (vimentin) or protein A gold conjugates (BioCell, Cardiff, UK). Sections were contrasted with 2% uranyl acetate (Agar) and lead citrate and were examined in a Joel 1010 electron microscope fitted with a digital camera (Gatan, CA, USA).
Flow cytometry
Isolated islet preparations (purity >60%) from 4 donors with T2DM (Supplementary Table 4) were dispersed into single cells using a 0.025% trypsin solution containing 10 µg/mL DNase (Pulmozyme, Genentech) at 37°C. Cells were fixed with 4% formaldehyde (Klinipath BV, The Netherlands) for 15 min at RT, and 0.1% saponin (Sigma) was added to permeabilise the cells for an additional 30 min. Primary antibodies against insulin (1:200; Santa Cruz Technologies) and vimentin (1:100; Dako) were used and incubated for 20 min at 4°C. Anti-mouse and anti-rabbit Alexa Fluor-488 and Alexa Fluor-647 (1:500, Molecular Probes) were used as secondary antibodies. Cells were analysed with LSRII Flow Cytometer (BD Biosciences).
Statistical analysis
All data are expressed as median and interquartile range (IQR), unless stated otherwise. Statistical significance of difference between groups was assessed by Mann–Whitney U test, or Kruskal–Wallis and Dunn’s Multiple Comparison Test when more than two groups were compared. P < 0.05 was considered to be statistically significant.
Results
General characteristics of the ND and T2D human tissue samples
Human pancreatic tissue samples were obtained from post-mortem subjects, including organ donors, diagnosed with T2DM (n = 28; Oxford n = 17, Leiden n = 11) or without diabetes (ND) (n = 38; Oxford n = 30, Leiden n = 8) (Supplementary Table 1). The characteristics of the subjects were not different in the two centres (Supplementary Table 1); T2DM group; age range 34–86 years, BMI 22–54 kg/m2, 9 females; ND group; age 27–78 years, BMI 19–38 kg/m2, 19 females. The median age was significantly less in the ND group compared to that in the T2DM group (56.0 vs 61.5 years, ND vs T2DM, P < 0.05), but the BMIs were similar between groups. The duration of diabetes ranged between 1 and 28 years and age of onset between 30 and 76 years, median 57.5 years. There were more organ donations from donation after brain death (DBD) than donation after circulatory death (DCD) in the ND group than that in the T2DM group (ND: DBD 23 and DCD 18; T2DM: DBD 13 and DCD 14).
Expression of vimentin is increased in α- and β-cells in T2DM
Confocal imaging showed that vimentin (vim+) is expressed in a subset of insulin-positive (ins+) and glucagon-positive (glu+) cells. Vimentin positivity was usually in small patches in the cytoplasm of the endocrine cells and rarely spread throughout the cell (Fig. 1A, B, C and D). Investigations were performed independently in Leiden and Oxford on different materials using slightly different methods; the characteristics of the patients and the data derived were not different and therefore the data sets were merged for analyses. The proportion of ins+ cells coexpressing vimentin (vim+ins+) was significantly higher in the T2DM group compared to that in the ND group (1.43 (0.74–2.32)% (median (IQR)) vs 0.98 (0.29–1.42)% respectively P < 0.05, Fig. 1E). The proportion of islets containing at least one vim+ins+ cell in a given subject was variable and was not different between ND and T2DM groups (28.0 (0–73.3)% (median (IQR)) vs 33.3 (4.0–93.3)% respectively).
The mesenchymal protein vimentin is expressed in a subset of human α- and β-cells in T2DM. (A, B) Islet from non-diabetic (ND) subject labelled for vimentin (white) and (A) insulin (red) and (B) glucagon (green). Insets show cells double labelled for the islet peptide and vimentin. (C, D) Islet from subject with type 2 diabetes (T2DM) labelled for vimentin (white) and (C) insulin (red) and (D) glucagon (green). Insets show cells double labelled for the islet peptide and vimentin. Scale bars = 50 µm. (E, F) Quantitative data of the frequency of vimentin+insulin+ cells (E) and vimentin+glucagon+ cells/subject (F) in pancreatic sections from subjects without (ND) and with T2DM. Graphs represent individual values (as scatter plots) and median values ± interquartile range (as box plots, *P < 0.05). Vimentin+glucagon+ cells were more frequent than vimentin+glucagon+ cells in both ND and T2DM subjects (note the differences in axis values).
Citation: Journal of Endocrinology 233, 3; 10.1530/JOE-16-0588
The frequency of glu+ cells coexpressing vimentin (vim+glu+) cells was also significantly higher in the T2DM than that in the ND group (4.53 (2.78–9.11)% vs 2.53 (1.39–5.71)% respectively P < 0.05, Fig. 1F). Moreover, the proportion of vim+glu+ cells was significantly higher than that in the number of vim+ins+ cells in both the ND and T2DM groups. The proportion of islets containing at least one glu+vim+ cell in a given subject was also highly variable but significantly higher in the T2DM group than that in the ND group (53.8 (9.5–93.3)% (median (IQR)) vs 72.0 (24.0–100)% total islets respectively, P < 0.05). In the whole cohort, no significant correlation was found between the proportion of either vim+ins+ or vim+glu+ cells and donor age, BMI, age of onset or duration of T2DM, the donor organ procurement technique (DBD or DCD and duration of ischaemic time) or gender in either the ND or T2DM groups (data not shown).
To demonstrate that this double labelling was not an artefact of post-mortem pancreatic material embedded in wax, the presence of vim+ins+ cells in human islets from 4 donors with T2D (n = 4) was confirmed by flow cytometry of human dispersed islet cells labelled for each peptide in an islet collagenase digest. A small population of vim+ins+ cells was identified in each preparation in the total population of cells labelled for insulin or vimentin (Supplementary Fig. 1A, B, C, D, E and F).
Ultrastructural identification of vimentin in human α- and β-cells
To determine the ultrastructural location of the vim+ labelling, human isolated islets were examined by electron microscopy. Immunogold labelling for vimentin was present in α- and β-cells in human isolated islets from donors with type 2 diabetes. No labelling for vimentin was found in islets from ND subjects (Fig. 2A and B). In addition, gold labelling for vimentin was present on intermediate filaments in residual islet capillary endothelial cells and islet perivascular macrophages (mesenchymal cells) in islets from ND and T2DM subjects as expected, confirming the antibody specificity (Supplementary Fig. 2A). The intermediate filaments were in bundles or small patches in the cytoplasm in islets isolated from both subjects with T2DM (Fig. 2C). In β-cells in T2DM, cytoplasmic vim+ labelling was less frequently observed than in α-cells and was located primarily in lysosomes (Fig. 2D). Primary lysosomes were identified by labelling with antibodies to the lysosomal enzyme cathepsin D (Supplementary Fig. 2B). The morphology of endocrine cells containing cytoplasmic vim+ intermediate filaments or vim+ lysosomes was not different from vim− cells in terms of the secretory granule population, apoptotic nuclei or swollen ER.
Ultrastructural identification of vimentin in human islets. Immunogold labelling for vimentin in islet from ND subject (A, B) and a subject with T2DM (C, D). Labelling was not detected in either α-cells (A) or β-cells (B) in islets from ND subjects although labelling was present in perivascular fibroblasts in all islets. (C) Vimentin was present as patches of cytoplasmic intermediate filaments (arrows) in α-cells in islets isolated from patients with T2DM. (D) In β-cells of islets from a diabetic subject, vimentin labelling was present in primary lysosomes (lys) (arrows) and infrequently in cytoplasmic intermediate filaments. g, glucagon granules; i, insulin granules; lb, lipofuscin body; m, mitochondrion; n, nucleus. Scale bars 500 nm.
Citation: Journal of Endocrinology 233, 3; 10.1530/JOE-16-0588
Vimentin-positive islet cells are not apoptotic and rarely express β-cell transcription factors
Immunofluorescence for active caspase 3 identified only rare islet cells undergoing apoptosis, and none of them expressed vimentin. This suggests that vim+ins+ and vim+glu+ cells are not undergoing apoptosis (Supplementary Fig. 3A, B, C and D). This was confirmed by electron microscopy where no vim+ cells had apoptotic nuclei.
Most ins+ cells (~70%) in both ND and T2DM groups were positive for the β-cell transcription factor Nkx6.1 in nuclei (Fig. 3A, white arrow head; Fig. 3B, left panel). However, cells coexpressing vimentin (vim+ins+ cells) rarely (<10%) expressed Nkx6.1 (Fig. 3A, white arrow, Fig. 3B, right panel, Supplementary Fig. 4A). Similarly, the transcription factor Pdx1 was identified in the nucleus of most ins+ cells in ND and T2DM islets. However, ins+ cells coexpressing vimentin and Pdx1 were very rare in both ND and T2DM (<0.1% of ins+ cells) (Fig. 3C, Supplementary Fig. 4B). In our previous study using many of the same samples of T2DM and ND subjects (Spijker et al. 2015), we were unable to find cells positive for the transcription factor neurogenin 3, which is associated with primordial β-cells in foetal pancreatic development and dedifferentiation (Rukstalis & Habener 2009). The mesenchymal protein, smooth muscle actin (SMA), was present in islets but not co-localised with vimentin in islet cells. However, the mesenchymal matrix protein, N-cadherin, was present in the extracellular matrix in islets and was expressed in vim+ins-negative cells as well as in vim+ ins-positive cells in islets (Supplementary Fig. 3E, F, G and H).
Vimentin+insulin+ cells rarely express the key β-cell transcription factor Nkx6.1. (A) Islet labelled for vimentin (white), insulin (red) and Nkx6.1 (green). Insets show a cell double labelled for insulin and vimentin (arrow), as well as an insulin+Nkx6.1+ (white arrow head) and insulin+Nkx6.1− cell (yellow arrow head). Scale bar 20 µm. (B) Quantification data on samples from the Leiden cohort. The majority of insulin+ cells express Nkx6.1 (B, left panel), whereas most vimentin+insulin+ cells were Nkx6.1 negative (B, right panel). (C) Islet labelled for vimentin (white), insulin (red) and Pdx1 (green). Most vimentin+insulin+ cells were Pdx1 negative. Scale bar 50 µm. Insets show a cell double labelled for insulin and vimentin. (D) Islet from a subject with T2DM labelled for vimentin (green), insulin (blue) and glucagon (red). Vimentin-positive cells (arrows) were positive for glucagon but not insulin. Scale bar 20 µm.
Citation: Journal of Endocrinology 233, 3; 10.1530/JOE-16-0588
No correlation between frequency of vimentin-positive cells and bihormonal endocrine cells or islet amyloid deposition in human T2DM
To determine whether the presence of vimentin in islet cells was associated with bihormonality, (insulin and glucagon granules in the same cell), vim+ins+glu+ (i.e. vimentin+bihormonal) cells were counted in eight subjects from the Oxford cohort (ND n = 3; T2DM n = 5). The frequency of ins+glu+ cells was, as expected, very low and variable (range 0.4–5% median 0.6%, of total insulin-positive cells (all subjects)). Importantly, of these bihormonal cells, less than 15% also expressed vimentin (range 0–14.3%, median 0.0% of ins+glu+ cells) (Fig. 3D). Finally, no correlation was found between vim+ins+ or vim+glu+ cells and bihormonal cell proportions in any subject. Altogether, these data suggest that vimentin expression (potential EMT) and dual expression of insulin and glucagon are independent processes.
Islet amyloid deposition in T2DM disrupts the islet architecture. We hypothesised that this could affect the extracellular matrix adjacent to endocrine cells and thereby trigger EMT. The proportion of vim+ cells associated with amyloid was assessed. There was no relationship between the prevalence of islet amyloid (determined as the proportion of islets affected throughout the pancreas (range 0–90%) and the proportions of vim+ins+ or vim+glu+ cells in eight donors with T2DM) (data not shown). In a set of 6 donors with T2DM possessing islet amyloidosis, the proportion of vim+ins+ or vim+glu+ cells was not significantly different in the amyloid-free compared to the amyloid-containing islets (amyloid-free islets: vim+ins+%ins, 1.14 ± 0.69%; vim+glu+%glu, 2.18 ± 1.08%; amyloid-containing islets: vim+ins+%ins, 0.45 ± 0.14%, vim+glu+%glu, 3.37 ± 2.274%, mean ± s.e.m.) (Supplementary Fig. 3I, J, K and L).
Frequency of vimentin-positive endocrine cells isincreased in diabetic but not in hyperinsulinaemic non-human primates
To determine whether the expression of vimentin in islet cells occurs before the onset of diabetes and could be an aetiological factor for islet dysfunction, pancreatic tissue specimens from primates were studied. These animals develop a syndrome of diabetes with characteristics similar to T2DM in humans, and the natural history of diabetes has been described (Hansen & Bodkin 1986, Guo et al. 2013).
Animal tissue of normoglycaemic and normoinsulinaemic (ND; n = 4), normoglycaemic and hyperinsulinaemic (HI; n = 4) and overtly diabetic (DM; n = 8) monkeys was analysed. The average age (17.1 ± 14.5 vs 22.8 ± 5.8 vs 21.8 ± 4.3, ND vs HI vs DM), weight (7.7 ± 3.0 vs 12.3 ± 3.7 vs 10.7 ± 2.1) and percentage of body fat (20.0 ± 2.8 vs 35.1 ± 5.7 vs 31.5 ± 11.1) were similar between the three groups (Supplementary Table 2).
The vim+ins+ cell proportion was larger in DM animals compared to that in the ND group (6.85 (5.03–17.45)% (median (IQR)) vs 0.50 (0.10–0.75)%, respectively, P < 0.05) (Fig. 4A, C and E). The frequency of these cells in HI animals, which are considered to be prediabetic, was not different from the ND group (0.40 (0.33–1.7)% (median (IQR)) vs 0.50 (0.10–0.75)%) (Fig. 4E). A trend towards an increased %vim+glu+ cells, although not statistically significant, was observed in the DM compared to that in ND and HI groups (Fig. 4B, D and F). In addition, we reported previously that the frequency of bihormonal cells (ins+glu+) was increased in DM compared to that in ND and HI animals (Spijker et al. 2015). Importantly, as observed for the human samples, the large majority of these bihormonal cells (all animals) did not coexpress vimentin (vim+/ins+/glu+ve, median <7% ins+/glu+ve), and the frequency of bihormonal cells expressing vimentin in DM was not different from ND or HI groups (data not shown). Finally, no association was found between the presence of amyloidosis and the frequency of vim+ins+ or vim+glu+ cells in the HI and DM groups (data not shown).
Vimentin expression in β-cells is higher in diabetic non-human primates. (A, B) Islets from non-diabetic non-human primates (ND) labelled for vimentin (white) and (A) insulin (red) and (B) glucagon (green). Insets show cells double labelled for the islet peptide and vimentin. (C, D) Islets from diabetic animals (DM) labelled for vimentin (white) and (C) insulin (red) and (D) glucagon (green). The islets from diabetic animals usually contained islet amyloid, as seen in (C) as an area devoid of cells/nuclei. Insets show cells double labelled for the islet peptide and vimentin. Scale bars = 50 µm. (E, F) Quantitative data of the frequency of vimentin+insulin+ cells (E) and vimentin+glucagon+ cells/subject (F) in pancreatic sections from non-diabetic (ND), hyperinsulinaemic (HI) and diabetic (DM) animals. Graphs represent individual values (as scatter plots) and median values ± interquartile range (as box plots, *P < 0.05).
Citation: Journal of Endocrinology 233, 3; 10.1530/JOE-16-0588
Discussion
The main findings of our observational study are that in diabetes there is an increased proportion of vimentin-positive α- and β-cells in pancreatic islets. Increased vimentin-positive islet cell proportions was associated with diabetes rather than with prediabetic conditions (obesity or insulin resistance) as shown in a cohort of spontaneously diabetic non-human primates. Vimentin positivity was not associated with islet amyloidosis or with cells expressing both insulin and glucagon.
The presence of vim+ islet cells in adult human histological specimens has been described in a few isolated cases (Gershengorn et al. 2004, Fanjul et al. 2010) including a recent examination of 3 subjects where it was concluded that vimentin islet cell positivity was absent in non-diabetic subjects and related to early-onset diabetes and metabolic stress (White et al. 2013). Our two-centred study, including data from 28 patients with T2DM and 38 normoglycaemic subjects of similar ages and BMI showed vim+ islet cells in ND subjects and increased vim+ endocrine cells to be associated with diabetes but with no correlation with age of onset or duration of diabetes. Human post-mortem or donor tissue has some degree of hypoxia, which could affect some islet features (Sullivan et al. 2015). However, there were no correlations of vim+ α- or β-cell proportions with any feature of tissue collection, age or BMI in the non-diabetic subjects. This suggests that the higher proportion of vim+ cells was associated only with diabetes. This was confirmed by data from the metabolically well-characterised non-human primates; the frequency of insulin+vimentin+ cells was increased in diabetic but not in hyperinsulinaemic (prediabetic) animals. Thus, we suggest that the elevated expression of vimentin is associated with, rather than a cause of diabetes. Recently, a single-cell transcriptome atlas of the human pancreas was generated by our group and others (Muraro et al. 2016, Segerstolpe et al. 2016). Segerstolpe and coworkers also examined a few cases of donors with T2DM (Segerstolpe et al. 2016). From their online database, it is evident that although expression is very variable; overall there was an increased expression of vimentin in donors with T2DM (Supplementary Fig. 5). This confirms our findings using an antibody-independent technique.
The presence of vimentin in endocrine cells could represent an attempt to regenerate islet cells, either from mesenchymal cells or pericytes resident in islets (Carlotti et al. 2010, 2011) or from duct cells through an EMT stage, as proposed by Collombat and coworkers (Al-Hasani et al. 2013, Pfeifer et al. 2013). However, in this case, insulin expression would be preceded by or accompanied with Nkx6.1 and Pdx1 expression (Russ et al. 2011, Pagliuca et al. 2014, Rezania et al. 2014). Thus, the lack of β-cell-specific transcription factors in vim+ins+ cells suggests that these cells are losing rather than acquiring their endocrine phenotype. This is in line with the findings from a transgenic mouse study that associated expression of vimentin expression and β-cell dedifferentiation in diabetes (Talchai et al. 2012). β-cell dedifferentiation has been associated with expression of neurogenin 3 (Ngn3) (a marker for β-cell precursor cells) in a number of mouse models (Talchai et al. 2012, Wang et al. 2014). We did not find any Ngn3 positivity in adult human islets although cells were positive in foetal tissue (Spijker et al. 2015). However, both the downregulation of key transcription factors (Hay & Docherty 2006, Butler et al. 2012, Szabat et al. 2012, Guo et al. 2013, Schaffer et al. 2013, Gao et al. 2014) and the expression of vimentin in islet cells (Gershengorn et al. 2004, Russ et al. 2009) have been associated with β-cell dysfunction and/or diabetes in several studies. Altogether, we propose that the presence of vimentin in endocrine cells is likely to represent a dedifferentiation stage that could contribute to islet cell dysfunction in T2DM.
Vimentin+ cell proportion was significantly higher (×4) in α-cells than that in β-cells. It has been postulated that the potential to alter identity is greater in α-cells than that in β-cells (Bramswig et al. 2013, van der Meulen & Huising 2015). The influence of aberrant expression of transcription factors in α-cells has been less extensively studied than that for β-cells, but T2DM is associated with dysfunctional stimulus secretion coupling of glucagon (Zhang et al. 2013). Whether these vim+ endocrine cells undergo the full transition to mesenchymal cells by EMT cannot be determined in this snapshot of cells in vivo. Mesenchymal cells typically express not only vimentin but also smooth muscle actin and N-cadherin. However, vim+ endocrine cells in islets did not express smooth muscle actin, and N-cadherin was present in insulin-positive cells in the absence of vimentin (Parnaud et al. 2011). Our data suggest that epithelial endocrine cells can express the so-called specific mesenchymal proteins (Ye & Weinberg 2015) and that if vimentin expression is a component of EMT, the process may be only partial and possibly reversible as in some cancer cells (Nieto 2013).
We previously identified that bihormonal cells were increased in T2DM (Spijker et al. 2015). In the current study we show, both in humans and in non-human primates, that bihormonal cells were not associated with vimentin positivity and that the infrequent bihormonal cells expressing vimentin were not present in higher proportions in diabetic vs non-diabetic groups. Altogether, these data indicate that vimentin positivity and bihormonality are two independent processes. Whether the higher number of vim+glu+ cells could represent previous β-cells that have transdifferentiated cannot be excluded from this study.
Conversion of β-cells towards mesenchymal cells in vitro is induced by changes in the extracellular matrix induced by isolation and monolayer culture (Gershengorn et al. 2004, Russ et al. 2009). However, extracellular matrix changes induced by islet amyloid formation in vivo in T2DM were not associated with significantly increased vim+ islet cells. Therefore, it appears that human islet cells can undergo several different forms of identity changes in T2DM in vivo. Functional studies are not possible on post-mortem fixed samples. Studies on human isolated islets in vitro can act as surrogates for some in vivo observations. However, the artificial environment including the process of isolation and culture (particularly in relation to EMT) precludes direct translation of experiments in vitro to the in vivo situation (Gershengorn et al. 2004, Russ et al. 2009, Negi et al. 2012).
To conclude, we propose that vimentin expression in islet cells is increased in T2DM. If this aberrant protein expression results in irreversible EMT, the potential loss of islet cell identity could contribute to the loss of islet cell function that occurs in diabetes. It is apparent that the cellular phenotype of mature islet cells can be challenged by several different mechanisms many of which are increased in type 2 diabetes.
Supplementary data
This is linked to the online version of the paper at http://dx.doi.org/10.1530/JOE-16-0588.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
Funding
Funding for the Leiden Laboratory was obtained from the Diabetes Cell Therapy Initiative including the Dutch Diabetes Research Foundation, DON Foundation, JDRF, and Bontius Foundation. H W was funded by a Clarendon Scholarship (University of Oxford). M F B holds a Wellcome Trust OXION Training Fellowship.
Acknowledgements
The authors are grateful to Stephen Hughes of the Diabetes and Wellness Foundation Human Islet Isolation facility in OCDEM, Oxford for provision of tissue. They thank Martijn Van de Bunt (OCDEM) for useful discussions. They thank Shankar Srinivas, Department of Physiology, Anatomy and Genetics, University of Oxford for advice and for use of his microscope and Patrik Rorsman and Stephen Gough (OCDEM) for their support.
References
Al-Hasani K, Pfeifer A, Courtney M, Ben-Othman N, Gjernes E, Vieira A, Druelle N, Avolio F, Ravassard P & Leuckx G et al. 2013 Adult duct-lining cells can reprogram into beta-like cells able to counter repeated cycles of toxin-induced diabetes. Developmental Cell 26 86–100. (doi:10.1016/j.devcel.2013.05.018)
Bramswig NC, Everett LJ, Schug J, Dorrell C, Liu C, Luo Y, Streeter PR, Naji A, Grompe M & Kaestner KH 2013 Epigenomic plasticity enables human pancreatic alpha to beta cell reprogramming. Journal of Clinical Investigation 123 1275–1284. (doi:10.1172/JCI66514)
Brereton MF, Iberl M, Shimomura K, Zhang Q, Adriaenssens AE, Proks P, Spiliotis II, Dace W, Mattis KK & Ramracheya R et al. 2014 Reversible changes in pancreatic islet structure and function produced by elevated blood glucose. Nature Communications 5 4639. (doi:10.1038/ncomms5639)
Butler AE, Robertson RP, Hernandez R, Matveyenko AV, Gurlo T & Butler PC 2012 Beta cell nuclear musculoaponeurotic fibrosarcoma oncogene family A (MafA) is deficient in type 2 diabetes. Diabetologia 55 2985–2988. (doi:10.1007/s00125-012-2666-2)
Butler AE, Dhawan S, Hoang J, Cory M, Zeng K, Fritsch H, Meier JJ, Rizza RA & Butler PC 2016 beta-cell deficit in obese type 2 diabetes, a minor role of beta-cell dedifferentiation and degranulation. Journal of Clinical Endocrinology and Metabolism 101 523–532. (doi:10.1210/jc.2015-3566)
Carlotti F, Zaldumbide A, Loomans CJ, van Rossenberg E, Engelse M, de Koning EJ & Hoeben RC 2010 Isolated human islets contain a distinct population of mesenchymal stem cells. Islets 2 164–173. (doi:10.4161/isl.2.3.11449)
Carlotti F, Zaldumbide A, Ellenbroek JH, Spijker HS, Hoeben RC & de Koning EJ 2011 beta-cell generation: can rodent studies be translated to humans? Journal of Transplantation 2011 892453. (doi:10.1155/2011/892453)
Cinti F, Bouchi R, Kim-Muller JY, Ohmura Y, Sandoval PR, Masini M, Marselli L, Suleiman M, Ratner LE & Marchetti P et al. 2016 Evidence of beta-cell dedifferentiation in human type 2 diabetes. Journal of Clinical Endocrinology and Metabolism 101 1044–1054. (doi:10.1210/jc.2015-2860)
Courtney M, Gjernes E, Druelle N, Ravaud C, Vieira A, Ben-Othman N, Pfeifer A, Avolio F, Leuckx G & Lacas-Gervais S et al. 2013 The inactivation of Arx in pancreatic alpha-cells triggers their neogenesis and conversion into functional beta-like cells. PLoS Genetics 9 e1003934. (doi:10.1371/journal.pgen.1003934)
Dhawan S, Georgia S, Tschen SI, Fan G & Bhushan A 2011 Pancreatic beta cell identity is maintained by DNA methylation-mediated repression of Arx. Developmental Cell 20 419–429. (doi:10.1016/j.devcel.2011.03.012)
Fanjul M, Gmyr V, Sengenes C, Ratovo G, Dufresne M, Lefebvre B, Kerr-Conte J & Hollande E 2010 Evidence for epithelial-mesenchymal transition in adult human pancreatic exocrine cells. Journal of Histochemistry and Cytochemistry 58 807–823. (doi:10.1369/jhc.2010.955807)
Gao T, McKenna B, Li C, Reichert M, Nguyen J, Singh T, Yang C, Pannikar A, Doliba N & Zhang T et al. 2014 Pdx1 maintains beta cell identity and function by repressing an alpha cell program. Cell Metabolism 19 259–271. (doi:10.1016/j.cmet.2013.12.002)
Gershengorn MC, Hardikar AA, Wei C, Geras-Raaka E, Marcus-Samuels B & Raaka BM 2004 Epithelial-to-mesenchymal transition generates proliferative human islet precursor cells. Science 306 2261–2264. (doi:10.1126/science.1101968)
Guo S, Dai C, Guo M, Taylor B, Harmon JS, Sander M, Robertson RP, Powers AC & Stein R 2013 Inactivation of specific beta cell transcription factors in type 2 diabetes. Journal of Clinical Investigation 123 3305–3316. (doi:10.1172/JCI65390)
Hansen BC & Bodkin NL 1986 Heterogeneity of insulin responses: phases leading to type 2 (non-insulin-dependent) diabetes mellitus in the rhesus monkey. Diabetologia 29 713–719. (doi:10.1007/BF00870281)
Hay CW & Docherty K 2006 Comparative analysis of insulin gene promoters: implications for diabetes research. Diabetes 55 3201–3213. (doi:10.2337/db06-0788)
Jonas JC, Sharma A, Hasenkamp W, Ilkova H, Patane G, Laybutt R, Bonner-Weir S & Weir GC 1999 Chronic hyperglycemia triggers loss of pancreatic beta cell differentiation in an animal model of diabetes. Journal of Biological Chemistry 274 14112–14121. (doi:10.1074/jbc.274.20.14112)
Kim-Muller JY, Fan J, Kim YJ, Lee SA, Ishida E, Blaner WS & Accili D 2016 Aldehyde dehydrogenase 1a3 defines a subset of failing pancreatic beta cells in diabetic mice. Nature Communications 7 12631. (doi:10.1038/ncomms12631)
Muraro MJ, Dharmadhikari G, Grun D, Groen N, Dielen T, Jansen E, van Gurp L, Engelse MA, Carlotti F & de Koning EJ et al. 2016 A single-cell transcriptome atlas of the human pancreas. Cell Systems 3 385–394. (doi:10.1016/j.cels.2016.09.002)
Negi S, Jetha A, Aikin R, Hasilo C, Sladek R & Paraskevas S 2012 Analysis of beta-cell gene expression reveals inflammatory signaling and evidence of dedifferentiation following human islet isolation and culture. PLoS ONE 7 e30415. (doi:10.1371/journal.pone.0030415)
Nieto MA 2013 Epithelial plasticity: a common theme in embryonic and cancer cells. Science 342 1234850. (doi:10.1126/science.1234850)
Nishimura W, Takahashi S & Yasuda K 2015 MafA is critical for maintenance of the mature beta cell phenotype in mice. Diabetologia 58 566–574. (doi:10.1007/s00125-014-3464-9)
Pagliuca FW, Millman JR, Gurtler M, Segel M, Van DA, Ryu JH, Peterson QP, Greiner D & Melton DA 2014 Generation of functional human pancreatic beta cells in vitro. Cell 159 428–439. (doi:10.1016/j.cell.2014.09.040)
Papizan JB, Singer RA, Tschen SI, Dhawan S, Friel JM, Hipkens SB, Magnuson MA, Bhushan A & Sussel L 2011 Nkx2.2 repressor complex regulates islet beta-cell specification and prevents beta-to-alpha-cell reprogramming. Genes and Development 25 2291–2305. (doi:10.1101/gad.173039.111)
Parnaud G, Gonelle-Gispert C, Morel P, Giovannoni L, Muller YD, Meier R, Borot S, Berney T & Bosco D 2011 Cadherin engagement protects human beta-cells from apoptosis. Endocrinology 152 4601–4609. (doi:10.1210/en.2011-1286)
Pfeifer A, Courtney M, Ben-Othman N, Al-Hasani K, Gjernes E, Vieira A, Druelle N, Avolio F, Faurite B & Mansouri A et al. 2013 Induction of multiple cycles of pancreatic beta-cell replacement. Cell Cycle 12 3243–3244. (doi:10.4161/cc.26357)
Piran R, Lee SH, Li CR, Charbono A, Bradley LM & Levine F 2014 Pharmacological induction of pancreatic islet cell transdifferentiation: relevance to type I diabetes. Cell Death and Disease 5 e1357. (doi:10.1038/cddis.2014.311)
Rezania A, Bruin JE, Arora P, Rubin A, Batushansky I, Asadi A, O'Dwyer S, Quiskamp N, Mojibian M & Albrecht T et al. 2014 Reversal of diabetes with insulin-producing cells derived in vitro from human pluripotent stem cells. Nature Biotechnology 32 1121–1133. (doi:10.1038/nbt.3033)
Rukstalis JM & Habener JF 2009 Neurogenin3: a master regulator of pancreatic islet differentiation and regeneration. Islets 1 177–184. (doi:10.4161/isl.1.3.9877)
Russ HA, Ravassard P, Kerr-Conte J, Pattou F & Efrat S 2009 Epithelial-mesenchymal transition in cells expanded in vitro from lineage-traced adult human pancreatic beta cells. PLoS ONE 4 e6417. (doi:10.1371/journal.pone.0006417)
Russ HA, Sintov E, Anker-Kitai L, Friedman O, Lenz A, Toren G, Farhy C, Pasmanik-Chor M, Oron-Karni V & Ravassard P et al. 2011 Insulin-producing cells generated from dedifferentiated human pancreatic beta cells expanded in vitro. PLoS ONE 6 e25566. (doi:10.1371/journal.pone.0025566)
Schaffer AE, Taylor BL, Benthuysen JR, Liu J, Thorel F, Yuan W, Jiao Y, Kaestner KH, Herrera PL & Magnuson MA et al. 2013 Nkx6.1 controls a gene regulatory network required for establishing and maintaining pancreatic Beta cell identity. PLoS Genetics 9 e1003274. (doi:10.1371/journal.pgen.1003274)
Segerstolpe A, Palasantza A, Eliasson P, Andersson EM, Andreasson AC, Sun X, Picelli S, Sabirsh A, Clausen M & Bjursell MK et al. 2016 Single-cell transcriptome profiling of human pancreatic islets in health and type 2 diabetes. Cell Metabolism 24 593–607. (doi:10.1016/j.cmet.2016.08.020)
Spijker HS, Ravelli RB, Mommaas-Kienhuis AM, van Apeldoorn AA, Engelse MA, Zaldumbide A, Bonner-Weir S, Rabelink TJ, Hoeben RC & Clevers H et al. 2013 Conversion of mature human beta-cells into glucagon-producing alpha-cells. Diabetes 62 2471–2480. (doi:10.2337/db12-1001)
Spijker HS, Song H, Ellenbroek JH, Roefs MM, Engelse MA, Bos E, Koster AJ, Rabelink TJ, Hansen BC & Clark A et al. 2015 Loss of beta-cell identity occurs in type 2 diabetes and is associated with islet amyloid deposits. Diabetes 64 2928–2938. (doi:10.2337/db14-1752)
Sullivan BA, Hollister-Lock J, Bonner-Weir S & Weir GC 2015 Reduced Ki67 staining in the postmortem state calls into question past conclusions about the lack of turnover of adult human beta-cells. Diabetes 64 1698–1702. (doi:10.2337/db14-1675)
Szabat M, Lynn FC, Hoffman BG, Kieffer TJ, Allan DW & Johnson JD 2012 Maintenance of beta-cell maturity and plasticity in the adult pancreas: developmental biology concepts in adult physiology. Diabetes 61 1365–1371. (doi:10.2337/db11-1361)
Talchai C, Xuan S, Lin HV, Sussel L & Accili D 2012 Pancreatic beta cell dedifferentiation as a mechanism of diabetic beta cell failure. Cell 150 1223–1234. (doi:10.1016/j.cell.2012.07.029)
Thorel F, Nepote V, Avril I, Kohno K, Desgraz R, Chera S & Herrera PL 2010 Conversion of adult pancreatic alpha-cells to beta-cells after extreme beta-cell loss. Nature 464 1149–1154. (doi:10.1038/nature08894)
van der Meulen T & Huising MO 2015 Role of transcription factors in the transdifferentiation of pancreatic islet cells. Journal of Molecular Endocrinology 54 R103–R117. (doi:10.1530/JME-14-0290)
Wang Z, York NW, Nichols CG & Remedi MS 2014 Pancreatic beta cell dedifferentiation in diabetes and redifferentiation following insulin therapy. Cell Metabolism 19 872–882. (doi:10.1016/j.cmet.2014.03.010)
White MG, Marshall HL, Rigby R, Huang GC, Amer A, Booth T, White S & Shaw JA 2013 Expression of mesenchymal and alpha-cell phenotypic markers in islet beta-cells in recently diagnosed diabetes. Diabetes Care 36 3818–3820. (doi:10.2337/dc13-0705)
Yang YP, Thorel F, Boyer DF, Herrera PL & Wright CV 2011 Context-specific alpha- to-beta-cell reprogramming by forced Pdx1 expression. Genes and Development 25 1680–1685. (doi:10.1101/gad.16875711)
Ye X & Weinberg RA 2015 Epithelial-mesenchymal plasticity: a central regulator of cancer progression. Trends in Cell Biology 25 675–686. (doi:10.1016/j.tcb.2015.07.012)
Zhang Q, Ramracheya R, Lahmann C, Tarasov A, Bengtsson M, Braha O, Braun M, Brereton M, Collins S & Galvanovskis J et al. 2013 Role of KATP channels in glucose-regulated glucagon secretion and impaired counterregulation in type 2 diabetes. Cell Metabolism 18 871–882. (doi:10.1016/j.cmet.2013.10.014)
