Author contributions: C.C.: conception and design, collection of data, data analysis and interpretation, and manuscript writing; A.J.K. and P.C.: provision of study material and final approval of manuscript; P.D.: collection of data; N.N.: provision of study material and final approval of manuscript; C.P.W.-J. and P.J.: conception and design, data interpretation, and manuscript writing. C.P.W.J. and P.M.J. contributed equally to this article.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS December 19, 2012.
Following islet transplantation, islet graft revascularization is compromised due to loss of endothelial cells (ECs) during islet culture. TGF-β signaling pathways are essential for vascular homeostasis but their importance for islet EC function is unclear. We have identified a population of multipotent mesenchymal stromal cells (MSCs) within islets and investigated how modulation of TGF-β signaling by these cells influences islet EC viability. Cultured islets exhibited reduced expression of EC markers (VEGFR2, VE-cadherin and CD31), which was associated with diminished but sustained expression of endoglin a marker of both ECs and MSCs. Double fluorescent labeling of islets in situ with the EC marker CD31 disclosed a population of CD31-negative cells which were positive for endoglin. In vitro coculture of microvascular ECs with endoglin-positive, CD31-negative islet MSCs reduced VEGFR2 protein expression, disrupted EC angiogenic behavior, and increased EC detachment. Medium conditioned by islet MSCs significantly decreased EC viability and increased EC caspase 3/7 activity. EC:MSC cocultures showed enhanced Smad2 phosphorylation consistent with altered ALK5 signaling. Pharmacological inhibition of ALK5 activity with SB431542 (SB) improved EC survival upon contact with MSCs, and SB-treated cultured islets retained EC marker expression and sensitivity to exogenous VEGF164. Thus, endoglin-expressing islet MSCs influence EC ALK5 signaling in vitro, which decreases EC viability, and changes in ALK5 activity in whole cultured islets contribute to islet EC loss. Modifying TGF-β signaling may enable maintenance of islet ECs during islet isolation and thus improve islet graft revascularization post-transplantation. STEM CELLS2013;31:547–559
Pancreatic islets are highly vascularized to enable precise sensing of blood glucose levels and rapid release of insulin into the circulation, and to provide the islet microenvironment with vasoactive mediators, growth factors, and cytokines (reviewed in ). Islet endothelial cells (ECs) have a clear paracrine role in maintenance of normal islet function [2, 3] but the mechanisms regulating adult islet ECs are poorly understood. Currently, a confounding factor for islet transplantation therapy is the loss of islet function during the early post-transplantation period, because of deleterious responses of the avascular transplanted islets to a hypoxic environment, which has detrimental effects on the outcome of individual grafts due to slow graft revascularization (7–14 days) from the donor tissue. Enhancing the rate at which islet grafts establish an adequate blood supply should improve the immediate function of the graft material. Animal models have demonstrated that the new blood vessels in engrafted islets are chimeric structures, containing both donor and host ECs [4, 5]. During islet isolation and culture, ECs are thought to die or differentiate, and it has been suggested that improvement in transplantation outcomes using freshly isolated versus cultured islets is a consequence of improved graft revascularization, reflecting the contribution of donor ECs to neoangiogenesis . The functional β cell mass correlates with islet vascular density [7–9], and EC-derived extracellular matrix proteins directly regulate β cell proliferation and insulin secretion . Thus, in addition to providing sufficient blood flow, an intact vasculature is essential for normal islet function, highlighting the importance of maintaining the viability of the remnant islet EC population prior to transplantation. Identifying mechanisms that promote reduced EC viability during islet culture should therefore allow for positive modulation of donor EC survival and inform procedures to enhance graft revascularization and function post-transplantation.
Transforming growth factor-β (TGF-β) is a multifunctional cytokine which regulates cell proliferation, differentiation, and apoptosis and has been implicated in islet homeostasis. The actions of TGF-β are mediated by an oligomeric receptor complex consisting of two transmembrane serine/threonine-kinases, the type I (TβRI) and type II (TβRII) TGF-β receptors. Within this complex, TβRII phosphorylates and activates TβRI (also known as activin receptor-like kinase 5; ALK5) which mediates signal transduction through a family of Smad proteins . Genetic studies have shown that modifications in TGF-β signaling results in vascular abnormalities (see ). In ECs, ALK5 impairs proliferation through downstream phosphorylation of Smad2/3. ECs also express the TGF-β accessory receptor, endoglin (CD105) and an alternate TβRI, ALK1, coupled to Smad1/5, and engagement of endoglin/ALK1 promotes EC proliferation and drives angiogenesis [13, 14]. In humans, mutations in endoglin and ALK1 cause hereditary hemorrhagic telangiectasia type 1 and 2, respectively, diseases characterized by vascular malformations . Current evidence suggests that a balance between TGF-β/ALK1 and TGF-β/ALK5 signaling pathways is required for normal angiogenesis [16, 17]. The effects of TGF-β on EC function are also highly context-dependent with low concentrations of TGF-β promoting angiogenesis and high TGF-β concentrations driving EC quiescence by inhibiting EC proliferation and migration .
Despite the ability of TGF-β to influence multiple EC functions, the roles of ALK5, ALK1, and endoglin in islet EC function are not defined. Here, we demonstrate that mouse and human islets contain two distinct populations of endoglin-expressing cells, ECs and mesenchymal stromal cells (MSCs), and provide evidence that modulation of TGF-β signaling between these cells by inhibition of ALK5 influences islet EC survival during islet isolation and culture. Modification of this pathway which regulates interactions between MSCs and ECs may offer an accessible point for therapeutic intervention to enhance islet revascularization and thus improve transplantation outcomes.
MATERIALS AND METHODS
Mouse and Human Islet Isolation and Culture
Islets were isolated from 8–10-week-old male ICR mice (Harlan, Bicester, U.K.) as previously described . Each individual experiment used islets combined from four to six mice. Human islets were isolated and cultured as described . Five separate batches of human islets (60%–85% purity) were used. All batches had 90% or greater viability and showed enhanced insulin output in response to elevations in extracellular glucose concentration (data not shown).
Islet MSC Isolation
Human islets were dissociated by incubation at 37°C in 0.02% EDTA solution (Sigma, Dorset, U.K., http://www.sigmaaldrich.com) for 5–10 minutes with pipetting every 2 minutes. Adhering cells were resuspended in MSC maintenance medium, comprising Dulbecco's modified Eagle's medium (DMEM) (Sigma) supplemented with L-glutamine (2 mM), penicillin (100 U/ml), streptomycin (100 μg/ml), and fetal calf serum (FCS) (10%). Cultures were used at passage 3–10.
Osteogenic, Adipogenic, and Chondrogenic Differentiation
MSCs were grown to 80%–90% confluence (∼2 days) in maintenance medium, and differentiation toward osteogenic or adipogenic phenotypes was performed for more than 28 days . Chondrogenic phenotype was induced using micromass cultures before addition of chondrogenic differentiation medium, comprising DMEM supplemented with 10 ng/ml TGF-β1 (Miltenyi, Surrey, U.K., http://www.miltenyibiotec.com), 0.1× ITS premix (Invitrogen, Paisley, U.K., http://www.invitrogen.com), 100 μM ascorbic acid, 1 μM dexamethasone, and 1.25 μg/ml bovine serum albumin (all from Sigma). Cells were maintained in this medium for 7 days, followed by 21 days in medium without FCS.
Human dermal microvascular ECs (HMVECs; Promocell, Heidelberg, Germany, http://www.promocell.com/) were cultured according to the manufacturer's guidelines and used at passage 5.
ECs were labeled with Cell Tracker (Invitrogen) a lysosomal dye, prior to the addition of unlabeled MSCs, as indicated in the manufacturer's guidelines.
MSC and HMVEC Coculture
MSCs and HMVECs were maintained in monolayer coculture and treated as outlined in the figure legends. In some experiments, MSCs and HMVECs were cocultured on Matrigel (BD Biosciences, Oxford, U.K., http://www.bdbiosciences.com) and capillary-like tube formation was monitored as described previously . For monolayer coculture experiments, HMVECs were plated directly onto tissue culture plastic and allowed to adhere overnight. The following day islet MSCs were added at equal density; islet MSCs alone, HMVEC alone, and islet MSC:HMVEC cocultures were then cultured in 50:50 MSC:HMVEC maintenance medium for the duration of the experiment.
Generation of MSC-Conditioned Medium
To avoid any detrimental effects of MSC maintenance medium on EC viability, we used EC growth medium to perform the conditioned medium (CM) experiments. Seven milliliters of EC growth medium (Promocell) was added to confluent dishes of MSCs and culture continued for 72 hours. Following collection, the MSC-CM was stored at −20°C and centrifuged (1,500 rpm) prior to use in EC apoptosis and viability assays.
Immunohistochemistry and Immunoassay
Paraffin-embedded sections of murine pancreas were stained for endoglin using Proteinase K antigen retrieval. For double labeling of endoglin and CD31, cryosections were fixed and incubated with primary antibodies followed by donkey anti-goat Alexa 594 and goat anti-rat Alexa (see Supporting Information Table 1A for antibody supplier details). Islet pellets were immunostained for insulin to assess β-cell area and for CD34 to quantify EC numbers as described previously . Immunostained islet area was assessed on micrographs using ImageJ software, and the islet EC density was assessed by manual counting of CD34-immunostained ECs. Islet insulin content was quantified by radioimmunoassay measurements of acidified ethanol extracts, as described [18, 19].
MSC cultures were resuspended at 105 cells per 50 μl for antibody staining. Cells were incubated at 4°C for 30 minutes with fluorescein isothiocyanate- or phycoerytrin-conjugated antibodies as described in figure legends.
Mouse and human islets (150) were isolated as described previously [18, 19], and RNA was extracted according to manufacturer's guidelines (Qiagen, Crawley, U.K., http://www1. qiagen.com).
Immunoblotting analyses were performed as previously described . Lysate protein levels were measured and blots were incubated with primary antibodies as detailed in Supporting Information Table 1A. Optical densities of bands were quantified using Image J software and normalized to α-tubulin.
Total cell RNA was reverse transcribed using a commercially available kit, according to the manufacturer's instructions (Invitrogen), and standard PCR was undertaken for amplification of mRNAs for a range of genes (see Supporting Information Table 1B for primer sequences).
Microarray analysis performed by the King's College Genomic Centre Services included generation of double-stranded cDNA, preparation and labeling of cRNA, and hybridization to 430 2.0 Mouse Genome Arrays (Affymetrix, Santa Clara, CA, http://www.affymetrix.com). Data were normalized, and genes were identified as differentially expressed if they showed a fold change of at least 1.5 with a p value lower than .05.
Viability of whole islets, ECs, or MSCs was assessed by measuring ATP using the CellTiter-Glo Luminescent Cell Viability Assay (Promega, Southampton, U.K., http://www.promega.com).
EC apoptosis was assessed using the Caspase-Glo 3/7 Luminescent Assay (Promega) according to the manufacturer's instructions.
Data were analyzed using Student's t test (paired and nonpaired) or ANOVA, as appropriate. Group data are presented as mean ± SEM unless otherwise stated.
Affymetrix Analysis of Cultured Islets
To determine how islet culture influences EC gene expression, RNA extracted from fresh and cultured (24–48 hours) mouse islets was analyzed using Affymetrix arrays. Expression of mRNAs for EC markers and for other genes normally expressed by ECs and generally associated with cytoprotection were reduced in cultured versus freshly isolated islets following 48 hours of culture (see Supporting Information Table 1C), including CD31 (−2.7-fold) and CD34 (−2.1-fold). Cultured islets also expressed reduced levels of vascular endothelial growth factor (VEGF) receptors VEGFR2 (−1.6-fold), VEGFR1 (−2.2-fold), and neuropilin (−2.8-fold) (Supporting Information Table 1C). In contrast, mRNAs for TGF-β type 1 receptor (TGF-β1R) and TGF-β-induced protein (TGF-βI) were upregulated in cultured islets and were maintained at +3 and +1.6-fold, respectively, after 48 hours. Islet endoglin mRNA levels were unchanged after 48 hours in culture whereas TβRII expression was marginally reduced (−1.1; Supporting Information Table 1C). Because TGF-β1R and TGF-βI mRNAs were elevated when EC markers were undetectable it is likely that non-EC populations are responsible for the increased TGF-β1R and TGF-βI mRNA expression in cultured islets.
Cultured Islets Retain Endoglin Expression
We verified the effects of islet culture on EC marker expression using end point reverse transcription PCR (RT-PCR), quantitative RT-PCR, and Western blotting. As expected, fresh adult mouse islets expressed the EC markers VEGFR2, CD31, vascular endothelial (VE)-cadherin, and CD34 which were markedly reduced or undetectable after 2 days of culture (Fig. 1A). In contrast, expression of endoglin mRNA was reduced after mouse islet culture for 2 days but remained detectable for up to 2 weeks (Fig. 1A). Similar results were obtained in human islets which lost VEGFR2 mRNA and protein expression in culture but retained expression of endoglin mRNA and protein after prolonged culture (Fig. 1B). Quantitative RT-PCR analysis revealed that endoglin expression decreased much more gradually than CD31 expression between 2 days and 1 week of culture (Fig. 1C).
Expression of Endoglin In Islets In Situ
To investigate the cellular expression pattern of endoglin in islets in situ, human and mouse pancreases were processed, sectioned, immunolabeled with endoglin-specific antiserum, and immunoreactivity was visualized using horse radish perioxidase (HRP)-3, 3′-diaminobenzidine (DAB) staining. In both human and mouse islets, endoglin expression was undetectable in the endocrine cells and most highly expressed around vessel-like structures within the islets and in the pancreatic matrix (Fig. 2A, 2B). Colabeling for endoglin (Texas red; Fig. 2C upper panel) and CD31 (fluorescein; Fig. 2C middle panel) was undertaken using cryosections of mouse pancreas. These studies revealed endoglin expression in islets which colocalized with CD31 in some cells (presumably ECs) and also showed endoglin expression in cells devoid of CD31. These observations provide evidence for the presence of a non-EC (CD31−) population of endoglin-expressing cells in islets in situ.
Characterization of Endoglin-Expressing Islet MSCs
As endoglin has been described as a marker of MSCs , we isolated endoglin-positive, non-EC populations by dispersion of human islets followed by expansion and detailed characterization of adherent cells in culture. Due to lack of a definitive marker for MSCs, cells (passage 3) were characterized by FACS, Western blotting, and RT-PCR analysis for a range of MSC markers using criteria for defining human MSCs proposed by the Mesenchymal and Tissue Stem Cell Committee of the International Society for Cellular Therapy  (Fig. 3). Thus, islet-derived cells were adherent to tissue culture plastic under standard culture conditions; the cells expressed a number of cell surface antigens associated with MSCs including CD105 (endoglin), CD13, CD44, CD73, and CD90 (Fig. 3A); and the cells did not express surface antigens associated with likely contaminating cells, including HLA-DR, CD31, CD34, CD69, CD117, and VEGFR2 (Fig. 3A–3C). In contrast, HMVECs were positive for CD34, CD31, VEGFR2, and endoglin mRNA and expressed VEGFR2 and endoglin protein (Fig. 3B, 3C). Islet-derived MSCs also highly expressed inhibitor of differentiation-1 (Id-1), a protein implicated in enhancing cell growth and limiting differentiation  with ECs expressing comparatively much lower levels of Id-1. Islet MSCs showed the expected proliferative capacity associated with MSC populations, with a linear growth curve over the passages 7–11 used in the current studies, and a doubling time of 3.9 days during this period in culture. Islet MSCs exhibited multipotency as they were capable of directed differentiation along adipogenic, chondrogenic, and osteogenic lineages, as evaluated by positive oil red O, alcian blue, and alizarin red staining, respectively (Fig. 3D). Human islet MSCs were distinct from ECs in their morphology and growth properties, displaying a fibroblastic appearance (Fig. 3D lower right panel). In keeping with cultured islets retaining an endoglin-expressing population of MSC-like cells, RT-PCR analyses showed that islets cultured for 2 weeks maintained their expression of other MSC markers (CD90, CD73, and CD44; Fig. 3E). In addition, cultured islets showed markedly enhanced Id-1 expression when compared to fresh islets (Fig. 3E); these findings are consistent with the high expression of Id-1 mRNA detected in islet MSCs but not ECs (Fig. 3B) and provide further evidence for retention of a population of MSCs in cultured islets.
Coculture of Human Islet MSCs with Microvascular ECs
There is evidence that bone marrow-derived MSCs exhibit proangiogenic properties and positively regulate angiogenesis through production of proangiogenic factors [25, 26]. Human islet MSCs strongly expressed VEGF mRNA splice variants (189, 165, and 121) (Fig. 4A) and differentiated after 1 hour in vitro to form two dimensional capillary-like structures on Matrigel, consistent with proangiogenic potential (Fig. 4B). However, in contrast to HMVECs, which formed networks at 3 hours (Fig. 4D) stable for up to 72 hours (data not shown), islet MSC networks disintegrated after 6 hours (Fig. 4C). To examine the potential for islet MSCs to influence the angiogenic functions of ECs, we coincubated MSCs with HMVECs on Matrigel. This showed that coculture of MSCs with HMVECs disrupted EC tube forming capacity after 6 hours (Fig. 4E). Further experiments using fluorescently labeled HMVECs (Cell Tracker) cultured alone or with islet MSCs confirmed that coincubation disrupted EC organization on Matrigel (Fig. 4F, 4G). Similarly, incubation of monolayer cultures of ECs with islet MSC-CM for 24 hours resulted in rounding and detachment of ECs (Fig. 4H) suggesting that soluble factors released by MSCs can exert detrimental effects on ECs. Incubation of ECs with medium preconditioned by MSCs for 72 hours also significantly (p < .01) reduced EC viability as assessed by luminescence measurements of total ATP content (Fig. 4I). Longer term mechanistic studies using HMVEC monolayers exposed to islet MSC-CM were not possible due to rapid EC detachment (data not shown). To further investigate the inhibitory effects of MSCs on EC function, we used monolayer cocultures of islet MSCs and HMVECs (Figs. 5 and 6). The phenotypic differences between islet MSCs and HMVEC were readily identifiable by light microscopy (Fig. 5A, 5B). Following 72 hours of coculture, the cobblestone morphology typified by ECs in monolayer culture became predominantly fibroblastic (Fig. 5C) and was associated with reduced VEGFR2 (Fig. 5D, 5E; p < .05 at 72 hours) and increased endoglin protein expression (Fig. 5D, 5E; p < .05 at 72 hours). To determine whether altered EC morphology and loss of VEGFR2 expression in the presence of MSCs reflected EC death, endothelial to mesenchymal transition, or overgrowth of ECs by MSCs, fluorescently labeled HMVECs (Fig. 5I; EC labeling after 72 hours) were cultured with and without MSCs. After 24–48 hours labeled ECs cocultured with islet MSCs exhibited a rounded morphology (Fig. 5F, 5G) and detached after 72 hours of coculture (Fig. 5H). Additional experiments showed that these detached, fluorescent ECs failed to readhere to tissue culture plastic, or to plastic coated with gelatin (1%) or cell attachment factor, consistent with an irreversible loss of EC viability (data not shown).
Inhibition of TβRI/ALK5 Activity Maintains EC Phenotype and Survival During EC:MSC Coculture
In ECs, TβRI/ALK5 is coupled to Smad2/3 phosphorylation and inhibition of proliferation . Basal phosphorylation of Smad2 was higher in monolayer cultures of MSCs versus HMVECs (Fig. 6A) and was significantly increased in MSC:EC cocultures after 48 hours (p < .05; Fig. 6A). MSCs also showed higher basal phosphorylation of Smad 1/5 than that observed in HMVECs but, in contrast to Smad2 (Fig. 6A), coculture of MSCs with ECs had no significant effect on Smad1/5 phosphorylation at any time point tested (24, 48, and 72 hours) (data not shown). These observations are consistent with the involvement of ALK5 in the loss of EC phenotype, which prompted us to examine the effects of SB431542 (SB), a small molecule inhibitor of ALK5 kinase activity  on islet MSC:EC interactions. As previous studies have shown a synergistic effect of SB and VEGF on angiogenesis , the effects of SB were examined in the absence and presence of VEGF165. As noted previously (Fig. 5) after 72 hours, cocultures appeared largely fibroblastic, consistent with the absence/reduced presence of ECs in the cultures (Fig. 6B). VEGF165 (20 ng/ml) had no effect on the phenotypic appearance of the cocultures (Fig. 6C) and did not significantly modify expression of VEGFR2 (Fig. 6F); there was also an increase in endoglin expression but this did not reach significance (Fig. 6G). However, cocultures exposed to SB (10 μM; 72 hours) exhibited clearly distinguishable islands of cells with an EC-like phenotype among the fibroblastic MSCs (Fig. 6D). SB-treated cocultures also expressed significantly greater levels of VEGFR2 protein than those exposed to vehicle alone (Fig. 6F; p < .001). Similar results were obtained in cocultures treated with a combination of VEGF165 and SB (Fig. 6E). ECs labeled prior to initiation of coculture were also clearly visible in cocultures exposed to SB for 72 hours (Fig. 6H).
We also found that SB had differential effects on monocultures of ECs and MSCs, significantly increasing EC viability but decreasing MSC viability as assessed by ATP measurements (Fig. 7A). Thus, the overall beneficial effect of SB on EC survival under coculture conditions most likely reflects actions on both ECs and MSCs. We then measured caspase 3/7 activity to confirm that the reduced EC viability reflected changes in EC apoptosis. Caspase activity was increased in ECs exposed to islet MSC-CM for 24 hours, and this effect was partially inhibited by pretreatment with SB (Fig. 7B). Similar results were observed when we used another pharmacological inhibitor of ALK5 kinase activity, A83-01 (10 μM), which also reduced the effects of MSC-CM on EC caspase 3/7 activity (caspase activity + CM, 100% ± 4%; + CM +A83-01, 43% ± 5%, n = 6; p < .01), confirming the involvement of ALK5 in the induction of EC apoptosis. Together, these results show that inhibition of ALK5 limits the loss of EC viability observed in the presence of islet-MSCs or of islet MSC-CM in vitro.
Treatment with SB431542 in Combination with VEGF164 Increases Islet Viability and SB431542 Prevents Loss of EC Markers During Islet Culture
To determine whether TβRI/ALK5 signaling also contributes to the loss of ECs observed in cultured islets, adult mouse islets were incubated with SB (10 μM) immediately following isolation and islet sensitivity to VEGF164 and expression of EC markers were measured after 48 hours in culture. SB alone had no deleterious effects on islet function. Thus, a 48-hour exposure to SB had no significant effect on the mean insulin-immunoreactive islet area (control, 7,231 ± 945 μm2; +SB, 9,005 ± 678, n = 110; p > .2), on islet insulin content (control, 23 ± 2 ng/islet; +SB, 21 ± 2, n = 12; p > .2), or on islet ATP content (Fig. 7C). Similarly, exposure of islets to exogenous VEGF164 (25–100 ng/ml) for 24 hours post isolation had no effect on islet viability as assessed by ATP production (Fig. 7C). However, ATP production by islets incubated with SB and VEGF164 (50 ng/ml) was significantly greater than that by islets exposed to VEGF164 alone (Fig. 7C; p < .05).
Finally, we measured the effects of SB treatment on mRNA expression of the EC markers VEGFR2, CD34, and CD31 using qRT-PCR. Treatment with VEGF164 (50 ng/ml) had no significant effect on marker expression compared to control (data not shown). In contrast, islets cultured in the presence of SB expressed significantly more VEGFR2 (p < .05; Fig. 7D), CD34 (p < .05; Fig. 7E), and CD31 (p < .01; Fig. 7F) than control islets. The SB-dependent increases in EC markers were associated with a decrease in endoglin mRNA expression (control, 92% ± 14% initial; +SB, 60% ± 8%, n = 3), consistent with the SB-induced reduction in immunoreactive endoglin in EC:MSC cocultures (Fig. 6G). The SB-dependent increase in VEGFR2 expression is consistent with VEGF164 enhancing whole islet viability, as assessed by ATP content, in the presence of SB but not in its absence (Fig. 7C), suggesting that the lack of effect of VEGF164 alone reflects reduced VEGFR2 expression. Direct immunohistochemical measurements of CD34-positive cells in islet sections demonstrated that incubation with SB alone caused a significant elevation in the numbers of CD34-positive ECs in islets (control, 0.8 ± 0.1 ECs per section; +SB, 1.2 ± 0.1 ECs per section, n = 110; p < .05). Thus, inhibition of ALK5 with SB limits the loss of EC markers that occurs in cultured islets.
The importance of TGF-β signaling in the cardiovascular system is underlined by the observation that genetic deletion of several TGF-β family members, their receptors, or downstream signaling molecules results in death of the mutants because of severe defects in the yolk sac vasculature . In this study, we have demonstrated that: (a) EC gene expression is diminished and TGF-βI/ TGF-β1R expression is upregulated in isolated islets maintained in culture; (b) islets contain two distinct populations of cells expressing the TGF-β accessory receptor endoglin, one of which coexpresses the EC marker CD31 in situ; (c) an endoglin-positive, MSC population expanded from human islets disrupts the proangiogenic properties of microvascular ECs and inhibits EC survival in vitro; these changes are associated with increased Smad2 phosphorylation, and (d) pharmacological inhibition of TβRI/ALK5 improves EC survival in cultured islets and reduces the detrimental effects of MSCs on cocultured ECs. Our studies provide evidence that resident MSCs influence EC survival in islets through engagement of TβRI/ALK5 and therefore identify TGF-β signaling elements as potential targets for positive manipulation of islet viability in vitro.
Our initial genomewide analysis of gene expression in fresh versus cultured islets showed a clear reduction in the expression of EC markers accompanied by elevated mRNA levels of TGF-β induced protein and TGF-β1R expression. TGF-β exerts diverse and complex actions on EC function, using either endoglin/ALK1 or ALK5 signaling  to promote or inhibit proliferation, respectively, and this prompted us to investigate whether TGF-β signaling may contribute to the loss of EC viability associated with islet culture. To determine whether altered signaling through endoglin contributed to the changes in EC marker expression and viability in cultured islets, we initially examined endoglin expression in whole islets in situ. Our measurements of the expression of endoglin and other EC markers (VEGFR2, VE-cadherin, CD31, and CD34) clearly demonstrated that both mouse and human islets contain at least two distinct populations of endoglin-expressing cells. The populations coexpressing endoglin, VEGFR2, VE-cadherin, and CD31 were progressively lost with time in culture, whereas the other endoglin-expressing population was readily detectable after prolonged culture, as was expression of the MSC markers CD44, CD73, CD90, and Id-1. Furthermore, colabeling experiments in situ revealed the presence of cells which were positive for endoglin but negative for CD31. Our identification of islet cell populations with overlapping gene expression profiles has implications for optimizing procedures used for isolating and characterizing islet ECs in the future. For example, using endoglin-coated immunomagnetic beads for extracting ECs from human islets has been described  and applied to several studies of islet EC function [31–34]. As described by the authors, cells isolated using this technique must be used at early passages due to possible contamination with nonendothelial, endoglin-expressing cells, consistent with our identification of two endoglin-expressing cell populations in human islets.
The identification of endoglin-positive cells that lack EC markers is in accordance with numerous reports of endoglin expression by diverse cell types . In particular, hemopoietic precursors from bone marrow and bone marrow-derived MSCs both express high levels of endoglin, with expression fluctuating according to cell activation status [22, 36, 37]. Our characterization of endoglin-positive, EC marker-negative cells isolated and expanded from dispersed human islets showed that they exhibited MSC-like properties and expressed a range of MSC marker genes . Islet MSCs also expressed high levels of Id-1 when compared to ECs, consistent with the high Id-1 mRNA expression detected in cultured islets. These islet-derived, endoglin-positive cells were also multipotent and could be driven toward adipogenic, osteogenic, and chondrogenic lineages under appropriate in vitro culture conditions, further confirming their MSC-like phenotype. Another recent study has also reported the presence of cells coexpressing endoglin and CD90 in isolated human islets  which, together with our findings, provides evidence for multiple endoglin-expressing cell populations in islets.
Our observations of the direct effects of islet-derived MSCs on cocultured microvascular ECs demonstrated that this particular population of MSCs negatively impact the angiogenic functions and survival of ECs in vitro. These effects were associated with enhanced Smad2 phosphorylation, suggesting that activation of TβRI/ALK5 is increased and that this pathway may therefore mediate the detrimental actions of islet MSCs on ECs. In ECs, endoglin associates preferentially with the TβRII/ALK1 complex (coupled to Smad1/5/8) and enhances TGF-β efficacy toward activating this complex [16, 13, 40] with ALK1 increasing and ALK5 inhibiting EC proliferation and survival . TGF-β promotes microvascular EC apoptosis in an ALK5-dependent manner  and suppresses EC expression of VEGFR2  consistent with our observation that SB treatment maintains VEGFR2 expression and thus enables the positive effects of VEGF164 on cell viability. It is clear, therefore, that in vitro blockade of ALK5 directly influences islet expression of EC-related genes. However, the levels of TGF-β required to maintain optimal islet EC function in vivo and the cell types involved in its production remain to be defined, although non-EC populations, including resident MSCs, appear likely candidates.
MSCs of various origins have generally been reported to exert beneficial effects on cell function and viability, rather than the deleterious effects that we have observed on EC survival. For example, bone marrow-derived MSCs enhance neoangiogenesis by secretion of growth factors [42, 43], and in vivo studies of islet transplantation have demonstrated beneficial effects, in terms of improved islet survival and glycemic control, of cotransplanting MSCs with islets . It is difficult to directly compare the deleterious effects on islet EC survival of an endogenous population of islet MSCs with the beneficial effects on islet endocrine function of MSC cotransplantation. Thus, the cotransplantation studies published to date have used exogenous (non-islet) MSC populations isolated from adipose tissue (adipocytes) and kidney . In addition, it is not clear in these in vivo models which of the many potential beneficial effects of MSCs (anti-inflammatory , immunosuppressive , matrix generating , proangiogenic ) are responsible for the reported improvements in islet survival and function.
It has been suggested that MSCs are derived from pericyte populations in vascularized tissues  and pericytes share some functional similarities with the islet MSCs characterized in our studies. Thus, pericytes express endoglin  and in common with the effects of islet MSCs observed in this study, early studies showed that pericytes also inhibit EC proliferation during coculture  via a TGF-β-dependent mechanism . In addition, it has been reported that pericyte-CM inhibits EC proliferation in vitro . These observations are consistent with our data from in vitro experiments using MSC-CM and MSC:EC cocultures in which pharmacological inhibition of ALK-5 using two different inhibitors improved EC survival, decreased endoglin expression, and reduced MSC viability; these results collectively suggest that limiting ALK5 activity alters TGF-β signaling in endoglin-expressing MSCs and that this is associated with EC survival. Thus, our in vitro data in ECs and MSCs are in keeping with our findings in whole islets and provide additional support for the suggestion that increased ALK5 activity in cultured islets involves changes in MSC TGF-β signaling which contribute to reduced EC viability.
The physiological role of this islet MSC population is uncertain, but TGFβ-dependent crosstalk between ECs and MSCs has previously been described in vivo. For example, studies in genetically modified mice revealed a TGF-β-mediated autoregulatory loop between ECs and MSCs, such that disrupted TGFβ signaling in ECs impaired TGFβ-driven ALK5 signaling in adjacent mesenchymal cells, inhibiting their differentiation into vascular smooth muscle cells and their association with developing endothelial tubes . Also, coculture of bone marrow-derived MSCs with adult ECs on Matrigel causes EC cytotoxicity, resulting in capillary network breakdown . These studies suggest that MSCs can suppress EC function and limit their survival and are consistent with our observations using islet-derived MSCs. Thus, it is possible that perturbed and unbalanced TGF-β signaling between islet ECs and MSCs results in detrimental effects on EC survival in isolated islets. Under normal conditions, there is little vascular remodeling or angiogenesis within adult pancreatic islets, but under physiological conditions where the β-cell/islet mass increases (e.g., obesity; pregnancy), the vasculature exhibits reversible expansion to accommodate the changes in islet endocrine mass . It is tempting to speculate that the interactions between islet MSCs and ECs observed in this study are involved in the regulation of these processes, and that the artificial situation of in vitro islet culture disrupts the fine balance of TGF-β signaling between MSCs and ECs that exists in vivo. It is likely that endothelial homeostasis within pancreatic islets is tightly coordinated by maintenance of appropriate levels of endoglin expression and TGF-β, as evident in other systems .
We have demonstrated that ALK5-mediated interactions between endogenous islet ECs and MSCs may be responsible for the loss of ECs in isolated, cultured islets, and that inhibition of TβRI/ALK5 activity maintains the phenotype of the EC population in isolated islets. Our observations suggest the existence of complex paracrine interactions involving TGF-β signaling between endoglin-expressing islet MSCs and ECs and highlight the TGF-β receptor signaling pathway as a novel therapeutic route through which to preserve intra-islet EC survival to enhance islet revascularization in procedures such as islet transplantation.
We thank Diabetes UK and The Diabetes Foundation for funding C.E.C.'s Training Fellowship, and the Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil for funding P.C.'s traveling scholarship. We also thank Professor Stephanie Amiel and Dr. Guo Cai Huang (Kings College Hospital human islet isolation facility) for human islets. We are grateful to Elaine Shervill (Royal Veterinary College) for histology and to Chloe Rackham (Kings College London) for isolating the islets used in the experiments shown in Figure 1C. We thank Dr. Helen Arthur and Dr. Marwa Mahmood (University of Newcastle) for the fluorescence images shown in Figure 2C.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors declare that they have no conflict of interest related to this study.