MiR-375 Promotes Redifferentiation of Adult Human β Cells Expanded In Vitro

In-vitro expansion of β cells from adult human pancreatic islets could provide abundant cells for cell replacement therapy of diabetes. However, proliferation of β-cell-derived (BCD) cells is associated with dedifferentiation. Here we analyzed changes in microRNAs (miRNAs) during BCD cell dedifferentiation and identified miR-375 as one of the miRNAs greatly downregulated. We hypothesized that restoration of miR-375 expression in expanded BCD cells may contribute to their redifferentiation. Our findings demonstrate that overexpression of miR-375 alone leads to activation of β-cell gene expression, reduced cell proliferation, and a switch from N-cadherin to E-cadherin expression, which characterizes mesenchymal-epithelial transition. These effects, which are reproducible in cells derived from multiple human donors, are likely mediated by repression of PDPK1 transcripts and indirect downregulation of GSK3 activity. These findings support an important role of miR-375 in regulation of human β-cell phenotype, and suggest that miR-375 upregulation may facilitate the generation of functional insulin-producing cells following ex-vivo expansion of human islet cells.


Introduction
Beta-cell replacement by regeneration or transplantation is considered a promising therapy for diabetes. Transplantation is greatly hindered by shortage of human islet donors. In-vitro expansion of β cells from adult human pancreatic islets could provide abundant insulinproducing cells for transplantation, however induction of islet cell replication in culture leads to loss of β-cell phenotype, in a process resembling epithelial-mesenchymal transition (EMT) [1][2][3]. Expanded human β-cell-derived (BCD) cells, which constitute~40% of cells in islet cell cultures [2], maintain open chromatin structure at β-cell genes [4], and can be redifferentiated in response to a combination of soluble factors termed Redifferentiation Cocktail (RC) [5]. These factors include activin A, exendin-4, nicotinamide, and high glucose concentrations, which have been shown to promote β-cell differentiation, in serum-free medium supplemented with B27 and insulin-transferrin-selenium. However, RC treatment leads to redifferentiation of only part of BCD cells. In search for improved redifferentiation approaches, we analyzed changes in microRNAs (miRNAs) during BCD cell dedifferentiation. miRNAs are endogenous short noncoding RNAs which bind to the 3 0 -untranslated regions of target mRNAs and act as negative regulators of gene expression [6]. miRNAs play important roles in regulation of islet development, β-cell differentiation and function [7,8], and human diabetes [9]. Among the miRNAs highly expressed in islets, miR-375 has been shown to be required for normal mouse glucose homeostasis [10] and zebrafish β-cell development [11], and expressed at high levels during human islet development [12], as well as in mature islets [13,14]. Using miRNA microarray analyses we identified miR-375 as one of the miRNAs greatly downregulated during BCD cell proliferation in vitro. We hypothesized that restoration of miR-375 expression in expanded BCD cells may contribute to their redifferentiation. Our findings demonstrate that overexpression of miR-375 alone activates BCD cell redifferentiation by affecting multiple targets.

Ethics statement
This study was conducted according to the principles expressed in the Declaration of Helsinki. The Institutional Review Boards of the following medical centers, which provided human islets, each provided approval for the collection of samples and subsequent analysis: University of Geneva School of Medicine; San Raffaele Hospital, Milan; Faculty of Medicine, Lille 2 University; Massachusetts General Hospital; Washington University; University of Pennsylvania; Scharp/ Lacy Institute; University of Illinois; University of Wisconsin; University of Miami; Southern California Islet Consortium. All donors provided written informed consent for the collection of all samples and subsequent analysis.

Virus production and cell infection
Pre-mmu-miR-375 was subcloned into pBABE-Bleomycin vector and co-transfected into human embryonic kidney 293T cells for virus production with the Ampo-helper packaging plasmid. The medium was replaced 6h post-transfection, and the virus was harvested 24h later and used fresh. 10 6 human islet cells were plated in 14-cm culture dishes in growth medium for 24h. Cells were infected at MOI of 3:1 in medium containing 8 μg/ml polybrene (Sigma-Aldrich) for 6h. The infection was repeated two more times in the following two days. Selection of bleomycin-resistant cells was initiated 2-3 days later with 4 μg/ml bleomycin for 5 days. Following selection (total of 10 days from the first infection), the cells were harvested for analysis.  [15]. Data was normalized to median hybridization intensity and analyzed using Genepix pro 4000b Axon and JMP statistical software.

qPCR analysis
Total RNA was extracted using ZR RNA MiniPrep kit (Zymo Research) or TRIzol (Sigma-Aldrich) with DNase I (Thermo Scientific). cDNA was prepared using High Capacity cDNA RT Kit (Applied Biosystems). qPCR was carried out using ABsolute blue qPCR Mix (Thermo Scientific) or FastStart Universal Probe Library Master Mix (Roche) in a 7300 real-time PCR instrument (Applied Biosystems). The results were normalized to transcripts of TATA-box-binding protein (TBP) and human large ribosomal protein (RPLPO). Table 2 lists primer sequences designed for Universal Probe Library (Roche). All reactions were performed with annealing at 60°C for 40 cycles. For undetectable transcripts, the cycle number was set to 40 for comparisons. cDNA for miRNA analyses was prepared and analyzed using Taqman MicroRNA Assay kit (Applied Biosystems) according to the manufacturer, with primers listed in Table 3.

Immunoblotting
Total protein was extracted from cells by incubation with a lysis buffer containing 0.5% NP-40 and protease inhibitor cocktail for 10 min. 20-25 μg protein were resolved by SDS-PAGE and electroblotted onto Immobilon-P 0.45-μm membrane (Millipore), followed by incubation overnight at 4°C with primary antibody ( Table 4). The bound antibody was visualized with the appropriate horseradish peroxidase-conjugated anti-IgG (Jackson Immunoresearch) and SuperSignal West Pico chemiluminescent substrate (Pierce). Quantification was done using TINA 2.0 software.

Immunofluorescence and cell proliferation analyses
Cells were trypsinyzed, spotted on slides using a Shandon Cytospin4 centrifuge (Thermo Scientific), and fixed for 15 min at RT in 4% paraformaldehyde (PFA). Slides were blocked for 30 min in PBS containing 1% BSA, 5% fetal goat serum and 0.2% saponin (blocking buffer). Slides were incubated overnight at 4°C with primary antibodies diluted in blocking buffer as follows: doi:10.1371/journal.pone.0122108.t002

Cell apoptosis assay
Apoptotic cells were detected by TUNEL staining using the In Situ Cell Death Detection Kit (Roche).

Proteomics
Cell pellets were solubilized in a buffer containing 6 M urea and 2 M thiourea in 50 mM Tris-HCl pH 7.5. Following protein reduction (1 mM DTT) and alkylation (5 mM iodoacetamide), proteins were digested with trypsin overnight at room temperature. Resulting peptides were purified on C18 tips. Liquid-chromatography mass-spectrometric analysis was performed on the EASY-nLC1000 UHPLC coupled to the Q-Exactive mass spectrometer (Thermo Scientific). Cells from each donor and treatment were analyzed in two technical replicates. MS files were analyzed in MaxQuant with an FDR threshold of 1% on the peptide and protein levels.

GSK-3 inhibition
Cells were incubated with SB-216763 (Sigma) at 3 or 6 μM for 48h and then harvested for analysis.

Statistical analyses
Significance of qPCR and immunoblotting data was determined by two-tailed t-test. To approach a normal distribution, logarithmic transformation was preformed. Significance of immunofluorescence cell counts was determined by χ2 test.

Changes in miRNA expression during BCD cell expansion in vitro
We used expression arrays to compare miRNA levels in expanded human islet cells following proliferation and dedifferentiation in culture, with those in isolated mature human islets. Twenty four miRNAs were downregulated (>2-fold) during the first two weeks of culture (equivalent to 2 passages), and 8 miRNAs were upregulated (Fig 1A). The miRNAs downregulated the most included the miR-141/miR-200 and miR-30 families, as well as miR-192, miR-204, and miR-215, which play key roles in maintaining epithelial cell phenotype [17,18], and their downregulation is in accordance with the EMT-like change occurring in cultured BCD cells. miR-7 participates in regulation of islet cell differentiation and function [19][20][21] and is abundant in mature islets [20,22], however its overexpression in developing pancreas inhibits αand β-cell differentiation [19]. Downregulation of specific miRNAs was confirmed in sorted GFP + BCD cells labeled by an insulin promoter-driven lineage tracing system [2] (Fig 1B). The expression array analysis demonstrated a >12-fold downregulation in miR-375 expression in expanded human islet cells, relative to islets. A much more pronounced (>380-fold) downregulation was revealed by qPCR in sorted GFP + BCD cells, indicating a β-cell-specific effect. Furthermore, miR-375 was readily detected by in-situ hybridization in β cells of isolated adult human islets, and co-localized with C-peptide immunostaining (Fig 2A). In view of the important developmental and functional roles of miR-375 in β cells, we evaluated the effect of miR-375 overexpression on BCD cell redifferentiation.

Effect of miR-375 overexpression on BCD cell redifferentiation
A pre-miR-375 retrovirus vector was used to overexpress pre-miR-375 and a bleomycin resistance gene in expanded BCD cells. miR-375 levels in bleomycin-resistant transduced cells were upregulated by several hundred-fold ( Fig 2B) and were comparable to the expression levels of miR-375 in islet cells prior to dedifferentiation (Fig 1B). Restoration of miR-375 resulted in a 2-fold decrease in transcripts encoding the mesenchymal markers N-cadherin and vimentin (Fig 2C), and a 3-fold increase in E-cadherin mRNA levels (Fig 2D), as well as a change in cell morphology (S1 Fig), suggesting the induction of mesenchymal-epithelial transition (MET). mRNAs of several key β-cell transcription factors, including PDX1, MAFA, NKX6.1, NEU-ROD1, and PAX4, were upregulated 3.4-7.6-fold, and INS and IAPP transcripts were induced 7.5-and 22-fold, respectively. Consistent upregulation was observed in sorted GFP + BCD cells ( Fig 2E). In addition, miR-375 overexpression in expanded islet cells upregulated MAFB and GCG transcripts (Fig 2D). However, since the miR-375-induced GCG, PPY, and SST transcript elevation in sorted GFP + BCD cells was insignificant (S2 Fig), we conclude that a distinct population of insulin-negative/glucagon-positive cells likely originates from non-BCD cells, in accordance with our previous results [5]. miR-375 further induced insulin protein formation, as judged by C-peptide immunofluorescence analysis (Fig 2F-2H). The vast majority (>98%) of C-peptide + cells co-stained for PDX1 (S3 Fig). Redifferentiation efficacy approximated 12.5% of GFP + cells (Fig 2G). This level of redifferentiation represents about half of that achieved with RC treatment [5]. In addition, redifferentiation was accompanied by a 4-fold increase in CDKN1A transcripts encoding the cell cycle inhibitor p21 (Fig 2D), and a >2-fold decrease in cell proliferation (from 12% to 5%), as judged by staining for Ki67 (Fig 2I). Minor changes in apoptosis rates were noted between uninfected cells (2.1±0.1%) and cells infected with miR-375 (2.5±0.2%) or empty viruses (S4 Fig). Taken together, these findings demonstrate that miR-375 induces profound changes in BCD cells and directs them towards redifferentiation.

miR-375 overexpression in expanded islet cells downregulates the PDPK1-AKT pathway
To unravel the mechanism underlying these effects, we analyzed changes in expression of established and predicted miR-375 targets. While overexpression of miR-375 did not cause a significant change in the expression levels of MTPN, HNF1B, PAX6, and NOTCH2, a small but significant 18% decrease was detected in transcripts encoding 3-phosphoinositide dependent protein kinase-1 (PDPK1) (S5 Fig). PDPK1 is a serine-threonine kinase which mediates signaling downstream of PI3-kinase and is directly targeted by miR-375 [23]. As seen in Fig 3A and 3B, PDPK1 protein levels increased by 50% during the first 3 weeks of human islet cell expansion in culture, whereas miR-375 overexpression resulted in a significant 30% reduction in PDPK1 levels (Fig 3C and 3D). Knockdown of PDPK1 by shRNAs was sufficient for induction of a significant increase in insulin transcripts in expanded islet cells (Fig 3E and 3F).
One of the main substrates of PDPK1 is AKT, which is activated by PDPK1-mediated Thr308 phosphorylation [24]. Dedifferentiation of BCD cells in the first three weeks of human islet cell expansion in culture was associated with a 15-fold elevation in phospho-AKT levels (Fig 3G and 3H). Accordingly, miR-375 overexpression resulted in a 5-fold reduction in phospho-AKT levels (Fig 3I and 3J), while total AKT protein levels slightly increased (Fig 3K and  3L). These findings position PDPK1 as an important functional target of miR-375 in a pathway that regulates BCD redifferentiation (Fig 3M).

miR-375 overexpression downregulates GSK3
Since target mRNA analysis may obscure miRNA effects manifested at the protein level, as a result of translation inhibition rather than transcript degradation, we performed mass spectrometric analyses for unbiased profiling of changes in gene expression induced in BCD cells by miR-375 overexpression. The analyses revealed changes in 49 proteins, 17 of them were upregulated, and 32 were downregulated (p<0.05) (Fig 4A). The protein downregulated to the largest extent was glycogen synthase kinase (GSK)-3α (1.6-fold difference between means of miR-375 overexpression and control). We therefore investigated changes in expression of both GSK-3α and GSK-3β during islet cell dedifferentiation and miR-375 overexpression. The levels of the active forms of both GSK-3α and GSK-3β were elevated >20-fold during the first 3 weeks of human islet cell expansion in culture, while the levels of the inactive forms decreased by 60% (Fig 4B-4D). miR-375 overexpression induced a significant decrease in total GSK-3α and GSK-3β protein levels (Fig 4E), as well as a decrease in the active forms of both proteins (Fig 4F). It also induced an increase in the inactive form of GSK-3β (Fig 4G), however levels of the inactive form of GSK-3α were not altered (Fig 4H). GSK-3α/β phosphorylates multiple substrates, including PDX1 [25,26] and MAFA [27], both of which are targeted for degradation following phosphorylation. MAFA protein levels were elevated in expanded islet cells overexpressing miR-375 (Fig 5A and 5B). To directly demonstrate that a reduction in GSK3 activity is involved in redifferentiation, we employed the GSK3 inhibitor SB-216763, which inhibits both GSK-3α and GSK-3β activity [28]. A 2-day treatment of expanded islet cells with SB-216763 induced a dose-dependent 60% increase in MAFA protein levels (Fig 5C and 5D). The decrease in GSK3 activity did not induce an increase in protein levels of β-catenin (Fig 6A and 6B), a key GSK3 target involved in regulation of cell proliferation [29]. Following miR-375 overexpression, β-catenin was predominantly localized near the plasma membrane, unlike control cells, in which it was detected throughout the cell (Fig 6C). As with miR-375 overexpression, SB-216763 did not increase total β-catenin protein levels (Fig 6D and 6E), and resulted in growth arrest (Fig 6F). Taken together, these findings suggest that GSK3 inhibition at least partially mediates the effect of miR-375 overexpression on BCD cell redifferentiation.

Synergistic effects of miR-375 overexpression and RC on BCD cell redifferentiation
We have previously shown that BCD cells can be redifferentiated by treatment with a combination of soluble factors in serum-free medium, termed Redifferentiation Cocktail (RC) [5]. RC treatment resulted in a detectable increase in miR-375 levels in expanded islet cells ( Fig  7A) and GFP + BCD cells (Fig 7B) during the first 4 days of treatment, and a further increase by 6 days (Fig 7A). Expanded islet cells subjected to a combined treatment of RC and miR-375 overexpression showed a 2-fold increase in β-cell transcripts, compared to RC treatment alone ( Fig 7C). Similar results were obtained in sorted GFP + BCD cells (Fig 7D). The combined treatment also resulted in a 70% increase in the number of C-peptide + BCD cells, compared with cells treated with RC alone (Fig 7E and 7F). Overall, these findings suggest that increased miR-375 levels interact with the pathways activated by RC and result in enhanced BCD cell redifferentiation.

Discussion
Our findings demonstrate that restoration of normal levels of a single miRNA, miR-375, in BCD cells is sufficient for induction of β-cell gene expression, reduced cell proliferation, and a switch from NCAD to ECAD expression, which is characteristic of mesenchymal-epithelial transition. These effects are reproducible in cells derived from multiple human donors. Our results support an important function of miR-375 in regulation of the differentiated human β-cell phenotype, and emphasize the roles of PDPK1 and GSK3 in mediating its effects.
PDPK1 has been shown to be a direct target of miR-375 in rodent islet cells [23]. Our findings suggest that it is modulated by miR-375 in human islet cells as well. Analysis of the PDPK1-AKT pathway revealed a reduction of 30% in PDPK1 protein levels following miR-375 overexpression, resulting in 80%-decrease in phospho-AKT levels. Reduced activity of the PDPK1-AKT pathway may cause a decrease in BCD cell proliferation and an increase in cell MiR-375-Induced Redifferentiation of Expanded Human β Cells differentiation. Indeed, mice deficient in PDPK1 in β cells manifest reduced β-cell numbers and hyperglycemia [30], while AKT overexpression under the Pdx1 promoter results in β-cell dedifferentiation [31]. One possible mechanism by which a decrease in phospho-AKT activity may lead to growth arrest is by induction of p21 (CDKN1A; Fig 2D) [32].
Our findings implicate for the first time GSK3 in miR-375 activity in human islet cells. miR-375 overexpression downregulated GSK3α/β levels and activity, and upregulated the inactive form of GSK3β. Since GSK3α and GSK3β transcripts do not contain miR-375 binding sites, these likely represent indirect effects. Given that AKT is a negative regulator of GSK3β [33], the reduction in phospo-AKT would be expected to result in an increase in active GSK3β, and a decrease in inactive GSK3β. However, it is conceivable that additional protein kinases and phosphatases are involved in balancing the different phosphorylated states of GSK3 [34] following miR-375 overexpression. Our results suggest that a decrease in GSK-3β activity is associated with reduced islet cell proliferation. This is supported by the findings that miR-375 overexpression or the GSK3 inhibitor SB-216763 did not significantly increase β-catenin levels in expanded islet cells, and resulted in growth arrest. Apparently, the residual GSK3β activity is sufficient for regulating β-catenin levels. In contrast, β-cell-specific GSK-3β knockout in mice resulted in increased β-cell mass [35]. Nonetheless, GSK-3β has been associated with increased cell proliferation in other systems [36,37]. miR-375 overexpression in expanded islet cells resulted in MET, as judged by downregulation of N-cadherin and upregulation of E-cadherin. Recent work has identified SHOX2, an MiR-375-Induced Redifferentiation of Expanded Human β Cells inducer of EMT in breast cancer cells, as a novel miR-375 target [38], suggesting a possible mechanism for restoration of the epithelial phenotype in BCD cells by miR-375.
Expression of miR-375 by itself induced detectable C-peptide expression in only 12% of BCD cells, making it difficult to assess cell function, such as glucose-stimulated insulin secretion (GSIS). However, miR-375 overexpression synergized with RC treatment in promoting BCD cell redifferentiation, as manifested by a 3.7-fold increase in insulin transcript levels, and a 1.8-fold increase in the number of C-peptide-positive cells, compared with RC alone, which were highly reproducible in cells from multiple human donors. This synergy occurred despite the significant miR-375 upregulation induced by RC alone, suggesting a quantitative correlation between miR-375 levels and insulin expression in BCD cells. miR-375 has been implicated in limiting GSIS under stress in MIN6 cells by downregulating myotrophin (Mtpn) transcripts [8]. However, the levels of human MTPN mRNA did not change following miR-375 overexpression in expanded islet cells, and miR-375 upregulation by RC did not inhibit GSIS in BCD cells [5].
Our results suggest that miR-375 expression may contribute to future approaches for cell replacement therapy of diabetes based on in-vitro expansion and redifferentiation of donor islet cells. Clinical application will require non-viral delivery of miR-375, functional assessment of the redifferentiated cells in vitro and in vivo, and development of effective immunoprotective approaches. Research Program). This work was performed in partial fulfillment of the requirements for a Ph.D. degree of GN.