The efficient differentiation of mesenchymal stem cells (MSCs) into functional insulin-producing cells (IPCs) provides an attractive approach to strategies of cell transplantation for curing diabetes. However, it is noticed that differentiated MSCs to IPCs do not behave like native beta cells in view of insulin secretion quantity and response to glucose. Here, gene expression profiling by DNA microarray technology was recruited to compare 10K genes between rat pancreatic beta cells with IPCs differentiated from MSCs. Moreover, gene expression profiles of both were compared after glucose stimulation. Data were confirmed by RT-PCR and insulin secretion assay. The results revealed great gene profile differences between beta cells and differentiated IPCs under basal and stimulatory glucose conditions. Although IPCs were responsive to glucose stimulation, the insulin output and stimulatory index were lower than that in beta cells. These results suggest that the applied HGF-EGF differentiation protocol is insufficient for inducing beta-cell-similar IPCs from MSCs.

Keywords: Mesenchymal stem cells, gene expression profiling, DNA microarray, stem cell differentiation, insulin producing cells, pancreatic islets.

Transplantation of insulin-producing cells (IPCs) derived from mesenchymal stem cells (MSCs) may represent an alternative to cure diabetes mellitus. Several studies have described the successful differentiation of bone marrow MSCs cells into IPCs [1-8]. MSCs are mostly differentiated into IPCs by one of 2 methods. The first depends on the chemical manipulation of the microenvironment around the cells by some modifications of the culture medium. In this method, which is mostly applied, the addition of some growth factors such as hepatocyte growth factor (HGF), epidermal growth factor (EGF) and nicotinamide was shown to be sufficient for the conversion of MSCs to IPCs [8-14]. In the other method, differentiation can be achieved by genetic manipulations. As a result of both differentiation protocols, IPCs were able to secrete insulin, but so far these cells were not as efficient as native beta cells. Differences in cell shape, phenotypic characters and in insulin response to glucose have been observed.

Here, gene expression profiling may help to investigate the similarity of the transcriptome between beta cells and differentiated stem cells. The insulin-secreting beta cells express specific genes that are essential for the development and/or function of these cells. In addition to insulin gene, the signal for insulin release from the β-cells is mediated by uptake of glucose via the low affinity glucose transporter type 2, which is encoded by the gene Glut2. For Insulin expression to be achieved, a number of transcription factors have been shown to play important roles during differentiation of pancreatic endocrine cells and in insulin secretion process. Nkx6.1 and Nkx2.2 act to assure β-cell function. In mice mutant for these genes the specification of β-cells are specified but fail to terminally differentiate [15, 16]. Pax4 is another transcription factor needed for the specification of both β-and δ-cells [17, 18]. Pax6, Isl1 and NeuroD are expressed in differentiated pancreatic endocrine cells. Deletions of any of these genes resulted in mice with perturbed pancreatic endocrine cell differentiation [19-21]. Also, lacking Ngn3 expression or function fail to generate any pancreatic endocrine cells and die postnatally from diabetes [22, 23]. Pdx1 has a dual role being required both for early pancreas development and for the proper function of insulin secreting β-cells [24, 25]. All of these beta cell marker genes must be upregulated during the differentiation of insulin-secreting cells. On the other side, other genes such as stem cell – specific and cell cycle activation genes should be downregulated during the differentiation process. DNA microarray technology has served in several studies to follow the gene expression profiling during the embryonic stem cell differentiation [26-28], but little is known about the gene profile of MSCs differentiation to IPCs.

In the present study, the transcriptome of differentiated MSCs to IPCs has been compared with that of beta cells and undifferentiated MSCs. It is shown that validating the differentiation status of MSCs to IPCs by microarray analysis and real-time RT-PCR using beta cell markers revealed that a dissimilarity exists between the transcriptome of beta cells and the MSCs-derived insulin-secreting cells in response to glucose stimulation.

Rat bone marrow was isolated by flushing femurs and tibias by DMEM as described by Zhang and Chan [29]. All bone marrow cells were cultured for 4 days and the plastic adhered cells were washed several times and cultured until reached confluency. MSCs in passage 3 were used for differentiation. The differentiation into insulin-producing cells (IPCs) followed the method applied by most laboratories [eg. 8-11] including ours [12-14]. The islet beta cell-conditioning medium was DMEM with 5.5 mmol/L glucose, and contained HGF (hepatocyte growth factor, Sigma-Aldrich, cat. H9661, 20 ng/ml), EGF (epidermal growth factor, Sigma-Aldrich, cat. E9644, 20 ng/ml) and nicotinamide (Sigma-Aldrich, 10 mmol/L). MSCs were cultured in this differentiation medium for 3 weeks, and then tested for insulin and other islet genes expression and insulin secretion. For investigating the effect of stimulation with glucose, cells (MSCs or IPCs) were cultured in RPMI media containing either 2.8 (basal) or 22.4 mmol/L (stimulatory) glucose for 24 hr.

Adult rat islets were isolated and cultured as described in details in our previous work [30]. For investigating the effect of stimulation with glucose, islets were cultured in RPMI media containing either 2.8 (basal) or 22.4 mmol/L (stimulatory) glucose for 24 hr.

Three tissues under 2 conditions were used for gene profiling. Tissues were rat MSCs, rat differentiated IPCs and rat islets, while conditions were culturing in either low (2.8 mmol/L) or high (22.4 mmol/L) glucose for 24 hr. Total RNA was extracted from tested tissues using a RNeasy Mini Kit (Qiagen). DNase I treatment of isolated RNA was used after extraction to exclude any DNA interference in the labeling reaction and during hybridization. First strand cDNA and dsDNA were synthesized and labelled using Express Art mRNA Amplification Kit (Micro Version, Amp Tec, Germany). During reverse transcription, fluorescent-labeled nucleotides were incorporated into the produced first strand cDNA. The first strand cDNA was then separated from the RNAse-degraded template RNA, primers, unincorporated nucleotides, and RNA debris. A microarray chip (Rat 10K OciChip, Ocimum BioSolutions), which can be loaded with 2 different samples have been applied. The two sets of differently labeled cDNAs were combined and co-hybridized to the same OciChip. After hybridization, unbound and non-specific fixed cDNA was removed by thoroughly washing the array. After scanning of the array using a Gene Array scanner into a microarray image, the fluorescence intensity of each spot, and the ratio of the expression levels between the two cell populations were analyzed by ImaGene software (Biodiscovery).

To confirm microarray data, the expression of some specific beta cell genes has been analyzed by RT-PCR as described elsewhere [31]. The following primers have been used: rat Ins1 5'-AGGCTCTGTACCTGGTGTGT-3' (forward) and 5'-AGTTGGTAGAGGGAGCAGATG-3' (reverse), Glucagon 5'-CTTCCCAGACAGAACCACTTG-3' (forward) and 5'-CTGGCCCTCCAAGTAAGAACT-3' (reverse), Glut2 5'-AGCACATACGACACCAGACG-3' (forward) and 5'-TCAAGAGGGCTCCAGTCAAC-3 (reverse), β-actin 5'-ACCGTGAAAAGATGACCCAGATC-3' (forward) and 5'-GACCAGAGGCATACAGGGACAAC-3' (reverse).

After the 24 hr incubation with either basal or stimulatory glucose, rat insulin was determined in the supernatant using ELISA kit from DRG diagnostics, Germany (EIA-2048).

Data are presented as mean±SEM. Analysis of variance (ANOVA) was applied for the statistical analysis followed by t-test as a post-hoc test. A p<0.05 was considered as significant in all cases.

The aim of the present study was to investigate the change in the gene expression profile of bone marrow – derived mesenchymal stem cells that are differentiated to insulin-secreting cells and compare this profile with that of the native insulin-secreting pancreatic beta cells. Thus, the transcriptome of three tissues (bone marrow mesenchymal stem cells MSCs, differentiated insulin-producing cells IPCs, and pancreatic islets) have been examined after culture with low (basal) or high (stimulatory) glucose concentrations. This transcriptome has been analyzed by the DNA microarray technology, which can describe, in a semi-quantitative way, 10,000 of actually active genes in a target cell.

Before the application of microarray analysis, IPCs were proven to be functional, regarding both glucose-stimulated insulin secretion and insulin gene expression. Although IPCs were able to secrete insulin in response to glucose, the present results showed that none of the beta cell marker genes have been found in the uppermost upregulated genes (Table 1) during the differentiation of MSCs to IPCs. Instead, the uppermost upregulated genes were related to stress tolerance, indicating that the differentiation of MSCs to IPCs is a stressful process. For example, ubiquitin C, Ubc, which expression increased about 20x in IPCs more than MSCs, is known to be induced during stress. It provides ubiquitin protein necessary to remove damaged or unfolded proteins, kinase activity, DNA repair and many other related biological processes [32-34]. Similarly, peroxiredoxin 6 and glutathione S-transferase participate in the protection against oxidative injury [35], and thioredoxins reduce oxidative stress through their response to reactive oxygen species [36]. Max protein represses MYC transcriptional activity from E-box elements and negative regulation of G0 to G1 transition [37], which is a sign of differentiation. Most other activated genes were related to metabolism, oxidative phosphorylation and mitochondrial function.

Upregulated genes of pancreatic islet development and function were found in the middle of the list of upregulated genes. Table 2 summarizes the upregulation level of some of these genes. Insulin 2 was the most upregulated gene and its expression increased by 3.17x in IPCs. The values of upregulation of insulin and other genes as glucagon, the glucose transporter Glut2 and also that of other transcription factors seem to be lower than expected. To confirm this, data of stimulation by glucose for the same cells in comparison with islet cell data were also included in Table 2. Glucose-stimulated insulin and glucagon gene expressions in islets 28.68 x and 21 x times that in IPCs, respectively. This great difference was not repeated in most of the transcription factor genes. However, the expression of the essential genes pdx1 and Glut2 was also 2.55 x and 3.36 x times higher in islets than in IPCs. Taken together, these data reveal the unequal response of islet-specific genes in islets and IPCs differentiated from MSCs. Also, it is logically expected that due to this comparatively weak response in these gene expressions, the insulin – and also glucagon – secretions would be lower from IPCs than from islets.

The effect of stimulation with glucose on insulin gene and other islet-specific gene expressions in IPCs, islets, and MSCs has been compared (Table 3). Samples of the 3 tissues have been cultured for 24 hr in either basal or stimulatory glucose concentrations. Labeled samples of the same tissue but from both culture conditions were arrayed on the same microarray chip. The results (Table 3) revealed that glucose-induced a moderate increase in insulin gene (4 times more than the basal value) and hardly any change in glucagon gene expression. In islets, glucose-stimulated 21 times increases in insulin gene expression than the basal value. It inhibited glucagon gene 15x and also induced a 12x increase in Glut2 expression. The effect on glucagon gene may be indirect. It was reported that glucose itself inducesglucagon secretion from isolated alpha cells, but the inhibition in islets is derived by the inhibitory effect of the secreted insulin and its paracrine action [38]. Thus, although islet gene expressions in IPCs were highly responsive to glucose, their response was weak, as compared to that of islets.

The results showed in Table 4 that glucose worked variably in both systems (IPCs and islets). The uppermost genes upregulated by glucose in islets were Insulin and Glut2, whereas most upregulated enzymes in IPCs were metabolic enzymes, including some involved in glucose metabolism, but not insulin secretion. These results revealed dissimilarities in the glucose-stimulated change in the gene profile of both islets and IPCs.

The obtained results have been confirmed by PCR amplification of some islet-specific genes, and also by the determination of glucose-stimulated insulin secretion (Figure. 1). The PCR (Figure. 1A) showed the lower scale of expressions of Ins1, Ins2, Glut2 and glucagon genes in IPCs than in islets cultured in basal glucose conditions.

Differentiated insulin-producing cells and pancreatic islets were cultured for 24 hr in either basal or stimulatory glucose media. Insulin secretion was measured in the supernatant and data were normalized by referring to the tissue proteincontent (Figure. 1B and 1C). The results showed a significant difference between different groups (ANOVA, p=0.000001). Glucose could stimulate insulin secretion significantly in both islets and IPCs. However, the secreted insulin quantity was far less in IPCs than that in islets. As well, the stimulatory index, calculated as stimulated secretion / basal secretion, was significantly higher in islets than in IPCs.

The present data demonstrate a different pattern of gene expression and different transcriptome in pancreatic islets and insulin-producing cells differentiated from bone marrow mesenchymal stem cells. More efforts are still required to develop differentiation protocols that make the transcriptome and the consequent glucose-stimulated insulin secretion in differentiated stem cells closer to that of pancreatic islets.