Loss of Langerhans’ islet occurs in both type I and type II diabetes [42, 43]. Currently, islet transplantation is the main way to cure diabetes completely. However, the lack of available donor islets has prevented extensive use of this method [44]. Thus, identification of alternative islet sources may provide a gateway to widespread use of this practice to treat diabetes.

Recent evidence suggests that adult pancreatic progenitors may provide a potential source for beta-cell neogenesis in adults [2]. In rodents, ductal epithelium has been proven to be a potential progenitor pool in vivo [7]. Partial duct ligation [7] or treatment with exendin-4 [8] induced ductal cell proliferation and differentiation into beta cells, suggesting that insulin-producing cells can be generated from adult pancreatic ductal cells. Just like embryonic pancreatic progenitors, in-vitro cultured murine adult pancreatic ductal cells can reacquire self-renewal capacity and pluripotency [3, 1113, 29]. These cultured ductal cells can differentiate into insulin-producing cells in vivo [10] and in vitro [9, 29]. Altogether, the adult pancreatic progenitors can serve as a promising beta-cell source for transplantation, although more research is needed.

Previously we reported that a 3D system was found to culture CD133+/SOX9high pancreatic progenitor-like colonies [11]. This subpopulation could be expanded for many passages without losing its potential to differentiate to endocrine/acinar lineages. In addition, these cells can response to Wnt signal agonist R-spondin1. However, even if this is an excellent system to enrich these colonies, it is not perfect to study the regulation because of the conditioned medium which has uncertain components. Also, CD133+/SOX9high were not proved to be all colony-forming units. To elucidate the mechanisms underlying the regulation of pancreatic progenitor cells, therefore, we modified the culturing system and used single cells from whole pancreas to generate pancreatic progenitor-like colonies and performed HTS to profile the transcriptome.

In our study, mouse pancreas was dissolved into single cells. In our modified system, these cells formed a cyst-like organoid. These colonies were PDX1+, CK7+, SOX9+ and weakly expressed NEUROG3. Also, they can differentiate to C-peptide secreting cells spontaneously. Following multiple passages, the cells did not lose self-renewal capacity and the ability to differentiate into other pancreatic lineages in a 2D system. Thus, our cultured pancreatic ductal epithelium exhibited pancreatic progenitor-like property and function.

To identify factors that regulate colony proliferation and differentiation, we profiled the colony cells’ transcriptome using HTS, a widely used technology to study the gene expression profile [45]. Total RNAs including mRNAs, miRNAs and lncRNAs are sequenced and annotated. As well as RNAs that have been discovered, new RNAs (mainly miRNAs and lncRNAs) can also be sequenced using this technology. By comparing the RNA expression profile between two samples, many candidate genes of differential expression can be revealed. In this study, total RNA of the colonies was extracted and sequenced, using whole pancreas as control. We captured 7266 significantly differentially expressed mRNAs along with 285 miRNAs and 183 lncRNAs. Expression of a subset of identified genes was confirmed by qPCR to verify HTS reliability. We found that CD133 was enriched in the colonies (Fig. 4b) and only CD133+ cells could form colonies (Fig. 5b). We also found that most of the cells in ring colonies were CD133+ (Fig. 5a) and had a similar gene expression pattern to CD133+ cell-derived colonies (Fig. 5c). Because CD133+ cells should represent all of the cells with the ability to form pancreatic progenitor-like colonies, this implied that our colonies were proper materials for HTS.

In addition, we noted that Mettl10 is the most changed gene in upregulated mRNAs. METTL10 is a methyltransferase-like protein that trimethylates eukaryotic translation elongation factor 1 alpha 1(EF1A1), a eukaryotic elongation factor, which implies its contribution to many biological processes on a translational level [30]. This finding suggests that epigenetic regulation might play an important role in pancreatic progenitors. In order to find new surface markers, we verified quite a few cell surface proteins including signal pathway receptors, cell adhesion molecules and clusters of differentiation (CDs).

The miRNA showed an important role in beta-cell development [39]. A previous study showed that miR-21 regulates beta-cell death by targeting the tumor-suppressing gene Pdcd4 [46]. Nevertheless, little is known of whether miRNA plays important roles in regulating pancreatic progenitors. In our results, miR-21a, miR-31, miR-200c and miR-155 were upregulated and miR-217, miR-802, miR-375 and miR-216 were downregulated (Additional file 9: Table S5). One differentially expressed miRNA of particular interest is miR-802, a transcriptional factor highly expressed in the liver and pancreas that targets HNF1B to regulate glycometabolism [38]. In addition, HNF1B is a crucial factor in pancreas development [47] and loss of HNF1B exhibits pancreas hypoplasia [48]. Also, HNF1B is upstream of SOX9 and NEUROG3 [48, 49], indicating its latent function to regulate pancreatic progenitors. In our study, HNF1B is upregulated in the colonies shown by HTS (log2 = 4.55, p < 0.05). Thus, miR-802 might be a potential miRNA regulating pancreatic progenitors. Another miRNA, miR-375, is a well-characterized miRNA regulating pancreas development [39]. miR-375 knockdown lead to reduced endocrine cells [50]. In conclusion, our profiling of the miRNA transcriptome provided a vast miRNA candidate pool for further research.

Our study also identified several differentially expressed lncRNAs, a group of noncoding RNAs which may regulate gene expression through various mechanisms [25]. Several lncRNAs play a role in pancreatic cancer [26], beta-cell biology [27] and pancreas development [28]. Recently a lncRNA was shown to regulate specification and function of beta cells by targeting multiple transcriptional factors [51]. Our profile identified 183 differentially expressed lncRNAs. In most upregulated lncRNAs, Malat1 was shown to promote pancreatic cancer proliferation by stimulating autophagy [41], indicating its regulation effect on cell proliferation.

To reveal the functions of differentially expressed RNAs in our colonies, GO and KEGG analyses were performed. We note that the term “metabolic” was enriched in the biological process of both mRNAs and miRNAs. Also, in KEGG analysis of mRNA, the metabolic pathway was found to be significant. Metabolic pathways have been shown to have an important role in regulating stem cell functions [52]. Stemness is regulated by metabolic pathways in stem cells [53]. It is likely that pancreatic progenitors may share the same metabolic condition with other kinds of stem/progenitor cells. Regulation of metabolic pathways may also affect their functions.