The pancreas is a vital organ with exocrine and endocrine cell functions, and a root of lethal human diseases including diabetes mellitus, pancreatitis, and pancreatic ductal adenocarcinoma. Exocrine acinar cells produce digestive zymogens that are delivered to the intestines by a branching network of exocrine ductal cells that secrete bicarbonate and other products. Pancreatic endocrine functions derive from clusters of epithelial cells (islets of Langerhans) called α-, β-, δ-, and PP-cells that respectively synthesize, store, and secrete the hormones Glucagon, Insulin, Somatostatin, and Pancreatic polypeptide (Benitez et al., 2012). Insulin production by islet β-cells is highly regulated: key features of mature β-cells include preproinsulin (INS) transcription, proinsulin processing by endo- and exo-peptidases and storage of the proinsulin cleavage products insulin and C-peptide in dense core vesicles. Likewise, cardinal β-cell functions regulate insulin release in response to glucose and other secretagogues, including glucose sensing and metabolism through the enzyme glucokinase, and use of ATP-dependent potassium channels (KATP) and voltage-gated calcium channels to induce insulin exocytosis (reviewed in Suckale and Solimena, 2010). Deficiency or malfunctioning of β-cells produces impaired glucose regulation and diabetes mellitus, a disease with autoimmune (type 1, T1DM) and pandemic forms (type 2; Ashcroft and Rorsman, 2012). Thus, replacement or regeneration of functional human β-cells is an intensely-sought goal.
Human islet transplantation can be used to replace β-cell function in T1DM (reviewed in Vardanyan et al., 2010), but a shortage of donors currently precludes broad use of human pancreatic islets for β-cell replacement. Because of their expandability and multipotency, human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs) have been explored as sources of replacement insulin-producing cells (reviewed in Hebrok, 2012). However, directing the differentiation of these developmentally ‘primitive’ cells through multiple sequential fates into β-cell-like progeny that synthesize, process, store, and secrete insulin while lacking tumorigenic potential has challenged investigators worldwide (Fujikawa et al., 2005;McKnight et al., 2010; Cheng et al., 2012). Moreover, different hESC and iPSC cell lines exhibit significant variability during development into insulin-producing cells (Nostro and Keller, 2012). Recent work demonstrated that differentiated cell types in adult organs, including the mouse pancreas, can be experimentally ‘reprogrammed’ into progeny resembling islet cells, suggesting a new strategy for β-cell replacement (Vierbuchen and Wernig, 2011). For example, adult mouse pancreatic acinar cells can be converted into insulin-producing cells in vitro and in vivo (Minami et al., 2005; Zhou et al., 2008). However, little progress has been made in reprogramming primary human epithelial cells into different cell types, including conversion of pancreatic non-β-cells toward a human β-cell fate (Vierbuchen and Wernig, 2011). Thus, systems permitting expansion and genetic modulation of human pancreatic cells could powerfully influence studies of β-cell biology and replacement.
Pancreatic ducts constitute 30–40% of human pancreas and have been proposed as a potential source of replacement β-cells (Bouwens and Pipeleers, 1998; Bonner-Weir et al., 2004). During pancreas development, fetal endocrine cells derive from primitive ductal epithelium (reviewed by Pan and Wright, 2011; Pictet and Rutter, 1972). In addition, some studies have suggested that in adult mice, β-cells may be produced from pancreatic ductal epithelium (Inada et al., 2008; Xu et al., 2008; Rovira et al., 2010). However, recent lineage tracing evidences have not supported this view (Solar et al., 2009; Furuyama et al., 2011; Kopp et al., 2011). In humans, prior studies have suggested that adult human primary ductal cells in heterogeneous cell mixtures may harbor the potential to generate endocrine-like progeny (Bonner-Weir et al., 2000; Heremans et al., 2002; Swales et al., 2012), but interpretation in these studies was limited by the probability of islet cell contamination. Therefore, the potential for conversion of pancreatic ductal cells toward an endocrine fate remains unclear. Moreover, prior studies have revealed only limited proliferative capacity of primary human pancreatic ductal cells in culture (Rescan et al., 2005). Thus, despite their relative abundance, multiple practical issues have prevented development of human pancreatic ductal cells as a source of replacement β-cells.
Here we report that normal human adult pancreatic duct cells can be sorted, clonally expanded, and genetically converted into endocrine cells. Human insulin-producing cells (IPCs) produced from sorted duct cells exhibited hallmark features of functional neonatal β-cells including high-level preproinsulin (INS) expression, proinsulin processing and dense-core granule formation. Moreover, secretion of insulin and insulin C-peptide from IPCs is stimulated by glucose and KATP channel stimulants in a calcium-dependent manner. Together these studies reveal a new system for investigating human pancreatic duct cell biology, genetics, and β-cell regeneration.
Purification and expansion of primary CD133+ human pancreatic ductal cells
To identify human pancreatic epithelial cells that can be grown and maintained in culture, we systematically screened cell isolation methods and culture conditions with dispersed adult human pancreatic cells obtained from cadaveric donors without known pancreatic cancer, diabetes mellitus, or other pancreatic diseases (Table 1). With primary cells plated at low density, we observed formation of multicellular epithelial spheres, when cultured in Matrigel with a serum-free culture medium without feeder cells (‘Materials and methods’, Figure 1—figure supplement 1A). The multicellular sphere formation suggested primary cell expansion, so based on this assay we fractionated cells by fluorescence-activated cell sorting (FACS) to isolate and characterize sphere-forming pancreatic cells. A survey of cell surface markers used for fetal mouse pancreatic cell isolation (Sugiyama et al., 2007) revealed that antibodies recognizing CD133 enriched sphere-forming cells by four fold, whereas sphere-forming cells were depleted in the CD133neg fraction (Figure 1A,B). Immunohistochemical analysis of the human adult pancreas revealed CD133 expression at the apical portion of duct epithelial cells that co-expressed keratin 19 (KRT19), whereas CD133 was undetectable in islet endocrine cells or acinar cells (Figure 1C, Figure 1—figure supplement 1B), consistent with prior reports (Lardon et al., 2008). We have achieved sphere formation from over 35 consecutive adult donors (Table 1); thus, the sphere formation of primary adult human pancreatic CD133+ cells was highly reproducible.
To assess the properties of FACS-purified adult pancreatic CD133+ cells, we performed quantitative reverse transcription PCR (qRT-PCR). This revealed that CD133+ cells expressed high levels of mRNA encoding ductal markers (KRT19 and CAR2), while mRNAs expressed in acinar (CPA1 and CEL) and endocrine (CHGA, INS, and GCG) cells were exclusively enriched in the CD133neg fraction (Figure 1D, Figure 1—figure supplement 1C). Immunostaining confirmed that >98% of sorted CD133+ cells produced KRT19, whereas CD133+ cells produced no detectable islet hormone (Figure 1E,F, Figure 1—figure supplement 1D). Thus, FACS efficiently eliminated islet endocrine and acinar cells, and enriched for a population of primary adult pancreatic duct cells that expanded as epithelial spheres in feeder- and serum-free culture.
Maintenance of ductal phenotypes by self-renewing human pancreatic CD133+ cells
After commencing in vitro cultures, the epithelial spheres from CD133+ ductal cells attained diameters ranging from 40 to 520 µm in 2 weeks (Figure 1A, Figure 1—figure supplement 1A and Figure 2—figure supplement 1A). Spheres 350–500 µm in diameter were composed of 1470 ± 310 cells (n = 5); thus, based on evidence of clonal expansion (see below), we calculated that spheres resulted from a minimum of 10 cell divisions in 2 weeks. Sphere epithelium maintained KRT19 protein expression and a polarized monolayer as indicated by apical localization of CD133 (Figure 2A, Figure 2—figure supplement 1A,D). Neither acinar (CPA1) nor islet endocrine (CHGA and insulin C-peptide) markers were detectable (Figure 3C and data not shown), suggesting epithelial cells in cultured spheres maintained ductal characteristics.
To assess whether sphere growth was achieved by cell proliferation or by other mechanisms like cell migration and aggregation, we analyzed spheres by immunostaining and time-lapse imaging. Immunohistochemistry revealed the proliferation marker Ki-67 in more than 25% of cells comprising 2-week-old spheres (Figure 2A, Figure 2—figure supplement 1B; labeling index 26.5 ± 5.1%), data further supported by detection of a second proliferation marker, phospho-histone H3 (Figure 2A). Time-lapse imaging revealed that spheres arose from single cells (Figure 2B), providing strong evidence that sphere formation resulted from CD133+ ductal cell proliferation, rather than through cell migration and aggregation. Enzymatic dispersion of 2-week-old G1 spheres and subsequent culture revealed that the spheres can be passaged up to seven generations (G7, 3 months) and that the total number of cells increased with each generation (Figure 2C,D, Figure 2—figure supplement 1C). After G7, ductal cell expansion was not achieved, and the spheres were not formed (Figure 2—figure supplement 1C and data not shown), supporting the view that ductal epithelial cells are not immortalized, and consistent with the origin of pancreatic cells from donors without neoplasia.
Neurog3 converts pancreatic duct cells into progeny expressing islet hormones
The endocrine potential of human or mouse pancreatic ductal cells remains controversial. To investigate the potential of purified human pancreatic ductal cells to achieve an endocrine fate, we used an adenovirus-mediated transgenic system. Neurog3 is a transcription factor necessary and sufficient for pancreatic endocrine cell differentiation in vivo (Gradwohl et al., 2000; Gu et al., 2002) and, combined with other factors, can induce pancreatic acinar-to-islet cell conversion in mice (Zhou et al., 2008). To test if Neurog3expression could respecify human duct cells toward an endocrine fate, we infected cultured spheres as well as primary CD133+ cells with recombinant adenovirus co-expressing red fluorescent protein and Neurog3(Ad-RFP-Neurog3), and assessed changes in gene expression by qRT-PCR (Figure 3A–C and 4C).Neurog3 induced the expression of NEUROD1, INSM1, and RFX6 (Figure 3C), genes whose mouse homologs are known direct targets of Neurog3 in pancreas development (Mellitzer et al., 2006; Smith et al., 2010). Ad-RFP-Neurog3 infection induced expression of the pan-endocrine markers chromogranin A (CHGA) and synaptophysin in both primary CD133+ duct cells and cultured spheres (Figures 3C and 4C, and data not shown). Ad-RFP-Neurog3 infection also induced expression of mRNA encoding PAX4 and NKX2.2, transcriptional regulators of pancreatic endocrine cell fate (Sosa-Pineda et al., 1997; Sussel et al., 1998), and mRNA encoding crucial β-cell factors such as the prohormone processing enzymes PCSK1(PC1/3) and PCSK2 (PC2), KATP channel components KCNJ11 (KIR6.2) and ABCC8 (SUR1), and glucokinase (GCK) (Figure 4D). Moreover, Ad-RFP-Neurog3 significantly induced mRNA encoding the pancreatic hormones ghrelin and somatostatin, but not mRNAs encoding insulin, glucagon, PPY or the intestinal hormones cholecystokinin and gastrin (Figures 3C and 4D, Figure 4—figure supplement 1A, and data not shown). These findings support the conclusion that human adult pancreatic ductal cells harbor pancreatic endocrine potential upon induction of Neurog3.
Immunostaining confirmed these qRT-PCR findings and demonstrated that only RFP+ cells produced by Ad-RFP-Neurog3 infection were immunostained with antibodies recognizing NEUROD1, NKX2.2, CHGA, SST or GHRL (Figure 3B,D, Figure 3—figure supplement 1). We also confirmed that no insulin-, glucagon- or PPY-positive cells were observed by immunostaining (data not shown). While only a subset of cells infected with Ad-RFP-Neurog3 (RFP+) expressed CHGA, we noted all GHRL+ or SST+ cells co-expressed CHGA (Figure 3D). Quantification of CHGA+ and hormone+ cells revealed that 30% of infected cells (RFP+) expressed CHGA. At least 45% of CHGA+ cells produced SST or GHRL, and less than 2% of CHGA+ cells expressed both hormones (Figure 3D,E). Thus, Neurog3 expression efficiently converted primary human ductal cells and cultured ductal epithelial spheres into hormone-expressing cells with cardinal features of endocrine pancreas.
In mice, Neurog3 gene dosage can determine commitment between exocrine and endocrine lineages in pancreas development (Wang et al., 2010). Therefore, we next assessed the possibility that the 70% of RFP+cells infected by Ad-RFP-Neurog3 failing to express CHGA may have achieved inadequate levels of Neurog3 expression. We fractionated cells produced by Ad-RFP-Neurog3 infection by RFP intensity and measured mRNA expression of Neurog3, CHGA, SST and GHRL by qRT-PCR (Figure 3F,G). We found that cell fractions with the highest levels of RFP expression (‘P4 and P5’, Figure 3F) had the highest levels of mouse Neurog3 mRNA, and only these cell fractions produced mRNA encoding CHGA, SST or GHRL (Figure 3G). These data suggest that relatively high threshold levels of Neurog3 may be necessary and sufficient for directing endocrine differentiation of human pancreatic cells.
Conversion of pancreatic duct cells into progeny that produce, process, and store insulin
The transcription factors MafA, Neurog3, and Pdx1 (a combination hereafter summarized as ‘MNP’) were sufficient to convert adult mouse acinar cells into insulin-producing cells (IPCs: Zhou et al., 2008). We constructed three adenoviruses expressing MafA, Neurog3, or Pdx1 (see ‘Materials and methods’; Figure 4A), and infected cultured spheres with this MNP combination. Within 5 days after infection, we reproducibly detected INS mRNA induction but at extremely low levels relative to adult human islet controls (0.0035 ± 0.0012% of islet levels; Figure 4B). Thus, we sought additional factors and discovered that mRNA encoding PAX6, an important regulator of mouse pancreatic endocrine cell development (Sander et al., 1997), was induced by MNP to only 0.03% of levels in control islets (Figure 4—figure supplement 1A). When combined with MafA, Neurog3, and Pdx1 (encoded in four viruses, ‘4V’), Pax6 induced INS expression in primary CD133+ ductal cells or spheres by over 30-fold relative to MNP (Figure 4A,C,D). We observed ductal conversion to IPCs with four consecutive, independent donors (INS, Figure 4D). We also detected substantially increased expression of other islet endocrine markers, including SST, GCK, PCSK1, KCNJ11, and ABCC8 (Figure 4D). Immunohistochemical analyses demonstrated that the number of Insulin+ cells was increased by 18 to 20-fold in spheres transduced by the four factor combination (4V) compared to the MNP combination (Figure 4F,G). ELISA studies quantified and confirmed this increase of proinsulin levels, showing that the spheres derived from 4V exposure contained proinsulin levels that averaged 0.7% of those in human islets (Figure 5E). Systematic removal of individual factors from this four virus combination revealed that omission of Neurog3 prevented expression of INS, CHGA or SST (Figure 4E). Omission of virus expressing MafA or Pax6 from this combination significantly reduced INS expression (Figure 4E–G), whereas omission of virus expressing Pdx1 did not significantly decrease INS expression. Thus, Neurog3-mediated endocrine cell conversion is required for the production of IPCs as well as other hormone-producing cells from ductal spheres.
0 comments:
Post a Comment