Multipotential Nestin-Positive Stem Cells Isolated From Adult Pancreatic Islets Differentiate Ex Vivo Into Pancreatic Endocrine, Exocrine, and Hepatic Phenotypes

  1. Henryk Zulewski1,
  2. Elizabeth J. Abraham2,
  3. Melissa J. Gerlach2,
  4. Philip B. Daniel2,
  5. Wolfgang Moritz3,
  6. Beat Müller4,
  7. Mario Vallejo5,
  8. Melissa K. Thomas2 and
  9. Joel F. Habener2
  1. 1Division of Endocrinology and Diabetes, University Hospital of Geneva, Geneva, Switzerland
  2. 2Laboratory of Molecular Endocrinology, Massachusetts General Hospital, Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts
  3. 3Department of Surgery, University Hospital of Zürich, Zürich
  4. 4Division of Endocrinology, Diabetology and Metabolism, Department of Internal Medicine, University Hospitals, Basel, Switzerland
  5. 5Institute for Biomedical Research, Superior Council of Scientific Research, Madrid, Spain

    Abstract

    The endocrine cells of the rat pancreatic islets of Langerhans, including insulin-producing β-cells, turn over every 40–50 days by processes of apoptosis and the proliferation and differentiation of new islet cells (neogenesis) from progenitor epithelial cells located in the pancreatic ducts. However, the administration to rats of islet trophic factors such as glucose or glucagon-like peptide 1 for 48 h results in a doubling of islet cell mass, suggesting that islet progenitor cells may reside within the islets themselves. Here we show that rat and human pancreatic islets contain a heretofore unrecognized distinct population of cells that express the neural stem cell–specific marker nestin. Nestin-positive cells within pancreatic islets express neither the hormones insulin, glucagon, somatostatin, or pancreatic polypeptide nor the markers of vascular endothelium or neurons, such as collagen IV and galanin. Focal regions of nestin-positive cells are also identified in large, small, and centrolobular ducts of the rat pancreas. Nestin-positive cells in the islets and in pancreatic ducts are distinct from ductal epithelium because they do not express the ductal marker cytokeratin 19 (CK19). After their isolation, these nestin-positive cells have an unusually extended proliferative capacity when cultured in vitro (∼8 months), can be cloned repeatedly, and appear to be multipotential. Upon confluence, they are able to differentiate into cells that express liver and exocrine pancreas markers, such as α-fetoprotein and pancreatic amylase, and display a ductal/endocrine phenotype with expression of CK19, neural-specific cell adhesion molecule, insulin, glucagon, and the pancreas/duodenum specific homeodomain transcription factor, IDX-1. We propose that these nestin-positive islet-derived progenitor (NIP) cells are a distinct population of cells that reside within pancreatic islets and may participate in the neogenesis of islet endocrine cells. The NIP cells that also reside in the pancreatic ducts may be contributors to the established location of islet progenitor cells. The identification of NIP cells within the pancreatic islets themselves suggest possibilities for treatment of diabetes, whereby NIP cells isolated from pancreas biopsies could be expanded ex vivo and transplanted into the donor/recipient.

    The mammalian pancreas consists of three distinct tissue types: the ductal tree, the exocrine acini that produce digestive enzymes, and the endocrine islets of Langerhans. Embedded in the exocrine tissue are the islets (which contain α-, β-, δ-, and PP-cells that produce the hormones glucagon, insulin, somatostatin, and pancreatic polypeptide, respectively) involved in the regulation of physiological nutrient homeostasis (1). Ductal cells of the adult pancreas include latent progenitor cells of the islet endocrine cells that can be induced to differentiate into islet endocrine cells given the appropriate morphogen stimuli—a process referred to as neogenesis (26). The differentiation of duct cells of the pancreas into endocrine hormone-producing cells is believed to recapitulate the embryonic development (ontogeny) of the pancreas, whereby the exocrine and endocrine pancreases arise from the differentiation and proliferation of patterned endodermal cells in the early embryonic foregut that first form a ductal tree by branching morphogenesis (1). During early embryonic development, neural and islet cells share many phenotypic properties. Developing islet cells express several neuronal-specific markers such as synaptophysins, nerve-specific enolase, the catechol-synthesizing enzymes tyrosine hydroxylase, dopamine decarboxylase, phenylethylnolamine methyl transferase (7), and the transcription factors Isl-1, Brain-4, Pax 6, Pax 4, Beta2/NeuroD, and IDX-1 (813).

    Recently, pluripotential stem cells have been identified in the brain that are capable of differentiating into either neuronal or glial tissues (4,12). A special characteristic of neural stem cells is that they express the protein nestin, an intermediate filament protein (14,15). Nestin is expressed in the neural tube of the developing rat embryo at embryonic day 11 (E11), reaches maximum levels of expression in the cerebral cortex at E16, and decreases in expression in the adult cortex, becoming restricted to a population of ependymal cells (14). Because of phenotypic similarities between developing neural and islet cells, we examined rat pancreatic islets for the presence of nestin-expressing cells. Here we demonstrate the existence of a distinct population of cells within islets and in focal regions of the pancreatic ducts and exocrine pancreas that express nestin and have an extended capacity for proliferation in vitro. These cells derived from islets have properties of stem cells and can differentiate in culture into cells with liver, pancreatic exocrine/ductal, and endocrine phenotypes. The differentiated cells express several liver and exocrine pancreatic markers, such as α-fetoprotein, transthyretin, carboxypeptidase, the ductal marker cytokeratin 19 (CK19), and neural cell adhesion molecule. They also secrete detectable levels of islet hormones, such as insulin, glucagon, and glucagon-like peptide 1 (GLP-1), as well as express the transcription factor IDX-1 (also known as PDX-1, IPF-1, and STF-1) (1619). Our observations described herein provide evidence that pancreatic islets themselves, apart from the ducts, also contain multipotential progenitor cells. These findings may have implications for enhancing engraftment of isolated islets in diabetic individuals by providing new insights into modifications of islet preparations used in ongoing clinical islet transplantation studies.

    RESEARCH DESIGN AND METHODS

    Animals.

    Male Sprague-Dawley rats, 8–9 weeks old and weighing ∼200 g (Taconics, Germantown, NY), were obtained for preparation of pancreatic tissue sections for immunocytochemistry and for the isolation of pancreatic islets for tissue culture. Timed pregnant female Sprague-Dawley rats (Charles River, Wilmington, MA) were obtained for isolation of fetal pancreases at E16.

    Isolation and culture of pancreatic islets.

    Rat pancreases were removed (postmortem) and dissected into 2- to 3-mm segments, and islets were isolated by the collagenase digestion method of Lacy and Kostianovsky (20). Human islet tissue was obtained from the islet distribution program of the Cell Transplant Center, Diabetes Research Institute, University of Miami School of Medicine (Miami, FL), and the Juvenile Diabetes Foundation Center for Islet Transplantation, Harvard Medical School (Boston, MA). Thoroughly washed islets were handpicked, suspended in modified RPMI 1640 media (11.1 mmol/l glucose) supplemented with 10% fetal bovine serum, 10 mmol/l HEPES buffer, 1 mmol/l sodium pyruvate, antibiotic-antimycotic (Gibco Life Technologies, Gaithersburg, MD), and 71.5 μmol/l β-mercaptoethanol (Sigma, St. Louis, MO), and added to 12-well tissue culture plates (Falcon 3043; Becton Dickinson, Lincoln Park, NJ) that had been coated with concanavalin A (ConA). The islet preparation was incubated for 96 h at 37°C with 95% air and 5% CO2. In these conditions, most of the islets remained in suspension (floated), whereas fibroblasts and other non-islet cells attached to the substratum. After 96 h of incubation, the media containing the suspended islets were carefully removed, and the islets were manually picked and resuspended in the modified RPMI 1640 media, now further supplemented with 20 ng/ml each of basic fibroblast growth factor (bFGF) and epidermal growth factor (EGF). The islet suspension (containing 20–30 islets per well) was added to 12-well plastic tissue culture plates not coated with ConA. The islets immediately adhered to the surfaces of the plates. Within several days, a monolayer of cells was observed growing out and away from the islets. Cells in the monolayers—nestin-positive islet-derived progenitor (NIP) cells—were repeatedly recloned and expanded: 10 times over 8.5 months (rat) and 7 times over 8 months (human). In certain instances, human NIP cells were cultured in modified RPMI media containing 2.5 mmol/l glucose and in several growth factor combinations that include activin-A (2 nmol/l), hepatocyte growth factor (HGF) (100 pmol/l), or betacellulin (500 pmol/l). In other instances, NIP cells were challenged with nicotinamide (10 mmol/l), exendin-4 (10 nmol/l), activin-A, and HGF in media (11.1 mmol/l glucose) containing no serum. Dexamethasone (10 μmol/l) treatments were also administered in modified RPMI media containing no serum.

    Antisera.

    We used mouse monoclonal antibodies to human CK19 (clone K4.627; Sigma). The rabbit polyclonal antisera to rat nestin and to IDX-1 were prepared by immunizations of rabbits with a purified GST rat nestin fusion protein or the last 12 amino acids of rat IDX-1, respectively (21). The anti-human nestin antiserum was a generous gift from Dr. C.A. Messam (National Institute of Neurological Disorders and Stroke [NINDS], National Institutes of Health, Bethesda, MD). Guinea pig anti-insulin and anti-pancreatic polypeptide antisera were obtained from Linco (St. Charles, MO). Mouse antiglucagon and rabbit antisomatostatin antisera were purchased from Sigma and DAKO (Carpinteria, CA), respectively. The mouse anti-human galanin and collagen IV antisera were purchased from Peninsula Laboratories (Belmont, CA) and Caltag Laboratories (San Francisco, CA).

    Immunocytochemistry.

    Cryosections (6 μm) prepared from E16 and adult (60-day) rat pancreases and cells were fixed with 4% paraformaldehyde in phosphate buffer. Cells were first blocked with 3% normal donkey serum for 30 min at room temperature and incubated with primary antisera overnight at 4°C. Sections and cells were rinsed off with phosphate-buffered saline (PBS) and incubated with the respective Cy3 and Cy2 labeled secondary donkey antisera for 1 h at room temperature. Slides were then washed with PBS and coverslipped with fluorescent mounting medium (Kirkegaard and Perry Laboratories, Gaithersburg, MD). Tissue sections were incubated overnight at 4°C with primary antisera. Primary antisera were then rinsed with PBS, and slides were blocked with 3% normal donkey serum for 10 min at room temperature before incubation with donkey anti-Cy3 (indocarbocyanine) or Cy2 and either anti–guinea pig (insulin), anti-mouse (glucagon), or anti-sheep (somatostatin) sera DTAF (Jackson ImmunoResearch Laboratories, West Grove, PA) for 30 min at room temperature. Slides were then rinsed with PBS and coverslipped with fluorescent mounting medium (Kirkegaard and Perry Laboratories). Fluorescence images were obtained using a Zeiss Epifluorescence microscope equipped with an Optronics TEC-470 CCD camera (Optronics Engineering, Goleta, CA) interfaced with a PowerMac 7100 installed with IP Lab Spectrum analysis software (Signal Analytics, Vienna, VA).

    Reverse transcription and polymerase chain reaction.

    Total cellular RNA prepared from rat or human islets and cell cultures was reverse transcribed and amplified by polymerase chain reaction (PCR) for 35 cycles as described previously (22). Oligonucleotides used as amplimers for the PCR and for subsequent Southern blot hybridization were as follows. Rat nestin: forward, 5′gcggggcggtgcgtgactac 3′; reverse, 5′aggcaagggggaagagaaggatgt 3′; hybridization, 5′aagctgaagccgaatttccttgggataccagagga 3′. Rat keratin 19: forward, 5′acagccagtacttcaagacc 3′; reverse, 5′ctgtgtcagcacgcacgtta 3′; hybridization, 5′tggattccacaccaggcattgaccatgcca 3′. Rat neural cell adhesion molecule (NCAM): forward, 5′cagcgttggagagtccaaat 3′; reverse, 5′ttaaactcctgtggggttgg 3′; hybridization, 5′aaaccagcagcggatctcagtggtgtggaacgatgat 3′. Rat IDX-1: forward, 5′atcactggagcagggaagt 3′; reverse, 5′gctactacgtttcttatct 3′; hybridization, 5′gcgtggaaaagccagtggg 3′. Human nestin: forward, 5′agaggggaattcctggag 3′; reverse, 5′ctgaggaccaggactctcta 3′; hybridization, 5′tatgaacgggctggagcagtctgaggaaagt 3′. Human keratin: forward, 5′cttttcgcgcgcccagcatt 3′; reverse, 5′gatcttcctgtccctcgagc 3′; hybridization, 5′aaccatgaggaggaaatcagtacgctgagg 3′. Human glucagon: forward, 5′atctggactccaggcgtgcc 3′; reverse, 5′agcaatgaattccttggcag 3′; hybridization, 5′cacgatgaatttgagagacatgctgaaggg 3′. Human E-cadherin: forward, 5′ agaacagcacgtacacagcc 3′; reverse, 5′cctccgaagaaacagcaaga 3′; hybridization, 5′ tctcccttcacagcagaactaacacacggg 3′. Human transthyretin: forward, 5′ gcagtcctgccatcaatgtg 3′; reverse, 5′ gttggctgtgaataccacct 3′; hybridization, 5′ ctggagagctgcatgggctcacaactgagg 3′. Human pancreatic amylase: forward, 5′gactttccagcagtcccata 3′; reverse, 5′ gtttacttcctgcagggaac 3′; hybridization, 5′ ttgcactggagaaggattacgtggcgttcta 3′. Human procarboxypeptidase: forward, 5′ tgaaggcgagaaggtgttcc 3′; reverse, 5′ ttcgagatacaggcagatat 3′; hybridization, 5′ agttagacttttatgtcctgcctgtgctca 3′. Human synaptophysin: forward, 5′ cttcaggctgcaccaagtgt 3′; reverse, 5′ gttgaccatagtcaggctgg 3′; hybridization, 5′ gtcagatgtgaagatggccacagacccaga 3′. Human HGF: forward, 5′ gcatcaaatgtcagccctgg 3′; reverse, 5′ caacgctgacatggaattcc 3′; hybridization, 5′ tcgaggtctcatggatcatacagaatcagg 3′. Human cMET (HGF receptor): forward, 5′ caatgtgagatgtctccagc 3′; reverse, 5′ ccttgtagattgcaggcaga 3′; hybridization, 5′ ggactcccatccagtgtctccagaagtgat 3′. Human XBP-1: forward, 5′gagtagcagctcagactgcc 3′; reverse, 5′ gtagacctctgggagctcct 3′; hybridization, 5′ cgcagcactcagactacgtgcacctctgca 3′. Human GLUT2: forward, 5′ gcagctgctcaactaatcac 3′; reverse, 5′ tcagcagcacaagtcccact 3′; hybridization, 5′ acgggcattcttattagtcagattattggt 3′. Human insulin: forward, 5′ aggcttcttctacaca 3′; reverse, 5′ caggctgcctgcacca 3′; hybridization, 5′ aggcagaggacctgca 3′.

    Primers were selected from two different exons and encompassed at least one intronic sequence. In addition, a reverse transcription (RT) minus control was run for most samples. PCR cycling was at 94°C for 1 min followed by 94°C for 10 s, 58/56/54°C for 10 s, 72°C for 1 min (35 cycles), and 72°C for 2 min. The annealing temperature was 58°C for rat nestin, 45°C for human insulin, and 56/54°C for the remaining primer pairs.

    For Southern hybridization, oligonucleotide probes were labeled with T4 polynucleotide kinase and [γ-32P]ATP. Radiolabeled probes were hybridized to PCR products that had been transferred to nylon membranes at 37°C for 1 h, then washed in 1 × sodium chloride–sodium citrate (SSC) + 0.5% SDS at 55°C for 10–20 min or 0.5 × SSC + 0.5% SDS at 42°C for the human PCR products.

    Radioimmunoassays.

    Insulin and glucagon concentrations in culture media were determined by ultrasensitive radioimmunoassay (RIA) kits purchased from Linco Research and Diagnostics Products, respectively. The antisera supplied in the respective kits are guinea pig anti-human insulin and rabbit anti-human glucagon. GLP-1 secretion was measured with an anti-human GLP-1(7-36) amide rabbit polyclonal antiserum raised by immunization of a rabbit with a synthetic peptide CFIAWLVKGR amide conjugated to keyhole limpet hemocyanin. The antiserum is highly specific for the detection of GLP-1(7-36) amide and only weakly detects proglucagon. The sensitivity levels for the insulin, glucagon, and GLP-1 assays are 8, 13, and 10 pg/ml, respectively. The interassay variability for the insulin RIAs was 15%.

    RESULTS

    Stem cell marker nestin expression within pancreatic islets.

    Nestin expression was observed by immunocytochemical staining of a distinct population of cells within developing islet clusters of E16 rat pancreas (Fig. 1A) and in islets of the adult rat pancreas (60 days postnatally) (Fig. 1B). The nestin-positive cells are distinct from β-, α-, δ-, and PP-cells because they do not costain with antisera to the hormones insulin (Figs. 1A and B), glucagon, somatostatin, or pancreatic polypeptide (data not shown). The nestin-positive cells also do not costain with antisera to collagen IV, a marker for vascular endothelial cells (Fig. 1C); with an antiserum to galanin, a marker for nerve cells (data not shown); or with a monoclonal antibody to CK19, a specific marker for ductal cells (Figs. 6A and B). Nestin-positive staining is associated with distinct cells within the islets clearly observed by nuclear costaining (Fig. 1D). To confirm the immunocytochemical identification of nestin expression in pancreatic islets, we performed an RT-PCR of the nestin mRNA using total RNA prepared from freshly isolated rat islets and human islet tissue. The RT-PCR generated products of the correctly predicted size (Fig. 1E, upper panels) that were confirmed by Southern blotting (Fig. 1E, lower panels) and by DNA sequencing of the products (data not shown). Thus, we identified a new cell type in pancreatic islets that expresses nestin and may represent an islet pluripotential or multipotential stem cell similar to the nestin-positive stem cells in the central nervous system.

    Isolation and proliferation of NIP cells in vitro

    Having established the presence of nestin-expressing cells within islets, we next sought to determine whether these cells have the potential to proliferate in vitro, another characteristic feature of stem cells. To pursue this notion, islets prepared from 60-day-old rats or obtained from a normal adult human were first plated on ConA-coated dishes and cultured in modified RPMI 1640 medium containing 10% fetal bovine serum for 4 days to purge the islet preparation of fibroblasts and other non-islet cells that adhered to the ConA-coated plates. The islets that did not adhere to the plates under these culture conditions were collected and transferred to 12-well plates (without ConA coating) containing the same modified RPMI 1640 medium now additionally supplemented with bFGF and EGF (20 ng/ml each). The growth factors bFGF and EGF together were selected because they are known to stimulate the proliferation of neural stem cells derived from ependyma of the brain (13). The islets attached to the plates and cells slowly grew out of the islet as a monolayer. The outgrowing monolayer of cells (Fig. 2A, panel 1) expressed nestin (Fig. 2A, panel 2). Rat cells were picked from the monolayer (batches of at least 5–10 cells), subcloned into 12-well plates, and incubated with the modified RPMI 1640 medium (11.1 mmol/l glucose) containing bFGF and EGF. Figure 2B shows a time lapse series of representative images of the same field of rat NIP cells taken every 24 h for 18 days. Only days 2, 5, 12, 15, and 18 are shown for sake of brevity. The subcloned cells were attached at 2 days, grew slowly up to 5 days, and then rapidly proliferated, dividing every 12–15 h, as determined by counting the cells in the wells, and became confluent by 10–11 days. After attaining confluence, the cells migrated to form wave-like structures, and after 15–18 days of culture, the cells formed spherical clusters (SCs) (Fig. 2B, bottom right panel). This proliferation behavior of the NIP cells is reminiscent of that of marrow stromal cells (described recently by Colter et al. [23]), which are pluripotential stem cells.

    Similar cells were cloned from human islets. The subcloned human cells expressed nestin (Fig. 2C) but not insulin (data not shown). By immunostaining, a small subpopulation of subcloned cells expressed the homeodomain protein IDX-1, possibly reflecting early stages of the differentiation process; however, the majority of the cells did not stain for IDX-1 (not shown). Upon reaching confluence (Fig. 2D, panel 1), the human cells migrated to form large vacuolated structures in the dish (Fig. 2D, panels 2 and 3). Then the cells lining the large spaces changed morphology, rounded, and aggregated together forming three-dimensional SCs (Fig. 2D, panels 4–6). It is important to note that monolayers of both rat and human NIP cells were cloned and recloned and expanded for 10 and 7 consecutive cycles, respectively.

    Differentiation of NIP cells toward endocrine or ductal/exocrine pancreatic phenotypes.

    We characterized indicators of differentiation of NIP cells that formed the SCs by RT-PCR and Southern blot and found that they express the endocrine marker NCAM (24) (Fig. 3A, right panel) and the ductal cell marker CK19 (2527) (Fig. 3A, left panels). At this stage of study, we concluded that when the NIP cells became confluent and aggregated into SCs, they began to express pancreatic genes (NCAM and CK19) but may have been limited in expression of islet genes because of the absence of growth factors essential for their differentiation to endocrine cells. We also recognized that the differentiation of a progenitor cell population typically requires first a proliferative phase and then quiescence of proliferation in the presence of differentiation-specific morphogen growth factors. Therefore, we modified the culture conditions in some instances by replacing the media containing 11.1 mmol/l glucose, bFGF, and EGF, with media containing lower glucose (2.5 mmol/l),HGF/scatter factor, betacellulin, activin-A, exendin-4, or nicotinamide. It is important to appreciate that glucose is a known proliferative factor for pancreatic islet β-cells (28,29) and that HGF/scatter factor, activin-A, and exendin-4 have been shown to differentiate the pancreatic ductal cell line AR42J into an endocrine phenotype that produces insulin, glucagon, and other pancreatic endocrine cell proteins (3032).

    We found that human NIP cultures containing SCs also expressed the pancreas-specific homeodomain protein IDX-1 by immunocytochemistry (Fig. 3B), RT-PCR, and Southern blot (Fig. 3C), and by Western immunoblot (Fig. 3D). The expression of IDX-1 is of particular importance because it is recognized to be a master regulator of pancreas development and to be required for the maturation and functions of the pancreatic islet β-cells that produce insulin (33).

    Some cultures of NIP cells containing SCs also expressed the mRNA encoding proglucagon and insulin, as seen by RT-PCR (Figs. 4A and B), and secreted small amounts of immunoreactive glucagon, GLP-1, and insulin (Table 1). Moreover, incubation of the cell clusters for 6 days in nicotinamide, as described by Ramiya et al. (34), or 3 days with a combination of growth factors containing activin-A, HGF, nicotinamide, and exendin-4 increased insulin secretion by two- to fivefold (Table 1 and Fig. 4C). Several additional pancreatic markers were expressed in differentiated NIP cells, such as GLUT2 (35), synaptophysin, and HGF (36), as shown in Fig. 5. To determine whether the differentiating NIP cells may have properties of pancreatic exocrine tissue, we used RT-PCR and detected the expression of amylase and procarboxypeptidase (Fig. 5).

    Differentiation of NIP cells toward hepatic phenotypes.

    Because of the reported apparent commonalties between hepatic stem cells (oval cells), hepatic stellate cells, and progenitor cells in the pancreas, and the observations that after some injuries, the regenerating pancreas undergoes liver metaplasia (1,3739), we performed RT-PCR to detect liver-expressed genes in the SCs. PCR products were obtained for XBP-1, a transcription factor required for hepatocyte development (40), and transthyretin, a liver acute-phase protein. Several other liver markers were also expressed, such as α-fetoprotein (41), E-cadherin (42), c-MET (43), HGF (44), and synaptophysin (35) (Fig. 5) The expression of proteins shared by the pancreas and liver, such as HGF and synaptophysin, may reflect their common origin from the embryonic foregut endoderm and represent differentiation toward either pancreatic or hepatic phenotypes.

    Nestin expression is also localized within limited focal regions of pancreatic ducts.

    Because the neogenesis of new islets is also known to occur by differentiation of cells in pancreatic ducts, particularly during the neonatal period (rats and mice) but to some extent throughout adult life (36), we looked for nestin expression in the pancreatic ducts of adult rats. By dual fluorescence immunocytochemistry with antisera to nestin and to CK19, a marker of ductal epithelium, nestin is strongly expressed in cells in localized regions of both the large and small ducts as well as in some centrolobular ducts within the exocrine acinar tissue (Figs. 6AC). Remarkably, the localized regions of nestin expression in the ducts are mostly devoid of staining for CK19. Further, the nestin-positive cells in the ducts appear to have a morphology that is distinct from that of the epithelial cells. The epithelial cells consist of a homogenous population of cuboid rounded cells, whereas the nestin-positive cells are nucleated, serpiginous, and appear to reside in the interstices among or around epithelial cells (Fig. 6C).

    Thus, CK19 is not expressed in the majority of ductal cells that express nestin, suggesting that these nestin-expressing cells located within the pancreatic ducts are a passenger population of cells distinct from the ductal epithelial cells and may represent stem cells that have not yet differentiated into a ductal or endocrine phenotype. The finding of localized populations of nestin-expressing cells within the pancreatic ducts (and islets) of the adult rat pancreas further supports the idea that rat pancreatic ducts contain cells that are progenitors of islet cells (neogenesis), but these progenitors may not be a subpopulation of ductal epithelial cells per se.

    DISCUSSION

    Here we demonstrate the presence of a distinct cell type that expresses the neural stem cell–specific marker nestin within rat and human pancreatic islets and ducts with an extended capacity to proliferate in vitro. These nestin-positive cells, when isolated from islets, can differentiate in vitro to cells that express pancreatic endocrine markers, such as GLUT2, insulin, glucagon, and the homeodomain transcription factor IDX-1 as well as a number of pancreatic exocrine and hepatic genes. These stem cell–like pluripotential cells may be similar to the islet progenitor stem cells (IPSCs) described by Pour (45), Cornelius et al. (46), and Ramiya et al. (34) and possibly related to the cultured human islet buds (CHIBs) described by Bonner-Weir et al. (47), all of which were derived from ducts. However, the cells that we identified in the ducts and islets in the pancreas and describe herein do not yet appear to have a ductal phenotype. Although nestin is expressed in some cells in the pancreatic ducts, the ductal marker CK19 is not expressed in the majority of these cells and also is not expressed in any of the nestin-positive cells located within the islets. These findings imply that these nestin-expressing cells located within the pancreatic ducts are undifferentiated cells that are distinct from ductal epithelial cells. We suggest that the nestin-positive cells may be distinct multipotential stem or progenitor cells that have not yet differentiated into either a ductal or an endocrine pancreatic or hepatic phenotype.

    We speculatively propose two possibilities for the origin (neogenesis) of endocrine cells from the pancreatic ducts (Fig. 7). One is that regionalized populations of ductal epithelial cells can become endocrine cells given the appropriate morphogenic stimuli. The other possibility is that the progenitors of endocrine cells consist, at least in part, of distinct stem cells apart from the ductal epithelial cells that reside within the interstices of the ductal epithelial cells and within the surrounding periductal lamina. We favor the concept that the stem/progenitor cells that can become islet endocrine cells are not a specialized form of ductal epithelial cells but rather are a distinct population of cells that reside within the interstices of the ductal epithelium and within the islets themselves. Notably, a now recognized property of stem cells is that they are highly motile (48). They move rapidly within tissues, in some aspects similar to macrophages. We propose that the pancreatic stem cells may likewise migrate within the islets and ducts and when they reach a specific mesenchymal niche and sense appropriate paracrine morphogen signals, are induced to differentiate into endocrine cells. Our bias is based on the observations of the distinct serpiginous morphology of the nestin-positive cells that is so unlike the cuboid morphology of the ductal epithelial cells or the rounded morphology of mature endocrine cells of the islets.

    It is possible that the stem cell–like multipotential cells reportedly derived from pancreatic ducts are actually the subpopulation of nestin-positive cells in the ducts (28,39). The conditions of ex vivo growth used by these investigators may have resulted in a more rapid differentiation to a ductal phenotype than we have found. Nonetheless, the finding of localized populations of nestin-expressing cells within the pancreatic ducts of the adult rat pancreas further supports the idea that cells located within pancreatic ducts and islets are progenitors of islet cells (neogenesis).

    An important property of stem cells is their ability for self-renewal and differentiation into specific cell lineages. We subcloned rat and human NIP cells, maintained them in culture for over 8 months (in the presence of bFGF and EGF and 11.1 mmol/l glucose), and found that many (a subpopulation) of these cells continued to express nestin. When the cells were allowed to become confluent and to form clusters (a process that was apparently accelerated by the removal of bFGF and EGF and adding the differentiation factors such as HGF, activin-A, exendin-4, and nicotinamide), we observed expression of NCAM, the ductal marker CK19, and the IDX-1 transcription factor. In fact, we are uncertain whether the HGF, activin-A, and nicotinamide per se are responsible for initiating the expression of pancreatic islet markers because in some of the culture wells, confluence alone appeared to initiate the production of these markers. A notable exception was transthyretin, a liver acute-phase protein (34) that was expressed only in NIP cell cultures treated with dexamethasone. In some cultures, we detected low levels of secreted islet hormones, such as insulin, glucagon, and GLP-1. In this regard, our cells are similar to the rat IPSCs described by Ramiya et al. (34), which also secrete small quantities of insulin in vitro (140 pg/300 islets) but are nevertheless able to reverse hyperglycemia in streptozotocin-induced diabetic mice when transplanted in vivo.

    Although it seems clear that the combination of bFGF and EGF is pro-proliferative for both neural stem cells and our nestin-positive islet progenitor cells (1214), it is less clear which factors are essential for the differentiation of NIP cells to an endocrine phenotype. We used HGF, activin-A, and exendin-4 because they have been shown to differentiate AR42J cells, derived from a rat ductal carcinoma, to glucagon- and insulin-producing endocrine cells. Moreover, exposure of the AR42J pancreatic ductal cell line to high doses of dexamethasone converts them to hepatocytes (49). However, the duct-derived IPSCs (34) and CHIBs (47) differentiated in response to the application of HGF/EGF/nicotinamide and keratinocyte growth factor/Matrigel, respectively. Further studies are warranted to optimize the in vitro conditions required to complete the pancreatic endocrine differentiation process and enhance hormone secretion from our NIP cell cultures.

    It remains unclear whether the NIP cells described in our studies are pluripotential stem cells, akin to bone marrow hematopoietic stem cells and neural ependymal stem cells, or are multipotential cells that can differentiate to more restricted cell lineages. Our speculation is that these intraislet cells are multipotential stem cells with the potential to differentiate into several pancreatic cell lineages (endocrine, exocrine, and ductal) or hepatic cell lineages (3739) given exposure to appropriate environmental growth factor stimuli.

    Our findings of IDX-1 expression in localized regions of human islet-like clusters suggest that these cells may be capable of differentiating into any pancreatic lineage. Similar to the CHIBs of Bonner-Weir et al. (47), we found weak cytoplasmic and nuclear IDX-1 staining in the NIP cells and strong nuclear staining in the SCs of cells. IDX-1 is a homeodomain protein expressed early in pancreas development (E8–9 in the mouse) (50) and the regenerating pancreas after partial pancreatectomy (51), which is required for the development of the pancreas (10,16). IDX-1 is thought of as a “master regulator” of pancreas development and an essential transactivator of islet cell–specific genes such as the insulin gene expressed in β-cells (33). The ectopic expression of IDX-1 in α-cell lines that do not normally express IDX-1 in vitro (52,53) and in hepatic cells in vivo (54) converts them into a β-cell phenotype. Remarkably, the administration to mice of an adenovirus-based vector expressing IDX-1 is sufficient in and by itself to convert a subpopulation of hepatic cells into insulin-producing β-cells that can also restore glucose homeostasis to streptozotocin-induced diabetic mice (54).

    It is also worth noting that pancreatic cells have the capacity to transdifferentiate into hepatic cells. During the regenerative phase of the rat pancreas after metabolic injury (37), portions of the reforming pancreas undergo liver metaplasia. It has also been reported that the pancreas contains hepatic oval stem cells (37). All of these lines of evidence suggest that a close relationship may exist between pancreatic stem cells, NIP cells, and hepatic/oval stem cells. Therefore, we speculate that transplantation of pancreatic stem cells to the liver may provide a favorable environment for successful engraftment, particularly if the cells are programmed to express the transcription factor IDX-1. It is also tempting to speculate that the recent success of islet transplantation in individuals with type 1 diabetes in which a much larger than usual number of islets were transplanted (55) may be attributable to the delivery of a suprathreshold number of stem cells contained within the islets.

    Further investigations of the properties of the pancreatic NIP cells may provide a means for successful islet cell transplantation without immunosuppression in patients with diabetes owing to an absolute or relative loss of β-cell mass. Pancreatic biopsies obtained from the diabetic individual could be used to prepare islets that could be used as a source for culture and expansion of intraislet progenitor cells ex vitro. These cells then could be genetically engineered to preclude autoimmunity reactions and transplanted back into the recipient host donor of the islets, thereby avoiding both immune intolerance (host vs. graft) and autoimmunity reactions. In this regard, successful transplantation of nonallogenic duct-derived pancreatic stem cells has already been accomplished in mice (34). Stem cells isolated from pancreatic ducts, expanded ex vivo, and implanted into nonobese diabetic mice cured diabetes. It is intriguing that in this study (34), the stem cell graft was allogeneic and there was neither immune intolerance nor autoimmunity, suggesting that the stem cells are immunologically blind and may be reprogrammed by the allogeneic host to be recognized as self. If these findings are substantiated by further studies, efforts to achieve successful transplantation of freshly isolated whole islets without immunosuppression in individuals with type 1 diabetes should be reconsidered and refocused on the transplantation of pancreatic stem cells or islets in which the population of stem cells has been expanded by ex vivo culturing.

    FIG. 1.

    Expression of the neural stem cell–specific marker nestin in a distinct cell population within pancreatic islets. Dual fluorescence immunocytochemical staining of (A) an islet cluster in an E16 rat pancreas and (B) a pancreatic islet in the pancreas of an adult 60-day-old (P60) rat. Immunostaining of nestin and insulin were developed with the fluorophores Cy3 (red) and Cy2 (green), respectively. Note that there is no coexpression of nestin and insulin in the cells, which, if present, would be seen as yellow cells. Typical representative islets are shown. C: Nestin (red) and collagen IV (green), a marker of vascular endothelium, are expressed in separate cell populations in a rat pancreatic islet. D: Nestin-positive (yellow in pseudocolor) cells have a distinct nucleus (blue stain with Hoechst). E: RT-PCR of RNA prepared from islets isolated from a P60 adult rat (left) or human islet tissue (right). RT-PCR was performed using oligonucleotide amplimers to rat and human nestin mRNA giving the predicted 326- or 495-bp product of the correct size (upper panels) and confirmed by Southern blot analysis hybridized with a 32P-labeled nestin probe (lower panels) (see research design and methods). Original magnification ×200.

    FIG. 2.

    Isolation and proliferation of NIP cells. A: Isolation of NIP cells. Panel 1: A rat islet cultured for 8 days in modified RPMI 1640 media containing 20 ng/ml bFGF and EGF showing outgrowth of the monolayer of cells. Original magnification ×100. Panel 2: Dual fluorescence immunostaining of the cell monolayer outgrowing from rat islets using antisera to nestin (Cy3) and insulin (Cy2). B: Proliferation and formation of SCs in rat cultures. Five to ten cells were picked from the monolayer shown in A and subcloned by cylindrical cloning. Two-day cultures show that the cells have attached. Then, cells proliferate and reach confluence by day 12. They then migrate (15 days) and form SCs by day 18. The last panel (bottom right) is a magnified image of an SC. C: NIP cells subcloned (passage 1; day 5) from human islets and immunostained with antisera to nestin (Cy3). D: Sequence of events leading to the formation of SCs in human NIP cell cultures. Cells were picked from the outgrowth surrounding a human islet and subcloned. The cells become confluent (panel 1), migrate (panel 2), and form vacuolated structures (panel 3). Next, cells around these spaces become round, forming dense bodies (panel 4), which become larger (panel 5), and eventually form SCs (panel 6).

    FIG. 3.

    Differentiation of rat and human NIP cells toward an endocrine or duct phenotype. A: The DNA products (upper panels) of the correct predicted sizes generated by RT-PCR of NIP RNA using amplimers to rat (690 bp) or human (1 kb) CK19 and rat NCAM (326 bp). The authenticity of the DNA products was verified by Southern blot hybridization using 32P-labeled CK19 and NCAM, respectively (lower panels). B: Expression of homeodomain protein IDX-1 in human SCs derived from NIP cells. Immunocytochemistry (Cy3 immunofluorescence) in SCs (after treatment with bFGF plus EGF for 17 days, then incubation with Activin-A plus HGF for 2 weeks) using an antiserum to IDX-1. Bright punctate structures are immunopositive nuclei throughout the SCs (inset; original magnification ×400). Preimmune serum control staining of human SCs were negative (not shown). C: RT-PCR of mRNA prepared from SCs. DNA product (553 bp) obtained by RT-PCR (top panel) was confirmed by Southern blot hybridization (bottom panel). D: Western immunoblot of protein extract (positive control) prepared from human SCs derived from NIP cells (left) compared with extract prepared from a rat clonal β-cell line (INS1) (right) using an antiserum to IDX-1 (upper panels) or preimmune serum (lower panels). The exposure was longer for the Western blot prepared for human NIP cells.

    FIG. 4.

    Expression of pancreatic endocrine hormones in NIP cultures containing SCs. A: RT-PCR of RNA prepared from human NIP cultures (bFGF plus EGF for 17 days, then incubated with activin-A plus HGF for 14 days) containing islet-like aggregates or freshly isolated islets was performed using oligonucleotide primers for human proglucagon (179 bp, top panels). The identity of bands were confirmed by Southern blotting with a 32P-labeled proglucagon oligonucleotide probe (lower panels; see research design and methods.) B: RT-PCR product of RNA prepared from human NIP cultures (bFGF plus EGF for 9 days, then incubated with activin-A, HGF, nicotinamide, and exendin-4 for 3 days) or islets using oligonucleotide primers for human insulin (∼100 bp, top panels). The authenticity of the bands were confirmed by Southern blotting with a 32P-labeled probe (lower panel). C: Insulin secretion values from NIP cultures that were subjected to the same treatment protocol as in B, or control media that contained no growth factors (n = 1). Note that the data in Table 1 and Fig. 4 are obtained from different clonal lines of human NIP cultures.

    FIG. 5.

    Expression of neuroendocrine, exocrine pancreatic, and hepatic markers in human NIP cultures containing SCs. RT-PCR of RNA prepared from human NIP cultures (cultured in the absence of serum for 9–12 days) was performed using oligonucleotides (see research design and methods) for human synaptophysin (SYN), HGF receptor (HGFR) or c-MET, GLUT-2, pancreatic amylase (AMY), procarboxypeptidase (CARB), transthyretin (TTR) (induced when cultures were treated with 10 μmol/l dexamethasone), HGF, E-cadherin (E-CAD), XBP-1, and α-fetoprotein (AFP). The identities of each band was confirmed by Southern blotting with 32P-labeled oligonucleotides (data not shown).

    FIG. 6.

    Nestin-positive cells are expressed in localized regions of the ducts in the rat pancreas. A: Phase contrast (panel 1) and dual immunofluorescence immunostaining (panel 2) using antisera to CK19 (green) and nestin (red) of a section of a pancreas from a 60-day-old rat (P60). (Original magnification ×100.) The arrows in panel 2 point to localized regions of the ducts that express nestin (red) or coexpress nestin and CK19 (yellow). Note that the islet contains cells are immunostained for cytokeratin (green) or nestin (red) but not both. Also note that there are nestin-positive cells scattered about in the exocrine parenchyma, probably consisting of cells in centroacinar ductules. B: Section of rat pancreas showing islets and ducts stained with antisera to nestin (red) and CK19 (green). Original magnification ×200. C: High magnification of a large duct in a section of a rat pancreas. The top panel shows immunostaining with antisera to nestin (yellow in pseudocolor), and the lower panel shows coimmunostaining for nestin (yellow in pseudocolor) and nuclei stained with Hoechst (blue).

    FIG. 7.

    Alternative models for the origin of pancreatic duct cells that are progenitors of islet endocrine cells. Model 1: Ductal epithelial cells contain specialized regions of prepatterned cells upon exposure to appropriate morphogens in the environment. Model 2: Stem cells distinct from epithelial cells reside in and around the ductal epithelium, are highly motile, and migrate until they find a mesenchymal niche that produces morphogens appropriate for their differentiation to endocrine cells. Our studies favor model 2 (Fig. 4C).

    TABLE 1

    Secretion of pancreatic endocrine hormones from differentiating human NIP cells

    Acknowledgments

    This work was supported in part by U.S. Public Health Service Grants DK 30457, DK 30834, and DK 55365 (to J.F.H.) and DK 02476 (to M.K.T.). H.Z. was supported in part by Deutsche Diabetes Stiftung. J.F.H. is an Investigator with the Howard Hughes Medical Institute.

    We thank H. Hermann, K. McManus, and S. McNamara for expert experimental assistance and T. Budde and R. Larraga for preparation of the manuscript. We thank Dr. C. Messam (NINDS, NIH, Bethesda, MD) for sharing the antibody against human nestin with us.

    Footnotes

    • Address correspondence and reprint requests to Joel F. Habener, Laboratory of Molecular Endocrinology, Massachusetts General Hospital, 55 Fruit St., WEL320, Boston, MA 02114. E-mail: jhabener{at}partners.org.

      Received for publication 25 January 2000 and accepted in revised form 15 November 2000.

      H.Z. and E.J.A. contributed equally to this work.

    REFERENCES

    | Table of Contents