Enriched Human Pancreatic Ductal Cultures Obtained from Selective Death of Acinar Cells Express Pancreatic and Duodenal Homeobox Gene-1 Age-Dependently

Abstract

Adult pancreatic ductal cells are believed to be islet precursors. Our aim was to obtain an enriched human ductal cell population in defined culture conditions, and to characterize these cultures for the presence of pancreatic developmental transcription factors. Non-endocrine adult human pancreatic digest was cultured for 4 days in serum-containing and serum-free media. During this time, analysis was done for phenotypic changes, cell death, and expression of islet and islet precursor markers. Culture in serum-supplemented and serum-free media gave similar recoveries of an enriched ductal population after 4 days. Extensive cell death due to apoptosis and necrosis was also observed over this time period. A donor-age dependent expression of pancreatic and duodenal homeobox gene-1 (PDX-1) in ductal cells was seen at 4 days whereby donors <25 yr expressed significantly more than donors >25 yr. Analysis of gene expression by RT-PCR showed the presence of islet developmental transcription factors neuroD, Nkx6.1, and PDX-1, as well as mature islet hormones. While acinar-ductal transdifferentiation of some cells cannot be ruled out, we provide evidence that the predominant mechanism for the derivation of enriched human ductal cultures in our culture conditions is selective acinar cell death. Furthermore, we have shown that ductal cultures from younger donors exhibit greater plasticity through expression of PDX-1, and may be of greater value in attempts to induce islet neogenesis. The presence, however, of insulin and glucagon mRNA indicates that contaminating endocrine cells remain in these cultures and underscores the need to use caution when assessing differentiation potential.

Introduction

With recent advancements in islet isolation and immunosuppressive therapy [1-3], islet transplantation is now an effective treatment for certain individuals with type 1 diabetes. Limiting this effectiveness, however, is the fact that it presently requires at least 10,000 islet equivalents/kg to consistently achieve insulin independence [1-3] necessitating the use of 2-3 donor organs for each recipient. For this reason, there exists the need for an increased supply of functional insulin-producing tissue in order to make islet transplantation a widespread treatment for patients with type 1 diabetes. Various alternative sources of insulin-producing tissue have been proposed including porcine tissue [4], engineered beta-cell lines [5], embryonic stem cells [6, 7], and pancreatic ductal tissue [8, 9], each with its own limitations. It is believed that, in the adult pancreas, cells of the ductal epithelium have the potential for differentiation to endocrine cells and may be one source of islet neogenesis throughout life [8, 9]. Bonner-Weir et al. reported the development of human islets from a ductal-enriched population in vitro [10], showing this to be an effective source; however to date a clinically significant number of islets has not been produced by this or similar methods.

The derivation of a ductal cell population through tissue culture of digested non-endocrine pancreatic tissue has been accomplished in several in vitro models [11-14]. Culture of human and rat exocrine-enriched cell preparations has been proposed to result in a conversion from a primarily amylase-expressing cell population into a cell population that no longer expresses amylase but rather the ductal markers cytokeratin 7 and 19 [11-15]. In other experiments, these ductal-like cells have been shown to be capable of expressing early endocrine markers [14, 16] or to have the capability to differentiate into rat β-cells in vivo [17]. These results suggest that ductal cell populations could potentially provide an abundant source of islets for transplant to type 1 diabetic patients.

Although transdifferentiation between phenotypes has been suggested as the mechanism for the derivation of these ductal cell populations, it has not been proven whether this actually occurs. We hypothesized that these cultures of predominately ductal cells arise from selective cell death of the exocrine component and preferential survival of the ductal population during tissue culture. To test this hypothesis, human pancreatic cultures were analyzed for overall cell survival, levels of apoptosis, and the presence of transitional cells (i.e. expressing both acinar and ductal markers) indicating a phenotypic intermediate between ductal and acinar. In addition, previous studies examining the preparation of enriched ductal populations have utilized serum-supplemented media [11-16] and it has been reported that for rat exocrine/ductal cultures to survive, serum must be added to the culture medium [14]. However, in order for islets created in the future from these cultures to be used clinically to treat diabetes, a culture environment free of xenoproteins will be desirable. Thus, in the present study, both serum-supplemented and novel, serum-free formulations were tested for efficiency in deriving an enriched population of ductal cells. As it has been proposed that human pancreatic ductal cultures obtained in this fashion may be used to create an abundant source of islets for transplantation via differentiation of endocrine progenitors, preparations were also analyzed for expression of genes involved in islet development and mature islet function.

The homeodomain transcription factor pancreatic and duodenal homeobox gene-1 (PDX-1) is expressed in mature β-cells ubiquitously [18] and has been proposed to play a role in islet development both during embryonic organogenesis [19, 20] and to affect islet turnover in the postnatal pancreas [21, 22]. Furthermore, other studies have shown that ectopic expression of PDX-1 in non-pancreatic cells is sufficient to induce differentiation to an insulin-producing phenotype [23, 24]. Since PDX-1 expression has been previously reported in human ductal cell cultures [14, 16] we assessed the levels of PDX-1 expression quantitatively with the hypothesis that cultures containing higher numbers of PDX-1-positive ductal cells may exhibit greater plasticity and have a higher potential for islet neogenesis.

Methods

Tissue preparation and culture

Human donor pancreases were removed from cadaveric donors who had previously given informed research consent, and processed according to the protocols described by this laboratory [1, 25]. Briefly, organs were cold-stored in University of Wisconsin solution and perfused via the duct with the enzyme solution Liberase (Roche, Indianapolis, USA). Once digested, islets were purified on continuous Ficoll gradients using the refrigerated Cobe 2991 (COBE BCT inc., Lakewood, USA). Immediately following islet purification, the predominately acinar cell fraction was collected from the Cobe bag and washed three times with Hanks balanced salt solution (HBSS; Sigma-Aldrich, Oakville, Canada) supplemented with 0.5% bovine serum albumin (BSA; fraction V, Sigma). After dithizone staining of representative samples to assess islet content, tissue was cultured in suspension with approximately 500 μl "pack tissue volume" per non-treated 15 cm plate (Fisher Scientific, Edmonton, Canada) in RPMI 1640 medium (Gibco/Invitrogen, Burlington, Canada) supplemented with either: i) 10% fetal calf serum (FCS), ii) 1% insulin-transferrin-selenium (ITS)/0.5% bovine serum albumin (BSA), or iii) 1% ITS/0.5% human serum albumin (HSA) (all supplements/sera from Sigma). Media changes were performed at day 1 and day 3 post-culture. Protocols used in this study were approved by the Research Ethics Board of the University of Alberta.

Characterization of cell preparations

After isolation, as well as 4 day tissue culture, cell preparations were assessed for cellular insulin (radioimmunoassay) and DNA content (picogreen dsDNA assay; Molecular Probes, Burlington, ON, Canada) as well as cell composition according to methods previously described [26]. To determine cellular composition, aggregates were dissociated into single cell suspensions to facilitate quantification of immunostaining. Single cell suspensions were obtained by mechanical disruption by pipetting at 37°C in Ca²⁺-free media supplemented with 1 mM EGTA and 0.5% BSA for 7 minutes before addition of trypsin (25 μg/ml) and DNAse (4 μg/ml) and further pipetting for 4 minutes. Cells were then allowed to adhere to histobond slides (Marienfeld, Germany) and fixed in Bouin's fixative for 12 minutes before storage at 4°C in 70% ethanol.

For morphological assessment, intact cellular aggregates were washed with phosphate-buffered saline (PBS), fixed immediately in 4% paraformaldehyde for 30 minutes, and stored in PBS. Samples were subsequently embedded in a 2% low melting point agarose solution (30-100 μl depending on sample size) and allowed to harden at 4°C before processing, paraffin embedding, and sectioning (3 μm) on to histobond slides.

Immunostaining of single cell and paraffin-embedded sections was performed using the ABC-DAB method. Sections or cells were quenched with a 20% H₂O₂/methanol solution. Microwave antigen retrieval for cytokeratins 7 and 19, PDX-1, and Ki 67 staining involved: 15 minutes on high power (Sanyo household model, 1260W) in 800 ml Na+ citrate for tissue sections and 5 sec. on high 6 times in 40 ml Na+ citrate for single cells. Blocking was performed with 20% normal goat serum (Fisher) for 15 minutes. Primary antibody concentrations were as follows: 1/100 rabbit anti-human amylase (Sigma), 1/1000 guinea pig anti-porcine insulin (Dako, Denmark), 1/50 mouse anti-human CK19 (Dako), 1/200 mouse anti-human CK7 (Dako), 1/1000 rabbit anti-human PDX-1 (gift from Dr. Tim Kieffer), 1/400 rabbit anti-human Mist-1 (gift from Dr. Christopher Pin, University of Western Ontario, London, Canada), and 1/50 rabbit anti-human Ki 67 (Santa Cruz Biotechnology inc., Santa Cruz, USA). Antibodies to cytokeratins 7 and 19 were used as markers for pancreatic ductal-epithelial-type cells as previously described [10, 11, 16]. Primary antibody incubations were 30 minutes at room temperature followed by a wash (3X) with PBS before addition of secondary antibody. All biotinylated secondary antibodies were obtained from Vector Laboratories (Burlingame, USA) and used at a concentration of 1/200 for 20 minutes. ABC complex (Vector) incubation time was 40 minutes and visualization was with the chromagen diaminobenzadine (Biogenex, San Ramon USA) for 5 minutes. For indirect immunofluorescence of cytokeratins, primary antibodies were used at the same concentration and donkey anti-mouse Cy3 conjugated secondary antibody (Jackson ImmunoResearch Laboratories Inc., Pennsylvania, USA) was used at 1/300. For detection of apoptotic cells, TUNEL staining (kit from Promega Corp., Madison, USA) was performed on dissociated cell samples from post-isolation, day 4 culture, and media change supernatant. Briefly, cells were fixed in 4% paraformaldehyde, permeabilized with 20 mg/ml proteinase K, and incubated with TdT enzyme and fluorescein-labeled dUTP for 1 h before visualization for fluorescence. All images were captured on a Zeiss Axioskop II fluorescent microscope with a Coolsnap camera and IP lab software (Scanalytics Inc., Fairfax, USA). Single cell counts were performed on a minimum of 500 cells per sample and percentages calculated as number of positive per 500. Negative controls were performed for all immunostaining procedures and consisted of the above protocol with elimination of the primary antibody (not included in figures).

For electron microscopy, aggregates and single cell suspensions were fixed in 2.5% (vol/vol) glutaraldehyde (Millonig's buffer, pH 7.2), post fixed in 1.5% (wt/vol) OsO₄, washed in distilled water, and dehydrated successively in 50, 70, 80, 90, and 100% ethanol, before embedding in aldarite. Sections were then stained with lead citrate and uranyl actetate and examined in a Hitachi H7000 (Hitachi Ltd., Tokyo, Japan) transmission electron microscope.

RT-PCR analysis

Samples from different donor cultures were snap-frozen dry in LN2 and stored at -80°C for molecular analysis. Samples were re-suspended in 1 ml trizol reagent and RNA extracted according to manufacturer’s protocol (Gibco). cDNA was constructed from 1 μg mRNA with 10 units (200 U/μl) Superscript reverse transcriptase in 1X buffer containing 0.01M DTT, 0.5 mM dNTPs, and 0.02 μg/μl oligo dT15. For each sample, 2 μl of cDNA was used per 25 μl reaction along with 1X PCR buffer, 2 mM MgCl, 0.2 mM dNTPs, 1 unit taq polymerase (5 U/μl), and 0.5 μM of each primer pair. All chemicals/reagents were from Gibco/Invitrogen and previously used primer sequences were obtained from Assady et al. [7] (PDX-1, insulin), and Heremans et al. [27] (neuroD, Pax4, glucagon). For all PCR reactions, 35 cycles were performed, with 30 seconds denaturation at 94°C, 30 seconds annealing at 58°C, and 30 seconds extension at 72°C. Final polymerization was at 72°C for 10 minutes followed by a 4°C hold. Products were separated on a 2% Ethidium bromide stained agarose gel and images captured on Alpha Digidoc software (Perkin-Elmer, Boston, USA). Primer sequences were as follows: 5'-CCC ATG GAT GAA GTC TAC C-3' (forward) and 5'-GTC CTC CTC CTT TTT CCA C-3' (reverse) (PDX-1-262 b.p. fragment), 5'-CCT GTA CCC CTC ATC AAG GA-3' (forward) and 5'-CTC TGT CAT CCC CAA CCA AT-3' (reverse) (Nkx6.1-182 b.p. fragment), 5'-ATC CCA ACC CAC CAC CAA CC-3' (forward) and 5'-CAG CGG TGC CTG AGA AGA TT-3' (reverse) (neuroD-439 b.p. fragment), 5'-AGG AGG ACC AGG GAC TAC CGT-3' (forward) and 5'-TTT AGG TGG GGT GTC ACT CAG-3' (reverse) (Pax4-496 b.p. fragment), 5'-CCT GCA GCC CTT GGC C-3' (forward) and 5'-GTT GCA GTA GTT CTC CAG GTG-3' (reverse) (insulin-102 b.p. fragment), 5'-CCC AAG ATT TTG TGC AGT GGT T-3' (forward) and 5'-GCG GCC AAG TTC TTC AAC AAT-3' (reverse) (glucagon-221 b.p. fragment), 5'-CTC GAG GGT AGA AAG GAT GAC GCC TC-3' (forward) and 5'-CCG AGT TGA GGT CGT GCA T-3' (reverse) (ngn3-313 b.p. fragment), 5'-TTT GTC ACC GTG GCC GTG TTT-3' (forward) and 5'-TTG CAT GTG TTC CCT GTC TGG-3' (reverse) (synaptophysin-253 b.p. fragment), 5'-CCA GCA GAG AAT GGA AAG TC-3' (forward) and 5'-GAT GCT GCT TAC ATG TCT CG-3' (reverse) (β-2 microglobulin-268 b.p. fragment). β-2 microglobulin primers were also used to verify the absence of genomic DNA contamination in samples as only the 268 b.p. fragment was seen and not the 900 b.p. genomic version of the gene. A previously characterized enriched human islet preparation was used as a positive control for several genes examined as well as for comparison with human ductal cultures. Positive controls for transiently expressed developmental factors (i.e. ngn3) were samples from previous experiments found to contain the mRNA for these genes. All products were verified through sequencing using a TOPO TA cloning kit (Invitrogen) and BLAST search against known GenBank sequences (Accession numbers: NM-020999 for ngn3, NM-004048 for β-2 microglobulin, NM-000207 for insulin, NM-002054 for glucagon, NM-002500 for neuroD, NM-006168 for Nkx6.1, NM-000209 for PDX-1, NM-003179 for synaptophysin, and NM-006193 for Pax4). Negative controls consisted of cDNA from another species and Sigma water (Sigma) in place of experimental cDNA.

Statistical analysis

Values are expressed as mean ± standard deviation. Statistical significance of differences was calculated by a one-way analysis of variance (ANOVA) and Scheffes test. Linear analysis of correlations was by Pearson’s correlation coefficient and all tests were performed on Statview (SAS Institute Inc, Cary, NC). All significance levels were set at p < 0.05.

Results

Recovery of human pancreatic tissue following 4 day culture

The digested acinar fraction of the pancreas from individual donors was obtained post-islet purification and divided into different culture conditions. Dithizone staining pre-culture showed the relative absence of islets in these preparations, although no culture was completely free from islet tissue. During suspension culture, extensive clumping and aggregation of tissue was observed until the second media change (day 3 culture), at which time the preparation became more dispersed. Cell survival, as measured by the change in total DNA content from immediately post-isolation to day 4 culture, was calculated from each individual donor and expressed as a mean for each culture condition (Table 1). When tissue was cultured in RPMI/FCS (n = 9), a mean survival rate of 33.1 ± 14.2% was seen over 4 days which was higher than, but not significantly different from, that observed in the serum-free conditions of RPMI/ITS/BSA (n = 12) (30.4 ± 12.5%) and RPMI/ITS/HSA (n = 8) (26.9 ± 18.2%).

Table 1. Recovery of human pancreatic tissue following 4 days culture vs. total cell number and insulin content obtained immediately post-isolation.

^a p < 0.0001 vs. post-isolation DNA recovery. ^b p < 0.0001 vs. post-isolation insulin content.

Changes in cellular insulin content as an estimation of β-cell content were also assessed immediately post-isolation to day 4 culture in each condition. The absolute amount of cellular insulin content in the acinar fraction immediately post-isolation ranged from 23 to 557 μg, which is significantly lower than the 3000 to 8000 μg obtained in purified islet preparations (unpublished observations). Following 4 day culture, the cellular insulin content was significantly lower in all culture conditions (<15%; p < 0.0001) than that prior to culture (Table 1).

Cellular composition of human pancreatic tissue preparations

Cellular phenotypic changes were assessed by immunostaining for the presence of β-cells (insulin), ductal-epithelial cells (CK7/19), and acinar cells (amylase), in dissociated tissue samples on the day of isolation and after day 4 culture (Table 2). Immediately post-isolation, the non-purified fraction was predominately amylase-positive exocrine cells (88.6 ± 3.1%), whereas the proportion of β-cells was only 1.8 ± 1.4% and the percentage of CK7 and CK19 expressing cells were 20.4 ± 21.0% and 9.9 ± 7.7%, respectively. Culture of this tissue markedly reduced the percentage of cells staining positive for amylase in all conditions (21.3 ± 24.0% RPMI/FCS, 16.1 ± 18.2% RPMI/ITS/BSA, and 9.5 ± 10.5% RPMI/ITS/HSA). In contrast, the proportion of cells expressing CK7 increased to 75.5 ± 10.0% (RPMI/FCS), 79.5 ± 12.6% (RPMI/ITS/BSA), and 86.7 ± 8.5% (RPMI/ITS/HSA) while expression of CK19 increased to 59.1 ± 15.2% (RPMI/FCS), 67.8 ± 19.2% (RPMI/ITS/BSA), and 77.7 ± 10.2% (RPMI/ITS/HSA). Consistent with the reduction in cellular insulin content (Table 1), immunostaining also revealed a decrease in the proportion of β-cells from 1.8 ± 1.4% at day 0 to 0.5 ± 0.6% (RPMI/FCS), 1.0 ± 0.9% (RPMI/ITS/BSA), and 1.1 ± 1.3% (RPMI/ITS/HSA). Moreover, the small fraction of endocrine cells in these preparations was also evidenced by the fact that in 12 different donor cultures at day 4 post-isolation, expression of the neuroendocrine marker synaptophysin was only 2.0 ± 1.4% (not shown in table). None of the changes seen in cellular phenotype between the three culture conditions at day 4 were statistically significant.

Table 2. Cellular composition of human pancreatic tissue preparations immediately following isolation and after 4 days culture.

^a p < 0.01 vs. post-isolation. ^b p < 0.0001 vs. post-isolation.

In addition to characterizing dissociated preparations, the cellular morphology of intact aggregates post-culture was assessed by immunostaining (Figure 1). It was shown that after 4 days culture, the majority of cells within the aggregates were positive for either CK7 or CK19. Amylase expression was present in some of the aggregates and very few insulin positive cells were observed.

Figure 1.

Representative immunohistochemistry for insulin (A, B), amylase (C, D), CK7 (E, F), and CK19 (G, H) in human pancreatic tissue after 4 days culture in RPMI/ FCS. Agarose-embedded tissue aggregates were immunostained using the immunoperoxidase method (brown) and counterstained with hematoxylin.

Cell death in human pancreatic cultures

In addition to assessment for cell recovery, samples from cultures were also analyzed for evidence of cell death/apoptosis using TUNEL staining and electron microscopy. Quantification of TUNEL-positive cells in dissociated samples showed that 27.5 ± 24.6% of the cells in culture were undergoing apoptosis at day 3-4 culture. This was significantly higher than the 4.7 ± 3.5% (p < 0.005) of apoptotic cells seen immediately post-isolation and represented a mean change of 30.7 ± 26.9% over the culture period (data not shown). During media change and wash at days 1 and 3 post-isolation, tissue was centrifuged at 1200 rpm for 2 minutes. At these times, samples of wash supernatant were taken from the media supernatant after centrifugation and were found to contain a high level of apoptotic cells (∼80%) indicating extensive cell death in culture. Furthermore, electron microscopy of these supernatant samples showed that a high proportion of cells were severely damaged, exhibiting condensed nuclei and membrane degradation (Figure 2).

Figure 2.

Electron microscopy of cell samples removed from the supernatant of human pancreatic cultures during media change at 2 days post-islet isolation. Magnification = 1200X.

Immunostaining to detect transdifferentiating acinar cells in culture

Double immunofluorescent staining was performed for CK7 and amylase at several time-points in order examine for possible transitional cells expressing both markers as suggestive evidence for transdifferentiation. No cells in any of the preparations (n = 9) tested were found to co-express ductal and exocrine markers. We also assessed the expression of the pancreatic acinar cell marker Mist-1 [28] in human ductal cultures in an attempt to identify possible transdifferentiating acinar cells that had lost expression of amylase and begun to express cytokeratin. Again it was found that no cells at any time-point showed co-expression of Mist-1 and either CK7 or CK19.

Expression of genes associated with islet neogenesis in human pancreatic ductal cultures

To assess for the possible presence of islet precursors, expression of several islet developmental-associated genes was examined by RT-PCR after 4 days culture (Figure 3). Cultures from 7 different donors of varying ages (lanes 1-7), as well as one previously characterized enriched human islet preparation (lane 8), were analyzed for the presence of mRNA for PDX-1, Nkx6.1, neuroD, Pax4, insulin, glucagon, neurogenin 3 (ngn3), and synaptophysin. Although little to no expression was observed of the islet developmental transcription factors ngn3 and Pax4 in the samples tested, expression of other downstream factors such as Nkx6.1 and neuroD was seen in almost all of the ductal cultures as well as in the islet preparation. Ductal samples were also found to express the genes for the mature islet proteins insulin, glucagon, and synaptophysin, probably confirming the presence of a small amount of contaminating islet-endocrine tissue remaining after purification.

Figure 3.

Semi-quantitative RT-PCR analysis comparing samples taken from human pancreatic ductal cultures at day 4 post-isolation (lanes 1-7) and an enriched human islet preparation (lane 8). Transcription of genes such as insulin, Nkx6.1, glucagon, and synaptophysin in the majority of ductal cultures indicates the presence of contaminating islets. PDX-1 (7/7 preparations), neuroD (6/7 preparations), and Pax4 (2/7 preparations) expression indicates the possible presence of precursor cells undergoing islet differentiation, although the important developmental transcription factor ngn3 was not detected in any of the samples. β-2 microglobulin was used as a control for RNA/cDNA quality.

Assessment of PDX-1 expression in ductal cultures

The proportion of cells expressing the PDX-1 protein was also assessed immediately post-isolation and after 4 days culture in all culture conditions. Immunostaining and quantification in dissociated samples showed expression of PDX-1 in post-isolation fractions very similar to the proportion of β-cells (1.7 ± 1.3% PDX-1 vs. 1.8 ± 1.4% β-cell). Following 4 days culture, the proportion of PDX-1 positive cells varied considerably from 0.2% to 22.6%. Due to this variation, we attempted to identify a correlation with donor or isolation characteristics including age, body mass index, and organ cold ischemia time that could explain these differences amongst donors. An inverse correlation was observed between expression levels post-culture and donor age (Figure 4). Changes in PDX-1 expression were equivalent regardless of culture condition, and the data presented is for the serum-supplemented condition. In young donors (< 25 yr; n = 7) PDX-1 expression after 4 day culture was significantly higher (p < 0.05) than that immediately post-isolation (1.7 ± 1.3% vs. 10.4 ± 6.3%; range post-culture 4.6% - 22.6%). Expression levels in older organs (>37 yr; n = 6) demonstrated a lower proportion of PDX-1 (1.7% ± 1.3 vs. 0.9 ± 0.5%; range post-culture 0.2% - 1.4%). When data for all donor ages was combined (age range of 11 - 65), a direct inverse correlation was observed (r² = -0.790, p < 0.05) with the level of PDX-1 expression in cultures at day 4. To further characterize the PDX-1 positive cells following 4 day culture, preparations were double-stained for co-localization with the ductal marker CK7. In donors younger than 25 yr, the majority of the PDX-1 positive cells were shown to also co-express CK7 (Figure 5 A-D). In contrast, the low numbers of PDX-1 positive cells present in the older donor cultures were shown not to co-express CK7 but to co-express insulin and were thus surviving 46;-cells (Figure 5 E-H). Sections from donor pancreatic biopsies corresponding to these preparations were also assessed for PDX-1 expression and localization to assess whether the protein is normally expressed in human pancreatic ducts. As expected, PDX-1 expression in these sections was almost exclusively restricted to islets and was rarely seen in cells of ductal phenotype, regardless of donor age.

Figure 4.

PDX-1 expression in cultured human pancreatic tissue after day 4 culture in RPMI/FCS correlates inversely with age of donor (r² = –0.790, p < 0.05) (n = 13).

Figure 5.

Characterization of PDX-1 expressing cells after 4 day culture of human pancreas through co-localization with insulin and the ductal marker CK7. Cytokeratin expression (bright red) and insulin expression (green) visualized with indirect immunofluorescence (A, C, E, G) and PDX-1 expression (brown) indicated by immunoperoxidase staining (B, D, F, H). A-D: Co-expression of PDX-1 and CK7 in a 17 year old donor preparation post-culture indicating many double positive cells (arrows) and only a single insulin positive/PDX-1 posi-tive cell. E-H: In contrast, no co-expression of PDX-1 and CK7 in older donor preparations could be detected (arrow indicates a PDX-1 positive but CK7 negative cell) while the few PDX-1 positive cells found were also insulin positive.

Cell proliferation levels are low in human pancreatic ductal cultures

We also assessed the proliferative capacity of these cells at day 4 culture using immunostaining for the cell cycle marker Ki 67. Quantification revealed low numbers of replicating cells (0.6 ± 0.7%), indicating low turnover rates for existing cells post-islet isolation.

Discussion

In order for islet transplantation to become a widespread therapeutic treatment for patients with diabetes, an abundant source of islet tissue must be identified. This may be facilitated through the use of methods to produce enriched cell populations that show characteristics of ductal cells, followed by differentiation into functional endocrine cells. Transdifferentiation from an acinar to ductal phenotype has been previously proposed in both rodent and human cell culture models [11-16] although conclusive evidence to support this phenomenon has remained elusive. In the present study we describe culture conditions (i.e. high density/serum free media) for the derivation of enriched ductal cell populations following culture of human pancreatic digest. We also show evidence in this model to favor the explanation of selective acinar cell death over transdifferentiation in these cultures and show that serum-free culture can be effective in deriving this primarily ductal population. Finally, we have shown that these cultures contain cells expressing certain markers of islet development, including PDX-1 in age-dependent manner, although the observed expression of mature-islet specific genes indicates that endocrine cells are still present and caution should be used when assessing the potential for islet neogenesis.

Due to cold ischemia of the donor pancreas followed by the stresses of enzymatic digestion, it is potentially difficult to reproducibly obtain highly viable pancreatic digest for subsequent tissue culture. This was reflected by the significant reduction in cellular DNA levels seen over 4 days culture whereby, even in the most optimal condition tested, a mean survival rate of 33.1 ± 14.2% could be achieved. One explanation for this cell reduction is the selective loss of exocrine cells, which we have previously described in cultures of porcine pancreatic tissue [26]. During the initial 24 hour culture period, many dead cells were observed, which is probably the result of deleterious effects of cold storage, collagenase digestion, and Ficoll exposure. Several studies have proposed that one reason for early islet graft failure post-transplant is due to apoptosis induced during isolation [29, 30], and methods have been suggested to decrease this cell death with varying results [30, 31]. Our results obtained using TUNEL staining and electron microscopy indicate that these pancreatic cells do not survive well in culture, and an extensive amount of apoptosis occurs in the days following isolation. These findings are in agreement with another study using human islet cultures that showed ∼32% of islet cells to be apoptotic by 5 days post-isolation [29]. The large variability in these experiments indicates that some donor preparations survived in culture much better than others, although a strong correlation to explain this phenomenon could not be found.

Previous reports have attributed the derivation of a ductal cell preparation from a primarily exocrine population to the phenomenon of transdifferentiation, whereby amylase- positive cells de-differentiate into a phenotype resembling ductal-epithelial cells [11-16]. Although this explanation seems attractive, definitive evidence for a transition of this type has remained elusive due to the difficulty in tracking cells in a heterogenous population through multiple days in culture. Another explanation for the change observed is a selective death of the exocrine cells and survival of pre-existing ductal cells resulting in a primarily ductal population after 4 days in culture. When using DNA recovery to calculate the total number of cells recovered (6.6 pg DNA/cell) and the phenotypic composition, in our cultures an overall increase in the number of ductal cells of 170 ± 112% (CK7) and 185 ± 127% (CK19) could be observed. Concomitantly, there was a decrease in the number of amylase-positive exocrine cells to only 6.8 ± 6.7% of the original number after 4 day culture. Considering that the overall cell population (which is predominately exocrine) is reduced by ∼60% over 4 days, clearly the majority of the exocrine cell component is undergoing cell death rather than transdifferentiation. Also, because about 1% of the cells present at day 4 expressed the cell cycle marker Ki67, it is possible that the increase in number of ductal cells is at least in part due to proliferation of existing ductal cells over the total span of 4 days in culture. Thus, the final ductal cell preparation may arise primarily from survival and some proliferation of pre-existing ductal cells without ruling out the possibility of a small contribution from the transdifferentiation of a selective population of exocrine cells. However, the fact that we observed no cells at any time-point in culture co-expressing the acinar markers amylase or Mist-1 with the ductal cytokeratin markers suggests against a direct transdifferentiation event occurring in these preparations.

The changes in phenotypic composition associated with pancreatic exocrine/ductal cell cultures have been characterized for some time [11-13]. Our results agree with these previous reports in that we observe a decrease in amylase-positive cells with concomitant increase in the expression of cytokeratin-positive cells when culture is in serum-supplemented media. Gmyr et al. obtained a population containing approximately 46% CK19 and 63% CK7 expressing cells after 7 days culture and reported a yield of approximately 4.9x10⁷ CK7/19 positive cells per gram of tissue [12]. When considering the total number of cells recovered per initial gram of digested tissue (1.81x10⁸), and the percentage of CK7-positive cells, our total number of CK7-positive ductal cells recovered after 4 days culture is approximately 1.45x10⁸ per gram. The difference in our higher recoveries with that of Gmyr et al. [12] may be because our preparations are more enriched in CK19-(70%) and CK7 (80%)-positive cells. Furthermore, in the present study the culture time was 4 days whereas Gmyr et al. cultured for 7 days, and this longer period is most likely associated with greater cell loss.

It has been reported that the derivation of a ductal cell population from primarily exocrine cultures does not occur when serum is omitted from the culture media [14]. Furthermore, all published reports utilizing this model have also utilized FCS in the media [12, 13, 16]. Although, in the present study, the standard RPMI/FCS condition yielded the highest mean survival rate, this was not significantly different from the serum-free RPMI/ITS/BSA and RPMI/ITS/HSA conditions, thereby indicating that this tissue can be maintained in a serum-free environment. Moreover, our results demonstrate the successful recovery of a ductal cell population in a serum-free media. The availability of serum-free conditions will therefore allow the selection of more defined media to initiate differentiation of ductal cells into functional islets. It has also been proposed that coating of human islet grafts with xenoproteins from FCS in pre-transplant media may accelerate alloimmune destruction [1], and for this reason serum-free culture is desirable for cells which may eventually be used in a clinical application.

It has been proposed that pancreatic ductal cells are islet progenitors and that these cultures can be used to create an abundant supply of new islets for transplantation [9, 10, 27]. Recently, Heremans et al. showed the differentiation of a proportion (∼12%) of these cells to insulin-producing cells using viral transduction of the transcription factor ngn3 [27]. These types of studies, however, are based on the assumption that pre-existing islets have been removed during isolation and observed neogenesis is not due either to β-cell replication or de-differentiation and subsequent re-differentiation of mature islet cells. We have shown, however, using immunostaining and RT-PCR, that islet-endocrine cells are still present in these cultures. Most notably, the presence of the hormones insulin and glucagon indicates that differentiated islet cells exist and must be considered when assessing subsequent neogenesis. For this reason, care should be taken to use only the donor preparations most free of contaminating islets as starting material to induce ductal to islet differentiation. The presence, however, of transcription factors known to be expressed transiently in islet development and not in mature islet cells (e.g. neuroD), combined with the observed re-expression of PDX-1 in a select population of ductal cells suggests that the potential for islet neogenesis from true precursor cells does exist in these cultures.

The donor age-dependent re-expression of PDX-1 in culture by ductal cells is an important finding when considering plasticity in the adult pancreas. PDX-1 is expressed in all pancreas-dedicated cells during embryonic development regardless of future phenotype [32] and is maintained primarily in β-cells of the mature organ [33]. PDX-1 is necessary for proper pancreatic formation as mice that are homozygous for the null mutation are born without a pancreas and die within a few days [19]. Using semi-quantitative PCR techniques, Gmyr et al. report a 10.5-fold increase in PDX-1 mRNA levels in human pancreatic cultures after 2 days with a 3.2-fold increase in protein expression [16] while Bouwens et al. describe a similar up-regulation in rat tissue [14]. Our results using quantitative immunohistochemistry to detect PDX-1 positive cells show an increase following 4 day culture in both conditions but only in certain donor preparations. We therefore examined various donor characteristics and found that the level of increase of PDX-1 in these cultures is dependent on donor age. This is an important finding to take into account when assessing the potential of future donor organs for islet neogenesis. In particular, all organs from donors under 25 yr showed an increased proportion of PDX-1 positive cells post-culture while those from older donors maintained or demonstrated decreased levels of expression. We also confirmed, as reported previously [16], that the cells expressing PDX-1 are ductal and co-express CK7/19. We also, however, demonstrate that PDX-1/CK7 co-expression is present only in tissue prepared from donors <25 yr, since in older donors this co-localization was not observed. The increase in PDX-1 expression seen in this study may indicate a higher degree of plasticity in younger organs. These findings are of particular interest in light of a recent study by Pipeleers and co-workers [34], which suggests that human ductal cultures derived from young donors (<10 yr) exhibit a greater potential for islet neogenesis when transplanted in a nude mouse model. While the authors of this study do not provide a clear mechanism by which younger cells undergo increased levels of differentiation, it can be proposed from our results that a greater proportion of the ductal population from young pancreases are capable of expressing PDX-1 under the appropriate conditions, and thus initiating the pathway to islet cell phenotype.

Due to conflicting reports regarding the levels and ductal localization of PDX-1 in the adult pancreas [35, 36], we assessed pancreatic biopsies from donors of all cultures represented in this study for PDX-1 expression. Immunostaining of corresponding biopsies collected pre-islet isolation showed that PDX-1 is rarely expressed in ductal cells in vivo regardless of donor age, reinforcing the proposal that young donor tissue may have greater plasticity under abnormal conditions (i.e. culture). These findings are significant because a recent report by Gu et al. [37] suggests that ductal progenitor cells, which transiently express PDX-1 during embryogenesis then turn off PDX-1 and become ductal-epithelial cells, may not necessarily be endocrine/exocrine precursors in the developing pancreas. We propose that mature ductal cells which are able to re-express PDX-1 may be capable of reverting back to stages of embryonic development and may in fact be able to change lineage and act as islet precursors. Further studies are being undertaken to assess the fate and plasticity of these PDX-1 expressing ductal cells under different conditions.

This conclusion that ductal cells act as islet precursors is not in agreement with a recent paper by Melton and co-workers [38], in which they proposed that β-cell neogenesis in the adult pancreas is due to the replication of pre-existing β-cells and not from the differentiation of a precursor. Our data is not in disagreement with the Melton results for two reasons. Firstly, we clearly demonstrate that PDX-1 re-expression in the ductal cells is age-dependent. Therefore it is possible that, in the young pancreas, ductal cells have a greater potential for differentiation into β-cells that is reduced or absent in adult stages. Secondly, we show the presence of islet β-cells even in cultures derived from the most dense gradient layer (exocrine) after islet purification. This suggests that in all studies to date reporting ductal cell differentiation, the possibility of only β-cells giving rise to new β-cells cannot be ruled out. However, although the results from Melton and colleagues are compelling, convincing evidence from other experiments, including our's, has emerged to suggest that ductal cells are capable of differentiating into β-cells [9, 10, 27, 34]. At this time, it is reasonable to assume from all available evidence that β-cell turnover in the adult pancreas may occur from β-cell replication and from unipotent or multipotent ductal cell precursors.

In conclusion, we have characterized the transition from a primarily exocrine cell pancreatic culture to a highly enriched ductal-epithelial-like cell preparation. The novel ability to culture this tissue in a serum-free environment allows the consideration of possible future clinical applications involving this tissue. In addition, we have shown that ductal-like cell populations prepared from younger donors contain a higher proportion of PDX-1 positive cells and propose that this tissue may have a greater potential to differentiate into functional islet endocrine cells. However, problems such as extensive cell death may hamper these efforts and it remains to be demonstrated whether this tissue can be converted to an abundant supply of functional islets for transplantation.