Culture of Adult Human Islet Preparations WithHepatocyte Growth Factor and 804G Matrix IsMitogenic for Duct Cells but Not for (S-CellsVeronique H. Lefebvre, Timo Otonkoski, Jarkko Ustinov, Mari-Anne Huotari, Daniel G. Pipeleers,and Luc Bouwens

It has recently been reported that human adult (J-cellsproliferate during culture on an extracellular matrixprepared from rat 804G cells and in the presence ofhepatocyte growth factor (HGF) (6). The present studycompares the mitogenic effect of this condition onhuman p-cells and on neighboring non-endocrine ductcells. Islet cell-enriched fractions were prepared fromadult human organ donors and cultured in suspension oron 804G matrix, with or without HGF. The combinationof 804G matrix and HGF increased the number of 5-bromo-2'-deoxyuridine-positive (BrdU+) cells within 48h reaching a maximum after 4 days. In sections, virtuallyall BrdU+ cells were negative for insulin or glucagon andfor preproinsulin mRNA but expressed the ductal cellmarkers cytokeratin 19 and 7, carbonic anhydrase-II, andcarbohydrate antigen 19-9. After 4 days of culture, thecytokeratin 19+ ductal cells exhibited a BrdU-labelingindex of 30% (P < 0.01 vs. 2% without HGF and matrix),whereas <0.1% of insulin-positive and <1% of glucagon-positive cells were labeled. Formation of bilayers withductal cells covering the endocrine cells may causeerroneous interpretation on double positivity in unsec-tioned tissue. It is concluded that culture of human isletcell preparations with HGF and 804G matrix stimulatesthe proliferation of the duct cells but not of the under-lying p-cells. Diabetes 47:134-137, 1998

Islet transplantation represents a potential treatment forinsulin-dependent diabetes. Development and appli-cation of this technique is hampered by the limitedavailability of human donor tissue. In vitro expansion

of human p-cells may provide an adequate source for p-cellgrafts. So far, attempts to expand adult rodent p-cells havebeen unsuccessful. In different experimental conditions, thiscell type has a low rate of replication both in vivo and in vitro(1,2). Nutrients and growth factors can stimulate fetal orneonatal P-cell replication, but most adult p-cells appearunresponsive to growth factor stimulation in vitro (1). In

From the Diabetes Research Center (V.H.L., D.G.P., L.B.), Vrije UniversiteitBrussel, Brussels, Belgium; and the Transplantation Laboratory and Chil-dren's Hospital (T.O., J.U., M.-A.H.), University of Helsinki, Helsinki, Finland.

Address correspondence and reprint requests to Veronique H. Lefebvre,Diabetes Research Center, Vrije Universiteit Brussel, Laarbeeklaan 103,B-1090 Brussels, Belgium. E-mail: velefeb@mebo.vub.ac.be.

Received for publication 29 August 1997 and accepted in revised form9 October 1997.

BrdU, 5-bromo-2'-deoxyuridine; HGF, hepatocyte growth factor .

vivo, the main mechanism leading to increased p-cell numbersappears to involve neogenesis from ductal progenitor cells(2,3). Human p-cells exhibit a very low proliferative activityboth during fetal development (9) and in adult life (4).According to a recent report, the proliferation of fetal humanp-cells can be stimulated by hepatocyte growth factor (HGF)(5). Furthermore, also adult human p-cells appear to prolif-erate when cultured in the presence of HGF and an extra-cellular matrix prepared from rat 804G cells (6). The presentstudy compares the mitogenic effect of this condition on thedifferent cell types that compose a human islet preparation.Our findings contradict the conclusion of Hayek et al. (6), butindicate, instead, that HGF stimulates the proliferation ofadult pancreatic duct cells.

RESEARCH DESIGN AND METHODSIslet isolation and culture. Human islets from ten heart-beating organ donorswere isolated at the Central Unit of p-cell Transplant, Medical Campus, VrijeUniversiteit, Brussels. The mean age of the organ donors was 41 ± 5 years (means± SE [range 19-64]) and the mean organ preservation time 11 ± 2 h (range 4-20h). Pancreases were processed by ductal distension with collagenase, gentle dis-sociation, and Ficoll gradient purification of islets. Islet-enriched fractions werecultured in serum-free Ham's F10 medium (Gibco, Life Technologies, Paisley, Scot-land, U.K.) with 7.5 mmol/1 glucose, 1% bovine serum albumin, 0.075 ing peni-cillin/ml, 0.1 mg streptomycin/ml, and 2 mmol/1 glutamine (7). After 4-16 days ofculture, these preparations contained <5% damaged cells, no acinar cells, 25 ± 5%ductal cells, 63 ± 5% insulin-positive cells, and 12 ± 2% glucagon-positive cells. Theywere recovered within this culture period and distributed into 35-mm dishes foreither suspension culture or monolayer culture on 804G matrix (6) for a subse-quent 1-7 days culture in RPMI1640 medium (Gibco, Life Technologies) with 5.5mmol/1 or 11.1 mmol/1 glucose, 10% pooled AB human serum (BioWhittaker,Walkersville, MD, or Finnish Red Cross, Helsinki, Finland), and antibiotics (0.075mg penicillin/ml and 0.1 mg streptomycin/ml) with or without human recombinantHGF (10-15 ng/ml) (Sigma, St. Louis, MO, or R&D Systems, Abingdon, U.K.).Medium was replaced every 2 days. Parallel samples were carried in 24-wellplates (Falcon, Becton Dickinson, Lincoln Park, NJ) and on Thermanox coverslips(Nunc, Naperville, IL) coated with the 804G matrix.

Proliferation assay and immunocytochemistry. During the last hour or 24 hof culture, 5-bromo-2'-deoxyuridine (BrdU) (Sigma) 0.1 mmol/1 was added and thepreparations were then removed with trypsin/EDTA (Gibco) or dispase(Boehringer Mannheim, Mannheim, Germany), washed, fixed overnight in 4% neu-tral-buffered formaldehyde, and embedded in paraffin. Immunohistochemicalstaining of sections was performed by the streptavidin-biotin method. For analy-sis of BrdU and cytokeratin-19 staining, sections were pretreated with trypsin toretrieve antigenicity (8,9). For double-staining, peroxidase/diaminobenzidine(brown color) and alkaline phosphatase/New Fuchsin (red color) were used forBrdU and cytokeratin 19 or insulin, respectively. Negative controls consisted ofomission of the primary antibody. Electron microscopy and in situ hybridizationfor insulin mRNA were performed as previously described (10). The percentpositive cells is presented as a mean ± SE for n independent experiments. Sta-tistical significance of differences was calculated by one-way analysis of variance(ANOVA) and Scheffe test.

Antibodies. Monoclonal antibody to cytokeratin 19 was from Dako (Glostrup,Denmark; clone RCK108), monoclonal antibody to cytokeratin 7 from Biogenex

134 DIABETES, VOL. 47, JANUARY 1998

V.H. LEFEBVRE AND ASSOCIATES

(San Ramon, CA), monoclonal antibody to carbohydrate antigen 19-9 from Cis BioInternational (Gif-sur-Yvette, France), and monoclonal anti-BrdU from Eurodi-agnostics (Apeldoorn, The Netherlands). Polyclonal guinea pig anti-insulin andrabbit anti-glucagon were kindly provided by Dr. C. Van Schravendyk (DiabetesResearch Center, Free University Brussels, Belgium), and polyclonal rabbit anti-carbonic anhydrase II was a gift from Dr. A.-K. Parkkila (Department of Anatomy,University of Oulu, Finland). Biotinylated secondary antibodies and ABC kits werepurchased from Dako.

RESULTS

After 1,2,4, or 7 days of culture on 804G matrix, <0.1% of theinsulin-positive cells had incorporated BrdU, while manyinsulin-negative cells became BrdU+, starting within 48 h andreaching a maximum after 4 days. The decrease in prolifera-tion after 7 days is probably caused by contact inhibition,since the cultures were nearly confluent at this time. Themitogenic effect on the insulin-negative cells was higher in thepresence of HGF (Fig. 1). The BrdLT cells were negative forpreproinsulin mRNA in situ hybridization but expressed duc-tal cell markers such as cytokeratin 19 (Fig. 2B), cytokeratin7, carbonic anhydrase-II, and carbohydrate antigen 19-9 (notshown). These four markers have been previously shown tobe specifically localized in pancreatic duct cells (9,11-14).

The proliferative activity of the different cell types wasassessed by determining the percentages of insulin-posi-tive, glucagon-positive, and cytokeratin 19+ cells that werealso positive for BrdU after a 4-day culture in the fourselected test conditions: suspension or matrix with andwithout HGF. Less than 0.1% of the insulin-positive cellsand <1% of the glucagon-positive cells were found to beBrdU+, irrespective of the presence of matrix or HGF (Table1). On the other hand, 30% of the cytokeratin 19+ cells wereBrdlT after culture on matrix plus HGF, which is significantlymore than after culture in suspension or on matrix withoutHGF (Table 1). These observations were made after detach-ing the cell layers from the bottom of the culture dish withdispase. Also, when these preparations were completelydissociated into single cells, no insulin BrdU double-positivecells were detected, but many cells were positive for bothBrdU and ductal markers.

Islet cell preparations cultured on matrix plus HGF werefound to be organized as bi- or multilayers with the insulin-pos-itive cells frequently covered by a thin layer of cytokeratin 19+

cells. This particular organization could only be discerned insections (Fig. 2A-B,E-F). Consequently, when these prepara-tions are viewed unsectioned from the top or the bottom, theywill generate images of BdrU+ non-endocrine cells that aresuperposed on insulin-positive cells , and hence may suggestthe presence of cells that seem double-positive for BrdU andinsulin (Fig. 2CJD). At the end of the 24-h culture period withBrdU, the preparations presented couples of adjacent BrdU+

cells, suggesting the formation of daughter cells (Fig. 2A, E).

DISCUSSIONThe present study questions the recent statement thathuman p-cells proliferate on an extracellular matrix inmedium supplemented with HGF (6). This culture condi-tion was indeed found to stimulate the mitotic activity inhuman islet cell preparations, but the effect was almostexclusively located in the ductal cells, which are consis-tently associated with isolated human endocrine islet cells(unpublished observations). These ductal cells can be iden-tified by specific markers such as cytokeratin 19, cytokeratin

30 r

20

•a 1 0

Time (days)

FIG. 1. Percent BrdlT cells in human islet cell preparations (n = 5) cul-tured for 1-7 days at 5.5 mmol/1 glucose on 804G matrix with or with-out HGF 15 ng/ml; BrdU was added during the last hour of culture.Comparison of BrdU positivity in insulin-positive (•; results formatrix alone and matrix + HGF are identical) and insulin-negative cells(• , matrix; D, matrix +HGF). *P < 0.05 matrix + HGF versus matrixwithout HGF.

7, carbonic anhydrase-II, or carbohydrate antigen 19-9(9,11-14). In sections from detached cell cultures, wenoticed the formation of a ductal cell monolayer on top ofclusters of insulin-positive cells. This particular arrange-ment may create false impressions of double-positivity inunsectioned material that is viewed from the top or the bot-tom. We assume that this phenomenon may have caused anerroneous interpretation on the identity of the BrdU+ cells inthe study of Hayek et al. (6).

Double immunocytochemistry on sectioned materialclearly indicated that virtually all BrdU+ cells in the matrixplus HGF condition were negative for insulin or glucagon andpreproinsulin mRNA, but positive for the ductal cell markers.This was the case in preparations both with a higher (88%) orlower (44%) percentage of endocrine cells at the start of cul-ture. Analysis of the time course of cell-specific prolifera-tion confirmed that only insulin-negative cells proliferated atany time point, thus excluding the possibility that the BrdU-labeled cells would represent transdifferentiated (3-cells.

Stimulation of duct cell proliferation was only significant forthe combination of the 804G matrix and HGF. That HGF is amitogen for human pancreatic epithelial cells has alreadybeen reported by Vila et al. (15). HGF-induced proliferation offetal human (B-cells has also been clearly demonstrated (5). Inthe present study, we failed to demonstrate a HGF-inducedproliferation of adult human pancreatic 3-cells. Irrespectiveof the presence of HGF and/or the 804G matrix, (3-cells exhib-ited an extremely low proliferative activity in vitro. Less than0.1% of p-cells were found to be BrdU+, confirming previousreports on the negligible mitotic potential of differentiatedhuman (3-cells, both in vitro (4) and in vivo (9).

DIABETES, VOL. 47, JANUARY 1998 135

CULTURE OF ADULT HUMAN ISLET PREPARATIONS

FIG. 2. Human islet cell preparations cultured for 4 days at 11.1 mmol/1 glucose with 804G matrix and HGF 10 ng/ml; BrdU was added duringthe last 24 h of culture. A and B: consecutive sections from detached monolayers (420 x ); A is double immunostained for BrdU (brown) andinsulin (red), and B for BrdU (brown) and cytokeratin 19 (red). All BrdU* cells are insulin-negative and cytokeratin 19*. C and D: unsectionedpreparation after double immunostaining for insulin (red) and BrdU (brown) (C) and for cytokeratin 19 (red) and BrdU (brown) (Z)) (210x) .The BdrU* cells in the insulin-positive clusters (arrows) correspond to duct cells, which have spread over the endocrine cells. E and F: elec-tron micrographs showing nongranulated cells with microvilli (arrows) covering an a- (2?) (4800x) or p-cell (F) (19,200x).

136 DIABETES, VOL. 47, JANUARY 1998

V.H. LEFEBVRE AND ASSOCIATES

TABLE 1Effect of matrix and/or HGF on percent BrdU+ cells (n = 4)

SuspensionSuspension +804G matrix804G matrix

HGF 10 ng/ml

+• HGF 10 ng/ml

Cytokeratin

% of total

25 ±629 ±834 ±940 ±10

19-positive% with BrdUpositivity*

2 ± 15 ± 3

10 ±530±6t

Cell typeInsulin-positive

% with BrdU% of total positivity*

48 ±5 <0.148 ±6 <0.148 ±6 <0.142 ± 7 <0.1

Glucagon-positive% with BrdU

% of total positivity*

13 ±2 <113 ±3 <115 ±3 <112 ±3 <1

Data are means ± SE. *BrdU was added 24 h before analysis, medium containing 11.1 mmol/1 glucose. tP < 0.01 vs. suspensions andP < 0.05 vs. matrix.

ACKNOWLEDGMENTSThis work was supported by grants from the European Com-munity (BMH4-CT 95-1561), the Juvenile Diabetes FoundationInternational (DIRP 1995-2000 and grant 196987), a Con-certed Action with the Flemish government (GOA 92/94-12),the Scientific Research Council (Belgium) (3.0057.94), theSigrid Juselius Foundation, the Foundation for DiabetesResearch (Finland), and the Research Fund of the HelsinkiUniversity Central Hospital.

V.H.L. is Postdoctoral Fellow of the Fund for ScientificResearch-Flanders (Belgium). T.O. is the recipient of a Juve-nile Diabetes Foundation International Career DevelopmentAward.

We thank the personnel of the Central Unit of p-Cell Trans-plant for preparing the human islet cells and Emmy De Blayfor technical assistance in the morphology analysis.

REFERENCES1. Hellerstrom C, Swenne I: Functional maturation and proliferation of fetal pan-

creatic p-cells. Diabetes 40 (Suppl. 2):89-93,19912. Bouwens L, Kloppel G: Islet cell neogenesis in the pancreas. Virchows Arch

427:553-560,19963. Bonner-Weir S, Baxter LA, Schuppin GT, Smith FE: A second pathway for

regeneration of adult exocrine and endocrine pancreas: a possible recapitu-lation of embryonic development. Diabetes 42:1715-1720, 1993

4. Tyrberg B, Eizirik DL, Hellerstrom C, Pipeleers DG, Andersson A: Human pan-creatic 3-cell deoxyribonucleic acid-synthesis in islet grafts decreases with

increasing organ donor age but increases in response to glucose stimulationin vitro. Endocrinology 137:5694-5699,1996

5. Otonkoski T, Cirulli V, Beattie GM, Mally MI, Soto G, Rubin JS, Hayek A: A rolefor hepatocyte growth factor/scatter factor in fetal mesenchyme-inducedpancreatic (3-cell growth. Endocrinology 137:3131-3139,1996

6. Hayek A, Beattie GM, Cirulli V, Lopez AD, Ricordi C, Rubin JS: Growth fac-tor/matrix-induced proliferation of human adult p-cells. Diabetes44:1458-1460,1995

7. Ling Z, Pipeleers DG: Prolonged exposure of human 3-cells to elevated glu-cose levels results in sustained cellular activation leading to a loss of glucoseregulation. J Clin Invest 98:2805-2812, 1996

8. Bouwens L, Wang RN, De Blay E, Pipeleers DG, Kloppel G: Cytokeratins asmarkers of ductal cell differentiation and islet neogenesis in the neonatal ratpancreas. Diabetes 43:1279-1283,1994

9. Bouwens L, Lu WG, De Kryger R: Proliferation and differentiation in thehuman fetal endocrine pancreas. Diabetologia 40:398-404,1997

10. Wang RN, Bouwens L, Kloppel G: Beta-cell proliferation in normal and strep-tozotocin-treated newborn rats: site, dynamics and capacity. Diabetologia37:1088-1096, 1994

11. Ichihara T, Nagura H, Nakao A, Sakamoto J, Watanabe T, Takagi H: Immuno-histochemical localization of CA 19-9 and CEA in pancreatic carcinoma andassociated diseases. Cancer 61:324-333,1988

12. Trautmann B, Schlitt H-J, Hahn EG, Lohr M: Isolation, culture, and charac-terization of human pancreatic duct cells. Pancreas 8:248-254,1993

13. Ramaekers F, Huysmans A, Schaart G, Moesker O, Vooys P: Tissue distribu-tion of keratin 7 as monitored by a monoclonal antibody. Exp Cell Res170:235-249, 1987

14. VM MR, Lloreta J, Real FX Normal human pancreas cultures display functionalductal characteristics. Lab Invest 71,423-431,1994

15. Vild MR, Nakamura T, Real FX: Hepatocyte growth factor is a potent mitogenfor normal human pancreas cells in vitro. Lab Invest 73:409-418,1995

DIABETES, VOL. 47, JANUARY 1998 137