EXPERIMENTAL

CELL

RESEARCH

191,

305-312 (19%))

Differences in Aggregation Properties and Levels of the Neural Cell Adhesion Molecule (NCAM) between Islet Cell Types DOMINIQUE Laboratoires

G. ROUILLER,’

VINCENZINO

de Recherche Louis Jeantet,

Centre Mkdical

Academic

Press.

AND PHILIPPE

Universitaire,

A. HALBAN

CH-121 I Geneva 4, Switzerland

PP-cells, secreting insulin, glucagon, somatostatin, and pancreatic polypeptide, respectively. These cells are not randomly distributed within the islets. Rather, they are typically organized as a core of B-cells surrounded by the three other non-B-cell types in various proportions [ 11. There is good evidence suggesting that this characteristic three dimensional organization is of physiological significance. Thus, the microcirculation of the islet is also well organized, with arterial flow irrigating the B-cell core before drainage into the venous system via the non-B-cell mantle [2]. At the functional level, dissociation of adult rat islets into single cells results in a loss of regulated function in vitro, while spontaneous reaggregation of the cells in culture is associated with restoration of normal basal and glucose-stimulated insulin release [3, 41. Interestingly, it was observed that these “pseudoislets” display a cellular organization similar to that of native islets, i.e., with B-cells at the center, surrounded by a discontinuous layer of non-B-cells [4, 51. Transformed rat B-cells of the RIN2A line, however, when cocultured with primary rat islet cells, segregate at the periphery of aggregates [6]. Thus, adult rat islet cells seem to retain the potential for presenting the appropriate signals required to reassume their typical three-dimensional organization. This property of Bcells is lost by transformation. Taken together, these findings suggest that a specific deployment of adhesion molecules characterizes each islet cell type. To test this hypothesis, the aggregation properties of purified islet B-cells, non-B-cells, and RIN cells have now been studied. The results show that unlike B cells, islet non-Bcells and RIN cells are able to aggregate in the absence of calcium, suggesting the presence on the surface of these cells of biologically significant concentrations of Ca2+-independent cell adhesion molecules (CAMS).~ We therefore looked for quantitative or qualitative differences in expression of the Ca2+-independent neural CAM (NCAM) among primary islet B-cells, non-B-

Cells within rat islets of Langerhans are typically organized as a core of B-cells, surrounded by the other cell types. When mixed in culture, primary islet cells and insulinoma (RINZA) cells form aggregates where Bcells are centrally located, surrounded by non-B-cells, while RIN-cells segregate as the outermost layer. To gain insight into the molecular basis underlying this nonrandom cellular organization, the aggregation properties of the three cell populations were studied. Isolated islet cells were separated into B-cells and non-Bcells by autofluorescence-activated cell sorting (FACS). In a short-term aggregation assay, primary B-cell aggregation in the absence of calcium was only 19 t 3.7%, compared to the 67 k 2.9% seen in the presence of calcium (mean ? SEM; P < 0.001; n = 7). By contrast, non-B-cell aggregation and RIN cell aggregation in the absence of calcium (62 + 2 and 66 f 2%, respectively) were only slightly less than with calcium (70 + 3 and 76 t 3%). The surface density of the Ca2+-independent neural CAM (NCAM) was therefore measured by flow cytometry and found to be 2.64 + 0.82-fold higher in non-B-cells, compared to that in B-cells (P < 0.01; n = 3). Even higher levels were found on RIN cells. In the three cell types, NCAM-140 was the only molecular form detected by immunoblotting. In conclusion, differences in the calcium dependency of aggregation and in the levels of NCAM are demonstrated among islet Bcells, non-B-cells, and RIN cells. Because cell-cell adhesion is crucial for the maintenance of adult tissue, these aggregation specificities might contribute to the concentric segregation of islet cell types in culture and to the nonrandom distribution of cells within rat islets. Q 1990

CIRULLI,

Inc.

INTRODUCTION

Islets of Langerhans are complex endocrine microorgans scattered throughout the pancreas of mammals. They are mainly composed of endocrine B-, A-, D-, and

’ Abbreviations used: CAMS, cell adhesion molecules; NCAM, neural cell adhesion molecule; BSA, bovine serum albumin; FLS, forward light scatter; HBSS, Hanks’ balanced salt solution; FCS, fetal calf serum; KRB, Krebs-Ringer-bicarbonate buffer; LIGFL, log integral green fluorescence; RIN cells, Rat Insulinoma cells.

’ To whom reprint requests should be addressed at Laboratoires de Recherche Louis Jeantet, Centre Medical Universitaire, 1, rue Michel Servet, CH-1211 Geneva 4, Switzerland. 305

All

0014-4827190 $3.00 Copyright 0 1990 by Academic Press, Inc. rights of reproduction in any form reserved.

306

ROUILLER,

CIRULLI,

cells, and RIN cells and found higher levels of this molecule in the cell types displaying Ca’+-independent aggregation.

AND

HALBAN

cubation was carried out in a siliconized mature aggregation. Flow Cytometric

METHODS Cell Preparation

and Culture

Islet isolation. Pancreatic islets were isolated from Sprague-Dawley rats (weighing 2009240 g) by a modification of the method of Sutton et al. [7]. Briefly, six overnight-fasted rats were anesthetized with 0.2 mg pentothal i.p./lOO g body wt (Abbot, Cham, Switzerland), their pancreas was exposed, and a cannula was inserted downstream into the bile duct. After clamping the pancreatic duct at its duodenal outlet, 16 mg of collagenase (type 1, Cat. No. C-0130, Sigma, St. Louis, MO), diluted in 6 ml ice-cold Hanks’ balanced salt solution (HBSS), pH 7.2, containing 6 mM CaCl, was injected through the cannula. The distended pancreas was immediately dissected out and stored at 2-4°C in a 50-ml plastic tube until all six rats had been processed. Each pancreas was then washed twice with 50 ml of HBSS, prewarmed at 37°C. Digestion was for 20 min, under static conditions, at 37°C. The reaction was stopped by adding ice-cold HBSS, containing 0.35% (wt/vol) bovine serum albumin (BSA, fraction V, Cat. No. A-2153, Sigma) (HBSS-BSA). The digested pancreas was then homogenized by five passages through a 14.gauge (3” length) syringe needle and the homogenate was washed 2X with buffer. Grossly undigested pancreatic tissue was removed by filtering through a plastic tea-strainer. The filtrate of each pancreas was split into two 25.ml plastic tubes and centrifuged for 10 s at 450g. The pellet was then taken up in 10 ml Histopaque 1077 (Sigma) and overlaid with another 10 ml HBSS-BSA [8]. The purified islets were recovered at the interface after a 20.min centrifugation (15”C, 1OOOg)and washed 3~ with ice-cold HBSS-BSA, pH 7.4. They were finally hand-picked with a Gilson pipet under the stereomicroscope and kept at 4°C. The yield was about 800-1000 islets/rat. Preparation of islet cells. Freshly isolated islets were washed once with M$+, Ca’+-free phosphate-buffered saline containing 0.5 mM EDTA and once with the same buffer without EDTA. They were then resuspended into 1.5 ml of Puck’s buffer containing 0.16 mg/ml of trypsin (activity against casein, 1:250) and 0.1 mM EDTA. Digestion was carried out until only few doublet cells remained (6-7 min at 37°C). The reaction was ended by adding 10 ml ice-cold KrebsRinger-bicarbonate (KRB) buffer, pH 7.4, containing 0.5% BSA, 2.5 mM glucose, and 10 mA4 Hepes. Following centrifugation for 8 min at 4°C and 6OOg,cells were taken up in the same buffer to a final concentration of 3 X lo6 cells/ml. The yield was l-l.5 X lo6 cells/rat. RIN cell culture. RIN-m5F cells were obtained from Dr Gazdar (Georgetown University, Washington, DC) [9]. A subline (RlN2A) was subsequently cloned in Geneva (P. A. Halban, C. B. Wollheim, and B. Blondel, unpublished data). This subline has a somewhat higher insulin content and secretory response to glyceraldehyde than the original line. The cells were grown in RPM1 1640 (GIBCO-Irvine, Scotland), containing 10% heat-inactivated fetal calf serum (FCS, GIBCO) and supplemented with antibiotics (250 U/ml penicillin; 0.11 mg/ml streptomycin). They were trypsinized and replated weekly at a 1:lO dilution. Primary cell culture. For immunoblot analysis, 5 X lo5 cells of either nonsorted islet cells or sorted B-cells and non-B-cells were resuspended in 5 ml Dulbecco’s minimum essential medium (DMEM from GIBCO), containing 10% FCS and 8.3 mM glucose, and seeded in 35-mm petri dishes to which they do not adhere (Cat. No. 1007, Falcon, Oxnard, CA). They were kept for 20 h in a humidified atmosphere of 95% air, 5% CO,, at 37°C to allow them to fully regenerate any lost or damaged cell surface proteins. Cells destined for the aggregation assay were diluted further (to lo4 cells/ml) in medium supplemented with 25 mM Hepes and only 5% FCS, and the 20-h prein-

Analysis

spinner flask to prevent pre-

and Sorting of Islet and RIN Cells

Dispersed islet or RIN cells were analyzed in an Epics-V flow cytometer connected to an MDAS microcomputer (Coulter Electronics, Hialeah, FL). Light scatter, which relates to cellular size [lo], was measured as near forward light scatter intensity (FLS). Cellular autofluorescence or cell-bound FITC antibody fluorescence was excited by an argon laser beam tuned to 488 nm at 500-600 mW output power. As reported [ll], at 2.5 mM glucose, two islet cell populations become apparent when particle flavin adenine dinucleotide (FAD) autofluorescence (510-550 nm) was plotted against light scatter. Sorting “windows” were then externally applied around both populations in order to deflect viable cells into one or the other collecting tube containing sterile KRB-BSA. The distribution of insulin and glucagon cells was assessed by classical double antibody cytochemistry [12]: in brief, cells were applied to poly(t-lysine)-coated cover glasses, fixed in Bouin’s solution, and permeabilized by dehydration. They were then sequentially incubated first with either a guinea pig anti-insulin antiserum (Cappel/Dynatech, Kloten, Switzerland, dilution 1:400), a rabbit anti-glucagon antiserum (Novo Industries, Bagsvaerd, Denmark; dilution l:lOO), or a nonimmune serum, and then with FITC-labeled anti-species antisera (Biosys, Compiegne, France, and Genofit, Geneva, Switzerland; dilution, 1:200). Cells were finally stained with Evans’ blue (Fluka AG, Buchs, Switzerland), allowing the detection of nonfluorescent cells under fluorescence microscopy. A minimum of 600 cells were counted per sample. Aggregation

Assay

At the end of a 20-h culture in spinner flasks, cells (either primary or transformed) were washed in cold Ca*+, Mg2+-free KRB containing 10 mM glucose, 0.5% BSA, 0.5 mMEDTA, resuspendedin 4 ml of this same buffer, and applied on top of a discontinuous Percoll gradient (3.5 ml each of isotonic 30 and 60% Percoll; Pharmacia, Denmark) in order to eliminate dead cells and cell debris. Viable cells (harvested from the 30%/60% Percoll interface after centrifugation for 10 min at 9OOg; room temperature) were then divided in two pools and washed with KRB containing 10 mM glucose, 0.5% BSA, and either 1 mM CaCl, or 0.5 mM EDTA, as previously described [ 131. EGTA was used occasionally and gave identical results. Cells were resuspended in 150 ~1 of the same buffers at a concentration of 5 X lo5 cells/ml in lo-ml polycarbonate conical tubes (Nunc, Roskilden, Denmark). In perturbation experiments, 75 fig of anti-NCAM Fab fragments (prepared by Dr U. Rutishauser) or nonimmune fragments were added and cells were preincubated for 45 min at 4”C, under static conditions. The tubes were then placed at a fixed angle (30”) in a shaking waterbath (100 cycleslmin) at 37°C for 30-45 min, a time in which a plateau of aggregation is reached (data not shown). Aggregation was assessed both qualitatively under the optic microscope and quantitatively by comparing the number of events before and after the aggregation period as measured in a ZM Coulter counter (Coulter Electronics). The following formula was then applied,

% of aggregation

= -(b - a) x 100, b

where b is the number of events before aggregation ber of events counted after the 45-min aggregation Flow Cytometric

Analysis

of NCAM

and a is the numperiod.

Levels

Isolated islet cells suspended in KRB-BSA were sequentially immunolabeled first with an anti-NCAM antiserum (kindly provided by Drs U. Rutishauser, Cleveland, OH [14], or C. Goridis, Marseille,

ISLET

CELL

TYPE

307

AGGREGATION RESULTS

Islet Cell Sorting

cytometric analysis of rat islet cells. FIG. 1. Biparametric Freshly isolated islet cells were preincubated with KRB-BSA containing 2.5 mM glucose. Cells (106) were then analyzed on a FACS V system with argon-ion laser excitation (500 mW at 488 nm). The log autofluorescence (510-550 nm) vs forward light scatter (FLS) histogram of lo6 cells is presented. They axis is presented as 126 channels which span 3 decades. The r axis is represented as 126 channels linearly correlated to cell size. The display threshold was set to 300 events/channel. Two populations of islet cells can be separated, the compositions of which are presented in Fig. 2.

France [15]) or a nonimmune serum, and second, after washing cells twice, with an FITC-labeled anti-rabbit IgG antiserum. Labeled cells were again washed twice and analyzed by flow cytometry as described for cell sorting. The mean log integral fluorescence (LIGFL) and FLS of each islet cell population were displayed on a 64 X 64 channel scale (channel units). NCAM-specific fluorescence of a given population was calculated as the mean fluorescence of the population after incubation of islet cells with anti-NCAM antiserum, minus the mean fluorescence after incubation with nonimmune serum. Cell surface was calculated from FLS data of islet cells compared to calibration beads. The relative NCAM surface density was calculated as the ratio of specific NCAM LIGFL to cell surface [16].

Immunoblot

Analysis

Cultured cells were homogenized by sonication (3 X 2 s) in 10 vol lysis buffer (50 mM Tris-HCl, pH 8.5, 1% NP-40,5 mM EDTA, 150 mM NaCl, 0.1% BSA, 350 pg/ml PMSF, 10 pg/ml leupeptin) [17], followed by a lo-min rest at 4°C. The supernatant of a 20-min centrifugation at 10,OOOgwas concentrated by ultrafiltration at 4°C through a Centriconmembrane (Cat. No. 4205, Amicon, Danvers, MA). Total protein was measured using the kit from Bio-Rad (Munich, FRG). Equal amounts of proteins were mixed 2:3 in sample buffer (9% SDS, 15% glycerol, 30 mM Tris, pH 7.8, 0.05% bromophenol blue, 6% mercaptoethanol), boiled for 5 min, and separated by electrophoresis on 0.1% SDS/7.5% polyacrylamide gel, accordingto a modification of Laemmli [18], as previously described [19]. High molecular weight markers were obtained from Bio-Rad. Proteins were then transferred to BA85 nitrocellulose filters (Schleicher & Schuell, Dassel, West Germany) in 96 mM glycine, 12 mM Tris, 20% methanol for 6 h at a field strength of 6 V/cm. Both NCAM and an internal standard, actin, were revealed by sequential incubation of the filters first with a rabbit anti-NCAM antibody (dilution 1:500 [14]), combined with a rabbit anti-actin antibody (dilution 1:500, kindly provided by Dr G. Gabbiani, Geneva), and second with 0.5 &i/ml of a 1251-labeled goat antiserum (sp act, 8.6 &i/fig; New England Nuclear, Boston, MA) raised against rabbit IgGs [20]. Autoradiography was for 1-3 days at -70°C.

Islet cells were obtained by mild trypsin digestion of freshly isolated rat islets. After being washed and resuspended in low glucose KRB-BSA, they were analyzed by flow cytometry. Two populations of cells could be easily recognized on a biparametric dot plot display, where FAD autofluorescence was one parameter (log scale) and forward light scatter was the other parameter (Fig. 1). Gating was then applied externally around both populations so that each analyzed event was deflected accordingly into one or the other collecting vessel. The resulting cell type distribution can be assessed by double antibody immunofluorescence and shows that as a mean of 12 experiments, population 1 contains 93 f 1.6% non-B-cells, while population 2 contains 95 ? 0.7% B-cells (Fig. 2). Roughly 8 million islet cells were usually obtained from six rats. The distribution in the initial population was 71.8 ? 2.2% B-cells and 27 ? 1.7% non-B-cells (n = 7). After cell sorting, the yield was usually 0.9 million viable non-B-cells and 2.5 million B-cells. Short-Term

Aggregation Assay

CAMS are commonly divided into two main classes according to their functional dependence on calcium [21]. Awareness of the calcium-dependence of aggregation of cells under consideration might indicate which adhesion molecules are involved. We therefore studied the role of calcium in aggregation of defined islet cell types. After a 20-h culture in spinner flasks, sorted primary islet B-cells, primary non-B-cells, and transformed Bcells (RINBA line) were concentrated and incubated for

BEFORE SORTING

POPLkATlON

1

POPULATlCN

2

FIG. 2. Islet cell type distribution before and after fluorescenceactivated cell sorting. Nonsorted cells and the two sorted cell populations marked in Fig. 1 were collected and seeded on poly(L-lysine)coated coverglasses. Classical double antibody immunofluorescence was then applied to count the proportion of insulin-containing cells (B-cells) and glucagon-containing cells. Population 1 contains 93 i 1.6% non-B-cells, while population 2 contains 95 * 0.7% B-cells (mean f SEM, n = 12).

308

ROUILLER,

B-CELLS

CIRULLI,

NON

AND

HALBAN

RIN-CELLS

B-CELLS

+

FIG. 3. Effect of calcium on islet cell type aggregation. I. Qualitative differences. After a 20-h culture in a spinner flask, sorted primary islet B-cells, primary non-B-cells, and transformed B-cells (RIN2A line) were concentrated and incubated for 45 min under standardized conditions (see text) in the presence of either 1 mMCaC1, (+) or 0.5 mMEDTA (6). Aggregation was evaluated qualitatively under phase-contrast microscopy. By contrast to B-cells, non-B-cells and RIN cells do aggregate in the absence of calcium. Note that while cell limits are difficult to identify in the presence of calcium, they remain well delimited at this magnification in the absence of calcium. Bar = 50 Frn.

45 min under standardized conditions in the presence of either 1 mM calcium or 0.5 mM EDTA. Aggregation was evaluated both qualitatively under the microscope and quantitatively by comparing the number of particles at the end of incubation with the starting conditions [22], using a Coulter counter. In the presence of calcium, islet B-cells, non-B-cells, and RIN cells form dense aggregates in which individual cell limits are barely recognized (Fig. 3, upper panels), suggesting the presence of calcium-dependent CAMS on the three cell types. In the absence of calcium (lower panels) B-cell aggregation is minimal (left lower panel), while islet non-B-cells and RIN cells form big aggregates (lower middle and right panels). Cell limits within aggregates are well recognized, suggesting a more focal type of adhesion, compared to the aggregation seen in the presence of calcium. These results would indicate the presence of calcium-independent CAMS on non-Bcells and RIN cells only. This was confirmed by quantitative measurements of aggregation in a Coulter counter: seven independent experiments were performed; 100% aggregation represents the theoretical sit-

uation where cells are all included in a single big aggregate. Aggregation in the presence of calcium was not statistically different among the three cell types. In the absence of calcium, B-cell aggregation was only 19 + 3.7% (mean * SEM), compared to the 67 + 2.9% seen in the presence of calcium (Fig. 4, left pair of bars; P < 0.001). By contrast, for non-B-cells (Fig. 4, middle pair) and RIN cells (Fig. 4, right pair), aggregation in the absence of calcium (62 * 2 and 66 + 2%, respectively) was only slightly, albeit significantly (P < 0.005 for both groups), less than with calcium (70 +- 3 and 76 t 3%). Taken together these results suggest biologically significant differences in expression of calcium-independent CAMS between islet cell types. NCAM

Flow Cytometry

The aggregation assay suggested the presence of Ca*+ -independent CAMS at the surface of non-B-cells and RIN-cells, but not of non-B-cells. We therefore next compared the expression of the Ca*+-independent NCAM among the three cell types. Cells were immuno-

ISLET

CELL

TYPE

309

AGGREGATION

100

5

80

s 2

60

k? ‘f

40

8 90 20

0

FLS B-CELLS

NON B-CELLS

FIG. 4. Effect of calcium on islet cell type aggregation. II. Quantitative differences. The short-term aggregation seen in Fig. 3 was evaluated quantitatively by comparing the number of particles at the end of incubation with the starting conditions, using a Coulter counter. Bars represent the mean + SEM of seven experiments. One hundred percent aggregation would represent the theoretical situation where all cells are included in a single big aggregate. Although the absence of calcium resulted in a significant decrease in aggregation of the three cell types, the effect on sorted primary B-cells was much more pronounced.

labeled first with either nonimmune serum or a specific rabbit anti-NCAM antiserum (provided by Dr U. Rutishauser) and second with an FITC-labeled goat antiserum raised against rabbit IgG. Labeled cells were analyzed by flow cytometry, as before. When incubated with nonimmune first antiserum, both islet cell populations were found at the usual location on a size vs log fluorescence plot (Fig. 5A, compared with Fig. 1). Incubation of cells with the anti-NCAM antiserum resulted in a shift of both populations on the log fluorescence scale (Fig. 5B, compared to Fig. 5A). NCAM-specific fluorescence, however, was 1.96 t- O.ll-fold higher in non-B-cells, compared to B-cells (mean & SEM, P

B NON B-CELLS B-CELLS

2 B-CELLS ’ FLS

NON B-CELLS (CELL

SIZE)

SIZE)

FLS

(CELL

SIZE)

FIG. 6. Flow cytometric analysis of NCAM levels on rat insulinoma cells (RINBA Line). Isolated RIN cells were immunolabeled and analyzed by flow cytometry as described for primary islet cells in the legend to Fig. 5. (A) Nonimmune serum. (B) Anti-NCAM antiserum. Very high levels of NCAM are present on RIN cells.

< 0.01, n = 3). The mean surface of non-B-cells being 40% that of B-cells (results derived from FLS data), the relative surface density of NCAMs on non-B-cells is 2.64 + 0.82-fold higher than that on B-cells (mean * SEM, P < 0.01, n = 3). Identical results were obtained with another specific anti-NCAM antiserum, provided by Dr C. Goridis. These results suggested that NCAMs are expressed on both populations, but to a much higher degree on non-B-cells. The proportion of the total events recovered in each population was identical whether cells were preincubated with nonimmune or anti-NCAM antiserum. RIN cells, which also aggregated independently of calcium, displayed a marked increase in fluorescence when immunostained with antiNCAM antiserum (Fig. 6B), in comparison to cells incubated with nonimmune first antiserum (Fig. 6A). The relative surface density of NCAM on RIN cells is about 56% higher even than on non-B-cells. Perturbation

A

(CELL

MN-CELLS

FLS

(CELL

SIZE)

FIG. 5. Flow cytometric analysis of NCAM levels on primary islet cell types. Isolated islet cells were immunolabeled first with either nonimmune serum (A) or a specific rabbit anti-NCAM antiserum (B) and second with an FITC-labeled goat antiserum raised against rabbit IgG. Labeled cells were then analyzed by flow cytometry, as described in the legend to Fig. 1, except that only 50,000 cells and a 64 X 64 multichannel analyzer were used. The display threshold was set to 110 events/channel. NCAM-specific immunofluorescence was much greater on non-B-cells, compared to that on B-cells. See text for statistical analysis of the results.

Studies

To assess whether NCAM was the exclusive calciumindependent molecule on islet cell types, cells were preincubated with anti-NCAM or nonimmune Fab fragments followed by an aggregation assay, as described before. Preincubation of RIN cells, non-B-cells, or Bcells with anti-NCAM Fab had no effect on short-term aggregation (data not shown), whether calcium was present or not, suggesting the presence of other calcium-independent CAMS, in addition to NCAM, on islet cell types. These unknown calcium-independent CAMS are sufficient to provide a normal primary cell-cell adhesion in the absence of active NCAMs. Immunoblot

Analysis

Three main forms of NCAM have been reported, of molecular weights 180, 140, and 120 kDa [23]. The proportion of each depends upon the cell type studied. De-

310

ROUILLER,

CIRULLI,

6 rv-rv&l

206

NCAM-180 NCAM-140 NCAM-120

116 97

66

ACTIN

-

42

FIG. 7. Immunoblotting analysis of NCAM on islet cell types. To assess whether qualitative differences in the NCAM molecule might exist between islet cell types, sorted islet non-B-cells and B-cells, RIN cells, and rat brain were solubilized and the clarified material was electrophorezed on a polyacrylamide gel, under reducing and denaturing conditions. Cells from a total of 10 experiments were pooled to obtain enough non-B-cell material. Proteins were then transferred to nitrocellulose and immunolabeled first with a mixture of rabbit anti-NCAM and anti-actin antisera and second with an T-labeled anti-rabbit IgG antibody. Actin levels were used as an internal standard for a more reliable comparison of NCAM levels between islet cell types. Lanes 1-4 and lanes 5 and 6 were from separate experiments. 1, RINZA cells; 2, nonsorted primary islet cells; 3, rat brain; 4, purified actin; 5, primary islet B-cells; 6, primary islet non-B-cells.

velopment-related modifications with functional relevance are also known to occur post-translationally, resulting in altered electrophoretic mobility of the three species [24]. To assess whether such qualitative differences in NCAM might exist between islet cell types, sorted islet non-B-cells and B-cells, RIN cells, and rat brain were solubilized and the clarified material was electrophoresed on a polyacrylamide gel, under reducing and denaturating conditions. Frozen sorted cells from a total of 60 rats were pooled to obtain a sufficient amount of non-B-cell material. Proteins were then transferred to nitrocellulose and immunolabeled first with a rabbit anti-NCAM antiserum and second with an 1251-labeled anti-rabbit IgG antibody. As described by others, the three molecular weight species were easily recognized in brain (Fig. 7, lane 3). In islet cell types, however, whether primary (lanes 2, 5, and 6) or transformed (lane l), only the 140-kDa form was expressed. Further, there did not seem to be any variation in the apparent molecular weight of this band among B-cells, non-B-cells, and RIN cells. Thus no apparent qualitative difference in NCAM could explain the differential aggregation properties of islet cell types. A rabbit antiserum raised against actin was added to the anti-NCAM antiserum, to provide an internal stan-

AND

HALBAN

dard for NCAM level quantification (Fig. 7, lane 4). This approach is commonly used in nucleic acid level analysis [25]. The results show that despite minimal changes in actin, NCAM labeling was much more intense in RIN cells (lane 1) and non-B-cells (lane 6), compared to total nonsorted islet cells (lane 2) and sorted B-cells (lane 5). Assuming that actin concentrations are identical between islet cell types, this confirms the FACS results demonstrating higher levels of NCAM at the surface of cell types displaying Ca2+-independent aggregation properties. DISCUSSION

In the present work, we show that in contrast to Bcells, islet non-B-cells and transformed B-cells of the RINBA line do not need calcium for aggregation. This suggests the presence of functional Ca2+-independent CAMS only at the surface of non-B-cells and RIN cells. The demonstration of higher levels of the Ca2+-independent NCAM at the surface of RIN cells and non-B-cells, compared to B-cells, supports this assumption. No conspicuous qualitative difference in the apparent molecular size of NCAM was observed between the three islet cell types. Pancreatic islets or pseudoislets of Langerhans are arranged as a core of B-cells surrounded by a discontinuous mantle containing the other endocrine cell types (non-B-cells). To gain insight into the possible mechanism underlying this nonrandom cell distribution, the aggregation properties of both cell types need to be studied separately and compared. It is now well established that cell-cell adhesion and sorting in collectives are mediated by two mechanisms: the Ca2+-dependent and the Ca2+-independent systems [26]. Both systems can coexist in one cell. The Ca2+-independent system is resistant to trypsin under the conditions we used for islet cell isolation [27]. It can therefore be investigated immediately following islet cell isolation. By contrast, the Ca2+-dependent system is highly sensitive to trypsin in the absence of calcium [28] and requires a minimum of 8 h of cell recovery in culture before its contribution to cell aggregation can be assessed. For separating B-cells and non-B-cells, the FACS was used, as described [ll]. Sorted cells or freshly trypsinized RIN cells were then preincubated for 20 h in spinner flasks to let them regenerate their normal surface distribution of Ca2+-dependent CAMS before studying their aggregation properties in the presence or absence of calcium. In the presence of calcium, there was no qualitative or quantitative difference in homotypic adhesion among the three islet cell types. Furthermore, all three cell types show significantly better aggregation in the presence than in the absence of calcium. This suggests a similar expression of calcium-dependent CAMS on Bcells, non-B-cells, and RIN cells. As expected from

ISLET

CELL

TYPE

AGGREGATION

311

previous studies on a total population of islet cells [29], primary CAMS or that NCAM is not involved in the aggregation of B-cells is strictly dependent upon the ex- primary adhesion of islet cells which we are testing in the short-term aggregation assay. Even if NCAM is not tracellular presence of calcium. This calcium dependence appears to be lost by transformation, inasmuch as implicated in the early events of cell-cell recognition and adhesion, the differences in levels between cell an insulin-secreting line (the RINBA line) aggregates almost as well in the absence as in the presence of cal- types might still be responsible, at least in part, for the secondary arrangement of cells within islets or pseucium, emphasizing the limited usefulness of mutant doislets, since it has been shown that a S-fold increase in lines for gaining insight into primary islet cell physiolsurface density of a CAM can result in a greater than ogy. Most interesting is our finding that primary non30-fold change in cell aggregation forces [32]. Another B-cells also aggregate in the absence of calcium. The attractive hypothesis would be that NCAM might be calcium independence of a subclass of islet cells might involved in attachment of nerve terminals to islet cells. have been unnoticed in the past because of the small It is indeed well recognized that nerves enter the islets proportion of these cells in a mixed islet cell population zone [33], particularly rich in non(see Fig. 2). These results suggest that although islet cell in the heterocellular B-cells. types may all carry an identical panel of calcium-depenThree main forms of NCAM have been described, of dent CAMS, they seem to show differences in expression molecular weights 180,140, and 120 kDa. The three speof calcium-independent CAMS. This would support our working hypothesis suggesting specific deployment of cies are the products of a single gene and evolve from CAMS at the surface of islet cell types as the origin of alternate splicing of RNA [23]. The proportion of each modificathe nonrandom distribution of cells within islets. It also varies between tissues. Development-related tions with functional relevance are also known to occur may provide an explanation for the concentric segregapost-translationally, resulting in altered electrophoretic tion of B-cells, non-B-cells, and RIN cells when mixed mobility of the three species [24]. To assess whether in culture [6]. such qualitative differences in NCAM might exist It was formerly believed that cell type-specific moleamong islet B-cells, non-B-cells, and RIN cells, immucules might be responsible for the genetically deternoblot analysis was performed. A rat brain preparation mined segregation of cells in collectives [26]. According was also studied for comparison. Although the three to more recent studies, an alternative theory now species of NCAM were readily recognized in brain, only emerges, suggesting that quantitative or qualitative modulation of more ubiquitous molecules is sufficient to the 140-kDa form could be detected in primary or transformed islet cells. Thus no apparent qualitative differprovide the necessary range of binding activities [30]. ences in NCAM were detected between islet cell types, Depending upon the type of adhesion they provide, in variance from a recent study suggesting the presence CAMS are classified as calcium independent or calcium of the 180-kDa form on another rat insulinoma cell line dependent. Since our aggregation assay suggested the [34]. The 140-kDa species is devoid of the cytoplasmic presence of functional Ca2+-independent CAMS at the surface of non-B-cells and RIN cells but not B-cells, we domain connecting the 180-kDa molecule to the cytoskeleton. It is unknown whether this might be relevant compared the concentration of the Ca’+-independent for the apparent lack of participation of NCAM in islet NCAM at the surface of the three cell types. Among cell primary adhesion. immunofluorescence techniques, the FACS analytic The immunoblot analysis confirms the flow cytometsystem offers the greatest sensitivity and allows for objective quantification [31]. Using this technique, we ric data by showing higher levels of NCAM in RIN cells and non-B-cells, compared to B-cells. Actin was imfound NCAM to be present on all three islet cell types, thus confirming recent data on various endocrine tismunolabeled simultaneously as an internal control for a more accurate NCAM quantification. Despite this presues using conventional immunocytometric techniques caution and because the blotting experiment was per[34]. The data show however that NCAM fluorescence is much higher on non-B-cells and RIN cells, compared formed only twice (such experiments demanding proto B-cells, especially if one considers that the surface of hibitive numbers of the relatively precious islet non-Bcells), we prefer to rely on the flow cytometric data for a non-B-cells is about 40% that of B-cells. The absence of an effect of anti-NCAM Fab fragmore accurate determination of the relative levels of ments on the short-term aggregation of RIN cells and NCAM on islet cell types. non-B-cells in the absence of calcium reveals the exisIn conclusion, in the present work we demonstrate tence of other calcium-independent adhesion moledifferent aggregation properties of primary islet B-cells, cules, in addition to NCAM, on the surface of islet cell non-B-cells, and transformed B-cells, with reference to types. These unknown calcium-independent CAMS ap- calcium requirements. Quantitative differences in the pear to provide normal primary cell-cell adhesion, de- Ca2+-independent NCAM also characterize the three spite NCAM inactivation. This finding could mean that cell types. Because cell-cell adhesion is crucial for the either there is a redundance of calcium-independent maintenance of adult tissue, the differences might con-

312

ROUILLER,

CIRULLI,

tribute to the concentric segregation of these cell types when mixed in culture and to the nonrandom distribution of cells within rat islets. We are indebted to Florence Theraulaz for skillful technical assistance. We also thank Drs. U. Rutishauser, C. Goridis, and G. Gabbiani, who generously supplied the specific antisera, and D. Wohlwend from the Flow Cytometry Unit, who provided excellent maintenance of the cell sorter. This work was supported by Swiss National Science Fund (FNRS) Grants 32-8917.86 and 3100-009394.

AND

15. Gennarini, 16. 17.

1.

21.

Orci, L., and Unger, R. (1975) Lancet 2, 124331244.

S., and Orci, L. (1982) Diabetes 31, 883-889. 2. Bonner-Weir, 3. Halban, P., Wollheim, C. B., Blondel, B., Meda, P., Niesor, E. N., and Mintz,

D. H. (1982) Endocrinology

111, 86694.

4. Halban, P. A., Powers, S. L., George, K. L., and Bonner, W. S. (1987) Diabetes 36,783-790. 5. Hopcroft, D. W., Mason, D. R., and Scott, R. S. (1985) In Vitro Cell. Dev. Biol. 21, 421-427. 6. Halban, P. A., Powers, S. L., George, K. L., and Bonner, W. S. (1988) Endocrinology 123, 113-119. 7. Sutton, R., Peters, M., Mcshane, P., Gray, D. W., and Morris, P. J. (1986) Transplantation 42,689-691. 8. Tze, W. J., Wong, F. C., and Tingle, A. J. (1976) Transplantation

22,201-205. 9. Gazdar, A., Chick, W., Oie, H., Sims, H., King, D., Weir, G., and 10. 11. 12.

Lauris, V. (1980) Proc. Nutl. Acud. Sci. USA 77, 3519-3523. Loken, M., and Herzenberg, L. (1975) Ann. N. Y. Acud. Sci. 254, 163-171. Van De Winkel, M., Maes, E., and Pipeleers, D. (1982) Biochem. Biophys. Res. Commun. 107,525-532. Coons, A., Leduc, E., and Connolly, J. (1955) J. Exp. Med. 102,

49-63. 13. 14.

Takeichi, M. (1977) J. Cell Biol. 75, 464-474. Rutishauser, U., Watanabe, M., Silver, J., Troy, F. A., and Vimr, E. R. (1985) J. Cell Biol. 101, 1842-1849.

G., Rougon, G., Deagostini-Bazin, H., Hirn, M., and Goridis, C. (1984) Eur. J. Biochem. 142, 57-64. Malin-Berdel, J., Valet, E., Thiel, E., Forrester, J. A., and Gtirtler, L. (1984) Cytometry 5, 204-209. Earp, H. S., Austin, K. S., Blaisdell, J., Rubin, R. A., Nelson, K. G., Lee, L. W., and Grisham, J. W. (1986) J. Biol. Chem. 261,

4777-4780. 18. 19.

20. REFERENCES

HALBAN

22. 23. 24. 25. 26. 27.

Laemmli, U. (1970) Nature (London) 227, 680-685. Rouiller, D. G., McElduff, A., Hedo, J. A., and Gorden, P. (1985) J. Clin. Invest. 76, 645-649. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Nutl. Acud. Sci. USA 76, 4350-4354. Brackenbury, R., Rutishauser, U., and Edelman, G. M. (1981) Proc. Nutl. Acud. Sci. USA 78, 387-391. Brackenbury, R., Thiery, J. P., Rutishauser, U., and Edelman, G. M. (1977) J. Biol. Chem. 252, 6835-6840. Cunningham, B., Hemperly, J., Murray, B., Prediger, E., Brackenbury, R., and Edelman, G. (1987) Science 236, 799-806. Schlosshauer, B., Schwartz, U., and Rutishauser, U. (1984) Nuture (London) 310, 141-143. Rouiller, D., McKeon, C., Taylor, S., and Gorden, P. (1988) J. Biol. Chem. 263, 13,185-13,190. Sperry, R. W. (1963) Proc. Nutl. Acud. Sci. USA 50, 703-710. Pizzey, J. A., Jones, G. E., and Walsh, F. S. (1988) J. Cell Biol.

107,2307-2317. 28. Takeichi, M., Ozaki, H. S., Tokunaga,

K., and Okada, T. S. (1979) Dev. Biol. 70, 195-205. 29. Maes, E., and Pipeleers, D. (1984) Endocrinology 114, 2205-

2209. 30. Edelman,

G. M. (1984) Proc. N&l. Acud. Sci. USA 81, 14601464. 31. Herzenberg, L., Sweet, R., and Herzenberg, L. (1976) Sci. Amer. 234, 108-117. 32. Hoffman, S., and Edelman, G. M. (1983) Proc. Nutl. Acud. Sci. USA 80, 5762-5766. 33. Fujita, T., Yanatori, Y., and Murakami, T. (1976) In Endocrine, Gut and Pancreas (Fujita, T., Ed.), pp. 347-357. Elsevier, Amsterdam. 34. Langley, 0. K., Aletsee-Ufrecht, M. C., Grant, N. J., and Gratzl, M. (1989) J. Histochem. Cytochem. 37, 781-791.