IN VITRO CELLULAR ~ DEVEIX)PMENTAL BIOLOGY

Volume 21, Number 8, August 1985

9 1985 Tissue Culture Association. Inc.

INSULIN SECRETION FROM PER IFUSED R AT PANCREATIC PSEUDOISLETS

D. W. HOPCROFT, D. R. MASON, ANDR. S. SCOTT'

Department of Medicine. Christchurch ClinicalSchool of Medicine.

Christchurch Hospital. Christchurch, New Zealand

{Received 13 November 1984; accepted 14 February 1985l

SUMMARY
Isolated adult rat pancreatic islets were dispersed into single cells and cultured free-floating for 3 to 4
d, during which time islet cells reaggregated spontaneously into spherical clusters or pseudoislets.
The gross morphology of these tissues resembled nondissociated islets. Electron microscopy revealed
well-preserved cell ultrastructure and intercellular membrane connections. Immunofluorescent
localization of islet cell types showed that A cells tended to be peripherally distributed around a B cell
core, with D cells scattered throughtout the aggregate mass. The dynamics of insulin release from
pseudoislets were evaluated in vitro by perifusion techniques. Pseudoislets exhibited clear biphasic
dose-dependent insulin responses to 30 rain glucose stimulation over the range 5.5 to 30 raM. Repeated
2-rain pulses with 22 mM glucose elicited brief monophasic spikes of insulin release of consistent
magnitude. L-Arginine (5 to 20 mM) evoked biphasic insulin release but these responses were not
dose-dependent. These data indicate that islet cells reaggregate into structures with close morphologic
similarities to intact islets, and that pseudoislet B cells continue to secrete insulin in response to
nutrient secretagogues, comparable to that seen with islets in vitro and in situ.
Key words: pancreas; pseudoislets; reaggregated islet cells; biphasic insulin secretion.

INTRODUCTION MATERIALS AND METHODS

Under certain conditions in vitro, dissociated mammali-
an cells show a propensity to reaggregate and reestablish
muhicellular, histotypic structures (14,18). With respect
to the endocrine pancreas this phenomenon has been
observed after enzymatic dispersion of whole pancreas
{4,5,16,28,29) and isolated pancreatic islets (20,30) where,
in the absence of an adherent substrate, the islet cells
spontaneously and selectively form free-floating aggre-
gates. These clusters have been shown to consist of the
four main islet cell types, viz. A, B, D, and PP cells, and
are comparable in size and appearance to intact islets;
they are therefore frequently referred to as "pseudo-
islets" (28).
Increasing interest has been shown in these tissues as a
potential source of islet tissue for transplantation
15,28,30) and in studies of islet cell interactions (23~.
Although their ability to secrete insulin has been
demonstrated in static culture (4,5,16,20,28,29,30) the
dynamics of insulin release from pseudoislets have not
been investigated. The present work was undertaken to
characterize in vitro pseudoislet B cell responses to
physiologic stimuli and to evaluate further the structural
organization of these tissues.
To whom requests for reprints should be addressed.
PREPARATIONOF PSEUDOISLETS
Pancreatic islets were isolated intact from 200 to 300 g
male Wistar rats using collagenase digestion, Ficoll
separation techniques ~27~. Batches of 200 islets were
hand-picked into 15 )< 85 mm nonadherent bacteriologic
dishes IGIBCO, Grand Island, NY) containing 20 ml
tissue culture medium, and cultured free-floating for 2 to
3 d in a CO2 incubator (5% CO2:95% air) at 37 ~ C. Tissue
culture medium was R P M I 1640 (GIBCOk supplemented
with 25 mM H E P E S buffer, 10% heat-inactivated fetal
bovine serum {GIBCOk 2 m M L-glutamine, 100 U/ml
penicillin, and 0.1 mg/ml streptomycin sulfate.
Between 1000 and 2000 islets were dispersed into single
cell suspensions by the procedure of Lerumark et al. (13).
This involved collagenase digestion, shaking in Ca-free
Krebs-Ringer containing [ethylene his(oxyethylenenitri-
lo)]-tetraacetic acid, and centrifngation through R P M I
1640 containing 4% bovine serum albumin. Dispersed
islet cells were resuspended in fresh tissue culture
medium and an aliquot taken for a cell count, using
trypan blue dye. Cell yield and viability averaged 1100
cells/islet and 93%, respectively. Usually not more than
70% of the yield consisted of single cells, with
approximately 25% as doublets or triplets and the
remainder as clumps of four to eight cells; larger clusters
were seen rarely.

421
422 HOPCROFT ET AL.

The cell suspension was transferred to a 15 X 85 mm
bacteriologic dish at a final density of 2 X l0 s cells/ml of
tissue culture medium and placed in a COz incubator. Cell
aggregation occurred spontaneously over 3 to 4 d. Apart
from movement incurred during daily microscopic
observation, the dishes remained stationary during this
culture period.
PERIFUSION STUDIES
For thin section electron microscopy (EM), samples
were fixed in Karnovsky's fixative and postfixed with 1%
osmium tetroxide in 0.2 M caeodylate buffer. After
staining with uranyl acetate, tissues were dehydrated and
embedded in Spurr's resin. Sections, 50 to 70 nm, were
then stained with lead citrate and examined with a
JEM-100B electron microscope (Jeol, JapanL Intercellu-
lar membrane connections were examined after impreg-
nation of tissues with 1% lanthanum hydroxide (26).

Pseudoislet insulin secretion was studied using a

multichannel perifusion system, modified from Ginger-
ich et al. (10). Batches of 30 to 40 pseudoislets were
perifused at 37 ~ C in minicolumns containing a
supporting matrix of 2.5 mg Cytodex-I microearrier
beads (Pharmacia, Sweden) through which medium was
pumped at 1.0 mL/min. Perifusion medium was Krebs-
Ringer bicarbonate (24 mM) buffered salt solution (KRB)
supplemented with 20 mM H E P E S buffer, 0.1% beat-
inactivated fetal bovine serum, and either 2.8 or 5.5 mN/
D-glucose, as specified, at pH 7.30.
Glucose dose-response studies. Alter a 30-min equili-
bration with KRB -{- 2.8 m M glucose, tissues were
perifused for 30 min with KRB containing 5.5, 11.0, 22, or
30 m M glucose.
Arginine dose-response studies. After a 30-rain equili-
bration with KRB + 5.5 m M glucose, tissues were
perifused for 30 min with this medium supplemented
with 5, 10, 15, or 20 m M L-arginine HC1.
Short pulse stimulation. Tissues were perifused for 30
min with KRB + 2.8 m M glucose before stimulating with
2-min pulses of KRB containing 22 mM glucose at 10-min
intervals over 60 min, with interposing KRB containing
2.8 nL~ glucose. One-minute effluent fractions were
collected at regular intervals and stored at --20 ~
C. After perlfusion pseudoislets were retrieved and their
DNA content measured by fluorometric assay (12), using
the fluorochrome reagent Hoechst 33258 (Sigma Chemical
Co., St. Louis, MO) at a final concentration of 0.1 pg/ml,
and calf thymus D N A (Sigma) as standards. Inter- and
intraassay coefficients of variation were 8.2 and 3.7%
respectively, and the minimum detection limit was 5.6
• 0.9 ng DNA (n -~ 15).

MORPHOLOGY

Pseudoislets cultured for 3 d were fixed for 4 h in

Bouin's solution, dehydrated, and embedded in paraffin.
Sections stained with Hematoxylin and eosin (H&E), were
viewed by light microscopy. Topography of component
islet cell types was examined by indirect immunofluo-
rescence (7). For the first incubation serial sections were
exposed to guinea pig antiporeine insulin serum, rabbit
antiporcine glucagon serum, and rabbit antisynthetic
cyclic somatostatin (SRIF) serum (J. Oliver, Australia),
diluted with phosphate buffered saline containing 0.3% FIG. 1. Photomicrographs showing formation of rat pancrea-
human serum albumin. For the second incubation tic pseudoislets. Culture dishes were viewed with an inverted
microscope at specified times after dispersion of intact islets: {1)
sections were exposed for I h to antirabbit or antiguinea pig
freshly dispersed; {2) 6 h; (3) 18 h; (4~ 24 h; q5} 48 h; (6) and {7D
IgG labeled with fluorescein isothiocyanate (Tago Inc., 72 h. Final magnification of each photograph: ~1) to (3) X 130; (4) and
Burlingame, CA) and examined for fluorescence. {51 X64; (6) X40; (7) XI00.
 1NSUI.IN SECRETION FROM PSEUDOISLETS 423

RADIOIMMUNOASSAYAND DATA PRESENTATION
Intracellular insulin, glucagon, and S R I F contents of
3-d cultured, similar-sized pseudoislets were measured
by radioimmunoassay ~RIA), after extraction of hor-
mones by sonicating and freeze-drying in 0.1 N acetic
acid containing 0.1% human serum albumin and 10 m M
EDTA.
Insulin and glucagon were measured by charcoal-
separation R I A methods 13}. Rat insulin and porcine
glucagon INovo Research Institute, Denmark) were used
as standards. S R I F was measured by a modification of the
R I A procedure of Patel et ai. ~24) using synthetic cyclic
S R I F (Sigma} standards, rabbit anticyclic S R I F serum
(courtesy of Y. Patel~, and polyethylene glycol separa-
tion.
Insulin secretion rates of perifused tissues were
expressed as picomoles of insulin per microgram D N A
content per minute ~12). Statistical significance between
groups was tested using the two-tailed Student's t test on
nonpaired data.
RESULTS
MORPHOLOGY OF PSEUDOISLETS
Thirty-two cultures were prepared containing dis-
persed islet cells ranging in number from 1.11 X 106 to
4,91 X 10~ cells/dish. Cultures were examined at regular
intervals to monitor the progress of cell aggregation ~Fig.
1L Within a few hours of dispersion, single cells began
to cohere as small chains, enlarging into loosely packed
clusters and then into irregularly shaped, larger (50 to 100
~m) aggregates. After 48 h cultures contained few single
cells. Cell clusters continued to merge until by Day 3 of
culture aggregates were compact, spherical structures
within the size range 80 to 350 gm.
H & E stained sections showed well preserved component
ceils with vesicular nuclei and pale, slightly granular and
faintly eosinophilic cytoplasm. Cell borders were clearly
defined, peripheral cells formed a continuous smooth
boundary, and component cells were closely packed (Fig. 2.1~
although occasional larger aggregates were seen to have small
extracellular spaces where cells had failed to cohere. No areas
of focal degeneration were seen. Thin section E M revealed
normal ultrastructural features. A, B, and D cells were
identified by their characteristic secretory granules ~22} but PP
cells were unable to be identified conclusively. B cells con-
rained both immature and mature granules IFig. 2.2K No
capillary endothelial ceils, fibroblasts, leucocytes, ductal cells,
or acinar cells were seen. Spot desmosomes were seen at
frequent intervals along with membranes of adjacent cells.
Many regions of narrowed intercellular spaces were observed
tFig. 2.3~; impregnation of these spaces with lanthanum
revealed regions of partial and total tracer exclusion indicative
21 i of gap junctions and tight j unctions, respectively.
Immunofluorescence staining of pseudoislets con-
firmed the presence of A, B, and D cells. B cells formed

Fit;. 2. Photomicrographs of 3-d cultured rat pancreatic pseudoislets, ill Section stained with H&E. viewed by light
microscopy. X400. 12~ Thin ~ction EM of pseudoislet B cells. X6(d)0. The bar represents 1.0/Jm. 131 Thin section EM,
showing membrane features of two apposing B cells. X 78 000. The bar represents 0.1 ~m.
424 tIOPCROFT ET AL.
FIG. 3. Immunofluorescence staining of serial sections of 3-d cultured rat pancreatic pseudoislets. Final
magnification X140. A. Pseudoislets stained with fluoroscein-labeled antiglucagon serum. B. Pseudoislets stained with
fluoroscein-labeledanti-SRIF serum.
the central mass whereas the great majority of glucagon-
containing A cells were located near or at the periphery of
each section (Table I, Fig. 3.1). Small numbers of D cells
were scattered throughout each aggregate (Table I, Fig.
3.2). The presence of A, B, and D cells was also shown by
hormone content measurements. Values (mean +
SEM) for insulin, glucagon and SRIF contents were 38
+ 4, 0.07 + 0.0], and 0.03 + 0.003 ng/psen-
doislet {n = I0), respectively.
FUNCTIONAL STUDIES
Basal insulin secretion rates stabilized within 15 min of
the start of perifusion, at 0.1 + 0.02 pmol//ag DNA
per rain (n ---- 19) in the presence of 2.8 mM glucose, and
0.22 + 0.03 pmol/~g DNA per min (n = 18} in the
presence of 5.5 m M glucose {P <0.005). Peak first-phase
and second-phase insulin responses of per[fused pseu-
doislets to glucose and arginine stimuli are
listed in Table 2. Pseudoislets exhibited dose-dependent
biphasic insulin responses to glucose stimulation (Fig. 4).
Insulin release was characterized by a rapid spike
reaching peak values within I to 2 min of the onset of
stimulation, declining to a nadir after 8 to 9 min, and then
rising more gradually to a second phase. Pseudoislet insulin
responses to arginine stimulation were biphasic but were not
dose-related {Fig. 5). Peak flrst-phase responses were rapid,
occurring within the Ist min of stimulation, then declining to a
nadir 4 to 5 rain after the onset of stimulation. Peak second-
phase release was more variable in location. After cessation of
glucose and arginine stimulation, insulin release declined
rapidly to prestimulus levels (not shown). Pseudoislets
responded with brisk monophasic spikes of insulin release to
repeated 2-min pulses of 22 m M glucose {Fig. 6), comparable
in magnitude to first-phase responses seen with prolonged
stimulation with this level of glucose.
DISCUSSION
Reaggregated islet cell preparations have aroused
interest in two areas. The first is as a source of
insulin-secreting tissue for transplantation. Standardized
techniques {27) for rodent islet isolation are much less
TABLE 1
LOCATION OF A AND D-CELLS IN PSEUDOISLETS ~
A Cells I) Cells
Percentage of ceils on periphery 67 • 1.4 31 + 5
Percentage of cells among outermost 86 + 1.3 ,56 + 4
three layers of cellsb
~ count was made of the total number of A cells and D cells in 73
pseudoislets from two tissue preparations stained with antiglucagon
and anti-SRIF sera. Results are mean + SEM percent.
bDepending on the size and shape of the pseudoislet, this area
corresponds to approximately 40 to 60% of the total cross-sectional
area.
 INSULIN SECRETION FROM PSEUDOISLETS 425

TABLE 2

PEAK FIRST AND SECOND-PHASE INSULIN RELEASE FROM PERIFUSED PSEUDOISLETS ~

D - G lucose 4m M I L- A rginine qm M

5.5 I1 22 30 5 10 15 20

Peak 1st-
phase
insulin
release 0.86 -- 0.19 1.08 -4- 0.16 1.36-4-0.17 1.88-4" 0.25 2.09 -4-0.22 1.96-4"0.13 2.38-4"0.92 2.61----_0.67
Peak 2nd-
phase
insulin
release 0.33 -4-0.06 0.51 -4-0.07 0.62-4-0.11 0.92 -4-0.18 0.82-f-0.14 0.68--4-0.06 0.70+-0.10 1.20-4"0.22
No of
experiments 4 4 6 5 5 5 3 6

~ are given as mean -4- SEM picomoles insulin per microgram DNA per minute.

efficacious when applied to pancreata of higher mammals
(particularly humanh with correspondingly lower islet
yields (2,19,28~. However, after enzymatic dispersion of
whole porcine, canine, and rodent pancreas it has been
observed that the endocrine components rapidly and
selectively reassembled into islet-like structures, raising
the possibility that such procedures may provide a means
of generating larger quantities of islet B cells (5,28h It has
also been suggested that these tissues, by virtue of having
fewer contaminating nonendocrine ceils, may be less
antigenic and therefore a more suitable islet tissue
preparation for allogeneic transplantation (5,30}. The
second area of interest in reaggregated islet tissues
reflects their usefulness in the study of structure-
function relationships between islet cells, comparing
insulin secretion from islet cells with and without
membrane connections present ~6,11,22,23,25 ~.
Despite the potential applications of these tissues, their
insulin secretory characteristics have received little
attention. Several studies ~4,5,16,20,28,29,30) have meas-
ured insulin release from pseudoislets in long-term static
cultures and acute insulin responses to stimulatory
substances added to culture media. Chertow et al. (6)
perifused rat islet cell clumps formed by exposure of
dispersed cells to 13-cis retinoic acid, and showed
monophasic insulin responses to glucose provocation.
However, it has not been determined whether B cells of
spontaneously formed islet cell aggregates are able to
respond to physiologic stimuli in a manner resembling
their parent tissue.
Considering these aspects, we found that pseudoislets not
only bore close structural similarities to nondissociated islets
but behaved in a qualitatively similar secretory fashion, viz
low, stable basal insulin release and marked biphasic secretory
profiles, characteristic of B cell secretion of other islet
preparations in vitro (9,10,25~. In contrast to glucose
stimulation, pseudoislet insulin responses to L-arginine were
not dose-related, raising the possibility that their B cells had
an increased sensitivity to this amino acid.
A pilot study was undertaken to determine whether
pseudoislets were capable of sustained insulin secretion
in vivo. Pseudoislets were transplanted (8) into the
splenic pulp of seven streptozotocin-diabetic allogeneic
rats resulting, in each case, in significant reductions in
nonfasting plasma glucose levels. Seven to sixteen days
after transplantation recipients returned to pretransplant
glucose levels, which was presumed to be a consequence
of tissue rejection because transplanted pseudoislets have
been shown (23) to normalize blood glucose levels of
syngeneic diabetic rats for as long as the recipients were
followed ~120 d). This preliminary data suggests that
pseudoislets retain antigenicity despite prolonged incu-
bation and apparent loss of nonendocrine cells.
The well-preserved morphology of pseudoislet cul-
tures in this study was consistent with other reports of
aggregate morphology ~5,20,30~. No areas of focal
degeneration were seen in these pseudoislets, irrespective
of their size, in contrast to our observations (12~ and other
reports (1,15,30) of necrotic central regions of larger
nondissociated islets in culture. Likewise, Ono et al. (20~
found no evidence of necrosis in rat islet cell aggregates
9 - 2 /2.8mMJ Glucose Stimulus J
a~
~Z
t~ o~ I
.t: D
D
ta~
_c-6
E
CI.O
~tt I I I I
0 20 40 60
Perifusion Time
Fro. 4. Insulin release from perifused pseudoislets. Tissues
were equilibrated for 30 min with KRB containing 2.8 mM
glucose before stimulation over 30 min with 5.5, 11, 22, or 30 mM
glucose. The figure shows mean responses and standard error
bars for each glucose concentration. The n value for treatments
A to D were 4, 4, 6, and 5, respectively.
426 HOPCROFT ET AL.

[ L-ARGININE STIMULUS i tended to be localized near or at the periphery of the

C~/ 5:.5 rnM GLUCOSE pseudoislets, resembling the distribution of A cells of
15 mM (3).~,- L intact islets (21,23L D cells were located throughout the
i ~ ' - - - 2 0 rnM (6)
.E aggregates and, unlike those of intact islets (22,23),
20 showed no preferential association with the A cells.
<
Z
tn These data reinforce other reports in suggesting that
free-floating dispersed rat islet cells undergo at least
partial reorganization of A, B, and D cells to form
-6
three-dimensional microorgans resembling intact islets,
and show that the B cells of these tissues exhibit
functional similarities to those of intact islets. This
suggests that they have retained their ability to
z
_J discriminate between and respond appropriately to
nutrient stimuli, supporting the suitability of these
0 tissues for further use in transplantation procedures and
~, t , i i i studies of islet cell interactions. Because they are
0 20 30 40 50 60 relatively large and free-floating in culture, pseudoislets
PERIFUSIONTIME (ram) are well-suited for manipulative procedures essential to
FIG. 5. Insulin release from pez~fused pseudoislets. Tissues such work.
were equilibrated for 30 rain with KRB containing 5.5 mM glucose
before stimulation over 30 rain with this medium supplemented with 5,
10, 15, or 20 m]14arginine. The figure shows mean responses for each REFERENCES
arginine concentration, with n values in brackets. 1. Andersson, A. Tissue culture of isolated pancrea-
tic islets. Acta Endocrinol. 205: 283-293; 1976.
2. Andersson, A.; Borg, H.; Groth, C-V.; Gunnarseon, R.;
cultured in petri dishes for up to 80 d and Heilerstrom, C.; Lundgren, G.; Westman, J.; Ostman, J.
attributed this in part to wider intercellular spaces of Survival of isolated human islets of Langerhans maintained
in tissue culture. J. Clin. Invest. 57: 1295-1301; 1976.
these tissues. The less compact nature of these avascular
3. Arcus, A. C.; Beaven, D. W.; Hart, D. S.; Holland, G. W.
microorgans may well facilitate nutrient and waste Long-term hormonal secretion from the autotransplanted sheep
transport to and from centrally located cells and thereby pancreas. Diabetologla 16: 325-330; 1979.
extend the viable lifetime of these tissues in culture. 4. Britt, L. D.; Scharp, D. W. Formation of pseudo-islets from
The identification of membrane regions indicative of neonatal pig pancreas. Surg. Forum 31: 354--356;1980.
5. Britt, L. D.; Stojeba, P. C.; Scharp, C. R.; Greider, M. H.:
gap and tight junctions in pseudoislets confirms other Scharp, D. W. Neonatal pig pseudo-islets: A product of
reports (20,30}. Clearly, intercellular junctions can be selective aggregation. Diabetes 30: $80-583; 1981.
reestablished rapidly after proteolytic dissociation of 6. Chertow, B. S.; Baranetsky, N. G.; Sivitz, W. I.; Meda, P.;
islets {6,11), but whether these structures are functionally Webb, M. D. Shih, J. C. Cellular mechanisms of insulin
release. Effects of retinoids on rat islet cell-to-cell adhesion,
competent and what their roles are with respect to insulin
reaggregation and insulin release. Diabetes 32: 568-574;
secretion remains unknown (23). Pseudoislet topography 1983.
reported in the literature has been quite varied; a 7. Coons, A. H.; Leduc, E. H.; Connolly, J. M. Studies on
preferential distribution of both A and D cells was found antibody production. I. A method for the histochemical
at the periphery of neonatal rat islet cells reaggregated in demonstration of specific antibody and its application to a study
of the hyperimmune rabbit. J. Exp. Med. 102: 49-63; 1955.
a three-dimensional collagen matrix (17}, whereas others 8. Finch, D. R. A.; Wise, P. H.; Morris, P. J. Successful
have found quite marked variations of topography intra-splenic transplantation of syngeneic and allogeneic
(5,16,28,30). In this study, glucagon-secreting A cells isolated pancreatic islets. Diabetologia 13: 195-
199; 1977.
9. Gerich, J. E.; Charles, M. A.; Grodsky, G. M. Regulation of
al 9 9 9 9 9 pancreatic insulin and ghcagon secretion. Ann. Rev. Physiol.
15 38: 353-388; 1976.
c_
10. Gingerich, R. L.; Aronoff, S. L.; Kipnis, D. M.; Lacy, P. E.
E
< Insulin and glucagon secretion from rat islets maintained in a
Z 1.0 tissue culture-perifusion system. Diabetes 28: 276-281; 1979.
11. Halban, P. A.; Wollheim, C. B.; Blondel, B.; Meda, P.;
Niesor, E. N.; Mintz, D. H. The possible importance of
| 0.5
contact between pancreatic islet cells for the control of insulin
0 release. Endocrinology 111: 86-94; 1982.
E 12. Hoperoft, D. W.; Mason, D. R.; Scott, R. S. Standardization
Q.
of insulin secretion from pancreatic islets: Validation of a DNA
Ij~l I I I I assay. Horm. Metab. Res.: In press.
0 20 40 60 80 13. Lernmark, A.; Kanatsuna, T.; Patzelt, C.; Diakoumis, K.;
PERIFUSION TIME (min) Carroll, R.; Rubenstein, A. H.; Steiner, D. F. Antibodies
FIG. 6. Insulin release from perlfused pseudoislets. Tissues directed against the pancreatic islet cell plasma membrane.
were equilibrated for 30 9 with KRB containing 2.8 mM Diabetologia 19: 445-451; 1980.
glucose followed by repeated 2-min pulses with 22 mM glucose. 14. Lissitzky, S.; Fayet, G.; Geraud, A.; Verrier, B.; Torresani, J.
The figure shows the mean responses and standard error bars of Thyrotropin-induced aggregation and re-organization into folli-
four experiments. cles of isolated porcine-thyroM cells. I. Mechanism o! action of
 INSULIN SECRETION FROM PSEUDOISLETS 427

thyrotropin and metabolic properties. Eur. J. Biochem. 24:
88-99; 1971.
15. Logothetopoulos, J.; Valiquette, N.; Cvet, D. Glucose
stimulation of beta-cell DNA replication in the intact rat and in
pancreatic islets in suspension culture. Diabetes 32: 1172-1176;
1983.
16. Merrell, R. C.; Scharp, D. W.; Gingerieh, R.; Feldmeier, M.;
Greider, M.; Downing, R. New approaches to separation of
islet cells. Surg. Forum 3: 303-305; 1979.
17. Montesano, R.; Mouron, P.; Amherdt, M.; Orci, L. Collagen
matrix promotes reorganization of pancreatic endocrine cell
monolayers into islet-like organoids. J. Cell Biol. 97: 935-939;
1983.
18, Moscona, A. A. Recombination of dissociated cells and the
development of cell aggregates. Willmer, E. N., ed. Ceils and
tissues in culture, vol. 1. New York: Academic Press; 1965:
489-529.
19. Najarian, J. S.; Sutherland, D. E. R.; Steffes, M. W. Isolation
of human islets of Langerhans for transplantation. Transplant.
Proc. 8: 611-613; 1975.
20. Ono, J.; Takaki, R.; Okano, H.; Fukuma, M. Long-term
culture of pancreatic islet cells with special reference to the
B-cell function. In Vitro 15: 95-102; 1979.
21. Orei, L.; Malaisse-Lagae, F.; Ravazzola, M.; Rouiller, D.;
Renold, A. E.; Perrelet, A.; Unger, R. A morphological basis
for the intercellular communication between A and B-cells in
the endocrine pancreas. J. Clin. Invest. 56: 1066-1070; 1975.
22. Orci, L. Morpho-funetional aspects of the islets of
Langerhans. The microanatomy of the islets of Langerhans.
Metabolism 25: 1303-1313; 1976.
23. Orci, L. Macro- and micro-domains in the endocrine
pancreas. Diabetes 31: 538-565; 1982.
24. Patel, Y. C.; Amherdt, M.; Orci, L. Somatostatin secretion
from monolayer cultures of neonatal rat pancreas. Endocrinolo-
gy 104: 676-679; 1979.
25. Pipeleers, D.; Veld, P.; Maes, E.; Van de Winkel, M.
Glucose-induced insulin release depends on functional co-
operation between islet cells. Proc. Natl. Acad. Sci. USA 79:
7322-7325; 1982.
26. Revel, J. P.; Karnovsky, M. J. Hexagonal array of subunits in
interceUularjunctions of the mouse heart and liver. J. Cell Biol.
33: C7-C12; 1967.
27. Scharp, D. W.; Kemp, C. B.; Knight, M. J.; Ballinger, W. F.;
Lacy, P. E. The use of Ficoll in the preparation of viable islets
o1 Langerhans from the rat pancreas. Transplantation 16:
686-689; 1973.
28. Scharp, D. W.; Downing, R.; Merrell, R. C.; Greider, M.
Isolating the elusive islet. Diabetes 29: 19-30; 1980.
29. Schroder, D.; Wegner, U.; Besch, W.; Zuhlke, H. Characteri-
zation of pseudo-islets formed from pancreatic islet cell
suspensions of neonatal rats. Mol. Cell. Endocrinol. 32:
179-193; 1983.
30. Tze, W. J.; Tai, J. Preparation of pseudoislets for
morphological and functional studies. Transplantation 34:
228-231; 1982.

We thank Ms. Christine Kempkers and Ms. Fay McCone for skilled technical assistance, and Ms.
Helen Morrin and Mr. Tony Day for the electron microscopy. Mrs. Dorothy Neal is thanked for
preparation of the manuscript. This work was supported by grants from the Medical Research Council
of New Zealand. D. W. H. was the recipient of a Novo Diabetes Research Scholarship.