Content uploaded by Eva Hammar
Author content
All content in this area was uploaded by Eva Hammar on Apr 23, 2015
Content may be subject to copyright.
Extracellular Matrix Protects Pancreatic -Cells
Against Apoptosis
Role of Short- and Long-Term Signaling Pathways
Eva Hammar,
1
Ge´raldine Parnaud,
1
Domenico Bosco,
2
Nadja Perriraz,
1
Kathrin Maedler,
3
Marc Donath,
3
Dominique G. Rouiller,
1
and Philippe A. Halban
1
We have shown previously that culture of -cells on
matrix derived from 804G cells and rich in laminin-5
improves their function. The purpose of this study was
to investigate whether this matrix protects -cells
against apoptosis and to elucidate signaling pathways
involved. Matrix protected sorted rat -cells against
apoptosis under standard conditions (11.2 mmol/l glu-
cose, 10% serum), after serum deprivation (1% serum),
and in response to interleukin-1 (IL-1; 2 ng/ml),
compared with control (poly-
L-lysine [pLL]). Caspase-8
activity was reduced in cells cultured on matrix, where-
as focal adhesion kinase (FAK), protein kinase B (PKB,
or Akt), and extracellular signal–regulated kinase
(ERK) phosphorylation was augmented. Treatment (4
h) with an anti-1 integrin antibody, with the ERK
pathway inhibitor PD98059, and/or with the phosphati-
dylinositol 3-kinase inhibitor LY294002 augmented cell
death on 804G matrix but not on pLL. In long-term
assays (48 h), PD98059 but not LY294002 drastically
augmented cell death on 804G matrix but did so to a
lesser extent on pLL. The protein inhibitor of nuclear
factor-B(IB␣) was overexpressed in cells cultured
18 h on matrix with partial blockade by PD98059. In
summary, this study provides evidence for activation of
signaling pathways and gene expression by extracellu-
lar matrix leading to improved -cell survival. Diabetes
53:2034 –2041, 2004
E
pithelial cells deprived of matrix attachment
undergo programmed cell death, a form of apo-
ptosis termed anoikis (1). The role of cell an-
chorage in cell survival has been demonstrated
in several cell types (reviewed in 2), including pancreatic
islet cells (3). Disengagement from extracellular matrix
(ECM) is known to prejudice islet cell survival, notably by
inducing apoptosis and necrosis (4). It has been reported
that reestablishment of appropriate cell-matrix contacts
reduces cell death (3,5); however, the molecular basis for
such pro-survival signaling remains largely unknown. Re-
cent evidence implicates integrins in mediating pro-sur-
vival signals emanating from the ECM (reviewed in 6).
Several signal transduction components activated by
integrins, including focal adhesion kinase (FAK) (7,8),
phosphatidylinositol (PI) 3-kinase (9,10), and the mito-
gen-activated protein (MAP) kinase/extracellular signal–
regulated kinase (ERK) (reviewed in 11) have been
implicated in mechanisms underlying anoikis (reviewed in
12). Recently, it has emerged that PI 3-kinase and its
downstream effector protein kinase B (PKB, or Akt) play
key roles in the regulation of pancreatic -cell survival
(13–17). The role of the MAP kinase ERK cascade in -cell
survival is controversial. Some studies have suggested an
antiapoptotic effect of this cascade (18), whereas others
claim that it serves as a mediator of cytokine-induced
apoptosis (19 –21), this claim in itself being controversial
(22). Nevertheless, a possible involvement and impact of
PKB/Akt and ERK in outside-in signaling from ECM to
-cells remains to be investigated.
ECM produced by the rat bladder carcinoma cell line
804G is rich in laminin-5 (epiligrin) and also contains
fibronectin (23–25). This matrix is able to induce attach-
ment and spreading of many epithelial cell types, including
pancreatic -cells (24,26). Furthermore, our group has
shown that this matrix improves -cell function (27). In
the present study, we examined whether the 804G matrix
could rescue primary pancreatic -cells from apoptosis,
and we investigated the intracellular pathways involved.
RESEARCH DESIGN AND METHODS
Reagents and antibodies. The following materials were used: PD98059 and
LY294002 (Calbiochem, Darmstadt, Germany); In Situ Cell Death Detection
Kit (POD; Roche Molecular Biochemicals, Rotkreuz, Switzerland) for termi-
nal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labeling
From the
1
Department of Genetic Medicine and Development, University
Medical Center, University Hospital, Geneva, Switzerland; the
2
Cell Isolation
and Transplantation Center, University Hospital, Geneva, Switzerland; and the
3
Division of Endocrinology and Diabetes, University Hospital, Zurich, Swit-
zerland.
Address correspondence and reprint requests to Eva Hammar, Department
of Genetic Medecine and Development, University Medical Center, 9th Floor,
Rue Michel Servet 1, 1211 Geneva 4, Switzerland. E-mail: eva.hammar@
medecine.unige.ch.
Received for publication 21 January 2004 and accepted in revised form 12
May 2004.
ECM, extracellular matrix; ELISA, enzyme-linked immunosorbent assay;
ERK, extracellular signal–regulated kinase; FAK, focal adhesion kinase;
GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HRP, horseradish per-
oxidase; IB␣, inhibitor of nuclear factor B; IL-1, interleukin-1; MAP,
mitogen-activated protein; MEK1, MAP kinase/ERK kinase; NF-B, nuclear
factor-B; PI, phosphatidylinositol; PKB, protein kinase B; pLL, poly-
L-lysine;
TUNEL, terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-
end labeling.
© 2004 by the American Diabetes Association.
2034 DIABETES, VOL. 53, AUGUST 2004
(TUNEL) assay with substrate 3,3⬘-diaminobenzidine (DAB) from Dako
(Carpinteria, CA); cell death detection ELISA
PLUS
kit (Roche); CaspaTag
caspase-8 (LETD) activity kit (Intergen, Oxford, U.K.); Hoechst 33342 (Sigma
Fluka, Buchs, Switzerland); propidium iodide (Intergen, Purchase, NY);
recombinant rat interleukin-1 (IL-1; RnD Systems, Abingdon, U.K.) hamster
anti-rat CD29 (integrin 1 chain) and hamster IgM control antibodies (Becton
Dickinson Biosciences, San Jose, CA); polyclonal anti-phospho-FAK (Tyr-397)
and anti-FAK (Biosource International, Camarillo, CA); polyclonal anti-phos-
pho-ERK1/2 (Thr-202/Tyr-204), anti-ERK1/2, anti–phospho-PKB/Akt (Ser-473)
and anti-PKB/Akt (Cell Signaling Technology-Bioconcept, Allschwil, Switzer-
land); monoclonal anti-actin (Chemicon International, Temecula, CA); C-21
anti–inhibitor of nuclear factor (NF)-B␣ (IB␣; Santa Cruz Biotechnology,
Santa Cruz, CA); anti-mouse horseradish peroxidase (HRP) and anti-rabbit
HRP (Amersham Pharmacia Biotech, Du¨ bendorf, Switzerland).
Islet isolation and -cell purification. All experiments were performed on
primary pancreatic -cells sorted from adult rat islet cells by autofluores-
cence-activated flow cytometry. Islets of Langerhans were isolated by colla-
genase digestion of pancreas from male Wistar rats (weighing 150 –200 g),
followed by Ficoll purification (28). Islets were digested with trypsin and
sorted by flow cytometry to obtain -cells as described (28). This purification
procedure yields a population consisting of ⬎95% -cells (27).
Cell culture. Sorted -cells were washed and incubated overnight in suspen-
sion for recovery as previously described (27). Cells were then resuspended at
4 ⫻ 10
5
cells/ml, and aliquots of 50 l were plated as droplets on noncoated
plastic dishes or on plastic dishes coated with poly-
L-lysine (pLL), with 804G
matrix, or with laminin-5 purified from 804G (provided by N. Koshikawal and
V. Quaranta, San Diego, CA). To assess the effect of the antibody directed
against 1 integrin subunit, cells were pretreated for1hinsuspension with 0.5
g/ml hamster IgM (control) or with 0.5 g/ml anti–1 integrin antibody
(Ha2/5) before plating them on pLL- or 804G-coated dishes. When indicated,
cells were cultured for 48 h in culture medium containing 1% FCS (serum
deprivation). To study the effect of IL-1, cells were cultured for 48 h in
culture medium containing 10% FCS and then treated for 4 h with 2 ng/ml
IL-1. To assess the effects of inhibitors, cells were pretreated for 15 min with
50 mol/l PD98059 and/or 50 mol/l LY294002, before plating them on pLL- or
804G-treated dishes. Cells were then cultured in standard medium (10% FCS)
for4horfor48h(fresh inhibitor added at 24 h). DMSO was added to control
cells at the same final concentration as that used for the inhibitors. For the
protein phosphorylation analysis by Western blot, cells were serum starved
for 1–4 h in suspension in serum-free medium containing 0.1% BSA. They were
then plated on pLL- or 804G-coated dishes for 30 min in the same medium
before performing the protein extraction. For gene expression analysis by
low-density gene arrays and by quantitative real-time PCR analysis, cells were
serum starved (1% FCS) in suspension for 2 h before plating them on pLL- or
804G-coated dishes (or on noncoated dishes as control for the gene arrays).
Cells were then cultured for 18 h in medium containing 1% FCS before mRNA
extraction. For the IB␣ protein analysis by Western blot, cells were treated
as for gene expression analysis and cultured for 24 or 48 h in medium
containing 1% FCS before protein extraction. Fresh medium (2 ml) was added
every 24 h.
804G matrix preparation. 804G cells (the kind gift of Desmos, San Diego,
CA) were grown in Dulbecco’s modified Eagle’s medium containing 10% FCS
and 5.6 mmol/l glucose. Conditioned medium (referred to hereafter as 804G
matrix) was prepared as previously described (27).
Coating of plastic dishes with pLL, 804G matrix, and laminin-5. Aliquots
(60 l) of pLL (0.1 mg/ml), crude 804G matrix, or laminin-5 purified from 804G
matrix (4 g/ml in PBS) were layered at the center of 35-mm culture petri
dishes (adherent dishes for mammalian cell culture). Dishes were kept in a
damp box at 37°C for 18–20 h before being rinsed three times with sterile H
2
O
and air dried. Uncoated Petri dishes (plastic) or dishes coated with 0.1 mg/ml
pLL were used as controls.
Analysis of cell death. To quantify cell death by TUNEL, attached cells were
washed with PBS and fixed with 4% paraformaldehyde (20 min at room
temperature). After permeabilization with 0.5% Triton X-100 (4 min at room
temperature), the TUNEL assay (detecting the free 3-OH strand breaks
resulting from DNA degradation) was performed with the In Situ Cell Death
Detection Kit, according to the manufacturer’s instructions. The preparations
were then rinsed with PBS and incubated (10 min at room temperature) with
the substrate DAB, and the quantification of dead cells was performed using
a light microscope. Unless stated otherwise, results are means ⫾ SD of three
independent experiments, with a minimum of 500 -cells examined for each
condition in each experiment. The type of cell death detected by TUNEL was
assessed as described by others (29). Cells were stained for 20 min at 37°C
with 1 g/ml Hoechst 33342 (to determine the morphology of the nuclei) and
with 1.25 g/ml propidium iodide (which stains the necrotic nuclei) before
fixation and staining with TUNEL reagent. Cells lacking propidium iodide
uptake and with condensed nuclei were counted as apoptotic. Late apoptosis
(or secondary necrosis) was defined by the uptake of propidium iodide and
the presence of nuclear condensation, and necrosis was defined by the uptake
of propidium iodide and normal nuclear morphology. Only TUNEL-positive
cells were considered for this assessment. The quantification by enzyme-
linked immunosorbent assay (ELISA) of mono- and oligonucleosomes present
in the cytoplasm of apoptotic cells was performed using the cell death
detection ELISA
PLUS
kit according to the manufacturer’s instructions (Roche).
Quantification of caspase-8 activity. To quantify caspase-8 activity, at-
tached cells were treated for 1 h with the fluorescein-labeled peptide
FAM-peptide-FMK (which irreversibly binds to active caspase-8) as indicated
by the manufacturer (Intergen). At 5 min before the end of the incubation, 25
mol/l of Hoechst 33342 (final concentration) was added to visualize the
nuclei. At the end of the incubation, cells were washed and fixed with 4%
paraformaldehyde (20 min at room temperature). The quantification of the
number of cells containing active caspase-8 as a percent of total was
performed using a fluorescence microscope (Axiocam).
Western blot analysis. To analyze FAK, PKB/Akt, and ERK protein phos-
phorylation, attached cells were washed with ice-cold PBS without Ca
⫹2
/Mg
⫹2
supplemented with 1 mmol/l sodium vanadate, and they were lysed in sample
buffer 1⫻ (62 mmol/l Tris-Cl, pH6.8, 2% SDS, 5% glycerol, and 1% 2-mercap-
toethanol). Protein concentrations were determined with the amido black
method (30), and equal amounts of total protein were loaded for SDS-PAGE.
To analyze IB␣ protein expression, cells were washed with ice-cold PBS and
lysed in a buffer containing 150 mmol/l NaCl, 50 mmol/l Tris-HCl, 1% Nonidet
P40, 0.25% deoxycholate, 0.1% SDS, 1 mmol/l ditiothreitol, and 1 mmol/l
phenylmethylsulfonyl fluoride. After determination of protein concentrations
by the method of Lowry, sample buffer 3⫻ was added to the proteins, and the
samples were analyzed on an 8% SDS-PAGE gel. All samples, after separation
on an SDS-PAGE gel, were electroblotted onto nitrocellulose membranes
(Schleicher & Schuell, Dassel, Germany) for immunoblotting with the appro-
priate antibody. An enhanced chemiluminescence protein detection kit (Am-
ersham Biosciences) and a Kodak image station were used for visualization of
the bands.
Gene expression analysis with microarrays. Low-density membrane
cDNA microarrays (Mouse Cancer Pathway Finder GEArray; SuperArray,
Bethesda, MD) were used to compare gene expression between cells cultured
on plastic and those cultured on 804G matrix. After cell culture, RNA from
attached cells was isolated using the QIAshredder and RNeasy Mini Kit
(Qiagen, Basel, Switzerland) according to the manufacturer’s instructions.
The RNA quality was verified performing agarose gel electrophoresis, and 1.5
g of the total RNA (for each condition) was used as the template for the
32
P-labeled cDNA probe synthesis performed according to the manufacturer’s
instructions (SuperArray). The membranes were prehybridized and hybrid-
ized with the probe, as indicated by the manufacturer, before exposure to
X-ray film for various lengths of time at ⫺80°C.
Quantitative real-time PCR. The results for select genes obtained with the
microarrays were verified by real-time PCR. cDNA was synthesized with
Superscript II (Invitrogen, Basel, Switzerland) using 1 g of total RNA in a
20-l reaction volume. For real-time PCR, the cDNA was amplified using a
GeneAmp 5700 sequence detection system (PE Applied Biosystems, Foster
City, CA). For this purpose, primers were designed according to Primer
Express software (PE Applied Biosystems). The following oligonucleotide
primer sets were used: IB␣ forward: 5⬘ TGCTGAGGCACTTCTGAAAGC 3⬘;
IB␣ reverse: 5⬘ TCCTCGAAA GTCTCGGAGGTC 3⬘. The dsDNA-specific dye
SYBR Green I was incorporated into the PCR buffer (PE Applied Biosystems)
to allow for quantitative detection of the PCR product. The results were
analyzed using ABI Prism 7000 SDS software (PE Applied Biosystems). The
house-keeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
was used as internal control.
RESULTS
Effect of 804G matrix on -cell apoptosis induced by
serum deprivation or IL-1. To determine whether the
804G matrix is able to serve as a survival ligand for
primary rat pancreatic -cells, TUNEL and ELISA were
performed on cells cultured on 804G matrix or on pLL (a
nonspecific adhesive polymer)-coated Petri dishes. -Cells
were cultured for 48 h on pLL-coated (control) or 804G
matrix-coated dishes (hereafter called pLL or 804G, re-
spectively) in medium containing either 10% (control
conditions) or 1% FCS (low-serum conditions). The num-
ber of apoptotic cells as a percent of the total was
E. HAMMAR AND ASSOCIATES
DIABETES, VOL. 53, AUGUST 2004 2035
determined under each condition by TUNEL (Fig. 1A).
Approximately 4% of -cells cultured on pLL were apopto-
tic under control conditions. Similar results were obtained
when noncoated plastic tissue culture dishes were used
instead of pLL-coated dishes (not shown). Culture under
low-serum conditions significantly increased the number
of TUNEL-positive cells on pLL (9%). When cells were
cultured on 804G, ⬍1% of cells were TUNEL positive
under control conditions, and serum deprivation did not
significantly affect this number.
We next investigated whether the 804G matrix protects
-cells from apoptosis induced by IL-1, a proinflamma-
tory cytokine that has been shown to be involved in
pancreatic -cell apoptosis leading to the development of
type 1 and type 2 diabetes (31,32). For this purpose, the
cells were cultured for 48 h on pLL or 804G before being
treated for 4 h with 2 ng/ml IL-1. The apoptosis of -cells
under each condition was measured by ELISA. There was
a slight but highly reproducible increase of apoptosis in
cells cultured on pLL treated with IL-1, as compared with
nontreated cells (P ⬍ 0.003). When cells were cultured on
804G, the apoptotic index was significantly decreased as
compared with cells cultured on pLL, and there was no
difference between nontreated and IL-1–treated cells
(Fig. 1B), indicating that this matrix also protects against
IL-1–induced cell death. These results have been con-
firmed by TUNEL (not shown).
Differential staining of nuclei of TUNEL-positive cells
with propidium iodide and Hoechst 33342 distinguished
between apoptotic and necrotic cells on pLL (Fig. 1C).
Under basal conditions (10% FCS), the majority of TUNEL-
positive cells were apoptotic or late apoptotic (secondary
necrotic). Serum deprivation induced apoptosis and, to a
lesser extent, necrosis, whereas IL-1 induced only apo-
ptosis (Fig. 1C).
To begin to elucidate the pathway leading to apoptosis
on pLL, the activity of caspase-8 in cells cultured for 4 or
48 h on pLL or on 804G was analyzed (Table 1). Caspase-8
activity was higher in cells on pLL compared with those on
804G, and this difference increased with time. These
results suggest that the 804G matrix might block caspase-8
activity in turn, explaining the pro-survival effect of the
804G matrix.
Effect of purified laminin-5 on rat islet -cell survival
and spreading. The 804G matrix is known to be rich in
laminin-5. However, it also contains fibronectin and may
also contain other signaling molecules. To investigate
whether laminin-5 is the extracellular molecule responsi-
ble for the effect of the 804G matrix, we compared the
effects on both cell survival and cell spreading of purified
laminin-5 and 804G matrix after 48 h of culture under
standard conditions (Figs. 2A and B). As shown before,
there was a significant difference in the number of TUNEL-
positive cells between pLL (11%) and 804G (0.5%). The
number of TUNEL-positive cells on pure laminin-5 was
similar to 804G matrix (Fig. 2A). It has been reported that
-cells rapidly flatten and spread on 804G matrix as on
pure laminin-5, indicating a strong adhesion of -cells on
these two matrices (27). Here we show that the morphol-
ogy of -cells after 48 h of culture on 804G is highly similar
to that of cells cultured on pure laminin-5 (Fig. 2B). By
contrast, rat -cells do not adhere well and/or do not
spread on vitronectin and fibronectin (unpublished re-
sults). These results suggest that laminin-5 is the major
component of 804G matrix responsible for its pro-survival
effect. Therefore, we decided to pursue our studies with
the 804G matrix, given its ready availability and conve-
nience of use.
The 1 integrin subunit is involved in the effects of
804G matrix on -cell survival and spreading. We have
previously reported that the laminin-5 receptors integrins
␣31 and ␣61 are expressed on rat -cells (27,33), and
that the latter is involved in the spreading of cells on the
804G matrix (27). To assess whether the 1 integrin sub-
unit is involved in the pro-survival effect of the matrix, we
TABLE 1
Active caspase-8 –positive cells
Time pLL 804G
4 h 2.6 ⫾ 0.9 0.8 ⫾ 0.4*
48 h 4.1 ⫾ 0.5 0.5 ⫾ 0.1†
Data are means ⫾ SD of at least three independent experiments. Rat
-cells were plated on pLL or 804G, and caspase-8 activity was
measured after 4 and 48 h. *P ⬍ 0.005 compared with pLL (4 h); †P ⬍
0.0002 compared with pLL (48 h).
FIG. 1. 804G matrix protects primary pancreatic -cells from apoptosis. A: Cells were cultured for 48 h in medium containing either 10 or 1% FCS
before identification of apoptotic (plus necrotic) cells by TUNEL. The number of TUNEL-positive cells as a percent of total was evaluated by
optical microscopy. *P < 0.02 compared with pLL 10% serum; **P < 0.001 compared with pLL 1% serum. B: After 48 h of culture, cells were treated
or not treated with 2 ng/ml IL-1 for 4 h. Apoptosis was measured by ELISA, and apoptotic index is the absorbance [A
405 nm
ⴚ A
490 nm
]. Results
are means ⴞ SD for n ⴝ 5 replicates from three independent experiments. *P < 0.003 relative to pLL (control). C: After 48 h of culture and
treatment as described above, cells cultured on pLL were stained for 20 min with 1 mg/ml Hoechst 33342 and 1.25 mg/ml propidium iodide for 20
min at 37°C, before fixation and TUNEL analysis. Death was analyzed as described in
RESEARCH DESIGN AND METHODS. 䡺, apoptosis; p, secondary
necrosis; f necrosis.
EXTRACELLULAR MATRIX AND -CELL APOPTOSIS
2036 DIABETES, VOL. 53, AUGUST 2004
tested the effect of an anti-1 antibody (Ha2/5) that blocks
the adhesion of this integrin to its ECM on the survival of
-cells after4hofculture (Fig. 3A). On pLL, there was no
difference between cells treated with the control antibody
IgM and cells treated with Ha2/5. However, on the 804G
matrix, the number of TUNEL-positive cells was signifi-
cantly higher when treated with Ha2/5 (4.9%) compared
with control cells (2.3%). Furthermore, the spreading of
cells after4hofexposure to the 804G matrix was reduced
by Ha2/5 (Fig. 3B). The control antibody had no effect on
cell spreading and cell death at the concentration used
(not shown). These results suggest that the 1 integrin
subunit mediates the signal from the 804G matrix into the
cell, leading to increased spreading and survival.
Effect of 804G matrix on phosphorylation of signal-
ing proteins FAK, PKB/Akt, and ERK. FAK, which is
activated by phosphorylation upon integrin ligand binding,
mediates survival signaling downstream of integrins and
suppresses anoikis and serum withdrawal–induced apo-
ptosis (7,34,35). The PI 3-kinase–PKB/Akt pathway is
known to block various apoptotic stimuli. We therefore
investigated whether 804G matrix induces activation of
FAK and/or PKB/Akt. Cells were serum-starved in suspen-
sion for1htoreduce constitutive (serum-induced) phos-
phorylation before plating them for 30 min on pLL or 804G
in serum-free medium. Phosphorylation of Tyr397-FAK
and Ser473-PKB/Akt was analyzed by Western blot. Adhe-
sion of cells to 804G induced an increased phosphoryla-
tion of both proteins compared with adhesion to pLL,
whereas the total amount of actin was the same on both
substrates (Figs. 4A and B). These results suggest that this
matrix does indeed activate integrin-binding signaling
pathways. The amounts of total FAK and total PKB/Akt
were slightly decreased in cells on pLL compared with
those on 804G (Figs. 4A and B). It has been reported that
caspases cleave specific signaling proteins (including FAK
and PKB/Akt) during apoptosis (36). Therefore, the slight
decrease of total FAK and PKB/Akt protein amounts in
cells cultured on pLL could be due to augmented caspase
activity on this substrate compared with matrix.
The FAK signaling complex is known to mediate the
activation of the MAP kinase ERK pathway (37,38), a
well-known pro-survival pathway in a number of cell types
(11). Exposure of serum-starved cells to 804G matrix for
30 min induced higher phosphorylation of ERK1 and ERK2
(Thr-202/Tyr-204 phosphorylation) than pLL (Fig. 4C).
Attachment of cells to pLL, however, induced higher
phosphorylation of ERK proteins as compared with cells
in suspension (Fig. 4C). In summary, the 804G matrix
activates FAK (hallmark of integrin activation) as well as
PKB/Akt and ERK pathways.
Signaling pathways involved in pro-survival effects of
the 804G matrix. To investigate whether PI 3-kinase–
PKB/Akt and ERK pathways are involved in the antiapop-
totic effect of 804G matrix, we examined the effects of
LY294002 and PD98059. These are specific inhibitors of
both PI 3-kinase and MAP kinase/ERK kinase (MEK1),
respectively, and were found to be effective in rat -cells at
the concentration used (data not shown). In the short term
(4 h), blocking PI 3-kinase and MEK1 activities signifi-
FIG. 2. The spreading and survival of -cells on purified laminin-5 is similar to that of cells cultured on 804G matrix. A: Rat -cells were cultured
for 48 h under standard conditions on pLL-, 804G matrix– or purified laminin-5 (Ln-5)-coated dishes. Apoptotic (plus necrotic cells) were
identified by TUNEL. *P < 0.001 relative to pLL (control). B: Phase-contrast microscopy after 48 h of culture as in panel A.
FIG. 3. The 1 integrin subunit is involved in the spreading and improved survival of -cells induced by the 804G matrix. A: Cells were pretreated
with control antibody (IgM) or with anti-1 integrin antibody (Ha2/5) and then attached on pLL- or on 804G-coated dishes. After4hofculture
under standard conditions, cells were fixed, and cell death was analyzed by TUNEL. *P < 0.03 relative to 804G (IgM), pLL (IgM), and pLL (Ha2/5).
B: Phase-contrast microscopy after4hofculture as in panel A. A group of spread cells is shown by the arrow.
E. HAMMAR AND ASSOCIATES
DIABETES, VOL. 53, AUGUST 2004 2037
cantly augmented cell death on 804G to the same extent
(Fig. 5A). This treatment did not affect cell death on pLL.
The combination of both inhibitors did not alter the effect
of each individual inhibitor, suggesting that they block a
common pathway activated by the matrix. However, cells
on 804G survived better than cells on pLL, even in pres-
ence of inhibitors, suggesting that both pathways are only
partially involved in the short-term antiapoptotic effect of
804G, or that the inhibitors were only partially effective
under these conditions. Blockade of the MAP kinase ERK
pathway with PD98059 during 48 h induced a striking
increase of the number of apoptotic cells on 804G and also
affected cell survival on pLL, albeit to a lesser extent (Fig.
5B). In the presence of this inhibitor, attachment to the
804G matrix no longer exerted any protective effect com-
pared with control (pLL). The Western blot analysis (Fig.
4C) showed that the attachment on pLL itself induces ERK
phosphorylation to a limited extent as compared with
suspended cells, although to a lesser extent than on 804G,
which may explain why 48 h of treatment with PD98059
also induced higher cell death in cells on pLL. It was also
noted that 48 h of culture with PD98059 did not affect the
well-spread morphology of the cells cultured on 804G, and
that DMSO (used for the dilution of inhibitors) alone had
no effect on cell survival and cell spreading at the concen-
tration used (not shown). Treatment of cells with the PI
3-kinase inhibitor LY294002 did not significantly affect cell
death on either pLL or 804G after 48 h (Fig. 5B), and
combined treatment with both inhibitors gave similar
results as treatment with PD98059 alone (Fig. 5B). These
results suggest that the MAP kinase ERK pathway is the
dominant one for long-term survival of primary pancreatic
-cells induced by the ECM.
Effect of 804G matrix on gene expression. Differential
expression on 804G matrix of genes involved in specific
signal transduction pathways such as those implicated in
apoptosis and cell cycle was investigated using low-
density cDNA microarray membranes (GEArray). Only
one gene present on this array was significantly overex-
pressed in cells cultured for 18 h on the 804G matrix
compared with control cells: IB␣ (not shown). IB␣ has
been reported to have an antiapoptotic function in pancre-
atic -cells (39). This overexpression was confirmed by
real-time quantitative PCR: the IB␣ mRNA (normalized to
GAPDH mRNA levels) was at least three times more highly
expressed in cells cultured on 804G matrix compared with
cells cultured on pLL (Fig. 6A). IB␣ protein levels were
measured by Western blot. IB␣ protein was increased
slightly on 804G vs. pLL after 24 h of culture, with a clear
increase by 48 h (Fig. 6B). There were no differences in
IB␣ expression (mRNA and protein) between noncoated
and pLL-coated dishes (data not shown). Although this
augmented protein expression correlates with the in-
creased gene transcription, we cannot exclude that this
difference might be a consequence (partly or completely)
of an enhanced degradation of this protein in cells cul-
tured on pLL-coated dishes compared with 804G.
The MAP kinase ERK pathway is involved in IB␣
overexpression induced by the 804G matrix. Given
that the 804G matrix stimulates the ERK cascade and that
this signaling pathway is involved in the pro-survival
effects of this matrix, we hypothesized that this cascade
might mediate the effect of the matrix on IB␣ gene
expression. This was tested using the inhibitor PD98059.
Treatment of cells with PD98059 did not significantly affect
IB␣ expression (mRNA) in cells cultured on pLL, but it
did reduce by ⬃60% the IB␣ overexpression induced by
the 804G matrix (Fig. 6A). Thus, these quantitative PCR
FIG. 4. 804G matrix induces phosphorylation of signaling proteins FAK, PKB/Akt, and ERK. FAK (A), PKB/Akt (B), and ERK1/2 (C)
phosphorylation was determined by Western blotting with specific anti-phospho antibodies compared with blotting with antibodies specific for
each corresponding protein. Each figure is representative of three independent experiments. Susp., suspension.
FIG. 5. Involvement of both ERK and Akt/PKB pathways in the pro-survival effect of the matrix. Cells were pretreated with inhibitors of PI
3-kinase (LY294002: LY) and/or of MEK1 (PD98059: PD) and then attached on pLL- or 804G-coated dishes. After4h(A)or48h(B), cell death
was analyzed by TUNEL. Results are shown as means ⴞ SEM of a minimum of three independent experiments. *P < 0.04 compared with 804G
control; **P < 0.01 compared with 804G control; ***P < 0.005 compared with control conditions (for pLL and 804G). 䡺 pLL; f, 804G.
EXTRACELLULAR MATRIX AND -CELL APOPTOSIS
2038 DIABETES, VOL. 53, AUGUST 2004
analyses show that blocking the ERK pathway significantly
reduces the effect of the 804G matrix on IB␣ gene ex-
pression. This result supports the notion that the activa-
tion of the MAP kinase ERK cascade by the 804G matrix
might contribute to long-term survival of pancreatic -cells
by inducing the expression of antiapoptotic genes.
DISCUSSION
We examined whether ECM can rescue primary pancreatic
-cells from apoptosis and investigated the intracellular
pathways involved. The 804G matrix protects pancreatic
-cells from apoptosis under three different conditions:
“standard” culture conditions, after serum deprivation, and
in response to the cytokine IL-1. These results demon-
strate the importance of adequate ECM for -cell survival.
It is interesting to note that signaling pathways induced by
IL-1 in islet -cells differ when this cytokine is adminis-
tered in vivo compared with in vitro treatment (40). It has
been suggested that one primary difference between the in
vivo and the in vitro model systems might be the absence
of ECM in the latter (41).
It has been shown that laminin is a component of the
basement membrane surrounding islets in vivo, and that
the loss of this basement membrane occurring during islet
isolation leads to increased cell death (4,42). Experiments
performed with laminin-5 purified from 804G matrix (Fig.
2A and B) suggest that it is the major component of this
matrix responsible for its pro-survival effect. The fact that
this matrix induces higher phosphorylation of FAK com-
pared with attachment to pLL (Fig. 4A) indicates that it
engages and activates integrins. Our group has shown that
rat pancreatic -cells express ␣61 and ␣31 integrins
(27,33), which have both been identified as receptors for
laminin-5. It was shown previously that function-blocking
antibodies directed against the ␣6 integrin subunit blocked
cell spreading induced by 804G matrix (27). Here we show
that a blocking antibody directed against the 1 integrin
subunit reduces both the spreading and the improved sur-
vival induced by the 804G matrix (Fig. 3A and B). Collec-
tively, these data lead us to suggest that ␣61 integrin is a
major conduit for outside-in signaling pathway in -cells
established on this particular matrix.
The intracellular molecule(s) responsible for transduc-
ing cell-matrix signaling in -cells have not been studied
previously in detail. We now show that compared with
pLL, attachment of cells to the 804G matrix induces higher
levels of phosphorylation of both MAP kinase ERK and
PKB/Akt (Figs. 4B and C). Furthermore, apoptosis of cells
cultured on 804G was increased in the presence of the
MEK1 inhibitor PD98059 (4 and 48 h) and/or in the
presence of the PI 3-kinase inhibitor LY294002 (4 h only)
(Figs. 5A and B). These results strongly suggest that the
short-term pro-survival effect emanating from the 804G
matrix is mediated by both of these pathways, whereas
longer term effects are mediated at least in part by MAP
kinase ERK but not by PKB/Akt.
We propose that the role of both the PI 3-kinase–PKB/
Akt and MAP kinase ERK pathways in mediating the
short-term (4 h) pro-survival effects of the 804G matrix
may be related to their effects on preventing the cleavage
and subsequent activation of procaspase-8. Indeed, it has
been reported that activated PKB/Akt prevents apoptosis
by inhibiting death-inducing signaling complex assembly,
thus preventing procaspase-8 cleavage (43), whereas MAP
kinase activation deflects DISC signaling from activating
caspase-8 (44). This would also explain the nonadditive
effect of both inhibitors on short-term cell survival. During
the short-term studies, cells on 804G matrix survived bet-
ter than cells on pLL, even in the presence of inhibitors,
suggesting that the ERK and PKB/Akt pathways may be
only two components of the short-term antiapoptotic ef-
fect of the matrix. A potential mechanism contributing to
cell death on pLL, presumably unrelated to ERK and PKB/
Akt pathways, is “integrin-mediated death,” whereby unli-
gated integrins in cells adhering to substrates devoid of
appropriate ECM ligands can act as negative regulators of
cell survival (45,46). Unligated integrins are reported to
promote apoptosis by recruitment of caspase-8 to the
plasma membrane independent of death receptors or
Fas-associating protein with a death domain. Our results
showing that caspase-8 activation is significantly reduced
in cells cultured on 804G compared with cells cultured on
pLL (Table 1) support both of the above hypotheses.
The present work shows for the first time the impor-
tance of the ERK pathway in promoting survival of pan-
creatic -cells induced by the ECM. The fact that the ERK
FIG. 6. IB␣ expression is induced by the 804G matrix, and this overexpression is reduced by treatment with the ERK inhibitor PD98059. A: Cells
were treated or not treated with PD98059 before plating them on pLL- versus 804G-coated dishes. After 18 h, RNA was extracted, and real-time
PCR was performed. The results (means ⴞ SD from two independent experiments) show the IB␣/GAPDH relative mRNA expression, which was
normalized to the control (pLL). B: The IB␣ protein levels in cells cultured for 24 and 48 h, on pLL or on 804G, were analyzed by Western
blotting. This Western blot is representative of three independent experiments.
E. HAMMAR AND ASSOCIATES
DIABETES, VOL. 53, AUGUST 2004 2039
pathway is involved in the antiapoptotic effect of the 804G
matrix on pancreatic -cells is intriguing because activa-
tion of this pathway has been reported to be involved in
cytokine-induced apoptosis in -cells (19,47). One poten-
tial explanation for this discrepancy is that cytokine- and
ECM-induced activation of ERKs may be significantly dif-
ferent in terms of duration and intensity, which may lead
to opposite effects on life-or-death decisions by the cell (48).
The finding that the 804G matrix induces overexpres-
sion of IB␣ is most interesting because IB␣ has been
shown to inhibit NF-B nuclear translocation and tran-
scriptional activity (49). In general, NF-B is reported to
mediate antiapoptotic signals. However, NF-B is consid-
ered as an important transcription factor, mediating IL-1–
induced signal transduction and regulating groups of
genes contributing to death in pancreatic -cells (31).
Overexpression of a nondegradable IB mutant, which
specifically blocks cytokine-induced NF-B activation,
prevents -cell apoptosis (39). The overexpression of the
IB␣ gene induced by the 804G matrix may thus contrib-
ute to its antiapoptotic effect. It is well established that
ECM can influence gene expression via the MAP kinase
ERK pathway (11). In the present work, we show that
activation of the ERK pathway by the 804G matrix medi-
ates the overexpression of IB␣ in -cells. This suggests
that activation of the MAP kinase ERK cascade by the
804G matrix might contribute to long-term survival of
pancreatic -cells by inducing the expression of antiapo-
ptotic genes.
Inadequate culture conditions have been suggested to
be responsible, at least in part, for the important loss of
islet cell viability before transplantation in diabetic pa-
tients. A prerequisite for the improvement of islet cell
maintenance in vitro, and consequently for increased
survival of the grafted cells, is to gain better knowledge of
both the intracellular pathways involved in cell survival
and factors regulating these signaling pathways. In this
context, these results are particularly relevant in providing
an insight into important signaling pathways activated by
ECM, which lead to improved pancreatic -cell survival
before transplantation.
ACKNOWLEDGMENTS
This work was supported by grants 3200BO-101902/1 and
3200BO-100656/1 from the Swiss National Science Foun-
dation, grants 4-1999-844 and 1-1998-213 from Juvenile
Diabetes Foundation International, and an educational
grant from Novo Nordisk A/S.
We thank Dr. Jean-Claude Irminger and Katharina Rick-
enbach for useful discussions.
REFERENCES
1. Frisch SM, Francis H: Disruption of epithelial cell-matrix interactions
induces apoptosis. J Cell Biol 124:619 –626, 1994
2. Grossmann J: Molecular mechanisms of “detachment-induced apoptosis—
Anoikis.” Apoptosis 7:247–260, 2002
3. Wang RN, Rosenberg L: Maintenance of beta-cell function and survival
following islet isolation requires reestablishment of the islet-matrix rela-
tionship. J Endocrinol 163:181–190, 1999
4. Rosenberg L, Wang R, Paraskevas S, Maysinger D: Structural and func-
tional changes resulting from islet isolation lead to islet cell death. Surgery
126:393–398, 1999
5. Ris F, Hammar E, Bosco D, Pilloud C, Maedler K, Donath MY, Oberholzer
J, Zeender E, Morel P, Rouiller D, Halban PA: Impact of integrin-matrix
matching and inhibition of apoptosis on the survival of purified human
beta-cells in vitro. Diabetologia 45:841– 850, 2002
6. Miranti CK, Brugge JS: Sensing the environment: a historical perspective
on integrin signal transduction. Nat Cell Biol 4:E83–E90, 2002
7. Frisch SM, Vuori K, Ruoslahti E, Chan-Hui PY: Control of adhesion-
dependent cell survival by focal adhesion kinase. J Cell Biol 134:793–799,
1996
8. Ilic D, Almeida EA, Schlaepfer DD, Dazin P, Aizawa S, Damsky CH:
Extracellular matrix survival signals transduced by focal adhesion kinase
suppress p53-mediated apoptosis. J Cell Biol 143:547–560, 1998
9. Khwaja A, Rodriguez-Viciana P, Wennstrom S, Warne PH, Downward J:
Matrix adhesion and Ras transformation both activate a phosphoinositide
3-OH kinase and protein kinase B/Akt cellular survival pathway. Embo J
16:2783–2793, 1997
10. Lee JW, Juliano RL: Alpha5beta1 integrin protects intestinal epithelial cells
from apoptosis through a phosphatidylinositol 3-kinase and protein kinase
B-dependent pathway. Mol Biol Cell 11:1973–1987, 2000
11. Howe AK, Aplin AE, Juliano RL: Anchorage-dependent ERK signaling–
mechanisms and consequences. Curr Opin Genet Dev 12:30 –35, 2002
12. Frisch SM, Screaton RA: Anoikis mechanisms. Curr Opin Cell Biol
13:555–562, 2001
13. Contreras JL, Smyth CA, Bilbao G, Young CJ, Thompson JA, Eckhoff DE:
Simvastatin induces activation of the serine-threonine protein kinase AKT
and increases survival of isolated human pancreatic islets. Transplanta-
tion 74:1063–1069, 2002
14. Liu W, Chin-Chance C, Lee EJ, Lowe WL Jr: Activation of phosphatidyl-
inositol 3-kinase contributes to insulin-like growth factor I-mediated
inhibition of pancreatic beta-cell death. Endocrinology 143:3802–3812,
2002
15. Srinivasan S, Bernal-Mizrachi E, Ohsugi M, Permutt MA: Glucose promotes
pancreatic islet beta-cell survival through a PI 3-kinase/Akt-signaling
pathway. Am J Physiol Endocrinol Metab 283:E784–E793, 2002
16. Tuttle RL, Gill NS, Pugh W, Lee JP, Koeberlein B, Furth EE, Polonsky KS,
Naji A, Birnbaum MJ: Regulation of pancreatic beta-cell growth and
survival by the serine/threonine protein kinase Akt1/PKBalpha. Nat Med
7:1133–1137, 2001
17. Wrede CE, Dickson LM, Lingohr MK, Briaud I, Rhodes CJ: Protein kinase
B/Akt prevents fatty acid-induced apoptosis in pancreatic beta-cells (INS-
1). J Biol Chem 277:49676 –49684, 2002
18. Paraskevas S, Aikin R, Maysinger D, Lakey JR, Cavanagh TJ, Hering B,
Wang R, Rosenberg L: Activation and expression of ERK, JNK, and p38
MAP-kinases in isolated islets of Langerhans: implications for cultured
islet survival. FEBS Lett 455:203–208, 1999
19. Pavlovic D, Andersen NA, Mandrup-Poulsen T, Eizirik DL: Activation of
extracellular signal-regulated kinase (ERK)1/2 contributes to cytokine-
induced apoptosis in purified rat pancreatic beta-cells. Eur Cytokine Netw
11:267–274, 2000
20. Larsen CM, Wadt KA, Juhl LF, Andersen HU, Karlsen AE, Su MS, Seedorf
K, Shapiro L, Dinarello CA, Mandrup-Poulsen T: Interleukin-1beta-induced
rat pancreatic islet nitric oxide synthesis requires both the p38 and
extracellular signal-regulated kinase 1/2 mitogen-activated protein kinases.
J Biol Chem 273:15294 –15300, 1998
21. Mandrup-Poulsen T: -Cell apoptosis: stimuli and signaling. Diabetes 50
(Suppl. 1):S58 –S63, 2001
22. Ammendrup A, Maillard A, Nielsen K, Aabenhus Andersen N, Serup P,
Dragsbaek Madsen O, Mandrup-Poulsen T, Bonny C: The c-Jun amino-
terminal kinase pathway is preferentially activated by interleukin-1 and
controls apoptosis in differentiating pancreatic -cells. Diabetes 49:1468 –
1476, 2000
23. Baker SE, DiPasquale AP, Stock EL, Quaranta V, Fitchmun M, Jones JC:
Morphogenetic effects of soluble laminin-5 on cultured epithelial cells and
tissue explants. Exp Cell Res 228:262–270, 1996
24. Hormia M, Falk-Marzillier J, Plopper G, Tamura RN, Jones JC, Quaranta V:
Rapid spreading and mature hemidesmosome formation in HaCaT kera-
tinocytes induced by incubation with soluble laminin-5r. J Invest Dermatol
105:557–561, 1995
25. Langhofer M, Hopkinson SB, Jones JC: The matrix secreted by 804G cells
contains laminin-related components that participate in hemidesmosome
assembly in vitro. J Cell Sci 105:753–764, 1993
26. Hayek A, Beattie GM, Cirulli V, Lopez AD, Ricordi C, Rubin JS: Growth
factor/matrix-induced proliferation of human adult -cells. Diabetes 44:
1458 –1460, 1995
27. Bosco D, Meda P, Halban PA, Rouiller DG: Importance of cell-matrix
interactions in rat islet -cell secretion in vitro: role of ␣61 integrin.
Diabetes 49:233–243, 2000
28. Rouiller DG, Cirulli V, Halban PA: Differences in aggregation properties
EXTRACELLULAR MATRIX AND -CELL APOPTOSIS
2040 DIABETES, VOL. 53, AUGUST 2004
and levels of the neural cell adhesion molecule (NCAM) between islet cell
types. Exp Cell Res 191:305–312, 1990
29. Kelly KJ, Sandoval RM, Dunn KW, Molitoris BA, Dagher PC: A novel
method to determine specificity and sensitivity of the TUNEL reaction in
the quantitation of apoptosis. Am J Physiol Cell Physiol 284:C1309–C1318,
2003
30. Schaffner W, Weissmann C: A rapid, sensitive, and specific method for the
determination of protein in dilute solution. Anal Biochem 56:502–514, 1973
31. Eizirik DL, Mandrup-Poulsen T: A choice of death–the signal-transduction
of immune-mediated beta-cell apoptosis. Diabetologia 44:2115–2133, 2001
32. Maedler K, Sergeev P, Ris F, Oberholzer J, Joller-Jemelka HI, Spinas GA,
Kaiser N, Halban PA, Donath MY: Glucose-induced beta cell production of
IL-1beta contributes to glucotoxicity in human pancreatic islets. J Clin
Invest 110:851– 860, 2002
33. Kantengwa S, Baetens D, Sadoul K, Buck CA, Halban PA, Rouiller DG:
Identification and characterization of alpha 3 beta 1 integrin on primary
and transformed rat islet cells. Exp Cell Res 237:394– 402, 1997
34. Hungerford JE, Compton MT, Matter ML, Hoffstrom BG, Otey CA: Inhibi-
tion of pp125FAK in cultured fibroblasts results in apoptosis. J Cell Biol
135:1383–1390, 1996
35. Almeida EA, Ilic D, Han Q, Hauck CR, Jin F, Kawakatsu H, Schlaepfer DD,
Damsky CH: Matrix survival signaling: from fibronectin via focal adhesion
kinase to c-Jun NH(2)-terminal kinase. J Cell Biol 149:741–754, 2000
36. Widmann C, Gibson S, Johnson GL: Caspase-dependent cleavage of
signaling proteins during apoptosis: a turn-off mechanism for anti-apo-
ptotic signals. J Biol Chem 273:7141–7147, 1998
37. Renshaw MW, Price LS, Schwartz MA: Focal adhesion kinase mediates the
integrin signaling requirement for growth factor activation of MAP kinase.
J Cell Biol 147:611– 618, 1999
38. Parsons JT: Focal adhesion kinase: the first ten years. J Cell Sci 116:1409–
1416, 2003
39. Heimberg H, Heremans Y, Jobin C, Leemans R, Cardozo AK, Darville M,
Eizirik DL: Inhibition of cytokine-induced NF-B activation by adenovirus-
mediated expression of a NF-B super-repressor prevents -cell apoptosis.
Diabetes 50:2219 –2224, 2001
40. Todaro M, Di Gaudio F, Lavitrano M, Stassi G, Papaccio G: Islet beta-cell
apoptosis triggered in vivo by interleukin-1beta is not related to the
inducible nitric oxide synthase pathway: evidence for mitochondrial
function impairment and lipoperoxidation. Endocrinology 144:4264– 4271,
2003
41. Lowe WL Jr: “The matrix unloaded”: implications for cytokine signaling in
islets? Endocrinology 144:4262– 4263, 2003
42. Wang RN, Paraskevas S, Rosenberg L: Characterization of integrin expres-
sion in islets isolated from hamster, canine, porcine, and human pancreas.
J Histochem Cytochem 47:499 –506, 1999
43. Jones RG, Elford AR, Parsons MJ, Wu L, Krawczyk CM, Yeh WC, Hakem R,
Rottapel R, Woodgett JR, Ohashi PS: CD28-dependent activation of protein
kinase B/Akt blocks Fas-mediated apoptosis by preventing death-inducing
signaling complex assembly. J Exp Med 196:335–348, 2002
44. Holmstrom TH, Schmitz I, Soderstrom TS, Poukkula M, Johnson VL, Chow
SC, Krammer PH, Eriksson JE: MAPK/ERK signaling in activated T cells
inhibits CD95/Fas-mediated apoptosis downstream of DISC assembly.
Embo J 19:5418 –5428, 2000
45. Cheresh DA, Stupack DG: Integrin-mediated death: an explanation of the
integrin-knockout phenotype? Nat Med 8:193–194, 2002
46. Stupack DG, Puente XS, Boutsaboualoy S, Storgard CM, Cheresh DA:
Apoptosis of adherent cells by recruitment of caspase-8 to unligated
integrins. J Cell Biol 155:459 –470, 2001
47. Donath MY, Storling J, Maedler K, Mandrup-Poulsen T: Inflammatory
mediators and islet beta-cell failure: a link between type 1 and type 2
diabetes. J Mol Med 81:455– 470, 2003
48. Rul W, Zugasti O, Roux P, Peyssonnaux C, Eychene A, Franke TF,
Lenormand P, Fort P, Hibner U: Activation of ERK, controlled by Rac1 and
Cdc42 via Akt, is required for anoikis. Ann N Y Acad Sci 973:145–148, 2002
49. Dixit V, Mak TW: NF-kappaB signaling: many roads lead to Madrid. Cell
111:615– 619, 2002
E. HAMMAR AND ASSOCIATES
DIABETES, VOL. 53, AUGUST 2004 2041