(Received for publication, February 20, 1997, and in revised form, April 21, 1997)
From the Department of Biochemistry, Cornell
University Medical College, New York, New York 10021 and the
¶ Departments of Physiology, Medicine, and Surgery, College of
Physicians and Surgeons of Columbia University,
New York, New York 10032
Advanced glycation end products (AGEs) exert
their cellular effects on cells by interacting with specific cellular
receptors, the best characterized of which is the receptor for AGE
(RAGE). The transductional processes by which RAGE ligation transmits signals to the nuclei of cells is unknown and was investigated. AGE-albumin, a prototypic ligand, activated p21ras in rat
pulmonary artery smooth muscle cells that express RAGE, whereas
nonglycated albumin was without effect. MAP kinase activity was
enhanced at concentrations of AGE-albumin, which activated p21ras and NF-B. Depletion of intracellular glutathione
rendered cells more sensitive to AGE-mediated activation of this
signaling pathway. In contrast, signaling was blocked by preventing
p21ras from associating with the plasma membrane or mutating
Cys118 on p21ras to Ser. Signaling was
receptor-dependent, because it was prevented by blocking
access to RAGE with either anti-RAGE IgG or by excess soluble RAGE.
These data suggest that RAGE-mediated induction of cellular oxidant
stress triggers a cascade of intracellular signals involving
p21ras and MAP kinase, culminating in transcription factor
activation. The molecular mechanism that triggers this pathway likely
involves oxidant modification and activation of p21ras.
In the presence of aldoses, proteins become nonenzymatically glycated and oxidized (1-3). This initially reversible glycation is followed by further irreversible rearrangements leading to a class of permanently modified proteins known as advanced glycation end products (AGEs).1 Although glycated proteins are found at low levels in normal individuals during aging, significantly higher levels are found in certain disease states such as diabetes and renal failure (4, 5). We have identified a cellular receptor for AGEs, termed RAGE, which exhibits a wide tissue distribution (6-9). We have recently demonstrated the enhanced presence of RAGE in vascular smooth muscle of diabetic vasculature (renal arterial vessel) compared with a similar sized vessel from a nondiabetic age-matched control. These areas of enhanced RAGE immunoreactivity co-localize with enhanced immunostaining for AGE-reactive epitopes (10). Our previous data in endothelial cells and in vivo demonstrated that interaction of AGEs with RAGE results in triggering a range of cellular responses, including transcription factor activation and changes in gene expression (11-14). However, the means by which a signal reflecting AGE engagement of RAGE is transmitted to the nucleus is not known. Given the enhanced expression of AGE and RAGE in diabetic vascular smooth muscle, we focused on elucidating the signaling pathways in smooth muscle cells that are triggered upon ligation of RAGE by AGE-albumin, a prototypical ligand.
Recent evidence supports a role for reactive oxygen species in
mediating signaling by several receptor systems (15-21). For example,
platelet-derived growth factor has recently been shown to stimulate
H2O2 production in vascular smooth muscle cells
(15). When production of oxidants was blocked, platelet-derived growth factor-induced enhancement of mitogen-activated protein (MAP) kinase
activity, chemotaxis, and DNA synthesis was prevented (15). Others have
found that induction of c-fos expression by tumor necrosis
factor- and basic fibroblast growth factor requires production of
reactive oxygen intermediates (16), as does activation of the MAP
kinase cascade in NIH-3T3 cells (17) and in neutrophils (18).
Our previous data indicate that interaction of AGEs with endothelial
RAGE and vascular RAGE in vivo results in the generation of
significant cellular redox stress manifested by the appearance of
malondialdehyde-reactive epitopes, increased mRNA for heme oxygenase-1 (12) and activation of the transcription factor NF-B
(12). Furthermore, we have recently identified a p21ras/MAP
kinase signaling cascade that is regulated by redox stress (22).
Therefore, we performed studies to elucidate the molecular signals that
result from AGE-RAGE interaction.
The farnesyltransferase inhibitor
-hydroxyfarnesylphosphonic acid was obtained from Biomol (Plymouth
Meeting, PA), and
L-buthionine-(S,R)-sulfoximine was
from Sigma.
Bovine serum albumin (Sigma) was
glycated by incubation with glucose (0.5 M) or ribose (0.25 M) at 37 °C for 6 weeks (6, 7). Controls consisted of
the same initial preparations of albumin incubated in the same manner
in the absence of aldose. Rat RAGE and monospecific rabbit anti-RAGE
IgG were prepared and characterized as described previously for bovine
RAGE (6, 12). The 35 kDa extracellular domain of rat RAGE was termed
soluble RAGE (sRAGE).
Rat pulmonary artery smooth
muscle cells and PC12 cells were maintained in Dulbecco's modified
Eagle's medium containing 10% fetal bovine serum (and 5% horse serum
for PC12 cells) and 2% L-glutamine. Cells treated with
inhibitors were resuspended in serum-free medium containing
-hydroxyfarnesylphosphonic acid (10 µM) or
L-buthionine-(S,R)-sulfoximine (100 µg/ml) 24 h prior to the addition of AGE-albumin.
Nuclei were isolated,
mixed with the 32P-labeled NF-B consensus sequence, and
assayed exactly as described previously (23). To confirm the
specificity of the assay, positive samples were incubated with 100-fold
excess unrelated NF-
B-specific DNA as well as 100-fold excess
unlabeled nonspecific DNA. In all cases, the NF-
B-specific DNA
effectively competed with the sample, whereas the unrelated DNA had no
effect.
The assay used to measure the GTP/GDP ratio on immunoprecipitated p21ras was essentially that of Downward et al. (24) with minor modifications as described previously (25).
MAP Kinase AssaySerum-starved cells (24 h, 5 × 106) were treated for the indicated times at 37 °C prior to MAP kinase assay as we previously described (22).
Our previous
studies have revealed activation of p21ras by reactive oxygen
and reactive nitrogen species (22, 25). Because the ability of AGEs to
induce NF-B activation has been shown to be dependent upon the
generation of cellular redox stress (12), we examined whether AGEs
initiate signaling to the nucleus by activating p21ras.
Addition of AGE-albumin (100 µg/ml) to rat pulmonary artery smooth
muscle cells (SMCs), in which the intracellular nucleotide phosphate
pools were labeled to equilibrium with
[32P]orthophosphate, yielded enhanced levels of GTP-bound
p21ras upon immunoprecipitation (Fig. 1,
solid circles). In contrast, cells treated with nonglycated
albumin (Fig. 1, open circles and open triangles)
had p21ras with basal levels of GTP associated with it
(13.2 ± 2.5%). Furthermore, cells in which endogenous
glutathione levels were greatly reduced by pretreatment with
L-buthionine-(S,R)-sulfoximine (20,
22) showed an enhanced response on subsequent addition of AGE-albumin (Fig. 1, solid triangles). We have previously demonstrated
that this concentration of AGE-albumin was optimal for
AGE-RAGE-mediated cellular oxidant stress (12). These data indicate
that endogenous p21ras was stimulated by addition of AGE and
that this activation is under redox regulation.
MAP Kinase Activation by AGEs
One of the known critical
intermediates in signal transduction by p21ras is activation of
a family of Ser/Thr kinases termed MAP kinases. We investigated whether
two of these kinases, ERK1 and ERK2, were activated by AGE-albumin.
Treatment of SMCs with AGE-albumin for 10-15 min resulted in
concentration-dependent activation of immunoprecipitated ERK1 and ERK2 (Fig. 2, open circles).
Addition of nonglycated albumin had no effect (data not shown).
Depletion of glutathione by exposure of SMCs to
L-buthionine-(S,R)-sulfoximine
enhanced AGE-albumin stimulation of MAP kinase activity (Fig. 2,
solid circles). We have previously reported an increase of
1.5-2-fold in MAP kinase activity upon addition of free radical
generators (22). Although the amplitude of stimulation seen in our
experiments was very consistent, others have found greater MAP kinase
activation with different stimuli and cells. This is likely due to the
relatively high level of basal MAP kinase activity in our SMC cultures
compared with other systems in which basal activity is minimal.
Activation of MAP kinases is typically an early event in cellular
activation. Therefore, we investigated the time course of AGE-RAGE-mediated activation of MAP kinase. Peak activity was observed
by 10 min after exposure of cells to AGE-albumin (100 µg/ml) (Fig.
3). MAP kinase activity increased as early as 2 min and
returned to basal levels at 60 min. Thus, stimulation of ERK1 and ERK2
activities by AGE-albumin resulted in an early, transient stimulation
of MAP kinase activity.
Receptor Dependence of AGE Stimulation
Generation of free
radicals and cellular oxidant stress has been attributed to AGEs
themselves (26, 27) and as a consequence of their interaction with RAGE
(12, 13). Therefore, we investigated whether AGE interaction with RAGE
on SMC was necessary for MAP kinase activation. Receptor blockade was
achieved by the addition of sRAGE, a truncated form consisting of only
the extracellular domain of the receptor, or monospecific, polyclonal
anti-RAGE IgG (Fig. 4). Addition of increasing
concentrations of sRAGE prevented AGE-enhanced MAP kinase activity in a
concentration-dependent manner (Fig. 4, open
bars). One likely explanation for the effect of sRAGE lies in its
ability to bind exogenous AGEs, thereby inhibiting their interaction
with cell surface RAGE and preventing generation of oxidant stress.
Anti-RAGE IgG, previously shown to prevent AGE binding to RAGE (6),
also prevented AGEs from stimulating MAP kinase activity in SMCs in a
concentration-dependent manner (Fig. 4, hatched
bars). In contrast, nonimmune IgG was without effect (data not
shown). Thus, activation of the MAP kinase cascade required AGE-RAGE
interaction at the cell surface.
The Role of p21ras in the Signaling Cascade
The
critical test of our hypothesis involved determining whether
AGE-induced stimulation of p21ras was linked to MAP kinase and
NF-B activation in SMC. To evaluate this, cells were pretreated with
-hydroxyfarnesylphosphonic acid. This compound specifically inhibits
farnesyltransferase, the enzyme responsible for the lipid modification
of p21ras (28). Without this lipid, p21ras does not
associate with the plasma membrane and thus cannot activate its
effectors and signal downstream (29). We found that SMC pretreated with
this inhibitor no longer responded to AGE-albumin by activating ERK1
and ERK2 kinases (Fig. 5). Furthermore, AGE-mediated nuclear translocation of NF-
B was also prevented by inhibition of
farnesyltransferase (Fig. 6). These data indicate that
p21ras or a related low molecular weight G protein is required
for signal transduction following AGE binding to cell surface RAGE.
A molecular target of reactive free radicals on p21ras has been
defined as Cys118 (30). To determine if this site is
targeted by reactive oxygen species generated by AGE-albumin, we
created a mutant of p21ras in which Cys118 was
changed to a Ser (30) and overexpressed it in PC12 cells, which also
express RAGE (9). These cells, termed PC12 p21rasC118S, were
compared with wild-type parental cells for their ability to respond to
AGE-albumin by activation of p21ras-dependent MAP
kinase activity. As seen in Fig. 7, wild-type cells responded to AGE-albumin by activating ERK1/2 kinases, whereas PC12
p21rasC118S cells were refractory to AGE-albumin. These data
provide a molecular basis for RAGE signal transduction.
Recently, a role for reactive oxygen intermediates in mediating
signal transduction has become apparent. For example, the ability of
platelet-derived growth factor to stimulate MAP kinase activity, DNA
synthesis, and chemotaxis in vascular smooth muscle cells was
completely blocked by the addition of catalase or antioxidants (15).
These cells were also found to produce H2O2
upon growth factor addition (15). Others have found that antioxidants
blocked the ability of a variety of unrelated stimuli to trigger
NF-B activation (31). Thus, a role for free radicals in mediating signal transduction is emerging.
We have previously demonstrated that interaction of AGEs with RAGE
induces cellular oxidant stress, probably as a consequence of the
generation of reactive oxygen intermediates (11, 12). In this study our
aim was to determine if cellular oxidant stress consequent to AGE-RAGE
interaction would trigger the p21ras/MAP kinase pathway. Our
observations indicate that this is indeed the case because
p21ras, MAP kinase activity, and NF-B nuclear translocation
are stimulated by AGE-albumin, enhanced by glutathione depletion, and
prevented by blockade of RAGE. The molecular mechanism may be due to
triggering of p21ras exchange activity by reactive oxygen
modification of Cys118 on p21ras. There are likely
to be many signaling pathways initiated by AGE-RAGE interaction, but
the one outlined herein is apparently important because inhibition of
p21ras blocked NF-
B activation. This transcription factor is
critical for stimulation of many acute phase response genes, and thus
the induction of many genes is likely to be controlled by this pathway. In fact, the heme oxygenase gene is under control of NF-
B (32) and
is induced upon AGE-RAGE interaction (12). In addition, our previous
work has indicated that AGE-mediated activation of NF-
B results in
binding of this transcription factor to specific sites within the
promoter of the VCAM-1 gene (33). Enhanced expression of this
endothelial cell adhesion molecule alters cellular phenotype,
potentially resulting in the adherence of mononuclear phagocytes to the
vessel wall (34).
In this work, we have examined signaling pathways likely to be of importance when AGEs ligate vascular smooth muscle cell RAGE. One of the consequences of this interaction might be enhanced production of monocyte and smooth muscle cell chemoattractant factors (14). Identification of the molecular basis underlying these events, possibly linked to the accelerated vascular disease observed in diabetes, is likely a central step toward understanding the consequences of AGE-RAGE interaction not only in vascular smooth muscle cells but also in diverse cell types such as endothelial cells and mononuclear phagocytes. The present report advances our knowledge of the processes set in motion when AGEs bind cellular RAGE, an event that culminates in transcription factor activation and alteration of cellular properties.
Past studies have not identified the source of the reactive oxygen intermediates generated by the AGE-RAGE complex. On one hand, AGEs by themselves produce reactive oxygen species that could interact directly or indirectly with critical cellular targets. Alternatively, AGE engagement of RAGE could initiate a series of events resulting in intracellular generation of reactive oxygen species, possibly by a system analogous to that of the NADPH oxidase system of neutrophils. Detailed studies examining these possibilities are ongoing.
AGEs are prevalent in certain pathophysiological conditions such as diabetes (4, 5) and renal failure or Alzheimer's disease, in which delayed protein turnover favors irreversible nonenzymatic glycation (35, 36). Therefore, understanding how AGEs transmit their signal to the nucleus will likely yield important insights into the mechanisms of these pathophysiological processes and provide targets for intervention.