| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Division of Nephrology and Hypertension, Oregon Health and Science University and the Portland Veterans Affairs Medical Center, Portland, Oregon 97201
Submitted 21 January 2003 ; accepted in final form 30 March 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
hypertonicity; heparin-binding epidermal growth factor; kidney; cylooxygenase
Cyclooxygenases (COX) are oxidoreductases that catalyze the conversion of membrane arachidonic acid to PGH2, a precursor of all prostaglandins, thromboxanes, and prostacyclins (39). There are at least two [and perhaps more (6)] isoforms of COX; COX-1 is constitutively and nearly ubiquitously expressed, whereas the inducible COX-2 isoform is primarily expressed in kidney and brain. COX-2, as a key mediator of inflammation, may be upregulated by a variety of cell activators, including mitogens, hormones, and environmental stressors (3, 21). Although ambient tonicity has long been known to influence prostaglandin synthesis in the renal medulla (11, 12) and gastrointestinal tract (2, 26), the ability of hypertonicity to increase expression of COX-2 was first noted in liver macrophages (51). In the kidney medulla, upregulated COX-2 (but not COX-1) expression was described in rodent models of chronic salt loading and water restriction (48, 49); in vitro, experimental hypertonic stress correspondingly increased COX-2 but not COX-1 expression in a collecting duct cell line (47, 48).
The signaling mechanism through which hypertonicity increases COX-2
expression is of great interest. Yang et al.
(47) observed MAPK dependence
of this phenomenon in cultured cells derived from the inner medullary
collecting duct; specifically, p38, ERK, and JNK axes were all implicated,
either pharmacologically or through a dominant negative approach. In renal
medullary interstitial cell and cortical thick ascending limb models (perhaps
most consistent with the in vivo pattern of renal COX-2 expression; see Ref.
20), tonicity also regulated
COX-2 expression but did so in an NF-
B-dependent fashion
(7,
19).
In an unbiased screen, we identified COX-2 as one of the genes most prominently upregulated by hypertonicity in the renal medullary mIMCD3 cell model (41). Because other tonicity-inducible phenomena are potentially dependent on transactivation of the EGF receptor (EGFR) kinase (e.g., Refs. 24, 35, and 38) and because at least partial dependence on the EGFR kinase effector, ERK, had previously been reported for tonicity-inducible COX-2 expression in this epithelial cell model (47), we investigated the role of the EGFR kinase in tonicity-dependent COX-2 expression.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
200 mosM (200 mM urea vs. 100 mM NaCl). All reagents were purchased from
Sigma unless otherwise specified. The following pharmacological inhibitors
were used: AG-1295 (100 nM-1 µM; Calbiochem); AG-1478 (100 nM; Calbiochem);
PD-153035 (100 nM);
N-{DK-[2-(hydroxyaminocarbonyl)methyl]-4-methyl-pentanoyl}-L-3-(2'-naphthyl)-alanyl-L-alanine
2-aminoethyl-amide (TAPI; 3–10 µM); doxycycline (100 µM); PD-98059
(50 µM); and U-0126 (10 µM). Inhibitors were applied 30 min before
solute or ligand treatment, unless otherwise indicated; all inhibitors were
present for the duration of the solute or ligand treatment interval
(additional 5 min-16 h, depending on assay).
Biochemical and molecular biological assays. Immunoblot analysis
was performed as previously described
(8,
50) using anti-COX-2 primary
antibody (no. 160106; Cayman Chemical) at 1:1,000 and goat anti-rabbit
horseradish peroxidase-conjugated secondary antibody (Pierce) at 1:4,000
dilution. Anti-phosphorylated (P)-ERK (recognizing phosphorylated
Thr202/Tyr204 of ERK1/2), anti-P-JNK (recognizing
phosphorylated Thr183/Tyr185 of JNK1/2), anti-P-p38
(recognizing phosphorylated Thr180/Tyr182 of p38), and
anti-P-EGFR (recognizing phosphorylated Tyr1068) were purchased
from Cell Signaling Technologies and used in accordance with the
manufacturer's directions; 60 kDa anti-heat shock protein (HSP60) was from
Santa Cruz Biotechnologies. Anti-phosphokinase immunoblots were controlled
with subsequent or parallel immunoblotting with the corresponding anti-kinase
antibody. In no instance did kinase expression change during the brief
intervals (0–30 min) over which kinase-dependent events were explored.
Stable transfection and luciferase reporter gene analyses were performed as
previously described (9) using
sequences from –724 to +7 of the murine COX-2 promoter subcloned into
pXP2 (43). Briefly, mIMCD3
cells were transfected with the indicated plasmid via lipofection and
subjected to selection pressure with G418. A pool of clones was generated in
this fashion and used in aggregate over several passages with equivalent
results in each passage. As the plasmid was integrated, there was no
normalization for transfection efficiency. In transient transfection
experiments performed in parallel, the degree of induction by hypertonicity
was less pronounced (data not shown), consistent with the nature of the COX-2
minimal promoter (5). Induction
was more apparent after normalization for cotransfected 
-galactosidase
reporter gene; however, normalization of studies of this sort via
cotransfection is problematic because of a general transcriptional inhibitory
effect of hypertonicity (e.g., see Ref.
41). Still other
"housekeeping" genes, such as actin, are immediate-early genes and
are susceptible to the nonphysiological superinduction
(15) that accompanies a
tonicity-dependent decrement in protein synthesis (Ref.
10 and references therein).
Depicted data are means ± SE of at least three separate experiments
(see legends for Figs. 1,
2,
3,
4,
5,
6,
7,
8,
9), with the exception of
immunoblot data wherein a representative figure is shown. Statistical
significance was ascribed to P < 0.05 via t-test
(VassarStats;
http://vassun.vassar.edu/~lowry/VassarStats.html).
For RNase protection assay to detect COX-2 mRNA, murine kidney
poly(A)+ mRNA was subjected to RT-PCR using primers COX-1411
(5'-TGTACAAGCAGTGGCAAAGG-3') and COX-1726
(5'-GCTCGGCTTCCAGTATTGAG-3') to generate a 316-bp partial cDNA,
which was T/A-subcloned into pCR4. The resultant plasmid was confirmed by
sequencing (Vollum Institute for Advanced Biomedical Research, Oregon Health
and Science University); clone CT005 exhibited identity with murine COX-2 in
BLAST analysis
(http://www.ncbi.nlm.nih-.gov/BLAST/).
This plasmid was linearized with NotI, and transcribed with T3 RNA
polymerase to generate a specific antisense COX-2 riboprobe. ELISA for
PGE2 production (R&D Systems) was performed according to the
manufacturer's directions, using 10-fold dilutions of cell-conditioned
medium.
|
|
|
|
|
|
|
|
|
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Because we have observed other potentially tonicity-dependent signaling events to be mediated via transactivation of the EGFR, we determined the effect of pharmacological inhibition of the EGFR on tonicity-dependent COX-2 expression. Two highly specific and potent EGFR kinase inhibitors, PD-153035 and AG-1478, both dramatically inhibited tonicity-dependent COX-2 mRNA expression (Fig. 2A).
Because upregulation at the mRNA level in response to hypertonic stress is frequently nonspecific, a consequence of so-called "superinduction" in the setting of the inhibition in protein synthesis that often accompanies hypertonic stress (Ref. 10 and references therein), we sought to determine the effect on COX-2 expression at the protein level. Again, consistent with the data of others, hypertonic stress dramatically increased COX-2 expression at the protein level (Fig. 2B), suggesting physiological relevance. Because the degree of cell confluence may influence signaling events, experiments were performed in parallel using subconfluent cells with comparable results (data not shown). As in the mRNA studies, both of the EGFR inhibitors markedly decreased the tonicity-dependent COX-2 upregulation (Fig. 2B).
We sought to confirm that the degree of hypertonicity used in this model was sufficient to activate the EGFR kinase. Hypertonic NaCl (100 mM) induced tyrosine phosphorylation of this receptor (on residue Tyr1068) to a degree comparable to that induced by the bona fide EGFR ligand EGF (Fig. 2C).
To address the upregulation of COX-2 expression in a more mechanistic
fashion, we examined the effect of hypertonicity on reporter gene activity of
an expression plasmid encoding luciferase and under control of the proximal
0.7 kb (–724 to +7, relative to the transcriptional start site) of
the human COX-2 promoter (43).
When stably transfected in mIMCD3 cells, the reporter gene exhibited
approximately fivefold greater activity after 6 h of hypertonic stress (100 mM
NaCl; Fig 3). This effect was
also dose dependent within the range of 50–150 mM NaCl. The effect of
urea was much more modest (27% increase) but achieved statistical
significance. The tonicity-dependent upregulation was partially blocked when
cells were pretreated with the EGFR kinase inhibitor PD-153035
(Fig. 3). The effect of the
impermeant solute mannitol was equivalent to that of NaCl in this assay, as
was its sensitivity to PD-153035 (data not shown). Because the most widely
recognized tonicity-responsive enhancer element is TonE, we sought this
element in the COX-2 promoter. Using the reported canonical sequence for TonE
(YGGAANNNYNY; see Ref. 31), we
were unable to identify this cis-acting element in the proximal 0.7
kb of the COX-2 5'-flanking sequence (data not shown).
We assessed PGE2 generation as an index of COX-2 function. Hypertonic NaCl increased PGE2 release in cell supernatants in a time-dependent fashion, and this effect was blocked by EGFR kinase inhibition (Fig. 4). Of note, the large error bar in the 16-h treatment in the absence of PD-153035 is a consequence of the variability in maximal upregulation by hypertonicity (i.e., 3-fold vs. 20-fold) in these unnormalized data; in all experiments, the ordinal relationship was identical.
ERK and JNK have been implicated in COX-2 regulation, particularly in medullary epithelial cells (47) where they may function downstream of EGFR in a signaling cascade. We therefore assessed the effect of EGFR kinase inhibition on activation of all three of the principal MAPK families by hypertonic stress. As anticipated and as previously shown in this model, hypertonicity increased ERK activation as assessed via anti-P-ERK immunoblotting (Fig. 5). Consistent with our earlier observation (52), EGFR kinase inhibition markedly blunted the effect of tonicity on ERK activation. In contrast, there was virtually no effect of EGFR kinase inhibition on tonicity-dependent JNK or p38 phosphorylation.
Because ERK appeared to be a principal effector of EGFR in this and other contexts, we tested the effect of pharmacological inhibition of ERK activation on tonicity-inducible COX-2 expression. We used two different mitogen/extracellular signal-regulated kinase (MEK) inhibitors (and inhibitors of ERK activation), PD-98059 and U-0126, in part to permit discrimination between the anticipated ERK1/2-dependent phenomenon and a potential ERK5-dependent phenomenon (23) recently reported in another model (17). Both MEK inhibitors substantially blocked the effect of NaCl on this end point, consistent with a role for MEK (and hence ERK) activation in tonicity-responsive COX-2 expression. Of note, the inhibitors were equally effective when applied at 50 µM for PD-98059 and 10 µM for U-0126, strongly suggesting that ERK5 was not playing a major role (23). Consistent with our observations with tonicity-dependent COX-2 transcription, the effect of hypertonic mannitol vis-à-vis COX-2 expression was equivalent to that of equiosmolar NaCl, and this effect was equally sensitive to MEK inhibition (Fig. 6).
EGFR activation generally occurs via metalloproteinase-dependent cleavage of an EGF-like ectodomain that then acts in an autocrine or juxtacrine fashion (reviewed in Ref. 16). To test this possibility in preliminary fashion in the present context, we investigated the ability of two metalloproteinase inhibitors to abrogate the effect of tonicity on COX-2 expression. The metalloproteinase inhibitor TAPI, which blocks ectodomain cleavage of the EGFR ligand, heparin-binding (HB)-EGF (28), inhibited tonicity-dependent COX-2 expression in a dose-dependent fashion, whereas vehicle exerted no effect (Fig. 7). The nonspecific metalloproteinase inhibitor doxycycline (e.g., Refs. 14 and 40) also blocked the effect, albeit to a lesser extent (consistent with its reported efficacy). We were unable to use the specific hydroxamate-based metalloproteinase inhibitors (e.g., ilomastat) because the vehicle concentration required by these relatively insoluble compounds (i.e., DMSO at 0.5–2%) was sufficient to interfere with COX-2 expression (data not shown). In support of our alternative approach, a recent report documented functional equivalence between TAPI and a hydroxamate-based inhibitor in at least one context (42).
HSP60 expression induces COX-2 expression (4), and hypertonic stress has been associated with increased expression of HSPs in renal epithelial cells (10). We investigated the possibility that HSP60 may mediate the effect of hypertonicity on COX-2 expression. Abundant HSP60 expression was detected in mIMCD3 cells (Fig. 8); however, this level was unaffected by hypertonic stress or by the presence of EGFR kinase inhibitors.
Last, we sought an EGFR-null model in which to test for the presence of tonicity-dependent signaling. Although the EGFR is nearly ubiquitous among cultured cell lines, CHO cells reportedly lack this tyrosine kinase (29). We compared CHO cells with two additional cell lines known to express the EGFR: 3T3 cells and the renal epithelial MDCK cell line. We first assessed the ability of EGF to activate ERK in each cell line. As anticipated, EGF induced a marked increase in ERK phosphorylation in both the 3T3 and MDCK cells at 5 and 30 min of treatment but produced no effect in the CHO cells (Fig. 9A). This confirmed the EGF nonresponsiveness of the CHO line. With this validated model in hand, we next assessed the ability of hypertonic stress to upregulate COX-2 expression in each cell line. Similar to the effect in mIMCD3 cells, hypertonic stress dramatically increased COX-2 expression in both 3T3 and MDCK cells (Fig. 9B); in marked contrast, however, there was essentially no effect in the EGFR kinase-null CHO cells. For comparison, we also determined the effect of the medullary solute, urea, on COX-2 expression in each of these models. Like NaCl, urea produced no effect in CHO cells. The urea effect was also much less prominent than that of hypertonic NaCl in both 3T3 and MDCK cells. These data were consistent with the relatively modest effect of urea vis-à-vis the COX-2 promoter (Fig. 3). Of note, introduction of an expression vector encoding EGFR in CHO cells failed to recapitulate tonicity-inducible COX-2 expression. We infer that EGFR expression is necessary but not sufficient in this context; CHO cells may lack either the EGFR ligand (i.e., HB-EGF) or the relevant ligand-directed metalloproteinase, both of which are required for an intact signaling module.
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Although medullary COX-2 expression and action likely play a pivotal role in renal physiology, particularly in response to dehydration, much of this effect has been attributed to actions of the medullary interstitial cells and adjacent cells of the macula densa and cortical thick ascending limb (reviewed in Ref. 20). Nonetheless, recent reports have emphasized expression and regulation of COX-2 in other renal epithelial cells in vitro (13, 47, 48), particularly in the context of hypertonic stress (47, 48).
It is of interest that Yang et al.
(47) did not detect an effect
of the tyrosine kinase inhibitors, tyrphostin-A23 (AG-18) and tyrphostin-A51
(AG-183), on tonicity-dependent COX-2 expression, although they did note an
inhibitory effect of the general tyrosine kinase inhibitor genistein
(47). We, in contrast, found a
profound effect in the mIMCD3 cell line at the mRNA and protein level and at
the level of PGE2 production. The study of Yang et al.
(47) was performed using the
mIMCD-K2 cell line (25),
whereas the present study employed the mIMCD3 line
(34). Tyrphostin-A23 and
tyrphostin-A51 exhibit IC50 for EGFR kinase of 40 µM and
800 nM, respectively, and were applied at 10 µM
(47); it is conceivable that
insufficient cellular levels were achieved, particularly for tyrphostin-A23.
While the present studies were being completed, Guo et al.
(17) reported the EGFR
dependence of COX-2 expression in a rat intestinal epithelial cell model in
response to the peptide hormone gastrin, in further support of this mode of
regulation.
Importantly, these are not the first data implicating a role for the EGFR
in transactivating tonicity-dependent signaling. Rosette and Karin
(35) reported aggregation of,
and signaling by, the EGFR using extreme hypertonic stress in a cultured cell
line, although specific genetic targets of this process were not examined. In
contrast, King et al. (24)
reported activation of the EGFR by osmotic shock in the absence of receptor
dimerization. Sheikh-Hamad and co-workers
(38) described
tonicity-dependent association of the following three proteins: the
osmotically inducible CD9
(36),
1-integrin
(37), and the EGFR ligand
HB-EGF. A related phenomenon was observed recently in normal human renal
tissue (32). Moreover,
expression of HB-EGF appears to be itself osmotically regulated
(1,
27). Interestingly, HB-EGF and
CD9 also interact with A disintegrin and metalloprotease domain 10 (also known
as Kuzbanian), and this metalloproteinase may mediate the ectodomain shedding
and autocrine action of EGFR ligands such as HB-EGF
(46). This ternary association
is enhanced by G protein-coupled receptor activation
(46), a phenomenon that has
not been observed in the context of hypertonic stress.
The EGFR ligand conferring transactivation in the present context is
unknown. Potential ligands, including EGF, amphiregulin, epiregulin,
transforming growth factor-
, and HB-EGF, all exist as precursors with a
single membrane-spanning domain and exhibit autocrine or juxtacrine action
after cleavage by one or more metalloproteinases (reviewed in Ref.
33). Our data using the
metalloproteinase inhibitors TAPI and doxycycline support a role for this
mechanism in tonicity-dependent COX-2 regulation, but they do not permit
discrimination among potential EGFR agonists. Neutralizing antibodies specific
for EGFR and EGFR ligands have seen widespread use in other models, but none
are effective in murine systems; we have thus far been unable to identify a
human renal epithelial cell line suitable for studying this phenomenon.
In our studies, the proximal 0.7 kb of the murine COX-2 promoter was
sufficient to confer tonicity responsiveness to a heterologous reporter gene.
Although lacking a canonical tonicity-responsive TonE element, this gene
region contains several cis-acting elements validated in other
contexts. For example, a cAMP-response element (CRE) is instrumental in the
COX-2 response to endotoxin, as are tandem CCAAT/enhancer-binding protein
sites (22,
30,
43); the CRE site, however, is
also responsive to platelet-derived growth factor
(44), in a JNK-dependent
fashion. An NF-B binding site was implicated in the COX-2 response to
tumor necrosis factor-
(45). Although most of these
data were acquired in heterologous expression systems, studies have emerged
using more native models. For example, both in cultured medullary interstitial
cells exposed to hypertonicity
(19) and in cultured thick
ascending limb cells exposed to low-salt or low-chloride conditions
(7), the NF-B site
mediated COX-2 transcriptional induction. Our data clearly support a role for
EGFR kinase signaling in tonicity-dependent upregulation of COX-2 expression
in the well-studied mIMCD3 model; whether this effect requires the CRE [as
might be suggested by earlier data with another agonist of a receptor tyrosine
kinase, PDGF (44)], the
NF-B element [consistent with the interstitial cell and thick ascending
limb models (19)], or still
another element, it clearly occurs in a TonE-independent fashion. Consistent
with these data, EGFR kinase inhibition failed to influence TonE- and
TonEBP-dependent transcription
(52).
In summary, we have shown that the well-described ERK-dependent upregulation in COX-2 expression accompanying hypertonic stress in medullary epithelial cells is mediated via transactivation of the EGFR. The EGFR agonist mediating this effect and the additional upstream activating events remain obscure.
|
DISCLOSURES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked ``advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Kuper, D. Steinert, M.-L. Fraek, F.-X. Beck, and W. Neuhofer EGF receptor signaling is involved in expression of osmoprotective TonEBP target gene aldose reductase under hypertonic conditions Am J Physiol Renal Physiol, May 1, 2009; 296(5): F1100 - F1108. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Steinert, C. Kuper, H. Bartels, F.-X. Beck, and W. Neuhofer PGE2 potentiates tonicity-induced COX-2 expression in renal medullary cells in a positive feedback loop involving EP2-cAMP-PKA signaling Am J Physiol Cell Physiol, January 1, 2009; 296(1): C75 - C87. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Kuper, H. Bartels, M.-L. Fraek, F. X. Beck, and W. Neuhofer Ectodomain shedding of pro-TGF-{alpha} is required for COX-2 induction and cell survival in renal medullary cells exposed to osmotic stress Am J Physiol Cell Physiol, December 1, 2007; 293(6): C1971 - C1982. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. B. Burg, J. D. Ferraris, and N. I. Dmitrieva Cellular Response to Hyperosmotic Stresses Physiol Rev, October 1, 2007; 87(4): 1441 - 1474. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. L. Pucci, S. Endo, T. Nomura, R. Lu, C. Khine, B. S. Chan, Y. Bao, and V. L. Schuster Coordinate control of prostaglandin E2 synthesis and uptake by hyperosmolarity in renal medullary interstitial cells Am J Physiol Renal Physiol, March 1, 2006; 290(3): F641 - F649. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Wu, R. A. Silbajoris, Y. E. Whang, L. M. Graves, P. A. Bromberg, and J. M. Samet p38 and EGF receptor kinase-mediated activation of the phosphatidylinositol 3-kinase/Akt pathway is required for Zn2+-induced cyclooxygenase-2 expression Am J Physiol Lung Cell Mol Physiol, November 1, 2005; 289(5): L883 - L889. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Rojek, J. Nielsen, H. L. Brooks, H. Gong, Y.-H. Kim, T.-H. Kwon, J. Frokiaer, and S. Nielsen Altered expression of selected genes in kidney of rats with lithium-induced NDI Am J Physiol Renal Physiol, June 1, 2005; 288(6): F1276 - F1289. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Pihakaski-Maunsbach, S. Tokonabe, H. Vorum, C. J. Rivard, J. M. Capasso, T. Berl, and A. B. Maunsbach The {gamma}-subunit of Na-K-ATPase is incorporated into plasma membranes of mouse IMCD3 cells in response to hypertonicity Am J Physiol Renal Physiol, April 1, 2005; 288(4): F650 - F657. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Xu, W. Tian, J. N. Lindsley, T. T. Oyama, J. M. Capasso, C. J. Rivard, H. T. Cohen, S. M. Bagnasco, S. Anderson, and D. M. Cohen EphA2: expression in the renal medulla and regulation by hypertonicity and urea stress in vitro and in vivo Am J Physiol Renal Physiol, April 1, 2005; 288(4): F855 - F866. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
).
