EphA2: expression in the renal medulla and regulation by hypertonicity and urea stress in vitro and in vivo

Hongshi Xu, Wei Tian, Jessie N. Lindsley, Terry T. Oyama, Juan M. Capasso, Christopher J. Rivard, Herbert T. Cohen, Serena M. Bagnasco, Sharon Anderson, David M. Cohen


EphA2, a member of the large family of Eph receptor tyrosine kinases, is highly expressed in epithelial tissue and has been implicated in cell-cell and cell-matrix interactions, as well as cell growth and survival. Expression of EphA2 mRNA and protein was markedly upregulated by both hypertonic stress and by elevated urea concentrations in cells derived from the murine inner medullary collecting duct. This upregulation likely required transactivation of the epidermal growth factor (EGF) receptor tyrosine kinase and metalloproteinase-dependent ectodomain cleavage of an EGF receptor ligand, based on pharmacological inhibitor studies. A human EphA2 promoter fragment spanning nucleotides −4030 to +21 relative to the putative EphA2 transcriptional start site was responsive to tonicity but insensitive to urea. A promoter fragment spanning −1890 to +128 recapitulated both tonicity- and urea-dependent upregulation of expression, consistent with transcriptional activation. Neither the bona fide p53 response element at approximately −1.5 kb nor a pair of putative TonE elements at approximately −3 kb conferred the tonicity responsiveness. EphA2 mRNA and protein were expressed at low levels in rat renal cortex but at high levels in the collecting ducts of the renal medulla and papilla. Water deprivation in rats increased EphA2 expression in renal papilla, whereas dietary supplementation with 20% urea increased EphA2 expression in outer medulla. These data indicate that transcription and expression of the EphA2 receptor tyrosine kinase are regulated by tonicity and urea in vitro and suggest that this phenomenon is also operative in vivo. Renal medullary EphA2 expression may represent an adaptive response to medullary hypertonicity or urea exposure.

  • osmotic
  • cell volume regulation
  • receptor tyrosine kinase
  • signal transduction
  • rat
  • kidney

the eph family comprises the largest group of receptor tyrosine kinases in higher eukaryotes. Their well-described role in axon guidance and neuronal migration has expanded dramatically in recent years to encompass diverse aspects of cell-cell interaction including migration, adhesion, and attachment to matrix components (reviewed in Ref. 24). In addition, increasing evidence supports a role for these tyrosine kinases in the regulation of cell survival and growth (reviewed in Ref. 49). Unique among receptor tyrosine kinases, Eph receptor signaling is bidirectional. Information is conveyed in the “forward” direction by regulation of tyrosine phosphorylation of the receptor itself, and in “reverse” fashion through regulated phosphorylation of their cognate ligands, the ephrins (24). Broadly speaking, each of nine EphA receptors may interact with each of six distinct A-class ephrin ligands. Similarly, EphB receptors are activated by B-subclass ephrin ligands. The EphA receptor, EphA2, is expressed in many adult human epithelia, although a role in renal function or physiology has not been postulated.

Anisotonicity regulates expression of a subset of proteins essential for the movement of water, ions, and organic osmolytes into and out of cells (21). In addition to the transport proteins governing these processes, a number of potentially tonicity-responsive signaling intermediates are also regulated at the level of protein expression including the enzyme cyclooxygenase-2 (reviewed in Ref. 56), and the cell surface proteins HB-epidermal growth factor (EGF) (3, 28), integrin β-1 (44), and CD-9 (43). In an unbiased screen for genes regulated by hypertonic stress and urea in cultured epithelial cells derived from the mouse kidney medulla, we observed a marked upregulation in mRNA encoding the receptor tyrosine kinase, EphA2 (47). Nahm et al. (37) in the same cell culture model detected upregulated EphA2 mRNA expression in response to hypertonic NaCl and raffinose but not to urea. We sought to determine the molecular mechanism through which tonicity governed EphA2 expression and to identify regions in the kidney that express EphA2.


General reagents and cell culture.

Cells [murine inner medullary collecting duct (mIMCD3)] were maintained and passaged as previously described (41). Solutes were added as concentrated aliquots to achieve the final osmolality; unless otherwise indicated, solute treatment resulted in the increment of final osmolality by 200 mosmol/kgH2O. To achieve this, the following concentrations were used (in mM): 200 urea, 100 NaCl, and 200 mannitol. For chronic studies, mIMCD3 cells were propagated for many passages in medium supplemented with NaCl to achieve final osmolalities of 600 or 900 mosmol/kgH2O as previously described (7, 8). In all studies, pharmacological inhibitors were generally added 30 min before solute treatment, unless otherwise indicated, and were present for the duration of the solute treatment interval. The following agonists and inhibitors were used (from Calbiochem unless otherwise indicated): 100 nM AG-1478, 100 nM-1 μM AG-1295, 30 nM EGF, 50 μM ilomastat, and 50 μM N-(R)-[2-(hydroxyaminocarbonyl)methyl]-4-methylpentanoyl-l-naphthylalanyl-l-alanine, 2-aminoethyl amide (TNF-α Protease Inhibitor-1 or TAPI-1; Peptides International, Louisville, KY). Ilomastat required DMSO vehicle control at 0.5–2.0% DMSO (vol/vol), which influenced EphA2 expression under some conditions (see results and legends for Figs. 19).

Fig. 1.

Urea and hypertonic stress increase EphA2 mRNA expression. A: RNA was harvested from murine inner medullary collecting duct (mIMCD3) cells receiving control (C) treatment or treatment with urea (U; 200 mM) or NaCl (N; 100 mM) for the indicated interval (6 or 16 h). Top, open arrowhead: EphA2-protected fragment. Bottom, filled arrowhead: protected actin fragment. Left: migration of undigested actin and EphA2 riboprobes (“Probe”). B: graphical depiction of data in A where n = 3 separate experiments. *Pidi* for comparisons with respect to control. †Pidi* with respect to the same time point vs. the other solute (e.g., NaCl at 16 h vs. urea at 16 h).

Transfection and reporter gene assays.

The ∼100-kb genomic clone RP11–276H7 (http://www.chori.org/bacpac/home.htm) containing the promoter of the human EphA2 gene was amplified with the Nucleobond BAC Maxi Kit (Clontech) and then restriction digested with BssHII. The ∼4-kb fragment containing the proximal 5′-flanking sequence was gel-purified and subcloned into MluI-digested, gel-purified pGL3-Basic to generate plasmid −4030-EphA2-Luc. This EphA2 promoter fragment encodes nucleotides −4030 to +21 relative to the putative EphA2 transcriptional start site based on the cDNA clone with the longest 5′-UTR deposited in the public domain (BC037166). Putative TonE elements (YGGAANNNYNY; Ref. 35) were identified at −3115 and −2909 relative to the putative transcriptional start site. A deletion mutant lacking these two elements was prepared by partially digesting the parent vector, −4030-EphA2-Luc, with BsrGI to drop out promoter sequence between −3371 and −2366, and then religating the backbone to create ΔTonE-4053-EphA2-Luc. The EphA2 human genomic DNA fragment subcloned into pGL2-Basic (18), which we refer to as −1890-EphA2-Luc, spans 1,916 bp from −1890 to +128 relative to the putative transcriptional start site based on EphA2 cDNA clone BC037166 and was kindly provided by Dr. X. Chen. The translational start site for this clone is +128.

Initially, cells were transiently transfected with the −4030-EphA2-Luc plasmid; luciferase reporter gene activity was normalized to that of a cotransfected β-galactosidase reporter under the control of the highly active cytomegalovirus long terminal repeat promoter. Changes in response to tonicity were evident but of low amplitude (<100% increase). However, because it is not uncommon for such regulation to be conformation and chromatin dependent (9), a stable transfection approach was adopted. To eliminate inherent bias in integration site of plasmid, we used pooled stable transfectants and noted highly reproducible results, even when such pools were generated de novo for multiple experiments. After lipofection of pGL2- or pGL3-based plasmid in conjunction with pcDNA3.1 (to confer neomycin resistance), pools of stable cell lines were generated under G418 selection pressure and generally achieved confluence after ∼2 wk. Morphologically, the pooled clones appeared homogeneous. We found pools such as these to exhibit a stable and reproducible phenotype and expression level for at least several subsequent passages. There is a consensus p53 element (cacCatGttggccaggCatGtct; nucleotides critical to the binding site are capitalized) in the human EphA2 promoter beginning approximately −1556 relative to the putative transcriptional start site of BC037166 (18). For the p53 promoter mutant, Δp53-EphA2-Luc, the p53 consensus, was subjected to site-directed mutagenesis (QuikChange; Stratagene) to cacAatTttggccaggAatTtct using the following pair of oligonucleotides (in 5′ to 3′ orientation): forward primer EphA2-p53–5′ (gtagagacggggtttcacAatTttggccaggAatTtctggagctcctgacc); reverse oligo EphA2-p53–3′ (ggtcaggagctccagaAatTcctggccaaAatTgtgaaaccccgtctctac). Stable transfectants were initially generated in the mIMCD3 cell line. Corroboratory experiments were performed in the human embryonic kidney (HEK) cell line to confirm the utility of the human promoter in the context of expression in a human cell line (see results), as there are no widely used renal medullary epithelial continuous cell lines of human origin. In addition, stable transfectants were also made using the BGT-2X-Luc luciferase reporter gene (55) under control of two tandem repeats of a canonical TonE element to test the tonicity responsiveness of the TonE element in the context of stable transfection.

RNase protection assay and immunoblotting.

For RNase protection assay using murine cell lines, riboprobe was reverse-transcribed from sequence-verified I.M.A.G.E. clone 459170 (Research Genetics) encoding partial cDNA for murine EphA2. For RNase protection assay using rat kidney tissues, a 275-bp fragment of rat EphA2 was PCR amplified from reverse-transcribed cDNA prepared from rat kidney mRNA using the following primer pair (written in 5′ to 3′ orientation, and derived from accession XM_238428): forward primer EphA2-Rn-3–5′ (gggacctgatgcaaaacatc) and EphA2-Rn-278–3′ (tgtcgatcttggtgaactgg). For murine and rat experiments, biotinylated riboprobe was prepared from this partial cDNA and used for RNase protection assay with RNA harvested from rat tissue. For RNA preparation, rat kidney was grossly dissected into cortex, medulla, and papilla, and then snap-frozen in liquid nitrogen. RNA and protein extracts were isolated in parallel using TRIzol reagent in accordance with the manufacturer's instructions (Life Technologies). Protein extracts were used for immunoblotting with commercial polyclonal rabbit anti-EphA2 antibody (Santa Cruz Biotechnology) at 1:1,000 dilution as previously described. Secondary antibody was goat anti-rabbit-horseradish peroxidase at 1:4,000 and visualization was via Chemiluminescence Plus reagent (Perkin-Elmer Life Science). For cultured cells, lysates were prepared with cell culture lysis buffer (20 mM Tris, pH 7.4, 1% Triton X-100, 10% glycerol, 137 mM NaCl, 2 mM EDTA, 25 mM β-glycerophosphate, 1 mM orthovanadate, 2 mM pyrophosphate, 25 μg/ml AEBSF, and 10 μg/ml leupeptin) and ∼20 μg were resolved via SDS-PAGE. For demonstration of tyrosine phosphorylation of EphA2, lysates were prepared using cell culture lysis buffer (above) and were subjected to immunoprecipitation with anti-EphA2 antibody and Protein A/Protein G agarose beads (Pharmacia). Immunoprecipitates were washed extensively with lysis buffer and then boiled in 1× Laemmli loading buffer before loading for SDS-PAGE. After transfer, the PVDF membrane was immunoblotted with anti-PY99 antibody recognizing phosphotyrosine. For RNase protection assay and immunoblotting, data depicted in Figs. 19 are representative of three separate experiments, except where explicity noted.

Immunohistochemistry and in vivo studies.

Kidneys were excised from male Sprague-Dawley rats (∼300 g; Harlan), sliced into three to four fragments in transverse fashion (with one cut passing through the papilla), and immersed in 10% formalin. All studies were approved by the Portland V.A. Institutional Animal Care and Use Subcommittee. Fixed kidneys were dehydrated through a graded series of ethanols, embedded in paraffin, sectioned at 4-μm thickness, and placed onto glass slides (Portland Tissue Processing, Portland, OR). For immunohistochemical demonstration of EphA2, the protocol of Komers et al. (29) was followed with minor modification. Rabbit polyclonal anti-human EphA2 antibody “C20” proved superior to “H77” (both antibodies from Santa Cruz Laboratories), although both antibodies produced presumably nonspecific staining of the kidney cortex, which we have not observed with antibodies directed against other antigens under identical conditions (e.g., Ref. 50). Sections were deparaffinized in xylene, rehydrated through graded ethanols to water, and pretreated by steaming in CITRA buffer (BioGenex, San Ramon, CA). After being treated with protein-blocking solution, the slides were incubated overnight with primary antibody (diluted 1:200) at 4°C. Endogenous peroxidase activity was blocked with 3% H2O2 solution in methanol. The first antibody was localized by using the Vectastain ABC-Elite peroxidase detection system (Vector Laboratories, Burlingame, CA). This was followed by reaction with diaminobenzidine (Dosindo, Japan) as chromogen and counterstaining with hematoxylin (Sigma). For water deprivation studies, male Sprague-Dawley rats (∼300 g body wt) were provided ad libitum access to water (control group) or were water deprived for 48 h (water deprivation group). Body weights were recorded daily. After 48 h, three control rats and four water-deprived rats were euthanized and kidneys were excised. Excised kidneys were sagitally divided and cortex and papilla were rapidly dissected and snap-frozen in liquid nitrogen, and then processed with TRIzol reagent for extraction of RNA and protein. Because of the exceedingly small amount of tissue provided by each renal papilla, specimens from all animals in each treatment group were pooled before processing. Serum [Na+] was measured via flame photometry, and urine osmolarity was measured via freezing point depression (Advanced Micro Osmometer, model 330, INC Lab Products) in samples diluted 10-fold. For urea feeding experiment, male Sprague-Dawley rats (∼150–200 g) received standard diet treatment (n = 4) or diet supplemented with 20% urea (n = 4) for 12 days. Urine urea concentration was measured using the Infinity BUN reagent (Sigma) and blood urea nitrogen was measured with urea nitrogen kit (535-B; Sigma) according to the manufacturer's directions.

Statistical analysis.

Statistical analysis was performed using Student's t-test (Microsoft Excel); data are reported as means ± SE. Statistical significance is assigned to P < 0.05. For experiments involving multiple comparisons (e.g., Fig. 1), the false discovery rate procedure (15) has been applied and statistical significance is ascribed to comparisons wherein Pidi* in accordance with published guidelines for this journal (16).


EphA2 expression is upregulated by hypertonic and urea stress.

We and others (37, 47) noted that EphA2 was among the genes most highly inducible by both hypertonic stress and elevated urea concentrations (10- and 13-fold, respectively, Ref. 47). As both of these conditions can be found in the medulla of the mammalian kidney in vivo, we investigated this phenomenon to provide insight into regulation of genes by these two stimuli. Consistent with our prior findings, elevated concentrations of NaCl and urea both increased expression of EphA2 at the mRNA level, as assessed via RNase protection assay (Fig. 1, A and B). The effect of both solutes was evident at 6 h, but only urea increased expression at 16 h. Specifically, at the 6-h time point, urea and NaCl increased EphA2 mRNA expression by 3.7 ± 0.2-fold (P = 0.00004) and 5.1 ± 1.3-fold (P = 0.02), respectively. At 16 h, urea increased expression 2.8 ± 0.5-fold (P = 0.01); the effect of NaCl was not significantly different from control at this time point and was significantly less than that of urea (P = 0.01).

In their screen of multiple genes, Nahm et al. (37) noted that the increase in EphA2 mRNA expression in mIMCD3 cells in response to hypertonic stress was potentiated by the inhibitor of protein synthesis, cycloheximide. Protein abundance was not addressed in their study. Because hypertonic stressors may inhibit protein synthesis and thereby lead to “superinduction” of mRNAs, particularly those encoding immediate early genes (Ref. 13 and references therein), the effect of these solute stimuli on EphA2 expression at the protein level was examined in parallel. We first examined a panel of renal cell lines. EphA2 was expressed at low levels in the human embryonal HEK-293 cell line, but at higher levels in murine distal convoluted tubule (20) and terminal mIMCD3 (41) cell lines (Fig. 2A). The mIMCD3 line was used for further study. Consistent with the mRNA data presented in Fig. 1, a variety of solute stressors rapidly and dramatically increased EphA2 immunoreactivity (Fig. 2B). Urea treatment (200 mM) produced a sustained and time-dependent increase in expression beyond 40 h. The hypertonic stressors, NaCl and mannitol, in contrast, affected dramatic but relatively short-lived upregulations in EphA2 abundance, which returned to pretreatment levels by 16 h. Coapplication of urea and NaCl, which abrogated tonicity-dependent gene regulation in some contexts (46, 47), failed to inhibit hypertonic induction of EphA2 expression. The peptide growth factor, EGF, exerted only a modest effect on EphA2 expression. Therefore, of the stimuli tested, only urea induced a marked and protracted increase in EphA2 protein expression.

Fig. 2.

Multiple solute stressors increase EphA2 protein expression. A: anti-EphA2 immunoblot of detergent lysates prepared from the indicated renally derived cell lines. B: mIMCD3 cells treated for the indicated intervals (in h) with the indicated solutes or agonists (Mann, mannitol; EGF, 30 nM epidermal growth factor; U+N, 100 mM NaCl following 30-min pretreatment with 200 mM urea). Left: migration of molecular mass standards (in kDa). HEK, human embryonic kidney cells; mDCT, murine distal convoluted tubule.

Because transactivation of the EGF receptor is likely pivotal to both urea- and tonicity-inducible signal transduction (27, 42, 47, 56, 57), we assessed the effect of inhibition of this tyrosine kinase on solute-dependent EphA2 protein expression. Whereas the tyrphostin-based receptor tyrosine kinase inhibitor AG-1295, which exhibits specificity for the PDGF receptor kinase, failed to influence EphA2 expression in response to either urea or hypertonic stress (Fig. 3A), the highly EGF receptor-specific inhibitor AG-1478 inhibited both (Fig. 3A). When quantified explicitly (n = 3–4 per treatment condition), AG-1478 decreased urea-inducible EphA2 expression from 1.77 ± 0.10 to 1.06 ± 0.04 normalized arbitrary units (P = 0.0005; Fig. 3C) and decreased NaCl-inducible EphA2 expression from 2.38 ± 0.11 to 1.90 ± 0.18 units (P = 0.02). These data were consistent with the effect of AG-1478 on EphA2 expression at the mRNA level (data not shown). AG-1478 was more effective at inhibiting urea- rather than NaCl-inducible EphA2 expression (Fig. 3, A and C).

Fig. 3.

Tonicity-dependent EphA2 expression is AG-1478 and metalloproteinase sensitive. Anti-EphA2 immunoblot of detergent lysates prepared from mIMCD3 cells receiving control treatment (C) or treatment with urea (U; 200 mM) or NaCl (N; 100 mM) for 6 h, following pretreatment with DMSO vehicle, the EGF receptor kinase inhibitor AG-1478, the PDGF receptor kinase inhibitor AG-1295 (A), the metalloproteinase inhibitors ilomastat (50 μM) or TAPI-1 (50 μM), or DMSO vehicle (B). (DMSO vehicle was applied at 10-fold higher concentration in B than in A, owing to the more limited solubility of ilomastat.) C: data from 3–4 separate immunoblots (3 for each treatment condition and 4 in the absence of pretreatment) derived densitometrically. *P < 0.05 with respect to control treatment in the absence of pretreatment (none). †P < 0.05 with respect to no pretreatment within a given treatment group (e.g., within the group of bars labeled “+ Urea”).

EGF receptor transactivation in the setting of urea stress and anisotonicity (57), as in other models (40), requires ectodomain cleavage followed by autocrine or juxtacrine action of EGF or a related EGF receptor ligand. This phenomenon may be interrupted by pharmacological inhibition of metalloproteinases. Consistent with such a model, the metalloproteinase inhibitors ilomastat and TAPI-1 blocked the effect of hypertonic NaCl on EphA2 expression (Fig. 3B). Specifically, ilomastat decreased urea-inducible EphA2 expression from 1.77 ± 0.10 to 1.19 ± 0.27 units (P = 0.03; Fig. 3C) and decreased NaCl-inducible EphA2 expression from 2.38 ± 0.11 to 1.59 ± 0.10 units (P = 0.001; Fig. 3C); however, the effect of ilomastat in the context of urea could not be adequately addressed because of the partial inhibitory effect of DMSO vehicle alone vis-à-vis urea treatment (Fig. 3B). Therefore, it can only be concluded that metalloproteinase inhibitors block EphA2 induction in response to hypertonic NaCl.

Although it does not require tyrosine phosphorylation, activation of the EphA2 receptor may be associated with either tyrosine dephosphorylation or tyrosine phosphorylation. When immunoprecipitated EphA2 receptor was immunoblotted with anti-phosphotyrosine antibody, sustained dephosphorylation was observed in the setting of acute hypertonic NaCl treatment; the effect of urea was much more modest (Fig. 4A). There was no change in the level of total EphA2 expression over this brief interval. To assess the effect of more long-term exposure to hypertonicity, mIMCD3 cells that had been chronically adapted to either 600 or 900 mosmol/kgH2O (supplemented with either 150 or 300 mM NaCl, respectively, Refs. 7, 8) were examined (Fig. 4B). Of note, the effect of urea could not be examined as these cells did not tolerate such extremes of urea exposure or multiple passages (8). Cells chronically adapted to 600 or 900 mosmol/kgH2O displayed only a modest increase in EphA2 expression (∼40 and ∼80%, respectively, n = 2; Fig. 4B). EphA2 tyrosine phosphorylation, however, was dramatically increased at both 600 and 900 mosmol/kgH2O (6- and 8-fold, respectively, n = 2). Under these extreme conditions, EphA2 migrates as a doublet on SDS-PAGE (Fig. 4B), which we suspect is a consequence of immunodetection of both the phosphorylated and unphosphorylated forms. Of note, an EphA2 coimmunoprecipitating protein migrating at >200 kDa and seen only in the chronically adapted cells may represent an associated phosphoprotein. These data indicate that acute hypertonic stress is associated with a decrease in EphA2 tyrosine phosphorylation, whereas a more chronic increase is associated with a pronounced increase in tyrosine phosphorylation.

Fig. 4.

Hypertonic stress is associated with regulation of tyrosine phosphorylation of EphA2. A: acute hypertonic stress. Anti-phosphotyrosine (top) or anti-EphA2 (bottom) immunoblots of anti-EphA2 immunoprecipitates prepared from detergent lysates of mIMCD3 cells treated for the indicated interval (in min) with urea (200 mM) or NaCl (100 mM). B: chronic hypertonic stress. Anti-phosphotyrosine (top) or anti-EphA2 (bottom) immunoblots of anti-EphA2 immunoprecipitates prepared from detergent lysates of control-treated mIMCD3 cells (300) or from mIMCD3 cells that had been chronically adapted to final osmolality of 600 mosmol/kgH2O (600, medium supplemented with 150 mM NaCl) or 900 mosmol/kgH2O (900, medium supplemented with 300 mM NaCl). EphA2 migrates as a doublet with the top most likely representing the phosphorylated form; the identity of the phosphoprotein migrating at >200 kDa is unknown. Duplicate samples are shown for B. Exposures (ECL) were selected to highlight decrease (A) or increase (B) in tyrosine phosphorylation.

EphA2 is likely regulated by hypertonic stress at the level of transcription.

We investigated the mechanism of upregulation of EphA2 expression at the mRNA level. We excised ∼4 kb of the human EphA2 promoter (−4030 to +21, relative to the putative transcriptional start site; see methods) and subcloned it into the luciferase reporter vector, pGL3-Basic, to create −4030-EphA2-Luc. Our preliminary sequence analysis revealed two canonical tonicity-responsive enhancer (TonE) elements at ∼3,000 bp upstream of the putative transcriptional start site in this clone (Fig. 5A). Because we hypothesized that these sites might be essential for gene induction by tonicity, we made a second construct in which these sequences were deleted (ΔTonE-4030-EphA2-Luc). On stable transfection into mIMCD3 cells, pronounced constitutive reporter gene activity was observed. Hypertonic NaCl (100 mM × 6 h) increased this reporter gene activity more than twofold (120 ± 9%, P = 0.00008), although urea stress (200 mM × 6 h) exerted no effect (Fig. 5B). Importantly, deletion of the ∼1 kb of the EphA2 5′-flanking sequence containing the two putative TonE elements (Fig. 5A) did not blunt the response to hypertonic stress (Fig. 5B). In this context, reporter gene activity was increased 190 ± 60% (P = 0.02). Because of this unexpected result, we sought to confirm that TonE elements were operative (i.e., accessible) in the context of stable transfection. The mIMCD3 cell line was stably transfected with a luciferase reporter gene driven by two tandem repeats of the well-studied TonE element (35) from the tonicity-responsive BGT1 (betaine/gamma-aminobutyric acid transporter) gene. This plasmid was called BGT-2X-Luc and had been used previously in similar studies (55). Six-hour treatment with hypertonic stress (100 mM NaCl) increased reporter gene activity 910 ± 60% (n = 3; data not shown), consistent with the effect of this plasmid in transient transfection (e.g., Ref. 55). These data indicated that although classical TonE was active in the context of stable transfection in this cell culture model, the putative TonE elements in the EphA2 proximal promoter did not account for the hypertonicity responsiveness in this context.

Fig. 5.

Hypertonic stress increases EphA2 promoter activity independent of TonE and p53 response element. A: diagram of EphA2 human genomic sequence included in each EphA2 luciferase reporter gene construct (x-axis depicts scale in nucleotides where 0 represents putative EphA2 transcriptional start site). Gray shading represents 5′-flanking sequence; black shading represents mRNA sequence (i.e., downstream of transcriptional start site). The location of 2 canonical putative TonE elements is shown in white boxes; these were deleted en bloc from ΔTonE-4030-EphA2-Luc. B: luciferase reporter gene activity (expressed relative to control treatment) in mIMCD3 cells stably transfected with the indicated plasmids, following control treatment or treatment with urea (200 mM) or NaCl (100 mM) for 6 h. Data are not normalized to coexpressed β-galactosidase (see results). The x-axis is interrupted to indicate that, although the same scale is used for these data, −4030-EphA2-Luc and ΔTonE-4030-EphA2-Luc plasmids were based on parent vector pGL3-Basic, whereas −1890-EphA2-Luc was based on pGL2-Basic. C: luciferase reporter gene activity in mIMCD3 cells stably transfected with the indicated plasmids (−1890-EphA2-Luc or Δp53–1890-EphA2-Luc), following control treatment or treatment with NaCl (100 mM) for 6 h. Data are normalized to coexpressed β-galactosidase (see results) and then to −1890-EphA2-Luc under control conditions. Of note, fold-induction of the −1890-EphA2-Luc vector is unaffected by normalization for β-galactosidase. B: data are depicted as arithmetic means ± SE of 3 independent experiments. *Statistical significance with respect to control treatment for each plasmid. Individual P values are provided in the text. C: data from 1 of 2 independent experiments.

Because our EphA2 promoter construct contained sequence extending only 21 bp downstream of the putative transcriptional start site (Fig. 5A), we investigated another fragment of human EphA2 5′-flanking sequence. For these studies, we used the proximal 1.9 kb of the human EphA2 promoter subcloned into the luciferase reporter vector, pGL2-Basic (18). This fragment included sequence information extending down to about +128 relative to the putative transcriptional start site (ending immediately upstream of the translational start site) but lacks ∼2 kb of extreme upstream sequence present in −4030-EphA2-Luc (Fig. 5A). mIMCD3 cells were stably transfected with this plasmid, referred to here as −1890-EphA2-Luc. The effect of hypertonic NaCl on this promoter fragment was more pronounced (600 ± 130% increase, P = 0.005; Fig. 5B). Importantly, urea treatment also increased transcription from this vector (200 ± 40% increase, P = 0.005; Fig. 5B). Therefore, the −1890-EphA2-Luc more closely recapitulated the regulation of EphA2 mRNA and protein by both of these medullary solutes.

Of note, transcription data in Fig. 5B were not normalized with respect to cotransfected β-galactosidase. Similar studies performed with normalization of data to coexpressed β-galactosidase were identical with respect to the effect of hypertonic NaCl; however, urea treatment upregulated transcription of β-galactosidase in these studies, obscuring the effect on the EphA2-Luc vectors. The promoter conferring constitutive expression of transfected β-galactosidase in these studies was the cytomegalovirus long terminal repeat; in preliminary studies, we were unable to discern any appreciable similarity between this sequence and the urea-responsive −1890 EphA2 promoter (data not shown). Importantly, urea is not a nonspecific activator of transcription because only a small subset of gene products (fewer than 1%) are upregulated in response to urea treatment in this mIMCD3 cell model (47). To eliminate this potential confounder, subsequent studies focused on the −1890-EphA2-Luc construct and the effect of only hypertonic NaCl.

To establish that this effect of hypertonicity on the human EphA2 promoter was also evident in the native context of a human renal cell line, human embryonal HEK cells (of undifferentiated renal origin but expressing endogenous EphA2; see Fig. 2A) were stably transfected with the −1890-EphA2-Luc vector. Hypertonic NaCl (100 mM NaCl × 6 h) increased reporter gene activity greater than twofold (120 ± 10%, n = 3, data not shown), confirming the effect of this promoter in the context of a human cell line. We continued to employ the murine mIMCD3 cell line for subsequent studies because of the lack of a widely used human renal medullary cell line.

In addition to the two putative TonE consensus elements (Fig. 5A), a number of cis-acting elements have been reported in the more proximal EphA2 promoter, including a p53 motif ∼1,550 bp upstream of the transcriptional start site with confirmed activity in other contexts (18). Because p53 signaling has been implicated as an effector arm of hypertonic stress signaling (17), we investigated the role of this cis element in the EphA2 transcriptional response to solute stress. Site-directed mutagenesis was used to inactivate the p53 motif (see methods). Tonicity responsiveness of the mutant promoter, however, was unaffected (Fig. 5C), indicating that the p53 cis element was dispensable for tonicity signaling to EphA2 transcription. Of note, data using these stable transfectants were normalized to cotransfected β-galactosidase and were indistinguishable from the unnormalized data (Fig. 5B; −1890-EphA2-Luc); NaCl treatment failed to influence transcription of stably transfected β-galactosidase.

EphA2 is highly expressed in the medulla in vivo.

Cells of the renal medulla are constitutively exposed to elevated concentrations of urea and NaCl in vivo. We hypothesized that renal medullary EphA2 expression would exceed that of the cortex. Via RNase protection assay, EphA2 expression was detectable in rat renal cortical RNA (Fig. 6A). EphA2 mRNA expression in the medulla was far greater than that of the cortex. Papillary expression was intermediate between these values. Actin mRNA expression did not differ among these tissues (data not shown). We sought to confirm this finding at the protein level. The profile of rat renal EphA2 protein expression mirrored that of EphA2 RNA expression; abundance was greatest in the medulla, followed by papilla, and cortex (Fig. 6B).

Fig. 6.

EphA2 is highly expressed in rat renal medulla and papilla in vivo. A: EphA2 RNAse protection assay using RNA prepared from rat kidney tissues. Top, closed arrowhead: incompletely digested EphA2 riboprobe. Bottom, open arrowhead: EphA2-protected fragment. B: anti-EphA2 immunoblot prepared from mechanical lysates of rat renal tissues. Open arrowhead depicts EphA2.

Immunohistochemistry was then used to examine EphA2 expression in rat kidney. Anti-EphA2 immunoreactivity was abundant in inner medullary and papillary collecting duct (Fig. 7, A and C). Cortical expression was difficult to evaluate reliably because relatively high background was encountered with both specific antibody and control serum. Interestingly, EphA2 was highly expressed in the transitional epithelium lining the renal pelvis (Fig. 7E).

Fig. 7.

EphA2 is expressed in rat renal medulla, papilla, and transitional epithelium. Immunohistochemical analysis of EphA2 expression (A, C, E) vs. immunoreactivity of control serum (B, D, F) in formalin-fixed rat kidney medulla (A, B), papillary tip (C, D), and renal pelvis (E, F). Arrowheads denote medullary collecting duct (A), papillary collecting duct (C), and transitional epithelium (E).

EphA2 is physiologically regulated by osmotic and/or urea stress in vivo.

Because EphA2 is highly expressed in the most osmotically challenging renal milieu, and because it is regulated by hypertonic stress and urea stress in renal epithelial cells in vitro, we examined the effect of increasing medullary tonicity on EphA2 expression. Rats received either control treatment or water deprivation for 48 h. Water deprivation caused an 11 ± 0.3% decrement in body weight (vs. weight gain of 4 ± 0.7% in controls, P = 2 × 10−6; Fig. 8A), an increment in plasma [Na+] from 140 ± 0.5 to 148 ± 0.9 meq/l (P = 0.0006; Fig. 8B), and an increase in urine osmolality from 1,200 ± 300 to 2,500 ± 100 mosmol/kgH2O (P = 0.003; Fig. 8C). Via RNase protection assay, mRNA encoding EphA2 was markedly increased in the papillae pooled from the osmotically stressed rats but not in cortical tissue pooled from these same animals (Fig. 8D). In corresponding fashion, EphA2 protein expression was also increased in papilla only (Fig. 8E). Owing to the exceedingly small amount of RNA recovered from individual rat papillae, and the relatively large amount of RNA required for RNAse protection assay, samples were pooled before this analysis. Only cortex and papilla were studied because these represented the tissue extremes of osmoregulation in vivo in response to water deprivation.

Fig. 8.

Renal EphA2 expression is regulated in vivo by water restriction. Male Sprague-Dawley rats (∼300 g) received control treatment (n = 3) or water deprivation for 48 h (n = 4). Water deprivation resulted in ∼11% decrement in body weight (A) and ∼7.5 meq/l increment in plasma [Na+] (B). Via RNase protection assay (C), EphA2 mRNA expression was upregulated in renal papillae but not cortices of water-deprived (D) rats relative to control-treated rats (C). Uosm, urinary osmolality. In D, the first and last gel lanes contain digested (P*) and undigested (P) riboprobe controls, respectively, in the absence of mRNA hybridization. Open arrowhead denotes protected fragment corresponding to EphA2. Via immunoblotting, EphA2 was correspondingly upregulated at the protein level in papillae but not cortices harvested from water-deprived rats (E). Filled arrowhead denotes EphA2. *P < 0.05 with respect to control treatment.

In a related model, rats were fed conventional diet or chow supplemented with 20% urea for 12 days. This regimen increases serum urea concentration and renal medullary parenchymal urea concentration without dramatically affecting urine urea concentration (26). Accordingly, urea supplementation increased serum urea concentration (20.4 ± 1.0 vs. 6.1 ± 0.4 mM, P = 0.00004; Fig. 9A) but did not increase urine urea concentration (969 ± 20 vs. 906 ± 111 mM, P = 0.3; data not shown). As expected, urea feeding induced an osmotic diuresis (urine volume 110 ± 6 vs. 17.5 ± 0.9 ml/day, P = 0.000003; Fig. 9B). Consistent with its action as an osmotic diuretic, urea feeding decreased urine concentration (1,200 ± 40 vs. 2,120 ± 150 mosmol/kgH2O, P = 0.0004) but dramatically increased daily urinary urea excretion (107 ± 4 vs. 16.0 ± 2.3 mmol/day, P = 10−6; Fig. 9C) owing to the marked increase in urine volume. In response to urea treatment, EphA2 protein expression was markedly increased in the outer medulla (P = 0.0004), whereas no change was detected in inner medulla and papilla (P = 0.06 and 0.3, respectively; Fig. 9, D and E). In conjunction with data from the water deprivation model, these data are consistent with regulation of EphA2 expression by hypertonic and urea stress in vivo.

Fig. 9.

Renal EphA2 expression is regulated in vivo by urea feeding. Male Sprague-Dawley rats (∼150–200 g) received standard diet treatment (n = 4) or diet supplemented with 20% urea (n = 4). Serum urea concentration was markedly increased by urea feeding (A), as was daily urine volume (B) and total daily urinary urea excretion (C). In response to urea feeding, EphA2 protein expression increased in outer medulla but not in inner medulla or papilla (D). These data are quantified in E. *P < 0.05 with respect to control treatment.


We show that EphA2 is upregulated at the level of mRNA and protein expression by two environmental stressors present in the renal medulla and papilla. This effect appears to be mediated, at least in part, through increased transcription based on reporter gene studies. Consistent with these data, EphA2 is expressed most abundantly in the renal tissues exposed to elevated ambient concentrations of NaCl and urea. Expression in papilla might be expected to exceed that in medulla, based on the renal corticomedullary concentration gradient. This is perhaps suggested by the immunohistochemistry data (Fig. 7), where expression is primarily in the collecting duct, but is not consistent with the immunoblotting data (Fig. 6). It is possible that the EphA2 antibody exhibits incomplete specificity under either immunohistochemistry or immunoblotting conditions. RNase protection assay, which affords complete specificity but detects EphA2 mRNA and not protein, was consistent with more abundant EphA2 expression in the medulla. This approach may also fail to detect an EphA2-related splice variant, although similar findings were observed with multiple riboprobes (data not shown). Data obtained with water-deprived rats corroborated these findings. The increased papillary tonicity and urea concentration associated with the water-deprived state were accompanied by increased EphA2 mRNA and protein expression. Whereas these data do not confirm transcriptional regulation of EphA2 expression in vivo in response to hypertonic or urea stress, they are entirely consistent with this model.

The regulation of EphA2 expression in the model of urea feeding also warrants careful interpretation. In a model that closely paralleled the urea feeding studies presented here, Kim et al. (26) measured tissue urea content in renal papilla, inner medulla, and outer medulla from control rats and rats fed a diet supplemented with 20% urea. They noted a doubling of tissue urea content in all zones; however, the increment in urea content varied markedly, increasing by ∼100 μmol/g dry tissue wt in the outer medulla and by ∼340 μmol/g dry tissue wt in the papilla. With other urea-dependent transcriptional phenomena, there was a urea concentration above which enhanced transcription was not seen and even decreased transcription was observed (12). The in vivo regulation of EphA2 expression in the urea-feeding context may reflect this “window” of urea concentration required for upregulation, and the criteria for enhanced transcription may only be fulfilled in the outer medullary milieu. For example, urea feeding obligates an osmotic diuresis (Fig. 9B) and the attendant polyuria-driven decrease in overall inner medullary and papillary tonicity may partially offset the isolated effect of increasing urea concentration in these tissues. EphA2 expression in each kidney zone in vivo only reflects the “net” effect of urea feeding.

Much recent attention has focused on the molecular mechanism through which tonicity regulates gene transcription and the role of the TonEBP/NFAT5 transcription factor in this process (21). The proximal promoter of the human EphA2 gene contains two canonical TonE elements with which TonEBP might be expected to interact. The EphA2 promoter sequence has not been cloned from a range of species, however, so the degree of conservation, and hence potential functional significance, of these sites remains unproven. Unexpectedly, deletion of these elements from the human EphA2 promoter failed to influence tonicity-dependent EphA2 reporter gene activity.

A bona fide p53 response element was identified in the EphA2 promoter and its function was confirmed in at least one experimental context (18). Because p53 has been implicated in the cell response to hypertonic stress (17), we investigated the role of this element in tonicity-dependent EphA2 promoter activity. On stable transfection into mIMCD3 cells, a robust tonicity-dependent increment in reporter gene activity was evident; however, site-directed mutagenesis of the p53 site to a nonfunctional mutant failed to influence the effect of tonicity, suggesting a different mechanism of regulation. Further supporting a p53-independent mechanism of EphA2 transcription in response to hypertonicity, DNA damage sufficient to induce p53 upregulation increased EphA2 tyrosine phosphorylation (18), rather than diminishing it as in the present study.

In addition to the well-characterized TonE/TonEBP interaction and the effect of p53, other transcription factors (and their cognate cis elements) are potentially responsive to elevated ambient solute concentrations including HSF (2, 10), CREB (6), AP-1 (51), STAT (5, 11), NF-κB (22, 38), Sp1 (4), SRE/Ets (12), as well as possible novel elements (48). Which, if any, of these elements confers tonicity responsiveness to the EphA2 promoter remains to be established.

There appears to be a distinction between the effects of hypertonic stress, per se, and urea stress on EphA2 expression and reporter gene activity. At the level of EphA2 mRNA expression, the urea effect is more protracted and more robust than that of NaCl treatment and is more sensitive to EGF receptor inhibition. At the protein level, the effect of urea is again more prolonged, akin to the effect of EGF treatment, and is again more sensitive to the inhibitory effect of AG-1478. The consequence of metalloproteinase inhibition, in contrast, could not be compared directly because of the inhibitory effect of DMSO vehicle alone with respect to the urea effect. The proximal ∼4 kb of EphA2 5′-flanking sequence recapitulated tonicity responsiveness in a luciferase reporter vector but did not confer a response to urea. Interestingly, a smaller (∼2 kb) piece of 5′-flanking sequence, in conjunction with an additional 100 bp of sequence downstream of the putative EphA2 transcriptional start site, exhibited both urea and NaCl responsiveness. From these data, we conclude that the transcriptional responses to urea and NaCl are likely mediated via distinct cis-acting elements and by distinct but perhaps overlapping signaling events. We have begun investigating whether this transcriptional regulation is conferred via upstream or downstream sequences in this reporter plasmid. However, the relatively modest increase in EphA2 transcription by urea or NaCl in vitro (i.e., <10-fold), although entirely consistent with physiological significance in vivo, makes dissection of the operative promoter elements extremely challenging.

The EphA2 receptor has been implicated in the regulation of cell growth, survival, migration, and angiogenesis (reviewed in Ref. 49). Interestingly, whereas most Eph family receptors are subject to developmental regulation, EphA2 is found in adult tissues where its expression appears to be restricted to epithelia. Aberrant EphA2 expression and regulation have been demonstrated in numerous malignant tissues as well as cell culture models of malignant transformation (reviewed in Ref. 49). To the best of our knowledge, EphA2 expression in the kidney has not been described. There is, however, a precedent for Eph receptor/ephrin signaling and water movement. EphB2 and EphB3 are required for normal function of the vestibular system (14). In this system, aquaporin-1 associates with EphB2, perhaps via an intervening adapter protein in a PDZ domain-dependent fashion; in addition, these two proteins colocalize in murine proximal tubules (14). Cowan et al. (14) speculate that these proteins may associate with other PDZ domain-containing transport proteins including anion exchanger-1 and -2 in a macromolecular signaling complex. In similar fashion, we speculate that EphA2 may play a role in indirectly regulating water or ion fluxes in response to anisotonicity in the kidney medulla and papilla. Alternatively, as a receptor tyrosine kinase, it may promote renal epithelial cell survival in the harsh medullary environment.

The relationship between EphA2 phosphorylation and activation is complex. For most Eph receptors, tyrosine (auto)kinase activity is only manifest following ligand binding. The EphA2 receptor, in contrast, is constitutively active and does not require ligand binding (52, 53). Consistent with this model, and unlike all other Eph receptors, EphA2 does not require tyrosine phosphorylation for activation. Ligand binding, however, will induce tyrosine phosphorylation. It was earlier reported that activation of EphA2 was associated with dephosphorylation of EphA2 tyrosine residues. We show here that upregulation of EphA2 expression in response to acute hypertonicity is associated with dephosphorylation. Perhaps paradoxically, chronic exposure to hypertonicity was associated with a dramatic increase in EphA2 tyrosine phosphorylation. It is possible to reconcile these disparate findings in the context of ligand engagement of EphA2 inducing tyrosine phosphorylation of this receptor. Specifically, ligand-receptor interactions may be disrupted by the acute decrement in cell volume (hence the greater effect of NaCl relative to urea) and these interactions may be adaptively reinforced under conditions of chronic hypertonic stress wherein cell volume has been restored. The phosphatase responsible for EphA2-regulated dephosphorylation, LMW-PTP, appears to constitutively interact with the receptor. Interestingly, overexpression of this phosphatase alone results in dephosphorylation of EphA2, upregulation of EphA2 expression, and assumption of the malignant phenotype (25, 53). A second phosphatase, SHP2, also interacts with EphA2 in a regulated fashion. Within minutes of EphA2 binding to its ligand, ephrinA1, SHP2 is recruited and likely mediates the dephosphorylation that follows ligand-dependent EphA2 tyrosine phosphorylation (33).

Several effectors of EphA2 have been identified and a role in tonicity signaling can readily be envisioned for each. EphA2 has been described as both a positive (39) and negative (34) regulator of ERK MAPKs. ERK is clearly an effector of both hypertonic and urea stress, where its activation is likely dependent on transactivation of the EGF receptor tyrosine kinase (57). There is also a complex interaction between EphA2 and Ras signaling. In a Ras-transformed mammary epithelial cell model, and in mammary tumors of mice transgenic for H-Ras, EphA2 expression is upregulated (53). Somewhat paradoxically, EphA2 activation also appears to inhibit Ras activation (34). We showed that hypertonic stress and urea stress increase Ras activity in renal epithelial cells (45).

EphA2 activation has also been shown to suppress function of integrins, the principal receptors for extracellular matrix components, and to cause dephosphorylation of FAK, the cytoplasmic kinase found in focal adhesions (33). Hypertonic stress activates FAK (32, 54), and changes in cell volume influence integrin expression (44) and function (1). In addition, members of the Src family of cytoplasmic tyrosine kinases associate with various Eph receptors and their ephrin ligands; in addition, they may positively regulate ephrin phosphorylation (reviewed in Ref. 30). Src family kinases have similarly been implicated in osmotic signal transduction where their activation in response to anisotonicity regulates various membrane transport processes (31, 36, 50).

EphA receptor ligands of the ephrin A family stably interact with the metalloproteinase, ADAM10 (also known as Kuzbanian). ADAM10 cleaves these membrane-anchored ligands, liberating a soluble form (23). Interestingly, the same protease has been implicated in an analogous ectodomain cleavage of the EGF receptor ligand, heparin-binding-EGF, that follows renal epithelial cell exposure to urea (57). We also show here, via pharmacological inhibitor studies, that the upregulation of EphA2 expression accompanying hypertonic stress requires transactivation of the EGF receptor kinase and is likely dependent on metalloproteinase action. In the response to urea (57) and many other agonists (40), metalloproteinases mediate the ectodomain cleavage responsible for liberating soluble peptide cytokines and growth factors so they may act in an autocrine and paracrine fashion.

In their study of 12 genes prospectively identified by virtue of their regulation by hypertonic NaCl, Nahm et al. (37) showed that pharmacological inhibition of the MAPKs p38 and ERK and of tyrosine phosphorylation partially blocked induction of EphA2 mRNA by salt stress. The epistatic relationship between these signaling events and metalloproteinase-dependent signaling through the EGF receptor has been the subject of intense investigation. Fischer et al. (19) recently reported that, in the stress response of a variety of cell lines, ERK activation is likely downstream of EGF receptor activation, whereas p38 activation is potentially an upstream event. Our data reported herein, and those of Nahm et al. (37), are entirely consistent with this postulated mechanism.

In summary, EphA2 expression and phosphorylation are regulated by tonicity and urea stress. Consistent with a physiological role for this protein under anisotonic conditions, EphA2 is highly expressed in medullary and papillary collecting duct and is upregulated in the papilla in vivo in the setting of water deprivation. Effectors of EphA2 may play a role in the maintenance or regulation of medullary anisotonicity or in the response to this stressor.


These studies were supported by National Institutes of Health Grants DK-52494 (to D. M. Cohen) and AG-14699 (to S. Anderson) and by the American Heart Association, the Juvenile Diabetes Research Foundation, and the Department of Veterans Affairs.


The authors thank X. Chen for the kind gift of the 2-kb human EphA2 promoter fragment subcloned into pGL2-Basic and for the corresponding p53 mutant.


  • 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.


View Abstract