Ischemia and Reperfusion Enhance ATF-2 and c-Jun Binding to cAMP Response Elements and to an AP-1 Binding Site from the c- jun Promoter*

The transcription factors controlling the complex ge- netic response to ischemia and their modes of regulation are poorly understood. We found that ATF-2 and c-Jun DNA binding activity is markedly enhanced in post-ischemic kidney or in LLC-PK 1 renal tubular epi- thelial cells exposed to reversible ATP depletion. After 40 min of renal ischemia followed by reperfusion for as little as 5 min, binding of ATF-2 and c-Jun, but not ATF-3 or CREB (cAMP response element binding protein), to oligonucleotides containing either an ATF/cAMP response element (ATF/CRE) or the jun 2TRE from the c- jun promoter, was significantly increased. Binding to jun 2TRE and ATF/CRE oligonucleotides occurred with an identical time course. In contrast, nuclear protein binding to an oligonucleotide containing a canonical AP-1 element was not detected until 40 min of reperfusion, and although c-Jun was present in the complex, ATF-2 was not. Incubating nuclear extracts from reperfused kidney with protein phosphatase 2A markedly re- duced binding to both the ATF/CRE and jun 2TRE oligonucleotides, compatible with regulation by an ATF-2 kinase. An ATF-2 kinase, which phosphorylated both the transactivation and DNA binding domains of ATF-2, was activated by reversible ATP depletion. This kinase co-eluted on Mono Q column chromatography with a c-Jun amino-terminal kinase and with the peak of stress-acti-vated protein kinase, but not p38, immunoreactivity.

The transcription factors controlling the complex genetic response to ischemia and their modes of regulation are poorly understood. We found that ATF-2 and c-Jun DNA binding activity is markedly enhanced in post-ischemic kidney or in LLC-PK 1 renal tubular epithelial cells exposed to reversible ATP depletion. After 40 min of renal ischemia followed by reperfusion for as little as 5 min, binding of ATF-2 and c-Jun, but not ATF-3 or CREB (cAMP response element binding protein), to oligonucleotides containing either an ATF/cAMP response element (ATF/CRE) or the jun2TRE from the cjun promoter, was significantly increased. Binding to jun2TRE and ATF/CRE oligonucleotides occurred with an identical time course. In contrast, nuclear protein binding to an oligonucleotide containing a canonical AP-1 element was not detected until 40 min of reperfusion, and although c-Jun was present in the complex, ATF-2 was not. Incubating nuclear extracts from reperfused kidney with protein phosphatase 2A markedly reduced binding to both the ATF/CRE and jun2TRE oligonucleotides, compatible with regulation by an ATF-2 kinase. An ATF-2 kinase, which phosphorylated both the transactivation and DNA binding domains of ATF-2, was activated by reversible ATP depletion. This kinase coeluted on Mono Q column chromatography with a c-Jun amino-terminal kinase and with the peak of stress-activated protein kinase, but not p38, immunoreactivity. In conclusion, DNA binding activity of ATF-2 directed at both ATF/CRE and jun2TRE motifs is modulated in response to the extreme cellular stress of ischemia and reperfusion or reversible ATP depletion. Phosphorylation-dependent activation of the DNA binding activity of ATF-2, which appears to be regulated by the stress-activated protein kinases, may play an important role in the earliest stages of the genetic response to ischemia/reperfusion by targeting ATF-2 and c-Jun to specific promoters, including the c-jun promoter and those containing ATF/CREs.
Acute renal failure is a major cause of morbidity in hospitalized patients and often results from transient cessation of blood flow (ischemia) followed by restoration of flow (reperfusion) (1). Reperfusion of ischemic tissue activates a complex genetic program, which may induce cells to dedifferentiate, to proliferate, or, possibly, to undergo apoptosis (1,2). Very little is known about the transcription factors governing these responses to ischemia and how they are activated.
Transcription of c-jun, c-fos, and Egr-1 is increased after reperfusion of ischemic kidney (2)(3)(4)(5). The transcription of these and other genes is regulated in part by transcription factors whose phosphorylation state determines their DNA binding to specific cis-acting elements or their trans-acting activity (6 -16). For example, the trans-acting activity of c-Jun, which autoregulates c-jun transcription (17), and ATF-2, a member of the family of cAMP response element (CRE) 1 binding proteins that mediates gene expression induced by the adenoviral E1a protein (18 -20), is increased by phosphorylation of residues in their amino-terminal transactivation domains (6,7,(21)(22)(23). These sites are efficiently phosphorylated by the stress-activated protein kinases (SAPKs), a subfamily of the mitogenactivated protein kinases, which are minimally activated by growth factors but markedly activated by inflammatory cytokines, heat shock, inhibitors of protein synthesis, and, as we have reported, reperfusion of ischemic kidney or reversible ATP depletion in renal tubular epithelial cells in culture (24 -26). Phosphorylation of ATF-2 by a SAPK in vitro at one or more sites outside the transactivation domain increases the DNA binding activity of ATF-2 (27), but it has been reported that ATF-2 DNA binding activity is not regulated in vivo in response to cellular stresses such as short-wavelength UV irradiation or exposure to other genotoxic agents, which are potent activators of the SAPKs (21,23).
We report that ATF-2 and c-Jun DNA binding activity is highly regulated in the intact animal in response to the extreme, albeit common, pathophysiologic stress of reperfusion of the ischemic kidney, and ATF-2 DNA binding activity is regulated by an ischemia-activated ATF-2 kinase, probably a SAPK. Phosphorylation may serve to target ATF-2 and c-Jun to various promoter sites, including the jun2TRE within the c-jun promoter, and this targeting may be an important initial step in the transcriptional responses necessary for repair and proliferation of cells that have survived the ischemic insult to restore tissue functional integrity.

MATERIALS AND METHODS
Animal Protocol-Male Sprague-Dawley rats weighing 220 -350 g were anesthetized with sodium pentobarbital (55 mg/kg) and then ad-* This work was supported by U. S. Public Health Service (USPHS) Grant DK01986, an American Heart Association grant-in-aid, and a grant from the National Kidney Foundation of Massachusetts and Rhode Island (to T. F.) and by USPHS Grants DK41513 and GM46577 (to J. M. K.) and DK39773, DK38452, and NS10828 (to J. V. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
For the binding reactions, nuclear extracts (3-20 g) were mixed with 20,000 cpm of the appropriate 32 P-labeled oligonucleotide in 20 l of buffer containing 10 mM Hepes (pH 7.9), 10% glycerol, 50 mM KCl, 0.1 mM EDTA, 1 mM DTT, 1 mM PMSF, and 1 g of poly[d(I⅐C)]. After 20 min at room temperature, the reaction products were separated on a 4% non-denaturing polyacrylamide gel. Gels were then dried and subjected to autoradiography. Competition experiments were performed by preincubating the extracts with unlabeled ATF/CRE, jun2TRE, or AP-1 oligonucleotides for 5 min. Labeled oligonucleotide was then added (20,000 cpm) for 20 min prior to electrophoresis.
Antibody supershift assays were performed by adding 2 g of the appropriate antibody to the extract after a 20-min preincubation of extract with labeled oligonucleotide. After 4 h at 4°C, samples were run on 4% non-denaturing gels. Alternatively, incubation with the antibody was performed for 1 h prior to addition of the labeled probe. All antibodies were obtained from Santa Cruz Biotechnology, Inc.
Phosphatase Experiments-To determine the role of phosphorylation in the binding activity directed at the ATF/CRE and jun2TRE oligonucleotides, extracts were incubated for 30 min at 30°C with the catalytic subunit of protein phosphatase 2A, which was purified from rabbit skeletal muscle (31) (generously provided by Dr. David L. Brautigan). To ensure that any change in binding activity was due to phosphatase activity, extracts were also exposed to phosphatase 2A in the presence of the phosphatase inhibitor, okadaic acid (100 nM), or to heat-inactivated phosphatase 2A (10 min at 55°C; heat inactivation was employed because a second phosphatase 2A inhibitor, microcystin, directly inter-fered with the gel shift assay). Following this, gel shift assays were performed as above. The effect of the phosphatase on ATF/CRE and jun2TRE binding was quantified by densitometry using an IS 1000 digital imaging system.

RESULTS
Rats were subjected to 40 min of unilateral renal ischemia followed by no reperfusion or reperfusion for 5, 40, or 90 min. Nuclear extracts were made from these kidneys and from sham-operated controls as well as contralateral, non-ischemic control kidneys. Binding of extract proteins to an oligonucleotide (5Ј-GATCCAGCTTGATGACGTCAGCCG-3Ј) containing the consensus palindromic ATF/CRE octanucleotide (in boldface type) was evaluated. Binding activity directed at the ATF/ CRE oligonucleotide in nuclear extracts from ischemic kidneys that had not been reperfused was not different from control kidney activity (Fig. 1A). Reperfusion for as little as 5 min, however, markedly increased ATF/CRE binding activity. Binding activity remained greater than control in extracts from kidneys reperfused for 90 min (Fig. 1A). The DNA binding activity was specific to the ATF/CRE oligonucleotide, since 5-fold excess unlabeled oligonucleotide markedly reduced and 125-fold excess oligonucleotide eliminated binding to the labeled oligonucleotide (Fig. 1B, left), but 125-fold excess of an unlabeled oligonucleotide (5Ј-GATCCAGCTTGATCATGTCA-GCCG-3Ј), containing base pair changes at two sites within the ATF/CRE palindrome (underlined), did not reduce binding (Fig. 1B, right).
We also examined DNA binding activity in nuclear extracts from renal tubular epithelial cells (LLC-PK 1 ) exposed to reversible ATP depletion, one component of ischemia. Consistent with previous observations in cultured cells, there was a high basal binding activity in control cells directed at the ATF/CRE oligonucleotide, but ATP repletion resulted in a significant increase in binding activity (Fig. 1C).
Antibody interference and supershift assays were performed to identify the proteins binding to the ATF/CRE oligonucleotide after reperfusion. ATF-2 is a member of the group of bZip transcription factors. Heterodimer formation between members of the bZip group is common and is believed to add diversity to the cis-acting elements at which binding of the dimers is directed. Specifically, ATF-2 may dimerize with c-Jun, as occurs in response to E1a, and in so doing shift the binding preference of c-Jun toward ATF/CRE sites (35)(36)(37). Therefore, we determined whether ATF-2 or c-Jun were present in the complex binding to the ATF/CRE oligonucleotide. Preincuba-

FIG. 1. Reperfusion enhances ATF/CRE binding activity in kidney and LLC-PK1 nuclear extracts.
A, time course of reperfusioninduced ATF/CRE binding activity. Nuclear extracts were made from both kidneys of animals subjected to 40 min of unilateral renal ischemia followed by no reperfusion (0 min reperfusion) or reperfusion for 5, 40, or 90 min or from sham-operated controls. Extracts were incubated with the labeled ATF/CRE-containing oligonucleotide as described under "Materials and Methods" and then run on a non-denaturing 4% polyacrylamide gel followed by autoradiography. Ischemic kidneys (I), reperfused for the times noted, or sham-operated control kidney (S) are grouped together with their respective contralateral control kidneys (C). In this and all subsequent figures, lanes labeled P contain labeled probe without nuclear extract. The more slowly migrating band (top arrow) reflected specific ATF/CRE binding activity since it, in contrast to the more rapidly migrating band, was observed in all experiments and was specifically eliminated by unlabeled ATF/CRE oligonucleotide (see below). The more rapidly migrating band (bottom arrow) was clearly apparent only in this experiment (compare to subsequent figures) and was not eliminated by 125-fold excess unlabeled oligonucleotide (data not shown). B, competition experiment for binding to the ATF/CRE oligonucleotide. Left, nuclear extracts from kidneys of animals subjected to 40 min of ischemia followed by 40 min of reperfusion were preincubated with no unlabeled ATF/CRE oligonucleotide (Ϫ) or 5, 25, or 125-fold excess of unlabeled ATF/CRE oligonucleotide followed by incubation with labeled ATF/CRE oligonucleotide. Also shown is ATF/CRE binding activity in the nuclear extract from the contralateral (non-ischemic) kidney (C). Right, nuclear extracts from kidneys of animals reperfused for 40 min after 40 min of ischemia were preincubated with no oligonucleotide (none), with 125-fold excess unlabeled ATF/CRE oligonucleotide (ATF/CRE) or 125-fold excess of an ATF/CRE oligonucleotide (ATF/CRE Mutant) containing a 2-base pair change in the ATF/CRE palindrome (see "Results"), followed by incubation with the labeled ATF/CRE oligonucleotide. C, ATF/CRE binding activity in LLC-PK 1 cells exposed to reversible ATP depletion. ATF/CRE binding activity was determined in nuclear extracts from control cells (C), cells depleted of ATP with cyanide and 2-deoxyglucose (D), or cells re-exposed to dextrose to allow cellular ATP levels to recover (D/R). tion of extracts with anti-ATF-2 or anti-c-Jun, but not anti-CREB or anti-ATF-3, markedly decreased apparent binding activity directed at the ATF/CRE oligonucleotide, suggesting ATF-2 and c-Jun account for an important percent of total ATF/CRE binding activity in post-ischemic kidney or postanoxic LLC-PK 1 cells (Fig. 2A, left and right). Although the c-Jun antibody decreased the apparent binding activity more than the ATF-2 antibody, this does not necessarily mean that c-Jun accounts for a larger percent of DNA binding activity and may be due to differences in the antibodies and/or to access of the antibodies to their respective epitopes. When extracts were first incubated with labeled ATF/CRE oligonucleotide and then one of the monoclonal antibodies, a supershift was induced by anti-ATF-2 and anti-c-Jun, but not anti-CREB or anti-ATF-3, even after prolonged exposure of the gels (Fig. 2B, left and  right). These data suggest that reperfusion of ischemic kidney specifically activates DNA binding activity of ATF-2 and c-Jun directed at an ATF/CRE motif. Although these data do not prove conclusively that ATF-2 and c-Jun bind to the ATF/CRE motif as a dimer after reperfusion of ischemic kidney, it is highly probable since c-Jun homodimers or heterodimers of c-Jun and a c-Fos family member bind with low affinity to ATF/CRE sites. This reperfusion-induced dimerization with ATF-2 appears to be specific to c-Jun since, as noted, neither ATF-3 nor CREB, with which ATF-2 may also dimerize (32,35,36,38,39), was detected in the complex.
The c-jun promoter is one of only a very few target regions of ATF-2 that have been identified. ATF-2 and c-Jun heterodimers appear to be responsible, at least in part, for the induction of c-jun by adenoviral E1a protein (37). The heterodimer binds to a cis-acting element in the c-jun promoter called the jun2TRE (TTACCTCA, positions Ϫ190 to Ϫ183 of the c-jun promoter), which is a variant of the canonical AP-1 element (TGA(C/G)TCA). Since c-jun is induced following reperfusion of ischemic kidney (3, 4), we determined whether reperfusion also enhanced binding activity of ATF-2 and c-Jun directed at a jun2TRE-containing oligonucleotide (5Ј-AGCT-AGCATTACCTCATCCC-3Ј, position Ϫ194 to Ϫ179 of the c-jun promoter underlined), as a possible mechanism of c-jun induction. Reperfusion enhanced a jun2TRE binding activity in kidney nuclear extracts (Fig. 3A), which was eliminated by incubation with 125-fold excess unlabeled jun2TRE oligonucleotide (Fig. 3B). The time course of activation of the jun2TRE binding activity and that of the ATF/CRE binding activity were similar. Binding to the jun2TRE was evident as early as 5 min postreperfusion. Preincubation of kidney nuclear extracts with antibodies to either ATF-2 or c-Jun, but not ATF-3 or CREB, prior to gel shift assay markedly reduced the intensity of the shifted band and induced a supershift (Fig. 3C). These data confirm the presence of both ATF-2 and c-Jun in the complex and suggest that together they account for a significant percent of the jun2TRE binding activity in post-ischemic kidney. These data suggest that ATF-2 and c-Jun may play a critical role in the induction of c-jun following ischemia and reperfusion, as they appear to do following E1a transformation, and that the initiating mechanism of c-jun induction may be a reperfusioninduced increase in DNA binding activity of ATF-2 and c-Jun.
When compared to the time course of binding activity directed at ATF/CRE and jun2TRE oligonucleotides, the binding to a canonical AP-1 motif-containing oligonucleotide (5Ј-AGCT-TGGTGACTCATCCG-3Ј) was delayed (Fig. 4A). AP-1 binding activity was not apparent until 40 min of reperfusion. Binding increased further at 90 min. Binding was eliminated by incubation with 125-fold excess of unlabeled AP-1 oligonucleotide (data not shown). c-Jun was present in the complex binding the AP-1-containing oligonucleotide, but in contrast to the complexes binding the ATF/CRE and jun2TRE-containing oligonucleotides, ATF-2 was not present (Fig. 4B). These data indicate that c-Jun DNA binding activity is increased after reperfusion of ischemic kidney, but the dimer binding to AP-1 motifs does not include ATF-2 and is likely to be composed of more typical AP-1 dimers such as c-Jun⅐c-Jun and c-Jun⅐c-Fos.
We performed gel shift assays after nuclear extracts had been exposed to the protein serine phosphatase, phosphatase

FIG. 2. ATF-2 and c-Jun are present in the protein complex binding to the ATF/CRE oligonucleotide.
A, antibody interference assays demonstrating that anti-ATF-2 and anti-c-Jun antibodies reduce binding to the ATF/CRE oligonucleotide. Nuclear extracts from kidneys of animals rendered ischemic and reperfused for 40 min (I/R) or sham-operated controls (C) (left), or from control LLC-PK 1 cells (C) or cells depleted of ATP and then re-exposed to glucose to increase ATP (D/R) (right) were preincubated with antibodies to ATF-2, c-Jun, ATF-3, or CREB followed by incubation with labeled ATF/CRE probe. B, prolonged exposures of autoradiograms from supershift assays demonstrating presence of ATF-2 and c-Jun but not ATF-3 (left) or CREB (right) in the complex binding to the ATF/CRE oligonucleotide. Nuclear extracts from kidneys of animals subjected to 40 min of ischemia followed by 40 min of reperfusion were first incubated with the ATF/CRE oligonucleotide and then antibodies to ATF-2, c-Jun, or ATF-3 (left) or ATF-2, c-Jun, or CREB (right) prior to gel electrophoresis. Supershifted complexes are identified with an arrow. To enhance the supershifted bands, the autoradiograms were exposed for 48 h, accounting for the apparent increase in the intensity of the band in extracts from control kidney (C). In the experiment performed for the left panel, anti-ATF-2 or anti-c-Jun was also incubated with probe (P) in the absence of nuclear extract.
2A, to determine whether protein phosphorylation played a role in the enhanced ATF/CRE binding activity. Pretreatment of kidney nuclear extracts with phosphatase 2A prior to gel shift assay reduced the intensity of the shifted band (Fig. 5A). This effect of phosphatase 2A was prevented by heat inactivation of the phosphatase (55°C for 10 min) (Fig. 5A, left) or by co-incubation with the phosphatase 2A inhibitor, okadaic acid (100 nM) (Fig. 5A, right). Incubation of nuclear extracts with okadaic acid alone had no effect.
We also explored the role of phosphorylation in the ischemiainduced increase in binding to the jun2TRE. Pre-incubation of nuclear extracts with phosphatase 2A markedly reduced the intensity of the shifted band (Fig. 5B). This effect of phosphatase 2A was inhibited by okadaic acid. These data suggest the FIG. 3. Reperfusion enhances a jun2TRE binding activity in kidney nuclear extracts. A, time course of jun2TRE binding activity. Nuclear extracts from kidneys of animals subjected to unilateral renal ischemia (I) and then reperfused for the times indicated above the lanes, sham control (S), or control (contralateral) kidneys (C) were incubated with labeled jun2TRE oligonucleotide and run on a non-denaturing polyacrylamide gel. B, competition experiment for binding to the jun2TRE oligonucleotide. Nuclear extracts from kidneys of animals subjected to 40 min of ischemia followed by 40 min of reperfusion were preincubated with no unlabeled jun2TRE oligonucleotide (Ϫ) or 5, 25, or 125-fold excess of unlabeled jun2TRE oligonucleotide followed by incubation with labeled oligonucleotide. C, ATF-2 and c-Jun are present in the protein complex binding to the jun2TRE oligonucleotide. Nuclear extracts from kidneys of animals subjected to 40 min of unilateral renal ischemia followed by 40 min of reperfusion (I/R) or from contralateral control kidney (C) were preincubated with antibodies to ATF-2, c-Jun, CREB, or ATF-3 prior to gel shift assay with the jun2TRE oligonucleotide. Supershifted bands are identified with an arrow. phosphorylation state of ATF-2, presumably mediated by an ATF-2 kinase, is the critical determinant of binding of the putative ATF-2⅐c-Jun dimer to both the ATF/CRE and jun2TRE motifs since dephosphorylation of any of three sites within the COOH-terminal region of c-Jun by phosphatase 2A should have increased DNA binding activity of c-Jun.
We then performed experiments to identify the ATF-2 kinase(s) activated by ischemia and reperfusion or reversible ATP depletion. Because phosphorylation in vitro of residues outside the transactivation domain may increase ATF-2 DNA binding activity, we used GST-ATF-2 (94 -505), containing the DNA binding and dimerization domains as substrate to assay Mono Q anion exchange column fractions from LLC-PK 1 cells exposed to chemical anoxia. The ATP depletion-activated ATF-2 DNA binding-dimerization domain kinase eluted as a single major peak (fractions 10 -12) (Fig. 6A). This kinase activity co-eluted with the peak of ATP depletion-activated ATF-2 transactivation domain kinase activity (Fig. 6B). Thus, reversible ATP depletion activated a single dominant peak of ATF-2 kinase activity. This kinase phosphorylated both the transactivation and DNA binding domains of ATF-2, suggesting it may modulate both the enhanced DNA binding activity of ATF-2 we observed and transactivating activity. The kinase activity coeluted with an ATP depletion-activated c-Jun amino-terminal (transactivation domain) kinase, suggesting it is a SAPK (Fig.  6C). Immunoblots of column fractions confirmed that the kinase activity co-eluted with SAPK immunoreactivity. The 46-kDa isoform eluted in fractions 10 and 11 and the 54-kDa isoform in fractions 11 and 12. A small amount of the 46-kDa isoform eluted earlier in the gradient (fractions 4 -6, apparent on longer exposure of the audioradiogram), which probably accounted for the small peak of ATF-2 kinase activity in those fractions.
The other major stress-activated mitogen-activated protein kinase identified to date, p38, which is activated by another ATP-depleting agent, sodium arsenite (34), and has been reported to be an ATF-2 transactivation domain kinase (40,41), eluted as a sharp peak in fraction 20 (data not shown). No ATF-2 kinase activity was detected in fraction 20, suggesting p38 either is not activated by cyanide-induced chemical anoxia or ATF-2 is not a physiologic substrate.
Erk-1 and Erk-2, which can also phosphorylate the DNA binding-dimerization domain of ATF-2 in vitro and enhance its DNA binding activity (27), eluted just after the peak of ATF-2 kinase activity in fractions 14 and 12, respectively (not shown). This, plus our prior observation that the Erks are not activated by the reversible ATP depletion protocol used in these experiments (26), suggests the Erks do not modulate ATF-2 DNA binding activity in response to reversible ATP depletion. Taken together, our data support the hypothesis that the SAPKs are the ATP depletion-activated kinases that modulate DNA binding and, possibly, transactivating activity of ATF-2.

DISCUSSION
Reperfusion of ischemic kidney activates a genetic program that culminates in profound phenotypic changes, particularly in tubular epithelial cells, which are very vulnerable to ischemia (1)(2)(3)(4)(5). Tubular epithelial cells that survive the ischemic insult may dedifferentiate and enter the cell cycle to replace irreversibly injured cells. Reperfusion induces the transcription of several genes, which may play a role in the dedifferentiation and proliferative responses, including c-jun, c-fos, Egr-1, and PCNA.
Induction of many genes is controlled by transcription factors, trans-acting activity of which is often modulated by phosphorylation catalyzed by protein kinases (reviewed in Refs. 9 and 42). We have recently reported that the SAPKs are the predominant c-Jun amino-terminal (transactivation domain) kinases activated by reperfusion of the ischemic kidney (26). The SAPKs also phosphorylate ATF-2 (27, 32), a member of the FIG. 4. Reperfusion enhances an AP-1 binding activity in kidney nuclear extracts. A, time course of AP-1 binding activity. Nuclear extracts from ischemic kidneys (I) reperfused for the times indicated, from contralateral kidneys (C), or from sham-operated controls (S) were incubated with an AP-1-motifcontaining oligonucleotide prior to gel electrophoresis. B, c-Jun, but not ATF-2, is present in the protein complex binding to the AP-1 oligonucleotide. Nuclear extracts from kidneys of animals reperfused for 90 min (the time of peak DNA binding activity to the AP-1 oligonucleotide) after 40 min of unilateral renal ischemia were incubated with the AP-1 oligonucleotide followed by either anti-ATF-2 or anti-c-Jun prior to gel electrophoresis. ATF/CREB family of transcription factors, which can mediate transcriptional responses to the adenoviral E1a protein (18,19). The SAPKs, but not protein kinase A, phosphorylate ATF-2 in vitro at several sites within both the amino-terminal and carboxyl-terminal regions of the protein. Phosphorylation of two amino-terminal threonine residues appears to enhance trans-acting activity of ATF-2.
Abdel-Hafiz et al. (27) first proposed that a SAPK (then known as p54 mitogen-activated protein kinase) might regulate ATF-2 DNA binding activity by phosphorylating one or more residues in the DNA binding domain of ATF-2. Co-incubation of bacterial-or baculoviral-expressed ATF-2 and p54 mitogen-activated protein kinase purified from rat liver markedly increased DNA binding activity of ATF-2 directed at an ATF/CRE oligonucleotide. Prior to our studies, however, it did not appear that ATF-2 DNA binding activity was importantly modulated in vivo in response to cellular stresses. Gupta et al. (21) were unable to demonstrate any change in ATF-2 DNA binding activity in nuclear extracts from COS-1 cells irradiated with short-wavelength UV, a cellular stress that potently activates the SAPKs. Similarly, hypoxic and low-glucose stress, which presumably, like chemical anoxia (26), activates the SAPKs, was not associated with an increase in binding activity directed at an oligonucleotide containing the jun2TRE (43). In contrast, our experiments demonstrate that ATF-2 DNA binding activity is modulated in response to reversible ATP deple-tion of renal tubular epithelial cells and reperfusion of ischemic kidney in the intact animal.
One possible explanation for this apparent discrepancy is that regulators of ATF-2 DNA binding may be constitutively active in some cultured cells. This is supported by in vivo footprint analyses, which suggest that ATF-2 and c-Jun act as pre-bound heterodimers at the junTREs in many cultured cells and that these preformed complexes mediate induction of c-jun in response to cellular stress (23,44). We cannot determine directly by in vivo footprint analysis whether ATF-2⅐c-Jun are pre-bound in the intact animal. However, the extremely low basal ATF-2 DNA binding activity and the marked increase in response to reperfusion in kidney and brain, 2 compared to the high basal level of binding reported in some cultured cells, suggests regulation of ATF-2 DNA binding is fundamentally different in these cultured cells compared to the intact animal.
To date, very few physiologic activators of ATF-2 have been identified. ATF-2 is not activated by stimuli that activate some other members of the ATF/CRE family. In contrast to CREB and ATF-1, ATF-2 is not activated by increases in levels of cAMP and subsequent activation of protein kinase A or by increases in cytosolic free [Ca 2ϩ ] (20,45,46). ATF-2 has not been reported to be activated by growth factors (15). ATF-2 is activated by the adenovirus E1a protein and by overexpression of the retinoblastoma susceptibility gene product, Rb, both of which interact directly with ATF-2 and alter its DNA binding and, in the case of E1a, its trans-acting activity (18 -20, 37, 47-50). Reperfusion of ischemic kidney is the first physiologic or pathophysiologic stimulus, besides adenoviral transformation, that activates ATF-2 DNA binding.
Our data suggest regulation of ATF-2 DNA binding activity may be an important mechanism for targeting ATF-2 to various promoters, including the c-jun promoter and promoters with ATF/CREs. We also found that reperfusion enhanced binding of c-Jun to the ATF/CRE and jun2TRE motifs and postulate this occurs via heterodimerization with ATF-2. ATF-2 readily forms dimers with c-Jun in vitro. Ordinarily, c-Jun homodimers or heterodimers of c-Jun with other jun or fos family members bind poorly to ATF/CRE motifs (36). However, when c-Jun and ATF-2 form a heterodimer, the DNA binding specificity of c-Jun changes, causing the dimer to bind preferentially to modifications of consensus AP-1 sites including the ATF/CRE palindrome (5Ј-TGACGTCA-3Ј) and the jun2TRE (5Ј-TTACCTCA-3Ј) (35)(36)(37)39). Prior studies have demonstrated that c-Jun can form dimers with ATF-2, ATF-3, and ATF-4 in vitro (but not with CREB or ATF-1). While all three heterodimers can bind to an ATF/CRE-containing oligonucleotide, it is the ATF-2⅐c-Jun dimer that preferentially binds to ATF/CRE as compared to AP-1 sites (36). This, plus the absence of ATF-3 in the complex binding to the ATF/CRE oligonucleotide after reperfusion, suggests c-Jun is targeted to the ATF/CRE motif as an ATF-2⅐c-Jun heterodimer.
Transcription of genes containing canonical AP-1 sites in their promoters, such as vimentin, is also increased after reperfusion (2). c-Jun⅐c-Fos heterodimers or, to a lesser extent, c-Jun⅐c-Jun homodimers are considered to be the predominant AP-1 complexes regulating transcription from promoters with these AP-1 motifs. We found that binding to an AP-1 elementcontaining oligonucleotide was enhanced after reperfusion of ischemic kidney, and c-Jun was present in the complex but ATF-2 was not. The data suggest that modulation of induction of AP-1-controlled genes by ischemia and reperfusion in vivo involves enhanced DNA binding activity of c-Jun and, possibly, other components of the heterodimer.
Our data demonstrate that the reperfusion-induced increase in binding to both the ATF/CRE and jun2TRE motifs is, at least in part, protein phosphorylation-dependent, since pretreatment of nuclear extracts with the protein serine phosphatase 2A significantly reduced binding to both oligonucleotides. This was somewhat surprising since c-Jun was present in the complex binding to both motifs, and dephosphorylation by phosphatase 2A of the three residues in the carboxyl-terminal region of c-Jun should have enhanced its DNA binding activity (9,13,14). These data, and the previous work demonstrating enhanced binding of ATF-2 to an ATF/CRE oligonucleotide FIG. 6. SAPKs are the reversible ATP depletion-activated ATF-2 kinases. A, activation of the ATF-2 (94 -505) kinases. Mono Q column fractions from lysates of LLC-PK 1 cells exposed to reversible ATP depletion (ATP depletion/repletion) or mock depletion (control) were assayed as described under "Materials and Methods" with ATF-2 (94 -505), containing the DNA binding and dimerization domains, as substrate. B, activation of the ATF-2 (8 -94) kinases. Mono Q column fractions were assayed with ATF-2 (8 -94), containing the transactivation domain, as substrate. C, alignment of peaks of ATF-2 (94 -505), ATF-2 (8 -94), and c-Jun (1-135) kinase activity following Mono Q fractionation of lysates from LLC-PK 1 cells exposed to reversible ATP depletion. D, anti-SAPK immunoblot of Mono Q column fractions from lysates of LLC-PK 1 cells exposed to reversible ATP depletion. A sample of the cell lysate, which was loaded onto the column, was run in the lane labeled L. Molecular mass standards are on the left. after phosphorylation in vitro (27), support the contention that ATF-2 is the primary determinant of the binding of the putative ATF-2⅐c-Jun dimer to the ATF/CRE oligonucleotide.
Although there are many examples where phosphorylation decreases DNA binding activity of transcription factors, including c-Jun (13,14), there are few well documented instances in which phosphorylation increases binding (reviewed in Ref. 9). Phosphorylation of ATF-2 is postulated to enhance dimerization and DNA binding by inducing an allosteric change in the conformation of the protein such that the zinc finger motif of ATF-2 (residues 27-49), which is believed to inhibit DNA binding by interacting with the bZip domain (residues 350 -409), no longer interacts with the bZip domain (27,32). We found a single major peak of reversible ATP depletion-activated ATF-2 kinase activity, whether the transactivation domain or DNA binding and dimerization domains of ATF-2 were used as substrates. This kinase activity co-eluted with the c-Jun aminoterminal kinase activity and SAPK immunoreactivity but not with the peak of p38 immunoreactivity, another stress-activated kinase purported to be an ATF-2 kinase. Thus, in agreement with work done in vitro (27), the SAPKs appear to be the major candidate modulator of ATP depletion-activated ATF-2 DNA binding activity we observed.
E1a targets promoters by interacting with the bZip DNA binding domain of ATF-2 and other transcription factors, and then the transactivating region of E1a and ATF-2 cooperatively increase transcription (19). Our data, demonstrating phosphorylation-dependent enhancement of DNA binding activity, and the previous reports of phosphorylation-dependent trans-acting activity (21-23) present a picture similar to activation of ATF-2 by E1a and suggest the SAPKs may fulfill both roles of E1a: promoter targeting and transcriptional activation.
In conclusion, following reperfusion of ischemic kidney, the DNA binding activity of ATF-2 and c-Jun increases markedly. This activation process may be an important mechanism whereby the pathophysiologic stress of renal ischemia and reperfusion activates gene expression. The mechanism of reperfusion-induced activation of binding to two types of promoter elements, an ATF/CRE motif and the jun2TRE, appears to involve phosphorylation catalyzed by the SAPKs. The reperfusion-activated binding of ATF-2 and c-Jun to the jun2TRE may provide a mechanism of induction of c-jun potentiating a SAPK-induced increase in c-Jun and ATF-2-trans-acting activity.