EF Domains Are Sufficient for Nongenomic Mineralocorticoid Receptor Actions*

The mineralocorticoid receptor (MR) is important for salt homeostasis and reno-cardiovascular pathophysiology. Signaling mechanisms include, besides classical genomic pathways, nongenomic pathways with putative pathophysiological relevance involving the mitogen-activated protein kinases ERK1/2. We determined the MR domains required for nongenomic signaling and their potential to elicit pathophysiological effects in cultured cells under defined conditions. The expression of full-length human MR or truncated MR consisting of the domains CDEF (MRCDEF), DEF (MRDEF), or EF (MREF) renders cells responsive for the MR ligand aldosterone with respect to nongenomic ERK1/2 phosphorylation, whereas only full-length MR and MRCDEF conferred genomic responsiveness. ERK1/2 phosphorylation depends on the EGF receptor and cSRC kinase. MREF expression is sufficient to evoke the aldosterone-induced increase of collagen III levels, similar to full-length MR expression. Our data suggest that nongenomic MR signaling is mediated by the EF domains and present the first proof of principle showing that nongenomic signaling can be sufficient for some pathophysiological effects. The minimum amino acid motif required for nongenomic MR signaling and its importance in various effects have yet to be determined.

The mineralocorticoid receptor (MR) 2 is usually described as a ligand-inducible transcription factor that controls expression of target genes involved in the regulation of Na ϩ and K ϩ homeostasis as well as blood pressure regulation (1). MR also promotes cardiovascular and renal fibrosis caused by tissue remodeling as well as endothelial dysfunction, independent of its effects on blood pressure or NaCl homeostasis (2)(3)(4). The activation of MR modulates the expression of various proteins like the epithelial sodium channel, Na ϩ -K ϩ -ATPase, the SGK kinase, and the EGF receptor (1,(5)(6)(7). There is increasing evidence that not all biological effects of MR are mediated by direct DNA binding and control of target gene expression (8,9). Some actions of MR appear to be the result of a cross-talk with other signaling cascades, such as nongenomic regulation of intracellular calcium (10,11), protein kinase C, cSRC kinase, EGF receptor (EGFR) activity, or extracellular-regulated kinase (ERK1/2) (12)(13)(14)(15). Furthermore, there seems to exist a functional cross-talk between classical and nongenomic actions (9, 16 -18).
The classical receptors for estrogen (ER), progesterone (PR), androgens (AR), and glucocorticoids (GR) also contribute to the nongenomic effects (12, 14, 19 -26) of these hormones, in many cases via ERK1/2 kinases. Some nongenomic effects arise from classical receptors in or at the plasma membrane (e.g. ER, Ref. 13) but specialized membrane receptors, like mPR and GPR30 have also been described (27,28). However, the results for the specialized membrane receptors are controversial (29,30). The mechanism(s) of action for ER and PR as well as the receptor domains involved have been explored in more detail. Receptor domains D, E, and F seem to be of special importance (19,26,(31)(32)(33). For example, ER␣ interacts with the SH2 domain of cSRC via a phosphorylated tyrosine residue at position 537 in the EF domain, and this interaction is facilitated by MNAR (26,31). Razandi et al. (32) described an important role for serine 522 within the EF domain of ER␣ during nongenomic signaling. However, there are also reports ascribing a role to domain D (33). Recently, the direct interaction of ER␣ with G␣ i through domain C and G␤␥ through DEF domains has been reported (34). PR has the ability to interact with the SH3 domain of cSRC via a proline-rich region of domain D (19).
A possible role for MR with respect to nongenomic effects has not yet been investigated as extensively as for other steroid receptors. Pharmacological evidence for a physiologically relevant role of the MR in nongenomic actions was presented, for example, with respect to the modulation of the plasma membrane Na ϩ -transporters (35) or vascular reactivity and NO synthesis (36 -38). Furthermore, nongenomic MR actions were made responsible for pathophysiological effects of aldosterone (39 -44), although the final proof is missing. Recently, the involvement of MR in nongenomic signaling at the cellular level was investigated in detail using a heterologous expression system (9). The data from this study showed that MR contributes to rapid aldosterone-induced activation of the ERK1/2 pathway via cSRC and EGFR activation. For MR, no information regarding the domain(s) necessary for nongenomic actions are available. In the present study, we determined which MR domains are required for nongenomic signaling and whether this pathway has the potential to elicit a pathophysiological effect in cultured cells under defined conditions.

EXPERIMENTAL PROCEDURES
Cell Culture-Cell culture was performed as described previously (6,55,56). We used CHO-K1, HEK-293, and OK cells from ATCC. 24 -48 h prior to the experiment, serum was removed. For the experiments presented, the cells were cultivated either on Petri dishes (Becton Dickinson GmbH, Heidelberg, Germany) in 24-well plates (for reporter assay and collagen determination), in 96-well plates (for ELISA), or on glass coverslips (fluorescence).
Constructs and Transfection-Transfection of the cells was performed under serum-free conditions as described before (6,55) with the Qiagen Polyfect reagent (Qiagen, Hilden, Germany), according to the manufacturer's instructions. We used the hMR expression vector pEGFP-C1-hMR (kindly provided by Dr. N. Farman (57)) and pEGFP-C1 (Clontech). Truncated versions of the pEGFP-hMR, which lack either the N-terminal AB domain (hMR CDEF ), the AB domain and the DNA-binding domain (hMR DEF ), or additionally the hinge region (hMR EF ) were constructed by cutting pEGFP-hMR with BglII and Hin-dIII and inserting appropriate PCR fragments (Table 1). hMR AB was generated by restriction with EcoR1. To determine whether the EGFP tag changed the characteristics of the receptor, we compared its GRE activation and nuclear translocation properties with that of the untagged hMR and could not find significant differences (9,57).
Immunoprecipitation-Cells were washed, harvested, and lysed in radioimmune precipitation assay buffer{3551}. Lysates were then centrifuged at 11,000 rpm at 4°C for 10 min, and the supernatant was incubated overnight with EGFP antibody (sc-8334, Santa Cruz Biotechnology) with end-over-end rotation and then with A/G plus-agarose for another 24 h. After 10 min of centrifugation at 10,000 rpm at 4°C, the pellet was mixed with 40 l of Laemmli buffer and separated by an 8% SDS-PAGE gel.
EGFP Fluorescence-Cells were cultivated on glass coverslips. Images were obtained either using an inverted microscope (Zeiss IM 135) equipped with ϫ40 and ϫ100 fluorescence objectives. Fluorescence images were taken with an ICCD camera (Hamamatsu, Herrsching, Germany). Alternatively, cells were analyzed by confocal microscopy (Radiance 2000, Bio-Rad), and the images were processed using the software MetaMorph Imaging System (Microsoft).
Determination of EGFP-hMR Expression by ELISA-EGFP-hMR expression was determined basically by the same method as ERK1/2 phosphorylation (see above). HEK cells were transfected, made quiescent, and after 48 h, expression was determined with an anti-EGFP primary antibody (Clontech) and anti-mouse-horseradish peroxidase antibody. pcDNA3.1-transfected cells were used as negative controls.
Quantification of ERK1/2 Phosphorylation by ELISA-For the quantification of ERK1/2 phosphorylation, we performed ELISA according to Versteeg et al. (58) with minor modifications that were described previously (55). After stimulation as indicated, the cells were fixed with 4% formaldehyde in PBS and permeabilized with 0.1% Triton X-100. Cells were blocked with 10% fetal calf serum in PBS/Triton for 1 h and incubated overnight with the primary antibody. Subsequently, cells were incubated with secondary antibody (peroxidase-conjugated mouse anti-rabbit antibody) in PBS/Triton with 5% bovine serum albumin for 1 h at room temperature. Finally, the cells were incubated with 50 l of a solution containing 0.4 mg/ml o-phenylenediamine, 11.8 mg/ml Na 2 HPO 4 , 7.3 mg/ml citric acid, and 0.015% H 2 O 2 for 15 min at room temperature in the dark. The resulting signal was detected at 490 nm with a multiwell multilabel counter (Victor 2 , Wallac, Turku, Finland). The protein content was determined with Trypan Blue (55). The determination of total ERK1/2 in parallel experiments was used to correlate phosphorylated ERK1/2 to total ERK1/2.
Quantification of Phospho-ERK1/2 by Sandwich-ELISA-For the quantification of phospho-ERK expression, we used the sandwich-ELISA from R&D Systems, Inc., Minneapolis R&D (SUV1018) according to the manufacturer's instructions.
Determination of Extracellular Collagen III Abundance by ELISA-Collagen III abundance in the medium of CHO cells expressing HER1 (55) was determined by ELISA as described previously (51). We tested the cross-reactivity of the primary antibodies (1:1000; Biotrend, Köln, Germany) using collagen standards and did not observe any significant cross-reactivity. Determination of Gelatinase Activity-Gelatinase activity in cell culture media was determined by the ENZ Gelatinase/Collagenase Assay kit from Molecular Probes (Leiden, NL) using fluorescein-conjugated gelatin. The increase in fluorescence is a direct measure for gelatinase/collagenase activity. As a positive control, bacterial collagenase activity (Sigma) was measured, and the gelatinase values are expressed as collagenase-equivalents.
GRE Reporter Gene Assay-Transactivation was assessed by the Mercury TM Pathway Profiling reporter gene assay system from Clontech Inc. using secretory alkaline phosphatase (SEAP) as reporter, essentially as described earlier (9). In brief, the cells were cotransfected with pGRE-SEAP and receptor constructs or empty vectors. SEAP activity in the media was determined with the AttoPhos System from Promega (Mannheim, Germany) and normalized to a transfection control (␤-galactosidase or EGFP).
Statistics-The data are presented as mean values Ϯ S.E. Significance of difference was tested by paired or unpaired Student's t test or analysis of variance as applicable. Differences were considered significant with p Ͻ 0.05. Cells from at least two different passages were used for each experimental series. N represents the number of tissue culture dishes investigated.

RESULTS
Transcriptional Activity of hMR Deletion Constructs-We compared transcriptional activity of the MR deletion constructs using a GRE-SEAP reporter containing three canonical GRE elements. Only cells transfected with hMR or hMR CDEF responded to aldosterone with increased GRE-SEAP activity (Fig. 1). Dexamethasone, a ligand for the endogenous glucocorticoid receptor (GR, Ref. 18) stimulated GRE-SEAP activity under all conditions, demonstrating the functionality of the experimental system. For cells transfected with hMR or hMR CDEF , the effect of dexamethasone was stronger, because of the additional activation of the mineralocorticoid receptor. The relative transactivation potency of hMR CDEF was greater when compared with hMR, because the inhibitory domain located in AB is missing (45). Half-maximal activation was obtained with similar concentrations of aldosterone (Fig. 1).
Expression of hMR Deletion Constructs-Western blot analysis confirmed the expression of proteins of the expected size ( Fig. 2A). The expression level of hMR was slightly lower compared with deletion constructs when determined by ELISA (Fig.  2B). The simplest explanation for this difference is the size of the protein. However, differences in the delivery to degradation sites cannot be excluded.
Using fluorescence microscopy, we investigated the cellular distribution. As expected, hMR is located mainly in the cytosol in the absence of hormone (Fig. 2C). In the presence of aldosterone (10 nmol/liter), the receptor translocates almost completely to the nucleus. hMR CDEF behaves in a similar way. hMR DEF and hMR EF , which are located in the cytosol in the absence of hormone, translocate partially into the nucleus after the addition of aldosterone.
ERK1/2 Phosphorylation in Cells Transfected with hMR Deletion Constructs-We used ERK1/2 phosphorylation as a readout for rapid, nongenomic effects, because it has been described continuously in different experimental systems. First, we investigated whether expression of the different hMR deletion constructs affects ERK1/2 phosphorylation in the absence of aldosterone. For this purpose, ERK phosphorylation was determined by two independent techniques, i.e. sandwich-ELISA and Western blot. Fig. 3 shows that expression of hMR EF led to a ligand-independent increase in ERK1/2 phosphorylation. The expression levels of total ERK1/2 or HSP90 were not altered significantly (Fig. 3). This difference was not observed in the presence of 10 mol/liter U0126 (inhibitor of ERK1/2 phosphorylation; Fig. 3A). Similar results were obtained with a different cell line, OK cells, where hMR EF also enhanced the basal pERK1/2 level (145 Ϯ 9% of control, n ϭ 18, p Ͻ 0.05).
Next, we tested by Western blot whether aldosterone is able to induce a rapid (10 min) increase in ERK1/2 phosphorylation (Fig. 3). In control cells (i.e. transfected with EGFP) aldosterone did not enhance ERK1/2 phosphorylation, in contrast to the positive control EGF. Transfection with MR, MR CDEF , MR DEF , or MR EF rendered cells responsive with respect to rapid ERK1/2 phosphorylation (Fig. 3), indicating that the domains A, B, C, and D are not required. To confirm these data by an independent method and to obtain quantitative data we repeated the experiments with an ELISA technique. Fig. 4A confirms that cells transfected with MR, MR CDEF , MR DEF , or MR EF respond to aldosterone with a rapid phosphorylation of ERK1/2, whereas cells transfected with the empty vector or MR AB did not. When ERK1/2 phosphorylation was prevented by U0126 (10 mol/liter), the ELISA signal was erased almost completely (Fig.  4B). Aldosterone-induced ERK1/2 phosphorylation was completely abolished by the EGFR kinase inhibitor AG1478 (100 nmol/liter, Fig. 4A) or the cSRC kinase inhibitor PP2 (100 nmol/liter, Fig. 4A), in accordance with the previously demonstrated EGFR and cSRC dependence of rapid MR signaling (9,15,39). Fig. 4C shows representative immunoblots of aldosterone-induced cSRC phosphorylation confirming the involvement of this kinase.
To gain further information on the cellular localization of MR EF , confocal microscopy was applied to HEK-293 cells transiently transfected with EGFP-MR EF . As shown in Fig. 4D, EGFP-MR EF is distributed throughout the cytosol reach- ing the plasma membrane without any detectable gap. Next, EGFP-MR EF -transfected cells were incubated with the membrane marker di-8-ANEPPS (red fluorescence). The yellow patches indicate an overlay of green and red fluorescence and therefore close proximity of MR EF to the plasma membrane (Fig. 4E). Preliminary coimmunoprecipitation experiments with MR EF and EGFR indicate an interaction of these two receptors but not of the EGFP tag alone with EGFR (data not shown). Taken together, these data suggest that MR EF -EGFR interaction takes place at the plasma membrane. This question as well as the detailed molecular mechanism of interaction has to be addressed in more depth in future studies.

Potential Pathological Effects of hMR Deletion Constructs-
The deleterious effects of MR activation in renal and cardiovascular tissue comprise to a certain extent inflammatory features, such as activation of the transcription factors AP1 or NFB (46,47) and fibrotic features, such as enhanced collagen III deposition (41, 42, 48 -50). The fibrogenic action of MR most probably depends on a pathogenic milieu, especially increased abundance of reactive oxygen species, which can be generated by NaCl loading of experimental animals. Recently, it was shown that human aortic smooth muscle cells in primary culture respond with enhanced collagen III abundance to MR activation in the presence of low H 2 O 2 concentrations but not in the absence of H 2 O 2 . This effect was EGFR-dependent (51).
We tested the effect of MR and the deletion constructs on AP1 and NFB activation as well as on collagen III abundance. In cells expressing full-length MR, aldosterone enhanced AP1 activity and to a small but significant extent NFB activity (Fig.  5A). In contrast, none of the deletion constructs rendered cells responsive to aldosterone with respect to AP1 or NFB activation. Interestingly, MR DEF and MR EF exerted a ligand-independent inhibitory effect on baseline NFB and AP1 activity. The underlying mechanism is not known. Fig. 5B shows the effect of mild oxidative stress (10 mol/ liter H 2 O 2 ) and/or aldosterone (10 nmol/liter) on collagen III abundance in cells expressing the vector, MR, or one of the deletion constructs. Mild oxidative stress enhanced collagen III abundance in all cases significantly, whereas aldosterone alone exerted virtually no effect. However, in the presence of mild oxidative stress, aldosterone enhanced collagen III abundance significantly not only in cells expressing MR, as already  described previously (51), but also in cells expressing MR CDEF , MR DEF , or MR EF , but not in cells expressing the empty vector (Fig. 5C). U0126 prevented the action of aldosterone in MR EF cells (collagen III abundance was 101 Ϯ 6% of control with 10 mol/liter U0126 and 88 Ϯ 9% of control in the presence of H 2 O 2 ϩ aldosterone ϩ U0126; n ϭ 4). Additionally, we assessed collagenase activity in the media and cellular (pro)collagen III protein content. Collagenase activity was not affected significantly (108 Ϯ 7% of control in the presence of aldosterone ϩ H 2 O 2 in cells expressing MR EF ; n ϭ 4). In contrast, cellular (pro)collagen III showed a similar pattern as extracellular collagen III abundance (Fig. 5C). These data are in agreement with the results of several other studies showing enhanced cellular collagen III synthesis, with little or no changes in collagen III degradation or secretion per se.
Thus, the profibrotic action of the mineralocorticoid receptor seems not to depend on its classical DNA binding capacity but on its nongenomic effects. MR EF , which elicits a certain ligand-independent activation of pERK1/2, also rendered cells more sensitive to H 2 O 2 with respect to collagen III abundance (Fig. 5D), supporting the importance of this signaling pathway for at least some pathophysiological effects of MR.

DISCUSSION
Nowadays the existence of nongenomic effects of the mineralocorticoid receptor, as well as of other steroid hormone receptors, is well accepted (12,14). Future key issues are the evaluation of their physiological and/or pathophysiological relevance as well as the underlying mechanisms. It has been shown that MR elicits its nongenomic effects, at least in part, via the activation of cytosolic SRC kinase and the EGF receptor (9). Therefore, the MR employs a signaling cascade similar to ER and PR (13,14,19,26,31,52), for which the importance of the C-terminal domains D, E, and F in this nongenomic signaling cascade had been shown, as mentioned in the Introduction. Our data in the present study provide strong evidence that the MR also employs the C-terminal domain EF, which comprises the ligand-binding site of the receptor for nongenomic signaling. MR EF sufficed for aldosterone-induced ERK1/2-activation via cSRC and EGFR. In analogy to the full-length MR, we suggest that MR EF transactivates EGFR at the cell membrane, employing cSRC, and thereby inducing the phosphorylation of ERK1/2 (9,15,39). At the moment, the minimum amino acid motif required for nongenomic MR signaling and its precise  mode of interaction with the signaling cascade are not known. Surprisingly, MR EF expression per se enhanced ERK1/2 phosphorylation in the absence of aldosterone, pointing at a ligandindependent action of the domain EF. Possibly MR EF spontaneously acquires a conformation sufficient for partial ERK1/2 activation, which is normally prevented by domain D in the absence of ligand. Altogether, these data open up the possibility for a model of domain-specific genomic-nongenomic interaction of MR. Domain EF is necessary for nongenomic activation of ERK1/2, which in turn seems to modulate MR transcriptional activity via the domain AB (9,18). MR DEF and MR EF elicit nongenomic but no classical genomic effects of aldosterone. Therefore, these truncated variants represent a tool to differentiate the consequences of nongenomic and genomic signaling and may help us to learn more about the pathophysiological relevance of nongenomic signaling. It has been shown that the EGFR-ERK1/2 signaling pathway plays a central role in cardiovascular fibrosis (53,54). Thus, the activation of this pathway by nongenomic MR signaling could contribute to its pathophysiological effects on cardiovascular tissue (4). We assessed the effect of aldosterone on two pathophysiological relevant parameters, i.e. activation of proinflammatory transcription factors (AP1, NFB) and collagen III abundance (41, 42, 46 -50). Aldosterone-induced transcription factor activation was observed only in cells expressing full-length MR, indicating the necessity of domain AB. In contrast, the synergistic action of aldosterone with low H 2 O 2 on (pro)collagen III abundance (i.e. sensitizing the cells for H 2 O 2 ) was observed in cells expressing MR, MR CDEF , MR DEF , and MR EF but not in cells expressing the vector alone. These data allow for the first time the ascription of a pathophysiological event to nongenomic MR signaling. The domain EF is necessary and sufficient to mediate the effect of aldosterone on ERK1/2 phosphorylation and collagen III abundance. In addition, the ligand-independent ERK1/2 phosphorylation in cells expressing MR EF already confers a basal sensitization for H 2 O 2 .
The data presented here were obtained in a heterologous expression system, and therefore the transfer to native tissue or the situation in vivo needs a note of caution. On the other hand, the heterologous expression system offers the possibility to compare the MR variants individually without perturbing endogenous MR. Through this, it was possible to perform a proof of principle and show for the first time that pathophysiological relevant effects of MR can be mediated by the nongenomic signaling pathway. Future challenges consist of further identifying the mechanisms underlying nongenomic signaling as well as in translating these molecular finding to the in vivo situation. An attractive perspective is the differential modulation of genomic and nongenomic signaling to develop therapeutical tools with fewer side effects.