Characterization of Rat NDRG2 (N-Myc Downstream Regulated Gene 2), a Novel Early Mineralocorticoid-specific Induced Gene*

The early phase of the stimulatory action of aldosterone on sodium reabsorption in tight epithelia involves hormone-regulated genes that remain to be identified. Using a subtractive hybridization technique on isolated renal cortical collecting ducts from rats injected with a physiological dose of aldosterone, we have identified an early response cDNA highly homologous to human and murine NDRG2 (N-Mycdownstream regulated gene2), which consists of four isoforms and belongs to a new family of differentiation-related genes. NDRG2 mRNA was expressed in classical aldosterone target epithelia, and in the kidney, it was specifically located in the collecting duct, the site of aldosterone-regulated sodium absorption. NDRG2 mRNA was increased within 45 min by aldosterone in the kidney and distal colon, whereas it was unaffected in the heart. In the RCCD2 collecting duct cell line, NDRG2 mRNA was enhanced as early as 15 min after aldosterone addition by transcription-dependent effects. NDRG2 was induced by aldosterone concentrations as low as 10−9 m, and a maximal effect was observed at 10−8 m. In contrast, the glucocorticoid dexamethasone was ineffective in NDRG2expression, whereas the glucocorticoid-regulated gene sgkwas induced. Taken together, these results indicate thatNDRG2 regulation by aldosterone is an early mineralocorticoid-specific effect. Interestingly, NDRG2 is homologous to Drosophila MESK2, a component of the Ras pathway, suggesting that activation of the Ras cascade may play a significant role in mineralocorticoid signaling.

Vertebrates must regulate salt and water excretion to maintain extracellular fluid volume, homeostasis, and blood pressure. The mineralocorticoid hormone aldosterone plays a major role in regulation of sodium absorption in tight epithelia such as the renal collecting duct, the distal colon, and the salivary and sweat glands (1,2). Its effects are mediated through the mineralocorticoid receptor, a member of the nuclear receptor superfamily that modulates transcription of mostly unknown target genes. This results in stimulation of sodium reabsorp-tion by increasing both the passive luminal entry of sodium into epithelial cells through the amiloride-sensitive epithelial sodium channel (ENaC) 1 and its active extrusion into the blood by the basolateral Na/K-ATPase. However, the hormone-elicited sodium reabsorption occurs before aldosterone-induced transcriptional regulation of ENaC and Na/K-ATPase subunits. Indeed, aldosterone stimulation of transepithelial sodium transport is a complex response (3). An enhanced apical sodium permeability is observed 30 -90 min after aldosterone addition, with a maximal effect several hours later. The early phase is thought to involve post-translational regulation such as methylation or phosphorylation of a pre-existing transport machinery. In particular, the activation of "silent" apical sodium channels mainly contributes to the initial increase in apical sodium influx. The late phase of aldosterone response (several hours to days) correlates with enhanced transcription/ translation of sodium channel subunits and Na/K-ATPase molecules. The aldosterone-induced regulatory proteins involved in the early phase of hormone response are mostly unknown.
The search for early corticosteroid-induced genes is the subject of intensive investigations. In amphibian A6 cells, induction of ASURs (adrenal steroid-up-regulated RNA) was evidenced. Among them, ASUR5, corresponding to the oncogene K-ras2, stimulates ENaC activity in the Xenopus laevis oocyte expression system (4) and increases sodium transport across A6 cell monolayers (5); however, its role in mammalian cells has not been documented. Attali et al. (6) identified in rat colon a dexamethasone-induced gene (CHIF, for channel-inducing factor) related to other regulatory proteins such as phospholemman and the ␥-subunit of Na/K-ATPase, but its function remains to be elucidated. Its expression is controlled by aldosterone in the colon, but not in the kidney (7). Substantial recent evidence has pointed to sgk (serum-and glucocorticoid-regulated kinase) as an aldosterone-induced gene in amphibian (8) as well as mammalian (8 -12) aldosterone target cells. Induction of sgk mRNA was reported as early as 30 min after treatment with a low dose of aldosterone as well as dexamethasone (10 -12). The encoded protein may play a role in mediating early aldosterone effects by stimulating ENaC-mediated sodium transport because coexpression of ENaC and Sgk in X. laevis oocytes increases the channel activity (8 -10). Serial analysis of gene expression of aldosterone-induced genes in a mouse collecting duct cell line revealed that a 4-h exposure to aldosterone results in large changes in the transcriptome be-cause 34 transcripts were induced and 29 were repressed (13). Among induced genes, GILZ (glucocorticoid-induced leucine zipper) appears to be an interesting induced candidate in this cellular model.
This study was designed to identify early target genes for aldosterone in native kidney cells and in a physiological context. For this purpose, we used a rat model in which glucocorticoid and mineralocorticoid plasma levels are controlled; such a model has been used for decades to describe early effects of aldosterone in vivo on renal Na ϩ and K ϩ excretion (14). Rats were adrenalectomized, supplemented with glucocorticoid hormone to maintain renal glomerular filtration rate at normal values, and injected acutely with a low dose of aldosterone. Then, a search for early aldosterone-regulated genes was performed by a subtractive hybridization approach conducted on a renal collecting duct preparation.
We have identified a member of the NDRG (N-Myc downstream regulated gene) gene family, NDRG2, as a putative effector of early effects of aldosterone. We show that the NDRG2 mRNA level was rapidly (45 min) enhanced in vivo by aldosterone in target epithelia such as the distal colon and kidney. In an aldosterone-sensitive cellular model derived from rat cortical collecting duct (RCCD2 cells), the NDRG2 mRNA level was increased as early as 15 min after aldosterone addition via transcription-dependent effects. Dose-dependent effects of aldosterone on NDRG2 expression showed that NDRG2 was induced with 10 Ϫ9 M aldosterone, and a plateau was reached at 10 Ϫ8 M hormone. In contrast, NDRG2 expression was not affected by the glucocorticoid dexamethasone, whereas the sgk mRNA level was strongly increased. Taken together, these results indicate that NDRG2, a member of a gene family related to cell differentiation, represents a novel mineralocorticoid-specific mediator of early physiological response to aldosterone in target tight epithelia.

EXPERIMENTAL PROCEDURES
Animals-Experiments were performed on Sprague-Dawley male rats weighing 160 -180 g. To select the conditions for differential cloning of early aldosterone-induced genes, urinary sodium excretion was measured after intravenous injection of a unique physiological dose of aldosterone. Rats were adrenalectomized and received a substitutive dose of dexamethasone (1 g/100 g of body weight/day) for 7 days via osmotic minipumps (1007D Alzet, Alza Corp., Palo Alto, CA) to prevent a fall in glomerular filtration rate. They were anesthetized with Inactin (Sigma), tracheotomized, and perfused (36 l/100 g/min) via a catheter introduced in the jugular vein for 1 h with a saline solution (65 mM NaCl, 4 mM KCl, 5 mM NaHCO 3 , 100 mM mannitol, and 52 mM glucose, pH 7.4). Aldosterone (0.2 g/100 g of body weight) was then injected as a bolus in the tail vein. Urine was collected via a catheter placed in the bladder before aldosterone injection and then every 30 min for 3 h and analyzed for Na ϩ content. As shown in Fig. 1, this experimental protocol resulted in a progressive decrease in the urinary sodium excretion rate, which was apparent 60 -90 min after hormonal injection and was maximal at 1.5-2 h. Because of these results, we chose to search for aldosterone-induced genes in rats injected with either aldosterone or vehicle (referred to as the control situation) and killed after 45 min. Other rats were killed 2 h after aldosterone injection.
Plasma levels of aldosterone from adrenalectomized rats, supplemented with dexamethasone and treated or not with aldosterone, were determined by radioimmunoassay (Table I). The plasma aldosterone concentrations from animals used either for subtractive hybridization or for RNase protection assays are given. 45 min after hormone injection, the plasma aldosterone concentration was increased compared with control rats (adrenalectomized and dexamethasone-supplemented). This rise was within the physiological range of hormone level variations (91-145 pg/ml, i.e. 0.2-0.4 nM) and was transient because it was no longer apparent 2 h after aldosterone injection.
Renal Cortical Collecting Duct Isolation and PCR-based Suppression and Subtractive Hybridization-Cortical collecting ducts (CCDs) were isolated using a Percoll gradient technique as previously described (16). Briefly, kidneys from control or aldosterone-treated rats (eight rats in each group, killed 45 min after intravenous injection of aldosterone or vehicle) were removed, and cortex slices were digested with collagenase (0.75 mg/ml for 1 h at 37°C; Serva) and then submitted to differential centrifugation on a Percoll gradient (40%). The lowest density band corresponded mainly to CCD fragments (25-40 mg), with minor contamination by glomeruli or proximal tubules. Poly(A) mRNAs were extracted from hormone-treated or control CCDs using oligo(dT) 25 covalently bound to magnetic beads (Dynal A.S., Oslo, Norway).
To identify up-regulated sequences, the two pools of mRNA fragments derived from control and aldosterone-treated animals were submitted to PCR-based subtractive hybridization and suppression PCR (PCR-Select TM cDNA subtraction kit, CLONTECH) according to the manufacturer's protocol. Double-stranded cDNA was synthesized and digested with RsaI. Adaptors were then linked to the aldosteronetreated cDNA pool, and two successive hybridizations followed by extension to fill in ends were performed in the presence of an excess of cDNA without linkers from untreated CCDs. The first PCR amplification using suppression PCR amplified exponentially only differentially expressed sequences. The second PCR amplification allowed us to reduce background levels and further enriched differentially expressed cDNA fragments. "Forward" subtraction was performed when linkers were added to cDNA fragments obtained from hormone-treated CCDs, whereas "reverse" subtraction was performed when linkers were added to cDNA fragments obtained from control CCDs. PCR products were cloned into the pT-Adv vector using the AdvanTAge TM PCR cloning kit (CLONTECH). Differential screening of the subtracted library was performed to eliminate false positives by hybridization with 32 P-labeled probes prepared from forward-and reverse-subtracted cDNAs (PCR-Select TM differential screening kit, CLONTECH). Clones showing signal ratios of Ͼ5:1 (forward-subtracted versus reverse-subtracted probes) were further analyzed by DNA sequencing (GENOME Express, Paris, France).
Cloning of Full-length Rat NDRG2-The full-length sequence of one expressed sequence tag (447 bp) was established by screening a rat kidney cDNA library (kindly provided by M. C. Lecomte, INSERM U409, Faculté de Médecine Xavier Bichat) by anchored PCR. We performed the first series of PCR with an antisense oligonucleotide (AS1) localized in the sequence of the expressed sequence tag and a sense oligonucleotide (S1) specific to the pGAD793 plasmid containing the library. To improve the specificity of the PCR products, we also carried out nested PCR with a second set of oligonucleotides, AS2 and S2. The different oligonucleotides were designed as follows: S1, 5Ј-CGATGAT-GAAGATACCCCACC-3Ј; AS1, 5Ј-TTAGGAGCATTGCCTTCAGAG-3Ј; S2, 5Ј-AGAGATCCTAGAACTAGTCGG-3Ј; AS2, 5Ј-AGGCACTGTGT-CAGCTTTGGGG-3Ј. PCRs were performed in 50 l of a mixture containing 1ϫ Advantage-2 PCR buffer, 200 M dNTPs, 50 pmol of each oligonucleotide (S1/AS1 or S2/AS2), 10 ng of rat kidney library, and 1ϫ Advantage-2 polymerase mixture (Advantage ® -2 cDNA polymerase mixture, CLONTECH). Cycles were as follows: 94°C for 5 min and then 94°C for 1 min, 54°C for 1 min, and 72°C for 1 min for 20 cycles. PCR products (ϳ740 bp) were sequenced and used to design an antisense oligonucleotide (GSP1, 5Ј-AGAGGGTCGCTCAACAGGAGACTT-3Ј) to clone cDNA 5Ј-ends by rapid amplification with the SMART TM RACE cDNA amplification kit (CLONTECH). Briefly, first-strand cDNA synthesis was performed with 1 g of total RNA from rat heart (a tissue strongly expressing NDRG2; see Fig. 4) with the SMART II oligonucleotide and Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's protocol. 5Ј-RACE/PCRs were realized with GSP1 and Universal Primer Mix oligonucleotides (cPontech) as indicated by the manufacturer's protocol. 5Ј-RACE products were characterized by sequencing (GENOME Express).
In Situ Hybridization-Rat tissues were fixed by perfusion via the aorta with 4% paraformaldehyde in phosphate-buffered saline (130 mM NaCl, 7 mM Na 2 HPO 4 , and 3 mM NaH 2 PO 4 , pH 7.4). Organs were dehydrated in graded alcohol solutions and embedded in paraffin (Paraplast). Hybridization was then carried out on tissue sections as previously described (17). The 3Ј-untranslated region of NDRG2 cDNA (nucleotides 1500 -1946 of NDRG2b 2 , DDBJ/GenBank TM /EBI accession number AJ426427) was subcloned into the XbaI-BamHI site of the pBluescript KS vector. 35 S-Labeled antisense and control sense probes were synthesized using 35 S-labeled UTP (specific activity of 1000 Ci/ mmol; Amersham Biosciences) and the Riboprobe ® combination system T3/T7 kit (Promega). Sense cRNA was synthesized after linearization with BamHI from the T7 promotor, and the antisense probe was synthesized after linearization with XbaI from the T3 promotor. Probes were then hybridized on tissue sections, and slides were developed after 27 days of exposure.
Real-time Quantitative PCR-Total RNA was extracted from RCCD2 cells with Trizol reagent, and 2 g were treated with DNase I (Invitrogen). The total RNA concentration was measured using the ultrasensitive fluorescent nucleic acid stain RiboGreen (RiboGreen ® RNA quantitation kit, Molecular Probes, Inc.) and a Model F-2000 spectrofluorometer (Hitachi, Ltd., Tokyo, Japan), which allow for accurate determination of the RNA concentration and standardization between samples (19). Reverse transcription was performed with 500 ng of total RNA using Superscript II reverse transcriptase and 500 ng of random hexamers (Amersham Biosciences) in a final volume of 20 l. Real-time PCR analysis of NDRG2 was carried out on an ABI7700 sequence detector (Applied Biosystems, Foster City, CA). The Taqman probe and primers (MWG Biotech) for NDRG2 were as follows: upper primer, 5Ј-TCCATTCCCCCAAAGCTG-3Ј; lower primer, 5Ј-CATCCAT-TTAGGAGCATTGCC-3Ј; and Taqman probe, 5Ј-FAM (6-carboxyfluorescin)-AACAGTGCCTTGACGTTTAAGGCCTCTGA-TAMRA (6carboxytetromethylrhodamine)-3Ј. In some experiments, the 18 S RNA level was measured as an internal control (upper primer, 5Ј-CCCTGC-CCTTTGTACACACC-3Ј; lower primer, 5Ј-CGATCCGAGGGCCTCACT-A-3Ј; and Taqman probe, 5Ј-FAM-CCCGTCGCTACTACCGATTGGAT-GGT-TAMRA-3Ј). PCR was performed with one-fifth (NDRG2) or onetwentieth (18 S RNA) of the reverse transcription reaction, 5 mM (NDRG2) or 3 mM (18 S RNA) MgCl 2 , 200 M dNTPs and 1.25 units of Taq polymerase. The final primer and probe concentrations were 400 and 100 nM, respectively. PCR reagents were from Eurogentec (qPCR core kit). The thermal cycling conditions comprised an initial denaturation step at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. Standard curves were generated using serial dilutions of a purified restriction fragment of NDRG2 (nucleotides 1500 -1946 of NDRG2b 2 ) or 18 S RNA (nucleotides 1428 -1788), covering 5 orders of magnitude and yielding correlation coefficients of at least 0.98 in all experiments. Each standard and sample values were determined in triplicate in three independent experiments. In each experiment, NDRG2 expression under a given experimental condition was calculated relative to the control condition.
Statistical Analysis-Data are expressed as means Ϯ S.E. (n ϭ number of animals). Statistical analysis was performed using Student's t test after analysis of variance.

Cloning of Rat NDRG2, a Member of the NDRG Gene Fam-
ily-PCR-based subtractive hybridization was used to establish a library of cDNAs representing mRNAs rapidly (45 min) up-regulated by a physiological dose of aldosterone in isolated rat CCDs. Selected clones were sequenced and analyzed for sequence homology using the BLAST program. About 50% of them corresponded to genes involved in oxidative metabolism (cytochrome c oxidase subunits) and the cytoskeleton (actin). One of the clones (clone 2, 447 bp) that was homologous to expressed sequence tag 239096 (DDBJ/GenBank TM /EBI accession number AI410803) was further studied.
The full-length sequence of clone 2 was obtained using RACE. Four different cDNAs were isolated. The cDNA sequences of rat NDRG2a 1 , NDRG2a 2 , NDRG2b 1 , and NDRG2b 2 have been deposited in the DDBJ/GenBank TM /EBI Data Bank under accession numbers AJ426424, AJ426425, AJ426426, and AJ426427, respectively. They share a common sequence over 1927 nucleotides and were shown to correspond to different mRNA isoforms. A new BLAST search revealed ϳ90% sequence homology to mouse and human NDRG2 cDNAs, mem-TABLE I Plasma aldosterone concentrations in control and aldosterone-treated rats Animals were adrenalectomized, supplemented with dexamethasone (1 g/100 g of body weight/day) for 7 days, and received a unique intravenous injection of vehicle (Control) or aldosterone (Aldo; 0.2 g/100 g of body weight) 45 min or 2 h before death. Values from rats used for subtractive hybridization (Series 1) or RNase protection experiments (Series 2) are shown (means Ϯ S.E., (n) ϭ number of animals). Series  bers of the NDRG family. The four isoforms have an identical 846-bp 3Ј-untranslated sequence containing a polyadenylation signal located 14 bp from the poly(A) tail, but they differ in their 5Ј-untranslated sequences, which are either 87 bp (NDRG2a) or 50 bp (NDRG2b) in length; furthermore, a 42-bp insertion in the coding sequence is either present (NDRG2a 1 and NDRG2b 1 ) or absent (NDRG2a 2 and NDRG2b 2 ) in these clones (Fig. 2). For NDRG2b 1 and NDRG2b 2 , the presumed initial ATG codon was assigned to the first Met codon because of an in-frame stop codon located 39 bp upstream from this ATG codon and a Kozak consensus sequence (ACCATGG) around this translation initiation site. For NDRG2a 1 and NDRG2a 2 , the same translation initiation site was predicted. The open reading frame is either 1113 bp (NDRG2a 1 and NDRG2b 1 ) or 1071 bp (NDRG2a 2 and NDRG2b 2 ) long. The 42-bp insertion encodes an additional 14 residues; the putative proteins contain 357 or 371 amino acids and present a high degree of homology to the murine and human NDRG2 proteins (ϳ90% identity). A search for the presence of consensus motifs in the amino acid sequences was performed using the Prosite, Pfam, and PSORTII programs. Fig. 3 shows a putative tyrosine kinase phosphorylation site (residues 302-309) and an ␣/␤-hydrolase fold predicted between amino acid 95 and 311. Interestingly, one putative nuclear localization signal was identified (residues 58 -64), and a nuclear localization for NDRG2 was predicted with a 56.5% versus 13% chance in the cytoplasm.
Tissue-specific Expression of NDRG2-We next investigated NDRG2 mRNA expression in different rat tissues by RPA using a probe common to the four isoforms. NDRG2 mRNA was strongly expressed in the heart, liver, skeletal muscle, and aorta and more weakly in the lung, distal colon, kidney, and bladder (Fig. 4A). NDRG2 mRNA was not detected in the small intestine and parotid. Fig. 4B shows the relative expression of NDRG2 in a large variety of tissues, including aldosterone target organs. These results indicate that NDRG2 is relatively ubiquitously expressed, although at different levels. Interestingly, two patterns of expression were identified: tissues with low levels of NDRG2 expression, including the lung, kidney, distal colon, and bladder, and tissues in which expression was ϳ10-fold higher such as the heart, aorta, skeletal muscle, and liver. To determine the NDRG2 mRNA localization at the cellular level, in situ hybridization experiments were performed in different tissues (Fig. 5). In the kidney cortex, the NDRG2 mRNA signal (antisense probe) was clearly seen in the distal tubule and CCD, with a low signal in the proximal convoluted tubule, as shown in Fig. 5 (A, C, and D). Glomeruli were negative. The outer medullary portion of the collecting duct (but not its papillary part) also showed positive labeling. Because the majority of collecting duct cells were positive (Fig. 5, C and D), NDRG2 is presumably expressed by principal cells, which are the most typical aldosterone-sensitive cells. NDRG2 mRNA was also found all along the epithelium of the distal colon (Fig. 6A), but not in the proximal colon (Fig. 6C). Among other epithelia, pulmonary bronchioles were positive, whereas alveoli were negative (Fig. 6E). Sense probes (Figs. 5B and 6, B, D, and F) gave background signals. A low signal was also present in the jejunum; in salivary glands, the ducts expressed NDRG2 mRNA, whereas acini were negative (data not shown).

NDRG2 Is Rapidly Induced by a Physiological Dose of Aldosterone in
Hormone-sensitive Epithelia-To examine aldosterone effects on NDRG2 expression, we performed RPAs in the kidney and distal colon from adrenalectomized and glucocorticoid-supplemented rats injected with aldosterone or vehicle (i.e. control condition). We were also interested in non-epithelial tissues such as the heart in which the hormone exerts effects by unknown mechanisms. Interestingly, aldosterone displays differential tissue-specific effects on NDRG2 mRNA expression. Fig. 7A illustrates the results from a typical RPA obtained with the NDRG2 probe hybridized with kidney RNA from control and aldosterone-treated rats for 45 min and 2 h. Fig. 7 (B-D) illustrates NDRG2 expression normalized to GAPDH and expressed as a percentage of controls in the kidney, distal colon, and heart, respectively. In the kidney, aldosterone induced a significant increase in NDRG2 mRNA expression 45 min after injection, and this increase was maintained at 2 h. A similar induction was observed in the distal colon, where NDRG2 expression was increased 45 min and 2 h after aldosterone injection. In contrast, NDRG2 mRNA expression was not modulated by aldosterone in the heart 45 min or 2 h after hormonal injection.
The mineralocorticoid receptor (MR) and the glucocorticoid receptor (GR) are coexpressed in aldosterone target cells (2,20), and aldosterone binds to both type of corticosteroid receptors (with a higher affinity for the MR than the GR). To address the question of the specificity of aldosterone effects on NDRG2 expression and to better characterize these effects, RPA and quantitative real-time PCR experiments were carried out in the rat cortical collecting duct cell line RCCD2. Characterization of the RCCD2 cell line has been recently reported (15). This cellular model has maintained many characteristics of the CCDs, including aldosterone-induced sodium transport. It expresses the MR and the GR, and low concentrations of aldosterone (10 Ϫ9 to 10 Ϫ8 M) induce an increase in the short-circuit FIG. 2. Nucleotide and amino acid sequences of the rat NDRG2 N-terminal region. NDRG2 consists of four isoforms: NDRG2a 1 , NDRG2a 2 , NDRG2b 1 , and NDRG2b 2 . Their first 45 amino acids are presented. Their 5Ј-UTRs are divergent: the 5Ј-UTR for NDRG2a 1 and NDRG2a 2 is 87 bp, whereas the 5Ј-UTR for NDRG2b 1 and NDRG2b 2 is 50 bp. The coding sequence is identical for the four different isoforms, except that NDRG2a 1 and NDRG2b 1 present an additional 42-bp insertion (14 amino acids; boxed), which is absent in the NDRG2a 2 and NDRG2b 2 sequences. Boldface numbers and numbers in parentheses correspond to the positions of nucleotides for NDRG2a 1 / NDRG2a 2 and NDRG2b 1 /NDRG2b 2 , respectively. Numbers on the right indicate the positions of amino acid residues relative to the initial methionine. Asterisks corresponds to the stop codon, and the Kozak sequence is underlined. current, a measure of transepithelial ion fluxes. Fig. 8A shows the aldosterone effects on NDRG2 expression in RCCD2 cells examined by RPA. The NDRG2 mRNA level was significantly enhanced in RCCD2 cells 45 min after treatment with 10 Ϫ8 M aldosterone. This rise appeared to be transient because 2 h after aldosterone treatment, NDRG2 mRNA expression returned to control cell values. Interestingly, treatment of RCCD2 cells with 5 ϫ 10 Ϫ8 M dexamethasone did not affect NDRG2 mRNA expression after 45 min or 2 h of hormonal treatment (Fig. 8A). To assess the GR functionality in this cellular model, we investigated the regulation of another gene (sgk) by dexamethasone. RCCD2 cells were treated with 5 ϫ 10 Ϫ8 M dexamethasone for 45 min, and NDRG2 and sgk expres-sion was then examined by real-time PCR (for NDRG2) and RPA (for sgk) on the same samples (Fig. 8B). Although the NDRG2 mRNA level was not modified by dexamethasone, sgk was strongly induced, and this effect was abolished in the presence of the GR antagonist RU486. These results indicate that although the GR is functional in RCCD2 cells, the glucocorticoid dexamethasone does not regulate NDRG2 expression. In addition, the dose dependence of the aldosterone effects on NDRG2 shows that NDRG2 was induced at physiological concentrations because the NDRG2 mRNA level was increased by 10 Ϫ9 M aldosterone, and a plateau was obtained with 10 Ϫ8 M hormone (Fig. 8C). Taken together, these results suggest that NDRG2 stimulation by aldosterone is a mineralocorticoidspecific response. Fig. 9A shows that NDRG2 is an early induced gene because its mRNA level was significantly increased as early as 15 min after aldosterone addition. This increase was still present at 1 h, and a return to the basal level was observed after 2 h. We examined whether transcription was required in the NDRG2 response to aldosterone using the transcription inhibitor actinomycin D (Fig. 9B). Preincubation of RCCD2 cells with actinomycin D 60 min before aldosterone addition prevented the rise in the NDRG2 mRNA level observed in the presence of aldosterone alone. Thus, NDRG2 regulation by aldosterone is transcription-dependent. DISCUSSION Early aldosterone effects on transepithelial sodium transport in tight epithelia are mediated by hormone-induced regulatory proteins, which are mostly unknown. The main finding of this study is that a physiological dose of aldosterone rapidly increases the mRNA level of NDRG2, a differentiation-related gene, in rat epithelial cells via a mineralocorticoid-specific effect.
In our search for rapid aldosterone-induced mRNAs, we have cloned and characterized four different cDNAs for rat NDRG2, the homolog of mouse and human NDRG2. These cDNAs code for the same putative NDRG2 protein except for the presence or absence of 14 residues in the N-terminal region, which are encoded by a 42-bp insertion (the calculated molecular mass is either 39.3 or 40.8 kDa). Comparison of the cDNAs with sequences contained in the Human Genome Database 2 indicates that this insertion corresponds to the inclusion of exon 3 of the human NDRG2 gene, suggesting an alternative splicing event. These two variants were also identified in the rat, as illustrated in Fig. 2. Interestingly, besides NDRG2b 1 and NDRG2b 2 , which share a common 5Ј-untranslated region (5Ј-UTR), we have identified two new NDRG2 variants, NDRG2a 1 and NDRG2a 2 , which contain a new 5Ј-UTR of 87 nucleotides. Sequences highly homologous to the two rat NDRG2 5Ј-UTRs were found on different exons in the human NDRG2 gene, indicating that they might correspond to alternative 5Ј-untranslated exons. This genomic organization strongly suggests the presence of alternative promotors that could direct expression of NDRG2 in a tissue-specific and developmentally regulated manner (21). It is noteworthy that a glucocorticoid-responsive element half-site (TGTTCT) was found in the human 2 Available at www.genome.ucsc.edu. NDRG2 promotor, corresponding to the sequence flanking the NDRG2 5Ј-UTR.
NDRG2 belongs to a new family of differentiation-related genes, the NDRG family. This family includes four recently identified related members: NDRG1-4 (22). NDRG1 was first seen in various tissues in different species (23)(24)(25)(26)(27)(28). Two other members of the family were then identified in the mouse: Ndr2 and Ndr3 (29). A recent report by Zhou et al. (22) documents the human NDRG gene family with characterization of three isoforms of NDRG4, the equivalent of rat BDM1 (30). Human NDRG members are highly homologous except in their C-and N-terminal regions. They exhibit distinct patterns of expression during development and adult life. NDRG1 is a widely expressed gene; NDRG2 and NDRG3 are essentially expressed in the brain, heart, skeletal muscle, and kidney, and NDRG4 expression is restricted to the brain and heart. This variable tissue-specific expression in the NDRG gene family suggests that each member may exert a particular role in different organs.
The functional role of NDRG genes remains to be established. Phylogenetic analysis revealed that NDRG genes are highly conserved in plants, invertebrates, and mammals, sug-FIG. 5. In situ hybridization of rat NDRG2 mRNA in the kidney cortex. Tissue sections were hybridized with a 35 S-labeled antisense (A, C, and D) or sense (B) probe. The hybridization signal (antisense probe) was apparent in the distal tubules and collecting ducts, whereas glomeruli (G) were negative. A low signal was present in the proximal convoluted tubules (pt). The background signal (sense probe) was very low. Asterisks indicate distal nephron segments (distal tubules, connecting tubules or CCDs), which cannot be distinguished by their morphology on such preparations; tubules at the close vicinity of the glomerulus are likely distal tubules. Bar ϭ 20 m. A and B, distal colon; C and D, proximal colon; E and F, lung sections with bronchioles and alveoli. A positive hybridization signal (antisense probe) was apparent in the epithelium of the distal (not proximal) colon (arrow), with a similar intensity in crypt cells and the surface epithelium. In the respiratory tract, the bronchiolar epithelium (*) was positive, whereas no specific signal was detected in alveolar cells. Sense hybridization showed low labeling in each of these tissues. Bar ϭ 20 m.
gesting important functions of this gene family (22). NDRG1 was initially designated as RTP (reducing agent-and tunicamycin-responsive protein), and its expression was shown to be regulated by homocysteine in cultured human umbilical vein endothelial cells (23). RTP was then isolated from various human and murine tissues (and referred to as Drg1, rit42, and Cap43 for the human homolog (24 -26) and TDD5 and Ndr1 for the murine homolog (27,28)). NDRG1 expression is regulated by chemical stimulation, which may impose a stress on the cells (23,26,31). Accumulated data suggest that NDRG1 plays a role in growth arrest and cell differentiation and could act as a signaling protein shuttling between the cytoplasm and the nucleus. Indeed, NDRG1 expression is up-regulated during cell differentiation (24,(31)(32)(33)(34) and repressed during cell transformation (24,25) and by N-myc or c-myc (28), known to inhibit terminal differentiation and to stimulate cell proliferation. Moreover, anti-oncogenic effects have been proposed: forced expression of NDRG1 in tumor cells reduces cell growth, increases cell differentiation, and reduces the metastatic potency of the cells (25,35). Interestingly, it has recently been shown that a mutation of the NDRG1 gene is causative for hereditary motor and sensory neuropathy-lom, suggesting a role of NDRG1 in the peripheral nervous system (36). Concerning the other members of the family, no information exists on NDRG2, NDRG3, and NDRG4 functions. Their pattern of expression is under spatiotemporal regulations that differ from NDRG1, suggesting specific functions (29).
In this report, we show that rat NDRG2 mRNA is expressed in several aldosterone-sensitive epithelia, including renal CCDs, epithelial cells of the distal but not proximal colon, pulmonary bronchioles, and urinary bladder and also in the heart, aorta, and skeletal muscle. In these latter two tissues, MR expression has been documented, and aldosterone effects have been proposed via unknown signaling cascades (37). Interestingly, aldosterone-dependent up-regulation of rat NDRG2 was evident only in epithelial target cells (kidney and distal colon) and not in the heart. This suggests that the early aldosterone response may be mediated by different mechanisms or effectors among target cells, as expected from their distinct functional characteristics. It is noteworthy that a similar lack of effect of the hormone on Sgk expression in the heart has been recently reported (12). The aldosterone-elicited increases in NDRG2 mRNA in the kidney and colon are relatively moderate (45-65%). Aldosterone-related changes of such limited amplitude have been previously described for transcripts encoding different subunits of the sodium channel (18,38,39) or the Na/K-ATPase (17,18). Such effects may be amplified at the protein level. It is also likely that other regulatory mechanisms or post-transcriptional effects of the hormone may be involved in the physiological response to aldosterone. We also provide evidence that the aldosterone-mediated increase in NDRG2 mRNA is transcription-dependent in the RCCD2 cell line (Fig. 9B). This transcription effect is rapid, as it was observed as early as 15 min after aldosterone addition (Fig.  9A). Such rapid transcriptional effects have been previously reported for Sgk induction by aldosterone.
Aldosterone target cells coexpress the MR and the GR (20). Aldosterone binds to the MR with high affinity (K d ϭ 10 Ϫ10 to 10 Ϫ9 M) and also to the GR with a lower affinity (K d ϭ 10 Ϫ9 to 10 Ϫ8 M). Thus, low concentrations of aldosterone (Ͻ10 Ϫ9 M) will bind to the MR, whereas higher concentrations of the hormone (Ͼ10 Ϫ9 M) will occupy both the MR and the GR. These two receptors regulate transcription of different (but partially overlapping) networks of genes. For this reason, it has often been difficult to distinguish between MR-and GR-mediated effects, especially because early receptor-specific responses are largely unknown. In many studies devoted to the search for early aldosterone-regulated genes, high doses of the hormone were used (4,8); and thus, the specificity of the response was difficult to evaluate. The experiments performed in the RCCD2 cell line FIG. 7. Aldosterone effects on NDRG2 mRNA expression in the kidney, distal colon, and heart. A, RPA was performed in kidneys from control rats (Control; i.e. adrenalectomized and supplemented with dexamethasone) or from rats injected with a unique physiological dose of aldosterone (0.2 g/100 g of body weight) 45 min (Aldo 45 min) or 2 h (Aldo 2h) before they were killed. Total RNA from kidney and yeast tRNA (40 g/lane) were hybridized with NDRG2and GAPDH-specific probes. Each lane corresponds to a different rat. MW, molecular weight markers. The protected fragments have expected sizes of 447 nucleotides for NDRG2 and 164 nucleotides for GAPDH. No background signal was detected with tRNA. B-D, quantification of NDRG2 mRNA expression in the kidney, distal colon, and heart, respectively, from control and aldosterone-treated rats was performed using an Instant Imager. NDRG2 mRNA expression was normalized to the internal standard GAPDH and expressed as a percentage of the control condition (without hormone). Values are means Ϯ S.E. of three independent experiments (n ϭ three rats in each group). *, p Ͻ 0.05; §, p Ͻ 0.025 (aldosterone versus control). allowed us to progress in the characterization of aldosterone effects on NDRG2 expression. Here, we have demonstrated that the glucocorticoid dexamethasone did not affect NDRG2 expression, although the GR was functional in the RCCD2 cells because sgk transcripts were induced by dexamethasone (Fig.  8, A and B). Although dexamethasone binds to the MR with an affinity close to that of aldosterone, this glucocorticoid hormone appears to be poorly efficient in promoting MR transactivation activity (40). Thus, the dexamethasone-induced increase in sgk mRNA is likely a GR-mediated (not MR-mediated) event. We also showed that NDRG2 mRNA was up-regulated by low doses of aldosterone (Fig. 8C). The absence of NDRG2 regulation by dexamethasone and the capacity of physiological concentrations of aldosterone to induce NDRG2 strongly suggest that the aldosterone-induced modulation of NDRG2 is a mineralocorticoid-specific response. Thus, this is the first characterization of an early mineralocorticoid-specific regulated gene in epithelial cells.
The existence of this family of genes and their conservation through evolution suggest an important biological role; however, their involvement in cell function and disease develop-ment remains to be elucidated. It is noteworthy that in A6 epithelia, aldosterone down-regulates the mRNAs of the protooncogenes c-myc, c-jun, and c-fos, which play a major role in stimulation of cell growth and inhibition of terminal differentiation (41). These facts may be related to our results demonstrating that aldosterone up-regulates the mRNA of a gene that belongs to a family referred to as "N-Myc downstream regulated genes," putatively involved in growth arrest and induction of cell differentiation. Along this line, a recent study reports that aldosterone favors very early adipocyte differentiation in a brown adipocyte cell line (derived from a hibernoma of transgenic mice overexpressing the large T antigen of SV40 under the control of the proximal human MR promotor) (42). Hence, new aspects of the pleiotropic action of the hormone are emerging, especially in terms of regulation of target cell differentiation status. The ability of aldosterone to favor cell differentiation can be understood in view of the fact that enhanced transepithelial sodium transport capacity requires multiple cellular processes occurring only in highly differentiated cells.
Interestingly, a search for known structural or functional domains within the deduced NDRG2 amino acid sequence re-  ). B, shown is NDRG2 and sgk expression after dexamethasone treatment in RCCD2 cells. RCCD2 cells were grown on filters and incubated with dexamethasone (5 ϫ 10 Ϫ8 M) in the absence or presence of RU486 (10 Ϫ6 M) for 45 min. NDRG2 was measured by quantitative real-time PCR, normalized to the internal standard 18 S RNA, and expressed as a percentage of the control condition (in the absence of any treatment). sgk was evaluated by RPA, normalized to GAPDH, and expressed as a percentage of the control condition. NDRG2 and sgk were determined on the same filters, and values are means Ϯ S.E. of three independent experiments. §, p Ͻ 0.025 (dexamethasone versus control). C, shown is the dose dependence of aldosterone effects on NDRG2 mRNA. RCCD2 cells grown on filters were incubated with or without various concentrations of aldosterone (10 Ϫ11 to 10 Ϫ7 M) for 30 min. NDRG2 mRNA was measured by quantitative real-time PCR, normalized to 18 S RNA, and expressed as a percentage of the control condition (in the absence of aldosterone). Values are means Ϯ S.E. of three independent experiments, with each experiment performed in triplicate. §, p Ͻ 0.025; **, p Ͻ 0.010 (aldosterone versus control).
vealed homology to the ␣/␤-hydrolase fold (InterPro entry IPR000073), which is common to a number of hydrolytic enzymes of widely differing phylogenetic origin and catalytic function, suggesting that NDRG2 may have an enzymatic function. It is noteworthy that the human NDRG2 gene has 34% identity to MESK2 (misexpression suppressor of dominantnegative KSR (kinase suppressor of Ras)), a gene recently identified in a misexpression screen conducted in Drosophila to search for genes that modulate the Ras1 signaling pathway (43). MESK2 may represent a new component of the Ras pathway or of other signaling pathways that can modulate signaling by Ras. The MAPK cascade is a central signaling module through which the small GTPase Ras, in response to diverse stimuli, exerts control over cell growth, cell differentiation, or apoptosis (44,45). Activated MAPK (via phosphorylation processes) regulates the activities of cytoplasmic and nuclear tar- FIG. 9. Time course of aldosterone effects on NDRG2 expression in rat RCCD2 cells. A, cells grown on filters were incubated with or without 10 Ϫ8 M aldosterone for various times. Expression of NDRG2 was studied by quantitative real-time PCR. The total RNA concentration was measured using the ultrasensitive fluorescent nucleic acid stain RiboGreen, which allows an accurate determination of RNA concentration and standardization between samples. Relative induction of NDRG2 was expressed as a percentage of the control condition (in the absence of aldosterone). Values are means Ϯ S.E. of three independent experiments, with each experiment performed in triplicate. *, p Ͻ 0.025; **, p Ͻ 0.005; ***, p Ͻ 0.001 (aldosterone versus control). B, shown is the influence of actinomycin D on NDRG2 induction by aldosterone in RCCD2 cells. Cells grown on filters were preincubated with or without actinomycin D (Actino; 2 ϫ 10 Ϫ6 M) for 1 h and then treated or not with aldosterone (Aldo; 10 Ϫ8 M) for 30 min (in the presence or absence of actinomycin D). NDRG2 was measured by quantitative real-time PCR, normalized by 18 S RNA, and expressed as a percentage of the control condition. Values are means Ϯ S.E. of three independent experiments, with each experiment performed in triplicate. **, p Ͻ 0.005 (aldosterone versus control). gets, which in turn lead to specific cellular responses. It is interesting to note that two recently identified early corticosteroid-induced regulatory proteins, K-Ras2 and Sgk, are directly or indirectly part of the Ras signaling pathway. Indeed, Sgk is a target of 3-phosphoinositide-dependent kinase-1, which itself is activated by the products of phosphatidylinositol 3-kinase. Because Ras is an upstream regulator of phosphatidylinositol 3-kinase, in some circumstances, Sgk activity may depend on Ras activity. Although K-ras2 mRNA was not found in rat collecting duct 3 other members of the Ras family may be involved in aldosterone action in mammalian cells. Thus, the regulatory network controlled by aldosterone is highly complex. Its early response involves induction of a number of effectors that are the site of convergence of different intracellular cascades and that contribute in synergy to sodium transport regulation. The elucidation of these signaling pathways should bring new insights into the understanding of corticosteroid hormone action in epithelia and its dysfunctions in diseases.