D. Kültz, Physiological Genomics Group, Department of Animal Science, University of California, Davis, One Shields Avenue, Davis, CA 95616, USA Fax: +1 530 752 0175 Tel: +1 530 752 2991 E-mail: firstname.lastname@example.org
We recently cloned a novel osmotic stress transcription factor 1 (OSTF1) from gills of euryhaline tilapia (Oreochromis mossambicus) and demonstrated that acute hyperosmotic stress transiently increases OSTF1 mRNA and protein abundance [Fiol DF, Kültz D (2005) Proc Natl Acad Sci USA102, 927–932]. In this study, a genome-wide search was conducted to identify nine distinct mouse transforming growth factor (TGF)-β-stimulated clone 22 domain (TSC22D) transcripts, including glucocorticoid-induced leucine zipper (GILZ), that are orthologs of OSTF1. These nine TSC22D transcripts are encoded at four loci on chromosomes 14 (TSC22D1, two splice variants), 3 (TSC22D2, four splice variants), X (TSC22D3, two splice variants), and 5 (TSC22D4). All nine mouse TSC22D transcripts are expressed in renal cortex, medulla and papilla, and in the mIMCD3 cell line. The two TSC22D3 transcripts (including GILZ) are upregulated by aldosterone but not by hyperosmolality in mIMCD3 cells. In contrast, TSC22D4 is stably upregulated by hyperosmolality in mIMCD3 cells and increased in renal papilla compared with cortex. Moreover, all four TSC22D2 transcripts are transiently upregulated by hyperosmolality and resemble tilapia OSTF1 in this regard. All TSC22D2 transcripts depend on hypertonicity as the signal for their upregulation and are unresponsive to increases in cell-permeable osmolytes. mRNA stabilization is the mechanism for TSC22D2 upregulation by hyperosmolality. Overexpression of TSC22D2–4 in mIMCD3 cells confers protection towards osmotic stress, as evidenced by a 2.7-fold increase in cell survival after 3 days at 600 mOsmol·kg−1. Based on variable responsiveness to aldosterone and hyperosmolality in kidney cells we conclude that mouse TSC22D genes have diverse physiological functions. TSC22D2 and TSC22D4 are involved in adaptation of renal cells to hypertonicity suggesting that they represent important elements of osmosensory signal transduction in mouse kidney cells.
In the mammalian kidney, the papilla is routinely exposed to severe hyperosmolality and to large changes in interstitial osmolality. These stressful conditions are a prerequisite for operation of the urinary concentrating mechanism and maintenance of systemic salt and water balance. Thus, renal papillary (and outer medullary) cells have special mechanisms to adapt to variable and severe hyperosmolality. Cellular adaptation to hyperosmotic stress is controlled via a complex array of cellular signaling mechanisms that modify gene and protein expression and protein function to promote osmoprotection . Such signaling mechanisms stimulate accumulation of the compatible organic osmolytes glycine-betaine, myo-inositol, taurine, sorbitol, and glycerophosphorylcholine [2–4]. Accumulation of glycine-betaine, inositol, and sorbitol is transcriptionally regulated and depends, at least in part, on the transcription factor tonicity-response element binding protein (TonEBP) . TonEBP also activates additional genes that are important for osmotic stress adaptation, including HSP70 and UT-A urea transporter [6,7]. In addition to the TonEBP pathway, hyperosmolality activates a very complex network of intracellular signaling pathways in renal medullary cells, including MAP kinase pathways , the p53 pathway , DNA-dependent protein kinases , and protein kinase A-dependent pathways . Thus, the response of mammalian kidney cells to hyperosmotic stress is highly complex and involves many different pathways and elements. Proper understanding of the cellular hyperosmotic stress response enabling computational modeling of this response is highly desirable because it would open avenues for manipulating stress-resistance networks of cells in states of renal disease and disorders of water and electrolyte balance. However, better knowledge about key elements of osmosensory signal transduction pathways and their interactions within osmotic stress signaling networks is required before in silico models that correctly reflect and predict cellular responses to osmotic stress can be devised.
We recently cloned a novel immediate early gene osmotic stress transcription factor 1 (OSTF1) that is involved in the cellular osmotic stress response of gill cells of euryhaline tilapia . In this fish, OSTF1 mRNA and protein levels rapidly and transiently increase in response to hyperosmotic stress, peaking at 2 and 4 h, respectively. The rapid and transient activation kinetics is characteristic of immediate early genes. OSTF1 belongs to the TSC22D family of leucine zipper proteins that are thought to be transcription factors in mammalian cells. In mouse tissues, TSC22D genes are regulated by glucocorticoids and transforming growth factor β (TGF-β) [13,14]. However, nothing is known about the osmotic regulation of any mouse TSC22D isoform. In addition, a systematic genome-wide analysis of mouse TSC22D gene products, identifying all family members, is lacking.
In this study, we identified nine murine TSC22D transcripts and investigated their regulation by hyperosmolality and aldosterone, which is a mineralocorticoid hormone important for modulation of the urinary concentrating mechanism. Moreover, TSC22D2 was identified as the closest functional mouse ortholog of tilapia OSTF1 and the mechanism and physiological significance of hyperosmotic upregulation of this gene was analyzed.
Identification of TSC22D family members in the mouse genome
We recently cloned tilapia OSTF1 and showed that it is a rapidly induced osmotic stress transcription factor . To identify possible functional homologs of tilapia OSTF1 in mammals, we carried out an exhaustive search of the complete annotated mouse genome using the ENSEMBL database (http://www.ensembl.org) . This search yielded six gene products with expectation values ranging from 6.1e-69 to 3.2e-21. These proteins are the products of transcripts encoded at four different loci (Table 1). In order to avoid ambiguity, we follow the recently updated and unified MGD nomenclature guidelines for TSC22D proteins in this study (Mouse Genome Informatics) . TSC22D1-1 and TSC22D1-2 are splice variants that are located on chromosome 14, TSC22D2 is located on chromosome 3, TSC22D3-1 and TSC22D3-2 are splice variants that are located on chromosome X, and TSC22D4 is located on chromosome 5 (Table 1). Although two of these proteins have been previously described as TSC-22 (TSC22D1-2) and glucocorticoid-induced leucine zipper (GILZ) (TSC22D3-2), the other four have not been characterized or only referred to as TSC22-like or GILZ-like proteins. Multiple sequence alignment shows that the six mouse proteins and tilapia OSTF1 share a conserved region of ≈ 70 amino acids, which comprises the TSC22D family signature motif and a leucine-zipper domain. The N- and C-termini are least conserved in all proteins. In particular, N-termini are highly heterogeneous, accounting for variability in total protein lengths ranging from 124 to 1057 amino acids (Table 1, Fig. 1). The protein with the highest overall sequence similarity to tilapia OSTF1 is TSC22D3-1, based on highest degree of conservation of the N-terminus (Fig. 1).
Expression of TSC22D family members in kidney mouse and mIMCD3 cells
We analyzed the expression of the six mouse TSC22D transcripts in kidney to learn whether any of them functionally resembles tilapia OSTF1. Levels of expression of the six transcripts were determined by quantitative PCR in three regions of the kidney that are characterized by increasing interstitial osmolality in the order from cortex (lowest) to medulla (intermediate) to papilla (highest). All six transcripts are expressed in all three regions of the kidney. Renal TSC22D2 is most abundant being expressed at levels that are between one and two orders of magnitude lower than that of the highly abundant ribosomal protein L32 (Fig. 2). The level of expression of TSC22D1-2 and TSC22D2 is similar in cortex, medulla, and papilla (Fig. 2). However, TSC22D3-1, TSC22D3-2, and TSC22D4 are significantly more abundant in papilla, whereas TSC22D1-1 is more abundant in cortex. The data suggest that hyperosmolality could potentially be responsible for altering the expression of four TSC22D transcripts. The level of expression of all six transcripts was also determined in mIMCD3 cells. All six transcripts are expressed in mIMCD3 cells and expression levels are similar to those in mouse kidney medulla in vivo (data not shown). Therefore, mIMCD3 cells are a good model for evaluating mechanisms of regulation of the mouse TSC22D transcripts.
Regulation of TSC22D transcripts in mIMCD3 cells by hyperosmotic stress and aldosterone
The responsiveness of TSC22D transcripts to hyperosmotic stress and/or aldosterone treatment was determined in mIMCD3 cells in 24-h time course experiments. Acute hypertonicity increases the expression of TSC22D2, TSC22D4 and TSC22D3-2. Of interest, TSC22D2 is elevated early and transiently, showing increases of 2.6- and 3.1-fold at 4 and 6 h of treatment, respectively, and returning to baseline levels within 12 h. In contrast, TSC22D3-2 and TSC22D4 show a slower but more stable upregulation, increasing three- and sixfold, respectively, after 24 h of treatment (Fig. 3). These results are in agreement with higher levels of TSC22D3-2 and TSC22D4 in renal papilla in vivo (see previous paragraph, Fig. 2). Aldosterone induces a rapid increase in TSC22D3-2 (4-fold at 1 h, 33-fold at 12 h, 10-fold at 24 h) and TSC22D3-1 (fivefold at 4–6 h hours) (Fig. 3). Of interest, a combination of hyperosmotic stress and aldosterone does not potentiate the transient increase in TSC22D3-2 (Fig. 3). By contrast, hyperosmotic stress and aldosterone in combination prevent transient short-term effects and offset each other. Taken together, the data on osmotic regulation of TSC22D transcripts implicate TSC22D2 as the closest functional homolog of tilapia OSTF1.
Identification of alternative TSC22D2 transcripts
Because of its similar osmotic regulation compared with tilapia OSTF1 we investigated mouse TSC22D2 in more depth. Two additional alternative transcripts encoding splice variants of TSC22D2 protein were identified that differed from the original cDNA ENSMUST00000029383 (TSC22D2-1; Fig. 1,Table 1). These two additional cDNAs (GENSCAN00000073255 = TSC22D2-2 and ENSMUSESTG00000010047 =TSC22D2-3) were predicted using the Ensembl database and gene prediction software genscan and genomewise/genewise. genscan is a bioinformatic tool that predicts gene loci and their exon/intron composition based on the genomic DNA sequence . genomewise/genewise gene-prediction software assembles cDNA sequences based on the analysis and integration of EST data . Taking advantage of information provided by these two complementary approaches we thoroughly examined the TSC22D2 gene for alternative splicing events. Alignment of the three identified TSC22D2 splice variants against the genomic TSC22D2 sequence revealed differences in exon composition. Two splice variants (TSC22D2-1/2) consist of three exons, whereas the third splice variant (TSC22D2-3) has four exons as a result of inclusion of an extra 72 bp exon in the second position (Fig. 4A). The length of the first and last exons is also variable in the three splice variants of TSC22D2 (Fig. 4A).
We then tested for expression of the newly predicted TSC22D2 transcripts (TSC22D2-2/3) in mouse kidney cells. Specific PCR primer pairs were designed to amplify TSC22D2-2 (primer pair E–F), TSC22D2-3 (primer pair A–C), and all splice variants (primer pairs A–B and A–D). We had already used primer pair A–B for previous quantification of overall TSC22D2 transcript abundance as it amplifies all possible splice variants (Fig. 4A, Table S1). Expression of TSC22D2-2 and TSC22D2-3 was confirmed based on the presence of RT-PCR products having the expected sizes (Fig. 4B, lanes A–C and E–F, respectively). In addition, using the primer pair A–D we detected three different PCR products of 493, 406 and 334 bp instead of the two products that we expected based on the primer design shown in Fig. 4A (amplicon ± exon 2). Therefore, the three PCR products obtained with primers A–D were purified, sequenced, and aligned to each other (Fig. 4C). The sequence of two of these PCR products matched the predicted sequence for TSC22D2-1/2 and TSC22D2-3 (Fig. 4C). These sequences differed by the presence of the 72-bp exon 2 in TSC22D2-3 as predicted.
Surprisingly, however, an additional unpredicted fragment was discovered by PCR analysis (TSC22D2-4, Fig. 4). Sequencing of the corresponding PCR product confirmed that TSC22D2-4 represents an entirely novel splice variant that was not predicted by any of the bioinformatics methods used in our study nor reported to exist previously. TSC22D2-4 included an alternative second exon of 159 bp but lacked the 72 bp exon 2. Schematic exon/intron structures of all four TSC22D2 splice variants are compared in Fig. 4D with emphasis on the two alternative exons 2A (72 bp) and 2B (159 bp), which are not present simultaneously in any TSC22D2 transcript in mIMCD3 cells (Fig. 4B).
Next, we analyzed the exon/intron regions flanking TSC22D2 exons 2A and 2B. All of these sequences match splice donor and acceptor consensus sites very well (5′-AG/GT AG/G-3′) (Table 2). In addition, the homologous intron/exon regions that flank exons 2A and 2B in human TSC22D2 are 95% identical to mouse sequences indicating a high degree of conservation of these critical areas compared with the overall much lower homology of TSC22D2 genomic sequence (< 50%; data not shown). Taken together, these observations strongly support alternative splicing events that give rise to TSC22D2 transcripts with different exon 2 sequences.
Table 2. Sequences corresponding to 3′ acceptor and 5′ donor exon/intron boundaries in TSC22D2 transcripts.
Exon 1 ...AGACAG/gtatgtaca...
...gtctcacag/GAATCC… Exon 2 A
Exon 1 …AGACAG/gtatgtaca…
…ctttgctag/AATTTT… Exon 2B
Exon 1 …AGACAG/gtatgtaca…
…tttttccag/TGCATC… Exon 3
Exon 2 A …GGATAG/gtatgatta…
…tttttccag/TGCATC… Exon 3
Exon 2B …AAATTG/gtaagactt…
…tttttccag/TGCATC… Exon 3
Exon 3 …GCAATG/gtaagtagg…
…tcttcacag/GATCTG… Exon 4
Protein products for TSC22D2-1 and TSC22D2-2 differ only by variable length of the first and last exons from each other (Fig. 4A). In contrast, TSC22D2-3 and TSC22D2-4 differ more substantially from the other TSC22D2 variants because of the presence of an additional exon (exon 2A/2B) (Fig. 4E). In particular, TSC22D2-4 differs greatly from the other variants because it lacks a large portion of the N-terminus due to the presence of four in-frame stop codons in exon 2B (Fig. 4C,E). An ATG codon following immediately after the last of these four stop codons may represent the transcription initiation site for a protein with a much shorter N-terminus (Fig. 4E). Each of the four possible TSC22D2 protein products also differs with respect to the presence of consensus phosphorylation sites for a number of stress-responsive protein kinases (Fig. 4E).
Response of the four TSC22D2 variants to hyperosmolality and aldosterone
We quantified abundances of individual TSC22D2 transcripts by quantitative PCR using the PCR primers depicted in Fig. 4A and Table 1. All four TSC22D2 variants are expressed at comparable levels in mouse kidney papilla, medulla, and cortex (data not shown). To analyze the regulation of the four TSC22D2 variants in response to hyperosmolality and aldosterone we exposed mIMCD3 cells to either of those stimuli alone and to a combination of both. All four TSC22D2 variants are transiently upregulated by hyperosmotic stress (Fig. 5). The highest degree of hyperosmotic upregulation was observed for TSC22D2-4. Aldosterone with or without hyperosmolality did not significantly affect the abundance of any individual TSC22D2 variant, consistent with the results obtained when all TSC22D2 variants were quantified together (Figs 3, 5).
Regulation of TSC22D2 variants by hyperosmolality
To identify the signal responsible for hyperosmotic upregulation of TSC22D2 we exposed mIMCD3 cells for 5 h to hyperosmotic media (550 mOsmol·kg−1) prepared by addition of NaCl, choline chloride, sodium gluconate, mannitol, urea or glycerol. TSC22D2-4 was always upregulated by hypertonic media (choline chloride, sodium gluconate, mannitol, NaCl) independent of the presence of Na+ or Cl– in such media (Fig. 6). In contrast, hyperosmolality due to nonhypertonic glycerol or urea did not alter TSC22D2-4 levels (Fig. 6). Similar results were obtained for the other three TSC22D2 variants (data not shown). These data demonstrate that neither Na+ nor Cl– nor hyperosmolality per se represent the signal for upregulation of TSC22D2. Instead, hypertonicity is the stimulus that triggers TSC22D2 upregulation.
mRNA stabilization of TSC22D2 during hyperosmotic stress
Next, we analyzed the mechanism of TSC22D2 upregulation in response to hyperosmotic stress. Transcription in mIMCD3 cells was completely blocked by a 1 h preincubation in 10 µm actinomycin D. Even 5 µm actinomycin was sufficient to effectively block transcription in mIMCD3 cells (Fig. S2). Cells were then exposed to hyperosmotic stress, aldosterone, and control conditions (isosmotic medium). The half-life of TSC22D2-4 was 2.8 ± 0.2 h for controls and aldosterone treatment but increased to > 20 h as a consequence of hyperosmolality (Fig. 7). TSC22D2-1/2 and TSC22D2-3 responded similarly, with half-lives increasing from 2.8 ± 0.3 to > 10 h and 2.3 ± 0.3 to > 15 h, respectively, in response to hyperosmolality (data not shown). These results indicate that mRNA stabilization is the mechanism responsible for hyperosmotic upregulation of TSC22D2 transcripts.
Osmoprotection of mIMCD3 cells by overexpression of TSC22D2-4
To evaluate whether TSC22D2 upregulation protects cells from hyperosmotic stress we generated stably transfected mIMCD3 cells that overexpress TSC22D2-4. We first generated a mIMCD3 cell line with a Flp recombinase target site stably integrated into the genome (mIMCD3FRT cells; Fig. S1) to generate a good control for future experiments with stably selected cells. We then cotransfected V5-epitope-tagged TSC22D2-4 in pcDNA5FRT vector together with a Flp recombinase expression vector to insert TSC22D2-4 into the FRT site in exchange for the LacZ gene. The transgenic TSC22D2-4 cell line expressed ≈ 100-fold higher levels of TSC22D2-4 compared with mIMCD3FRT control cells (Fig. 8A). A single protein with the expected molecular mass (17 kDa) was detected in TSC22D2-4 cells using V5 antibody (Fig. 8B).
The transgenic TSC22D2-4 cells showed significantly greater hyperosmotic stress tolerance than mIMCD3FRT control cells. We incubated these two cell lines for 24 h under hyperosmotic stress conditions that lead to a high frequency of apoptosis in wild-type mIMC3 cells (600–650 mOsm·kg−1) . Under these conditions, TSC22D2-4 cells had a significantly improved phenotype (Fig. 9A) and cell numbers were significantly higher compared with mIMCD3FRT control cells (Fig. 9B), indicating that high levels of TSC22D2-4 protect cells during hyperosmotic stress.
Mammals have four loci encoding at least nine TSC22D transcripts
We have identified four loci in the mouse genome that encode nine homalogs of the tilapia osmotic stress transcription factor OSTF1. All four genes belong to the TSC22D family of leucine zipper proteins that form homo- and heterodimers with other family members. Four TSC22D isoforms have previously been described: TSC22D1-2 (TSC-22), TSC22D3-2 (GILZ), TSC22D3-1 and TSC22D4.
TSC22D1-2 was first isolated based on rapid and transient transcriptional induction by TGF-β1 . It also increases in response to anticancer drugs, progesterone, and growth inhibitors  and has been implied in mechanisms of tumorigensis.
TSC22D3-2 was identified as a protein that is induced following the treatment of thymocytes with dexamethasone . Its mRNA increases threefold as early as 30 min and by more than 10-fold within 4 h of aldosterone exposure in principal cells of the renal cortical collecting duct . In contrast, it is downregulated by estrogen in MCF-7 human breast cancer cells  GILZ interacts with NF-κB and Raf and inhibits AP-1, FoxO3, and Raf-mediated apoptotic pathways [23–25]. This protein mediates aldosterone actions by stimulation of trans-epithelial sodium transport in kidney .
TSC22D3-1 was identified in porcine brain as a 77 kDa protein that shares immunoreactivity with the sequence-unrelated nonamer neuropeptide DSIP . It was later found to be the most highly glucocorticoid-induced cDNA among over 9000 tested in a cDNA gene chip array in human peripheral blood mononuclear cells .
TSC22D4 was identified in humans as a protein capable of forming heterodimers with TSC-22 . Its mouse homolog is involved in pituitary organogenesis .
In this study we identified five additional TSC22D transcripts that are encoded by genes located on chromosomes 3 (TSC22D2-1, TSC22D2-2, TSC22D2-3, TSC22D2-4) and 14 (TSC22D1-1). Although some of these novel transcripts have been previously described in the context of high-throughput cDNA sequencing projects [30,31] their functions are unknown. However, based on their sequence similarity to known TSC22D proteins they may be transcription factors that are involved in the regulation of cell proliferation, apoptosis, and stress response pathways.
All nine TSC22D transcripts are expressed in mouse kidney cells
Expression of all nine TSC22D transcripts was confirmed in mouse kidney and in the mIMCD3 cell line. The levels of expression of all nine TSC22D transcripts in mIMCD3 cells in vitro were comparable with renal tissue in vivo suggesting that mIMCD3 cells are a useful model for studying mechanisms of regulation and functions of TSC22D isoforms in mammalian kidney cells.
The lack of previous evidence for expression of several TSC22D transcripts identified in this study suggests that they may be particularly important for specific biological functions that are prevalent in renal cells. The multitude of alternatively spliced TSC22D2 gene products could be important for generating functional variability in response to different environmental cues. However, in the case of hyperosmolality all four splice variants of TSC22D2 are significantly upregulated even though the magnitude and kinetics of this upregulation was somewhat splice variant specific (see Discussion below) and the structures of the respective protein products are also different. Variable exon 2 usage produces proteins with different N-termini adjacent to the conserved TSC22D motif. This region is responsible for transactivation suggesting that TSC22D variants with truncated N-termini (in particular the novel TSC22D2-4 variant) may be transcriptional repressors that sequester other TSC22D family members .
TSC22D2 and TSC22D4 are regulated by hyperosmolality in kidney cells
In mIMCD3 cells exposed to hyperosmolality TSC22D2 and TSC22D4 transcripts increase significantly but with different kinetics. The increase in TSC22D2 transcripts is transient and closely resembles that observed previously for tilapia OSTF1 . Thus, despite the higher degree of structural homology of tilapia OSTF1 with murine TSC22D3-1, the novel murine TSC22D2 transcripts represent the closest functional homologs of tilapia OSTF1. The magnitude and kinetics of hyperosmotic upregulation of TSC22D2 splice variants shows some differences. TSC22D2-1 and TSC22D2-4 responded earlier and more robustly than TSC22D2-2 and TSC22D2-3.
Splice variants of other genes that respond differentially to osmotic stress have been reported before, e.g. for cyclooxygenase 1 in human intestinal epithelial cells . In addition such regulation has been observed for other types of stress. For instance, Drosophila heat shock transcription factor is regulated by alternative splicing in response to heat/cold stress . The splicing factor hSlu7 was reported to alter its subcellular distribution and thus modulate alternative splicing after UV stress . In fact, alternative splicing of pre-mRNA encoding transcription factors represents a common mechanism for generating the complexity and diversity of gene regulation patterns [35–38]. This mechanism produces a variety of functionally distinct isoforms from a single gene by use of different combinations of splice junctions. For example, alternative splicing within the DNA-binding domain of Pax-6 alters DNA-binding specificity of the resulting proteins . Alternative splicing of the transactivation domains in Pax-8 , the POU homeodomain family protein Pit-1  and the zinc finger transcription factor GATA-5  also results in protein isoforms with different transactivation properties. Deletion by splicing of the transactivation domain in AML1a  and CREB  produces proteins with dominant negative activity. This may also be the case for TSC22D2 splice variants with a truncated transactivation domain, in particular TSC22D2-4. Thus, alternative splicing of TSC22D2 may confer increased complexity of gene regulation in response to hyperosmotic stress. Our data indicate that TSC22D2-4 represents a survival factor for renal cells exposed to hyperosmolality suggesting that it promotes osmotic adaptation programs, possibly by acting as a transcriptional repressor of pro-apoptotic genes.
The time course of hyperosmotic induction of the murine TSC22D4 transcript is slower than TSC22D2, more stable, and more closely resembles that observed previously for TonEBP , although more transient hyperosmotic activation of TonEBP similar to that of TSC22D2 has also been reported recently . Moreover, significantly higher levels of TSC22D4 in renal papilla vs. cortex raise the possibility that this gene is stably upregulated by hyperosmolality not only in vitro but also in vivo. Of interest, AP1 (jun, fos) and NF-κB are transcription factors that are regulated by osmotic stress [47–52] and, intriguingly, they are known to interact with TSC22D3-2 (GILZ) [23–25].
TSC22D3 is regulated by aldosterone in kidney cells
Aldosterone is the major corticosteroid hormone regulating electrolyte and fluid homeostasis in all vertebrates [53,54]. The major action of the hormone on renal Na+ transport is localized to the collecting duct. Our results show that both TSC22D3 transcripts increase transiently in response to aldosterone treatment. Upregulation of TSC22D3-2 (GILZ) by a corticosteroid hormone has been observed previously in human lymphocytes , rat kidney , human peripheral blood mononuclear cells  and in a mouse kidney cortical collecting duct cell line . Upregulation of TSC22D3-1 by a corticosteroid hormone was observed in human peripheral blood mononuclear cells . Thus, TSC22D3 is a gene that is robustly upregulated by corticosteroid hormones in a wide variety of tissues, including kidney.
Our results indicate an antagonistic effect of aldosterone and hyperosmotic stress on the early responses of TSC22D2 and TSC22D3 transcripts. Aldosterone failed to induce TSC22D3 transcripts in the presence of hyperosmotic stress, and reciprocally, hyperosmotic stress failed to induce TSC22D2 transcripts in the presence of 1 µm aldosterone. This phenomenon seems restricted to the early responses (1–12 h) as long-term hyperosmotic effects on TSC22D3-2 and TSC22D4 transcripts were not altered by aldosterone. Consistent with these data, a hypertonic reduction of aldosterone-stimulated Na+ transport was reported in rat IMCD . The molecular mechanism of interaction between corticosteroid hormone-induced and hypertonic stress-induced pathways controlling expression of TSC22D isoforms remains unknown, but our results suggests that they involve common elements that are affected antagonistically by these two agents.
The TSC22D1 gene was unresponsive to either hyperosmolality or aldosterone treatment. Overall, our data suggest that the four TSC22D genes are not functionally redundant but involved in different aspects of cellular regulation that are triggered by distinct extracellular signals.
Hyperosmotic upregulation of TSC22D2 is triggered by hypertonicity and results from mRNA stabilization
We tested the effect of different hyperosmotic media to investigate the signal for TSC22D2 upregulation. Our results show that TSC22D2 is only elevated when hyperosmotic media are prepared with nonpermeable solutes. Hyperosmolality per se (resulting from elevation of cell-permeable solutes) was insufficient to elicit a response. Thus, we conclude that hypertonicity is the signal for TSC22D2 upregulation. We recently reported that tilapia OSTF1 hyperosmotic induction is also dependant on hypertonicity  and in mIMCD3 cells hypertonicity represents the signal for induction of mRNAs encoding the TonEBP transcription factor and multiple genes involved in compatible osmolyte accumulation, protein-, and DNA- stabilization [45,59].
The molecular nature of the hypertonicity signal is not yet known. Hypertonicity is known to cause many secondary effects including cell shrinkage, macro- and micromolecular crowding, changes in the organization of cell membranes, altered water movements across cell membranes (osmosis), and stress on the cytoskeleton . Such secondary effects are independent of the particular solutes responsible for hypertonicity and our results illustrate that there is no specific sodium or chloride ion requirement for TSC22D2 upregulation. Therefore, we conclude that one or more of the above-mentioned secondary effects associated with hypertonicity provide the sensory stimulus that triggers TSC22D2 upregulation.
Of interest, TSC22D2 isoforms responded to hypertonic stress even in the presence of the transcriptional repressor actinomycin D, indicating that they are regulated by mRNA stabilization. In concordance with these results, hyperosmotic upregulation of tilapia OSTF1 is also based on mRNA stabilization . In addition, mRNA stabilization was also observed in the regulation of GADD45 genes , TonEBP transcription factor  and aquaporin  in response to hypertonicity. mRNA stabilization is a regulatory mechanism involved in rapid responses to various forms of cellular stress, including heat shock , UV irradiation [63,64], hypoxia  and nutrient deprivation . This mechanism permits a rapid increase in steady state mRNA levels by preventing its degradation. It is characteristic of inducible transcription factors and other immediate early genes with high rates of mRNA turnover . Thus, stabilization of TSC22D2 mRNA during hypertonicity supports a regulatory role of its protein product for osmotic stress adaptation of renal cells.
Murine inner medullary collecting duct (mIMCD3) cells of passage 18 were used for all experiments . Cell culture medium consisted of 45% Ham's F-12, 45% DMEM, 10% fetal bovine serum, 10 mU·mL−1 penicillin, and 10 µg·mL−1 streptomycin (all reagents were from Invitrogen, Carlsbad, CA). Cells were grown at 37 °C and 5% CO2. Final medium osmolality of isosmotic medium was 300 ± 5 mOsmol·kg−1 of H2O. Hyperosmotic media were prepared by the addition of an appropriate amount of NaCl to isosmotic medium to yield the indicated osmolality. When specified, choline chloride, sodium gluconate, urea, mannitol or glycerol instead of NaCl were added for hyperosmotic media preparation. Final osmolality of all media was verified with a microosmometer (Model 3300, Advanced Instruments, Norwood, MA). Aldosterone (A9477; Sigma, St Louis, MO) was added when indicated to a final concentration of 1 µg·mL−1. Controls with vehicle (ethanol) were always run in parallel. In all experiments, medium was substituted 24 h before treatments with a hormone-free medium, where 10% dextran/charcoal-treated fetal bovine serum (Biosource, Rockville, MD, USA) replaced the 10% fetal bovine serum.
Animals and RNA isolation
C57/BL6 mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and a stock maintained at the ColeB small animal colony at UC Davis. Mice were kept on a normal mouse diet with water ad libitum. After culling mice using CO2, kidneys were dissected into papilla, medulla, and cortex and these tissues immediately snap-frozen in liquid nitrogen . All procedures were approved by the UC Davis Institutional Animal Care and Use Committee (IACUC).
Total RNA from mIMCD3 cells or renal tissues was extracted using Trizol reagent (Invitrogen) as specified by the manufacturer. RNA was treated with DNase (Turbo DNA free; Ambion, Austin, TX) and purity was confirmed and quantity determined by measuring absorbance of the samples at 260 and 280 nm (340 nm background values were subtracted) with a Beckman DU520 spectrophotometer.
cDNA synthesis and quantitative real-time PCR
RNA (2 µg for mIMCD3 cells and 0.5 µg for renal tissues) was reverse-transcribed using Superscript III first-strand synthesis reagents (Invitrogen) with a random hexamer/oligo(dT) mix (1 ng/μL : 1 μm) as primers. Abundance of all transcripts was quantified with a PRISM 7500 real-time thermal cycler (Applied Biosystems, Foster City, CA, USA). Reactions were performed in duplicate in 20 µL reaction volume using SYBR Green PCR Master Mix (Applied Biosystems) and 30 pmol of each primer. PCR conditions were 50 °C for 2 min and 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Data were collected at 60 °C. Efficiencies of individual PCR reactions were analyzed using LinRegPCR  and were always > 1.9. All data were normalized to abundance of L32 mRNA encoding a ribosomal protein, and expressed as fold change over controls as described previously . L32 mRNA was selected as a normalizer gene over 18S rRNA, β-actin mRNA and GAP-3-DH mRNA based on highly constant levels of expression during all conditions as determined in preliminaries assays. Gene-specific primer sequences were designed with primer express software (Applied Biosystems). The following sequences were used as templates for primer design: NM_172086.1(L32) from GenBank and AK007760 (TSC22D2), AF315352 (TSC22D4), AF201285 (TSC22D1-1), L25785 (TSC22D1-2), AF024519 (TSC22D3-2), AF201289 (TSC22D3-1) from EMBL. The absence of unwanted by-products was confirmed by automated melting curve analysis and agarose gel electrophoresis of PCR products.
Analysis of mRNA stability
After 24 h incubation in hormone-free medium, cells were pretreated for 1 h by adding actinomycin D (Sigma, A9415) to a final concentration of 5 µg·mL−1. After this 1 h preincubation period cells were dosed with either hyperosmotic medium or aldosterone or both. Cells were harvested at the times indicated for measurement of mRNA abundance by quantitative PCR as described above to measure the stability of transcripts in the absence of mRNA synthesis.
Overexpression of epitope-tagged TSC22D2-4 in mIMCD3 cells
TSC22D2-4 ORF was amplified with the primers: GAAATGTTGTCCACAAGAGTGTC (forward; initiation codon in bold-type) and TGCTGAGGAGACATTCGGCTG (reverse) and the correct sequence of the PCR product was confirmed by double-pass sequencing. pcDNA5/FRT/Tsc22D2i3 construct was created by cloning the PCR product in the vector pcDNA5/FRT/V5-His/Topo vector (Invitrogen). The construct was then propagated in Escherichia coli strain DH5 (Invitrogen). Endotoxin-free plasmid Mega-preps were performed using a kit as described by the manufacturer (Qiagen GmbH, Hilden, Germany). Stable cell lines were established by transfecting mIMCD3FRT cells (supplementary Fig. S1) with 2 µg of a 1 : 9 mix of pcDNA/FRT/Tsc22D2i3 plasmid DNA: pOG44 plasmid DNA and 4 µL of LipofectAMINE 2000 reagent (Invitrogen). Twenty-four hours after transfection cells were selected with medium containing 0.6 µg·mL−1 hygromycin (Invitrogen). After 2 weeks individual colonies were picked, expanded, and tested for expression of V5-His epitope-tagged TSC22D2-4 using quantitative PCR and western blot analysis.
Protein extraction and western blot analysis
For protein extraction, cells were lyzed in a buffer contained 50 mm Tris·HCl, pH 7.4, 1% Nonidet P-40, 0.25% sodium deoxycholate, 150 mm NaCl, 1 mm EDTA, 1 tablet of minicomplete protease inhibitor mixture (Roche Molecular Biochemicals, Indianapolis, IN) per 10 mL, 1 mm activated Na3VO4, and 1 mm NaF. Protein concentrations were determined by bicinchoninic acid protein assay according to the manufacturer's instructions (Pierce, Rockford, IL). Proteins were separated by SDS/PAGE. Equal amounts of protein (20 µg) were loaded in each lane of 12% Tris-glycine SDS/PAGE gels. Samples were electrophoresed at 125 V, gels briefly rinsed in transfer buffer (25 mm Tris, 200 mm glycine, 20% methanol), and proteins blotted onto PDVF membrane (Millipore Corp., Bedford, MA) at 1 mA·cm−2 for 90 min using a TransBlot SD semidry transfer cell (Bio-Rad Laboratories, Hercules, CA). Membranes were blocked for 30 min at room temperature in a solution containing 137 mm NaCl, 20 mm Tris, pH 7.6 (HCl), and 3% (w/v) nonfat dry milk. They were then incubated for 1 h in blocking buffer containing V5-HRP antibody (Santa Cruz Biotechnology, Santa Cruz, CA; 1 : 5000). Blots were developed with SuperSignal Femto (Pierce) and imaged with a chemiimager (Alpha Innotech, San Leandro, CA, USA).
Cells were grown in 12-well plates and harvested after being treated as indicated in the results section. Appropriate dilutions of cell suspensions were obtained in 0.2% methylene blue, incubated for 1 min, and viable (unstained) cells and dead (stained cells) were counted in Neubauer hemocytometer chambers.
Bioinformatics and statistical analysis
Multiple sequence alignments and phylogentic trees were constructed with alignx software (Informax, Bethesda, MD, USA). Data analysis was carried out with sigmaplot 9.0 (Systat, San Jose, CA, USA). Differences between pairs of data were analyzed by unpaired t-test. Differences in time series data sets were statistically evaluated using ANOVA. Significance threshold was set at P < 0.05 and data are presented as mean ± SEM.
We would like to thank Dr Devulapalli Chakravarty for assistance with the generation of the mIMCD3FRT cell lines. This study was supported by a grant from the National Institute of Diabetes and Digestive and Kidney diseases (NIH R01-DK59470). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.