Differential Transcriptional Regulation by Mouse Single-minded 2s*

Single-minded 1 and 2 are unique members of the basic helix-loop-helix Per-Arnt-Sim family as they are transcriptional repressors. Here we report the identification and transcriptional characterization of mouse Sim2s, a splice variant of Sim2, which is missing the carboxyl Pro/Ala-rich repressive domain. Sim2s is expressed at high levels in kidney and skeletal muscle; however, the ratio of Sim2 to Sim2s mRNA differs between these tissues. Similar to full-length Sim2, Sim2s interacts with Arnt and to a lesser extent, Arnt2. The effects of Sim2s on transcriptional regulation through hypoxia, dioxin, and central midline response elements are different than that of full-length Sim2. Specifically, Sim2s exerts a less repressive effect on hypoxia-induced gene expression than full-length Sim2, but is just as effective as Sim2 at repressing TCDD-induced gene expression from a dioxin response element. Interestingly, Sim2s bind to and activates expression from a central midline response element-controlled reporter through an Arnt transactivation domain-dependent mechanism. The differences in expression pattern, protein interactions, and transcriptional activities between Sim2 and Sim2s may reflect differential roles each isoform plays during development or in tissue-specific effects on other protein-mediated pathways.

The basic helix-loop-helix Per-Arnt-Sim (bHLH-PAS) 2 proteins comprise a growing family of transcription factors that play key roles during development and in sensing and adapting to changes in the environment. Individual PAS proteins are known to control morphogenesis, circadian rhythmicity, responses to hypoxia and toxin metabolism. These proteins contain a bHLH motif, which mediates dimerization with other bHLH proteins and contributes to determining DNA binding specificity. The PAS domain, named after the founding members of this family (period-arylhydrocarbon nuclear translocator-single minded), is a multifunctional protein surface responsible for such diverse activities as ligand binding, PAS protein dimerization, and non-PAS protein interactions (1).
In addition to environmental adaptation, some members of the bHLH-PAS family regulate development. In Drosophila, single-minded (sim) acts as the master regulator of central nervous system midline development by controlling expression of many genes required for differentiation. Similar to other bHLH-PAS proteins, sim functions as a heterodimer with arnt (2). This complex binds to central midline elements (CME) in the regulatory regions of target genes to activate expression of proteins required for proper central midline establishment (2,3).
Two mammalian homologs of sim, Sim1 and Sim2, have been identified (4 -6). These proteins share a high degree of similarity in their PAS domains, but little conservation is apparent in their carboxyl termini. Sim1 and Sim2 interact with Arnt, but differ from Drosophila sim, the aryl hydrocarbon receptor (AHR) and hypoxia inducible factor (HIF) by functioning as transcriptional repressors (7,8). Sim1 and Sim2 are expressed in a variety of tissues including brain, kidney, lung, and skeletal muscle where they play important developmental roles. Human SIM2 was first identified by exon trapping of a region on chromosome 21 known to be associated with Down syndrome (9). In addition, a splice variant of human SIM2, designated SIM2 short (SIM2s), has also been identified (9). This splice variant, which is missing exon 11 and therefore lacks a portion of the region implicated in mediating the repressive effects of SIM2, is reported to be involved in cancer susceptibility (10,11); however, functional differences between these two isoforms have not been reported.
Because the bHLH-PAS proteins share structural motifs and common binding partners, it is not surprising that cross-talk can occur between PAS-protein-mediated pathways. In the case of the HIF proteins, both Sim1 and Sim2 can compete with HIFs for Arnt binding, and interact with a prototypical hypoxia response element (HRE) to affect gene expression (7,8,12). Interestingly, Sim1/Arnt, but not Sim2/Arnt, can induce transcription of an HRE-controlled reporter gene via the COOH-terminal transactivation domain of Arnt (12). In contrast, Arntmediated transactivation of a CME-controlled reporter gene is severely impaired in the presence of Sim2 and is dependent upon the Sim2 dimerization domain and carboxyl terminus, which contains two separate repressive domains (7,8). Because repression by Sim2 is not specific for Arnt, as Sim2-Gal4 fusion constructs have repressive effects on a thymidine kinase promoter (9), it is thought that Sim2-mediated repression can also occur through direct interactions with the basal transcription machinery. Similar to Drosphilia sim, Sim1 and Sim2 also bind and regulate transcription through a consensus CME, which is not surprising because the CME core sequence (5Ј-ACGTG-3Ј) is identical to that of the HRE. As was seen for the HRE, Sim1 strongly activates transcription of a CME-controlled gene through the transactivation domain of Arnt, whereas Sim2 is repressive (7).
We have isolated a splice variant of mouse Sim2 that corresponds to the human Sim2s transcript, and have characterized its expression profile and transcriptional properties. Mouse Sim2s is expressed at high levels in adult kidney and skeletal muscle but in different ratios with respect to full-length Sim2. Like Sim2, Sim2s interacts with Arnt, but is slightly less efficient at binding Arnt2. These differences in expression pattern and binding partner preference may prove significant, as we have also determined that Sim2s and Sim2 differ dramatically in their ability to regulate expression of genes under control of hypoxia, dioxin, and central midline response elements.

MATERIALS AND METHODS
Animals and Cell Culturing-Female C57Bl/6J mice were housed under standard 12-h lights on/off conditions in a temperature and humidity controlled facility with food and water provided ad libitum. All animal housing and treatments were approved by and conformed to the Animal Use Protocols at Texas A&M University. All cell lines were maintained in a 37°C humidified incubator in a mixture of 95% air and 5% CO 2 . HEK-293T and HepG2 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal bovine serum and 1% penicillin-streptomycin.
RNA Isolation and Reverse Transcription-Total RNA was initially isolated from mouse tissues using TRIzol reagent (Invitrogen) and was further purified using RNeasy kits (Qiagen) with on-column DNase digestion (Qiagen). One microgram of total RNA was reverse transcribed using oligo(dT) and Superscript II reverse transcriptase (Invitrogen).
Plasmid Construction-The full-length mouse Sim2s plasmid (pcDNA-mSim2s) was made by amplifying the 5Ј most 1548-bp of Sim2 from pmSim2-GAL4-HA (kindly provided by Dr. Jerry Pelletier, McGill University, Montreal, Quebec, Canada) using HiFi Taq DNA polymerase (Roche Diagnostics) and the primers 5Ј-ATGAAGGAGAAGTC-CAAAAATGC-3Ј and 5Ј-AGCATTCACAGGAGAAGGCTCAGAA-3Ј. This fragment was cloned into pCR II-TOPO, and the 5Ј SpeI fragment was cut out and ligated to SpeI/XbaI cut pCRmSim2s-8 (original Sim2s 3Ј RACE clone) to create pCRFLmSim2s. After the sequence was confirmed, the entire insert was removed by EcoRI digestion, and cloned into EcoRI-cut pcDNA3. pmSim1-HA and pmSim2-HA were a kind gift of Dr. Jerry Pelletier.
All yeast two-hybrid plasmids utilized pGBK-T7 and pGADT7-Rec as the DNA-binding and activation plasmids, respectively (Clontech, Carlsbad, CA). For mouse Sim2s yeast two-hybrid plasmids, the fulllength insert of pCRFLmSim2s was removed by EcoRI digestion and ligated to EcoRI-cut vectors. Full-length Sim2L was removed from pmSim2-GAL4-HA by EcoRI digestion and cloned into EcoRI-cut vectors. Full-length Arnt and Arnt2 were amplified from mouse kidney and cloned into pCR II-TOPO. Once the sequences were verified, the inserts were removed and cloned into pGBK-T7 and pGADT7-Rec.
Reporter Plasmids-pDRE-TATA-luc, pGL2-TATA, and p␤-galactosidase were kindly given to us by Dr. Stephen Safe (Texas A&M University) (13). pHRE-TATA-luc was constructed by cloning annealed oligonucleotides corresponding to the top and bottom strands of the hypoxia-responsive region of the EPO promoter into pGL2-TATA (14). pCME-luc was a kind gift of Dr. Jerry Pelletier.
Yeast Two-hybrid Liquid Culture Assay-One hundred ml of YPDA were inoculated with 10 ml of an overnight AH109 yeast culture (Clontech) and grown to an A 600 of ϳ0.6. Cells were spun down and washed two times in 50 ml of sterile water and once in 2 ml of 1 M cold sorbitol. Cell pellets were resuspended in 2 ml of 100 mM LiAc/TE containing 25 mM dithiothreitol and incubated at room temperature for 1 h. Following an additional 1 M sorbitol wash, cell pellets were resuspended in 1 M sorbitol and mixed with plasmids on ice (50 l of cells plus 250 ng of each plasmid). Cells were transformed by electroporation with a 5-ms pulse at 1.5 kV, 50 microfarads, and 100 ohm. One ml of YPD was added to transformed yeast, which were then incubated at 30°C for 1 h. Cells were spun, and resuspended in 100 l of 1 M sorbitol and plated onto DOB/-Leu/-Trp plates and incubated at 30°C until colonies appeared. Double dropout broth liquid cultures of three colonies from each plate were grown overnight at 30°C with shaking. Fresh media was inoculated with 1 ml of overnight culture and incubated at 30°C with shaking until cells reached log phase (A 600 of the cultures between 0.5 and 0.8). At this point, the A 600 of each culture was recorded and 1.5 ml of yeast were spun and washed in Z buffer (10 mM Na 2 HPO 4 , 5 mM NaH 2 PO 4 ⅐H 2 O, 10 mM KCl, 1 mM MgSO 4 , pH 7.0). Cells resuspended in 100 l of Z buffer were subjected to three cycles of freezing/thawing for 60 s each in liquid nitrogen and a 37°C water bath. Seven hundred l of Z buffer containing 4 mM ␤-mercaptoehanol and 160 l of Z buffer containing 4 mg/ml o-nitrophenyl ␤-D-galactopyranoside was added to each tube and the reactions were placed at 30°C for 60 min. Reactions were stopped by addition of 400 l of 1 M Na 2 CO 3 and the time of reaction recorded. The OD 420 of each reaction was recorded and ␤-galactosidase activity calculated using the equation ␤-galactosidase units ϭ 1000 ϫ OD 420 /(t ϫ V ϫ A 600 ), where t ϭ elapsed time (in min) of incubation, V ϭ 0.1 ml ϫ concentration factor, and A 600 ϭ A 600 of 1 ml of culture. Data are expressed as the average of three separate colonies Ϯ S.E.
Co-immunoprecipitation-Full-length Arnt, Arnt2, Sim2, and Sim2s were made from pcDNA3-based expression plasmids using an in vitro translation kit (Promega). Sim2 and Sim2s were made in the presence of [ 35 S]methionine (Amersham Biosciences). Five l of Arnt or Arnt2 product were mixed with 15 l of radiolabeled Sim2 or Sim2s and allowed to incubate at room temperature for 2 h. Five g of anti-Arnt (Upstate), anti-Arnt2 (Santa Cruz Biotechnology, Inc.), or rabbit IgG (Upstate) were added, and the volumes were increased to 150 l with water and 2ϫ co-immunoprecipitation buffer to make a 1ϫ solution (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100 and 0.5% IGEPAL). Following a 2-h incubation at room temperature, 20 l of agarose A bead slurry was added and samples were incubated at room temperature for 1.5 h with gently mixing. Beads were recovered at 5000 rpm for 1 min, and washed 3 times in 500 l of 1ϫ co-immunoprecipitation buffer. Final pellets were resuspended in loading buffer, boiled, and separated on polyacrylamide gels. Gels were vacuum-dried and exposed to film for 1 week.
Transient Transfection-Sim2 and Sim2s expression was confirmed by Western analyses before large-scale experiments were conducted. For each transfection, cells were seeded at 4 ϫ 10 4 cells per well in 24-well plates the day before transfection. The following morning, 200 ng of the appropriate reporter plasmid was co-transfected with 100 ng of internal control (p␤-galactosidase) and various amounts of test plasmids using Lipofectamine and Plus reagent (Invitrogen). Twenty-four hours later, cells were incubated under hypoxic (1% O 2 ) or normoxic (21% O 2 ) conditions (pHRE-TATA-luc transfected HEK-293T cells) or in the presence of vehicle (Me 2 SO) or 10 nM 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (pDRE-TATA-luc transfected HepG2 cells) for 40 h. Cells were harvested, and the luciferase and ␤-galactosidase activities of the cell lysates were determined by dual luciferase assay using Luciferin (Molecular Probes) and Galacto-Light (Tropix, Applied Biosystems). Luciferase activities were normalized to the internal control values and are represented as the mean Ϯ S.E. for three wells per condition. Significant differences were determined using Student's t test.
Chromatin Immunoprecipitation Assay-Twenty-four h after transfection, formalin (270 l/10 ml of medium) was added to HEK-293 cells and allowed to incubate at 37°C for 10 min. Cross-linking was stopped by addition of glycine to a final concentration of 125 mM. Following a series of phosphate-buffered saline washes, cells were scraped and recovered by centrifugation. Cells (500,00 cells per ChIP assay) were resuspended in SDS lysis buffer (50 mM Tris-HCl, pH 8.1, 1% SDS, 10 mM EDTA) and sonicated on ice. Chromatin was recovered, and mixed with ChIP dilution buffer (16.7 mM Tris-HCl, pH 8.1, 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 167 mM NaCl) containing 1ϫ Complete Protease Inhibitor Mixture (Roche). Chromatin was cleared twice by incubation with agarose bead slurry at 4°C for 30 min, followed by centrifugation. Antibody was added (10 g for Sim2, 1 g for IgG) and incubated at 4°C overnight with mixing. Agarose bead slurry was added and reactions were incubated at 4°C for 1 h with mixing. The agarose beads were recovered and washed successively for 5 min at 4°C in low salt (20 mM Tris-HCl, pH 8.1, 150 mM NaCl, 0.1% SDS, 1% Triton X-100, 2 mM EDTA), high salt (20 mM Tris-HCl, pH 8.1, 500 mM NaCl, 0.1% SDS, 1% Triton X-100, 2 mM EDTA), LiCl (10 mM Tris-HCl, pH 8.1, 250 mM LiCl, 1% IGEPAL-CA630, 1% deoxycholic acid, 1 mM EDTA), and TE (10 mM Tris-HCl, pH 8.0, 1 mM EDTA) buffers. Chromatin was eluted at room temperature by incubation in freshly prepared elution buffer (1% SDS, 0.1 M NaHCO 3 ) for 15 min. Samples were then treated with RNase A in the presence of 200 mM NaCl and cross-linking was reversed by incubation at 65°C for 4 h. Chromatin was ethanol precipitated and resuspended in buffer containing 40 mM Tris-HCl, pH 8.1, 10 mM EDTA, and 200 g/ml proteinase K. DNA was purified using Qiaquick PCR purification columns (Qiagen) and eluted in 50 l of sterile water. PCR detection of pGL2-CME immunoprecipitation targets was performed in 25-l reactions containing 2 l of chromatin and the primers pGL2-ChIP-F1 (5Ј-CCCCCTGAACCT-  (3 RACE). A, structure of mouse Sim2 and Sim2s cDNAs. Numbered squares indicate exons. Numbered arrows indicate relative positions of mouse Sim2-specific primers used in 3Ј RACE PCR. Dotted lines indicate relative positions of restriction sites used to verify the identity of 3Ј RACE products. B, restriction enzyme analyses of mouse kidney and liver 3Ј RACE products. Two micrograms of total RNA from mouse kidney or liver was reverse transcribed using an oligo(dT)-based adapter primer (AP-dT 17 ). An initial round of PCR was performed on 1 l of a 1:100 dilution of the RT reaction using the adapter primer and a mouse Sim2-specific forward primer (P1). The products of this initial PCR were not visible (data not shown). Subsequent PCR were performed on 1 l of a 1:100 dilution of the previous reaction using the adapter primer and nested Sim2-specific primers (P2, P3, and P4 in Fig. 2A). For visualization and confirmation of identity, 10 l of PCR product plus (ϩ) or minus (Ϫ) restriction enzyme was analyzed on a 1.2% agarose gel. Mouse Sim2s 3Ј RACE products are predicted to have SpeI (P2/AP), BamHI (P3/AP), or HaeIII (P4/AP) sites. Sim2-specific products were only detected in kidney (Kid) RNA samples, and only Sim2s was detected. Liv, liver.

RESULTS
Previous studies have identified a splice variant of human SIM2; however, a similar isoform has not been reported in mice, nor have functional differences between Sim2 and Sim2s been reported. To better characterize biochemical and physiological differences between Sim2 and Sim2s in mice, we set out to identify the mouse Sim2s transcript. Total RNA from mouse kidney and liver tissues were used for 3Ј RACE experiments because Sim2 is expressed at high levels in the kidney, but not in liver. A unique adapter primer (AP-dT 17 ) was used to reverse transcribe total RNA from the tissues. This reaction was then used as a template for PCR using a primer directed to the AP region of the 3Ј adapter and a Sim2-specific forward primer (P1, Fig. 1A). Subsequent PCR were performed on diluted PCR products using the AP and nested Sim2-specific primers (P2-P4, Fig. 1A). The resulting PCR products from each reaction were submitted to restriction enzyme analyses for predicted sites to support the identity of the PCR products as Sim2 (Fig. 1B). Potential Sim2 clones with predicted restriction digest products were only obtained in kidney, and were too small to be full-length Sim2. In humans, the SIM2s splice variant is missing exon 11, and contains an extension of exon 10 (9). Based on the sequence of mouse chromosome 16 near the region of mouse Sim2 exon 10, a single HaeIII site is predicted to be present in the Sim2s transcript. This was confirmed in the fourth round 3Ј RACE PCR product (Fig. 1B, right  panel) suggesting that this clone is the mouse homolog of human SIM2s.
3Ј RACE clones were sequenced and compared with reported mouse Sim2 sequences and chromosome 16. Like the human form, mouse Sim2s is missing exon 11 and contains an extension of exon 10 ( Fig. 2A). The mRNA (Fig. 2B) is predicted to encode a protein of 579 amino acids that is missing the last 131 amino acids and contains an additional 53 amino acids encoded by the "s" exon (Fig. 2C). Mouse Sim2s still contains the Ser/Thr-and Pro/Ser-rich regions shown previously to harbor repressive activities, but is missing the Pro/Ala-rich repressor region present in full-length Sim2 (8).
Mouse and human Sim2s share 86% identity at the RNA level, and are 87% homologous at the protein level. Interestingly, the s portions of mouse and human Sim2s are not as homologous as the rest of the gene (Fig. 2C). The mRNA and protein sequences of mouse and human Sim2 from start to the end of exon 10 are 88 and 91% homologous, respectively. However, the DNA and protein sequences of the extended exon 10 regions of mouse and human Sim2s are only 45 and 20% homologous, respectively. Despite these differences, a few amino acid residues remain conserved, including 545 GGGW 548 and 561 SASK 564 .
Sim expression was analyzed in mouse tissues by RT-PCR and real time RT-PCR (Fig. 3). Sim1 was expressed at high levels in kidney, brain, lung, and skeletal muscle (Fig. 3A). In contrast, Sim2 and Sim2s were expressed at high levels in kidney and skeletal muscle. The ratio of Sim2s and Sim2 mRNA expressed in kidney and skeletal muscle appeared to FIGURE 3. Tissue-specific expression of mouse Sim genes. A, RT-PCR analysis of Sim1, Sim2, and Sim2s expression in various mouse tissues. Total RNA from the indicated tissues (BRN, brain; HRT, heart; KID, kidney; LVR, liver; LNG, lung; SKM, skeletal muscle; SPL, spleen; UTR, uterus) was subjected to RT-PCR as described under "Materials and Methods." Sim1 was expressed at relatively high levels in mouse kidney Ͼ brain Ͼ lung Ͼ skeletal muscle. Sim2s was expressed at high levels in kidney and was detectable in skeletal muscle. Full-length Sim2 was detected in skeletal muscle Ͼ kidney. B, co-amplification of Sim2 and Sim2s by RT-PCR in mouse kidney, liver, and skeletal muscle (Sk. M). RT-PCR were performed on total RNA from mouse tissues using a single forward primer and both Sim2s-and Sim2 long formspecific reverse primers. Sizes of DNA ladder are indicated to the left. Neither isoform of Sim2 was detectable in liver, but both Sim2 and Sim2s were detected in kidney and skeletal muscle. The relative amounts of Sim2 and Sim2s expressed in these tissues differed with kidney expressing more Sim2s than Sim2 and vice versa in skeletal muscle. C-E, quantitative real time PCR analyses of mouse kidney, liver, and skeletal muscle mRNA for total Sim2 (C), Sim2s (D), and full-length Sim2 (E). APRIL 21, 2006 • VOLUME 281 • NUMBER 16 differ, suggesting that their expression is controlled in a tissue-specific manner. Therefore, expression of both isoforms of Sim2 were determined in kidney, liver, and skeletal muscle by RT-PCR using a common forward primer located in exon 9, and both Sim2-and Sim2s-specific reverse primers located in exons 11 and s, respectively. As expected, neither Sim2 isoform was detected in liver, but kidney and skeletal muscle expressed both Sim2 and Sim2s in different relative amounts (Fig.  3B). In agreement with our initial analyses (Fig. 3A), kidney expressed slightly more Sim2s than Sim2, whereas skeletal muscle expressed significantly more Sim2 than Sim2s. Quantitative real time RT-PCR analyses of kidney, liver, and skeletal muscle RNA for total Sim2 (Fig. 3C), Sim2s (Fig. 3D), and full-length Sim2 (Fig. 3E) corroborated our PCR analyses and found that, overall, skeletal muscle contains significantly higher levels of full-length Sim2 mRNA than kidney.

Characterization of the Mouse Sim2s Gene
The utility of the yeast two-hybrid system in addressing functional interactions between Sim genes and other PAS proteins has been well established (6). To begin assessing if Sim2 and Sim2s differ in their transcriptional properties, we took advantage of the two-hybrid liquid culture assay. Chimeras containing the GAL4 DNA binding domain (pGBK-based plasmids) in-frame with mouse Arnt, Arnt2, Sim2, or Sim2s were transfected into AH109 yeast in various combinations with constructs containing these genes in-frame with the GAL4 activation domain (pGAD-based plasmids). Positive interactions between the introduced proteins "reconstitutes" the GAL4 transcription factor leading to increased expression of a LacZ reporter gene under control of a GAL4 activation sequence.
High levels of LacZ expression were observed when GBK-Arnt and GAD-Arnt were co-transfected into yeast (Table 1) supporting the observation that Arnt can homodimerize (15). Significantly less LacZ activity was seen in GBK-Arnt-GAD-Arnt2 transfected cells suggesting that ARNT does not interact as well with Arnt2. Both the long and short forms of mouse Sim2 interacted with Arnt, and to a lesser extent with Arnt2. Interestingly, the degree of these interactions was dependant upon which protein was coupled to the DNA binding domain of Gal4 as LacZ activity was less in GBK-based Sim constructs interacting with GAD-based Arnt and Arnt2 than GBK-based Arnt and Arnt2 constructs interacting with GADbased Sim constructs (Table 1). These observations may be due to the repressive nature of the Sim2 proteins, or steric hindrance induced by the tertiary structure of the Sim2-GAL4AD protein.
Interactions between Sim2 and Arnt proteins were confirmed biochemically by co-immunoprecipitation. [ 35 S]Methionine-labeled Sim2 and Sim2s proteins were mixed with cold Arnt or Arnt2 and complexes were precipitated with antibodies directed against Arnt or Arnt2. Anti-Arnt antibody was able to pull down both Arnt/Sim2 and Arnt/Sim2s complexes at relatively comparable levels (Fig. 4, lanes 1 and 2). In contrast, anti-Arnt2 antibody was able to pull down Sim2 slightly more efficiently than Sim2s (Fig. 4, lanes 3 and 4). Although IgG control samples show some Arnt binding (Fig. 4, lane 5), there was very little Sim2 or Sim2s pulled down by nonspecific binding (Fig. 4, lane 6). These results, coupled with the yeast two-hybrid data, suggest that Sim2s favors binding to Arnt over Arnt2 in comparison to full-length Sim2, which can interact with both Arnt and Arnt2 with relatively equal affinities.
Direct targets of mammalian Sim proteins have not been determined; however, indirect transcriptional activities of Sim1 and Sim2 have been reported. In those studies, it was found that Sim2 can inhibit the actions of other bHLH-PAS proteins through active repression or competition for Arnt binding (7,8,12). Overexpression of Sim2 blocked hypoxia-induced expression from HRE-controlled genes by directly binding to the HRE and actively repressing expression of adjacent genes. This effect was dependent upon the Sim2 dimerization domain and carboxyl terminus. Sim2 can also inhibit TCDD-induced expression from XRE-controlled genes, but does so by competing with AHR for Arnt binding. In contrast, Sim2 represses expression of a CME-driven reporter, whereas Sim1 can activate CMEmediated gene expression via the Arnt transactivation domain (7). To investigate if Sim2s can affect the function of other bHLH-PAS proteinmediated pathways, we performed co-transfection assays using various mouse Sim and Arnt expression plasmids with HRE-, XRE-, and CMEcontrolled luciferase reporter constructs.
Expression of Sim2 and Sim2s in transfected HEK293T cells was confirmed by Western blotting (Fig. 5A). HEK293T cells transfected with an HRE-controlled luciferase reporter and the pcDNA empty vector showed a significant increase in luciferase activity following a 40-h incubation under hypoxic conditions (Fig. 5B). Introduction of increasing amounts of mouse Sim2 or Sim2s significantly repressed hypoxiainduced reporter gene expression. Inclusion of 100 ng of mouse Arnt expression vector increased the response to hypoxia and slightly overcame both Sim2 and Sim2s-mediated repression. Addition of mouse Arnt2 expression vector did not alter the response to hypoxia and both Sim2 and Sim2s repressed hypoxia-induced expression of the HREcontrolled reporter in the presence of Arnt2. Increasing amounts of Sim or Sim2s did not repress in a concentration-dependent manner suggesting that both Sim2 and Sim2s repress hypoxia-induced gene expression through direct interaction with the HRE and not by competing with HIF factors for Arnt binding.
Previous studies have suggested that Sim2 can interfere with AHR-mediated gene expression through competition with AHR for Arnt binding (8). To determine whether Sim2s exerts a similar effect on AHR-mediated signaling, HepG2 cells were co-transfected with a DRE-controlled reported gene with various combinations of Sim2 or Sim2s and Arnt expression plasmids. AHR activation was accomplished by addition of 10 nM TCDD to growth medium 24 h before harvest. A robust TCDD response seen in pcDNA-transfected cells was significantly repressed (p Ͻ 0.0001) by inclusion of either Sim2 or Sim2s expression plasmids (Fig. 5C). The degree of this repression was not significantly different between Sim2 and Sim2s; however, increasing amounts of Arnt attenuated the repression (p Ͻ 0.01) in both Sim2 and Sim2s-transfected cells indicating that the mechanism of  this repression involves competition for Arnt. These results are consistent with previous studies (8) and suggest that Sim2 and Sim2s do not differ in their ability to repress AHR-mediated gene expression, which occurs through direct competition for Arnt binding.
In Drosophila, sim regulates transcription through the CME, which contains the core sequence 5Ј-ACGTG-3Ј also found in an HRE. Previous studies have shown that murine Sim1 and Sim2, in concert with Arnt, can bind and regulate expression of a CME reporter construct (7). Sim1 activated expression of a CME-controlled gene through the Arnt transactivation domain, whereas full-length Sim2 was repressive. Activation of a CME-controlled gene by Sim2 was accomplished when portions of the C-terminal repression domain were deleted. Because Sim2s is missing part of this repressive region, we anticipated that Sim2s would be less repressive than Sim2 on CME-mediated gene expression.
Cotransfection of HEK293 cells with a CME-controlled reporter gene and increasing amounts of Arnt or Arnt2 had no effect on reporter expression (Fig. 6A). Contrary to previous reports, we found that introduction of Sim2 with Arnt, but not Arnt2, resulted in a slight increase in luciferase activity (Fig. 6A). Co-expression of Sim2s with Arnt, but not Arnt2, resulted in significantly increased reporter gene expression. This effect appears to be Arnt-dependent as luciferase activity increased with increasing amounts of Arnt. This conclusion was further supported by experiments utilizing constant Arnt and increasing amounts of Sim2 expression plasmids (Fig. 6B). Although reporter gene expression increased with increasing amounts of Sim2 and Sim2s expression plasmids, this effect was only significant in cells receiving the highest amount of Sim2 and Sim2s expression plasmid. Confirmation of Arntdependent Sim2s transcriptional activation from a CME-controlled gene is presented in Fig. 6C. In these experiments, contransfection of a mutant Arnt (Arnt⌬TAD), which is missing the transactivation domain, abolished the ability of Sim1, Sim2, and Sim2s to increase expression from a CME. Expression of Arnt⌬TAD repressed basal CME-mediated gene expression in the absence of external Sim proteins (Fig. 6C). Co-expression of Sim1 and Arnt resulted in significant reporter gene expression that was abolished when Arnt was replaced with the Arnt⌬TAD expression vector. A similar effect was seen with Sim2, but the degree of gene activation was significantly lower than was seen with Sim1. Surprisingly, Sim2s was almost as potent as Sim1 in activating CME-controlled gene expression (Fig. 6C). As was seen with Sim1, this effect was completely abolished when Arnt⌬TAD was substituted for Arnt, implying that this effect is entirely mediated by the activation domain of Arnt. FIGURE 6. Transcriptional activity of mouse Sim2 and Sim2s on a CME-controlled reporter gene. A, effects of increasing Arnt and Arnt2 on mouse Sim2-and Sim2s-mediated expression of a CME-controlled reporter. HEK-293T cells were transfected with increasing amounts of Arnt or Arnt2 expression vector in the presence of absence of Sim2 or Sim2s expression vectors plus a CME-controlled luciferase reporter and ␤-galactosidase expression plasmid. The amount of each plasmid per transfection (in nanograms) is indicated under the figure. Luciferase activity was increased in both Sim2-and Sim2s-transfected cells, but only in the presence of Arnt, and not Arnt2. The degree of induction was significantly higher in Sim2s-transfected cells in comparison to Sim2 cells. Asterisk, luciferase expression is significantly higher in the presence of Sim and Sim2s in comparison to no Arnt controls ( p Ͻ 0.005). Double asterisk, luciferase expression is significantly higher in Sim2s cells in comparison to Sim2 cells ( p Ͻ 0.01). B, the effects of increasing amounts of mouse Sim2 or Sim2s on CME-mediated gene expression. HEK-293T cells were transfected with or without 100 ng of Arnt or Arnt2 expression vector with increasing amounts of Sim2 or Sim2s expression vector plus a CME-controlled luciferase reporter and ␤-galactosidase expression plasmids. The amount of each plasmid per transfection (in ng) is indicated under the figure. Luciferase activity was increased in both Sim2-and Sim2s-transfected cells, but only when Arnt was present. The degree of induction was Sim2-dependent as higher amounts of expression vector increased luciferase activity. Again, Sim2s was significantly better than Sim2 at inducing CME-controlled gene expression. Asterisk, luciferase expression is significantly higher in the presence of Sim and Sim2s in comparison to no Arnt controls ( p Ͻ 0.005). Double asterisk, luciferase expression is significantly higher in Sim2s cells in comparison to Sim2 cells ( p Ͻ 0.01). C, SIM-mediated regulation of a CME-controlled gene requires the transcriptional activation domain of Arnt. Control (pcDNA), Sim1, Sim2, or Sim2s expression plasmids were co-transfected with pCME-luc and a ␤-galactosidase expression vector along with full-length Arnt (white bars) or a mutant Arnt (black bars) expression plasmid. In the presence of full-length Arnt, robust luciferase expression was detected in Sim1-and Sim2s-transfected cells and to a lesser extent, in Sim2 cells. This effect was abolished when Arnt was replaced with the transactivation mutant Arnt (Arnt⌬TAD). Asterisk, luciferase expression is significantly repressed in the presence of Arnt⌬TAD ( p Ͻ 0.03). Double asterisk, luciferase expression is significantly induced in the presence of Sim1 ( p Ͻ 0.02). Triple asterisk, luciferase expression is significantly increased in the presence of Sim2 in comparison to pcDNA control cells ( p Ͻ 0.01). §, luciferase expression is significantly higher in the presence of Sim2s in comparison to pcDNA-transfected cells ( p Ͻ 0.0003).
†, luciferase expression is significantly inhibited in the presence of Arnt⌬TAD ( p Ͻ 0.001).
To confirm the interactions between Sim2 and Sim2s on a CME, ChiP were performed on CME-luc-transfected control and Sim2/Arnt or Sim2s/Arnt cells. Chromatin was immunoprecipitated with an anti-Sim2 antibody that recognizes both Sim2 and Sim2s, and was analyzed for CME binding using a set of PCR primers specific for the CME reporter plasmid. The presence of Sim2 on the CME was detectable in control cells, most likely due to endogenous Sim2 (data not shown). More importantly, the presence of both Sim2 and Sim2s on the CME was elevated in Sim2-transfected cells (Fig. 7). These data suggest that the differential outcomes of Sim2 isoform binding to a CME are not because of changes in DNA binding.

DISCUSSION
Transcriptional regulation occurs through multiple distinct mechanisms involving negative as well as positive interactions between regulatory factors. The mammalian Sim proteins are unique members of the bHLH-PAS family because they can exert negative effects on transcription. It has been determined that the repressive effects of Sim2 are mediated by two domains in its carboxyl terminus. One of these domains is rich in Pro and Ser residues, whereas the other is Pro and Ala rich. These repressive domains appear to be nonspecific as Sim2 can suppress activation of a Gal4 activation domain on a thymidine kinase promoter (6, 7) as well as Arnt-mediated transactivation. Similar hydrophobic domains are present in other transcriptional repressors including the Drosophila Kruppel transcription factors and Even-skipped, which inhibit transactivation by competing with TBP for TATA box binding thus preventing assembly of the preinitiation complex (16). In this paper, we have shown that mouse Sim2s, a splice variant of Sim2, has differential effects on CME-and HRE-mediated gene expression. Sim2s is less potent than full-length Sim2 at repressing Hif1␣ (Fig. 5B), and can activate expression of a gene controlled by the Drosophila toll gene CME via the transactivation domain of Arnt (Fig. 6).
The hypo-suppressive effects of Sim2s are not surprising given that Sim2s is missing the Pro/Ala-rich repression domain. What is surprising is that Sim2s is just as repressive as Sim2 on TCDD-mediated gene expression through a DRE (Fig. 5C) and is able to increase expression of a CME-controlled gene (Fig. 6). These data suggest a model in which the response element dictates transcription factor domain-dependent suppression or activation. On an HRE, both the Pro/Ser-rich and Pro/Alarich domains of Sim2 appear to exert repressive effects. This is based on the observation that Sim2s, which is missing the Pro/Ala-rich region, still exerts a repressive effect although it is not as strong as that observed with Sim2 (Fig. 5B). In the case of the DRE, both Sim2 and Sim2s can repress gene expression to an equal extent suggesting that only the Pro/Ser-rich domain mediates Sim2-mediated repression from a DRE. In contrast, Sim2s can activate expression from a CME apparently by acting as a docking protein for Arnt (Figs. 6 and 7). This implies that the Pro/Ala-rich sequence present in Sim2, but not Sim2s, exerts a negative effect on CME-mediated gene expression. As Sim2s lacks this domain, interactions between Sim2s and Arnt on a CME result in Arnt-mediated activation of Sim2s targets.
Interactions between transcription factors and their cognate response elements are influenced by sequence flanking the core binding motif. Whitelaw et al. (17) reported that a region of the AHR ligand-binding domain exerted different degrees of repression on different DNA targets thus, providing an example of such promoter-specific influence on transcription factor function in the bHLH-PAS family (17). These response element-specific effects are most likely due to DNA-dependent conformational changes in the interacting factor, which may influence the ability of the transcription factor to recruit co-regulatory proteins to the promoter. For example, the POU domain-containing transcription factor POU1F1 (e.g. PIT1) represses transcription when bound to its response element in the growth hormone gene, but induces expression through a similar element in the prolactin promoter. This was shown by crystallography to be due to DNA-induced allosteric changes in PIT1 confirmation resulting in differential coregulator recruitment (18). In the case of Sim2s, such dynamic, DNA-mediated changes in transcriptional outcome may reflect the ability of Sim2s to exert differential effects on similar response elements. Further studies are necessary to elucidate the mechanisms governing Sim2s-mediated gene repression and activation.
In Drosophila, several sim targets have been identified including slit, engrailed, breathless, and spitz (2,19). Not surprisingly, Hif1␣ regulates many of the mammalian homologs of these genes because Hif1␣ can bind the same core response element. Although definitive targets of Sim2 have not been identified in mammals, this study and others have shown that Sim2 can affect the actions of other bHLH-PAS proteins by active repression and interference (8,12).
Human SIM2s is expressed in normal kidney and tonsil as well as lung and testes (11,20). We have found that Sim2s expression in mice is comparable, with high levels of Sim2s mRNA detected in kidney and skeletal muscle (Fig. 3A). Interestingly, the ratio of Sim2 to Sim2s differs between these tissues with Sim2s being expressed at higher levels than Sim2 in kidney and vice versa in skeletal muscle (Fig. 3B). The significance of Sim2 isoform predominance in these tissues is unknown, but presumably could have substantial effects as we have demonstrated differences between binding partner specificity (Table 1 and Fig. 4) and transcriptional activities between Sim2 and Sim2s.
Due to the complex interaction potential and overlapping expression patterns of Sim proteins and Hif1␣, a hypoxic switch has been proposed to operate in cells expressing both genes (12). Such a switch could have profound implications for environmental regulation of developmental signaling pathways as developmental stage and organ-specific differences in response to systemic hypoxia have been reported (21 22). Presently, little is known about the functions of Sim2 proteins and direct targets of mammalian Sim2 have not been reported. We are actively investigating the role of Sim2 and Sim2s in development and characterizing downstream target genes. With a better understanding of the biochemical properties of Sim2s, and identification of bona fide Sim2s target genes, a better comprehension of its role in development will be achieved.