Usf1, a suppressor of the circadian Clock mutant, reveals the nature of the DNA-binding of the CLOCK:BMAL1 complex in mice

The Clock^Δ19 mutation results from an A→T transversion in a splice donor site that causes exon skipping and disruption of the C-terminal transactivation domain of CLOCK (King et al., 1997). Consequently, the levels of Per and Cry mRNA are much lower in Clock^Δ19 mutants than in wild-type mice (Lowrey and Takahashi, 2011). Given the nature of the suppressor phenotype, we hypothesized that the product of the gene responsible for the Soc QTL should activate transcription from E-boxes in a manner similar to CLOCK:BMAL1. We tested E-boxes from Per1 and Per2 for activation by the seven Soc candidates in HEK293T cells. Of these, only Usf1 significantly activated transcription from both Per1 and Per2 E-boxes (Figure 3B). We observed a clear dose response of Per promoter activation by USF1 (Figure 3B). Unlike CLOCK:BMAL1, activation of E-boxes by USF1 was not inhibited by the CRYs (Figure 3B). Although it has been reported previously that USF1 binds E-box sequences (Ferré-D'Amaré et al., 1994), our finding that, of the seven Soc candidates, only USF1 effects significant transactivation from circadian-relevant E-boxes, provided initial molecular evidence that Usf1 is a candidate for the Soc locus.

We next tested the circadian expression of the seven candidate genes in liver tissue collected from Clock^Δ19/+ (BALB x B6)F1 and B6 animals every 4 hr on the third day of DD exposure. Among the genes tested, only the Usf1 transcript exhibited significant up-regulation in F1 mice (Figure 3C). By western blot analysis, USF1 protein levels in the liver were ∼40% higher in BALB animals (Figure 3D). To test whether the suppressor strain activates E-box-containing circadian clock genes in vivo, we examined the effect of strain background (B6 or [BALB x B6]F1) on expression levels of Per1, Per2, Cry1 and Cry2 in liver using quantitative PCR (qPCR). We detected significant up-regulation of all four circadian genes tested in the F1 background (Figure 3E), consistent with elevated expression of an E-box transcription factor such as USF1. Taken together these results strongly suggest that Usf1 is a prime candidate gene for the Soc locus. Because there are no coding mutations in Usf1 between the B6 and BALB mouse strains (Figure 2—source data 1), we hypothesized that the elevated expression of Usf1 in the BALB suppressor strain is likely the cause of Clock suppression.

To determine whether the allele-specific expression difference is the result of promoter polymorphisms between B6 and BALB, we tested basal promoter activity of ∼2.3 kb upstream of the Usf1 transcription start site using a luciferase reporter assay. Usf1 promoter sequences from both B6 and BALB were cloned into the PGL4 luciferase reporter vector and transfected into HEK293T cells. Although both the B6 and BALB sequences activated the luciferase reporter, the BALB sequence resulted in a significantly higher signal than that from B6 (Figure 5D). By exchanging EcoRI-XhoI fragments between the B6 and BALB reporter clones, we narrowed the critical polymorphisms to the EcoRI-XhoI fragment (∼1000 bp) and identified seven common polymorphisms (five SNPs and two deletions) that match the phenotypic distribution of 16 mouse strains (Figure 5E).

We next amplified ∼100-bp fragments containing each of the seven polymorphisms from BALB and B6 genomic DNA and assayed their reporter gene activity to test whether the candidate SNPs affect promoter activity. Among the six fragments tested (SNPs 5 and 6 were contained within the same fragment), only two (SNPs 3 and 7) showed strong promoter activity and statistically higher activity from the BALB allele compared to the B6 allele (Figure 5F). SNPs 3 and 7 correspond to rs31538551 and rs31538547, respectively, in the NCBI SNP database. Next, we used site-directed mutagenesis on the B6 promoter (EcoRI-XhoI fragment) to introduce the corresponding BALB polymorphism at each of the two SNP loci and examined promoter activity. We observed a significant increase in luciferase signal by replacing the B6 allele of SNP7 (rs31538547) with the BALB allele (Figure 5G). This strongly suggests that SNP7 from the BALB Usf1 promoter region is a quantitative trait nucleotide (QTN) responsible for the Soc phenotype.

To explore our observation that CLOCK^Δ19:BMAL1 occupancy decreases in Clock^Δ19/Clock^Δ19 mice we determined the relative affinity of CLOCK:BMAL1, CLOCK^Δ19:BMAL1 and USF1 complexes to E-boxes using EMSA. Given our results demonstrating Usf1 expression differences between the B6 (non-suppressor) and BALB (suppressor) mouse strains (Figure 3), we used isogenic B6 animals in these experiments to insure that USF1 levels were constant as we examined E-box occupancy of CLOCK, CLOCK^Δ19, and USF1. We examined the same E-box sites in the Dbp, Per1, and Per2 genes used for ChIP analysis (Figure 6A), and confirmed the identity of gel shift complexes by supershift using specific antibodies to CLOCK, BMAL1 and USF1 (Figure 6D). At all E-boxes, we detected three types of DNA:protein complexes: CLOCK:BMAL1 tandem heterodimers (CB2), CLOCK:BMAL1 heterodimers (CB1), and USF (consisting of USF1:USF1 and possibly USF1:USF2 or USF2:USF2 dimers). As shown previously, these E-box sites consist of tandem E-boxes spaced 6–10 nucleotides apart, and binding to these tandem sites appears to be cooperative (Figure 6E; Rey et al., 2011). Interestingly, wild-type CLOCK:BMAL1 complexes bind primarily as a tandem complex (CB2) at all four E-box sites with the Dbp E_P and Per2 E′ E-boxes showing the most bias towards tandem binding (Figure 6D). By contrast, the mutant CLOCK^Δ19:BMAL1 heterodimer binds primarily as a single complex (CB1) at all four sites. Antibody supershift experiments affected both the CB1 and CB2 complexes, consistent with the hypothesis that these complexes, whether containing mutant or wild-type CLOCK, represent single or tandem CLOCK:BMAL1 binding complexes, respectively (Figure 6D).

In addition to these overall trends, specific E-box sites also provide further insight into the affinity of CLOCK:BMAL1 binding. In the case of the Dbp E_I2 site (Figure 7A), where wild-type CLOCK:BMAL1 binds both as single and tandem complexes, the tandem CB2 site has much higher affinity than CB1, consistent with the prediction of previous work (Rey et al., 2011). In contrast, at the Dbp E_P site (Figure 7A), where mutant CLOCK^Δ19:BMAL1 complex can be seen binding as both single CB1 and tandem CB2 complexes, the affinities of the two mutant complexes are similar and lower than wild-type CLOCK:BMAL1 complex. This suggests that the affinity of the mutant CLOCK^Δ19:BMAL1 complex is lower for two reasons: first, it does not preferentially bind as a tandem complex, thus binding may not be cooperative; and second, its affinity, even as a tandem complex, remains lower suggesting that the Clock^Δ19 mutation interferes both with cooperativity and with affinity. Thus, our saturation binding experiments clearly demonstrate that the affinity of CLOCK^Δ19:BMAL1 is significantly lower relative to the wild-type complex, and that this lower affinity is similar to the affinity of USF1 at the same sites. This suggests that under normal conditions, the wild-type CLOCK:BMAL1 complex binds with much higher affinity than the USF1 complex, but when mutant, the affinities of the CLOCK^Δ19:BMAL1 and USF1 complexes are comparable which allows USF1 to bind more effectively to E-box sites.

Heat map analysis of the 1916 sites that are co-occupied by both USF1 and CLOCK:BMAL1 in Clock^Δ19/Clock^Δ19 mutant animals reveals a dramatic increase in binding intensity of USF1 in mutant mice (Figure 8B,C). Not only do the number of USF1 peaks increase, but also the binding intensity at existing sites increases in mutant animals. Concomitantly, in the mutant animals there is a decline in CLOCK binding intensity at these 1916 common binding sites (Figure 8B,C). We quantitated the binding intensity of the top 10% of the binding sites for all three ChIP-Seq experiments and performed comparisons between wild-type and Clock^Δ19/Clock^Δ19 animals (Figure 8C). For USF1 and CLOCK there was a significant increase and decrease, respectively, in binding between wild-type and mutant animals, but no change in BMAL1 binding (Figure 8C). Motif analysis of each group of binding sites identified the canonical E-box binding site, CACGTG, as the most significant motif, indicating both USF1 and CLOCK:BMAL1 bind to the same consensus sequence in both wild-type and Clock^Δ19/Clock^Δ19 mice (Figure 8D).

UCSC genome browser profiles are shown for examples of four genes (Figure 8E). Per1 has at least five E-boxes, however ChIP-Seq analysis reveals two major peaks approximately 5 kb upstream of the transcription start site as well as two minor peaks further downstream (Figure 8E). The peak height of USF1 binding increases dramatically in Clock^Δ19/Clock^Δ19 animals, and there is a slight reduction in CLOCK binding in mutant animals. Within the Dbp gene there are also multiple E-boxes, and USF1 binding increases in Clock^Δ19/Clock^Δ19 animals on the promoter, and intron two E-boxes. On the Dbp promoter CLOCK and BMAL1 binding decreases in the Clock^Δ19/Clock^Δ19 animals. A similar pattern is observed for other clock genes such as Rev-erbα which contains multiple E-boxes (Figure 8E) and other target genes such as Ilf3 (Figure 8E).

We extended this analysis of ChIP-Seq data to 41 E-boxes in 15 ‘reference clock genes' that are thought to be core regulators of the circadian clock (Rey et al., 2011). We specifically analyzed these E-box sites in the two genotypes for USF1 binding changes. Of these 41 sites, 23 bound USF1 with a significant increase in USF1 binding in the Clock^Δ19/Clock^Δ19 animals (Figure 8F,G). Thus, USF1 binds to a majority of reference clock genes and its occupancy on these sites increases in the Clock^Δ19/Clock^Δ19 mice.

These data suggest an intriguing mechanism of rescue of the Clock^Δ19 heterozygous phenotype by USF1. In this model, USF1 expression is higher in BALB than in B6 strain owing to a promoter polymorphism. Our data show that USF1 can competitively bind to the same E-boxes in vivo as CLOCK:BMAL1. Furthermore quantitative analysis of binding kinetics indicates that the CLOCK^Δ19:BMAL1 complex has a much lower binding affinity for E-boxes than its wild-type counterpart. This lowered affinity combined with increased expression of USF1 in BALB mice accounts for the rescue of the Clock^Δ19/+ phenotype in this strain. We propose that USF1 acts as a partial agonist whose transcriptional regulation of circadian genes can compensate for the reduction in transcriptional activation by the CLOCK^Δ19:BMAL1 complex.

All animals used in this study were raised at Northwestern University or the University of Texas Southwestern Medical Center in a 12 hr light/12 hr dark cycle (LD12:12) from birth. All animal care and use procedures were in accordance with guidelines of the Northwestern University and UT Southwestern Institutional Animal Care and Use Committees. At 8–12 weeks of age, mice were transferred to individual cages equipped with running wheels and housed in LD12:12 conditions. After a minimum of 7 days, animals were transferred to DD conditions for 3 weeks. The free running period was calculated using a χ² periodogram as described previously (Shimomura et al., 2001).

Per2^Luc mice (Yoo et al., 2004) were euthanized by cervical dislocation between ZT11 and ZT12.5. Tissue was rapidly dissected as described in (Yamazaki and Takahashi, 2005). 300 μm coronal sections containing the SCN were isolated. Tissues were cultured on Millicell culture membranes (PICMORG50, Millipore, Billerica, MA) and were placed in 35 mm tissue culture dishes containing 2 ml DMEM media (90-013-PB, Gibco, Life Technologies, Grand Island, NY) supplemented with 352.5 μg/ml sodium bicarbonate, 10 mM HEPES (Gibco), 2 mM L-Glutamine, 2% B-27 Serum-free supplement (Invitrogen, Life Technologies, Grand Island, NY), 25 units/ml penicillin, 25 μg/ml streptomycin (Gibco), and 0.1 mM luciferin potassium salt (L-8240, Biosynth AG, Staad, Switzerland). Sealed cultures were placed in LumiCycle luminometer machines (Actimetrics, Wilmette, IL) and bioluminescence from the tissue was recorded continuously.

(BALB/cJ x C57BL/6J)F2 Clock^Δ19/+ mice were bred by intercrossing (BALB/cJ x C57BL/6J)F1 Clock^Δ19/+ animals. Only Clock^Δ19/+ mice were wheel-tested. Note that because Clock maps to chromosome 5, and the mutation was originally induced in the B6 background, it was not possible to scan for QTLs on this chromosome in the F2 cross. Genotyping primers used were: forward 5′-TACCAGCTGCTAATGTCCAGTG-3′; reverse 5′-TACATTGGGCTAGCCTTCCTAAG-3′. PCR conditions were 95°C for 2 min followed by 32 cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 15 s. Following amplification, PCR products were digested with 2 units of HincII for 2 hr at 37°C. After restriction digestion, PCR products were resolved on 4% agarose gels. The Clock mutant allele was identified as a ∼70-bp product, while wild-type animals were identified by the presence of bands at ∼50 and 20 bp.

SSLP markers used in this study are available upon request. The genotyping protocol has been described previously (Shimomura et al., 2001). QTL analysis was performed using MapmanagerQTX (http://www.mapmanager.org/↗).

(BALB/cJ x C57BL/6J)F1 Clock^Δ19/+ mice were backcrossed to C57BL/6J wild-type mice. Animals were genotyped with five SSLP markers (D1Mit22, D1Mit218, D1Mit33, D1Mit291 and D1Mit155) to identify individuals carrying BALB alleles of Soc. Following the N4 generation, we selected mice for breeding by both phenotype and genotype. After the N6 generation, selection was performed by genotyping with D1Mit452 and D1Mit155. Mice carrying smaller fragments of the Soc^BALB region and with suppressed phenotypes were crossed to C57BL/6J wild-type animals. After the 10th generation of phenotypic and genotypic selection, mice were intercrossed to produce the final congenic line.

All inbred strains were purchased from Jackson Laboratory (Bar Harbor, ME). These strains were crossed to C57BL/6J Clock^Δ19/+ females. For high-resolution mapping of the Soc locus, we used the 150 K SNP set from The Genomics Institute of the Novartis Foundation.

RNA was extracted using Trizol reagent (Invitrogen) according to the manufacture's protocol. RNA was then reverse-transcribed into cDNA by TaqMan RT kit. qPCRs were performed with SYBR green in an Applied Biosystems 7900HT Fast Real-Time PCR System (384 well). Primer sequences are described in supplementary file 1. Copy number estimates were performed as described previously (Uno and Ueda, 2007).

We used three different methods for detecting B6 and BALB allele specific expression in (BALB x B6)F1 Clock^Δ19/+ animals.

HEK293T cells were cultured at 37°C under a humidified atmosphere containing 5% CO₂ in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. We seeded cells into 24-well poly-D-lysine coated culture plates (∼2 × 10⁵ cells/well). After 24 hr, plasmid DNAs were transiently transfected into the HEK293T cells using Lipofectamine2000 reagent (Invitrogen) according to the manufacturer's protocol. After 24 hr of transfection, the cells were harvested and luciferase assay was performed and the raw numbers were normalized with a beta-actin-LacZ transfection control.

Mouse Usf1 cDNA clone (MGC:59,374) was obtained from the Mammalian Gene Collection (Open Biosystems) and was used to PCR-amplify Usf1 with forward (5′-GCCGCCACCATGAAGGGGCAGCAGAAAA-3′) and reverse primers (5′-TTAAAGAGCGTAATCTGGAACATCGTATGGGTAGTTGCTGTCATTCTTGATGACGACCT-3′). The PCR product containing a modified Usf1 with a Kozak consensus sequence and a 3′ HA tag was inserted into pBluescript II KS- using standard cloning procedures. Site-directed mutagenesis was performed with forward (5′- GCAGGGAGGGAGCCAGCGATC-3′) and reverse (5′- GATCGCTGGCTCCCTCCCTGC -3′) primers using Quikchange XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The product was reamplified with forward (5′-GCCGCCACCATGAAGGGGCAGCAGAAAA-3′) and reverse (5′-TTAAAGAGCGTAATCTGGAACATCGTATGGGTAGTTGCTGTCATTCTTGATGACGACCT-3′) primers and subcloned into the EcoRV restriction site in PMM-400 (provided by Mark Mayford). Transgenic mouse lines were generated by pronuclear injection using standard techniques as described (Hong et al., 2007). The linearized DNA fragment was injected at a concentration of 1 ng/μl into fertilized mouse oocytes isolated from crosses of C57BL/6J matings to produce transgenic mice that were isogenic on C57BL/6J to exclude the possibility of contamination by genetic suppressor backgrounds. Transgenic mice were identified by PCR analysis of genomic DNA prepared from tail biopsy samples. Two sets of primers were used for genotyping: USF1-5F: 5′-CCAAAAACGAGGAGGATTTG-3′ and USF1-5R: 5′-GTGGCAGGGTAACCACTGAT-3′, or USF1-3F: 5′-GCAGGGGTTAGATCAGTTGC-3′ and USF1-3R: 5′-TGCTCCCATTCATCAGTTCC-3′. PCR conditions were: 95°C for 2 min followed by 32 cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 30 s. Following amplification, PCR products were resolved on 2% agarose gels.

Livers from mice were immediately homogenized in 4 ml per liver of 1× PBS containing 2 mM EGS. The homogenate was incubated for 20 min at room temperature. Next, 110 µl of 36.5% formaldehyde was added followed by an 8 min incubation at room temperature. Cross-linking reactions were stopped by the addition of 250 µl of 2.5 M glycine and kept on ice. The homogenate was added to 10 ml of ice-cold 2.3 M sucrose containing 150 mM glycine, 10 mM HEPES pH 7.6, 15 mM KCl, 2 mM EDTA, 0.15 mM spermine, 0.5 mM spermidine, 0.5 mM DTT and 0.5 mM PMSF, and layered on top of a 3 ml cushion of 1.85 M sucrose (containing the same ingredients with 10% glycerol) and centrifuged for 1 hr at 24,000 rpm at 4°C in a Beckman SW32.1 rotor. The nuclei were resuspended in 1 ml of 20 mM Tris pH 7.5, 150 mM NaCl, 2 mM EDTA, transferred to a 1.5 ml microfuge tube, centrifuged for 3 min at 3000 rpm at 4°C, washed again, and stored –80°C until use. The nuclei were then resuspended in 0.8 ml per liver of lysis buffer (10 mM Tris pH 7.5, 1 mM EDTA, 0.5 mM EGTA, 0.2 M NaCl, 0.5% N-lauroylsarcosine, 0.1% sodium deoxycholate, 1 mM PMSF and EDTA-free protease inhibitor cocktail; Roche, Indianapolis, IN), and sonicated 5 s at 4°C 36 times with a Covaris S2. The fragmented chromatin was then diluted tenfold with IP buffer (10 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% triton X-100, 0.1% sodium deoxycholate, 1 mM PMSF, EDTA-free protease inhibitor cocktail). Approximately 120 µg of fragmented chromatin was pre-cleared by incubating with 40 µl of protein A-agarose (Sigma, St. Louis, MO) for 2 hr at 4°C with rotation. Pre-cleared chromatin was then incubated with antibody overnight at 4°C with rotation, followed by addition of 10 µl of Protein A/G Plus-agarose and incubation for 1.5 hr at 4°C. Agarose beads were then washed twice with IP buffer, twice with high salt wash buffer (20 mM Tris pH7.5, 500 mM NaCl, 2 mM EDTA, 1% Triton X-100, 1 mM PMSF), twice with LiCl wash buffer (20 mM Tris pH 7.5, 250 mM LiCl, 2 mM EDTA, 0.5% Igepal CA-630, 1% sodium deoxycholate, 1 mM PMSF), and once with TE. Co-immunoprecipitated DNA fragments were eluted with 100 µl of 20 mM Tris pH 7.5, 5 mM EDTA, 0.5% SDS, then reverse crosslinked at 65°C overnight, incubated with 10 µg of RNaseA for 30 min at 37°C, with 160 µg of proteinase K for 30 min at 55°C, and then purified using the QIAquick PCR purification Kit (Qiagen, Germantown, MD). ChIP products were analyzed by qPCR using SYBR green. Primer sequences are described in supplementary file 1.

ChIP-Seq libraries were prepared as described (Kim et al., 2010; Koike et al., 2012). SOLiD sequencing of ChIP-Seq libraries were performed on an ABI SOLiD4 instrument with 35-bp reads according to manufacturer's instructions (Life Technologies, Grand Island, NY) by the UTSW McDermott DNA Sequencing Core. Sequence reads were mapped to the mouse genome (NCBI m37) with Applied Biosystems BioScope v1.3. The peaks were identified from uniquely mapped reads without duplicates using MACS (Zhang et al., 2008) followed by PeakSplitter (Salmon-Divon et al., 2010). Sequence reads were checked for abnormal read content in input samples that led to global false positive peak detection that arise primarily from low complexity sequence and contamination from plasmid/cDNA sequences. Global false positive peaks were removed from peak lists derived from MACS and PeakSplitter, and tags derived from plasmids/cDNAs were not analyzed as described previously (Heinz et al., 2010). Results were further analyzed using HOMER (Heinz et al., 2010). Peak overlaps (peak summit ± 120 bp) were determined with HOMER. Genome browser views were normalized as uniquely mapped reads per 10 million reads.

Liver nuclear extract was prepared as described previously (Yoshitane et al., 2009). Mice were sacrificed by cervical dislocation at ZT8. Whole liver was quickly removed and briefly washed with ice-cold PBS and homogenized with seven strokes of a dounce tissue homogenizer (Pestle A only) at 4°C in 9 ml of ice-cold buffer A (10 mM HEPES-NaOH [pH 7.8], 10 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride [PMSF], 4 mg/ml aprotinin, 4 mg/ml leupeptin, 50 mM NaF, and 1 mM Na₃VO₄). The homogenate was centrifuged twice (5 min each, 700×g), and the precipitate was resuspended in 2 ml of ice-cold buffer C (20 mM HEPES-NaOH [pH 7.8], 400 mM NaCl, 1 mM EDTA, 5 mM MgCl₂, 2% [v/v] glycerol, 1 mM DTT, 1 mM PMSF, 4 mg/ml aprotinin, 4 mg/ml leupeptin, 50 mM NaF, and 1 mM Na₃VO₄). After gentle mixing at 4°C for 30 min, the suspension was centrifuged twice (30 min each at 21,600×g), and the final supernatant was used as a nuclear extract.

EMSA was performed in a manner similar to that described previously (Ripperger and Schibler, 2006). For oligonucleotide annealing, 1 nm of each forward and reverse oligonucleotide (supplementary file 1) were mixed with 1× T4 polynucleotide buffer at final volume 50 μl. The reaction mix was heated at 95°C for 10 min then slowly cooled to 25°C (1 °C/min). Oligonucleotide labeling was performed in 50 μl with 50 pmole of double-stranded oligo, 5 μl of T4 polynucleotide kinase buffer (NEB, Ipswich, MA), 2.5 μl of T4 polynucleotide kinase (NEB), 3.75 μl ATPγ³²P (Perkin Elmer cat# BLU002Z250UC, Waltham, MA), and 36.25 μl water. The reaction mix was incubated at 37°C for 30 min followed by 65°C for 10 min. After the kinase reaction, the sample was purified through a G50 spin column. EMSA reactions were performed with 15 μg of nuclear protein in 25 mM HEPES-KOH (pH 7.6), 150 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 200 ng/μl sheared salmon sperm DNA, 50 ng/μl poly(dI-dC), and 2 μl of radioactive probe in a volume of 12 μl. Binding buffer and nuclear protein were mixed and incubated at room temperature for 15 min. Then, 2 μl of radioactive probe was added and incubated at 16°C for 30 min. For supershifts, 1 μl of antibody was added immediately before the addition of the radio-labeled probe. The antibodies used were anti-BMAL1, anti-CLOCK (a gift from Dr. Choogon Lee) as described previously (Koike et al., 2012), and anti-USF1 (sc-8983×, Santa Cruz Biotech, Dallas, TX). 1 μl of 15% Ficoll was added, and the reaction mixes were loaded on 4% polyacrylamide gels (0.5× TBE with 2.5% glycerol). Gels (0.2 mm thickness, 4% polyacrylamide, Bio-Rad #161–0144) were run for 90 min at room temperature (7.5 V/cm) and exposed on phosphor imager screens for 18 hr at 4°C. For saturation binding assays, increasing amounts of probe were added to a fixed amount of nuclear extract in triplicate. A four-parameter nonlinear least squares curve fitting method (Prizm, Graphpad Software, La Jolla, CA) was used to estimate the B_max and K_d values and to conduct tests for whether each best-fit parameter differs among the four groups in Table 2.

Mouse embryonic stem cell gene trap line W233A07 and P116C05 were obtained from the German Gene Trap Consortium (http://www.genetrap.de↗). The ROSAbetageo+1 gene trap vector (incorporating fused beta galactosidase and neomycin) was introduced to 129S2/SvPas-derived TBV2 embryonic stem (ES) cells to generate mutant cell lines. Both W233A07 and P116C05 cells, carrying a vector insertion into the intron one of Usf1 gene, were used to generate mice which were backcrossed to C57BL/6J for four generations. In order to detect presence of trap vector, Usf1-TrapFwd: 5′-AGTGACAACGTCGAGCACAG-3′, Usf1-TrapRev: 5′-CGGTCGCTACCATTACCAGT-3′ were used. To discriminate +/− from −/− mouse, we used real time PCR using Usf1-Trap 129Fwd:5′- GGAGGTGGGGATTATAGTCTGAA-3′, Usf1-Trap B6Fwd:5′- GGAGGTGGGGATTATAGTCTGAG-3′ and Usf1-Trap Rev:5′-TGCTACAAGGAGGGGTTCTG-3′. Since mutant allele at Usf1 locus is derived from 129S2/SvPas, +/− mouse is 129/B6 heterozygous and +/− mouse is 129/129 homozygous at Usf1 locus.