Interleukin-like EMT inducer (ILEI) promotes melanoma invasiveness and is transcriptionally up-regulated by upstream stimulatory factor-1 (USF-1)

Interleukin-like EMT inducer (ILEI, FAM3C) is a secreted factor that contributes to the epithelial-to-mesenchymal transition (EMT), a cell-biological process that confers metastatic properties to a tumor cell. However, very little is known about how ILEI is regulated. Here we demonstrate that ILEI is an in vivo regulator of melanoma invasiveness and is transcriptionally up-regulated by the upstream stimulatory factor-1 (USF-1), an E-box–binding, basic-helix-loop-helix family transcription factor. shRNA-mediated knockdown of ILEI in melanoma cell lines attenuated lung colonization but not primary tumor formation. We also identified the mechanism underlying ILEI transcriptional regulation, which was through a direct interaction of USF-1 with the ILEI promoter. Of note, stimulation of endogenous USF-1 by UV-mediated activation increased ILEI expression, whereas shRNA-mediated USF-1 knockdown decreased ILEI gene transcription. Finally, we report that knocking down USF-1 decreases tumor cell migration. In summary, our work reveals that ILEI contributes to melanoma cell invasiveness in vivo without affecting primary tumor growth and is transcriptionally up-regulated by USF-1.

Interleukin-like EMT inducer (ILEI, FAM3C) is a secreted cytokine-like molecule that contributes to the epithelial-tomesenchymal transition (EMT) (16). EMT is a cell-biological process in which epithelial cells with apical-basal polarity undergo cytoskeletal rearrangement to become motile mesenchymal cells (17). This process is thought to contribute to chemoresistance and metastasis. Although melanoma cells do not undergo a traditional EMT, they utilize a similar process known as phenotype switching. This is a process in which melanoma cells interconvert between a proliferative MITF-high state and an invasive MITF-low state (18 -24). Recently, we have described a contribution of ILEI to the invasive MITF-low phenotype in vitro (25). Additionally, we showed that phenotype switching between the proliferative MITF-high and invasive MITF-low state modulates ILEI mRNA expression. The molecular regulation of ILEI has focused on post-transcriptional mechanisms including translational regulation by hnRNP-E1-TGF-␤ and proteolytic processing by plasmin (26 -30). However, not much is known about transcriptional regulation of ILEI.
Here, we report that ILEI contributes to melanoma cell line invasiveness in vivo and build on our previous finding that vemurafenib (BRAFi) decreases ILEI mRNA expression to characterize the molecular details of ILEI transcriptional regulation by USF-1 (25).

ILEI regulates lung colonization but not primary tumor growth
Previously we found that knockdown of ILEI contributes to melanoma invasiveness in vitro, whereas ILEI had no effect on proliferation (25). Here we used the same ILEI-modulated cell lines and conducted in vivo flank and tail vein injections to measure primary tumor formation or lung colonization, respectively. In the flank injection experiment we found no significant difference in primary tumor growth (Fig. 1, A and B), but we observed in the tail vein injection experiment that ILEI knockdown significantly attenuated lung colonization (Fig.  1C). We did not observe a difference in the size of the lung colonies formed by either shSCR or shILEI cell lines, suggesting that the ability to invade the lung was affected rather than the ability to grow metastatic colonies in the lung (Fig. 1D). Based on these results we conclude that ILEI specifically regulates melanoma invasiveness in vivo without affecting primary tumor growth.

Vemurafenib inhibits ILEI expression
To date there are three known mechanisms of ILEI regulation: the first is translational regulation of ILEI by TGF-␤-AKT2-hnRNP-E1, the second is degradation of ILEI by the ubiquitin/proteasome system, and the third is an autophagymediated increase in ILEI protein expression (16, 25, 26, 30 -33). Nothing has been reported about the mechanistic basis of ILEI transcription. None of the previously established mechanisms appears to be responsible for ILEI regulation by vemurafenib (25). Herein, we sought to determine the mechanism of ILEI transcriptional regulation.
First, we treated WM983B melanoma cell lines with vemurafenib, an inhibitor of V600E BRAF, and observed that ILEI protein and mRNA expression decreased at 24 h (Fig. 2, A and   B) (25). We also conducted RT-PCR using primers targeting an intronic sequence of ILEI to amplify ILEI pre-RNA (Fig. 2B). If ILEI mRNA levels are regulated post-transcriptionally by miRNAs, we expect that ILEI pre-RNA should not be affected by vemurafenib. However, if ILEI mRNA levels are regulated transcriptionally, we expect that ILEI pre-RNA should go down upon vemurafenib treatment. We saw that vemurafenib treatment decreased both ILEI total RNA and pre-RNA, suggesting that vemurafenib affected ILEI transcription (Fig. 2B). To confirm that these findings were not due to off-target effects of vemurafenib we conducted a control experiment in which we used the BRAF WT WM3918 melanoma cells. Vemurafenib is a specific inhibitor of mutant BRAF, and it does not affect WT BRAF (34). Therefore, we treated BRAF WT WM3918 melanoma cells with vemurafenib, which should retain all the nonspecific effects of vemurafenib without the BRAF-specific effects. We found that vemurafenib did not affect ILEI protein expression or ERK phosphorylation in BRAF WT WM3918 melanoma cells (Fig. 2C). We confirmed these findings by a quantitative method using real-time qPCR and found that vemurafenib decreased ILEI expression in BRAF mutant 501-Mel and 1205Lu cells but not in BRAF WT WM3918 cells (Fig.  2D). Considering that ILEI is a secreted cytokine, we wanted to confirm the physiological relevance of vemurafenib-mediated ILEI inhibition by conducting immunoblot analysis of the conditioned medium. These results confirmed that vemurafenib decreased secreted ILEI levels (Fig. 2E). Finally, to further determine whether this effect is specific for the MEK pathway

USF-1 regulates ILEI transcription
we used a second inhibitor of the MEK pathway (U0126, MEKi) and observed decreased ILEI expression by U0126 (Fig. 2F). Based on these results we conclude that vemurafenib inhibits ILEI mRNA expression, and that this effect depends on the presence of oncogenic BRAF V600E mutation.

Vemurafenib inhibits ILEI at the transcriptional level
We cloned the ILEI promoter from 2,300 bp upstream of the transcription start site (TSS) to 80 bp upstream of the TSS into pBV-Luc (Fig. 3A) (35). We transfected 501-Mel cells with the ILEI promoter luciferase reporter construct along with a control Renilla construct and treated the cells with vemurafenib ( Fig. 3B). We found that vemurafenib decreased the ILEI promoter activity in a dose-dependent manner. As further controls we tested the ILEI promoter construct with the MEK inhibitor U0126 and found that U0126 decreased promoter activity (Fig.  3C). However, when we tested the ILEI promoter construct with the PI3K inhibitor LY-294002, we found that promoter activity was not affected (Fig. 3D). Additionally, we cloned the ILEI 3Ј-UTR from the stop codon to the end of the mRNA at 1,620 bp into pmirGLO Dual Luciferase (Fig. 3E). We transfected 501-Mel cells with the ILEI 3Ј-UTR construct and treated the cells with vemurafenib (Fig. 3F). We found that vemurafenib did not affect the ILEI 3Ј-UTR construct. From

USF-1 regulates ILEI transcription
these experiments we concluded that vemurafenib regulates ILEI mRNA at the transcriptional level.
Because vemurafenib regulates ILEI transcription, we hypothesized that a cis element in the ILEI promoter region regulates this phenomenon. We analyzed a 2,300-bp sequence at the ILEI promoter from the TSS for various transcription factor motifs using the JASPAR database (36). Previously we had found that ILEI is highly expressed in MITF-low invasive melanoma cell lines, so we focused our transcription factor search on regulators of the MITF-low invasive state (E-box (ZEB1/2), JUN, and TEAD4) (20, 21, 23, 37-39). We found 11 putative E-box, 5 JUN, and 3 TEAD4 sites in the ILEI promoter, and generated successive 5Ј-deletions of the ILEI promoter reporter (Fig. 4A). We transfected 501-Mel cells with these constructs and observed that the truncation from Ϫ204 to Ϫ150 reduced the luciferase activity to empty promoter control levels. There was an E-box motif in this region, so we next wanted to know if this E-box was specifically important for ILEI promoter activity or if any E-box would suffice. Thus, we used our longest ILEI promoter reporter (Ϫ2300 to Ϫ80) and specifically mutated the E-box consensus site at Ϫ163 from CACGTG to CAAATG. Again, we observed a marked inhibition in luciferase activity  (Fig. 4B). From these experiments we concluded that an E-box 163 bp upstream of the ILEI TSS is critical to basal ILEI promoter activity.

USF-1 regulates ILEI transcription
Next, we wanted to know the role of this E-box motif in the vemurafenib-mediated regulation of ILEI expression. We transfected 501-Mel cells with the ILEI promoter truncation constructs and treated the cells with vemurafenib ( Fig. 5A). We found that truncation of bp Ϫ204 to Ϫ150, which includes an E-box motif, eliminated ILEI promoter vemurafenib responsiveness. The Ϫ204 to Ϫ150 region of the ILEI promoter contains 48 bp in addition to the E-box, so we mutated the E-box in the Ϫ300/Ϫ80 construct. We sequenced the construct and analyzed the sequence in the JASPAR database to find that our mutations abolished all putative binding sites in this region (data not shown). We conducted luciferase assays and observed that mutation of the E-box similarly eliminated vemurafenib responsiveness (Fig. 5B). From these experiments we concluded that vemurafenib inhibits ILEI at the transcriptional level through an E-box 163 bp upstream of the ILEI TSS.

USF-1 directly regulates ILEI transcription
Considering the importance of the E-box motif to ILEI expression, we wanted to identify the trans-acting factor regulating this effect. To this end, we overexpressed different transcription factors known to bind E-box motifs and found that USF-1 induced ILEI promoter activity, but not c-MYC, N-MYC, L-MYC, TFEB, TFE3, MITF, CREB3L2, and ID2 (Fig.  6A). USF-1 activates transcription through the E-box motif, but also through a pyrimidine-rich initiator sequence (Inr) (40).
The ILEI promoter sequence has an Inr sequence 6 bp downstream of the E-box (Fig. 6B). Thus, we generated Inr mutant ILEI promoter constructs in addition to our E-box mutants to see if USF-1 regulated ILEI promoter activity directly through the USF-1-binding sites (Fig. 6C). We observed that USF-1 induced ILEI promoter activity in the WT promoter, the E-box mutant promoter, the initiator mutant promoter, but not the E-box and Inr double mutant promoter (Fig. 6D). From these experiments we concluded that USF-1 directly regulates ILEI promoter activity.
We wanted to further prove a direct role of USF-1 in ILEI regulation so we pursued a potential interaction between the ILEI promoter and USF-1. We conducted streptavidin pulldown experiments using 5Ј-biotin-tagged ILEI promoter constructs either WT or mutant for the E-box (200 to 110 bp upstream of ILEI TSS, Fig. 7A) and 501-Mel melanoma cell nuclear extract. We found that USF-1 can be isolated from cell nuclear extract using the WT ILEI promoter, but not the E-box mutant ILEI promoter, or a control random 60-bp oligonucleotide (SCR, Fig. 7A).
Interestingly, the E-box mutation was sufficient to abolish binding by the streptavidin pulldown experiment, but the E-box mutation was not sufficient to abolish USF-1-mediated induction of the ILEI promoter reporter (Fig. 6D). We speculate either that 1) USF-1 still binds the ILEI promoter at levels that are undetectable by our streptavidin pulldown assay, or 2) endogenous USF-1 binds only to the E-box but overexpressed exogenous USF-1 provides excess USF-1 to bind both the E-box and the Inr. This is supported by the finding that basal ILEI

USF-1 regulates ILEI transcription
promoter activity is affected more by the mutation of the E-box than the Inr (Fig. 6D). This is also consistent with previously published findings that show the E-box is the high-affinity USF-1-binding site, whereas the Inr is the low-affinity binding site (41).
To confirm that the interaction between USF-1 and the ILEI promoter was physiologically relevant, we conducted ChIP using USF-1 antibody and PCR primers flanking the ILEI promoter E-box (Fig. 7B). We confirmed the efficacy of our assay by showing that USF-1 IP can detect a segment of the TYR promoter (15), and we detected an interaction between USF-1 and the ILEI promoter (Fig. 7B). However, we found that a segment of the HO1 promoter, which is a USF-1 interaction reported in epithelial cells, was not detected in our melanoma cells (42). From these experiments we concluded that USF-1 directly interacts with the ILEI promoter and this interaction is E-box dependent.

USF-1 in vemurafenib-regulated ILEI expression
Thus far we have shown that vemurafenib inhibits ILEI transcription through an E-box motif, and that USF-1 directly regulates ILEI transcription through the same E-box motif. Next, we wanted to understand the role of USF-1 in vemurafenibmediated regulation of ILEI transcription.
For this we employed the streptavidin pulldown assay using 5Ј-biotin-tagged ILEI promoter constructs WT or E-box mutant in cells treated with vemurafenib. First, we validated the binding of USF-1 to the ILEI promoter (DMSO condition) and showed that vemurafenib treatment abolished USF-1 interaction with the ILEI promoter (Fig. 8A). Importantly, we noticed by comparing the input lanes that USF-1 expression is decreased by vemurafenib treatment. We used qPCR to confirm that vemurafenib represses the mRNA levels of USF-1 (Fig. 8B).
Finally, we confirmed these results with the endogenous ILEI by overexpressing USF-1, treating with vemurafenib, and conducting PCR analysis. Although the rescue was not complete, we observed that USF-1 overexpression rescues vemurafenib inhibition of ILEI (Fig. 8D). From these experiments we concluded that the mechanism of vemurafenib-mediated inhibition of ILEI is through down-regulation of USF-1 mRNA.

Role of endogenous USF-1 in ILEI expression
We wanted to assess if endogenous USF-1 had a role in ILEI expression. USF-1 is known to be regulated by stress-mediated p38 MAPK. Mechanistically, p38 phosphorylates USF-1 on threonine 153, which activates its transcriptional activity for target genes, including TYR (15). Thus, we used UV treatment as a model of endogenous USF-1 activation and observed that UV increased ILEI mRNA both by RT-and qPCR (Fig. 9, A and  B). Furthermore, we wanted to know if UV-USF-1-ILEI regulation was intact in nonmelanoma cell lines. We used HMLE human mammary epithelial cells and observed that UV increased ILEI mRNA by qPCR (Fig. 9C). In addition to activation of endogenous USF-1, we wanted to test the effect of inhibiting endogenous USF-1. We used two different shRNA mole-
Finally, we wanted to know if USF-1 regulation of ILEI had any biological significance. Given the importance of ILEI to the invasive melanoma phenotype (Fig. 1), we tested the contribution of USF-1 to wound healing. We observed that USF-1 knockdown in 501-Mel melanoma cells attenuates wound healing (Fig. 10, A and B). From these experiments we concluded that USF-1 regulates migration in melanoma cell lines.
In summary, we have described the following novel findings: ILEI regulates melanoma invasiveness in vivo, vemurafenib inhibits ILEI at the transcriptional level through a specific E-box sequence, USF-1 directly interacts with this E-box in the ILEI promoter and that this interaction is abolished upon vemurafenib treatment, and finally that endogenous USF-1 contributes to ILEI expression and the invasive melanoma phenotype. Through these findings we have established a novel regulatory mechanism for ILEI expression by USF-1, and based on this previously unknown context for ILEI expression we speculate on the following new biological functions for USF-1 and ILEI (Fig. 11).

USF-1 regulates ILEI transcription
the impact of vemurafenib on BRAF mutant melanoma, the mechanisms of vemurafenib-mediated tumor killing have been thoroughly characterized. For instance, mouse models have shown that vemurafenib induces apoptosis of tumor suppressive regulatory T cells while simultaneously promoting CD8 T cell-mediated killing of tumor cells (46). Other models have shown that vemurafenib increases the recruitment of macrophages to BRAF mutant melanomas (47). Combining our finding that vemurafenib inhibits USF-1-ILEI with the established finding that vemurafenib regulates the tumor microenvironment, we suggest that both USF-1 and ILEI could be regulators of the tumor immune microenvironment. Future studies should seek to elucidate the role of USF-1/ILEI in this paracrine signaling context.
Additionally, USF-1 is primarily known as a regulator of glucose and lipid metabolism and is poorly described as an oncogene (48 -50). In melanoma USF-1 is known as a stress responsive transcription factor involved in UV-induced, but not MITF-mediated constitutive pigmentation (51). The non-overlapping function of USF-1 with MITF in pigmentation suggests that USF-1 could play other roles in melanoma biology when MITF is poorly expressed, such as in MITF-low invasive melanoma cells. These cells are characterized by the lack of MITF but not by the presence of any major transcription factor,

USF-1 regulates ILEI transcription
although ZEB1 and JUN have been proposed (20,21,23,37,38). Considering that ILEI is a regulator of MITF-low cells we speculate that USF-1 could be a novel transcription factor regulating the MITF-low invasive state (Fig. 11) (25). Future work should address the role of USF-1 in melanoma phenotype switching.

Animal studies
All procedures were approved by the Institutional Animal Care and Use Committee at the Medical University of South Carolina. Lung colonization experiments were conducted as follows: tail vein injection into 6 -8 -week-old mice were performed with 1205Lu cells expressing shSCR or shILEI 3 (1 ϫ 10 5 cells in 100 l of PBS). Cells were previously characterized (25). After 8 weeks, mice were sacrificed, and lungs were harvested. Organs were formalin-fixed and paraffin-embedded. Histopathological analysis was conducted by Hollings Cancer Center Biorepository & Tissue Analysis Shared Resource. Lung colonies were counted by two individuals.

Constructs
ILEI reporter constructs were cloned as follows: the ILEI promoter sequence from 2,300 bp upstream of the predicted transcription start site to 80 bp upstream was amplified by PCR from 501-Mel genomic DNA with 5Ј KpnI and 3Ј NheI primers and ligated into pBV-Luc (pBV-Luc vector (a gift from Bert Vogelstein (Addgene plasmid number 16539)) (35). Primers are listed in Table 1. Subsequent ILEI promoter truncation constructs were cloned by PCR using an alternate 5Ј KpnI primer and the same 3Ј NheI primer, or using a Q5 SDM kit from New England Biolabs (Ipswitch, MA). The ILEI 3Ј UTR construct from stop codon to 1,620 bp downstream of the stop codon was amplified by PCR from 501-Mel cDNA with 5Ј NheI and 3Ј SalI primers and ligated into pmirGLO Dual Luciferase vector (RAB14 3Ј UTR WT was a gift from Curt Civin (Addgene plasmid number 61489)) (52). The following constructs were used in this study: pcDNA3-cmyc (a gift from Wafik El-Deiry (Add-  31151)), and USF1, CREB3L2, and ID2 coding sequence constructs in pLX304 were generated by David Root and supplied by DNASU (57)(58)(59)(60). Corresponding empty vector for human L-MYC was constructed by digesting the L-MYC plasmid with NotI to remove the L-MYC coding sequence and religating using T4 ligase.
Stable cell lines were generated by lentiviral transduction with Polybrene (8 g/ml; Sigma). 24 h post-transduction the medium was changed, and 48 h post-transduction the cells were selected and cultured with 0.125-0.5 g/ml of puromycin (InvivoGen). Pools of stably transduced cells were analyzed. shRNA sequences are listed in Table 2. UV treatments were conducted using Fisher Scientific UV Cross-linker FB-UVXL-

USF-1 regulates ILEI transcription
1000 at 0, 5, or 50 J/m 2 in room air, and then incubating in normal cell culture conditions for 24 h.

Lentivirus
Lentiviral shRNAs were obtained from the MUSC Hollings Cancer Center shRNA Shared Resource Technology. All vectors used in this study are listed in Table 2. Lentivirus was generated by seeding 293T (1,000,000 cells; Takara Bio, Mountainview, CA) to a 60-mm cell culture dish, and transfecting with 6 l of Lipofectamine 2000 (ThermoFisher), 1 g of pLKO vector, 0.75 g of psPAX2, and 0.25 g of pMD2.G. 24 h posttransfection the media was changed, and 48 and 72 h posttransfection the media was harvested. Viral supernatant was cleared by centrifugation, filtered through a 0.22-m filter, and stored at Ϫ80°C until use.

Luciferase analysis
Cell lines were seeded at 50,000 -75,000 cells per well in 24-well plate in 0.5 ml of complete medium. At 24 h the cells were transfected using X-tremeGENE 9 (Roche Applied Science, Switzerland; 100 ng of firefly experimental luciferase, 5 ng of Renilla control luciferase, 0.

PCR analysis
Total RNA was isolated using TRIzol (ThermoFisher Scientific). Reverse transcription was performed using oligo(dT) primers and Moloney-murine leukemia virus Reverse Transcriptase (New England BioLabs). Semi-quantitative PCR was conducted on 10 ng of cDNA using Maxima Hot Start PCR Master Mix (ThermoFisher Scientific). Real-time quantitative PCR was conducted using iQ SYBR Green Supermix (Bio-Rad) using CFX384 Real-Time System (Bio-Rad). Reactions were conducted on 50 to 10 ng of cDNA. Primers are listed in Table  1. Relative gene expression was calculated using RFX Manager software, and genes were normalized to GAPDH internal control.

Chromatin immunoprecipitation (ChIP)
ChIP protocol was modified from Carey et al. (61). Briefly, 1.5 ϫ 10 7 cells were fixed in 1% formaldehyde for 10 min at room temperature and quenched in 125 mM glycine. Cells were harvested in lysis buffer (5 mM PIPES, pH 8, 85 mM KCl, 0.5% Nonidet P-40), and centrifuged for 10 min (3,000 rpm at 4°C). Supernatant (cytosolic fraction) was removed, and the pellet was resuspended in nuclei lysis buffer (50 mM Tris, pH 8, 10 mM EDTA, 1% SDS, protease and phosphatase inhibitor tablets). Nuclear extracts were sonicated for 10 min (30 s on and 30 s off), and cleared by centrifugation (2 times, 10 min, 13,000 rpm at 4°C). The supernatant was considered the chromatin fraction. 100 g of chromatin samples were precleared for 2 h at 4°C with 30 l of slurry ChIP-Grade Protein G-Agarose Beads (9007; Cell Signaling). The resulting supernatants were incubated overnight on a rotator at 4°C with 5 g of control Mouse (G3A1) mAb IgG1 Isotype Control (5415; Cell Signaling) or USF1 antibody (C-20). 30 l of slurry ChIP-Grade Protein G-Agarose Beads was added to each sample for 2 h on a rotator at 4°C. Beads were washed 4 times in high-salt wash buffer (50 mM HEPES, pH 7.9, 500 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, 0.1% deoxycholate) each time incubating for 10 min on a rotator at room temperature. Beads were next washed 2 times in TE buffer each time incubating for 10 min on a rotator at room temperature. Beads were resuspended in 300 l of elution buffer (50 mM Tris, pH 8, 10 mM EDTA, 1% SDS) ϩ 20 g of RNase A (R1253; ThermoFisher Scientific) and incubated on a 55°C heat block for 2 h. 1.6 milliunits of proteinase K (P8107S; New England BioLabs) were added prior to another incubation on a 55°C heat block for 2 h. The samples were transferred to a 65°C heat block for overnight elution. The samples were purified in 30 l of H 2 O using GeneJET PCR Purification Kit (K0702; ThermoFisher Scientific). PCR was conducted with primers used for FAM3C promoter cloning specific for human FAM3C promoter from 300 to 80 bp upstream of TSS, as listed in Table 1, human TYR promoter or human HMOX1 promoter (15,42).

Biotin DNA pulldown
DNA pulldown assays were conducted with 5Ј-biotinylated double-stranded annealed oligonucleotides corresponding to FAM3C promoter from Ϫ200 to Ϫ110 upstream of the TSS. A mutant E-box construct was used in which the E-box at Ϫ162 to Ϫ156 was mutated from CACGTG to CAAATG. A 5Ј-biotinylated double-stranded annealed oligonucleotide of a random sequence was used as a negative control (SCR).
Nuclear extracts were isolated by harvesting cells in lysis buffer (10 mM HEPES, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT, 0.05% Nonidet P-40, pH 7.9, protease and phosphatase inhibitor tablets) and pelleting cell nuclei. Pellets were incubated for 30 min on ice in nuclei extraction buffer (5 mM HEPES, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT, 26% glycerol, pH 7.9, protease and phosphatase inhibitor) and NaCl was added to a final concentration of 300 mM. Samples were homogenized by passing 20 times through a 28-guage needle. Samples were incubated on ice for 30 min and centrifuged. The supernatants were considered the nuclear extract. Nuclear extract (20 g) was reserved as 5% input.
Preclear beads were prepared as follows: 5Ј-biotinylated random oligonucleotide (SCR) was bound to streptavidin-agarose beads (50 l of slurry ϩ 10 g of oligo; ThermoFisher) for 1 h on a rotator at room temperature. IP beads were prepared as follows: 5Ј-biotinylated FAM3C promoter WT or E-box mutant (CACGTG to CAAATG) was bound to streptavidin-agarose beads (50 l of slurry ϩ 1 g of oligo; ThermoFisher) for 1 h on a rotator at room temperature. Preclear was conducted as follows: 400 g of nuclear extract was incubated with preclear beads for 1 h on a rotator at room temperature. The resulting sample was centrifuged and the supernatant was precleared again for a total of three preclears. The final supernatant was incubated on IP beads overnight on a rotator at 4°C. The beads were washed in PBS, 0.1% Triton X-100 and resolved by SDS-PAGE, transferred to nitrocellulose, and probed using antibody against USF1 (C-20; sc-229; Santa Cruz; 1:1,000).

Statistical analyses
Data are mean Ϯ S.D. unless indicated otherwise. p Ͻ 0.05 by unpaired two sample Student's t test is considered significant. Representative experiments are repeated at least twice.

Wound-healing assays
Cells (3 ϫ 10 5 , 0.5 ml of complete medium) were seeded in a 24-well plate, and a 1-ml pipette tip was used to scratch the cells. Images were recorded from 0 to 24 h, and analyzed using ImageJ (National Institutes of Health, Bethesda, MD).