GATA zinc finger protein p66β promotes breast cancer cell migration by acting as a co-activator of Snail

The transcriptional repressor Snail induces EMT during embryonic development and tumor metastasis. Growing evidence indicates that Snail functions as a trans-activator to induce gene expression; however, the underlying mechanism remains elusive. Here, we report that Snail cooperates with GATA zinc finger protein p66β to transactivate genes in breast cancer cells. Biologically, depletion of p66β reduces cell migration and lung metastasis in BALB/c mice. Mechanistically, Snail interacts with p66β and cooperatively induces gene transcription. Notably, a group of genes induced by Snail harbor conserved G-rich cis-elements (5′-GGGAGG-3′, designated as G-box) in their proximal promoter regions. Snail directly binds to G-box via its zinc fingers and transactivates the G-box-containing promoters. p66β enhances Snail binding affinity to G-box, whereas depletion of p66β results in a decreased binding affinity of Snail to the endogenous promoters and concomitantly reduces the transcription of Snail-induced genes. Taken together, these data demonstrated that p66β is critical for Snail-mediated cell migration by acting as a co-activator of Snail to induce genes containing G-box elements in the promoters.

Notably, growing evidence indicates that Snail can transactivate gene expression directly, but distinct models are proposed to explain the transactivation of Snail [23]. For example, Snail binds to the E-box-rich regions of ERCC1 and MMP15 promoters to induce their expression by a potential mechanism that Snail interacts with CBP to acetylate K146 and K187 residues, which results in preventing the formation of the repressor complex [24][25][26]. The Snail homolog CES-1 in C. elegans induces the expression of genes containing bHLH transcription factor binding elements and E-box elements [27]. Notably, new models suggest that transcriptional activation by Snail is dependent on novel transcriptional motifs or elements rather than the canonical E-box sequence. A novel Snail-responsive element (5′-TCACA-3′) was identified in the promoters of ZEB1, MMP9, and p15 INK4b genes activated by Snail in collaboration with EGR-1 and SP-1 [28,29]. p65NF-kB, PARP1, and Snail form multiple protein complexes at the Fibronectin 1 (FN) promoter to induce FN transcription [30]. In Drosophila Snail is also shown to induce the expression of genes critical for mesoderm development by collaborating with Twist [31]. Nevertheless, the exact mechanism by which Snail induces gene expression still remains elusive.
The p66α (GATAD2A) and p66β (GATAD2B) proteins are two members of the p66 family that contain a GATA zinc finger in conservative region two (CR2) in the C-terminal [32]. Classical vertebrate GATA transcription factors have two zinc fingers and comprise the general configuration of Cys-X2-Cys-X17-Cys-X2-Cys. While the zinc finger in the C-terminal specifically recognizes the consensus sequence A/T(GATA)A/G in the target promoters, the N-terminal zinc finger predominantly interacts with partner proteins such as the FOG family proteins to stabilize the DNAprotein complex [33,34]. However, the GATA zinc finger in p66α/ p66β has not been reported to be capable of binding to DNA or protein.
The p66 family was first identified as a component of the highly conserved, ATP-dependent, and nucleosome remodeling and deacetylase (NuRD) complex that dominates a transcriptional repression process on methylated CpG islands depending on its nucleosome remodeling and deacetylation activity [35]. While the X. Zou et al. CR1 domain of p66 mediates the interaction between p66 and other NuRD components, the CR2 domain of p66 is essential for targeting the NuRD complex to specific nuclear loci and for mediating histone tail interaction [32,36]. Thus, p66β proteins are usually reported to exert transcriptional repression function as a component of the NuRD complex to participate in various functions, such as regulating progesterone receptor and neurodevelopmental disorders [37,38]. However, Grzeskowiak et al. reported that p66β induces Myc expression in KRAS-mutant lung cancer cells, indicating that p66β could function as a transcriptional activator [39].
In this study, we report that Snail interacts with the CR2 domain of p66β via its zinc finger. p66β and Snail transactive genes that are activated but not repressed by Snail contain a conserved G-rich cis-element (5′-GGGAGG-3′) in the proximal target promoter regions. Snail cooperates with p66β to transactivate gene transcription by binding to G-rich cis-elements within its target promoters.

Snail interacts with p66β
To identify molecules that can cooperate with Snail to promote metastasis, we established HEK-293T cells stably expressing Flag-Snail and performed affinity purification assays. The co-eluents were resolved by SDS-PAGE followed by silver staining, and the specific band at 66 kDa was identified as p66β (Fig. 1A). To confirm the interaction between Snail and p66β, we co-expressed HA-Snail and Flag-p66β in HEK-293T cells and performed reciprocal co-IP assays. Indeed, Snail and p66β interacted with each other (Fig. 1B, C). Moreover, stably expressed Flag-Snail co-immunoprecipitated with the endogenous p66β protein (Fig. 1D). Reciprocally, p66β immunoprecipitated with Snail endogenously (Fig. 1E). To determine the subcellular localization of Snail and p66β, indirect immunofluorescence assays were performed in MDA-MB-231 cells using antibodies specifically against Snail or p66β (Fig. 1F). Notably, both Snail and p66β proteins co-localized in the nucleus.
Snail binds the CR2 domain of p66β through its zinc finger region To determine the regions in Snail responsible for p66β binding, full-length of GST-Snail, truncated zinc-finger structure (GST-Snail-CT), and SNAG domain (GST-Snail-NT) proteins were incubated with HA-p66β protein prepared from HEK-293T cells. The coprecipitation experiments were performed by using GST beads, and the co-eluted HA-p66β protein was examined by Western blotting assays using an anti-HA antibody. GST-Snail-CT and the full-length of GST-Snail showed similar binding affinities to p66β, whereas GST-Snail-NT showed no binding (Fig. 1G, H).
To identify regions in p66β that are critical for the interaction between p66β and Snail, p66β full-length or truncations were transiently co-expressed with HA-Snail in HEK-293T cells. Co-IP assays demonstrated that the CR2 domain displayed strong binding activity toward Snail, which was comparable to that of full-length p66β, whereas the CR1 domain showed weak binding activity (Fig. 1I, J). Taken together, these data indicated that p66β is a novel Snail-interacting protein and that its binding is mediated by the zinc finger region of Snail and the CR2 region of p66β.
p66β promotes breast cancer cell migration and metastasis To examine the potential function of p66β in breast cancer, we analyzed the expression of p66β, Snail, and EMT markers in breast cancer cells, and p66β was highly expressed in triple-negative cells (Fig. S1A). To determine the role of p66β in breast cancer cell migration, p66β was stably expressed in MCF-10A cells, and the resulting cells were subjected to transwell assays ( Fig. 2A and Fig. S1C, D). More importantly, depletion of p66β in luciferase-labeled MDA-MB-231 cells greatly decreased lung colonization, as indicated by the lower luciferase intensity and fewer and smaller micro-metastatic nodules in BALB/c female nude mice, in which the resulting cells were injected via the tail vein ( Fig. 2J-N). Collectively, these data demonstrated that p66β promotes breast cancer cell migration and metastasis.
p66β is required for Snail-mediated cell migration To determine whether p66β is critical for Snail-mediated cell migration and metastasis, we depleted p66β in MDA-MB-231-Snail or -Vector cells, respectively (Fig. 3A). The transwell assays showed that depletion of p66β markedly dampened Snail-mediated cell migration (Fig. 3B, C). Notably, knockdown of Snail also decreased p66β-induced cell migration ability (Fig. S2). Moreover, we analyzed the Kaplan-Meier plotter database and found that high expression of Snail and p66β was predictive of poor prognosis in patients with triple-negative (ER-/PR-/HER2-) breast tumors, respectively (Fig. 3D, E).

Snail and p66β cooperatively induce target gene transcription
To determine whether p66β participates in Snail-mediated gene transcription, we first examined the best-known Snail target genes such as CDH1, FN, as well as PAPSS2, and COL6A2 which were induced by Snail in our previous RNA-seq analysis [40]. Neither ectopic expression nor depletion of p66β affected the mRNA level of CDH1. However, the expression of p66β apparently increased the mRNA levels of the PAPSS2, COL6A2, and FN genes (Fig. 4A, B), suggesting that p66β may participate in Snail-mediated gene induction rather than repression.
Next, the CDH1-Luc, PAPSS2-Luc, or COL6A2-Luc reporters, together with plasmids encoding Flag-p66β, were transiently transfected into HEK-293T cells, and the luciferase activity was Fig. 1 Snail binds the CR2 domain of p66β through its zinc finger motifs. A Colloidal staining showed that p66β was a potential Snailinteracting protein purified from HEK-293 cells. B, C Plasmids encoding Flag-p66β and HA-Snail were transiently transfected into HEK-293T cells, the cell lysates were incubated with either anti-Flag M2 beads or HA antibody and the co-eluted proteins were blotted with anti-HA (B) or anti-Flag (C) antibodies. α, anti; WB, Western blotting. D Snail interacted with p66β at the endogenous level. Lysates prepared from MDA-MB-231-Flag-Snail stable cells were incubated with anti-Flag M2 beads, and the co-eluted proteins were detected with an anti-p66β antibody. E Lysates prepared from MDA-MB-231 cells were incubated with an anti-p66β antibody and protein A/G beads, and the co-eluted proteins were detected with a Snail antibody. F Co-localization of Snail and p66β in MDA-MB-231 cells. Endogenous Snail and p66β in MDA-MB-231 cells were detected for immunofluorescence staining using anti-Snail and anti-p66β antibodies, and the images were taken by confocal microscopy. Scale bar: 90 μm. G Schematic diagrams showed full-length and truncated Snail proteins. H The C-terminal zinc fingers of the Snail interacted with p66β. Bacterially expressed GST-Snail and its truncation proteins, together with HA-p66β proteins prepared from HEK-293T cells were incubated with GST beads, and co-eluted proteins were analyzed by anti-HA. *non-specific band. I Schematic diagrams showed full-length and truncated forms of p66β. The CR2 domain contains a GATA-type zinc finger motif. J The CR2 domain of p66β interacted with Snail. Lysates prepared from HEK-293T cells transfected with plasmids encoding Flag-p66β and its truncation mutants, together with HA-Snail were incubated with anti-Flag M2 beads, and co-eluted proteins were analyzed by anti-HA. . Images represent one microscopic field in each group (B). Scale bar: 200 μm. Quantification of migrated cells is shown (C) (n = 5 fields were randomly chosen and counted for statistical analysis). ***p < 0.001. D-F Depletion of p66β inhibited the migration of SUM-159 cells. Scale bar: 100 μm (n = 9 fields were randomly chosen and counted for statistical analysis). ***p < 0.001. G-I Depletion of p66β inhibited the migration of MDA-MB-231 cells. Scale bar: 100 μm (n = 9 fields were randomly chosen and counted for statistical analysis). ***p < 0.001. J, K MDA-MB-231-Luc cells stably expressing p66β shRNA or shVector were injected into the tail veins of female nude mice. The development of lung metastases was recorded on the 2nd, 4th, and 6th week using luciferase bioluminescence imaging and quantified by measuring the photon flux compared with the mock group (n = 9 and n = 10 mice, respectively). **p < 0.01. L, M Depletion of p66β decreased the number of metastatic lung nodules. **p < 0.01. N Lung metastatic nodules were detected in paraffin-embedded sections stained with hematoxylin and eosin. Images were taken at 10x magnification with a scale bar of 200 μm; images were taken at 40× magnification with a scale bar of 50 μm). All data are shown as the mean ± S.D. by Student's t-test.
normalized to β-galactosidase activity. Consistently, p66β showed no effect on the CDH1-Luc reporter activity (Fig. 4C), but stimulated PAPSS2-Luc and COL6A2-Luc reporter activities in a dosage-dependent manner (Fig. 4D, E). Moreover, Snail alone inhibited CDH1 promoter activity, and adding p66β did not apparently affect the inhibitory effect of Snail (Fig. 4F). Conversely, Snail or p66β alone weakly or modestly induced PAPSS2-Luc and COL6A2-Luc activity, whereas co-expression of Snail and p66β markedly induced PAPSS2-Luc and COL6A2-Luc activity (Fig. 4G, H), indicating that p66β cooperates with Snail to induce target gene expression.
To explain the paradox that the NuRD transcriptional repression complex component p66β is involved in Snail-mediated transcriptional activation, Flag-p66β, HA-MTA2, and/or HA-Snail were coexpressed in HEK-293T cells and co-IP assays were performed. The co-eluted bands showed that MTA2 and Snail could be pulldowned by Flag-p66β, respectively. The interaction between p66β and MTA2 weakened with the increased expression of HA-Snail (Fig. S3), indicating that Snail competes with the MTA2 protein for p66β.

Snail-induced genes contain G-box elements in their proximal promoter regions
By analyzing the promoter sequences of Snail-induced genes, a conserved G-rich cis-element (5′-GGGAGG-3′, designated as G-box herein) was identified in their proximal promoter regions (Fig. 5A). To strengthen this observation, we analyzed the published ChIP-seq data of Snail binding sites in tumor cells prepared from MMTV-PyMT mice using RSAT software and identified 137 G-box sites where Snail displayed peak binding activities ( Fig. 5B and Fig. S4A, B) [8]. Among the 137 G-box sites, we selected 14 genes conserved in humans and mice and then examined their mRNA levels in human and mouse mammary cells that stably expressed Snail, respectively (Table 1). Indeed, Snail induced the expression of the selected representative genes ( Fig. 5C and Fig. S4C). Conversely, depletion of Snail by using specific shRNAs decreased the mRNA levels of these genes (Fig. 5D). Notably, an antibody specific to Snail readily enriched the DNA fragments flanking the G-box elements in the representative promoters of JDP2, EMP3, COL6A2, and PAPSS2 examined by chromatin immunoprecipitation assays (Fig. 5E). Taken together, these data indicated that Snail can induce the transcription of genes containing G-box elements in their promoters and that the G-box element is a potential Snailresponsive element.

Snail induces transcriptional activity of promoter reporters driven by G-box element
To determine whether Snail directly modulates the activity of the G-box element, we first mutated the G-box of PAPSS2-Luc. The two reporters, together with plasmids encoding HA-Snail, were transiently transfected into HEK-293T cells and showed four-fold Snail-induced PAPSS2-Luc reporter activity but had no effect on the mutant reporter ( Fig. 6A left). Similarly, Snail induced the activity of the COL6A2-Luc reporter containing a G-box element, but not a mutant reporter (Fig. 6A right). Moreover, deletion of the SNAG domain of Snail did not obviously affect Snail-mediated induction of promoters with G-boxes (Fig. S5). These data demonstrated that Snail induces the transcriptional activity of promoters containing G-box elements. C Statistical analysis of the migrating cells from the transwell assays was shown in the bar graphs (n = 9 fields were randomly chosen and counted for statistical analysis). The data are shown as the mean ± S.D. by Student's t-test. **p < 0.01. D, E Kaplan-Meier plots of the relapsefree survival of patients with triple-negative (ER-/PR-/HER2-) breast cancer (TNBC) in the whole data set stratified by Snail and p66β expression. Data were acquired from the Kaplan-Meier plotter database. The p-value for Kaplan-Meier curve was determined using a log-rank test, Statistical significance was set at p < 0.05.
Next, we performed electrophoretic mobility shift assays (EMSA) to verify the direct binding between Snail and G-box elements. An up-shifted band was observed after incubation of GST-Snail protein prepared from bacterial cells with biotin-labeled G-box oligonucleotides of COL6A2 (Fig. 6B, lane 2). The shifted band disappeared with the addition of unlabeled wild-type oligonucleotides but not with unlabeled mutant oligonucleotides (Fig. 6B,   lane 3, 4). These data indicated that the Snail protein specifically binds to G-box elements in vitro.

Snail directly binds G-box elements via its zinc finger motifs
To determine the regions of the Snail protein that can directly bind to the G-box element, full-length GST-Snail, and its truncated mutants were incubated with biotin-labeled G-box Fig. 4 Snail and p66β cooperatively transactivate genes. A Stably expressed p66β didn't induce the mRNA level of CDH1 but markedly increased the mRNA level of PAPSS2, COL6A2, and FN. The relative mRNA expression of CDH1, PAPSS2, COL6A2, and FN was detected by qRT-PCR and normalized by the expression of β-actin. *p < 0.05, **p < 0.01, ***p < 0.001 and n.s.: no significance. B Depletion of p66β didn't affect the mRNA level of CDH1 but resulted in decreased mRNA levels of PAPSS2, COL6A2, and FN. **p < 0.01, ***p < 0.001, n.s.: no significance. C-E p66β didn't induce the CDH1-Luc (C) promoter activity, but the PAPSS2-Luc (D), and COL6A2 -Luc (E) promoter activity. The reporters were cotransfected with HA-p66β in HEK-293T cells. Reporter assays were performed and the luciferase activity was normalized to β-galactosidase activity. *p < 0.05, **p < 0.01 and n.s.: no significance. F-H p66β and Snail cooperatively induced the PAPSS2 and COL6A2 promoter activity, not the CDH1 promoter activity. **p < 0.01, ***p < 0.001 and n.s.: no significance. All data are shown as the mean ± S.D. from three independent experiments by Student's t-test.
oligonucleotides of COL6A2. The protein/DNA complex was coprecipitated with NeutrAvidin beads and the co-eluted Snail protein was examined by Western blotting. Both full-length Snail and the truncated C-terminal zinc finger mutant (GST-Snail-CT) readily bound to the COL6A2-G-box element, while the SNAG domain mutant (GST-Snail-NT) did not (Fig. 6C, D).
Moreover, we disrupted the zinc finger structure by simultaneous mutations of H176D, C180A, C210D, H230D, H252D, and C259A, and the mutant protein failed to bind to the COL6A2-Gbox element (Fig. 6E, F). Collectively, these data demonstrated that Snail directly binds to the G-box element via its zinc finger motifs.
p66β protein enriches on the promoters flanking G-boxes To determine whether p66β can directly bind to the G-box element, EMSA assays were performed by incubating the biotinylated COL6A2-G-box with cell extracts prepared from HEK-293T cells expressing wild-type Flag-p66β. Remarkably, adding p66β protein resulted in shifted bands containing the biotinylated oligonucleotide COL6A2-G-box elements, which disappeared upon the addition of unlabeled wild-type oligonucleotides, but not the unlabeled mutants (Fig. 7A). ChIP assays using p66β antibodies indicated that p66β enriched endogenous DNA fragments from the promoters of JDP2, EMP3, PAPSS2, and COL6A2 flanking G-box sites (Fig. 7B). Taken together, these data demonstrated that p66β enriches on the promoter flanking G-boxes.
p66β protein is indispensable for Snail to bind the G-box elements To examine the role of p66β in Snail-mediated transactivation, we performed DNA pull-down assays to examine the binding of Snail to the G-box elements in the presence of p66β. Surprisingly, p66β alone bound the G-box probes with a relatively high affinity and enhanced Snail binding to the G-box probes (Fig. 8A).
To further examine the binding activity of Snail and p66β at the endogenous target chromatins, we performed ChIP assays in SUM-159 cells bearing specific shRNAs to deplete p66β or Snail expression (Fig. 8B). Consistently, both Snail and p66β were associated with the promoter chromatin flanking the G-box elements of JDP2, EMP3, PAPSS2, and COL6A2 in SUM-159-shVector cells, whereas depletion of p66β or Snail, respectively, resulted in decreased binding of Snail and p66β to these promoters ( Fig. 8C-F). Collectively, these data demonstrated that p66β is required for Snail to efficiently bind G-box elements.

DISCUSSION
In this study, we demonstrated that Snail cooperates with p66β to transactivate target gene transcription by directly binding G-box elements in their promoters and that the induction of gene expression by Snail does not require the SNAG domain (Fig. 8G). Snail interacts with the CR2 domain of p66β via the zinc fingers. Snail by itself weakly binds to the G-box elements, while p66β can markedly enhance Snail binding affinity to G-box elements, and depletion of p66β results in diminished Snail binding to the endogenous promoters, indicating that p66β is critical for Snailmediated transactivities. More importantly, ectopic expression of p66β promotes breast cancer cell migration and metastasis, whereas depletion of p66β is detrimental to Snail-mediated cell migration and metastasis. Collectively, these data indicated that Snail cooperates with p66β to induce gene expression by directly binding to the G-box elements in the promoters.
Growing evidence has indicated that Snail functions as a transcriptional activator and directly induces gene expression. Earlier studies found that the Snail transcriptional activation recognition motif is still E-box. The E-boxes in the regulatory region of ERCC1 and MMP15 were recognized by Snail [24,25]. As the research unfolded, it was found that Snail recognition sequences were not limited to the E-box, the new binding sequences were discovered. A novel Snail-responsive motif, 5′-TCACA-3′, in the promoters of ZEB1, MMP9, and p15 INK4b , has been identified and activated by Snail in collaboration with EGR-1 and SP-1 [28,29]. In this study, we denote, G-box (5′-GGGAGG-3′), a new element recognized by zinc fingers in the C-terminal of Snail and important for transactivation. Interestingly, a fragment of the Fibronectin1 promoter containing G-box, 5′-GGGGGAGGAGAGG-GAACCCCAGGCGCGAGC-3′, was bound by Snail extraction, not the GST tagged Snail [30]. Whether there are similar consensus binding motifs for Snail to upregulate gene expression is worth investigating in the future.
The role of the SNAG domain in Snail-mediated gene induction is controversial. The SNAG domain of Snail is essential for both the suppression of E-cadherin and the activation of ERCC1 in head and neck squamous cell carcinoma cells [24], while in another report, the transactivation of β-catenin-dependent reporter by Snail does not require the SNAG domain but the zinc fingers [41]. In this study, we showed that deletion of SNAG does not affect the transactivation of target genes.
p66β protein was previously identified as a key component of the NuRD complex, which modifies chromatin structure to maintain gene repression [32,35,36]. Notably, Grzeskowiak et al. reported that p66β induces Myc expression in KRAS-mutant lung cancer cells, indicating that p66β could function as a transcriptional activator [39]. Our data demonstrated that p66β can directly bind G-box elements and induces G-box-containing promoter activity. In contrast, p66β does not bind E-box elements and consequently, does not affect E-box-driven promoter activity. The Snail/p66β complex and Mi-2/NuRD exist mutually exclusively in cells since Snail competes with the MTA2 protein for p66β.
Both Snail and p66β contain zinc finger motifs. The snail contains four C2H2 types of zinc fingers. The CR2 domain of p66β contains a GATA zinc finger protein. It is known that zinc fingers are not only responsible for specific DNA binding but also function as protein-protein interacting modules [42,43]. In this study, we demonstrated that Snail binds to the CR2 domain of p66β, which contains a GATA zinc finger, via its C2H2 zinc fingers. We speculate that Snail and p66β may also dimerize through their zinc finger motifs, resulting in the remodeling of their binding specificity and affinity toward G-box elements to activate gene expression. These studies have uncovered a novel mechanism for Snail-mediated transcriptional activation. The Snail/p66β complex binds to G-box elements to induce target gene expression. Biologically, p66β is critical for Snail to promote cell migration and metastasis.

MATERIALS AND METHODS Plasmids
The pLU-GFP-Flag-Snail, pCMV5-HA-Snail, and pCDNA3.1-Flag-Snail plasmids have been described previously [40]. GST-Snail and its truncated plasmids were subcloned into pGEX-4T-1(28-9545-49, GE Healthcare) by PCR between BamH I (R3136M, NEB) and EcoR I (R3101M, NEB) restriction enzyme sites. The GST-Snail-ΔZn was subcloned from GST-Snail using point mutation as indicated in Fig. 6E. The pCDNA-Flag-p66β plasmid was a kind gift from Dr. Renkawitz. Deletion mutants of Flag-p66β and pCDH-Flag-p66β plasmids were subcloned into the pCMV4-Flag-Vector (E7158, Sigma-Aldrich) and pCDH-CMV-MCS-EF1-Puro-Vector (CD510B-1, System   Fig. 5 Snail induces transcription of genes containing G-box elements in the proximal promoter regions. A The genes induced by Snail contained the 5'-GGGAGG-3' (G-box) motif in their promoters. B Prediction of Snail binding sequences by using RSAT software from a published Snail ChIP-seq data in tumor cells prepared from MMTV-PyMT mice and 137 G-box sites, where Snail displayed peak binding activities, were identified. C The mRNA levels of genes containing G-boxes were increased in Snail stably expressed MCF-10A. The relative mRNA level of genes was detected by qRT-PCR and normalized by the expression of β-actin. **p < 0.01, ***p < 0.001. D The mRNA levels of genes containing G-boxes were reduced in SUM159-shSnail cells. *p < 0.05, **p < 0.01, ***p < 0.001, and n.s.: no significance. E Snail bound the proximal promoter regions flanking the G-box motif. The ChIP assays were performed in SUM-159 and MDA-MB-231 cells with a specific antibody against Snail and the enriched DNA fragments were examined by qRT-PCR and normalized to input. *p < 0.05, **p < 0.01. All data are shown as the mean ± S.D. from three independent experiments by Student's t-test.

Co-immunoprecipitation (Co-IP), Western blotting (WB), immunofluorescence (IF), and antibodies
For Co-IP assays, HEK-293T cells were transfected as indicated: 36 h later, cells were lysed in IP buffer (1% Triton X-100, 150 mM NaCl, and 50 mM Tris/HCl (pH 7.5), 1 mM PMSF, and protease inhibitor cocktail) and incubated with anti-Flag M2 affinity gels for 4 h at 4°C. For reverse IP, HEK-293T cells were lysed and incubated with anti-HA antibody and protein A/G agarose for 12 h at 4°C. Agarose was washed five times with IP assay buffer, boiled in SDS sample buffer, and subjected to Western blot analysis. Western blotting and immunofluorescence (IF) were performed as previously [40].
The following antibodies were used: mouse anti-Flag (F3165, Sigma-Aldrich), rabbit anti-Flag (F7425, Sigma-Aldrich), rabbit anti-GST (2622, Cell Signaling Technology), mouse anti-β-actin (66009-1-Ig, Proteintech), rabbit Fig. 7 p66β enriches on the promoters flanking G-boxes. A EMSA assays were performed by incubating biotinylated COL6A2-G-box probes with Flag-p66β protein prepared from HEK-293T-Flag-p66β cells. Unlabeled wild-type and mutant probes were included as indicated. B p66β protein bound the promoter region flanking the G-box motif. The ChIP assays were performed in MDA-MB-231 and SUM-159 cells with a specificity antibody against p66β and the enriched DNA fragments were examined by qRT-PCR and normalized to input. *p < 0.05, **p < 0.01, and ***p < 0.001. The data are shown as the mean ± S.D. from three independent experiments by Student's t-test. Luciferase reporter assay, quantitative real-time PCR (qRT-PCR), and chromatin immunoprecipitation (ChIP) assay Luciferase reporter assays and quantitative real-time PCR (qRT-PCR) were performed as described [40].
ChIP assays were performed using 5 × 10 6 cells and antibodies against Snail (3879, Cell Signaling Technology) and p66β (ab76925, Abcam). Cells were cross-linked with 1% formaldehyde for 10 min at room temperature, followed by quenching with 125 mM glycine (50046, Sigma-Aldrich). After washing with PBS, cells were scraped and centrifuged. The cell pellets were sonicated in lysis buffer (150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 5 mM EDTA, NP-40 (0.5% vol/vol), Triton X-100 (1.0% vol/vol), 1 mM PMSF, and protease inhibitor cocktail) using a sonicator (Bioruptor Plus, Diagenode) and the fragment length (200-500 bp) was determined by agarose gel electrophoresis. After incubation with beads and antibodies, the ChIP DNA was eluted with elution buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 1.0% SDS) and incubated with RNAase (R6513, Sigma-Aldrich) at 37°C for 1 h and Proteinase K (p2308, Sigma-Aldrich) at 55°C for 2 h. The final ChIP products were purified using an EZNA Cycle-Pure kit (D6492-02, Omega) and analyzed by qRT-PCR using SYBR Green on a Roche system (LightCycler 480II). The comparative cycle threshold method was used to determine the enrichment relative to the input level. The assays were replicated three times. The primer sets used for the qRT-PCR and ChIP assays are listed in Table 2.

Primers for ChIP
Name Forward sequence Reverse sequence reaction was performed in a final volume of 20 μl. GST fusion proteins (12 μg) were mixed with 600 fmol of probes, 10× binding buffer (100 mM Tris, 500 mM KCl, 10 mM DTT; pH 7.5) (2 μl), 50% glycerol (1 μl), 100 mM MgCl 2 (1 μl), 1 mg/ml poly (dI.dC) (1 μl), 1% NP-40 (1 μl)), and the presence or absence of unlabeled probes. Electrophoresis was performed on a 6% non-denaturing polyacrylamide gel in 0.5× TBE buffer at 100 V and 4°C for 2 h, followed by electrophoretic transfer onto a nylon membrane (77016, Pierce). The membrane was then cross-linked with a CL-1000 ultraviolet crosslinker (UVP) and subjected to detection using the Chemiluminescent detection module provided in the kit, following the manufacturer's protocol. The sequences of the biotinylated probes are listed in Table 2. Potential Snail-DNA complex-binding abilities were verified using biotin-DNA pull-down assays. GST-Snail and /or Flag-p66β prepared from HEK-293T-p66β cells were incubated with biotinylated oligos of COL6A2-G-box and NeutrAvidin agarose resins (29202, Thermo Scientific) for 8 h at 4°C. The beads were washed five times with IP assay buffer, boiled in SDS sample buffer, and subjected to Western blotting.
Transwell, metastasis assays, and hematoxylin and eosin (HE) staining Transwell assays were performed to assess the cell migration capability. To evaluate migration, MCF-10A cells were harvested after serum-free starvation for 6 h and then resuspended in the plain DMEM/F12 medium. A total of 5 × 10 4 cells were applied into the upper 8-μm pore transwell filters (3495, Corning). Complete DMEM/F12 medium was added to the bottom chamber as attractants. After incubation for 20 h, cells were fixed with 4% paraformaldehyde and stained with coomassie brilliant blue. The non-migrated cells on the top of the chamber were removed gently with cotton swabs, and the migrated cells at the bottom of the filter were quantified by counting five (scale bar, 200 μm) randomly chosen fields using an inverted phase contrast microscope in each experiment. The Experiments were repeated in triplicates. The procedure of migration assays in SUM-159 and MDA-MB-231 cells was essentially the same as that in MCF-10A cells, but with 3 × 10 4 cells and complete DMEM media as attractants instead. The migrated cells at the bottom of the filter were quantified by counting nine randomly chosen fields (scale bar, 100 μm) using an inverted phase contrast microscope in each experiment. All animal experiments were performed under relevant guidelines and regulations and were approved by the Institutional Animal Care and Use Committee of Shanghai (IACUC: GBT 35892-2018).
Eight-week-old of female BALB/c nude mice were ordered from SLAC Laboratory Co. Ltd (Shanghai, China) and were assigned to experimental groups using simple randomization. 5 × 10 5 MDA-MB-231 cells (shVector and shp66β) labeled with luciferase were resuspended in 100 μl PBS buffer and were injected into the tail vein of mice (n = 10 for each group). Metastatic tumors were examined by an experimenter blinded to injection condition and experimental cohort using the Xenogen IVIS Imaging System biweekly after the mice were deeply anesthetized by ether. Mice were also deeply anesthetized by ether before decapitation, followed by lung enucleation. Of these, one mouse injected with shVector cells died because of overdue stimulation, resulting in 19 mice for analysis.
For hematoxylin and eosin (HE) staining, murine lung tissues were excised, washed in PBS, and fixed in formalin for 48 h. After fixation, the samples were dehydrated in a graded ethanol series, followed by xylene. Samples were then embedded in paraffin, sectioned onto slides (5 mm thick), and allowed to dry. The slides were deparaffinized using a standard procedure (xylene × 2 washes, 100% ethanol × 2 washes, 95% ethanol, 70% ethanol, and 50% ethanol) and stained with hematoxylin and eosin (HE) (HT110132, Sigma-Aldrich).

Statistical analysis
Data were shown as mean ± SD and were analyzed by the independent Student's t-test from three independent experiments. The postoperative survival of patients with triple-negative (ER-/PR-/HER2-) breast tumors was analyzed by the Kaplan-Meier estimator (https://kmplot.com/analysis/) and tested by the log-rank. Statistical significance was set at p < 0.05.