Transcription factor AP-2α activates RNA polymerase III–directed transcription and tumor cell proliferation by controlling expression of c-MYC and p53

Deregulation of transcription factor AP2 alpha (TFAP2A) and RNA polymerase III (Pol III) products is associated with tumorigenesis. However, the mechanism underlying this event is not fully understood and the connection between TFAP2A and Pol III–directed transcription has not been investigated. Here, we report that TFAP2A functions as a positive factor in the regulation of Pol III–directed transcription and cell proliferation. We found TFAP2A is also required for the activation of Pol III transcription induced by the silencing of filamin A, a well-known cytoskeletal protein and an inhibitor in Pol III–dependent transcription identified previously. Using a chromatin immunoprecipitation technique, we showed TFAP2A positively modulates the assembly of Pol III transcription machinery factors at Pol III–transcribed gene loci. We found TFAP2A can activate the expression of Pol III transcription-related factors, including BRF1, GTF3C2, and c-MYC. Furthermore, we demonstrate TFAP2A enhances expression of MDM2, a negative regulator of tumor suppressor p53, and also inhibits p53 expression. Finally, we found MDM2 overexpression can rescue the inhibition of Pol III–directed transcription and cell proliferation caused by TFAP2A silencing. In summary, we identified that TFAP2A can activate Pol III–directed transcription by controlling multiple pathways, including general transcription factors, c-MYC and MDM2/p53. The findings from this study provide novel insights into the regulatory mechanisms of Pol III–dependent transcription and cancer cell proliferation.

Deregulation of transcription factor AP2 alpha (TFAP2A) and RNA polymerase III (Pol III) products is associated with tumorigenesis. However, the mechanism underlying this event is not fully understood and the connection between TFAP2A and Pol III-directed transcription has not been investigated.
Here, we report that TFAP2A functions as a positive factor in the regulation of Pol III-directed transcription and cell proliferation. We found TFAP2A is also required for the activation of Pol III transcription induced by the silencing of filamin A, a well-known cytoskeletal protein and an inhibitor in Pol IIIdependent transcription identified previously. Using a chromatin immunoprecipitation technique, we showed TFAP2A positively modulates the assembly of Pol III transcription machinery factors at Pol III-transcribed gene loci. We found TFAP2A can activate the expression of Pol III transcriptionrelated factors, including BRF1, GTF3C2, and c-MYC. Furthermore, we demonstrate TFAP2A enhances expression of MDM2, a negative regulator of tumor suppressor p53, and also inhibits p53 expression. Finally, we found MDM2 overexpression can rescue the inhibition of Pol III-directed transcription and cell proliferation caused by TFAP2A silencing. In summary, we identified that TFAP2A can activate Pol IIIdirected transcription by controlling multiple pathways, including general transcription factors, c-MYC and MDM2/ p53. The findings from this study provide novel insights into the regulatory mechanisms of Pol III-dependent transcription and cancer cell proliferation.
RNA polymerase III (Pol III) products are a subset of abundant noncoding RNA molecules, including 5S rRNA, tRNA, U6 snRNA, and 7SL RNA. These RNAs, accounting for about 15% of transcription activity in eukaryotes, play essential roles in several fundamental cellular activities that include ribosomal assembly, translation, protein transportation, RNA processing, and cell growth (33)(34)(35)(36). Pol III-directed transcription initiation can be classified into three types of initiation modes and is tightly controlled by general transcription factors such as TFIIIA, TFIIIB, TFIIIC, and proximal sequence ‡ These authors contributed equally to this work. * For correspondence: Wensheng Deng, dengwensheng@wust.edu.cn; Huan Deng, denghuan@wust.edu.cn; Hongwei Ni, nihongwei@wust. edu.cn. element-binding factor SNAPc (33,34). Other factors, including MAF1, oncogenic factor c-MYC, tumor suppressors (p53 and retinoblastoma protein), and signaling factors (ERK, JNK, and PTEN), also modulate Pol III-mediated transcription by directly or indirectly interacting with general transcription factors (33,(37)(38)(39)(40)(41). It has been shown that the synthesis of Pol III products can be regulated by chromatin modification (35,42,43). For example, nc886 is a typical gene transcribed by Pol III and its expression is inhibited by CpG methylation (44). Deregulation of Pol III transcription-related factors or Pol III products is associated with a range of diseases such as neural degeneration diseases and cancers (36,(45)(46)(47). Our previous study revealed that cytoskeletal Filamin A (FLNA) can inhibit Pol III-dependent transcription in transformed cell lines (48). However, the factors that participate in this process remain to be identified. In this study, we showed that TFAP2A not only activates Pol III-directed transcription but also is required for the inhibition of Pol III-dependent transcription mediated by FLNA. We then elucidated the regulatory mechanisms of Pol III transcription mediated by TFAP2A.

FLNA inhibits TFAP2A expression
In our previous work, we found that cytoskeletal FLNA suppresses Pol III-dependent transcription in human transformed cell lines (48). To understand the regulatory mechanisms underlying this event, we recently performed RNA-Seq analysis using SaOS2 cell lines stably expressing FLNA shRNA or control shRNA (SRA accession number: SRP318361, https://www.ncbi.nlm.nih.gov/Traces/study/? acc=PRJNA726417). Based on the RNA-Seq data, we analyzed differential expression genes between two groups of samples and found that TFAP2A mRNA reads were drastically enhanced by FLNA depletion (Fig. 1, A and B). In addition, we found that FLNA knockdown in SaOS2 cells severely affected cell shapes of SaOS2 but did not affect the shapes of SaOS2 nuclei and nucleoli (Fig. S1), suggesting that cytoskeletal FLNA plays a key role in the maintenance of cell shape. Next, we verified TFAP2A expression by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and Western blot using SaOS2 cell lines stably expressing FLNA shRNA and control shRNA. Indeed, TFAP2A expression was enhanced by FLNA depletion in either mRNA or protein levels ( Fig. 1, C and D). To determine whether this result can be reproduced in other cell types, we generated HeLa cell lines stably expressing FLNA shRNA and control shRNA using FLNA-expressing lentiviral particles, and TFAP2A expression was assessed by RT-qPCR and Western blot using the resulting cell lines. As expected, FLNA silencing enhanced TFAP2A expression at both mRNA and protein levels (Fig. 1, E and F). This result is in agreement with that obtained from SaOS2 cells, suggesting that FLNA can inhibit TFAP2A expression. To confirm the inhibitory role of FLNA in TFAP2A expression, we generated a HeLa cell line with FLNA overexpression and its control cell line using a lentiviral transduction system. TFAP2A expression was analyzed by RT-qPCR and Western blot. We showed that FLNA overexpression reduced the expression of TFAP2A mRNA and protein in HeLa cells (Fig. 1, G and H). Taken together, these results indicate that FLNA may repress TFAP2A expression in SaOS2 and HeLa cells.

TFAP2A activates Pol III-directed transcription
Given that FLNA inhibits both Pol III-directed transcription and TFAP2A expression, we asked whether there is a connection between TFAP2A and Pol III-dependent transcription. Before answering this question, we initially investigated TFAP2A expression in several cell types, including SaOS2, 293T, and HeLa cell lines, by immunoblotting. Unexpectedly, TFAP2A expression level in SaOS2 cell line was very low and hard to detect by Western blot, but expression of TFAP2A in 293T and HeLa was quite abundant (Fig. 1, D and I). Thus, we next focused on the role of TFAP2A in Pol IIIdirected transcription in HeLa and 293T cell lines. The effect of TFPA2A depletion on the synthesis of Pol III products was first investigated by transfecting TFAP2A siRNA into 293T cells. RT-qPCR and Western blot analysis confirmed TFPA2A downregulation after transient transfection of TFAP2A siRNA (Fig. 2, A and B). Interestingly, TFAP2A depletion significantly inhibited the synthesis of Pol III products, including 5S rRNA, U6 RNA, 7SL RNA, and tRNA-Met (Fig. 2C). Meanwhile, we tested the effect of TFAP2A siRNA transfection on the expression of TFAP2A and Pol III products in HeLa cells; and the results are consistent with those obtained in the assays with 293T cells (Fig. 2, D-F), indicating that TFAP2A is required for maintaining the normal synthesis of Pol III products. To confirm this result, we generated 293T cell lines that stably express TFAP2A shRNA or control shRNA. As expected, TFAP2A shRNA stable expression dampened TFAP2A expression and the synthesis of Pol III products (Fig. 2, G-I). Consistent results were also achieved when HeLa cells with TFAP2A silencing were analyzed (Fig. 2, J-L). These results suggest that TFAP2A plays a positive role in Pol III-directed transcription. To validate the positive role of TFAP2A in Pol III-dependent transcription, we assessed the effect of TFAP2A overexpression on Pol III products by performing transient transfection with 239T cells. Notably, TFAP2A overexpression stimulated the expression of Pol III products in 293T cells (Fig. S2, A and B). Comparable results were obtained when HeLa cells transiently transfected were analyzed (Fig. S2, C and D). Next, 293T and HeLa cell lines with HA-TFAP2A stable expression were generated and similar experiments were performed. As expected, stable expression of HA-TFAP2A enhanced the expression of Pol III products in both cell types (Fig. S2, E-H). Altogether, we showed that TFAP2A can positively regulate Pol III-directed transcription.
Alteration of Pol III products induced by TFAP2A contributes to cell proliferative activity It is well established that Pol III products are essential to ribosomal assembly, protein synthesis, and cell growth. Thus, TFAP2A and Pol III-directed transcription we determined whether alteration of Pol III products caused by TFAP2A upregulation or downregulation affects cell growth. We first examined the effect of TFAP2A alteration on the proliferative activity by performing cell counting and MTT assays. Noticeably, TFAP2A knockdown reduced the proliferative activity of 293T and HeLa cells (Fig. 3, A-D). In contrast, TFAP2A overexpression enhanced them in these cell types (Fig. S3, A, B and E). Since EdU compound (5-ethynyl-2 0 -deoxyuridine) is widely applied to proliferation assays, we next investigated the effect of TFAP2A alteration on cell growth by performing EdU assays. EdU assays revealed that TFAP2A downregulation reduced the rate of EdU-positive cells (Fig. 3, E-H), whereas TFAP2A upregulation enhanced the rate of EdU-positive cells (Fig. S3, C, D, F and G). These data further confirmed that TFAP2A can promote the proliferative activity of 293T and HeLa cells. To understand whether the  A and B, effect of FLNA silencing on TFAP2A mRNA expression was analyzed using the mRNA-Seq data obtained from SaOS2 cell lines expressing control shRNA or FLNA shRNA. TFAP2A mRNA reads from three biological replicates and the log 2 fold change are presented in A and B, respectively. Log 2 fold change was obtained by comparing the mean of TFAP2A mRNA reads from FLNA shRNA samples with that from Ctrl shRNA samples. C and D, effect of FLNA knockdown on TFAP2A expression in SaOS2 cells; RT-qPCR (C) and Western blot (D) were performed using SaOS2 cells stably expressing control shRNA or FLNA shRNA. E and F, effect of FLNA knockdown on TFAP2A expression in HeLa cells. RT-qPCR (E) and Western blot (F) were performed using HeLa cell lines stably expressing control shRNA or FLNA shRNA. G and H, effect of FLNA overexpression on TFAP2A expression in HeLa cells. A HeLa cell line stably expressing HA-FLNA and its control cell line (empty vector) were used to extract total RNA. TFAP2A mRNA expression was analyzed by RT-qPCR using the cDNA synthesized from the total RNA (G). TFAP2A protein expression was detected by Western blot using the total cell lysate of these cell lines and the antibodies against the factors as indicated (H). I, expression of TFAP2A in SaOS2, 293T, and HeLa cells was analyzed by Western blot. Each column in histograms represents the mean ± SD of three biological replicates. *p < 0.05; **p < 0.01. p values were obtained by Student's t test. RT-qPCR, reverse transcription-quantitative polymerase chain reaction.   (D) and Western blot (E). Pol III products were detected by RT-qPCR (F). G-I, TFAP2A shRNA stable expression reduced the expression of Pol III products in 293T cells. 293T cell lines stably expressing control shRNA or FLNA shRNA were used to analyze TFAP2A expression by RT-qPCR (G) and Western blot (H), where Pol III products were detected by RT-qPCR (I). J-L, TFAP2A shRNA stable expression reduced the expression of Pol III products in HeLa cells. HeLa cell lines stably expressing control shRNA or FLNA shRNA were used to analyze TFAP2A expression by RT-qPCR (J) and Western blot (K), where Pol III products were detected by RT-qPCR (L). Each column in histograms represents the mean ± SD of three biological replicates. *p < 0.05; **p < 0.01. p Values were obtained by Student's t test. RT-qPCR, reverse transcription-quantitative polymerase chain reaction.

t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t t
U r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r rg g g g g g g g g TFAP2A and Pol III-directed transcription enhancement of cell proliferation is caused by the activation of Pol III-dependent transcription, we analyzed proliferative activity and Pol III-dependent transcription in the 293T cell line expressing HA-TFAP2A and its control cell line in the presence or absence of 54 μM ML-60218 (a Pol III transcription inhibitor). As shown in Figure 3, I and J, TFAP2A overexpression enhanced 293T cell proliferation and expression of Pol III products, which is consistent with the result as described above (Figs. S2B and S3, A and B). Strikingly, the presence of ML60218 reversed the activation of proliferative activity and Pol III-directed transcription induced by TFAP2A overexpression (Fig. 3, I and J). Western blot showed that TFAP2A expression was not significantly affected by the treatment with ML-60218 (Fig. S4). These results suggest that the increase of Pol III products induced by TFAP2A overexpression contributes to the enhancement of proliferation activity, although the contribution of other pathways to cell proliferation cannot be excluded.
TFAP2A regulates the assembly of the Pol III transcriptional machinery at Pol III target loci and the expression of multiple Pol III transcription-related factors To understand how TFAP2A regulates Pol III-directed transcription, we initially investigated whether TFAP2A binds to the promoters of Pol III target loci by performing chromatin immunoprecipitation (ChIP) assays. The results showed that TFAP2A did not bind to the Pol III target loci except the 5S rRNA gene locus (Fig. S5A). Next, we determined whether TFAP2A silencing affects the assembly of Pol III transcription machinery factors at Pol III target loci by ChIP assays. ChIP analysis showed that TFAP2A silencing reduced the occupancies of components of the Pol III transcription machinery at Pol III target loci, including 5S rRNA, 7SL RNA, and tRNA-Met loci (Fig. 4, A-D). BRF1 and GTF3C2 showed very low occupancy on the U6 snRNA gene locus. This result is rational because U6 snRNA transcription does not require the involvement of BRF1 and GTF3C2. However, the occupancy of POLR3A showed significant reduction at this locus. These data suggest that TFAP2A can regulate Pol IIIdependent transcription by controlling the recruitment of components of the Pol III transcription machinery to Pol IIItranscribed gene loci. To understand how TFAP2A expression alteration affects the assembly of the Pol III transcription machinery at the Pol III-transcribed loci, we analyzed the expression of Pol III transcription-related factors using Western blot and cell lines with TFAP2A deletion or overexpression. It is well established that Pol III general transcription factors play essential roles in Pol III transcription initiation (33). Thus, the expression of Pol III general factor subunits was first analyzed by Western blot using 293T cells with TFAP2A depletion or overexpression. Interestingly, TFAP2A silencing inhibited the expression of BRF1 (one subunit of TFIIIB) and GTF3C2 (one subunit of TFIIIC), but it did not affect TBP expression (Fig. 4E). Comparable results were obtained when HeLa cells were analyzed (Fig. 4F). In contrast, TFAP2A overexpression activated BRF1 and GTF3C2 expression but did not impact on TBP expression in these cell types (Fig. 4, G and H). These results suggest that TFAP2A can regulate the assembly of the Pol III transcription machinery at Pol III target loci by affecting BRF1 and GTF3C2 expression.
Pol III-dependent transcription is also controlled by oncogenic factors, tumor suppressors, and signaling transduction factors (33,34). Next, we investigated whether alteration of TFAP2A expression affects the expression of these factors in 293T and HeLa cells by Western blot. Interestingly, TFAP2A depletion reduced the expression of oncogenic MDM2 and c-MYC; meanwhile, it increased p53 expression in these cell types (Figs. 4, I and J and S6, A and B). Conversely, TFAP2A overexpression enhanced MDM2 and c-MYC expression in which p53 expression was reduced in the same cell types (Figs. 4, K and L and S6, C and D). Rho A is an upstream factor of several signal pathways that are required for Pol III-directed transcription (33). In these assays, we found that RhoA expression was not affected by TFAP2A silencing or overexpression (Figs. 4, I and J and S6, A and C). PTEN, a factor in the AKT/PTEN pathway, has been shown to regulate Pol III-directed transcription (41). In these experiments, however, PTEN expression remained unchanged when TFAP2A was upregulated or downregulated in 293T and HeLa cell lines (Figs. 4, I and J and S6, A and C). Taken together, these findings suggest that TFAP2A indirectly regulates Pol III-dependent transcription by affecting the expression of multiple factors, including transcription factor subunits BRF1 and GT3C2, oncogenic c-MYC and MDM2, and tumor suppressor p53.

TFAP2A regulates Brf1 and Gtf3c2 gene transcription by binding to their promoters
We showed that TFAP2A expression alteration can affect the expression of BRF1 and GTF3C2 at the protein level. However, how TFAP2A regulates BRF1 and GTF3C2 expression remains to be clarified. To address this question, we first analyzed BRF1 and GTF3C2 mRNA expression in 293T and HeLa cell lines with TFAP2A depletion or overexpression by RT-qPCR. Apparently, TFAP2A knockdown inhibited BRF1 and GTF3C2 mRNA expression, whereas TFAP2A overexpression activated their expression, suggesting that TFAP2A could regulate the expression of Brf1 and Gtf3c2 genes at the transcriptional step (Fig. 5, A-D). To determine whether TFAP2A can modulate the transcription of Brf1 and Gtf3c2 genes, we searched TFAP2A-binding sites from the Brf1 and Gtf3c2 gene promoters. Surprisingly, the Brf1 promoter 4 (Brf1P4) and the Gtf3c2 promoter 2 (Gtf3c2P2) contain a canonical TFAP2A binding motif (TFAP2A consensus sequence: GCC(N)3/4GGC) (Fig. 5E). Next, we examined whether TFAP2A can bind to these two promoters by performing ChIP assays using HeLa cells. ChIP qPCR revealed that TFAP2A displayed specific occupancy at Brf1P4 and Gtf3c2P2 (Fig. 5, F and G). Since TFAP2A binds to Brf1P4 and Gtf3c2P2, we investigated the effect of TFAP2A expression change on the activities of Brf1P4 and Gtf3c2P2 by performing reporter gene   assays. 293T or HeLa cell lines with TFAP2A silencing or overexpression and their corresponding control cell lines were transfected with the reporter gene vectors driven by Brf1P4 or Gtf3c2P2, respectively. Reporter assays showed that TFAP2A silencing reduced the activities of Brf1P4 or Gtf3c2P2 (Fig. 5, H and I), whereas TFAP2A overexpression enhanced the activities of these two promoters (Fig. 5J). In addition, mutations of TFAP2A-binding sites within these promoters inhibited the activities of Brf1P4 or Gtf3c2P2 (Fig. 5, K and L). These results indicate that TFAP2A can directly modulate the transcription of Brf1 or Gtf3c2 genes by binding to their promoters. Since the depletion of BRF1 or GTF3C2 has been shown to inhibit Pol III-directed transcription (49), the results obtained from the previous and present experiments suggest that TFAP2A can indirectly activate Pol III-directed transcription by enhancing Brf1 and Gtf3C2 gene transcription.

TFAP2A indirectly regulates Pol III-directed transcription by controlling the activity of the MDM2-p53 pathway
We have shown that TFAP2A can regulate the expression of MDM2, p53, and c-MYC in the above experiments, and both c-MYC and p53 have been reported to modulate Pol IIIdirected transcription (38,39). To understand how TFAP2A regulates expression of c-MYC and p53, we performed ChIP assays using an anti-TFAP2A antibody and the primers detecting p53 and c-MYC promoters. Unexpectedly, ChIP results showed that TFAP2A did not bind to the promoters of c-MYC and p53 genes (Fig. S5B), suggesting that TFAP2A indirectly regulates the expression of c-MYC and p53. We next focused on the role of MDM2 in Pol III-mediated transcription. First, we examined whether MDM2 overexpression can reverse the inhibition of Pol III-directed transcription caused by TFAP2A downregulation. To this end, both 293T and HeLa cell lines stably expressing TFAP2A shRNA were further infected by lentiviral particles expressing mCherry-MDM2, and the generation of cell lines expressing both TFAP2A shRNA and mCherry-MDM2 were verified by Western blot (Figs. 6A and S7A). We showed that mCherry-TFAP2A expression reversed the inhibition of Pol III-directed transcription caused by TFAP2A silencing in both 293T and HeLa cells (Figs. 6B and S7B). Interestingly, consistent results were obtained when human cervical epithelial cells expressing TFAP2A shRNA and mCherry-MDM2 were generated and used for similar assays (Fig. S8). Since alteration of Pol III products can affect cell proliferation, we next examined the effect of mCherry-MDM2 expression on the proliferative activity of the cell lines with TFAP2A silencing. Interestingly, the mCherry-MDM2 expression also reversed the inhibition of cell proliferation induced by TFAP2A silencing in 293T and HeLa cell lines (Figs. 6C and S7, C-E). How TFAP2A regulates MDM2 expression remains unclear. To address this question, MDM2 mRNA expression in the cell lines with TFAP2A silencing or overexpression was detected by RT-qPCR. We showed that TFAP2A can activate MDM2 mRNA expression (Figs. 6, D and E and S7, F and G), suggesting that TFAP2A regulates MDM2 expression at the transcriptional step. To support this assumption, we searched TFAP2A-binding sites within the Mdm2 gene promoter. Surprisingly, the Mdm2 promoter contains two putative TFAP2A-binding sites upstream of transcription starting site (Fig. 6F). We next examined whether TFAP2A binds to the Mdm2 promoter by performing ChIP assays. ChIP-qPCR results showed that TFAP2A exhibited specific occupancy at the Mdm2 promoter compared with that of control IgG (Fig. 6G), indicating that TFAP2A can bind to the Mdm2 promoter. We next asked whether TFAP2A expression change affects the Mdm2 promoter activity. To answer this question, we prepared the DNA fragments of the wildtype MDM2 promoter and its mutant containing TFAP2A-binding site mutations by PCR and inserted them into the pGL3-basic reporter vector (Fig. 6F). Luciferase assays were performed using the cell lines transiently transfected with the promoter-driven reporter vectors. The data from luciferase activity showed that TFAP2A downregulation inhibited the activity of the Mdm2 promoter activity, whereas TFAP2A overexpression activated its activity (Figs. 6, H and I and S7, H and I). Intriguingly, mutations of TFAP2A-binding sites reduced the Mdm2P1 activity (Figs. 6J and 4J). These data indicate that TFAP2A can regulate Mdm2 gene transcription by binding to the Mdm2 promoter. MDM2 was identified to be an inhibitor of p53 (50); moreover, MDM2 expression was opposite to p53 (Fig. 4, I and J). Thus, we determined the effect of mCherry-MDM2 expression on p53 expression in 293T and HeLa cell lines with TFAP2A silencing. As expected, mCherry-MDM2 expression inhibited the activation of p53 expression induced by TFAP2A silencing (Figs. 6K and S7K). However, mCherry-MDM2 expression in the TFAP2A shRNA-expressing cells did not rescue the expression of BRF1 and GTF3C2. Similar results were observed when mCherry-MDM2 was expressed in human cervical epithelial cells with TFAP2A silencing (Fig. S8, B and  C). Collectively, these results indicate that TFAP2A can indirectly regulate Pol III-directed transcription by controlling the MDM2-p53 axis.

TFAP2A is required for the inhibition of Pol III transcription mediated by FLNA
Previous work showed that FLNA downregulation enhances Pol III-directed transcription (48); furthermore, FLNA DNA (1 ng genomic DNA). E and F, TFAP2A downregulation inhibited BRF1 and GTF3C2 expression. Western blot was performed using 293T (E) and HeLa (F) cell lines stably expressing control shRNA or TFAP2A shRNA and antibodies against the indicated factors. Bottom panels in E and F represent the quantified results of Western blots in the corresponding upper panels. GAPDH c :GAPDH for control group. G and H, TFAP2A overexpression enhanced BRF1 and GTF3C2 expression. Western blot was performed using 293T (G) and HeLa (H) cell lines stably expressing HA-TFAP2A and their control cell lines and antibodies against the indicated factors. Bottom panels in G and H represent the quantified results of Western blots in the corresponding upper panels. GAPDH c :GAPDH for control group. I and J, TFAP2A depletion reduced MDM2 and c-MYC expression but increased p53 expression in 293T cells. Western blot was performed as described in E. J, represents the quantified results of Western blots in I. K and L, TFAP2A overexpression activated MDM2 and c-MYC proteins but inhibited p53 protein expression in 293T cells. Western blot was performed as described in G. L, represents the quantified results of Western blots in K. Each column in histograms represents the mean ± SD of three biological replicates. *p < 0.05; **p < 0.01. p Values were obtained by Student's t test. silencing also activated TFAP2A expression in SaOS2 and HeLa cells (Fig. 1). Whether TFAP2A is required for the inhibition of Pol III transcription induced by FLNA remains unclear. To address this question, we generated SaOS2 and HeLa cell lines that stably express both FLNA shRNA and TFAP2A shRNA using cell lines with FLNA depletion. RT-qPCR and Western blot results showed the achievement of these cell lines (Figs. 7, A and B and S9, A and B). Analysis of Pol III products by RT-qPCR revealed that TFAP2A silencing inhibited the activation of Pol III-directed transcription induced by FLNA depletion (Figs. 7C and S9C), indicating that TFAP2A is required for the activation of Pol III transcription induced by FLNA silencing. These data suggest that FLNA and TFAP2A can play opposite roles in Pol III-dependent transcription. FLNA has been shown to suppress tumor cell proliferation (48). Thus, cell proliferation assays were performed using cell lines established above. The results from cell counting and MTT assays revealed that TFAP2A depletion inhibited the enhancement of cell proliferation induced by FLNA silencing (Figs. 7, D and E and S9, D and E), suggesting that TFAP2A is required for the inhibition of cell proliferation mediated by FLNA. This result is in agreement with that obtained in the analysis of Pol III products (Figs. 7, C and D and S9, C and D). TFAP2A has been shown to regulate Pol IIIdependent transcription by controlling BRF1 and GTF3C2 expression (Fig. 5). Thus, we analyzed BRF1 and GTF3C2 expression by RT-qPCR and Western blot using cell lines obtained above (Fig. 7, A and B). Interestingly, TFAP2A depletion inhibited the activation of BRF1 and GTF3C2 induced by FLNA silencing in SaOS2 and HeLa cells (Figs. 7, E and F and S9, E and F). This result is consistent with that obtained using cell lines expressing TFAP2A shRNA only (Fig. 4, E and F), indicating that TFAP2A modulates FLNAmediated Pol III transcription by controlling BRF1 and GTF3C2 expression. Taken together, TFAP2A is required for the regulation of Pol III-directed transcription mediated by FLNA.

TFAP2A and Pol III-directed transcription
Based on the data obtained in this study, we proposed a model by which TFAP2A regulates Pol III-dependent transcription. Specifically, TFAP2A expression is inhibited by FLNA; however, TFAP2A upregulation activates the expression of BRF1, GTF3C2, c-MYC, and MDM2. On the one hand, the increase of BRF1 and GTF3C2 expression directly enhances Pol III transcription machinery assembly at target loci to activate the synthesis of Pol III products (Fig. 7H-①). On the other hand, the activation of oncogenic c-MYC (Fig. 7H-②) and MDM2 (Fig. 7H-③) increases the interaction between c-MYC and TFIIIB and decreases the interaction between p53 and TFIIIB, which subsequently activates Pol IIIdependent transcription. These regulatory pathways revealed by this study extend the understanding of the mechanism of Pol III-directed transcription.

Discussion
TFAP2A has been identified for decades as a transcription factor that regulates RNA polymerase II-directed gene transcription (1, 2, 5-7). In this study, we found a novel role of TFAP2A in Pol III-directed transcription, and TFAP2A positively regulates the synthesis of Pol III products (Figs. 2 and S1). We showed that TFAP2A is also required for the activation of Pol III transcription induced by FLNA silencing (Figs. 7 and S9). Thus, the findings of this study extend the functions of TFAP2A as a transcription factor in human cells. It has been reported that TFAP2A can play negative or positive roles in cancer cell proliferation (4). Especially, the positive role of TFAP2A in cancer cell proliferation was frequently revealed in recent years (4). However, the mechanism underlying the event remains elusive. It is well established that Pol III product levels correlates closely with the rate of cell growth (33,37), and abnormal high expression of Pol III products was observed in several cancer types (51). We showed that the presence of ML-60218 can reverse the activation of both cell proliferation and Pol III-directed transcription induced by TFAP2A overexpression in 293T cells (Fig. 3, I and J), indicating that Pol III-mediated transcription activation contributed to the promotion of cell proliferation by TFPA2A overexpression. These findings provide a novel explanation from about how TFAP2A promotes cancer cell proliferation.
General transcription factors TFIIIB and TFIIIC tightly control transcription initiation directed by Pol III (33,34). We previously showed that BRF1 and GTF3C2 silencing inhibited Pol III-dependent transcription (49). In this study, we found that TFAP2A can positively regulate the synthesis of Pol III products and the expression of BRF1 and GTF3C2. Furthermore, TFAP2A modulates Brf1 and Gtf3c2 gene transcription by binding to the promoters of Brf1 and Gtf3c2 genes (Figs. 4 and 5). These results suggest that TFAP2A regulates the expression of Pol III products to control BRF1 and GTF3C2 expression (Fig. 7H-①). It has been shown that Specificity represents a TFAP2A consensus sequence: GCC(N)3/4GGC/G). FP and RP represent primers used for ChIP-qPCR. FP, forward primer; RP, reverse primer. F and G, TFAP2A binds to Brf1P4 and Gtf3c2P2. ChIP assays were performed using HeLa cells and the anti-TFAP2A antibody and control IgG. The DNA quantity of Brf1P4 (F) and Gtf3c2P2 (G) was monitored by qPCR. The relative enrichment was obtained as described in Fig. 4, A-D. H and I, TFAP2A depletion by shRNA expression inhibited the activities of Brf1P4 and Gtf3c2P2. Reporter assays were performed by transfecting the Brf1P4-or Gtf3c2P2-driven reporter vectors into 293T (H) and HeLa (I) cell lines expressing control shRNA or TFAP2A shRNA. Relative luciferase activity (Rel luc act) was obtained by comparing the luciferase activity of treated samples (TFAP2A shRNA) with that of control samples (Ctrl shRNA), where the luciferase activity of the control samples was arbitrarily set as 1. J, TFAP2A overexpression activated the activities of Brf1P4 and Gtf3c2P2 in 293T cells. Luciferase assays were performed by transfecting the Brf1P4-or Gtf3c2P2-driven reporter vectors into the 293T cell line expressing HA-TFAP2A and the corresponding control cell line. K, a scheme showing the wildtype DNA fragment of Brf1P4 and Gtf3C2P2 and their mutants containing mutated TFAP2A consensus bases. The bold blue bases within the promoter sequences represent wildtype TFAP2A consensus sequence, whereas the bold red bases represent mutated TFAP2A consensus bases (mut). L, mutations of TFAP2A consensus sequence severely affected the activities of Brf1P4 and Gtf3c2P2 in 293T cells. Luciferase assays were performed by transfecting 293T cells with the reporter vectors driven by the wildtype promoters (Brf1P4 and Gtf3c2P2) or by their mutants. The relative luciferase activity (Rel luc act) was obtained as described in H. BSM, binding site mutations. Each column in histograms represents the mean ± SD of three biological replicates. *p < 0.05; **p < 0.01. p Values were obtained by Student's t test. ChIP, chromatin immunoprecipitation; RT-qPCR, reverse transcriptionquantitative polymerase chain reaction. TFAP2A and Pol III-directed transcription protein 1 (Sp1) regulates Pol III-directed transcription by controlling transcription of Brf1 and Gtf3C2 genes (49). Recently, we found that GATA4 is required for the regulation of Pol III-directed transcription mediated by Sp1 and can directly control Sp1 expression; therefore, FLNA suppresses Pol III-dependent transcription by the FLNA-GATA4-Sp1 pathway (52). Furthermore, both TFAP2A and Sp1 expression can be negatively regulated by FLNA (Fig. 1). Since the promoters of Brf1 and Gtf3C2 genes contain Sp1-and TFAP2Abinding sites (Fig. 5), it is possible that TFAP2A and Sp1 can independently regulate Pol III-directed transcription. Thus, FLNA may suppress Pol III-mediated transcription using several parallel pathways. Oncogenic factor c-MYC and tumor suppressor p53 have been reported to modulate Pol IIIdirected transcription by interacting with TFIIIB (39,40). In this study, we found that TFAP2A can positively regulate c-MYC and MDM2 expression but negatively affect p53 expression (Figs. 4 and 6). MDM2 has been identified to be an upstream effector of p53 (50), suggesting that TFAP2A can also indirectly modulate Pol III-mediated transcription by controlling the MDM2/p53 axis. Based on the previous and current findings, we conclude that TFAP2A can indirectly regulate Pol III-directed transcription by controlling multiple pathways (Fig. 7H). Moreover, TFAP2A participates in the inhibition of Pol III-directed transcription mediated by FLNA, perhaps, using similar mechanisms. These findings put extra weight on the understanding of the regulatory mechanism of Pol III-dependent transcription mediated by FLNA, where FLNA may modulate Pol III-directed transcription by a complicated regulatory network.

HeLa
In summary, we found that TFAP2A functions as a positive regulator in Pol III-directed transcription and promotes the proliferative activity of 293T and HeLa cells. TFAP2A is also required for the regulation of Pol III transcription mediated by FLNA. Mechanistically, TFAP2A modulates Pol III-directed transcription by controlling the activities of multiple pathways, including the expression of general transcription factor subunits BRF1 and GTF3C2, c-MYC, and MDM2/p53. Our findings provide novel insights into the regulatory mechanisms of Pol III-directed transcription and cell proliferation.

Plasmids, cells, and reagents
The lentiviral expression vectors such as pLV-U6-EGFP-Puro, pLV-U6-mCherry-Puro, and pLV-EF1α-EGFP-Puro were purchased from Inovogen Tech Co. FLNA shRNA lentiviral particles (SC-35374-V) and control shRNA lentiviral particles (SC-108080) were obtained from Santa Cruz Biotech. Human cell lines, including SaOS2, HeLa, and 293T, were derived from American Type Culture Collection (ATCC). SaOS2 cells were cultured in McCoy's 5A complete medium (HyClone), while HeLa and 293T cells were cultured in Dulbecco's modified Eagle's medium complete medium (HyClone). DNA and RNA oligonucleotides were purchased from Sangon Biotech. PCR DNA purification and gel extraction kits were obtained from Axygen Co. Biological reagents such as restriction enzymes and transfection reagents were obtained from Thermo Scientific. All chemicals were obtained from Sinopharm Chemical Reagent Co.

Gene cloning
A Filamin A gene was amplified by PCR using the cDNA synthesized with 1 μg of the total RNA extracted from 293T cells. The HA-Filamin A fusion gene was generated by PCR using the resulting Filamin A gene and inserted into the pcDNA 3.0(+). TFAP2A shRNA-coding DNA fragments were synthesized by Sangon Biotech. After annealing and phosphorylation, DNA fragments were loaded downstream of the U6 promoter at the pLV-U6-mCherry-Puro plasmid. MDM2 cDNA fragments were amplified by PCR and loaded downstream of the mCherry gene at the pLV-EF1α-mCherry-Puro plasmid to form a mCherry-MDM2 fusion gene. A TFAP2A gene was amplified by PCR using the cDNA synthesized as described above; thereafter, the HA-TFAP2A fusion gene was generated by PCR using the resulting TFAP2A gene and loaded downstream of the EF1α promoter at the pLV-EF1α-EGFP-Puro plasmid.

RT-qPCR and detection of Pol III products
SaOS2, 293T, and HeLa cell lines expressing FLNA shRNA or TFAP2A shRNA were grown in 12-well plates. At 90% confluence, cells were harvested and total RNA was prepared using an RNA extraction kit (Axygen). Reverse transcription was performed at 42 C in a 20-μL reaction mixture containing 0.5 μg of RNA and 10 U of reverse transcriptase (New England Lab). After incubating for 1 h, the reaction was terminated by heating the sample for 15 min at 85 C, followed by diluting it with 80 μl of ddH 2 O. Next, 1 μl of cDNA was used for quantitative PCR (qPCR) in which 20 μl of reaction mixture contained 10 μl of 2 × SYBR Green Master Mix (Vazyme Co) and 5 pmoles of the primers detecting Pol III target genes. Quantitative PCR was performed using a Bio-Rad real-time PCR detection system. The resulting data were processed with CFX Manager 3.1 software (Bio-Rad), where an Actin or GADPH gene was used as a reference gene.

Western blot analysis
SaOS2, 293T, and HeLa cell lines were cultured in six-well plates. At 90% confluence, cells were harvested and lyzed using 250 μl of 1 × SDS loading buffer. After heating for 10 min at 100 C, 20 μl cell lysate was loaded into wells within an SDS-PAGE gel for electrophoresis. After that, immunoblotting was performed using antibodies against the factors, including

Cell proliferation assays
Cell proliferation assays were performed using different methods, including cell counting, MTT, CCK8, and EdU. For cell counting and EdU assays, 293T and HeLa cell lines expressing TFAP2A shRNA or HA-TFAP2A and their control cell lines were cultured in 12-well plates. Proliferation assays were performed as described (49,53,54). For MTT assays, 293T and HeLa cell lines with TFAP2A knockdown or overexpression and their control cell lines were cultured in 96-well plates. MTT assays were performed as described (49). For CCK8 assays, 293T and HeLa cell lines stably expressing both TFAP2A shRNA and mCherry-AP2A and their control cell lines were cultured in 96-well plates. After culturing for 24 h, 10 μl of CCK8 solution (Vazyme) was added to each well and incubated for 4 h at 37 C. Absorbance values for individual samples at the 450 nm wavelength were measured using the SpectraMax i3 equipment (Molecular Device).

Chromatin immunoprecipitation assays
HeLa cell lines stably expressing TFAP2A shRNA or control shRNA were cultured in 10-cm dishes. At 90% confluence, cell lines were fixed for 10 min using 10 ml of 1% formaldehyde freshly prepared with 1 × PBS before adding 1 ml of 2.5 M glycine solution into the medium to quench the fixation. After washing twice with 1 × PBS solution, the fixed cells were lysed with 1 ml ChIP lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl [pH8.0]), followed by disrupting for 20 min in an ultrasonicator and centrifuging for 10 min at 12,000 rpm. The supernatant was retained and diluted 7 times with RIPA buffer (0.01% SDS, 1.1% TritonX-100, 1.2 mM EDTA, 16.7 mM Tris-HCl pH 8.0, and 167 mM NaCl). The rest of the procedures for ChIP assays were performed as described (49,54). Chromatin immunoprecipitation was performed using antibodies against BRF1 (SC-81405, Santa Cruz Biotech), BRF2 (Ab17011, Abcam), GTF3C2 (SC-81406, Santa Cruz Biotech), and POLR3A (SC-292119, Santa Cruz Biotech). After de-cross-linking, DNA from ChIP samples or input samples was recovered using a DNA purification kit (Axygen), where 40 μl of ddH 2 O was used to elute the DNA. One microliter of DNA samples (1/40) from a ChIP assay was used for one qPCR reaction; meanwhile, 1 ng of genomic DNA from input (0.05% input) was used as a positive control during qPCR. The relative enrichment of each Pol III gene locus was obtained by comparing the relative DNA quantity from 1 μl of ChIP sample to that from 1 ng input DNA.

Reporter assays
The DNA fragments for Brf1 gene promoter 4, Gtf3C2 gene promoter 2, and MDM2 gene promoter 1 were amplified (E) assays were performed using cell lines established in A and B and the corresponding control cell line. F, BRF1 and GTF3C2 expression in SaOS2 cell lines was detected by RT-qPCR. G, BRF1 and GTF3C2 expression in SaOS2 cell lines was detected by Western blot. H, a proposed model showing the regulatory mechanisms by which TFAP2A modulates Pol III-directed transcription. Each column in histograms represents the mean ± SD of three biological replicates. *p < 0.05; **p < 0.01. p Values were obtained by Student's t test. RT-qPCR, reverse transcription-quantitative polymerase chain reaction. from 293T genomic DNA by PCR. The resulting promoter DNA was inserted into pGL3-basic reporter vector (Promega). 293T and HeLa cells were cultured in 12-well plates for 24 h and were subjected to transient transfection with the Brf1P4 or Gtf3c2P2-driving reporter vectors and the vectors expressing β-galactosidase, where three biological replicates for treatment or control groups were designed. After 48 h, cell samples were harvested and lysed with 100 μl of lysis buffer provided by a dual-light luciferase detection kit (Promega). Five microliters of cell lysate was used to detect the activities of luciferase and β-galactosidase based on the manual provided by the kit. The luciferase activity of each sample was normalized by the activity of β-galactosidase within the same sample. Relative luciferase activity was obtained by comparing the luciferase activity from the treatment samples to that from the control samples, where the activity of the control sample was arbitrarily set as 1. For reporter assays using MDM2P1-driving reporter vectors, luciferase activity for treatment and control was normalized by the protein quantity used in each assay.

Statistical analysis
The experiments including RT-qPCR, cell proliferation assays, ChIP assays, and reporter assays were performed in three biological replicates. The mean and standard deviation (SD) for the data from gene expression analysis (RT-qPCR), cell proliferation assays, reporter assays, and ChIP assays (qPCR) were calculated using the GraphPad Prism 6.0 software. Each column or point in all digital graphs in this article represents the mean±SD of three biological replicates (n = 3). p Values were obtained by a Student's t test or two-way analysis of variance.

Data availability
All data were included in main figures in the article and figures in the supplementary file.