Tricho-rhino-phalangeal syndrome 1 protein functions as a scaffold required for ubiquitin-specific protease 4-directed histone deacetylase 2 de-ubiquitination and tumor growth

T47D, BT474, MCF7, MDA-MB-231, and HEK293T cell lines were purchased from American Type Culture Collection (ATCC) and were authenticated by the short tandem repeat (STR) typing. The cell lines were used for the current study within 6 months after cell authentication. BT474 and HEK293T cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Life Technologies, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS) (HyClone, NY, USA) and 1% penicillin-streptomycin solution (Life Technologies). T47D and MCF7 were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium (Corning Cellgro) supplemented with 10% FBS and 1% penicillin-streptomycin solution. To generate TRPS1 overexpression system in MDA-MB-231, the open reading frame (ORF) of TRPS1 was cloned into the lentivirus vector pCDH-CMV-MCS-EF1-copGFP (System Bioscience, CA, USA) and then transfected into HEK293T cells. The virus-containing supernatant was collected 48 h after transfection, and passed through the 0.45 μm filter to infect MDA-MB-231 cells.

The following plasmids were used in this study: pCDNA3.1-Flag-USP4, pCDNA3.1-Myc-HDAC2, pCDNA3.1-His-Ub, p3 × Flag-TRPS1, p3 × Flag-TRPS1-N(1–2640), p3 × Flag-TRPS1-ΔC(1–2940), p3 × Flag-TRPS1-C(2941–3885), p3 × Flag-TRPS1-ΔN(2641–3885), p3 × Flag-TRPS1-GATA(2641–2940), 4 × UAS-TK-luciferase, GAL4-HDAC2, and Renilla luciferase (pRL-SV40).

Antibodies used in this study and their sources are as follows: anti-TRPS1 (R&D Systems#AF4838), anti-HDAC2 (Cell Signaling Technology#5113), anti-H4K16ac (Millipore#39929), anti-H4 (Millipore#04–858), anti-USP4 (Cell Signaling Technology#2651), anti-actin (proteintech#60008–1-Ig), anti-HA (Biotool#B23402), anti-Flag (Sigma#F7425), anti-Myc (Biotool#B23402), anti-His (Cell Signaling Technology#12698), and anti-Gal4 (Santa Cruz#510).

Cells were collected and lysed for 15 min on ice in the lysis buffer (Beyotime) supplemented with a protease inhibitor. The cell lysates were incubated with antibodies and protein A/G agarose overnight at 4 °C. Unbound proteins were removed by washing three times with wash buffer. The immunoprecipitates from agarose beads were removed using the elution buffer (50 mM Tris–HCl (pH 7.4), 900 mM NaCl, 1 mM EDTA, 1% Triton X-100) for sequential co-immunoprecipitation (Co-IP). For immunoblotting following SDS-PAGE, immunoprecipitated proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Millipore) and probed with various antibodies. The ECL detection system (ImageQuant LAS4000) was used for detection.

To examine the ubiquitin-modified proteins, cells were lysed in the denaturing buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 4%SDS, 1 mM EDTA, 8% glycerol, 1 mM DTT, 1 mM PMSF and protein inhibitors) supplemented with 20 mM NEM and heated at 90 °C for 10 min. For immunoprecipitation, the lysates were further diluted to 0.1% SDS and immunoprecipitated with anti-Myc antibody at 4 °C overnight and then the ubiquitinated proteins were tested by western blotting.

For RT-qPCR analysis, total RNAs were extracted from various cell lines using the RNeasy kit (Qiagen, Hilden, Germany) and reverse transcription of RNA was performed using PrimeScript RT reagent kit (TaKaRa, Otsu, Shiga, Japan) according to the manufacturer’s instructions. The primers used for RT-qPCR are listed in Additional file 1: Table S1.

Small interfering RNAs (siRNAs) targeting TRPS1, HDAC2, USP4 or the non-targeting control siRNA (Genepharm, Shanghai, China) were transfected into MCF7, BT474, and T47D cells using Lipofectamine RNAi MAX (Invitrogen; Carlsbad, CA, USA) according to the manufacturer’s instructions. All plasmids were transfected using Lipofectamine®2000 (Invitrogen; Carlsbad, CA, USA). An MCF7 cell line with stable depletion of TRPS1 was generated using a lentivirus short hairpin RNA (shRNA) system. The sequences for siRNAs and shRNA are listed in Additional file 2: Table S2.

HEK-293 T cells were transfected with 4 × UAS-TK-luc, Gal4-HDAC2, pRL-SV40, and Flag-TRPS1/truncations or control vector as indicated. Cells were subjected to luciferase reporter assay according to instructions provided in the Promega dual luciferase reporter assay kit.

Cell proliferation was measured using the CCK-8 kit according to the protocol recommended by the manufacturer (Dojindo Laboratories, Kumamoto, Japan). Cells were seeded into 96-well plates. After treatment with siRNA, cells were grown for 24 h or 48 h. Absorbance was read at 450 nm using a Bio-Rad iMark plate reader.

MCF7 cells were transfected with non-targeting siRNAs (control siRNAs), TRPS1 siRNAs, and HDAC2 siRNAs. At 48 h later, total RNA was extracted from the same number of cells from each group with the RNeasy kit (OMEGA). Analysis of RNA-Seq data was performed using a standard TopHat-Cufflinks workflow.

Formaldehyde was added to the cell culture medium at a final concentration of 1%, then incubated at room temperature (RT) with shaking to create protein-DNA crosslinks. After 10 min, glycine was added to the cell culture medium to stop fixation. Subsequently, the cells were washed with ice-cold PBS, harvested in SDS lysis buffer containing the protease inhibitor, and sheared by sonication. Sheared chromatin was used for immunoprecipitation with IgG and anti-H4K14Ac antibodies. The immunoprecipitates were washed, reverse crosslinked, and eluted to obtain the purified DNA for q-PCR. Primer sequences used for ChIP-qPCR experiments are listed in Additional file 1: Table S1.

Four-to-six-week-old female athymic nude (Foxn/nu/nu) mice were purchased from the Model Animal Research Center of Nanjing University. MCF-7 cells (5 × 10⁶) suspended in 200 μl of the PBS–Matrigel mixture were injected into the mammary fat pads. A 0.72-mg E2 60-day release pellet (Innovative Research of America, Sarasota, FL, USA) was implanted subcutaneously on the dorsal side of each mouse a day before tumor cells injection. The length and width of tumors were examined weekly using a Vernier caliper and the volume was calculated by the formula:

π/6 × length×width².

Analysis of TRPS1 interactome had indicated that TRPS1 is associated with HDAC1 and HDAC2 [23], which belong to class I HDACs. To explore the function of TRPS1, we used luminal breast cancer cells MCF7, T47D, and BT474 with elevated TRPS1 as model cell lines. Silencing TRPS1 with two different siRNAs in all three cell lines consistently showed decreased HDAC2 but not HDAC1 protein levels indicating TRPS1 positively regulates HDAC2 and not HDAC1 (Fig. 1a–c and Additional file 3: Figure S1A–C). To further confirm this observation, we overexpressed TRPS1 in MDA-MB-231 cells with no detectable endogenous TRPS1 expression and detected increased HDAC2 protein levels with ectopic overexpression of TRPS1 (Additional file 3: Figure S1D). Since TRPS1, a member of the GATA transcription factor family, is a well-documented transcription factor, we first investigated whether TRPS1 transcriptionally regulated the expression of HDAC2. We found that neither knockdown nor overexpression of TRPS1 was able to significantly change HDAC2 mRNA levels (Fig. 1d–f and Additional file 3: Figure S1E). These observations suggested that TRPS1 regulated HDAC2 protein level independent of its transcription factor function. We then hypothesized that TRPS1 regulates HDAC2 protein level by affecting its protein stability. To test this idea, we first treated cells with cycloheximide (CHX), a eukaryotic protein synthesis inhibitor, and found that in the presence of CHX, silencing of TRPS1 significantly reduced HDAC2 protein stability (Fig. 1g–i). Conversely, overexpression of TRPS1 in cells increased HDAC2 stability (Additional file 3: Figure S1F). These results suggested that TRPS1 positively regulated and stabilized HDAC2 protein level.

It is well-known that the majority of intracellular proteins are degraded by UDPD [24]. To test whether TRPS1 stabilized HDAC2 levels by modulating UDPD, we used MG132, a specific 26-s proteasome inhibitor with the ability to reduce the degradation of ubiquitin-conjugated proteins in mammalian cells. Fig. 2a–c shows that HDAC2 protein levels were sustainable in the presence of MG132 if we silenced TRPS1. To further confirm if TRPS1 inhibited HDAC2 in a UDPD-dependent manner, we performed ubiquitination assay by co-transfecting HEK293T cells with Myc-HDAC2 vector plus His and Flag-empty vectors, or His-Ubiquitin and Flag empty vectors, or His-Ubiquitin and Flag-TRPS1 vector and compared these with Myc-empty vector plus His and Flag empty vectors. At 24 h later, we immunoprecipitated HDAC2 from whole cell lysates using an anti-Myc antibody, and evaluated HDAC2 ubiquitination (Ub-HDAC2) using anti-His antibody. We found that TRPS1 overexpression reduced HDAC2 ubiquitination level (Fig. 2d). Taken together, these results suggested that TRPS1 stabilized HDAC2 protein levels by reducing the UDPD of HDAC2.

To test whether TRPS1 inhibits UDPD of HDAC2 through directly binding to HDAC2, we performed Co-IP using anti-TRPS1 antibody. As displayed in Fig. 3a, b, the results confirmed that TRPS1 physically associated with HDAC2. Since TRPS1 is not a ubiquitin-specific peptidase that could directly de-ubiquitinate and stabilize its interacting partner HDAC2, we hypothesized that TRPS1 reduced HDAC2 ubiquitin level by a ubiquitin-specific peptidase to de-ubiquitinate HDAC2 by reducing UDPD of HDAC2. It has recently been reported that USP4 directly interacts and de-ubiquitinates HDAC2 resulting in its stability in colon cancer cells [22]. To test whether USP4 was responsible for TRPS1-mediated HDAC2 de-ubiquitination and stabilization, we first tested whether, like TRSP1, USP4 had a stabilizing effect on HDAC2. By silencing USP4 in T47D and MCF7 cells, we detected decreased HDAC2 protein levels as was the case with TRPS1 (Fig. 3c, d).

Since TRPS1 is well-documented as a transcription factor, we examined whether USP4 is transcriptionally regulated by TRPS1. We investigated USP4 expression upon silencing TRPS1 and found no change in either USP4 mRNA or protein levels (Additional file 4: Figure S2A, B). It appeared that both TRPS1 and USP4 interacted with and stabilized HDAC2 in a UDPD-dependent fashion. We hypothesized that TRPS1, USP4, and HDAC2 formed a complex, used Co-IP to test this notion, and found that TRPS1 co-immunoprecipitated with USP4 and HDAC2 (Fig. 3a, b). To further validate this point, we co-transfected Flag-USP4, Myc-HDAC2, and Flag-TRPS1 in HEK293T cells and performed Co-IP using anti-Myc antibody and found that TRPS1, USP4, and HDAC2 were co-purified as a complex (Fig. 3e).

To further confirm the interaction between TRPS1, USP4, and HDAC2, we generated serial TRPS1 truncates based on the domain structure of TRPS1 consisting of an N-terminal domain and GATA and C-terminal domains (Fig. 3f). We performed Co-IP experiments in HEK293T cells with ectopic overexpression of TRPS1 or TRPS1 truncation mutants. The results indicated that N-terminal or C-terminal domains of TRPS1 together with the GATA domain were sufficient to interact with HDAC2 and USP4 while neither TRPS1 N-terminal or C-terminal domain alone interacted with HDAC2 and USP4 (Fig. 3g). These observations indicated that GATA-zinc finger domain of TRPS1 was essential for its interactions with HDAC2 and USP4. To test whether the TRPS1 GATA domain alone was sufficient to interact with USP4 or HDAC2 or both, we carried out Co-IP experiments in HEK293T cells with ectopic overexpression of TRPS1-GATA truncate containing only the GATA-zinc finger domain. Our results indicated that the GATA domain of TRPS1 alone was unable to interact with HDAC2 and USP4 (Fig. 3h). These observations suggested that the GATA-zinc finger domain of TRPS1 was necessary but not sufficient for TRPS1, USP4, and HDAC2 complex formation. Interestingly, although ectopically overexpressed Flag-UPS4 could be co-immunoprecipitated with overexpressed Myc-HDAC2 using anti-Myc antibodies, overexpression of TRPS1 could increase the co-immunoprecipitated UPS4 levels in this experiment (Fig. 3e). These findings raised the possibility that TRPS1 functions as a scaffold protein to facilitate the UPS4 and HDAC2 interaction. To test this notion, we first performed a sequential immunoprecipitation assay and confirmed that TRPS1, USP4, and HDAC2 formed a ternary complex (Fig. 3i). Furthermore, when we silenced TRPS1 in cells and carried out Co-IP using anti-HDAC2 antibodies, the co-immunoprecipitated USP4 level with HDAC2 decreased significantly (Fig. 3j, k). We further verified that TRPS1 functioned as a scaffold protein for the complex formation of TRPS1-USP4-HDAC2 and was responsible for the reduction of HDAC2 ubiquitination level by facilitating interaction between USP4 and HDAC2. We carried out ubiquitination assays by co-transfecting HEK293T cells with a Myc-HDAC2 expression plasmid plus His and Flag empty vectors, His-Ubiquitin and Flag empty vectors, His-Ubiquitin, Flag-USP4 and Flag empty vectors,or His-Ubiquitin, Flag-USP4 and Flag-TRPS1 vectors and compared them with cells transfected with Myc plus His vector and Flag-empty vectors. Although USP4 alone could reduce HDAC2 ubiquitination level, additional overexpression of TRPS1 significantly enhanced the reduction of HDAC2 ubiquitination (Fig. 3l). Taken together, our results suggested that TRPS1 stabilized HDAC2 by functioning as a scaffold protein to bring UPS4 and HDAC2 together forming the TRPS1-UPS4-HDAC2 complex to enhance USP4-directed HDAC2 de-ubiquitination.

HDAC2, as a histone deacetylase, exerts its transcriptional repression activity by deacetylating histones [25]. H4K16ac, which is a key epigenetic marker and is necessary for transcriptional regulation [26], has been shown to be a major deacetylation target of HDAC2 [16]. Consistent with these notions, we observed that silencing of HDAC2 in T47D and MCF7 cells led to increased H4K16ac levels (Additional file 5: Figure S3A, B). Furthermore, silencing TRPS1 increased H4K16ac level as did silencing HDAC2 (Additional file 5: Figure S3C–E). These observations indicated that TRPS1-UPS4-HDAC2 formed a functional axis in controlling the acetylation status of H4K16 via HDAC2. To further confirm this notion, we first tested whether TRPS1 controlled the acetylation status of H4K16 via HDAC2. We silenced TRPS1 with siRNAs with or without rescuing overexpression of HDAC2 in MCF7 and found that silencing TRPS1 led to decreased HDAC2 and increased H4K16ac, whereas HDAC2 overexpression restored the H4K16ac level (Fig. 4a, b). Furthermore, when we silenced USP4 with siRNAs with or without HDAC2 overexpression in MCF7, we consistently found that USP4 silencing led to decreased HDAC2 and increased H4K16ac, whereas additional HDAC2 overexpression restored the H4K16ac level (Fig. 4c, d). Reduced H4K16ac has been shown to be an indicator of transcription repression [27‐29]. Gal4-TK-luciferase reporter assay is generally used to test HDAC2 transcriptional repression activity [30‐33]. To further investigate whether TRPS1 could affect HDAC2 transcriptional repression activity, we co-transfected the Gal4-TK-luciferase plasmid with Gal4-HDAC2 with or without Flag-TRPS1 and TRPS1 domain truncates. As shown in Fig. 4e–i, ectopic overexpression of full-length TRPS1 and TRPS1-△N truncate exhibited strong repression activity of luciferase expression while TRPS1-N, TRPS1-△C, and TRPS1-C truncates did not show significant effects. These observations indicated that TRPS1 regulated transcriptional repression activity of HDAC2. Furthermore, the C-terminal domain of TRPS1 was necessary but not sufficient for transcriptional repression activity of HDAC2. This could be due to the fact that C-terminal and GATA domains of TRPS1 were required for binding of TRPS1 to HDAC2 (Fig. 3). As we proposed, TRPS1 recruited UPS4 to mediate HDAC2 stability, and regulated its transcriptional repression activity (Additional file 5: Figure S3F).

To further identify the transcriptional output of HDAC2 mediated by TRPS1, we first investigated transcriptional alterations by RNA sequencing in MCF7 cells. Upon silencing of TRPS1 in MCF7 cells, 66 and 33 genes were up- and down-regulated, respectively (fold change ≥ 2 and q < 0.05), and upon silencing HDAC2 in MCF7 cells, 23 and 44 genes were up-regulated and down-regulated, respectively (fold change ≥ 2 and q < 0.05) (Additional file 6: Table S3A, B). There were 10 genes, ADAMTS7P1, AES, CASP7, GS1-44D20.1, IFT27, PCDH19, PERP, SHISA2, TPM4, and ZW10, that were consistently upregulated upon silencing either TRPS1 or HDAC2, while 8 genes, CC2D1B, CCDC146, CCT6P3, GAPDHP1, LINC01659, RORA, SMOX, and U1 were consistently down-regulated upon silencing either TRPS1 or HDAC2 (Fig. 5a). The genes with consistently up-regulated expression upon silencing either TRPS1 or HDAC2 were considered as candidate targets for TRPS1 transcriptional repressive function over HDAC2. RT-qPCR validated that AES, CASP7, IFT27, PERP, SHISA2, TPM4, and ZW10 were consistently up-regulated upon silencing either TRPS1 or HDAC2 (Fig. 5b, c). Furthermore, additional overexpression HDAC2 in cells with TRPS1 silencing restored the expression levels of these genes (Fig. 5d). Considering the function of the regulatory axis of TRPS1, USP4, and HDAC2 on transcriptional regulation, we further confirmed that silencing USP4 led to consistent up-regulation of AES, CASP7, IFT27, PERP, SHISA2, TPM4, and ZW10, and additional overexpression of HDAC2 restored the expression levels of these genes (Fig. 5e, f). To determine the H4K16ac level at the affected genes, we carried out ChIP-qPCR using H4K16ac antibody upon silencing TRPS1 or USP4 and found that silencing either TRPS1 or USP4 alone increased the H4K16ac level at these target genes (Fig. 5g, h). Taken together, these results support our claim that the TRPS1-UPS4-HDAC2 axis regulates transcriptional repression activity of HDAC2.

The set of genes consisting of AES, CASP7, IFT27, PERP, SHISA2, TPM4, and ZW10, repressed by the TRPS1-USP4-HDAC2 axis were implicated in cell proliferation [34‐40]. Thus, we postulated that the TRPS1-USP4-HDAC2 axis contributes to tumor growth. To test this idea, we first performed in vitro cell culture experiments. Silencing of TRPS1 reduced proliferation of MCF7 cells while overexpression of HDAC2 in cells with TRPS1 silencing rescued this phenotype (Fig. 6a). Similarly, silencing of USP4 also reduced proliferation of MCF7 while overexpression of HDAC2 in cells with USP4 silencing rescued this phenotype (Fig. 6b).

Consistent with the in vitro experiments, in xenograft mouse models, silencing TRPS1 significantly reduced both tumor volume and weight, and additional overexpression of HDAC2 rescued this phenotype of xenograft tumors (Fig. 6c–e). Thus, both in vitro and in vivo results demonstrated that the TPRS1-USP4-HDAC2 regulatory axis confers tumor growth.

Taken together these results led to a mechanistic scheme in which TRPS1 scaffolding recruits USP4 to de-ubiquitinate and stabilize HDAC2, leading to repression of a set of anti-growth genes and acceleration of tumor growth (Fig. 6f).