TSA was obtained from Selleckchem (Houston, TX, USA). TSA was dissolved in 70% ethanol to prepare a stock concentration of 500 nM, which was then stored at −20°C until use. Anti-baculoviral IAP repeat containing 5 (BIRC5) (cat #ab469), anti-cyclin-dependent kinases regulatory subunit 1 (CKS1B) (cat #ab72639), anti-NDC80 (cat #ab186839) antibodies, and anti-STAT5A (cat #ab32043) antibodies were obtained from Abcam (Cambridge, UK); anti-STAT5A (cat #94205, ChIP grade) was obtained from Cell Signaling Technology (Danvers, MA, USA); and a mouse monoclonal anti-β-actin antibody (cat #A5441) was obtained from Sigma-Aldrich Chemicals (St. Louis, MO, USA).

This study was registered with and approved by the Ethics Committee of the First Affiliated Hospital of Fourth Military Medical University (registration number: KY20213431-1). Primary ovarian carcinoma tissues, breast carcinoma tissues, and adjacent normal tissues were collected from patients in Xijing Hospital. HUVECs were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). HUVECs were maintained in EGM-2 endothelial cell growth medium (Lonza, Basel, Switzerland) at 37°C in a humidified incubator with 5% CO₂ and 95% O₂. For drug treatment, cells were grown to 70%–80% confluency and then treated with different doses of TSA for specific durations of time. The human ovarian carcinoma cell line SKOV3 and the human breast carcinoma cell line MCF-7 were purchased from the ATCC and maintained in DMEM supplemented with 10% FBS (6, 7).

Microarray data for differentially expressed genes in HUVECs treated with or without TSA were downloaded from the NCBI GEO database with the accession number GSE5856 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE5856). In particular, in differentially expressed genes (DEGs), we considered the false discovery rate (FDR) <0.05 and a log₂ fold change |log₂FC| >1. Based on the list of 214 DEGs, we performed gene ontology enrichment analysis of DEGs using the DAVID bioinformatics resources website (https://david.ncifcrf.gov/) (8). The ClueGO+CluePedia tool in Cytoscape was used to perform cluster-based analysis of the biological processes dysregulated in TSA-treated HUVECs ( Supplementary Figure 2 ). The structure of STAT5A was predicted on the Swiss-model website (https://swissmodel.expasy.org). Furthermore, we predicted the signal transducer and activator of transcription 5 A (STAT5A)-binding sites [gamma-activated site (GAS) 5′-TTCNNNGAA-3′] of the BIRC5, CKS1B, and NDC80 promoter regions using a tool from the NCBI website (9). The data were uploaded in github website (https://github.com/bwddcgzl/EC-STAT5A.git).

For isolation of endothelial cells from human ovarian carcinoma tissue, breast carcinoma tissue, and adjacent normal tissues, tissues were cut into pieces with a surgical knife and digested with collagenase 1 (Gibco, 17100017) in DMEM for 2 h at 37°C. The solution derived from the tissue was filtered, and red blood cells were eliminated using red blood cell lysis buffer. A single-cell suspension was prepared and resuspended to achieve a cell concentration of 1 × 10⁸ cells/ml. For labeling with Dynabeads (Invitrogen, 11155D), the beads were suspended in a vial, washed via the placement and removal of a magnet, and then resuspended. The prepared cell solutions were incubated with the resuspended Dynabeads, before washing as described to obtain the bead-bound CD31+ endothelial cells. Pellets of these cells were transferred to the preferred cell medium (10).

Cells of the human breast carcinoma cell line MCF-7 and ovarian carcinoma cell line SKOV3 were first seeded in a 10-cm dish in growth medium containing 10% FBS, separately. To prepare the tumor-conditioned medium (TCM), the growth medium was replaced with serum-free medium, and the breast and ovarian cancer cells were incubated for an additional 48 h. The TCM was collected and filtered through a 0.2-μm filter for use in culture experiments.

Total RNA was isolated from the HUVECs using TRIzol reagent (Invitrogen) and reverse transcribed into cDNA using the miScript Reverse Transcription Kit (Qiagen, Valencia, CA, USA) according to the instructions of the manufacturer. Quantitative real-time polymerase chain reaction (qRT-PCR) amplification was then performed with primers of human BIRC5, CKS1B, NDC80, and STAT5A and a housekeeping gene GAPDH (GenePharma, Shanghai, China). The primers were designed using the Premier 5.0 software (Biosoft International, Palo Alto, CA, USA; see Supplementary Table 1 ). The qRT-PCR conditions were set to 95°C for an initial 5-min incubation, then 35 cycles of 95°C for 10 s, 58°C for 30 s, and 72°C for 30 s, followed by a final 10-min incubation at 72°C. The samples were then stored at 4°C. GAPDH mRNA was used as an internal loading control. Relative mRNA levels were calculated using the 2^−[Δ][Δ]Ct method.

HUVECs in the logarithmic growth phase were seeded into a 96-well plate at a density of 2,000 cells per well and grown overnight. The cells were then treated with different doses of TSA for up to 5 days, and the cell numbers were calculated daily for 5 days using a Cellomics ArrayScan VTI reader (Cellomics, Inc., Pittsburgh, PA, USA) with analysis settings adjusted according to the input parameters. The experiments were performed in triplicate and repeated three times independently.

The STAT5A open reading frame (ORF) cDNA was amplified by PCR from HUVECs and subcloned into pcDNA3.0 (Invitrogen), while plasmids carrying human BIRC5, CKS1B, and NDC80 cDNA and promoter regions were amplified as described previously (7, 11). Plasmids carrying a mutated GAS site (5′-TTCNNNGAA-3′) within the distal BIRC5, CKS1B, and NDC80 promoter regions were constructed using the site-directed mutagenesis kit from Stratagene (La Jolla, CA, USA) as described previously (12). After confirmation of the DNA sequences, HUVECs were transfected with these plasmids. The PCR primers are listed in the Supplementary Table 1 . Furthermore, we chemically synthesized small interfering RNAs (siRNAs) and control RNAs and used them to knock down STAT5A, BIRC5, CKS1B, and NDC80 expression to study the effects on HUVEC proliferation. For transfection, HUVECs were seeded into six-well plates at a density of 2.5 × 10⁵ cells per well, grown for 24 h, and then transfected with plasmids using Lipofectamine 2000 (Invitrogen) for 48 h according to the protocol of the manufacturer. The cells were then subjected to the described assays. GAS site wild-type and mutation sequence was uploaded in github website (https://github.com/bwddcgzl/EC-STAT5A.git).

A chromatin immunoprecipitation (ChIP) assay was performed to assess in vivo DNA–protein interactions at the BIRC5, CKS1B, and NDC80 promoters using the ChIP-IT express enzymatic kit in accordance with the instructions of the manufacturer (Active Motif, CA, USA). Briefly, cells were grown, treated with TSA, and then cross-linked with 1% formaldehyde at room temperature for 10 min to induce the cross-linking of protein–DNA complexes. After that, cells were washed twice with ice-cold phosphate-buffered saline (PBS) containing protease and lysed using the ChIP lysis buffer containing cell lysis buffer containing protease inhibitor cocktail II. Cells were centrifuged at 4°C for 5 min at 800×g, and the supernatant was removed. Then, cell pellets were resuspended in nuclear lysis buffer containing protease inhibitor cocktail II. After sonicating on wet ice, cell lysates were centrifuged, and the supernatant was collected in fresh microfuge tubes. Dilution buffer was added into each tube containing 50 μl of chromatin. Then, 5 μl (1%) of the supernatant was collected for future use as the “input.” The immunoprecipitating antibody and 20 μl of fully resuspended protein A/G magnetic beads were incubated (with rotation) with the STAT5A antibody at 4°C overnight with end-over-end rotation. Samples were washed with wash buffer and eluted with SDS elution buffer. Following elution, samples were reverse cross-linked at 65°C for 4 h, treated with RNase A at 37°C for 2 h, and mixed with CaCl₂ and proteinase K at 37°C overnight. The immunoprecipitated DNA samples were isolated and used as templates for PCR amplification of BIRC5, CKS1B, and NDC80 promoters or the target region of the predicted GAS sites. The primers are listed in the Supplementary Table 1 . The PCR conditions were set to an initial 5-min incubation at 98°C followed by 32 cycles of 94°C for 30 s, 56°C–58°C (depending on the primer) for 30 s, and 72°C for 30 s, with a final 10-min incubation at 72°C, as described previously (13).

Total cellular protein was extracted from HUVECs using a radioimmunoprecipitation assay buffer and quantified. Samples of 50 μg protein were separated by 10% SDS-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride (PVDF) membranes. The membranes were then blocked in non-fat dry milk–PBS solution at room temperature for 1 h and incubated with primary antibody at a dilution of 1:800 at 4°C overnight. The next day, the membranes were washed with PBS-Tween 20 (PBS-T) three times briefly and then incubated with an appropriate anti-rabbit or mouse IgG conjugated with horseradish peroxidase (HRP) at a dilution of 1:2,000 for 1 h at room temperature in darkness. The membranes were subsequently incubated briefly with an enhanced chemiluminescence (ECL) solution (Amersham Biosciences, NJ, USA). β-Actin was used as a loading control by stripping the membranes and reblotting with an anti-β-actin antibody.

The luciferase reporter assay was performed in 48-well plates as described previously (14) with the protocol of the manufacturer using the dual luciferase assay kit from Promega (Madison, WI, USA). HUVEC cells were grown and co-transfected with 100 ng of a firefly luciferase reporter plasmid pGL3 carrying different wild-type or mutant constructs or the Renilla luciferase vector pRL-TK (5 ng; for normalization) from Promega using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, USA) according to the protocol of the manufacturer. After 48 h, the total cellular protein was extracted using a radioimmunoprecipitation assay buffer, quantified, and subjected to the luciferase reporter assay. Luciferase activity was determined by calculating the ratio of firefly/Renilla luciferase activity and recorded using GloMax 96 microplate luminometer (Promega, USA). The experiments were performed in triplicate.

HUVECs transfected with the specified plasmids were cultured in TCM and treated with or without TSA for 48 h. The cells were then seeded in 24-well plates coated with Matrigel (BD Biosciences, NJ, USA). After incubation at 37°C with 5% CO₂ for 24 h, five randomly selected fields in each well were photographed using a Nikon inverted microscope. Total tube length and branch number were determined using ImageJ image analysis software (NIH, Bethesda, MD, USA) (15).

All numerical data are expressed as mean ± SD of at least three independent replicates and were statistically analyzed using PRISM, version 9 (GraphPad Software, CA, USA). Data were two-tailed Student’s t-test for qRT-PCR, luciferase assay, cell proliferation assay, and tube formation assay. A P-value equal to or less than 0.05 was considered statistically significant.

The HDACi TSA has been reported to influence angiogenesis, but its effects vary in different disease models. Elucidating how TSA directly affects endothelial cells and the underlying mechanisms is important for its future clinical application. First, we screened potential HDAC genes using data from a previous genome-wide microarray analysis of human HUVECs (GEO accession number: #GSE5856) and then confirmed the findings via qRT-PCR. The data showed that HDAC1–10 genes were coexpressed, with HDAC1 and HDAC2 as the predominant transcripts ( Figure 1A ). Then, we analyzed the HDAC transcripts in HUVECs after TSA treatment within the microarray data (#GSE5856) and identified the top 214 differentially expressed genes ( Figure 1B ). Analysis of the differentially expressed genes using the online software Cytoscape indicated that most TSA-regulated genes were enriched in endothelial cell proliferation ( Supplementary Figure 1 ). We then performed GO analysis of these differentially expressed genes and found that the functions of these dysregulated genes were involved in several important cell processes, including cell division, proliferation, and nucleus division ( Figure 1C ).

We hypothesized that the HUVEC proliferation repression induced by HDACi treatment is dose- and time-dependent. From treatment of HUVECs with a dose gradient, the HUVEC proliferation rate was decreased by 76% upon treatment with 500 nM TSA ( Figures 1D, E ). From treatment of HUVECs over a time gradient, HUVEC proliferation was decreased by 86% after 24 h of TSA treatment ( Figures 1F, G ). We concluded that treatment with 500 nM TSA for 24 h repressed HUVEC proliferation significantly, and this concentration and time are consistent with those used to obtain the data in GSE5856. These data indicate that HDACi treatment inhibited HUVEC proliferation.

Previous studies have shown that BIRC5, CKS1B, and NDC80 are involved in cell proliferation in a variety of cell types (16–18). Indeed, the analysis of microarray data in this study also identified BIRC5, CKS1B, and NDC80 as differentially expressed genes in TSA-treated HUVECs ( Supplementary Figure 2 ). We then performed qRT-PCR and found that treatment of HUVECs with 500 nM TSA for 24 h significantly reduced the expression of these three genes ( Figures 2A, B ).

Furthermore, we knocked down the expression of BIRC5, CKS1B, and NDC80 in HUVECs using siRNAs, along with control siRNA ( Figures 2C, D ), and found that the knockdown of endogenous BIRC5, CKS1B, and NDC80 expression reduced HUVEC viability ( Figures 2E, F ). These data showed that reduced BIRC5, CKS1B, and NDC80 expressions were involved in TSA-induced viability suppression of HUVECs.

STAT5A protein is an important transcriptional activator to maintain expression of a great deal of genes in various biological processes including cell proliferation (19). So, we wanted to know whether STAT5A is involved in TSA-induced cell proliferation inhibition of HUVECs. In the microarray data, TSA significantly downregulated STAT5A expression ( Figure 3A ). Further qRT-PCR assay revealed that TSA treatment inhibited the level of STAT5A mRNA in a dose- and time-dependent manner ( Figures 3B, C ). Moreover, after treatment of HUVECs with 500 nM TSA for 24 h, the levels of STAT5A mRNA and protein were significantly lower than those in cells with control treatment ( Figures 3D, E ). These results suggested that STAT5A may contribute to TSA effects in HUVECs.

Next, we explored the functions of STAT5A in HUVECs by knocking down STAT5A expression using STAT5A siRNAs. qRT-PCR and Western blotting assay confirmed the knockdown effect of siRNA-STAT5A in HUVECs ( Figures 3F, G ). When STAT5A was knocked down, we observed a significant decrease of cell proliferation in HUVECs ( Figures 3H, I ). Thus, knockdown of STAT5A expression in vitro inhibited HUVEC proliferation. Then, we designed a rescue experiment to investigate whether STAT5A mediated TSA effects in cell proliferation of HUVECs. qRT-PCR and Western blotting assay confirmed the overexpression effect of STAT5A vector in HUVECs ( Figures 3J, K ). In cell proliferation assay, STAT5A expression could partially rescue the TSA-induced cell proliferation inhibition of HUVECs ( Figures 3L, M ). These results demonstrated that TSA suppressed the cell proliferation of HUVECs via downregulating STAT5A expression.

A previous study demonstrated that the STAT5A protein could recognize the target genes via GAS-binding sites and activated gene transcription (20). The interaction between STAT5A protein and the target genes through GAS-binding sites (TTCNNNGAA/AAGNNNCTT) is displayed in Figure 4A . We scanned the BIRC5, CKS1B, and NDC80 promoter regions and observed several imperfect STAT5A-binding sites ( Figure 4B ). So, we hypothesized that STAT5A mediated the transcription of BIRC5, CKS1B, and NDC80. qRT-PCR and Western blotting assay demonstrated that the knockdown of STAT5A downregulated BIRC5, CKS1B, and NDC80 expression in HUVECs ( Figures 4C, D ). Then, we constructed different plasmids carrying the BIRC5, CKS1B, and NDC80 promoters and performed a luciferase reporter assay in HUVECs. As shown in Figure 4E , knockdown of STAT5A significantly reduced the firefly/Renilla luciferase activity of BIRC5, CKS1B, and NDC80 promoters compared with the control siRNA. These results suggested that STAT5A played important roles in maintaining the transcription of BIRC5, CKS1B, and NDC80.

Furthermore, we investigated whether TSA treatments involved STAT5A-mediated transcription of BIRC5, CKS1B, and NDC80 in HUVECs. We treated the HUVECs with TSA and transfected the STAT5A overexpression vector simultaneously. Compared with the control treatments, STAT5A overexpression partially rescues BIRC5, CKS1B, and NDC80 expressions which were downregulated by TSA in HUVECs ( Figures 4F, G ). This suggested that STAT5A mediated TSA-induced downregulation of BIRC5, CKS1B, and NDC80. Furthermore, the luciferase reporter assay confirmed that TSA treatments reduced the firefly/Renilla luciferase activity of BIRC5, CKS1B, and NDC80 promoters. However, STAT5A overexpression blocked the reduction effect of TSA on the firefly/Renilla luciferase activity of BIRC5, CKS1B, and NDC80 promoters ( Figure 4H ). These results revealed that TSA downregulates BIRC5, CKS1B, and NDC80 transcription through STAT5A repression.

The data described above demonstrated that STAT5A mediated the transcription repression of proliferation-related genes including BIRC5, CKS1B, and NDC80 under TSA treatment. We next assessed whether STAT5A directly interacted with the promoters of BIRC5, CKS1B, and NDC80. We constructed luciferase reporter vectors containing different truncations of the BIRC5 promoter region ( Figure 5A ). Then, the responding luciferase vectors containing different fragments of BIRC5 promoter were co-transfected with STAT5A overexpression vector into the HUVECs under TSA treatment. As shown in Figure 5B , STAT5A activated the luciferase reporter activity of the full length of the BIRC5 promoter. However, when the −979- to −37-bp fragment of the BIRC5 promoter was deleted, STAT5A failed to upregulate the luciferase reporter activity. This suggested that the −979- to −37-bp fragment of the BIRC5 promoter may contain the binding sites of STAT5A. Luciferase reporter vectors containing different truncated CKS1B promoters are displayed in Figure 5C . The -1695 bp to -979-bp fragment of the CKS1B promoter didn’t result in STAT5A-induced transcription of the luciferase reporter gene ( Figure 5D ). This suggested that activation of CKS1B promoter by STAT5A was mainly dependent on the remaining fragment of CKS1B promoter, the -979 bp to 335-bp. Additionally, we also cloned different truncations of NDC80 promoter in to the luciferase reporter vector ( Figure 5E ). The dual luciferase reporter gene assay showed that STAT5A activated NDC80 through -983 bp to -98-bp fragment of the BIRC5 promoter ( Figure 5F ).

Next, we wanted to identify the direct binding sites of BIRC5, CKS1B, and NDC80 promoters for STAT5A interaction. The potential GAS sites from the responsible promoter regions identified above were scanned by sequence alignment and displayed ( Figure 5G ). Plasmids carrying the wild-type and mutated GAS sites of these three gene promoters were constructed and co-transfected with STAT5A into HUVECs to perform dual luciferase reporter gene assay ( Figure 5G ). Compared to the wild-type promoters of BIRC5, CKS1B, and NDC80 mutants of the predicted GAS sites in their promoter abolished the STAT5A-induced upregulation of the luciferase activity ( Figure 5H ). This indicated that the GAS sites including −264 to −256 bp of the BIRC5 promoter, −864 to −856 bp of the CKS1B promoter, and −460 to −452 bp of the NDC80 promoter were responsible for STAT5A-activated transcription. Finally, we performed the ChIP assay to confirm the direct DNA–protein interaction between STAT5A and BIRC5, CKS1B, and NDC80 promoters. The ChIP data revealed that IgG did not precipitate the identified promoter regions of BIRC5, CKS1B, and NDC80, whereas the antibody against STAT5A precipitated the identified promoter regions of BIRC5, CKS1B, and NDC80 ( Figures 5I, J ). This supported the hypothesis that STAT5A activated the transcription of BIRC5, CKS1B, and NDC80 by directly binding their promoter region. All of these results revealed that STAT5A directly bound to promoters of BIRC5, CKS1B, and NDC80 via GAS sites and activated their transcription.

The expression level of STAT5A was first quantitatively analyzed in a panel of human cancer-derived endothelial cells (CDECs) including breast and ovarian CDECs. The results showed that STAT5A expression was increased in both breast and ovarian CDECs compared with its expression levels in adjacent normal endothelial cells ( Figures 6A, B ). Furthermore, when HUVECs were treated with breast and ovarian cell-derived TCM, STAT5A, BIRC5, CKS1B, and NDC80 expressions were all increased ( Figure 6C ). These results support the notion that STAT5A and its downstream targets may play important roles in cancer angiogenesis. Next, we wanted to know whether TSA displayed the same effects on gene regulation in breast and ovarian CDECs. qRT-PCR and Western blotting assay showed that TSA decreased mRNA and protein levels of STAT5A in both breast and ovarian CDECs ( Figures 6D, E ). Furthermore, TSA also decreased the expression of STAT5A downstream targets including BIRC5, CKS1B, and NDC80 at the mRNA level, but STAT5A overexpression disrupted the suppression of BIRC5, CKS1B, and NDC80 expression induced by TSA in both breast and ovarian CDECs ( Figures 6F, G ). These results suggested that TSA-associated STAT5A-mediated transcription of BIRC5, CKS1B, and NDC80 also existed in CDECs. Finally, in the tube formation model, TSA reduced tube branch formation by the breast and ovarian CDECs cultured in TCM, while STAT5A overexpression rescued branch formation by TSA-treated breast and ovarian CDECs ( Figures 6H–K ). All of these results suggested that TSA also inhibited BIRC5, CKS1B, and NDC80 expression and angiogenesis via STAT5A in the CDEC angiogenesis model.