Elevation of HJURP in breast cancer subtypes carrying frequent P53 mutations
We first aimed to stratify the data of RNA-Sequencing available from The Cancer Genome Atlas (TCGA) according to four main molecular subtypes of breast cancer: Luminal A, Luminal B, Triple-negative/basal-like, HER2-enriched. Correlation between HJURP expression levels and gene copy numbers and specific oncogenic mutations was also detected in the TCGA and Gene Expression Omnibus (GEO) datasets. P53 mutational spectrum and the prognostic implications indeed are subtype-specific. Consistently with the amplified frequency of P53 mutations in HER2 and TNBC subtypes, higher expression of HJURP was detected in triple-negative/basal-like and HER2-enriched subtypes compared to luminal tumors (Fig. 1A, Supplementary Fig. 1A-C). Overall mutation frequency in the breast cancer cohorts (PanCancer Altas and METABRIC) also indicated p53 mutation as the most common alteration in the HJURP-altered group (Fig. 1B). Expression of HJURP was presented to be positively correlated with proliferation marker as MKI67, PCNA, BRCA1 and BRCA2, and negatively correlated with ER and PR (Fig. 1C, Table 1). Our analysis found that HJURP expression was enriched in breast tumors carrying P53 mutations in TCGA and GEO data sets (Fig. 1D, Supplementary Fig. 1D). Interestingly, this effect was most substantial in the luminal-type tumors, suggesting a subtype-related mechanism contributing to HJURP activation and P53 mutagenesis (Fig. 1E). Furthermore, HJURP expression is also elevated in various mutp53 human tumors types, including the lung, bladder, and stomach tumors (Supplementary Fig. 1E). Besides, HJURP was also highly-expressed in TNBC compared with normal breast tissue and non-TNBC breast cancers, and the elevation of HJURP expression occurred in late-stage breast cancers (Supplementary Fig. 1F, G).
We next sought to detect HJURP expression in a selected panel of mutp53 and wtp53 breast cancer cell lines. Consistent with our findings, up-regulation of HJURP transcript and protein was clearly examined in breast cancer cell lines carrying mutp53 (Fig. 1F). Immunofluorescence assay further confirmed the much higher expression of HJURP in mutp53 MDA-MB-231 and BT-549 cells than that of wtp53 MCF-7 cells, and HJURP mainly localized to the cellular nuclear (Fig. 1G, H). To evaluate associations between HJURP protein content and breast cancer patient outcomes, we stained and analyzed a human tissue microarray (TMA) with annotated P53 mutation status, which consisting of primary breast tumor specimens and normal breast tissue, and for which molecular subtypes were available. Examples of HJURP-stained images from cases displaying abundant (HJURP high) or reduced HJURP staining (HJURP-low) are shown in Fig. 1I. Analysis of the TCGA cohort revealed a significant association of HJURP overexpression with tumor patient age, tumor grade, tumor size, ER/PR/HER2 status and lymph node metastasis (Table 2). Collectively, HJURP is highly expressed in human breast cancer subtypes with frequent P53 mutations, with certain clinical characteristics of breast cancer patients.
Inhibition of HJURP impaired cell growth and sensitized cells to chemotherapy regimens in TNBC carrying mutP53
We next utilized a loss- and gain-of-function strategy to elucidate the function of HJURP in regulating breast cancer cell biology. BT-549 and MDA-MB-231 cells carrying mutP53 were selected for HJURP shRNA knock-down experiments (Fig. 2A, B). Silencing of HJURP in BT-549 and MDA-MB-231 cells suppressed cell growth over time, compared with lentivirus carrying control shRNA transduced cells (Fig. 2C). We next determined the effects of HJURP silencing on cell growth by clonogenic cell survival assay. Deletion of HJURP reduced colony formation in both BT-549 and MDA-MB-231 cells (Fig. 2D). The group of HJURP knock-down showed a significantly higher apoptosis rate than the control. Expression levels of apoptosis markers as cleaved-PARP, cleaved-caspase 3 and 7 were increased in the HJURP silencing group compared to the control group (Fig. 2E, F). The function of HJURP in regulating cellular migration and invasion was explored using a Matrigel-coated Transwell assay. We found that knock-down of HJURP significantly reduced the ability of TNBC to migrate and invade through Matrigel (Fig. 2G, H). Cancer cells undergoing epithelial-mesenchymal transition (EMT) acquire a cancer stem-like phenotypes and metastatic traits associated with malignant behaviours. Here we found that HJURP is a major inducer of EMT progress. HJURP ablation prevented EMT and maintained an epithelial morphology of TNBCs (Fig. 2I). These results showed that HJURP silencing not only reduced cancer cell viability but also inhibited the metastatic capacity and EMT, which is crucial for an effective treatment strategy.
We next determined the role of HJURP on tumor proliferation using transplanted tumor mouse models. The tumor formation and growth of transplanted tumors were monitored for four weeks and we found that knockdown of HJURP led to impaired growth ability of MDA-MB231-engrafted tumors (Fig. 2J, Supplementary Fig. 2A). The tumor weights of Lenti-shHJURP-group were significantly lower than the control group (Fig. 2K). Moreover, we performed HJURP, and KI67 staining of randomly selected mouse tumors, and decreased KI67 and HJURP expression was detected in shHJURP-group (Fig. 2L). Altogether, these data indicate that HJURP inhibition suppresses the proliferation of MDA-MB231–engrafted tumours.
As cancer cells carrying mutp53 strongly rely on P53 GOF for survival, we determined whether silencing of HJURP affected the mutp53 expression level. We found that HJURP silencing by specific siRNA exerted a limited effect on mutP53 expression level in BT-549 and MDA-MB-231 cells (Supplementary Fig. 2B). Recent work has indicated P53 as a predictor of chemotherapy response in cancer patients, and P53-positive TNBC tend to more sensitive to chemotherapy. To test the effect of HJURP loss in chemotherapy responsiveness of TNBC carrying mutP53, we knocked down HJURP in BT-549 and MDA-MB-231 cells using siRNA and measured the cell survival upon Adriamycin (DOX), Paclitaxel (PTX) or Fluorouracil (5-FU) treatment. We found that the cells transfected with HJURP siRNA had decreased cell viability upon chemotherapy regimens treatment (Fig. 2M). These attributes are accompanied by a change in cell aggressive basal-like morphology (Fig. 2N, Supplementary Fig. 2C). Taken together, these data demonstrate loss-function of HJURP is associated with decreased cell growth, induced cell apoptosis and enhanced sensitivity to chemotherapy regimens in TNBC carrying mutp53.
HJURP overexpression bypassed P53-mediated growth arrest in wtp53 breast cancer cell
The functional effects of HJURP overexpression were investigated in wtp53 breast cancer cell lines MCF-7 and ZR-75-1 using in vitro assays. Previously, RT-qPCR analysis among a list of breast cancer cell lines illustrated that MCF-7 and ZR-75-1 cells expressed HJURP at relatively low levels. Overexpression of HJURP in MCF-7 and ZR-75-1 cells after lentivirus vector infection was confirmed by RT-qPCR and western blot analysis, and the expression level of P53 showed no obvious difference (Fig. 3A). We next use cellular functional assays to reveal that HJURP overexpression alone exerted a limited effect on cell migration and invasion ability (Fig. 3B, C). HJURP overexpression caused increased cell proliferation and decreased cell apoptosis in ZR-75-1 and MCF-7 cells (Fig. 3D, E). Considering the essential tumor-suppressive function of wtp53 in MCF-7 and ZR-75-1, we used a small-molecule inhibitor of P53-mediated transcription, pifithrin α (Pthα), to test the role of wtp53 in mediating HJURP function and whether HJURP could bypass either P53-mediated growth arrest or apoptosis (Fig. 3F). Colony formation was observed in the absence of Pthα, or transduced into a provirus containing HJURP or a control provirus. In cells with P53 inhibition (Pthα-treatment group), HJURP overexpression rescued proliferation of MCF-7 and ZR-75-1 cells (Fig. 3G). It was concluded that wtp53 plays a central role in mediating growth arrest of wtp53 breast cancer cell, and HJURP was capable of bypassing wtp53 activity as a negative regulator of cell proliferation, when added as an independent factor.
Identification of gene expression signatures related to HJURP induction in breast cancer
To decipher the underlying mechanism of HJURP as they pertain to breast cancer cell proliferation and metastasis, we performed whole transcriptome RNA-seq analysis using BT-549 and MDA-MB-231 cells with HJURP inhibition or MCF-7 cells with HJURP overexpression. As shown in the volcano plot, we identified a total of 2456 HJURP silencing-related genes in BT-549 cells, with 1111 upregulated and 1345 downregulated genes in the HJURP deletion group compared to the control group. Besides, we identified 257 and 519 genes that were significantly up- or downregulated, respectively between the MDA-MB-231-silencing and control group (Fig. 4A). The expression levels of 86 and 115 genes were positively and negatively correlated with HJURP overexpression in MCF-7 cells, respectively (Fig. 4B). Among these genes potentially regulated by the HJURP, most were involved in certain signalling pathways, physiologic regulation, and cell growth process (Fig. 4C, D). To further narrow down transcripts related to cancer progression, upregulated or downregulated genes with at least a 1.5-fold change with HJURP overexpression or inhibition were selected The expression of several oncogenes, including GLO1, E2F7, SPARC, GSTO1, BSCL2, GSKIP and FTH1, was positively regulated by HJURP (Fig. 4E), and the expression of several tumor suppressor genes, as p21, TP63 and TP53I3, was negatively regulated by HJURP (Fig. 4F). Together, these data indicated that multiple tumor-related genes are involved in HJURP-driven breast cancer progression.
MDA-MB-231 and BT-549 cells carry mutations in P53, and we observed that mutp53 expression was down-regulated with HJURP inhibition. Pathway enrichment analysis demonstrated P53 signalling pathway was induced upon HJURP silencing in BT-549 cells (Fig. 4C). Then, gene set enrichment analysis (GSEA) revealed that HJURP primarily affected the P53 signaling pathway, glycolysis, the G2/M checkpoint and E2F targeting (Data not shown here). Since several studies have identified a potential relationship between HJURP and P53, we attempted to investigate HJURP’s effect on P53. We analysed breast cancer data sets (METABRIC and TCGA) to examined whether HJURP expression correlated with P53-induced or repressed genes. HJURP expression was positively correlated with most of P53-repressed genes (CCNB1, CDC25C, CDK1 and BIRC5) but not with P53-induced genes (APAF1, CDKN1A, MYOC, BBC3) (Fig. 4G, H). The mRNA levels of selected P53-repressed genes (CDKN1A, BAK1, BLCAP, FOS, TOB1) in the P53 signalling pathway were consistently affected by HJURP in breast cancer cells, which was confirmed by qRT-PCR analyses (Fig. 4I). Taken together, the above results suggested that HJURP is a P53-response target gene and probably exerts its biological function through the inactivate P53 pathway.
Genome editing of enhancer region affected HJURP transcription
We systematically profiled regulatory landscapes in a series of TNBC (MDA-MB-231, CAL-51, BT-549, SUM 149T, SUM 159T and MDA-MB-436), non-TNBC (MCF-7, T-47D, AU565 and SK-BR-3) and normal human mammary epithelial cell (MCF-10A). Notably, significantly elevated H3K27ac signals were found at the proximal regions within HJURP coding sites in all TNBC samples (chromatin region: chr2:234,762,042–234,764,060). In contrast, only background level signals were present in non-TNBC cell lines and MCF-10A (Fig. 5A, Supplementary Fig. 3A). Increased enrichment of MED1, and P53-related transcription factors as E2F1, FOXM1 signals were overlapped with H3K27ac-enriched enhancer region. In addition, high-throughput measure by DNase-Seq mapped a DHS site within the cis-regulatory element, which is necessary for the binding of proteins such as transcription factors (Fig. 5B). These findings indicated that the generation of new enhancer and the aberrant activation of HJURP is specific to TNBC. Enhancers are bound by aberrant high levels of transcription factor, including CDK7 and the bromodomain and extra-terminal domain (BET) protein family, and drive cell type-specific gene expression programs. Downregulation of HJURP upon a specific inhibitor targeting CDK7 (THZ1) treatment was determined by qRT-PCR and immunoblot in a dose-dependent manner (Fig. 5C, D). As non-TNBC cells show no prominent enhancer activity and HJURP overexpression, we observed no significant decreased of HJURP expression when ZR-75-1 and MCF-7 cells were treated with THZ1 for 24 hours (Data not shown here). Similarly, JQ1, a small-molecule inhibitor blocking BRD4 binding to active enhancer elements, leading to diminished the mRNA levels of HJURP (Fig. 5E, Supplementary Fig. 3B, C), and consistent with published datasets (GSE63584 and GSE65201) (Supplementary Fig. 3D). Knock-down of BRD4 also affected HJURP expression (Fig. 5F). In DMSO-treated BT-549 and MDA-MB-231 cells, ChIP-qPCR data showed the presence of the typical enhancer markers, H3K27ac at E1-E4 within the HJURP enhancer region After JQ1 (400 nM) or THZ1 (125 nM) treatment for 24 h, H3K27ac occupancy at the four randomly selected enhancer regions was examined. H3K27ac marks at all the four sites were noticeably reduced. The suppressive effect of THZ1 and JQ1 on HJURP expression partial contributed to the decreased abundance of H3K27ac mark on the enhancer chromatin (Supplementary Fig. 5E). The ability of enhancer to regulate luciferase activities was then tested. Two loci within enhancer region (pGL3-Enhancer_1 and pGL3-Enhancer_2) and a negative control named pGL3-basic outside enhancer site were constructed. We observed that the enhancer-transfected group showed significantly increased luciferase activity compared to the corresponding control, and the THZ1 or JQ1 treatment group presented decreased luciferase activity (Fig. 5G). Together, these results indicated that HJURP transcription was sensitive to transcriptional inhibition targeting specific enhancer.
We then employed CRISPR interference (CRISPRi) system by transgenic expression of doxycycline (Dox) inducible dCas9-KRAB, and single-guided RNAs (sgRNAs) were designed to target the active enhancer element (Fig. 5H, Supplementary Fig. 3F). RNA-guided epigenome editing reduced HJURP expression, showing that enhancer could positively regulate HJURP transcription (Fig. 5H). As expected, we observed that the reduction of the enrichment of H3K27ac signals within the enhancer genomic locus upon CRISPRi-mediated inactivation (Fig. 5I). Remarkedly, induced inactivation of the HJURP-enhancer led to a significant decrease in cell survival (Fig. 5J). These findings indicated that enhancer-mediated epigenetic programme contributed to HJURP activation in TNBC and is important for the malignant function of HJURP gene.
P53 regulated enhancer activity to shape HJURP transcription associated with E2F1 and FOXM1
We next tested whether P53 regulate HJURP expression in mutp53 or wtp53 breast cancer cells. Silencing of P53 using small interfering RNA (siRNA) promoted the expression of HJURP in wtp53 breast cancer cell lines (MCF-7 and ZR-75-1) (Fig. 6A). The small molecule Nutlin-3a inhibits the interaction between P53 and MDM2, to specifically restore P53 activity and represents an attractive approach for cancer therapy. To ensure proper pathway activation as well as RNA and protein accumulation of Nutlin-3a, P53 wild-type cancer cell lines MCF-7 and ZR-75-1 were cultured with Nutlin-3a or DMSO vehicle (untreated control) for 24 h treatment time. The expression level of MDM2, p21, and P53 was up-regulated in response to Nutlin-3a treatment. Accumulation of P53 protein resulted in decreased levels of endogenous HJURP protein in wtp53 breast cancer cells, but not in mutp53 MDA-MB-231 cells (Fig. 6B, C, Supplementary 4A). The changes in endogenous HJURP protein levels observed by immunoblotting following P53 expression correlate with the changes observed by RNA-seq for endogenous P53. P53-reactivating drugs RITA and 5-Fluorouracil (5-FU) induce its accumulation in tumor cells, with decreased expression of HJURP in breast cancer cells (Fig. 6D, E). We also assessed the effect of p53 status on HJURP gene expression in the HCT116 human colorectal cancer cell line. The HCT116 wild-type (wt), HCT116 p53+/+, and HCT116 p53-/- were employed, covering three different P53 variations. HJURP was found to be more strongly induced in HCT116 wt and HCT116 p53+/+ compared to other cell lines (Supplementary 4B). RNA-seq data presented that P53-reactivating drugs RITA induces HJURP accumulation in HCT116 parental cell but not in HCT116 P53-null cell (Supplementary 4C). We then measured enhancer luciferase activity in the presence and absence of wtP53. Induction of wtp53 (Nutlin-3a treatment group) repressed the − 5 kb HJURP enhancer activity in both MCF-7 and ZR-75-1 cells (Fig. 6F). ChIP-qPCR data showed the presence of typical enhancer markers in the HJURP enhancer region. After Nutlin-3a treatment at 20 nM for 48 h, the H3K27ac, BRD4, and RNA polymerase II occupancy at the four selected regions within the HJURP enhancer was examined, signals of active enhancer markers were noticeably decreased. These data suggested that Nutlin-3a treatment contributed to the disruption of the HJURP-enhancer in wtp53 breast cancer cells (Fig. 6G). To determine whether P53 directly binds to the HJURP enhancer to regulate gene transcription, we performed ChIP-qPCR in mutp53 and wtp53 breast cancer cells. However, no P53 occupancy at this regions was detected (Supplementary 4D). We applied a discovery P53BLD tool (p53 Binding Loci Database) for identifying the genome-wide binding loci of human P53. P53 ChIP-seq data of breast cancer cell lines (MCF-7, HCC70, BT-549 and MDA-MB-468) was re-mapped to visualize the regions with significant enrichment of P53 binding in the genome (Supplementary 4E). These two results demonstrated no physical association between the P53 and HJURP enhancer chromatin.
The initiation of individual enhancers activation generally requires multiple TFs binding to DNA response elements, which provide a flexible, robust and finely tuned mechanism for regulating gene expression programmes. To explore the underlying mechanism for p53-mediated transcription repression, we sought to identify P53 -interacting proteins. E2F1 acts upstream of P53, which links the Rb/E2F proliferation pathway and the P53 apoptosis pathway19. It has been proposed that P53 and E2F1 co-ordinately modulate FOXM1 transcription through an E2F-binding site located within the proximal promoter region20,21. To assess whether the E2F1 and FOXM1 are involved in the P53-dependent HJURP repression, we firstly analysed the expression pattern of E2F1 and FOXM1 with HJURP. A close positive correlation between the expression pattern of E2F1 and FOXM1 with HJURP was presented (Fig. 6H). We next determined the involvement of E2F1 and FOXM1 in P53-mediated accumulation of HJURP. Knock-down of E2F1 or FOXM1 significantly suppressed HJURP mRNA and protein expression in mutp53-containing cells (BT-549 and MDA-MB-231). Expression levels of CENP-A, MDM2 and p21 were also reduced upon E2F1 silencing (Fig. 6I - L). Besides, comparable decreased expression of HJURP as well as FOXM1-activated genes (CCNB1, CDC25B and PLK1) were detected in MDA-MB-231 cells treated with the FOXM1 inhibitor (FDI6) (Supplementary Fig. 4F). The expression of HJURP and other established transcriptional targets of FOXM1 or E2F1 were also down-regulated upon knockdown of E2F1 or FOXM1(Supplementary Fig. 4G). FDI6 as an inhibitor of FOXM1 binding, specifically down-regulated these FOXM1-activated genes with FOXM1 occupancy confirmed. The binding affinity of FOXM1 to the enhancer region (E1-4) of HJURP was suppressed after a 6 h treatment with the molecule (Supplementary Fig. 4H). Motif enrichment analyses of chromatin-defined enhancer sequences of HJURP identify the sequence-specific motifs (C/TAAACA) of the transcription factor FOXM1 as well as the binding motif of E2F1 (Supplementary Fig. 4I). To further explore the functional consequence of an interaction between 53, FOXM1 and E2F1, we detected the expression of E2F1 and FOXM1 with P53 silencing in MDA-MB-231 cells. Expression of E2F1 and FOXM1 were reduced after deletion of P53 (Fig. 6M). We then measured enhancer luciferase activity in the presence and absence of FOXM1 and E2F1. Suppression of E2F1 or FOXM1 repressed the HJURP enhancer activity (Fig. 6N). In TNBC cells harboring mutp53, E2F1 and FOXM1 were found to occupy the proximal enhancer region of HJURP, which could be inhibited by the depression of P53 (Fig. 6O). However, we didn’t detect an interaction of P53 with E2F1 and FOXM1 (Supplementary Fig. 4J). In summary, these observations supported a direct role of P53 in regulating HJURP transcription through association with E2F1 and FOXM1-modulated enhancer.
Increased HJURP expression correlated with the unfavorable prognostic effect of breast cancer patients
We next assessed the survival biomarkers value of HJURP by an online Kaplan-Meier Plotter, which incorporates microarray data with the prognosis information of breast cancer patients. Breast cancer patients were divided into two groups by auto-selection of best cut-off. There was a significant association of elevated HJURP mRNA level with unfavourable prognosis in patients with breast cancer. To determine whether HJURP expression status correlates with disease-free and overall survival, we analysed four independent breast cancer datasets, demonstrating that high HJURP mRNA level is a prognostic factor for poor survival in all datasets. (Fig. 7A, B).
To further examine whether the poor outcome associated with elevated HJURP mRNA levels depend on the promoting effect of HJURP in mutp53-subtype cancers, a cohort of 236 breast cancer patients from GEO dataset (GSE3494) with annotated P53 mutation status and information of clinical outcomes was used. We noted an inverse relationship between HJURP levels and outcomes of breast patients expressing wild-type, but not mutant, P53 (largely a consequence of the limited number of patients with mutp53 in these last groups). Significant associations were established between HJURP and overall survival in breast cancer patients carrying wtp53 (Fig. 7C). No significant prognostic relevance of HJURP expression levels with breast cancer patients with mutp53 was identified, which could be caused by the overall activation of HJURP in this subtype cancer. Segregating the breast cancer patient cohort into four major molecular subtypes (luminal A, luminal B, HER2-enriched and triple negative/basal-like), a significant association was identified between HJURP and clinical outcomes in luminal A/B, but not in basal or HER-2 subtypes. These results showed that increased HJURP mRNA levels are related to unfavourable prognosis, especially for tumors carrying wtp53 and luminal-breast cancer (Fig. 7D). Our findings identify HJURP acts as a molecular bypass that overcomes the anti-proliferative and pro-apoptotic effects exerted by wtp53 in breast cancer.