The majority of clinical deaths in patients with triple-negative breast cancer (TNBC) are due to chemoresistance and aggressive metastases, with high prevalence in younger women of African ethnicity. Although tumorigenic drivers are numerous and varied, the drivers of metastatic transition remain largely unknown. Here, we uncovered a molecular dependence of TNBC tumors on the TRIM37 network, which enables tumor cells to resist chemotherapeutic as well as metastatic stress. TRIM37-directed histone H2A monoubiquitination enforces changes in DNA repair that rendered TP53-mutant TNBC cells resistant to chemotherapy. Chemotherapeutic drugs triggered a positive feedback loop via ATM/E2F1/STAT signaling, amplifying the TRIM37 network in chemoresistant cancer cells. High expression of TRIM37 induced transcriptomic changes characteristic of a metastatic phenotype, and inhibition of TRIM37 substantially reduced the in vivo propensity of TNBC cells. Selective delivery of TRIM37-specific antisense oligonucleotides using antifolate receptor 1–conjugated nanoparticles in combination with chemotherapy suppressed lung metastasis in spontaneous metastatic murine models. Collectively, these findings establish TRIM37 as a clinically relevant target with opportunities for therapeutic intervention.

Significance:

TRIM37 drives aggressive TNBC biology by promoting resistance to chemotherapy and inducing a prometastatic transcriptional program; inhibition of TRIM37 increases chemotherapy efficacy and reduces metastasis risk in patients with TNBC.

Triple-negative breast cancer (TNBC) is an aggressive breast cancer subtype that accounts for approximately 20% of all breast cancer cases with the annual incidence rate in the United States estimated to be 40,000 (1). Patients with TNBC are disproportionately associated with the highest frequency of chemoresistance, relapse, and metastasis. Consequentially, the 5-year survival rate of TNBC is 77% relative to 93% for other breast cancer subtypes (1). Despite the high mortality rate in women worldwide, chemotherapy remains the standard of care for patients with TNBC. Although chemotherapy is effective initially in patients with TNBC, it is often accompanied by resistance, relapse, and severe side effects. Therefore, new and effective targeted therapies to prevent and ultimately cure TNBC are a clinical priority. Although lifestyle, epidemiologic, and cultural factors shape TNBC clinical outcome, the disease etiology is also dependent on biogeographical ancestry (2). Thus, lack of targeted therapies for TNBC is fraught with multiple challenges attributed to limited understanding of genetic complexities, metastatic biology, and drivers of metastatic traits.

An unresolved question in cancer biology is what drives a primary tumor to become metastatic? This is a clinically relevant question because metastatic, not primary, tumors are fatal. In general, numerous oncogenes and tumor suppressors are genetically or epigenetically altered in cancer and accumulate during tumorigenesis. But whether drivers of tumorigenesis are also the causal factor of the metastatic transition remains to be addressed. To this end, extensive transcriptomic and genetic scans of evolving carcinomas revealed mutations that were represented in premalignant biopsies but not in tumor biopsies, suggesting divergence of genetic alterations during the transition from primary to regional metastases (3, 4). In addition, dynamic epigenetic mechanisms are also intimately linked to metastatic transitions (5, 6). For example, TNBC tumors harbor a high frequency of hypermethylated promoters in commonly targetable drivers, such as TP53, BRAF, KRAS, and EGFR (7). Alterations in epigenetic factors causing neomorphic mutations (e.g., EZH2, DNMT3A) or translocations (e.g., NSD2, MMSET) are also frequent in patients with cancer (6). As such, several small molecules targeting epigenetic regulators have entered clinical trials, for example, Estinosat, Belinostat, and Panobinostat (8). Understandably, these drug treatments are not mutation-specific and thus pose a significant toxicity risk to untransformed cells, underscoring the critical need for targeted therapies.

We have originally described tripartite motif-containing protein 37 (TRIM37) as a breast cancer oncoprotein that can epigenetically silence tumor suppressors (9, 10). Clinically, high-TRIM37 associate with poor overall survival (9). Mechanistically, TRIM37 monoubiquitinates histone H2A at Lys119 (H2Aub) to downregulate target genes (9). Functionally, TRIM37 overexpression renders nontransformed breast cells tumorigenic, and inhibition of TRIM37 function reduces tumor growth (9). Although TRIM37 promotes tumorigenesis, its function in breast cancer metastasis and the therapeutic implications of TRIM37 targeting remain to be demonstrated.

Metastasis is a multistep process that includes pathways regulating the epithelial–mesenchymal transition, infiltration of distant sites, and metastatic growth (11). Given the majority of patients with TNBC receive chemotherapy, the ability to resist therapy-induced cell death is perhaps the first step toward a metastatic phenotype. Indeed, recent evidence revealed synchronized expression of genes involved in surviving the stress of chemotherapy as well as overcoming the natural barriers of metastatic growth. For example, a CXCL1/2 paracrine pathway (12) and MTDH (13) were recently identified with dual functionality in metastasis and chemoresistance. Here, we uncover that TRIM37 alters DNA damage response to prevent therapy-induced cell death, and enforces a transcriptional program favoring metastasis. In particular, we report that selective TRIM37 inhibition in TNBC tumors suppresses lung metastases in vivo. Together, these data reveal that TRIM37 is a new epigenetic driver of aggressive TNBC biology, which can be targeted to simultaneously increase chemotherapy efficacy and reduce metastasis risk in patients with TNBC.

Cell lines and cell culture

MDA-MB-231-luc-D3H2LN-BMD2b (231-b2; provided by Takahiro Ochiya), MDA-MB-231, MDA-MB-468, and HCC1806 (provided by Michael J. Lee) cells were maintained in RPMI medium supplemented with 10% FBS (Invitrogen) at 37°C and 5% CO2. MCF10A, MCF7, MCF10AT, HCC1806RR (provided by Sophia Ran), and TP53−/− MCF10A (provided by David Weber and Michele Vitolo) were cultured as described previously (9, 14, 15). Cells cultured at the same time were pooled together and then seeded after counting in a 6-well or 10-cm dish. Cells were then subjected, in a random order, to treatment with a control or test different biologics, which included short hairpin RNA (shRNA), sequence-specific guide RNA (sgRNA), vectors, and small-molecule inhibitors. Cells were routinely tested for mycoplasma using PlasmoTest kit from Invivogen.

Animal care

NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ and Balb/cJ (Jackson Laboratory) were housed in a specific pathogen-free facility accredited by the American Association of Laboratory Animal Care. All animal studies were approved (#4112 and #4222) by the Institutional Animal Care and Use Committee.

CRISPR/Cas9 targeting

The Gapdh sgRNAs were cloned in pLentiCrispr v2 plasmid (Addgene) and packaged into virus as recommended by the manufacturer. Cells were infected with packaged virus as described previously (9).

Recombinant antibody cloning

Farletuzumab (anti-human FOLR1) and LK26 (anti-mouse FOLR1) antibodies were cloned, engineered, expressed, and purified as described previously (16). To generate an antibody conjugate platform for nanoparticles linkage, a novel linker sequence [(X)3Cys(X)3 amino acid sequence] was engineered in continuation of the carboxy terminal of heavy chain (called Fc-Linkered). The knob chain is exactly similar to hole chain except for the presence of this unique cysteine residue.

Recombinant antibody expression

Free style CHO-S cells (Invitrogen) were cultured and maintained according to supplier's recommendations (Life Technologies). A ratio of 2:1 (light chain, VL: heavy chain, VH) DNA was transfected using 1 mg/mL polyethylenimine and cultured at 37°C. After 24 hours, transfected cells were cultured at 32°C for additional 9 days. Cells were fed every second day with 1:1 ratio of Tryptone feed and CHO Feed B, and antibodies were purified as described previously (16). An Autodesk Inventor Professional 2020 was used to draw the design of antibody.

Nanoparticle synthesis and packaging

Cholesterol, 1,2-dioleoyl-3-trimethylammonium-propane chloride salt (DOTAP), 1,2-dioleoyl-sn-glycero-3-phosphate (DOPA), and 1,2-distearoryl-sn-glycero-3-phosphoethanolamine-N-[maleimide-(polyethyleneglycol-2000)] ammonium salt (DSPE-PEG2000-Maleimide) were purchased from Avanti Polar Lipids, Inc.

TRIM37-ASO (IDT) and control-ASO (IDT) loaded bilayer nanoparticles were prepared as previously described with modification (17). Briefly, 150 mmol/L CaCl2 with 0.5 mmol/L ASO were dispersed in Cyclohexane/Igepal CO-520 (70:30 v/v) solution to form water-in-oil reverse microemulsion (Solution A). The phosphate phase was prepared by mixing 1.5 μmol/L NaHPO4 (pH = 9.0) and 6.9 μmol/L DOPA in Cyclohexane/Igepal CO-520 (70:30 v:v) solution (Solution B). Calcium phosphate-alginate (CaP) core was prepared by mixing Solutions A and B. For assembly of outer leaflet, CaP core was mixed with 0.5 μmol/L DOTAP/Cholesterol (1:1) and 0.15 μmol/L DSPE-PEG-2000-Maleimide. CaP-bilipid nanoparticles were mixed with 0.56 μmol/L engineered farletuzumab solutions and incubated overnight at 4°C. Finally, CaP-bilipid nanoparticles were sterile filtered for subsequent experiments. An Autodesk Inventor Professional 2020 was used to draw the design of smart nanoparticle.

NSG tumor xenograft studies

For indicated cell lines, weight and aged matched female NSG mice were injected s.c. with 2 × 106 TNBC cells in their right flank as described previously (9). For in vivo Dox-induced TRIM37 upregulation studies, mice bearing approximately 200 mm3 HCC1806 tumors weight matched animals were randomly assigned into groups and injected with 2 mg/kg Dox. Tumors were harvested 24 hours following Dox treatment. For smart nanoparticles' efficiency in vivo, mice bearing approximately 200 mm3 231–2b tumors weight matched animals were randomly assigned into groups and injected with control or smart nanoparticles. At the endpoint, RNA and protein lysates were prepared from isolated tumors.

Bioluminescent imaging

One hundred and fifty mg/kg luciferin (Perkin Elmer Inc.) was administered to mice intraperitoneally, and mice were imaged as described previously (16).

Spontaneous metastatic tumor studies

4T1 or 231–2b cells (2 × 106) were injected into the inguinal mammary fat pad or right flank of 6- to 8-week-old female Balb/cJ or NSG mice, and tumor growth was monitored as previously described (9). Where indicated, primary tumors were resected and animals were allowed to develop lung metastases. Mice bearing TNBC tumors were weight matched and randomly assigned into groups that received 1.2 mg/kg smart or control nanoparticles at the indicated times. The lung metastases in the animals were monitored by bioluminescent imaging (BLI). At the termination of the experiment, animals were euthanized and indicated tissues were harvested and processed for histologic examination and IHC staining or for qRT-PCR analysis. Metastatic burden was calculated either as the number of visible metastatic lesion in each organ or as the relative luminescence signal from gross organ tissue.

Experimental metastasis in vivo

231–2b cells (3 × 105) were inoculated directly into the left cardiac ventricle. Metastatic growth was monitored using BLI. Lungs, liver, femurs, and brains were harvested postmortem and processed for histologic examination and gross analysis.

Bioinformatic analysis

cBioPortal was used to obtain The Cancer Genome Atlas expression z-scores for genes in Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) cohort (18). FireBrowse (Broad Institute) was used to obtain normal breast tissue expression z-scores. Cooccurrence and mutual exclusivity analysis were performed with cBioPortal. Hazard ratios were assessed with Cox proportional hazard model for which TNBC and non-TNBC patients were stratified into high- and low TRIM37 as third and first quartiles. For Kaplan–Meier analysis, the P value was calculated with log-rank test. For boxplot analysis, TRIM37-regulated genes were stratified according to TRIM37 expression using z score thresholds (z > 0.5, z ≤ 0.5).

Statistical analysis

All experiments were performed at least in triplicate, and the results presented are the mean of at least three different biological replicates. The comparisons between the two groups were done by unpaired t test; comparisons between multiple treatment groups were done by one-way or two-way ANOVA with indicated multiple comparisons post hoc tests. The enrichment of genes positively correlating with TRIM37 was calculated with two-tailed Fisher exact test. The distributions in correlation between GO terms were calculated by Kolmogorov–Smirnov test. All statistical analyses were performed using R/Bioconductor (version 2.15.2).

TRIM37 associates with double-strand break repair machinery in TNBC

Several observations provided a rationale to explore the potential role of TRIM37 in promoting resistance to chemotherapeutic drugs that induce high level of DNA damage. TRIM37 catalyzes monoubiquitination of H2A (9), a chromatin modification enriched at transcriptionally repressed gene promoters (19) as well as at DNA damage sites (20). Furthermore, our analysis of the METABRIC cohorts revealed TRIM37 association with DNA repair genes (Fig. 1A; Supplementary Table SI), which was significantly stronger than genes involved in cellular proliferation (Supplementary Fig. S1A; Supplementary Table SI). Double-strand break (DSB) repair is one of the major pathways to repair damaged DNA in cancer cells, and its kinetics predicts resistance to therapy (21, 22). We therefore limited the analysis to repair proteins that participate in homologous recombination (HR) and nonhomologous end-joining repair pathways (NHEJ), the two subtypes of DSB repair. RAD51C, XRCC5 (Ku80), RNF8, XRCC6BP1 (Ku70 binding protein), RNF168, and MRE11 were among the DSB genes whose expression most strongly associated with TRIM37 (Fig. 1A). RAD51-associated proteins generally promote HR repair (23). XRCC5 and XRCC6BP1 are involved in NHEJ repair pathway (23). MRE11 forms complex with NBS1 and RAD50 to regulate response at DSB (24). RNF8 and RNF168 are E3 ligases that promote binding of DSB genes (23). In addition, TRIM37 also correlated with the overexpression of a family of other DSB factors, including RAD51AP1, SFR1, DDX1, RAD51, ERCC1, and CHEK2. Together, these findings identified 28 DSB genes that significantly correlated with TRIM37 in patients with breast cancer (Pearson's coefficient > 0.2; Supplementary Table SI).

Figure 1.

TRIM37 associates with DSB repair proteins in TNBC. A, Gene ranking of DNA repair genes according to cooccurrence with high TRIM37 in patients with breast cancer. The inset shows genes that have the highest correlation with TRIM37. A complete list of DNA repair genes that correlate with TRIM37 is presented in Supplementary Table SI. B, Heat map for the expression of TRIM37 and 28 DSB repair genes in patients with breast cancer stratified by TNBC (n = 299), non-TNBC (n = 1605), and NAT (n = 100). n, number of samples. C and D, Forest plot of HR in TNBC (C) and non-TNBC (D) patients stratified for high- and low-TRIM37 expression using METABRIC cohorts. E, Immunoblot monitoring of XRCC5, XRCC6, NBS1, RAD51C, and TRIM37 in protein complexes pulled down by either anti-TRIM37 (left), the indicated DSB proteins (right), or an IgG control. Input, ∼1%–5% of whole cell lysates. F, ChIP monitoring TRIM37 and H2Aub binding at Gapdh in MDA-MB-468 cells expressing either Cas9 alone or with Gapdh site-specific sgRNA. G, ChIP monitoring TRIM37 and BMI1 binding at FRA3B, HOXA3, and Actin in MDA-MB-468 cells treated with Dox. H, ChIP monitoring H2Aub, XRCC5, XRCC6, NBS1, and RAD51c binding at FRA3B and Actin in MDA-MB-468 cells expressing a nonsilencer (NS) or TRIM37 shRNA. I, HR (left) and NHEJ (right)-mediated DSB-repair activity in MDA-MB-468 cells expressing control or TRIM37 shRNA (#1, #2). Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Figure 1.

TRIM37 associates with DSB repair proteins in TNBC. A, Gene ranking of DNA repair genes according to cooccurrence with high TRIM37 in patients with breast cancer. The inset shows genes that have the highest correlation with TRIM37. A complete list of DNA repair genes that correlate with TRIM37 is presented in Supplementary Table SI. B, Heat map for the expression of TRIM37 and 28 DSB repair genes in patients with breast cancer stratified by TNBC (n = 299), non-TNBC (n = 1605), and NAT (n = 100). n, number of samples. C and D, Forest plot of HR in TNBC (C) and non-TNBC (D) patients stratified for high- and low-TRIM37 expression using METABRIC cohorts. E, Immunoblot monitoring of XRCC5, XRCC6, NBS1, RAD51C, and TRIM37 in protein complexes pulled down by either anti-TRIM37 (left), the indicated DSB proteins (right), or an IgG control. Input, ∼1%–5% of whole cell lysates. F, ChIP monitoring TRIM37 and H2Aub binding at Gapdh in MDA-MB-468 cells expressing either Cas9 alone or with Gapdh site-specific sgRNA. G, ChIP monitoring TRIM37 and BMI1 binding at FRA3B, HOXA3, and Actin in MDA-MB-468 cells treated with Dox. H, ChIP monitoring H2Aub, XRCC5, XRCC6, NBS1, and RAD51c binding at FRA3B and Actin in MDA-MB-468 cells expressing a nonsilencer (NS) or TRIM37 shRNA. I, HR (left) and NHEJ (right)-mediated DSB-repair activity in MDA-MB-468 cells expressing control or TRIM37 shRNA (#1, #2). Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Close modal

The analysis of METABRIC cohorts stratified by breast cancer subtypes revealed statistically significant correlation between TRIM37 and DSB genes in patients with TNBC but not in non-TNBC patients or histologically normal breast tissue adjacent to tumor (NAT, Fig. 1B; Supplementary Fig. S1B). As shown in Fig. 1C, the hazard ratio was approximately 2.3-fold higher for patients with TNBC with high- compared with low-TRIM37 expression, linking TRIM37 to poor prognosis in patients with TNBC. In contrast, no significant association between TRIM37 levels and survival was observed in non-TNBC patients (Fig. 1D). Likewise, higher expression of a subset of DSB genes analyzed predicted poor overall survival in patients with TNBC as determined by HR > ∼1 (Supplementary Fig. S1C).

We next performed a series of functional experiments to determine the molecular mechanisms underlying the biological activity of TRIM37 in DSB repair using several human TNBC cell lines (HCC1806, MDA-MB-468, and MDA-MB-231; Supplementary Fig. S1D). We first asked whether TRIM37 was physically associated with DSB genes. To test this idea, HCC1806 whole cell extract was fractionated by sucrose gradient sedimentation, and individual fractions were analyzed for TRIM37 and a representative subset of DSB proteins identified in Fig. 1A. Results shown in Supplementary Fig. S1E demonstrated that TRIM37 cosediments with XRCC5, XRCC6, RAD51c, MRE11, and NBS1. Physical interactions between TRIM37 and DSB proteins could be confirmed by coimmunoprecipitation (Fig. 1E; Supplementary Fig. S1F) and immunofluorescence analysis (Supplementary Fig. S1G) in MDA-MB-468 cells treated with doxorubicin (Dox), a first-line chemotherapeutic agent. No significant changes in the expression of DSB genes were observed in HCC1806 cells expressing TRIM37-specific shRNA, excluding TRIM37-mediated transcriptional regulation of these DSB genes (Supplementary Fig. S1H and S1I; Supplementary Table SII).

Prompted by these findings, we interrogated TRIM37 recruitment to DSB using two independent experimental systems. We first enzymatically induced DSB at Gapdh using sgRNA to promote Cas9-nuclease binding. Chromatin immunoprecipitation (ChIP) assay confirmed TRIM37 as well as H2Aub enrichment at the DSB in Gapdh (Fig. 1F). Next, we analyzed recruitment of TRIM37 and BMI1, a component of the polycomb complex that participates in DSB repair (25), to the endogenous fragile site, FRA3B. Consistently, ChIP analysis confirmed TRIM37 and BMI1 binding to FRA3B following Dox treatment (Fig. 1G). In marked contrast to the control cells, knockdown of TRIM37 substantially decreased H2Aub enrichment as well as DSB proteins binding to FRA3B in Dox-treated MDA-MB-468 cells (Fig. 1H). As expected, repair efficiencies of NHEJ and HR pathways were significantly reduced in TRIM37-knockdown cells compared with control cells (Fig. 1I). Together, these results demonstrate that TRIM37 interacts with DSB repair factors and functionally contributes to the repair of therapy-induced DNA damage.

TRIM37-catalyzed H2Aub is required for its function in chemoresistance

We next directly examined the impact of TRIM37 knockdown on chemotherapy-induced DNA damage and clonogenic growth. Knockdown of TRIM37 resulted in approximately 6-fold increase in median tail length in comet assay (Fig. 2A; Supplementary Figs. S1H and S2A) and approximately 5-fold increase in the median nuclear coverage of phosphorylated histone H2AX (γH2AX, Fig. 2B) in Dox-treated cells. TRIM37 knockdown in TNBC cell lines also markedly increased caspase-3 activity and PARP cleavage (Fig. 2C), hallmarks of cell death. By contrast, Dox-treated MCF7, a hormone receptor–positive breast cancer cell line, did not augment PARP cleavage or caspase-3 activity (Fig. 2C), suggesting TRIM37 function in chemoresistance was limited to TNBC cells. As expected, knockdown of TRIM37 sensitized MDA-MB-468 cells to chemotherapeutic stress without affecting proliferation as indicated by substantially decreased clonogenic growth relative to the control cells (Fig. 2D; Supplementary Fig. S2B).

Figure 2.

TRIM37-catalyzed H2Aub is required for chemoresistance in TNBC. A, Top, tail moment in DMSO or Dox-treated MDA-MB-468 cells expressing nonsilencer (NS) or TRIM37 shRNA (#1, #2). Bottom, representative images of the tails for each group are shown. Scale bars, 100 μm. B, Top, quantification of γH2AX foci in MDA-MB-468 cells expressing nonsilencer or TRIM37 shRNA (#1, #2) following treatment with Dox. Bottom, representative immunofluorescence images of γH2AX foci (green) in Dox-treated MDA-MB-468 cells expressing nonsilencer or TRIM37 shRNA (#1, #2). DAPI (blue) stains the nucleus. Scale bars, 50 μm. C, Caspase-3 activity assay (top) and immunoblot for PARP and cleaved PARP (c-PARP; middle) in DMSO or Dox-treated MDA-MB-468, MDA-MB-231, and MCF7 cells expressing nonsilencer or TRIM37 shRNA (#1, #2). Bottom, quantification of PARP cleavage relative to total PARP is shown. D, Quantification of the fold change in chemotherapeutic drug-resistant colonies obtained for MDA-MB-468 cells expressing either nonsilencer or TRIM37 shRNA (#1, #2) by a clonogenic assay. Cells were treated with DMSO, Dox, temozolomide (TMZ), etoposide (EPO), daunorubicin (Daun), and cisplatin (CPN). Results were normalized to the colony-forming unit (cfu) for DMSO. E, ChIP monitoring TRIM37 and H2Aub binding at FAR3B and Actin in MCF10AT cells expressing vector control (VC), TRIM37, or mutant TRIM37 [TRIM37(C18R)]. F, Top, tail moment in DMSO or Dox-treated VC-, TRIM37-, or TRIM37(C18R)-expressing MCF10AT cells. Bottom, representative images of the tail moment for each group are shown. Scale bars, 100 μm. G and H, Caspase-3 activity assay (G) and immunoblot for PARP and c-PARP (H, left) in DMSO or Dox-treated VC, TRIM37, or TRIM37(C18R)-expressing MCF10AT cells. Right, quantification of PARP cleavage relative to total PARP is shown. I, Quantification of the fold change in chemotherapeutic drug resistant colonies obtained for VC, TRIM37, or TRIM37(C18R) by clonogenic assay. Cells were treated with DMSO, Dox, temozolomide, etoposide, daunorubicin, and cisplatin. Results were normalized to the cfu for DMSO. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 2.

TRIM37-catalyzed H2Aub is required for chemoresistance in TNBC. A, Top, tail moment in DMSO or Dox-treated MDA-MB-468 cells expressing nonsilencer (NS) or TRIM37 shRNA (#1, #2). Bottom, representative images of the tails for each group are shown. Scale bars, 100 μm. B, Top, quantification of γH2AX foci in MDA-MB-468 cells expressing nonsilencer or TRIM37 shRNA (#1, #2) following treatment with Dox. Bottom, representative immunofluorescence images of γH2AX foci (green) in Dox-treated MDA-MB-468 cells expressing nonsilencer or TRIM37 shRNA (#1, #2). DAPI (blue) stains the nucleus. Scale bars, 50 μm. C, Caspase-3 activity assay (top) and immunoblot for PARP and cleaved PARP (c-PARP; middle) in DMSO or Dox-treated MDA-MB-468, MDA-MB-231, and MCF7 cells expressing nonsilencer or TRIM37 shRNA (#1, #2). Bottom, quantification of PARP cleavage relative to total PARP is shown. D, Quantification of the fold change in chemotherapeutic drug-resistant colonies obtained for MDA-MB-468 cells expressing either nonsilencer or TRIM37 shRNA (#1, #2) by a clonogenic assay. Cells were treated with DMSO, Dox, temozolomide (TMZ), etoposide (EPO), daunorubicin (Daun), and cisplatin (CPN). Results were normalized to the colony-forming unit (cfu) for DMSO. E, ChIP monitoring TRIM37 and H2Aub binding at FAR3B and Actin in MCF10AT cells expressing vector control (VC), TRIM37, or mutant TRIM37 [TRIM37(C18R)]. F, Top, tail moment in DMSO or Dox-treated VC-, TRIM37-, or TRIM37(C18R)-expressing MCF10AT cells. Bottom, representative images of the tail moment for each group are shown. Scale bars, 100 μm. G and H, Caspase-3 activity assay (G) and immunoblot for PARP and c-PARP (H, left) in DMSO or Dox-treated VC, TRIM37, or TRIM37(C18R)-expressing MCF10AT cells. Right, quantification of PARP cleavage relative to total PARP is shown. I, Quantification of the fold change in chemotherapeutic drug resistant colonies obtained for VC, TRIM37, or TRIM37(C18R) by clonogenic assay. Cells were treated with DMSO, Dox, temozolomide, etoposide, daunorubicin, and cisplatin. Results were normalized to the cfu for DMSO. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Given ubiquitin is critical for DSB factor recruitment to damaged DNA (26), we next asked whether TRIM37-catalyzed H2Aub is required for its function in chemoresistance. To test this idea, we ectopically expressed either wild-type TRIM37 (TRIM37) or catalytically-dead TRIM37 [TRIM37(C18R)] in MCF10AT, a premalignant K-RAS transformed triple-negative breast cell line (Supplementary Fig. S2C). Although Dox-treatment induced significant enrichment of H2Aub at FRA3B in TRIM37 overexpressing cells, TRIM37(C18R) failed to promote H2Aub enrichment at FRA3B (Fig. 2E). Consequentially, Dox treatment of TRIM37(C18R)-expressing cells caused substantially longer comet tails (Fig. 2F) as well as increased caspase-3 activity (Fig. 2G) and PARP cleavage (Fig. 2H) relative to TRIM37-expressing cells. Finally, substantially fewer colonies were observed for cells expressing TRIM37(C18R) compared with TRIM37 following chemotherapeutic stress (Fig. 2I; Supplementary Fig. S2D).

TRIM37 promotes chemoresistance in the absence of functional p53

Wild-type TP53 represents a barrier to chemoresistance by altering DSB repair responses, activating checkpoints, and the stress responses (27). Strikingly, TNBC tumors frequently harbor disrupting TP53 mutations (Supplementary Fig. S3A), which primarily cause the loss of its wild-type function (28). Surprisingly, analysis of representative TRIM37-associated DSB genes showed a striking correlation between TRIM37 and DSB genes expression in patients with TNBC carrying mutant TP53 but not in wild-type TP53 (Fig. 3A; Supplementary Fig. S3B–S3D). Consistently, TP53 mutant, but not wild-type, patients with TNBC with high TRIM37 were at approximately 2.3-fold higher risk of death relative to low TRIM37 (Fig. 3B and C).

Figure 3.

TRIM37 reduces cytotoxicity of chemotherapy in the absence of functional p53. A, Heat map for expression of TRIM37 and DSB genes in patients with TNBC stratified by TP53 status [wild-type (n = 238) and mutant (n = 61)]. n, number of patients. B and C, Forest plot of HR for patients with TNBC with mutant TP53 (B), or wild-type TP53 (C) stratified for high- and low-TRIM37 expression. D, Immunoblots in Dox-treated MDA-MB-468 and HCC1806 cells expressing nonsilencer (NS), TRIM37 shRNA, or TP53. Tubulin was the loading control. Bottom, quantification of c-PARP relative to total PARP. E, Caspase-3 activity assay in Dox-treated MDA-MB-468 and HCC1806 expressing nonsilencer, TRIM37 shRNA, or TP53. F, Quantification of the fold change in colony-forming unit (cfu) for Dox-treated MDA-MB-468 and HCC1806 cells expressing nonsilencer, TRIM37 shRNA, or TP53 by clonogenic assay. G, Immunoblots in Dox-treated p53−/− MCF10A cells expressing TRIM37. Tubulin was the loading control. Right, quantification of c-PARP relative to total PARP. H, Caspase-3 activity assay in Dox-treated p53−/− MCF10A cells expressing either vector control or TRIM37. I, Quantification of the fold change in cfu for p53−/− MCF10A cells expressing TRIM37 plated after Dox treatment by clonogenic assay. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Figure 3.

TRIM37 reduces cytotoxicity of chemotherapy in the absence of functional p53. A, Heat map for expression of TRIM37 and DSB genes in patients with TNBC stratified by TP53 status [wild-type (n = 238) and mutant (n = 61)]. n, number of patients. B and C, Forest plot of HR for patients with TNBC with mutant TP53 (B), or wild-type TP53 (C) stratified for high- and low-TRIM37 expression. D, Immunoblots in Dox-treated MDA-MB-468 and HCC1806 cells expressing nonsilencer (NS), TRIM37 shRNA, or TP53. Tubulin was the loading control. Bottom, quantification of c-PARP relative to total PARP. E, Caspase-3 activity assay in Dox-treated MDA-MB-468 and HCC1806 expressing nonsilencer, TRIM37 shRNA, or TP53. F, Quantification of the fold change in colony-forming unit (cfu) for Dox-treated MDA-MB-468 and HCC1806 cells expressing nonsilencer, TRIM37 shRNA, or TP53 by clonogenic assay. G, Immunoblots in Dox-treated p53−/− MCF10A cells expressing TRIM37. Tubulin was the loading control. Right, quantification of c-PARP relative to total PARP. H, Caspase-3 activity assay in Dox-treated p53−/− MCF10A cells expressing either vector control or TRIM37. I, Quantification of the fold change in cfu for p53−/− MCF10A cells expressing TRIM37 plated after Dox treatment by clonogenic assay. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Close modal

To investigate the relationship between p53 and TRIM37, we transiently expressed wild-type p53 in MDA-MB-468 (carries transcriptionally inactive p53 R273H) and HCC1806 (carries p53 T256Kfs*90 deletion) cells. For each cell line, p53-reconstituted TNBC cells showed significantly higher PARP cleavage (Fig. 3D) and caspase-3 activity (Fig. 3E) compared with the control cells following Dox treatment. A clonogenic assay confirmed that p53 overexpression sensitized MDA-MB-468 (∼4-fold) and HCC1806 (∼2-fold) cells to Dox despite high levels of TRIM37 (Fig. 3F; Supplementary Fig. S3E). Reciprocally, ectopic expression of TRIM37 in genetically ablated p53-null (p53-/-) MCF10A cells (14) substantially reduced PARP cleavage and caspase-3 activity relative to empty vector (Fig. 3G and H). As expected, TRIM37-expressing p53-/- MCF10A showed an approximately 4-fold increase in colony formation in comparison with control cells (Fig. 3I; Supplementary Fig. S3F). In summary, consistent with previous results for MDM2, KRAS, and ARID1 A (29), we find that TRIM37 requires loss of p53 to drive the chemoresistant phenotype in TNBC cells.

Chemotherapy amplifies a TRIM37 survival axis in TNBC

Most patients with TNBC receive chemotherapy, which is effective in early stages of the disease, but approximately 30% to 50% patients develop resistance (1). Although the exact mechanisms of chemoresistance remain to be understood, chemotherapeutic drugs often induce genomic and transcriptomic reprogramming of resistant signatures (30), including alterations in DNA repair capacity (31). Previous studies have suggested that accumulation of such changes accompanies selection and expansion of resistant TNBC cells (32). We therefore analyzed the expression of TRIM37 in MDA-MB-468, HCC1806, and MDA-MB-231 cells following chemotherapy. Surprisingly, we found that TRIM37 is upregulated in all the three TNBC cell lines tested in a time-dependent manner (Fig. 4A). In contrast, no significant increase in TRIM37 was observed in p53 wild-type MCF7 or MCF10A, an immortalized breast epithelial cell (Fig. 4A). The analysis of TRIM37 upregulation kinetics in MCF7 and MCF10a revealed quick and robust p53 activation following Dox treatment (Supplementary Fig. S4A), supporting our previous findings that p53 overrides TRIM37 function in chemoresistance (Fig. 3). Indeed, ectopic expression of p53 in MDA-MB-468 cells obliterated Dox-induced TRIM37 upregulation (Supplementary Fig. S4B). As expected, TNBC cells treated with additional chemotherapeutic drugs also increased TRIM37 expression after treatment (Fig. 4B). Finally, increased TRIM37 level in xenograft tumors following dox treatment revealed therapy-induced transcriptional upregulation of TRIM37 in vivo (Fig. 4C and D).

Figure. 4.

Chemotherapy amplifies oncogenic TRIM37 network in TNBC. A, Top, immunoblots in MDA-MB-468, HCC1806, MDA-MB-231, MCF7, and MCF10A cells treated with Dox for 0, 24, 48, and 72 hours. Tubulin was the loading control. Bottom, quantification of TRIM37 relative to tubulin. B, Immunoblot monitoring TRIM37 in HCC1806 (top) or MDA-MB-231 (bottom) cells treated with daunorubicin (Daun), cisplatin (CPN), etoposide (EPO), or temozolomide (TMZ). Tubulin was the loading control. Right, quantification of TRIM37 relative to tubulin. C, Schematic showing that NSG mice were treated with i.p. injection of 2 mg/kg Dox once tumor reached the size of approximately 200 mm3. Tumors were harvested postmortem. D, qRT-PCR monitoring TRIM37 expression in HCC1806 subcutaneous tumors derived from NSG mice following treatment with Dox. n = 6 animals per group. E, Schematic of ATM signaling with the downstream effectors of ATM, E2F1, and STAT. Specific small-molecule inhibitors of the ATM signaling are also indicated (red). F, qRT-PCR monitoring TRIM37 expression in MDA-MB-468 cells following treatment with KU555933, HLM006474, or AG490 in combination with Dox. G–I, ChIP analysis monitoring the binding of E2F1 (G), STAT1 (H), and STAT3 (I) to TRIM37 and Actin in MDA-MB-468 cells treated with HLM006474 or AG490 in combination with Dox. J, Top, immunoblots in MDA-MB-468 cells treated with Dox and the indicated inhibitors of ATM signaling. Tubulin was the loading control. Bottom, quantification of TRIM37 relative to tubulin. K, Heat map for TRIM37 expression in mutant and wild-type TP53 TNBC tumor tissue samples pre- and post-chemotherapy. The type of TP53 mutation is indicated on the right. NA, not available (n = 17). Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Figure. 4.

Chemotherapy amplifies oncogenic TRIM37 network in TNBC. A, Top, immunoblots in MDA-MB-468, HCC1806, MDA-MB-231, MCF7, and MCF10A cells treated with Dox for 0, 24, 48, and 72 hours. Tubulin was the loading control. Bottom, quantification of TRIM37 relative to tubulin. B, Immunoblot monitoring TRIM37 in HCC1806 (top) or MDA-MB-231 (bottom) cells treated with daunorubicin (Daun), cisplatin (CPN), etoposide (EPO), or temozolomide (TMZ). Tubulin was the loading control. Right, quantification of TRIM37 relative to tubulin. C, Schematic showing that NSG mice were treated with i.p. injection of 2 mg/kg Dox once tumor reached the size of approximately 200 mm3. Tumors were harvested postmortem. D, qRT-PCR monitoring TRIM37 expression in HCC1806 subcutaneous tumors derived from NSG mice following treatment with Dox. n = 6 animals per group. E, Schematic of ATM signaling with the downstream effectors of ATM, E2F1, and STAT. Specific small-molecule inhibitors of the ATM signaling are also indicated (red). F, qRT-PCR monitoring TRIM37 expression in MDA-MB-468 cells following treatment with KU555933, HLM006474, or AG490 in combination with Dox. G–I, ChIP analysis monitoring the binding of E2F1 (G), STAT1 (H), and STAT3 (I) to TRIM37 and Actin in MDA-MB-468 cells treated with HLM006474 or AG490 in combination with Dox. J, Top, immunoblots in MDA-MB-468 cells treated with Dox and the indicated inhibitors of ATM signaling. Tubulin was the loading control. Bottom, quantification of TRIM37 relative to tubulin. K, Heat map for TRIM37 expression in mutant and wild-type TP53 TNBC tumor tissue samples pre- and post-chemotherapy. The type of TP53 mutation is indicated on the right. NA, not available (n = 17). Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Close modal

We next sought to determine the mechanistic basis for chemotherapy-induced burst in TRIM37 levels in TNBC tumors. A previous study identified TRIM37 association with ataxia-telangiectasia-mutated (ATM) kinase, a DNA damage sensor (33). Moreover, TRIM37 promoter harbors regulatory elements for STAT and E2F1, downstream effectors of ATM kinase (Supplementary Fig. S4C–S4E). We therefore investigated the potential role of ATM signaling in the transcriptional regulation of TRIM37 by utilizing small-molecule inhibitors of either ATM or its downstream effectors, E2F1 and JAK (Fig. 4E; Supplementary Fig. S4F–S4J). JAK is a tyrosine kinase that phosphorylates STAT. qRT-PCR analysis revealed that pharmacologic inhibition of ATM or its downstream effectors blocks TRIM37 upregulation following Dox treatment (Fig. 4F). Similarly, E2F1- and STAT1/3 knockdown abolished Dox-induced TRIM37 upregulation (Supplementary Fig. S4K–S4L), whereas their overexpression significantly increased TRIM37 levels (Supplementary Fig. S4M and S4N). ChIP analysis confirmed Dox-induced STAT1, STAT3, and E2F1 recruitment to TRIM37, which substantially decreased following pharmacologic inhibition of JAK or E2F1 activation (Fig. 4GI). Consequently, inhibition of ATM signaling in MDA-MB-468 cells induced significantly higher DNA damage and cell death relative to control cells as determined by γH2AX and PARP cleavage (Fig. 4J).

Finally, to clinically validate chemotherapy-induced burst in TRIM37, we analyzed TRIM37 expression in a panel of matched pre- and postneoadjuvant chemotherapy-treated tumor biopsies from patients with TNBC (see also Materials and Methods). qRT-PCR analysis showed that chemotherapy treatment increased TRIM37 expression in approximately 82% of TNBC tumors carrying mutation in TP53 (n = 11, Fig. 4K). Consistent with TNBC cellular models, no significant change in TRIM37 was observed in TP53 wild-type TNBC tumors (n = 6; Fig. 4K). Collectively, our results show that chemotherapeutic stress increases TRIM37 expression in TNBC tumors in an ATM-dependent manner.

TRIM37 remodels transcriptional program favoring TNBC metastasis

Although the emergence of chemoresistance is closely related to metastasis, the ability of a cancer cell to survive and proliferate under continuous standard chemotherapy does not warrant metastasis. Our analysis of previously published expression profiling of PDX mammary fat pad primary and corresponding lung metastatic tumors (34) revealed higher TRIM37 in metastatic lesions (Supplementary Fig. S5A). These results indicated that higher TRIM37 levels are maintained throughout the metastatic transition.

TRIM37 in association with polycomb complex alters gene expression to promote tumorigenesis (9). To test whether TRIM37 causes transcriptional misregulation of genes involved in metastasis, we knocked down TRIM37 in MDA-MB-231-D3H2LN-2b (35), hereafter referred to as 231–2b, using TRIM37-specific ASO (TRIM37-ASO, Supplementary Fig. S5B) and performed transcriptomic analysis. Of the approximately 2,600 genes whose expression differed significantly between TRIM37 knockdown and control cells (GSE136617; Fig. 5A and B), approximately 71 tumor and metastases suppressors, such as KISS1 and BRMS1, were significantly downregulated by TRIM37 (Supplementary Table SIII). As robust metastatic suppressors, KISS1 and BRMS1 reciprocally correlate with increased tumor recurrence, metastatic foci, and reduced disease-free survival (36, 37). The antimetastatic function is mediated by altered gene expression through cell signaling pathways (38, 39) as well as transcriptional regulation (40, 41). In addition, gene set enrichment analysis (GSEA) revealed that TRIM37 knockdown downregulates hypoxia, epithelial–mesenchymal transition, glycolysis, angiogenesis, inflammatory, and immune response-related genes in TNBC cells, indicating TRIM37-dependent activation of a prometastatic transcriptional program (Fig. 5C; Supplementary Fig. S5C). Likewise, KEGG pathway analysis identified TRIM37 target genes that associated with focal adhesion, pathways in cancer, actin cytoskeleton, ECM interaction, and signaling pathways (Supplementary Fig. S5D).

Figure 5.

TRIM37 alters the transcription program to favor metastatic growth of TNBC tumors. A, MA plot illustrates differential gene expression in TRIM37-ASO–treated compared with control 231–2b cells. Red, significantly upregulated genes (n = 1,126); blue, significantly downregulated genes (n = 1,502); gray, genes not significantly changed (n = 12,440). FDR < 0.05. B, Hierarchical clustering of median-centered gene expression in control or TRIM37-ASO–treated 231–2b cells. Each colored line in the dendrogram identifies a different gene (n = 3). C, Pathways significantly downregulated (blue) or upregulated (red) in TRIM37-ASO–treated cells relative to control 231–2b cells identified by GSEA. D and E, qRT-PCR monitoring TRIM37-regulated metastasis suppressor genes in TRIM37 overexpressing p53−/− MCF10A cells relative to vector control (D) and TRIM37-ASO–treated 231–2b tumors relative to control tumors (E). F, ChIP monitoring BMI1, EZH2, TRIM37, and H2Aub binding at KISS1, BRMS1, and Actin in control and TRIM37-ASO–treated 231–2b cells. G and H, qRT-PCR monitoring TRIM37 target genes in 231–2b cells expressing KISS1 shRNA (G) and BRMS1 shRNA (H) relative to nonsilencer (NS) shRNA. I and J, qRT-PCR monitoring TRIM37 target genes in TRIM37 overexpressing p53−/− MCF10A cells relative to vector control (I) and TRIM37-ASO–treated 231–2b tumors relative to control tumors (J). K, Schematic showing that mice were injected with control or TRIM37-ASO–treated 231–2b cells intracardially and monitored for metastatic tumor burden. n = 8 animals per group. L, Representative ventral BLI of 231–2b expressing control or TRIM37 knockdown (TRIM37-ASO) at day 21. M, Analysis of metastatic tumor growth in mice tissues measured by relative luciferase signal for lungs and number of metastatic lesions in lungs, bones, brain, and liver. n = 8 animals per group. N, Kaplan–Meier survival curve for mice injected with control or TRIM37-ASO–treated 231–2b cells. n = 8 animals per group. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 5.

TRIM37 alters the transcription program to favor metastatic growth of TNBC tumors. A, MA plot illustrates differential gene expression in TRIM37-ASO–treated compared with control 231–2b cells. Red, significantly upregulated genes (n = 1,126); blue, significantly downregulated genes (n = 1,502); gray, genes not significantly changed (n = 12,440). FDR < 0.05. B, Hierarchical clustering of median-centered gene expression in control or TRIM37-ASO–treated 231–2b cells. Each colored line in the dendrogram identifies a different gene (n = 3). C, Pathways significantly downregulated (blue) or upregulated (red) in TRIM37-ASO–treated cells relative to control 231–2b cells identified by GSEA. D and E, qRT-PCR monitoring TRIM37-regulated metastasis suppressor genes in TRIM37 overexpressing p53−/− MCF10A cells relative to vector control (D) and TRIM37-ASO–treated 231–2b tumors relative to control tumors (E). F, ChIP monitoring BMI1, EZH2, TRIM37, and H2Aub binding at KISS1, BRMS1, and Actin in control and TRIM37-ASO–treated 231–2b cells. G and H, qRT-PCR monitoring TRIM37 target genes in 231–2b cells expressing KISS1 shRNA (G) and BRMS1 shRNA (H) relative to nonsilencer (NS) shRNA. I and J, qRT-PCR monitoring TRIM37 target genes in TRIM37 overexpressing p53−/− MCF10A cells relative to vector control (I) and TRIM37-ASO–treated 231–2b tumors relative to control tumors (J). K, Schematic showing that mice were injected with control or TRIM37-ASO–treated 231–2b cells intracardially and monitored for metastatic tumor burden. n = 8 animals per group. L, Representative ventral BLI of 231–2b expressing control or TRIM37 knockdown (TRIM37-ASO) at day 21. M, Analysis of metastatic tumor growth in mice tissues measured by relative luciferase signal for lungs and number of metastatic lesions in lungs, bones, brain, and liver. n = 8 animals per group. N, Kaplan–Meier survival curve for mice injected with control or TRIM37-ASO–treated 231–2b cells. n = 8 animals per group. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

To validate the RNA sequencing (RNA-seq) results, we analyzed expression of representative genes in p53-/- MCF10A cells ectopically expressing TRIM37. Expression of KISS1 and BRMS1 was significantly lower in cells ectopically expressing TRIM37 compared with empty vector (Fig. 5D). Conversely, TRIM37-knockdown tumors expressed KISS1 and BRMS1 at significantly higher levels relative to control xenograft tumors (Fig. 5E). To investigate the mechanism by which TRIM37 regulates KISS1 and BRMS1, we analyzed binding of polycomb complex components, BMI1 and EZH2, to BRMS1 and KISS1 by directed ChIP assays. Both the gene promoters were enriched for BMI1 and EZH2, which was diminished after TRIM37 knockdown (Fig. 5F). These gene promoters were also enriched for H2Aub, which was reduced after TRIM37 knockdown (Fig. 5F). As expected, knockdown of BMI1 and EZH2 resulted in increased expression of these genes (Supplementary Fig. S5E and S5F).

Our results raised the possibility that TRIM37-mediated repression of metastases suppressors induces transcriptional program favoring metastasis. To test this idea, we analyzed a representative set of 15 genes in the top GSEA categories based on statistical analysis and their known biological functions in multiple steps of metastasis (Supplementary Table SIV). For all 15 genes analyzed, knockdown of KISS1 or BRMS1 resulted in their increased expression (Fig. 5G and H; Supplementary Fig. S5G). Consistently, ectopic expression of TRIM37 in p53-/- MCF10A cells significantly increased expression of all the TRIM37 target genes compared with empty vector (Fig. 5I), indicating reciprocal relationship between TRIM37 and metastases suppressors. To validate RNA-seq results in vivo, a subset of TRIM37 target genes was analyzed in TRIM37-knockdown tumors. As expected, all the 15 genes analyzed were significantly reduced in TRIM37-knockdown tumors relative to control tumors (Fig. 5J). Moreover, knockdown of TRIM37 also decreased expression of TRIM37 target genes in MCF7 cells (Supplementary Fig. S5H).

To investigate directly the potential function of TRIM37 in metastasis, we compared the in vivo propensity of control and TRIM37-knockdown cells in NSG mice (Fig. 5K). Knockdown of TRIM37 showed dramatic reduction in the metastatic burden in comparison with control tumors that developed in sites comparable with human breast cancer metastases, such as the brain, lung, liver, lymph nodes, and bone (Fig. 5L; Supplementary Fig. S5I and S5J). TRIM37 knockdown reduced the metastatic tumor burden in lungs by approximately 2-fold, bone by approximately 2-fold, brain by approximately 3-fold, and liver by approximately 3-fold compared with the control animals 21 days after TNBC cell injection (Fig. 5M). Further, histologic analysis of tumors confirmed the distant tumor growth of the control and TRIM37-knockdown 231–2b cells in lung and liver (Supplementary Fig. S5K). Finally, TRIM37 knockdown in 231–2b resulted in a modest but significant improvement of post-injection survival (Fig. 5N). Collectively, these results demonstrate that TRIM37 overexpression enforces transcriptional program in TNBC tumors that promotes metastatic progression.

Design, construction, and characterization of molecularly targeted nanoparticles to deliver TRIM37-ASO in vivo

We show that TRIM37 alters chromatin modification to resist chemotherapy and enforces gene expression changes to favor metastatic transition. These results formed the underlying rationale for targeting TRIM37 as a therapeutic strategy for treating TNBC. To systematically assess the therapeutic effectiveness of inhibiting TRIM37, we engineered liposome-based nanoparticles consisting of DOPA, DOTAP, and DSPE-PEG2000-Maleimide with TRIM37-ASO encapsulated in the core (see also Materials and Methods). We functionalized the nanoparticles by incorporating an investigational FOLR1 antibody (farletuzumab). To this end, we site-specifically and covalently conjugated DSPE-PEG2000-Maleimide to a cysteine-containing Fc-linkered sequence at the C-terminus of a knob heavy chain in farletuzumab (Fig. 6A; Supplementary Fig. S6A–S6C). For convenience, these monodisperse complexes are referred to as “smart nanoparticles” (Fig. 6B; Supplementary Fig. S6D). Figure 6C confirmed that approximately 67% of the maleimide groups present on the outer surface of the smart nanoparticles were functionalized with farletuzumab. No significant differences in the binding affinity of Fc-linkered and monomeric farletuzumab were observed (Fig. 6D). As a control, we generated and characterized farletuzumab-conjugated nanoparticles with control-ASO as a payload; hereafter, referred to as “control nanoparticles.” As expected, the diameter of 106 ± 21 nm (Fig. 6E) and the overall charge of −4.72 ± 0.30 mV (Fig. 6F) were not significantly different between smart and control nanoparticles.

Figure 6.

Design, structural, and functional characterization of smart nanoparticles in TNBC cellular and xenograft mouse models. A, A scheme of site-specific covalent conjugation of farletuzumab to DSPE-PEG2000-Maleimide. An Fc-linkered sequence harboring a Cys at the extended C-terminal hole chain in farletuzumab and conjugated to DSPE-PEG2000-Maleimide. B, Structural model of smart nanoparticles with TRIM37-ASO encapsulated in the core. C, Density of farletuzumab on smart nanoparticles (Smart NP). D, Binding assay for relative avidity index of farletuzumab or nanoparticles. E and F, Dynamic light scattering (E) and zeta potential (F) for nanoparticles. G, Schematic of coculturing experiments described in H and I. HCC1806RR (red) and MCF7 or HCC1806RR (red) and MCF10A cells were cocultured and treated with the IR800-labeled smart nanoparticles. H and I, Representative images for the uptake of smart nanoparticles by HCC1806RR cocultured with either MCF7 or MCF10A as described in G. Scale bars, 50 μm. Results are quantified (I). J, Representative fluorescent images of tumor-bearing mice after treatment with IR800-labeled smart nanoparticles at indicated times. The color scale depicts the fluorescence counts emitted from the tumor cells. K, Necropsies from animals in J were analyzed by fluorescent imaging for detailed organ-specific distribution of IR800-labeled smart nanoparticles. L, qRT-PCR monitoring expression of TRIM37 in 231–2b cells treated with either TRIM37-ASO or control-ASO or nanoparticles for indicated times. M, qRT-PCR monitoring expression of TRIM37 in 231–2b tumors treated with control or smart nanoparticles. Error bars indicate SD and range of at least three biological replicates. n = 6 animals per group. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 6.

Design, structural, and functional characterization of smart nanoparticles in TNBC cellular and xenograft mouse models. A, A scheme of site-specific covalent conjugation of farletuzumab to DSPE-PEG2000-Maleimide. An Fc-linkered sequence harboring a Cys at the extended C-terminal hole chain in farletuzumab and conjugated to DSPE-PEG2000-Maleimide. B, Structural model of smart nanoparticles with TRIM37-ASO encapsulated in the core. C, Density of farletuzumab on smart nanoparticles (Smart NP). D, Binding assay for relative avidity index of farletuzumab or nanoparticles. E and F, Dynamic light scattering (E) and zeta potential (F) for nanoparticles. G, Schematic of coculturing experiments described in H and I. HCC1806RR (red) and MCF7 or HCC1806RR (red) and MCF10A cells were cocultured and treated with the IR800-labeled smart nanoparticles. H and I, Representative images for the uptake of smart nanoparticles by HCC1806RR cocultured with either MCF7 or MCF10A as described in G. Scale bars, 50 μm. Results are quantified (I). J, Representative fluorescent images of tumor-bearing mice after treatment with IR800-labeled smart nanoparticles at indicated times. The color scale depicts the fluorescence counts emitted from the tumor cells. K, Necropsies from animals in J were analyzed by fluorescent imaging for detailed organ-specific distribution of IR800-labeled smart nanoparticles. L, qRT-PCR monitoring expression of TRIM37 in 231–2b cells treated with either TRIM37-ASO or control-ASO or nanoparticles for indicated times. M, qRT-PCR monitoring expression of TRIM37 in 231–2b tumors treated with control or smart nanoparticles. Error bars indicate SD and range of at least three biological replicates. n = 6 animals per group. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Because higher levels of FOLR1 associate with advanced metastatic stage and recurrence in TNBC (42), we rationalized that smart nanoparticles will preferentially deliver TRIM37-ASO to TNBC cells. To test this idea, HCC1806RR (15) was cocultured with either MCF7 or MCF10A cells (Fig. 6G). Immunoblot analysis confirmed that FOLR1 is expressed at significantly higher levels in HCC1806RR compared with MCF7 or MCF10A cells (Supplementary Fig. S6E). Interestingly, we observed approximately 100% uptake of IR800-labeled nanoparticles by TNBC cells compared with MCF7 or MCF10A cells after mixing (Fig. 6H and I). Finally, to test the TNBC cells selectivity of smart nanoparticles in vivo, we injected smart nanoparticles into the 231–2b xenograft-bearing mice. As expected, the smart nanoparticles selectively accumulated in the xenograft tumors within 24 hours and remained localized to the tumor up to 96 hours (Fig. 6J). The accumulation of smart nanoparticles in xenograft tumors was confirmed by the detailed tissue distribution using mice necropsies (Fig. 6K).

Next, we evaluated the biological activity of the smart nanoparticles, which demonstrated sustained TRIM37-ASO release over a period of 48 hours with a biphasic release pattern (Supplementary Fig. S6F). Treatment of MDA-MB-231 cells with smart nanoparticles decreased TRIM37 expression by approximately 80% in a time-dependent manner relative to control nanoparticles (Fig. 6L). To investigate smart nanoparticles-mediated TRIM37 targeting in vivo, we challenged 231–2b-derived xenografts with smart nanoparticles. qRT-PCR and immunoblot analysis confirmed TRIM37 inhibition in smart nanoparticle–treated tumors compared with control nanoparticle–treated tumors (Fig. 6M; Supplementary Fig. S6G). Together, these results confirmed that smart nanoparticle–mediated delivery of TRIM37-ASO effectively decreases TRIM37 expression in TNBC tumors.

Targeting TRIM37 to prevent metastasis in TNBC

Next, we asked whether TRIM37 targeting will reduce metastatic lesions in syngeneic spontaneous metastasis murine model. To this end, Balb/c mice bearing mammary fat pad tumors derived from murine 4T1 cells were administered 1.2 mg/kg of control or smart nanoparticles (Fig. 7A). To inhibit TRIM37, we utilized smart nanoparticles conjugated with murine cross-reactive anti-FOLR1 (Supplementary Fig. S7A–S7C). Although approximately 90% of the control nanoparticle–treated mice developed overt lung metastasis, smart nanoparticle–treated animals showed a dramatic decrease in lung metastasis, with approximately 60% of smart nanoparticle–treated animals showing no detectable lung metastases (Fig. 7B; Supplementary Fig. S7D). Animals were sacrificed at day 30 after treatment due to the moribund condition of control nanoparticle–treated mice, which correlated with more rapid tumor growth at the primary and metastatic sites. The lung metastases were confirmed by hematoxylin and eosin (H&E) staining (Fig. 7C; Supplementary Fig. S7E), the gross lung tissue isolated postmortem (Fig. 7D and E), and a high proliferative index as determined by Ki67 staining (Fig. 7F). Furthermore, TRIM37 was significantly reduced in the metastatic lesions isolated from the smart nanoparticle–treated animals compared with control animals (Supplementary Fig. S7F). No significant changes in liver histology (Supplementary Fig. S7G) or serum aspartate aminotransferase and alanine aminotransferase levels (Supplementary Fig. S7H) between the control and smart nanoparticles–treated animals indicated a lack of hepatotoxicity.

Figure 7.

Targeting of TRIM37 suppresses metastatic lung tumors in vivo. A, Schematic showing that female Balb/c mice bearing mammary fat pad 4T1 tumors were treated with intranasal and intratumor injections of smart or control nanoparticles. n = 8 animals per group. B, Representative dorsal and ventral BLI images of tumor-bearing mice at day 30 after treatment with either control or smart nanoparticles. The color scale depicts the luminescence counts emitted from the metastasis cells. n = 4 animals per group. C, Representative 10× H&E staining images of the lung metastases for control and smart nanoparticle–treated animals. Scale bar, 0.5 mm. Arrows, lung metastatic nodules. D, Lung necropsies from animals in B were analyzed by fluorescent imaging for tumor burden.E, Quantification of metastasis incidence in the lung tissue after control or smart nanoparticles treatment. n = 8 animals per group. F, Ki67 staining of lung tumors derived from mice treated with control or smart nanoparticles. Scale bar, 0.5 mm. Arrows, highly proliferative lung metastatic nodules. G, Model depicting TRIM37 function in multiple steps of TNBC metastasis. H, Schematic showing that NSG mice bearing subcutaneous 231–2b tumors were treated with intranasal injections of smart or control nanoparticles in combination with Dox posttumor resection. n = 8 animals per group. I, Representative BLI images for tumor-bearing mice at day 1 and day 30 postprimary tumor resection. The color scale depicts the luminescence counts emitted from the metastasis cells. n = 4 animals per group. J, Representative 10× H&E staining images of the lung sections of control and smart nanoparticle–treated animals. Scale bar, 0.5 mm. Arrows, lung metastatic nodules. K, Lung necropsies from animals in I were analyzed for tumor burden. L, Quantification of accumulated luciferase signal from the lung tissue after control and smart nanoparticles treatment. n = 8 animals per group. M, Caspase-3 staining of lung tumors derived from mice treated with control or smart nanoparticles in combination with Dox. Scale bar, 0.5 mm. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Figure 7.

Targeting of TRIM37 suppresses metastatic lung tumors in vivo. A, Schematic showing that female Balb/c mice bearing mammary fat pad 4T1 tumors were treated with intranasal and intratumor injections of smart or control nanoparticles. n = 8 animals per group. B, Representative dorsal and ventral BLI images of tumor-bearing mice at day 30 after treatment with either control or smart nanoparticles. The color scale depicts the luminescence counts emitted from the metastasis cells. n = 4 animals per group. C, Representative 10× H&E staining images of the lung metastases for control and smart nanoparticle–treated animals. Scale bar, 0.5 mm. Arrows, lung metastatic nodules. D, Lung necropsies from animals in B were analyzed by fluorescent imaging for tumor burden.E, Quantification of metastasis incidence in the lung tissue after control or smart nanoparticles treatment. n = 8 animals per group. F, Ki67 staining of lung tumors derived from mice treated with control or smart nanoparticles. Scale bar, 0.5 mm. Arrows, highly proliferative lung metastatic nodules. G, Model depicting TRIM37 function in multiple steps of TNBC metastasis. H, Schematic showing that NSG mice bearing subcutaneous 231–2b tumors were treated with intranasal injections of smart or control nanoparticles in combination with Dox posttumor resection. n = 8 animals per group. I, Representative BLI images for tumor-bearing mice at day 1 and day 30 postprimary tumor resection. The color scale depicts the luminescence counts emitted from the metastasis cells. n = 4 animals per group. J, Representative 10× H&E staining images of the lung sections of control and smart nanoparticle–treated animals. Scale bar, 0.5 mm. Arrows, lung metastatic nodules. K, Lung necropsies from animals in I were analyzed for tumor burden. L, Quantification of accumulated luciferase signal from the lung tissue after control and smart nanoparticles treatment. n = 8 animals per group. M, Caspase-3 staining of lung tumors derived from mice treated with control or smart nanoparticles in combination with Dox. Scale bar, 0.5 mm. Error bars indicate SD and range of at least three biological replicates. *, P < 0.05; **, P < 0.01.

Close modal

Primary tumors are routinely treated with a combination of surgery and chemotherapy. Our findings revealed that TRIM37 augments the metastatic potential of TNBC tumors by promoting survival under chemotherapeutic stress, and by inducing metastatic effectors (summarized in Fig. 7G). These results raised the possibility that combining TRIM37 inhibition with chemotherapy will simultaneously increase chemotherapy efficacy and prevent metastatic progression of TNBC to increase the overall survival in patients with TNBC.

To test this idea in a clinically relevant setting, we generated primary tumors by subcutaneously implanting 231–2b lung-tropic cells in female NSG mice (Fig. 7H). Animals were treated with either control or smart nanoparticles intranasally in combination with a single dose of 2 mg/kg of Dox intraperitoneally after tumor resection. Significantly, animals treated with control nanoparticles developed lung metastases, whereas smart nanoparticles treatment dramatically reduced metastatic burden in the lungs (Fig. 7I; Supplementary Fig. S7I). H&E-stained lung sections (Fig. 7J) and luciferase signals from gross lung tissues revealed approximately 5-fold decrease in metastatic growth in animals treated with smart nanoparticles compared with control nanoparticle–treated animals (Fig. 7K and L; Supplementary Fig. S7J). Tumors from smart nanoparticles and Dox-treated animals showed decreased tumor growth as indicated by significantly higher staining for caspase-3 (Fig. 7M) and lower Ki67 staining (Supplementary Fig. S7K) in lung metastatic tumors in comparison with the control tumors.

This study identifies a new TRIM37 network, which is amplified by chemotherapeutic drugs, as a unifying mechanism that drives chemoresistant and metastatic phenotype in TNBC tumors. The results are relevant to approximately 80% of patients with TNBC that lack functional p53 and rely on systemic chemotherapeutic treatments due to the unavailability of any targeted TNBC therapy.

Chemoresistance, in general, is accompanied with extensive genetic and epigenetic alterations. However, whether selection of the clonal cancer cells or new mutations drive chemoresistant phenotype remains to be resolved (32). A recent genomic and phenotypic evolution profiling of TNBC tumors identified both pre-existing resistant genotypes as well as transcriptional reprogramming of resistant signatures in TNBC tumors (30). Our findings are in line with Kim and colleagues showing that pre-existing higher levels of TRIM37 in TNBC tumors promote resistance to chemotherapeutic stress and thus increase survival of TNBC cells. On the other hand, chemotherapy triggers ATM signaling to transcriptionally upregulate TRIM37, which could further select for aggressive TNBC cells. In summary, TRIM37-positive TNBC tumors are protected and thrive under continued chemotherapy to cause aggressive metastatic disease.

A major hurdle in finding cure for aggressive TNBC is the lack of known drivers of the metastatic transition, in part due to a lack of mechanistic insights in the development of metastatic tumors. A genomics-driven discovery of recurring genetic mutations and epigenetic aberrations in the breast cancer genome has revealed tumorigenic drivers (6, 43). However, whether these drivers that were primarily identified in the primary tumors are maintained throughout the chemotherapy regimen and the multistep process of metastasis remain to be evaluated. We used genomic and genetic approaches in relevant TNBC cellular, preclinical murine models, and tumor biopsies to establish TRIM37 function in reducing therapy-induced DNA damage, increasing cancer cell survival, and causing transcriptional aberrations (Fig. 7G). Our proof-of-concept results thus provide a rationale to target such common molecular effectors in combination with chemotherapy to prevent or significantly delay the metastatic progression in patients with TNBC.

Although targeted therapies are desperately needed to limit damage to healthy tissues, new delivery mechanism for cancer cell–specific targeting is also required to reduce detrimental side effects in healthy tissues. Molecularly targeted nanoparticles represent one such mechanism and are being aggressively explored for developing new treatment designs. As such, there are four nanoparticle-based therapies in clinical trials—BIND-014 for NSCLC and prostate cancer (44), CALAA-01 for solid tumors (45), SEL-068 for nicotine addiction (46), and Yale BNP for skin cancer (47).

Similar to mechanisms used in antibody–drug conjugates, we utilized a clinically investigative monoclonal antibody, which selectively delivers TRIM37-ASO into TNBC cells. The molecularly targeted nanoparticles offer a significant advantage over antibody–drug conjugates in terms of higher payload concentrations in tumor cells by enhancing retention times and permeability. An important consideration for nanoparticle delivery designs is clearance from the mononuclear phagocytic system. This is particularly critical for therapies designed to treat metastatic TNBC because premature elimination from circulation will prevent uptake by circulating tumor cells, decrease their accumulation in cancer cells, and minimize their therapeutic impact. To overcome these issues, we incorporated PEG into our design, which enables steric stabilization of nanoparticles and prevents interaction of the nanoparticles with immune cells (48). Functionalizing nanoparticles with “self” markers or homing molecules can further improve the systemic delivery of nanoparticles (49).

Notably, the majority of single-agent therapies tested to date for metastatic TNBC achieved an unimpressive response rate of less than 20%, with minimal impact on patient survival (50). The effective and selective delivery of TRIM37-ASO by farletuzumab-conjugated nanoparticles provides an excellent opportunity to test additional TNBC-enriched surface proteins using monoclonal or bispecific antibody formats. Some of the targets that can be exploited include TNBC-enriched MUC1, Trop-2, and VEGFR2. Our results also raise the possibility that molecularly targeted nanoparticles can deliver diverse payloads selectively to cancer cells.

In conclusion, our results identify a new driver of metastatic progression in patients with TNBC and provide a mechanistic link between the two clinically linked phenotypes: chemoresistance and metastasis. Our findings also raise the possibility of clinically targeting TRIM37 to diminish the resistance to therapy, reduce the dissemination of cancer cells, and infiltration of distant sites. We demonstrate that our therapeutic design selectively inhibits TRIM37 and attenuates metastatic progression of TNBC tumors in vivo.

B.B. Morris reports grants from NCI during the conduct of the study. M.W. Mayo reports grants from NIH/NCI during the conduct of the study and grants from NIH/NCI outside the submitted work. L. Teixeira reports grants from Pfizer (funding research in cancer), other compensation from Roche (principal investigator of clinical trial in triple-negative breast cancer), grants and other compensation from Novartis (principal investigator of clinical trial in triple-negative breast cancer), other compensation from Lilly (principal investigator of clinical trial in triple-negative breast cancer), and other compensation from Invectys (principal investigator of clinical trial in triple-negative breast cancer) outside the submitted work. S. Bhatnagar reports grants from Metavivor Foundation, grants from The Hartwell Foundation, and grants from Department of Defense during the conduct of the study; in addition, S. Bhatnagar has a patent for U.S. Provisional Patent Application Serial No. 62/963,883 pending. No potential conflicts of interest were disclosed by the other authors.

P. Przanowski: Data curation, formal analysis, validation, investigation, visualization, writing-original draft. S. Lou: Formal analysis, validation, investigation, methodology, writing-original draft, writing-review and editing. R.D. Tihagam: Investigation, methodology, writing-review and editing. T. Mondal: Investigation. C. Conlan: Investigation. G. Shivange: Validation, investigation. I. Saltani: Software, investigation. C. Singh: Investigation. K. Xing: Investigation. B.B. Morris: Formal analysis. M.W. Mayo: Resources. L. Teixeira: Resources. J. Lehmann-Che: Resources. J. Tushir-Singh: Conceptualization, resources, data curation, supervision, funding acquisition, methodology, writing-original draft, project administration, writing-review and editing. S. Bhatnagar: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

We thank Michael J. Lee, Michael R. Green, Takahiro Ochiya, David Weber, Michele Vitolo, and Sophia Ran for providing reagents, University of Virginia Tissue Histology Core, Biorepository and Tissue Research Facility (P30CA044579), Flow Cytometry Core (P30CA044579), Molecular Electron Microscopy Core, and The Office of Animal Welfare at University of Virginia. We also thank Edward Egelman, Ani W. Manichaikul, Marya Dunlap-Brown, and Stefan Bekiranov for technical and scientific discussion of statistical analysis; Christine Siu and Agnieszka Kokot for technical assistance with culturing cells; and David Auble and Kristine Zengeler for editorial assistance. M.W. Mayo is supported by NCI RO1CA192399. J. Tushir-Singh is an ovarian cancer Early Career Investigator. J. Tushir-Singh is supported by NCI/NIH grant (R01CA233752) and BCRP breakthrough level-1 award (BC17097) and OCRP funding award (OC180412). S. Bhatnagar is the Hartwell investigator. S. Bhatnagar is supported by Department of Defense Breast Cancer Research Breakthrough Award (BC170197P1, BC190343P1) and Metavivor Translational Research Award.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Pal
SK
,
Childs
BH
,
Pegram
M
. 
Triple negative breast cancer: unmet medical needs
.
Breast Cancer Res Treat
2011
;
125
:
627
36
.
2.
Bauer
KR
,
Brown
M
,
Cress
RD
,
Parise
CA
,
Caggiano
V
. 
Descriptive analysis of estrogen receptor (ER)-negative, progesterone receptor (PR)-negative, and HER2-negative invasive breast cancer, the so-called triple-negative phenotype: a population-based study from the California cancer Registry
.
Cancer
2007
;
109
:
1721
8
.
3.
Bertucci
F
,
Ng
CKY
,
Patsouris
A
,
Droin
N
,
Piscuoglio
S
,
Carbuccia
N
, et al
Genomic characterization of metastatic breast cancers
.
Nature
2019
;
569
:
560
4
.
4.
Yates
LR
,
Knappskog
S
,
Wedge
D
,
Farmery
JHR
,
Gonzalez
S
,
Martincorena
I
, et al
Genomic evolution of breast cancer metastasis and relapse
.
Cancer Cell
2017
;
32
:
169
84
.
5.
Ryan
RJ
,
Bernstein
BE
. 
Molecular biology. genetic events that shape the cancer epigenome
.
Science
2012
;
336
:
1513
4
.
6.
Shen
H
,
Laird
PW
. 
Interplay between the cancer genome and epigenome
.
Cell
2013
;
153
:
38
55
.
7.
Stephens
PJ
,
Tarpey
PS
,
Davies
H
,
Van Loo
P
,
Greenman
C
,
Wedge
DC
, et al
The landscape of cancer genes and mutational processes in breast cancer
.
Nature
2012
;
486
:
400
4
.
8.
Cheng
Y
,
He
C
,
Wang
M
,
Ma
X
,
Mo
F
,
Yang
S
, et al
Targeting epigenetic regulators for cancer therapy: mechanisms and advances in clinical trials
.
Signal Transduct Target Ther
2019
;
4
:
62
.
9.
Bhatnagar
S
,
Gazin
C
,
Chamberlain
L
,
Ou
J
,
Zhu
X
,
Tushir
JS
, et al
TRIM37 is a new histone H2A ubiquitin ligase and breast cancer oncoprotein
.
Nature
2014
;
516
:
116
20
.
10.
Bhatnagar
S
,
Green
MR
. 
TRIMming down tumor suppressors in breast cancer
.
Cell Cycle
2015
;
14
:
1345
6
.
11.
Lambert
AW
,
Pattabiraman
DR
,
Weinberg
RA
. 
Emerging biological principles of metastasis
.
Cell
2017
;
168
:
670
91
.
12.
Acharyya
S
,
Oskarsson
T
,
Vanharanta
S
,
Malladi
S
,
Kim
J
,
Morris
PG
, et al
A CXCL1 paracrine network links cancer chemoresistance and metastasis
.
Cell
2012
;
150
:
165
78
.
13.
Hu
G
,
Chong
RA
,
Yang
Q
,
Wei
Y
,
Blanco
MA
,
Li
F
, et al
MTDH activation by 8q22 genomic gain promotes chemoresistance and metastasis of poor-prognosis breast cancer
.
Cancer Cell
2009
;
15
:
9
20
.
14.
Weiss
MB
,
Vitolo
MI
,
Mohseni
M
,
Rosen
DM
,
Denmeade
SR
,
Park
BH
, et al
Deletion of p53 in human mammary epithelial cells causes chromosomal instability and altered therapeutic response
.
Oncogene
2010
;
29
:
4715
24
.
15.
Volk-Draper
LD
,
Rajput
S
,
Hall
KL
,
Wilber
A
,
Ran
S
. 
Novel model for basaloid triple-negative breast cancer: behavior in vivo and response to therapy
.
Neoplasia
2012
;
14
:
926
42
.
16.
Shivange
G
,
Urbanek
K
,
Przanowski
P
,
Perry
JSA
,
Jones
J
,
Haggart
R
, et al
A single-agent dual-specificity targeting of FOLR1 and DR5 as an effective strategy for ovarian cancer
.
Cancer Cell
2018
;
34
:
331
45
.
17.
Lou
S
,
Zhao
Z
,
Dezort
M
,
Lohneis
T
,
Zhang
C
. 
Multifunctional nanosystem for targeted and controlled delivery of multiple chemotherapeutic agents for the treatment of drug-resistant breast cancer
.
ACS Omega
2018
;
3
:
9210
9
.
18.
Pereira
B
,
Chin
SF
,
Rueda
OM
,
Vollan
HK
,
Provenzano
E
,
Bardwell
HA
, et al
The somatic mutation profiles of 2,433 breast cancers refines their genomic and transcriptomic landscapes
.
Nat Commun
2016
;
7
:
11479
.
19.
Zhou
W
,
Zhu
P
,
Wang
J
,
Pascual
G
,
Ohgi
KA
,
Lozach
J
, et al
Histone H2A monoubiquitination represses transcription by inhibiting RNA polymerase II transcriptional elongation
.
Mol Cell
2008
;
29
:
69
80
.
20.
Shanbhag
NM
,
Rafalska-Metcalf
IU
,
Balane-Bolivar
C
,
Janicki
SM
,
Greenberg
RA
. 
ATM-dependent chromatin changes silence transcription in cis to DNA double-strand breaks
.
Cell
2010
;
141
:
970
81
.
21.
Lord
CJ
,
Ashworth
A
. 
The DNA damage response and cancer therapy
.
Nature
2012
;
481
:
287
94
.
22.
Goldstein
M
,
Kastan
MB
. 
The DNA damage response: implications for tumor responses to radiation and chemotherapy
.
Annu Rev Med
2015
;
66
:
129
43
.
23.
Ciccia
A
,
Elledge
SJ
. 
The DNA damage response: making it safe to play with knives
.
Mol Cell
2010
;
40
:
179
204
.
24.
Williams
RS
,
Williams
JS
,
Tainer
JA
. 
Mre11-Rad50-Nbs1 is a keystone complex connecting DNA repair machinery, double-strand break signaling, and the chromatin template
.
Biochem Cell Biol
2007
;
85
:
509
20
.
25.
Ismail
IH
,
Andrin
C
,
McDonald
D
,
Hendzel
MJ
. 
BMI1-mediated histone ubiquitylation promotes DNA double-strand break repair
.
J Cell Biol
2010
;
191
:
45
60
.
26.
Cohn
MA
,
D'Andrea
AD
. 
Chromatin recruitment of DNA repair proteins: lessons from the fanconi anemia and double-strand break repair pathways
.
Mol Cell
2008
;
32
:
306
12
.
27.
Kastenhuber
ER
,
Lowe
SW
. 
Putting p53 in Context
.
Cell
2017
;
170
:
1062
78
.
28.
Gasco
M
,
Yulug
IG
,
Crook
T
. 
TP53 mutations in familial breast cancer: functional aspects
.
Hum Mutat
2003
;
21
:
301
6
.
29.
Mina
M
,
Raynaud
F
,
Tavernari
D
,
Battistello
E
,
Sungalee
S
,
Saghafinia
S
, et al
Conditional selection of genomic alterations dictates cancer evolution and oncogenic dependencies
.
Cancer Cell
2017
;
32
:
155
68
.
30.
Kim
C
,
Gao
R
,
Sei
E
,
Brandt
R
,
Hartman
J
,
Hatschek
T
, et al
Chemoresistance evolution in triple-negative breast cancer delineated by single-cell sequencing
.
Cell
2018
;
173
:
879
93
.
31.
O'Connor
MJ
. 
Targeting the DNA damage response in cancer
.
Mol Cell
2015
;
60
:
547
60
.
32.
Navin
NE
. 
Tumor evolution in response to chemotherapy: phenotype versus genotype
.
Cell Rep
2014
;
6
:
417
9
.
33.
Wu
G
,
Song
L
,
Zhu
J
,
Hu
Y
,
Cao
L
,
Tan
Z
, et al
An ATM/TRIM37/NEMO axis counteracts genotoxicity by activating nuclear-to-cytoplasmic NF-kappaB signaling
.
Cancer Res
2018
;
78
:
6399
412
.
34.
Echeverria
GV
,
Powell
E
,
Seth
S
,
Ge
Z
,
Carugo
A
,
Bristow
C
, et al
High-resolution clonal mapping of multi-organ metastasis in triple negative breast cancer
.
Nat Commun
2018
;
9
:
5079
.
35.
Tominaga
N
,
Kosaka
N
,
Ono
M
,
Katsuda
T
,
Yoshioka
Y
,
Tamura
K
, et al
Brain metastatic cancer cells release microRNA-181c-containing extracellular vesicles capable of destructing blood-brain barrier
.
Nat Commun
2015
;
6
:
6716
.
36.
Ulasov
IV
,
Borovjagin
AV
,
Timashev
P
,
Cristofanili
M
,
Welch
DR
. 
KISS1 in breast cancer progression and autophagy
.
Cancer Metastasis Rev
2019
;
38
:
493
506
.
37.
Kodura
MA
,
Souchelnytskyi
S
. 
Breast carcinoma metastasis suppressor gene 1 (BRMS1): update on its role as the suppressor of cancer metastases
.
Cancer Metastasis Rev
2015
;
34
:
611
8
.
38.
Cicek
M
,
Fukuyama
R
,
Welch
DR
,
Sizemore
N
,
Casey
G
. 
Breast cancer metastasis suppressor 1 inhibits gene expression by targeting nuclear factor-kappaB activity
.
Cancer Res
2005
;
65
:
3586
95
.
39.
Wu
P-Y
,
Yu
I-S
,
Lin
Y-C
,
Chang
Y-T
,
Chen
C-C
,
Lin
K-H
, et al
Activation of Aryl hydrocarbon receptor by kynurenine impairs progression and metastasis of neuroblastoma
.
Cancer Res
2019
;
79
:5550–62
.
40.
Cicek
M
,
Fukuyama
R
,
Cicek
MS
,
Sizemore
S
,
Welch
DR
,
Sizemore
N
, et al
BRMS1 contributes to the negative regulation of uPA gene expression through recruitment of HDAC1 to the NF-kappaB binding site of the uPA promoter
.
Clin Exp Metastasis
2009
;
26
:
229
37
.
41.
Hurst
DR
,
Xie
Y
,
Vaidya
KS
,
Mehta
A
,
Moore
BP
,
Accavitti-Loper
MA
, et al
Alterations of BRMS1-ARID4A interaction modify gene expression but still suppress metastasis in human breast cancer cells
.
J Biol Chem
2008
;
283
:
7438
44
.
42.
Ginter
PS
,
McIntire
PJ
,
Cui
X
,
Irshaid
L
,
Liu
Y
,
Chen
Z
, et al
Folate receptor alpha expression is associated with increased risk of recurrence in triple-negative breast cancer
.
Clin Breast Cancer
2017
;
17
:
544
9
.
43.
Cancer Genome Atlas Network
. 
Comprehensive molecular portraits of human breast tumours
.
Nature
2012
;
490
:
61
70
.
44.
Sanna
V
,
Pala
N
,
Sechi
M
. 
Targeted therapy using nanotechnology: focus on cancer
.
Int J Nanomedicine
2014
;
9
:
467
83
.
45.
Zuckerman
JE
,
Gritli
I
,
Tolcher
A
,
Heidel
JD
,
Lim
D
,
Morgan
R
, et al
Correlating animal and human phase Ia/Ib clinical data with CALAA-01, a targeted, polymer-based nanoparticle containing siRNA
.
Proc Natl Acad Sci U S A
2014
;
111
:
11449
54
.
46.
Desai
RI
,
Bergman
J
. 
Effects of the nanoparticle-based vaccine, SEL-068, on nicotine discrimination in squirrel monkeys
.
Neuropsychopharmacology
2015
;
40
:
2207
16
.
47.
Deng
Y
,
Ediriwickrema
A
,
Yang
F
,
Lewis
J
,
Girardi
M
,
Saltzman
WM
. 
A sunblock based on bioadhesive nanoparticles
.
Nat Mater
2015
;
14
:
1278
85
.
48.
Suk
JS
,
Xu
Q
,
Kim
N
,
Hanes
J
,
Ensign
LM
. 
PEGylation as a strategy for improving nanoparticle-based drug and gene delivery
.
Adv Drug Deliv Rev
2016
;
99
:
28
51
.
49.
Alqaraghuli
HGJ
,
Kashanian
S
,
Rafipour
R
. 
A review on targeting nanoparticles for breast cancer
.
Curr Pharm Biotechnol
2019
;
20
:
1087
107
.
50.
Vidula
N
,
Bardia
A
. 
Targeted therapy for metastatic triple negative breast cancer: The next frontier in precision oncology
.
Oncotarget
2017
;
8
:
106167
8
.

Supplementary data