E3 Ubiquitin Ligase UBR5 Drives the Growth and Metastasis of Triple-Negative Breast Cancer

Triple-negative breast cancer (TNBC), as defined by the negative expression of the estrogen receptor (ER) and progesterone receptor (PR) as well as human epidermal growth factor receptor-2 (HER-2), represents approximately 12% to 17% of all invasive breast carcinomas (1). TNBCs are highly aggressive with large tumor mass, high nuclear grade, increased lymph node involvement at the time of diagnosis, contributing to the highest risk of recurrence and metastasis of all breast cancer types at an early time (2). Particularly, because of the heterogeneity and no well-defined molecular targets, patients with metastatic TNBC have a poor survival for 5 years despite standard treatment (3). It is urgent to discover new targets that are involved in shaping the highly aggressive behavior of TNBC.

Tumor invasion and metastasis is responsible for as much as 90% of cancer-associated mortality (4). To form distant metastasis, epithelial tumor cells must invade the surrounding extracellular matrix (ECM), disseminate into the systemic circulation, and then establish secondary tumors in distant sites (5, 6). In this process, epithelial-to-mesenchymal transition (EMT), in which tumor cells lose cell polarity and E-cadherin expression, is thought to promote tumor cell migration and invasion and to be an early event of metastatic dissemination (7, 8). However, some studies have yielded conflicting results in different models of metastasis, which show that tumors are quite heterogeneous and induction of EMT is not always associated with increased invasive behavior (9–11). Similarly, maintenance of the epithelial phenotype is not necessarily detrimental to invasion and metastasis (12). Histologic analyses have revealed morphologic similarities between primary tumors and metastatic lesions (13), and that E-cadherin expression was elevated in lymph node metastases compared with matched primary tumor, both suggesting that EMT in primary tumors may be followed by mesenchymal-to-epithelial transition (MET) at distant sites in cancer metastasis process (14, 15). Despite these correlative clinical findings, rigorous functional studies linking MET with metastatic colonization ability are scarce and the mechanism of MET is unclear.

UBR5, emerging as a key regulator of the unfolded protein response (UPS) in development and cancer, was originally identified in a screen for progestin-regulated genes in human breast cancer cells as a tumor suppressor gene (16), and several studies reported that UBR5 expression is deregulated in many cancer types (17, 18). Provisional data from The Cancer Genome Atlas (TCGA) unveil that UBR5 gene amplification is a common alteration in many cancer types particularly in ovarian and breast cancers (18). Breast cancer patients with aberrant UBR5 expression have a poor overall survival, indicating that UBR5 may be a critical factor driving the aggressive behavior of certain breast cancers (18). UBR5, a 300-kDa nuclear phosphoprotein, is highly overexpressed in many TNBC samples (17) and modulates ERα-induced gene expression and proliferation through its ubiquitin ligase activity in human breast cancer cell lines (19). UBR5 has also been shown to mediate therapeutic resistance in ovarian cancer through modulation of the DNA-damage response (20). Collectively, UBR5 appears to be a key regulator of cell signaling relevant to broad areas of cancer biology and may have the potential to be a new cancer therapeutic target. However, most of our knowledge about the functions of UBR5 in cancer derives from cell biology studies, the mechanism by which UBR5 contributes to tumor initiation and progression in vivo remains poorly defined. More recently, it is reported that delivery of UBR5 siRNA to ovarian tumors via (dioleoylphosphatidyl-choline) DOPC liposomal nanoparticles resulted in significant reduction in tumor burden and enhanced cisplatin efficacy (21).

In this study, for the first time, we demonstrated by whole-exon sequencing, RNAseq and tissue microarray that the UBR5 gene was amplified and overexpressed in many primary TNBC tumors. We then investigated the role of cell-autonomous UBR5 in tumor growth and metastasis in a syngeneic murine model of transplanted TNBC. We showed that ubr5 gene deletion not only suppressed primary tumor growth with decreased angiogenesis and increased apoptosis, but also abolished metastatic colonization of breast cancer by regulating cell-intrinsic epithelial traits through targeting E-cadherin. Moreover, the tumor growth-promoting effects were completely dependent on UBR5′s interaction with immune cells, whereas its metastasis-mediating activities appeared to be cell autonomous. These findings provide new insights into the role of UBR5 in defining the aggressiveness and metastasis of TNBC and may also provide a new therapeutic target for TNBC.

WT female BALB/c mice and 8-week-old female NOD.CB17-Prkdcscid/J (NOD/SCID) mice were purchased from the Jackson Laboratories. BALB/c mice were maintained in a pathogen-free facility and NOD/SCID mice were maintained in filtered-air laminar-flow cabinets under specific pathogen-free conditions, supplying with sterile food and water. All animal experiments were performed in accordance with National Institutes of Health guidelines for housing and care of laboratory animals after protocol (protocol Number 0701-569A) approved by IACUC at Weill Cornell Medicine.

A total of 21 pairs of TNBC samples and adjacent noncancer tissue samples were obtained from patients who underwent modified radical mastectomy in the Affiliated Xiangya Hospital of Central south University in China. The matched noncancer adjacent tissues were harvested at least 5 cm away from the tumor site. The histologic subtype was determined according to the World Health Organization classification. Written informed consent was obtained from all participants, and research protocols for the use of human tissue were approved by and conducted in accordance with the policies of the Institutional Review Boards at Central South University. Total DNA were extracted from clinical tissues and cell lines using PureLink Genomic DNA Kits (Invitrogen, K182000). DNA from tissues were proceed to sequencing and DNA from cell lines were proceed to PCR for genotype test. Next generation of whole-exon sequencing was performed by Genomics Core Facilities at Weill Cornell Medicine.

All murine and human cancer cell lines used in this study were acquired through ATCC, which provided authentications of each cell line with the commercial purchase orders. ATCC authenticates cell lines utilizing Short Tandem Repeat (STR) profiling via a standardized PCR-based method. STR profiling helps to detect misidentified, cross-contaminated, or genetically drifted cells. The mouse 4T1 breast cancer cell line was cultured in RPMI medium with 10% FBS and maintained under these conditions at 37°C in humidified atmosphere containing 95% air and 5% CO₂ with medium change every second day. To establish a stable UBR5 knockout in 4T1 and B16, cells were transfected with UBR5 CRISPR/Cas9 KO plasmid (Santa Cruz Biotechnology, sc-42781) and UBR5 HDR plasmid (Santa Cruz Biotechnology, sc-42781-HDR) using lipofectamine 2000 reagent (Invitrogen, 11668-019) as per the manufacturer's protocol. Three primers targeting UBR5 genome locus were designed and used for genotyping detection: TTACCCAATTCCAGTCTGTC (strand1), TCACGCCGCGTTTCTTCTTG (strand2), and GCCTGCTCCAGTACATTCAG (strand3). 4T1 and B16 cells were transfected with the TurboGFP control plasmid (Sigma, SHC003). All transfected cells were selected using puromycin. Stable UBR5 knockout monoclones were picked from single-cell colonies and knockout efficiency was determined after selection and propagation.

Human breast cancer cell lines MCF-7 and MDA-MB-231 were cultured in DMEM with 10% FBS. For RNAi-mediated UBR5/EDD expression silencing, these cells were transfected with 20 μmol/L of ubr5-siRNA duplex (5′-CAACUUAGAUCUCCUGAAA-3′) and Lipofectamine RNAiMAX (Invitrogen, 13778075) as per the manufacturer's instructions. Nonspecific siRNA oligo (Sigma, SIC002) was used as a negative control.

To generate WT and/or ubiquitin ligase-dead mutant UBR5/EDD-reconstituted 4T1/ubr5^−/− cell line, cells were transfected with pCMV-Tag2B EDD1 and/or pCMV-Tag2B EDD1 C2768A (from Addgene, #37188 and #37189, respectively) using lipofectamine 3000 reagent (Invitrogen, L3000008) as per the manufacturer's protocol. The stable cell lines (4T1/ubr5^−/− + EDD and 4T1/ubr5^−/− + EDD-C2768A) were selected using G418 (500 μg/mL) for 4 weeks and confirmed by q-PCR and Western blot.

Total RNAs were extracted with the RNeasy plus Mini Kit (Qiagen, 74136), and cDNA was synthesized using the SuperScript VILO cDNA Synthesis Kit (Invitrogen Life Technology, 11754250). RT-qPCR was performed on ABI PRISM 7900HT (Applied Biosystem). Target gene primers were synthesized by Integrated DNA Technology, and amplification of endogenous β-actin was used as an internal control.

Cells were lysed in RIPA buffer (Thermal Scientific) and sonicated briefly. Cell lysates were centrifuged at 10,000 × rpm for 20 minutes at 4°C, and supernatant was collected. Protein concentration was quantified by Bio-rad protein assay (Bio-rad, 5000006). Protein (50 μg) was resolved on a 4%–12% SDS PAGE gel (Genescript) and transferred to the PVDF membrane, blocked with 3% BSA and probed for monoclonal antibodies against UBR5 (NBP2-19051) or E-cadherin (NBP2-19051; Novus Biologicals), or anti-GAPDH (Santa Cruz,sc-FL335).

Cells were seeded into 24-well tissue culture plates with 1 × 10⁴ cells per well and cultured for 72 hours to measure cell proliferation by the sulforhodamine B (SRB) assay (22). Briefly, cells were fixed with 10% trichloroacetic acid for 30 minutes at 4°C and stained with 0.4% (w/v) SRB (Sigma) in 1% acetic acid solution for 30 minutes. SRB was removed and plates were washed 5 minutes with 1% acetic acid. Bound SRB was solubilized with 10 mmol/L Tris buffer, and absorbance (OD) was measured at 510 nm using a microplate reader. Independent experiments were carried out in triplicates.

A wound-healing assay was performed to detect cell migration as per a previous protocol (23). Briefly, when cells were grown to 95% confluence in 6-well plates, they were scratched with a pipette tip and washed three times with PBS. Images of the same field were collected right after the scratch with an attached digital camera mounted on an inverted microscope at indicated time points. The width of the wounds was quantified using Image software. Independent experiments were carried out in triplicate. The gap size (%) was determined and used to evaluate cell migration. Independent experiments were carried out in triplicates.

The Transwell assay was carried out to determine cell migration and invasion as previously described (24, 25). Boyden chambers without Matrigel bedding (Corning, 354578) were used to determine cell migration, and chambers with Matrigel bedding (Corning, 356234) were used to determine cell invasion. Tumor cells (3 × 10⁴) starved overnight in serum-free medium were placed in the upper chambers and 500 μL of complete medium was added to the lower chambers. After 18 hours of incubation for migration assay and 24 hours of incubation for invasion assay, the cells remaining in the upper chambers were removed, and the Transwell membrane was fixed with 4% paraformaldehyde and stained with 0.5% crystal violet. To count the fixed cells, five random fields of vision were captured and counted using an inverted microscope with an attached digital camera. Independent experiments were performed in triplicate.

The adhesion assay was determined with the SRB assay as previously described (26). Briefly, exponentially growing cells were detached from culture plates with cell dissociation buffer (Gibco). After washing, cells were resuspended in serum-free medium. Equal numbers of cells (100,000) were seeded into 96-well plates precoated with Matrigel (200 μg/mL) and incubated for 30 minutes at 37°C. Nonadherent cells were removed by washing with serum-free medium. One hundred microliters of complete culture medium was added, and cells were incubated at 37°C for 4 hours. The number of adhered cells was quantified by the SRB assay. All assays were conducted in triplicate at least three times.

A single-cell suspension was seeded in triplicate 6-well culture plates as previously described (26). Cells were cultured for 1 to 2 weeks and fixed with 4% paraformaldehyde and then stained with 0.5% crystal violet. The number of colonies formed in each well was counted and photographed under the microscope. All assays were conducted in triplicate.

Fifteen microliters drops of cell suspension with 2.5 × 10⁵ cells per mL were seeded onto the inner surface of a 12-well culture plate lid. The lid was quickly flipped over and then placed on the plate so that the drops were hanging from the lid with the cells suspended within them. To limit evaporation, 2 mL of serum-free culture medium was added in each well. After 24-h incubation, the lid of the plate was inverted and photographed using an inverted microscope. The number of cell clusters was determined and used to evaluate the cell aggregation. At least 5 drops were analyzed per experiment and repeated at least three times.

A total of 5 × 10⁵ 4T1/GFP or 4T1/ubr5^−/− tumor cells were subcutaneously injected into the inguinal mammary fat pad of 8- to 10-week-old female BALB/c mice (27).Tumor sizes were measured every other day using an electronic caliper. Tumor volume was calculated using the equation (L × W²)/2, where “L” = length and “W” = width. The mean values of added Individual tumor volumes were used to calculate the tumor burden of each group. Tumor weights were taken at termination of each study. To produce experimental lung metastases, a total of 2 × 10⁵ 4T1/GFP or 4T1/ubr5^−/− tumor cells were intravenously (i.v.) injected into mice. Lungs were harvested on day 21 after tumor cell i.v. injection.

The number of overt macrometastases was counted manually on fixed lungs. To quantify micrometastases, mice were sacrificed at day 21 after tumor cell i.v. injection. Left lungs were excised, minced, digested with tissue dissociation buffer [0.25% collagenase IV (384 unit/mg, Worthington), 0.2% Dipase II (Roche), and 0.01% DNase I (Sigma) in HBSS] with periodic votexing for 1 hour in 37°C water bath, washed and strained with 70-μm strainer and plated in 60 μmol/L 6-thioguanine selection (serve as duplicates). Following 1 to 2 weeks of selection, tumor colonies were stained with crystal violet for 10 minutes, rinsed with ultrapure water and dried overnight prior to counting (27).

Tissue microarray (BR1503d) was purchased from Alena biologicals of Xi'an China. Animal specimens (tumors and lungs) and 18 human breast benign diseases specimens were collected, fixed in 10% buffered formalin, embedded in paraffin, and stained with hematoxylin and eosin (H&E) according to standard protocols or based on the manufacturer's instructions. Monoclonal antibodies against UBR5 (Abcam, ab70311), E-cadherin (Novus Biologicals, NBP2-19051), HIF-a (Novus Biologicals, NB100-105), CD34 (Biolegend, 119301), and Ki67 (Vector laboratories, vp-rm04) were used for immunohistochemistry. Tissue sections (5 μm) were dewaxed in xylene and rehydrated with distilled water. After antigen unmasking with citric antigen retrieval reagent (pH 6.0), immunoreactivity in sections was demonstrated using a horseradish peroxidase-3′-, 3′-diaminobenzidine kit (R&D Systems, CTS017) according to the manufacturer's instructions, followed by counterstaining with hematoxylin, dehydration, and mounting. Sections without primary antibodies were enrolled as negative controls for staining. Tissue apoptosis assay was measured by using a TUNEL assay kit (Promega, G7360).

Two pathologists, both blinded to the results of other markers, performed the analysis of immunohistochemistry independently. Each section was randomly selected for five fields at 400 × magnification. For semiquantitative analysis of immunoreactivity of CD34, the number of vessels was counted according to Weidener's method and its further modifications (28).For semiquantitative analysis of immunoreactivity of E-cadherin, HIF-a, ubr5, and Ki67, the multiple of intensity score and positive cell percentage score was determined as previously described (29). The intensity of staining scored from 0 to 3 (0, none; 1, weak; 2, moderate; 3, strong) and the percentage of positive cells scored from 0 to 4 (0, none; 1, <10% positive; 2, from 10% to 50%; 3, from 51% to 80%; 4, >80%). The multiple score was 0–1, negative (−), 2–3 +, 4–5 ++, >6 +++.

The χ² contingency test Yates correction or Fisher exact test was used to determine the relationship of UBR5 expression and clinical pathologic parameters of breast cancer patients. All other values were expressed as mean ± SEM, and the Student t test was used to determine statistical differences between groups. Values of P < 0.05 were considered statistically significant. These analyses were carried out using the SPSS 19.0 for statistical software.

To identify potential “driver” oncogenes for TNBC development and/or pathogenesis, we applied next-generation sequencing technology to analyze TNBC patients' whole-exon sequence data initially in 21 surgical tumor samples in comparison to the surrounding normal tissues. Many genes showing significant amplifications (fold change >1.2) in these samples across multiple patients are presented in Fig. 1A, including RAB2A (7/21), DPY19L4 (8/21), OSR2 (8/21), FBXO43 (8/21), UBR5 (8/21), and MYC (9/21). One of the most striking genes is UBR5 (ubiquitin protein ligase E3 component n-recognin 5, also known as EDD), a member of the E3 ligase family. UBR5 is highly conserved in mammals and essential for mammalian development. We verified the genomic DNA amplification of UBR5 at the mRNA level in seven randomly selected TNBC samples by real-time RT-PCR (Fig. 1B). We also analyzed UBR5 protein expression in different breast cancer subtypes using tissue microarrays (a representative is shown in Fig. 1C). There were no or weak UBR5 expression in 26 benign breast tissue samples, while all 28 TNBC samples expressed UBR5 and over 85% samples moderately or highly expressed UBR5 (Table 1). The data revealed that the more advanced clinical stage or nuclear grade of the breast cancer, the more UBR5 expression; and the luminal A or B subtypes showed lower UBR5 expression than the HER2 or TNBC breast cancer subtypes (Table 1), suggesting that UBR5 expression may be associated with local breast cancer recurrence and distal metastasis.

To investigate the functional role of UBR5 in TNBC, we studied this gene in a syngeneic murine model of transplanted TNBC in lieu of Ubr5-null mice because of the embryonic lethality. We used the highly metastatic and triple-negative mouse mammary carcinoma 4T1, which expresses very high levels of UBR5. We knocked out UBR5 expression in tumor cells using the CRISPR/Cas9 system in a selected clone named 4T1/ubr5^−/− (Fig. 2A), and confirmed that the gene was inactivated by more than 95% at the levels of both mRNA (Fig. 2B) and protein (Fig. 2C). We then compared a series of phenotypic features of this 4T1 clone with control 4T1 (4T1/GFP) and found an increased expression of several mesenchymal markers, such as MMP-9 and vimentin in 4T1/ubr5^−/− (Fig. 2D). In contrast, the expression of several epithelial markers, such as E-cadherin and β-catenin, was decreased in 4T1/ubr5^−/− compared to the control 4T1/GFP. Meanwhile, the morphology of 4T1/ubr5^−/− was altered from a cuboidal epithelial shape to a more elongated mesenchymal shape (Fig. 2E), which is characteristic of EMT. During EMT, epithelial cells acquire migratory characteristics and become more invasive. We performed wound healing and Transwell invasion assays to determine the migratory and invasive capacity of the 4T1 tumor cells. Consistently, compared with 4T1/GFP, 4T1/ubr5^−/− cells healed the wound faster (P < 0.01, Fig. 2F) and displayed an increased invasive capacity in Transwells (P < 0.001, Fig. 2G). Taken together, these data suggest that loss of ubr5 endows 4T1 cells with enhanced migratory and invasive properties by inducing EMT.

To assess the effects of cell-autonomous UBR5 on tumor growth and metastasis, female BALB/c mice were inoculated in the mammary gland with 4T1/GFP and 4T1/ubr5^−/− tumors. Both types of 4T1 cells showed similar degrees of in vitro growth rates (Supplementary Fig. S1). However, in vivo, the ubr5^−/− tumor growth was very much arrested from day 10 onward, and the tumors shrank gradually over time and some even disappeared completely 2 weeks after tumor cell inoculation (Fig. 3A and B). At that time, lung metastases were compared between 4T1/GFP and 4T1/ubr5^−/− tumor. As Fig. 3C shows, there were far less lung metastatic foci in ubr5^−/− tumor-bearing mice (P < 0.001). To evaluate the spontaneous lung metastasis of these two types of 4T1 tumors, we intravenously injected the same number of tumor cells into mice and observed a strong decrease of lung metastasis in mice bearing 4T1/ubr5^−/− with much lower incidences of lung metastatic nodules (P < 0.001, Fig. 3D) and far fewer tumor colonies grown in vitro as measured by the 6-thioguanie clonogenicity assay (P < 0.001, Fig. 3E). These data demonstrate that although 4T1/ubr5^−/− tumors could form lung micrometastases, they failed to progress to macroscopic metastases.

Moreover, to check against “off-target” effects of the CRISPR knockout method, we reconstituted UBR5 expression in 4T1/ubr5^−/− cells with the human ortholog of UBR5, EDD, or its ubiquitin ligase-deficient mutant C2768A (30) to levels similar to that of the WT tumor (Fig. 3F). Upon inoculation in mice (Fig. 3G), the EDD-reconstituted 4T1 tumor completely restored its growth to the WT level whereas the C2768A mutant only did so partially. Given the very similar levels of EDD and C2768A expression, these data suggest that the tumor growth-promoting effect of UBR5/EDD is not entirely dependent on the E3 ubiquitin-ligase activity of UBR5/EDD.

To form distant metastasis, tumor cells need to undergo a phenotypic transformation, MET, which enables them to colonize in the secondary organ environment to form micrometastases and progress to macroscopic metastases (31). Because E-cadherin is one of the most important hallmarks of MET whose aberrant expression results in the acquisition of invasiveness and more advanced tumor stage for breast cancer, we measured E-cadherin expression both in vitro and in vivo. Strikingly, the protein expression of E-cadherin was completely abrogated in ubr5-deleted 4T1 cells (Fig. 4A), as well as in tumor sections from mice bearing 4T1/ubr5^−/− tumor (P < 0.001, Fig. 4B). Mechanistically, the ability of cells to aggregate (P < 0.001, Fig. 4C), to form clones (Fig. 4D and E), and to adhere to Matrigel (P < 0.001, Fig. 4F) were all strongly decreased in 4T1/ubr5^−/− cells. These data demonstrate that deletion of ubr5 in 4T1 is causative in its loss of E-cadherin expression and highly impaired MET and capacity to colonize in secondary organs such as the lung.

To further determine if the phenotype in 4T1/ubr5^−/− cells was peculiar to this cell line, we silenced the expression of UBR5 in the human breast cancer cell line MCF-7 via siRNA transfection. The silencing effect was confirmed at the protein level (∼90% reduction) for UBR5 and E-cadherin (Supplementary Fig. S2A), and at the mRNA level for β-catenin (Supplementary Fig. S2B). Increased migratory (Supplementary Fig. S2C) and invasive (Supplementary Fig. S2D) activities of the modified MCF-7 cells were confirmed by Transwell assays. Enhanced gap-closing activity of the modified MCF-7 cells was verified by the “wound-healing” assay (Supplementary Fig. S2E). Decreased cellular “clustering” capacity and ability to form clones were demonstrated for the UBR5-silenced MCF-7 cells (Supplementary Fig. S2F andS2G, respectively). The modified cells were also less able to adhere to Matrigel-coated plates (Supplementary Fig. S2H). We have also analyzed a second human breast cancer cell line, MDA-MB-231, which revealed very similar properties of UBR5. These results indicate that UBR5′s role in EMT phenotype changes is more general and not confined to the murine mammary tumor cell 4T1.

To further elucidate the effects of autologous UBR5 on tumor progression, we examined the tumor microenvironment more closely. Histologic analysis revealed an early emergence of ischemia and necrosis in 4T1/ubr5^−/− tumors at day 8 post tumor inoculation (Fig. 5A). There were a substantial number of infiltrating cells within each type of tumors, with a high percentage of polymorphonuclear leukocytes in 4T1/ubr5^−/− tumor, and mostly of mononuclear leukocytes in 4T1/GFP tumor. This was associated with decreased angiogenesis within the 4T1/ubr5^−/− tumor, as measured by immunohistochemistry staining for CD34⁺ endothelial microvessels (P < 0.001, Fig. 5B) and the critical angiogenesis-promoting transcription factor hypoxia-inducible factor-1α (HIF1α; P < 0.001, Fig. 5C). Consistent with a tumor growth arrest, decreased expression of the proliferation marker Ki67 was observed within the 4T1/ubr5^−/− tumor (day 8, P < 0.01; day 12, P < 0.001, Fig. 5D) while TUNEL staining indicated increased apoptosis (day 12, P < 0.001, Fig. 5E). Together, these results demonstrate a strongly altered tumor microenvironment in mice bearing 4T1/ubr5^−/− tumor: decreased angiogenesis and proliferation, increased immune cell infiltration, and tumor cell apoptosis.

To explore the involvement of immune system in tumor growth arrest of ubr5-deleted 4T1 cells, we subcutaneously transplanted 4T1/GFP and 4T1/ubr5^−/− cells into the nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice, which lack functional T/B cells and circulating complement, as well as reduced natural killer (NK) cell and macrophage functions. In sharp contrast to the shrinkage/disappearance of ubr5-deleted tumors in BALB/c mice from day10 onward in WT mice, 4T1/ubr5^−/− and 4T1/GFP tumors displayed little difference in growth in NOD/SCID mice (Fig. 6A). Strikingly, however, 4T1/ubr5^−/− tumors in NOD/SCID mice showed a similarly impaired lung, spleen, and liver metastases compared with its response in WT mice (Fig. 6B).

To further investigate the cellular mechanisms involved in the antitumor effect of UBR5 depletion, we analyzed the major subsets of leukocytes in the spleen and lymph node of BALB/c mice bearing these two types of 4T1 tumors on day 10 before 4T1/ubr5^−/− tumor growth was arrested. There were more splenic effector CD4⁺ T cells (Fig. 6C) that produced higher levels of IFNγ (Fig. 6D), and increased CD8⁺ T cells (Fig. 6E) that produced lower levels of IFNγ (Fig. 6F). In the tumor-draining lymph nodes, we found decreased tumor-infiltrating CD25⁺ Foxp3⁺ regulatory T cells (Treg; Fig. 6G), and increased CD11c⁺, MHC II⁺ mature DCs (Fig. 6H) in mice bearing 4T1/ubr5^−/− tumor, compared with mice bearing control tumors, indicating a more active immune response to ubr5-deleted 4T1 tumor. These results demonstrate that tumor-derived UBR5 promotes tumor growth via suppressing immune leukocyte-mediated control, whereas it facilitates metastasis, at least partially, in a manner independent of the immune cells.

In this study, we showed that deletion of ubr5 in 4T1 mammary tumor suppressed metastatic lung colonization through mechanisms that go beyond cell-intrinsic regulation of epithelial traits. UBR5 knockout also blocked tumor growth through mechanisms that affect immune microenvironment in vivo, associated with decreased angiogenesis, leading to decreased proliferation, increased apoptosis and necrosis, and arrest of tumor growth.

In the process of tumor metastasis, primary tumor cells would acquire fibroblast-like properties and exhibit reduced cell–cell adhesion and increased motility via EMT, which represents a fundamentally important process conducive to tumor dissemination (32–34). In the present study, ubr5 deletion in 4T1 cells resulted in changes in EMT properties and increased their motility and invasive properties in vitro. Surprisingly, the 4T1/ub5^−/− cells failed to form lung macroscopic metastases when directly injected into the blood circulation, which suggested that a complete EMT may not necessarily increase the metastasis in this tumor model. Although 4T1/ubr5^−/− cells displayed decreased adhesion properties on ECM surfaces compared with 4T1/GFP cells, we still observed 6-thioguanine-resistant tumor cell colonies that had metastasized to the lung, indicating that 4T1/ubr5^−/− cells can survive in circulation and may be able to attach to the vessel wall, extravasate, and form micrometastases, but failed to progress to macroscopic metastases in the lung. In the last step of metastasis process, tumor cells should further undergo MET in the secondary organ environment to form micrometastases and progress to macroscopic metastases (31, 35). Several studies have proposed that expression of E-cadherin is an important hallmark of MET (32, 36). Dykxhoorn DM and colleagues reported that among the four isogenic mouse breast cancer cell lines (67NR, 168FARN, 4TO7, and 4T1) with different ability to metastasize implanted into the mammary fat pad to model the steps of metastasis, only 4T1 that highly expressed E-cadherin and cytokeratin-18 formed macroscopic lung and liver metastases (12). Yao D and colleagues showed that when the mesenchymal-like TNBC cell line MDA-MB-231, which has no E-cadherin expression, were implanted into the mammary fat pad of mice, E-cadherin–positive metastatic foci were detected, indicating that reexpression E-cadherin is essential for metastatic colonization (32). In our study, E-cadherin expression in 4T1 cells was almost totally abolished in vitro and in vivo following ubr5 deletion, these cells may not accomplish the final metastatic step of colonization to the secondary organs.

Although an inverse correlation between E-cadherin and invasiveness has been reported for many cancers (37, 38), abundant examples are inconsistent with this general assumption. TNBC patients with high E-cadherin expression in their tumors had a poor clinical outcome and reduced expression of E-Cadherin led to a dramatic reduction of the in vivo growth capability of SUM149, Mary-X, and 4T1 tumor cells (9). E-cadherin status of breast carcinomas showed weak or no correlation with vascular invasion (39), nodal status (40, 41), the presence of metastases (39, 40), and disease recurrence or survival (39–41). Our findings also support the dynamic roles for EMT and MET in the 4T1 model during different steps of metastasis: whereas EMT and low levels of UBR5/E-cadherin may promote invasion and extravasation, MET and high UBR5/E-cadherin expression may be required for efficient colonization of distal organs by the tumor cells.

Tumor angiogenesis is considered to be one of the most critical steps in tumor growth and metastasis, providing essential nutrients and growth factors, and sustained angiogenesis is thought to be a “hallmark” of cancer (42). Angiogenesis is often established by hypoxia-induced expression of vascular endothelial growth factor-α (VEGF-α) and other angiogenesis-inducing molecules (43, 44). We observed early ischemia and necrosis in 4T1/ubr5^−/− tumors with significantly decreased CD34 and HIF1α expression compared with 4T1/GFP tumors. Studies in Drosophila have shown that Hyd (a mutant of the Drosophila UBR5 homolog), in controlling Hh and Dpp signaling, may be involved in yolk sac vascular development (45). Homozygous deletion of ubr5 in mice leads to embryonic lethality around E10.5 due to failed yolk sac and allantoic vascular development (46). The molecular mechanisms underlying these phenotypes remain to be described. A cluster of UBR5-regulated genes were recently identified to be associated with vascular development. Specifically, vascular endothelial cell expression of UBR5 is necessary for normal vessel formation through inhibiting the angiogenic factor ACVRL (47). Our data also showed that there were abundant immune cells infiltrating tumors in each group, but the morphology differed, indicating immune factors may affect tumor angiogenesis.

The observation of a total dependence of the control of tumor growth on the immune system as a result of UBR5 deficiency suggests that tumor-derived UBR5 must affect the immunocompetence of the host directly or indirectly. The accumulation of Tregs at tumor sites has been correlated with biomarkers of accelerated angiogenesis such as VEGF overexpression and increased microvessel density in endometrial (48) and breast cancers (49), providing clinical cues for an association between Tregs and angiogenesis (50). The FACS data also demonstrated that 4T1/GFP tumors recruited more Tregs at the tumor environment and draining lymph nodes compared with 4T1/ubr5^−/− tumors. Tregs can promote angiogenesis by suppressing the activities of Th1 effector T cells, releasing angiostatic cytokines like TNFα and IFNγ, as well as interferon-induced chemokines such as CXCL9, 10, and 11 (50, 51). Tregs can also promote tumor angiogenesis by specifically inhibiting tumor-reactive T cells (52). Facciabene and colleagues have demonstrated that tumor hypoxia in ovarian cancer leads to the recruitment of Tregs via CCL28 upregulation, resulting in significantly increased blood vessel development, while deletion of CD25⁺ or CCR10⁺ cells eliminates Tregs significantly suppressed VEGF expression and angiogenesis at these sites (44). Last but not least, we observed that the TNFα and IFNγ mRNA levels were elevated and the VEGFα mRNA level was decreased in 4T1/ubr5^−/− tumors (Supplementary Fig. S3). The enhanced level of IFNγ could, among other things, induce the assembly of immunoproteasomes (53), which would facilitate processing and presentation of tumor antigens by antigen-presenting cells to host T cells, triggering specific immune responses to the tumor. This is consistent with our observation of increased CD11c⁺, MHC II⁺ DCs in the tumor-draining lymph nodes. Taken together, our results demonstrate that the ubr5 deletion in 4T1 tumor cells induces the activation of DCs, via a yet-to-be identified mechanism(s). The activated DCs may subsequently activate naïve CD4⁺ and CD8⁺ T effector cells in tumor-draining lymph nodes, resulting in the effectors' recruitment to the tumor where they suppress the tumor growth via inhibiting angiogenesis and inducing tumor cell death.

We also investigated the regulatory mechanism of E-cadherin expression by UBR5. Promoter hypermethylation, signaling molecules, transcriptional factors, and microRNAs are involved in the regulation of E-cadherin levels (54). We found that inhibition 4T1/ubr5^−/− cells' methylation with 5-Aza-dC did not result in reexpression of E-cadherin (Supplementary Fig. S4), indicating that UBR5-regulated E-cadherin expression is not through DNA hypermethylation. As an E3 ubiquitin protein ligase, UBR5 may regulate E-cadherin through targeting substrates for proteasome-dependent degradation. Indeed, treating 4T1/GFP cells with Velcade resulted in reduction of E-cadherin levels, demonstrating the importance of the proteasome in this process. We speculated that UBR5 may regulate E-cadherin expression through ubiquitinating certain transcriptional repressors. We then analyzed the expression of some E-cadherin transcription repressors and found that Snai l and Twist levels were slightly elevated in 4T1/ubr5^−/− cells while those of ZEB1 and ZEB2 were not significantly affected (Supplementary Fig. S5).

In conclusion, we combined clinical and experimental approaches to establish a novel and fundamental role for UBR5 in tumor growth and metastasis. Deletion of ubr5 in tumor cells inhibits metastatic lung colonization by reducing cell-intrinsic epithelial traits via the UBR5/E-cadherin axis and blocks tumor growth by inhibiting the tumor angiogenesis via altering the tumor microenvironment, particularly the immune cell populations. UBR5 seems to play dichotomous roles in metastasis, enhancing early steps of migration and invasion while inhibiting the late step of metastatic colonization. How UBR5 regulates immune activation is of great interest to further explore. UBR5′s vital role in promoting tumor growth and metastasis renders itself a tangible therapeutic target for TNBC and others such as ovarian and prostate cancers.

No potential conflicts of interest were disclosed.

The contents are solely the responsibility of the authors and do not necessarily represent the official views of the Qatar National Research Fund.

Conception and design: L. Liao, M. Song, X. Li, L. Tang, X. Ma

Development of methodology: L. Liao, M. Song, X. Li, L. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Liao, M. Song, X. Li, X. Ma

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Liao, M. Song, X. Li, T. Zhang, L. Chouchane, X. Ma

Writing, review, and/or revision of the manuscript: L. Liao, M. Song, L. Chouchane, X. Ma

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Liao, M. Song, X. Li, Y. Pan

Study supervision: X. Ma

This work was supported by New York State Department of Public Health grant C028251 (X. Ma), National Natural Science Foundation of China No. 31370903 (X. Ma), Qatar National Research Foundation NPRP 7-136-3-031 (L. Chouchane and X. Ma).

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.