Ubiquitin-conjugating enzyme complex Uev1A-Ubc13 promotes breast cancer metastasis through nuclear factor-кB mediated matrix metalloproteinase-1 gene regulation

Human breast cancer cell lines MDA-MB-231, MCF7, MDA-MB-468, MDA-MB-361, MDA-MB-453, MDA-MB-436 and SK-BRIII were obtained from the American Type Culture Collection (ATCC, Manassan, VA, USA). The cells were cultured in Dulbecco’s minimum essential medium (DMEM) (Invitrogen, Burlington, ON, Canada) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin (Invitrogen) in a 5% CO₂ atmosphere at 37°C. The MCF10A immortalized human mammary epithelial cells were obtained from ATCC and cultured in DMEM/F12 medium supplemented with 10% horse serum, 100 units/ml penicillin, 100 μg/ml streptomycin (Invitrogen), 10 μg/ml insulin (Sigma, St. Louis, MO, USA), 100 ng/ml choleratoxin (Sigma), 0.5 mg/ml hydrocortisone (Sigma), and 20 ng/ml EGF (Peprotech). MDA-MB-231-TR stable cell lines were created by transfecting MDA-MB-231 cell lines with pLenti6-TR-lentivirus (Invitrogen) and selecting with 10 μg/ml blasticidin (Invitrogen).

Human MMS2, UEV1A, and UEV1C open reading frames (ORFs) were PCR-amplified as KpnI-XhoI fragments and cloned into the pcDNA4.0/TO/HA(+) plasmid vector. The 1.9-kb human MMP1 promoter sequence [GenBank: AJ002550.1] was PCR-amplified as a KpnI-HindIII fragment and then cloned into the same sites of pGL4.2 (Invitrogen). The NF-кB target site was subsequently mutated by site-directed mutagenesis using a quick-exchange method (Stratagene, La Jolla, CA, USA). The sense primer for creating the NF-кB binding site mutation is 5′-AAAGG CAGAA GGGAA CCTCA AGAGG TTTTG AAGAG CCACC GTAAA GTGAG-3′ (mutated sequence italicized). The mutated Ubc13-binding site (F38E) in Uev1A was designed based on a previous study with Mms2-F13E [9]. The modified sequence for UEV1 small hairpin RNA (shRNA) delivered by lentiviral particles was from Santa Cruz Biotechnology, Inc (Dallas, Texas, USA). The lentiviral particle infection of breast cancer cells was performed following instructions of the supplier. The MMP1 and MMP9 small interfering RNAs (siRNAs) were purchased from Genepharma Co Ltd (Shanghai, China). The sequence for MMP1 siRNA is 5′- GCGUGUGACAGUAAGCUAATT-3′ and that for MMP9 siRNA is 5′-CGCUCAUGUACCCUAUGUATT-3′.

Total RNA was prepared from cultured breast cancer cells by using TRIzol reagent (Invitrogen). First-strand cDNA was synthesized from total RNA with SuperScript (Invitrogen) according to manufacturer’s instructions. The human breast cancer cDNAs TissueScan™ cancer qPCR Arrays (#BCRT102) were purchased from Origene (Beijing, China). The clinical information is shown on the website [38] and Additional file 1. qRT-PCR analysis was performed on the iQ5 cycler (Bio-Rad, Hercules, CA, USA). The specific primer sets were as follows: GADPH, 5′- GAAGGTGAAGGTCGGAGTC-3′ and 5′- GAAGATGGTGATGGGATTTC-3′; UEV1, 5′- TCTCCACAGCAATCCTATGAGGTTGA-3′ and 5′- CCAACAGTCGGAAATTGCGAGGG-3′; UEV1A, 5′- GAGAGGTTCAAGCGTCTTACCTGAA-3′ and 5′-ACTGTGCCATCTCCTACTCCTTTCT -3′; UEV1C, 5′-GCAGCCACCACGGGCTCG-3′ and 5′- CAATTATCATCCCTGTCCATCTTGT-3′; MMS2, 5′- CGCTTGTTGGAAGAACTTGA-3′ and 5′- CGGAGGAGCTTCTGGGTAT-3′; MMP1 5′- AAATGCAGGAATTCTTTGGG-3′ and 5′-ATGGTCCACATCTGCTCTTG-3′; MMP9 5′-CATCGTCATCCAGTTTGGTG-3′ and 5′- TCGAAGATGAAGGGGAAGTG-3′. The relative expression levels were calculated using the comparative cycle threshold (CT) method (2^-ΔCT) on the Bio-Rad iQ5 (Bio-Rad).

Cells were seeded in 24-well plates at a density of 1 × 10⁵. After 24 hr, the cells were transfected using X-tremeGENE HP DNA Transfection Reagent (Roche, Indianapolis, IN, USA). Briefly, luciferase reporter gene constructs (400 ng), pcDNA-Uevs plasmids (400 ng) and the pRL-SV40 Renilla luciferase construct (5 ng) (for normalization) were co-transfected into the wells. Cell extracts were prepared 48 hr after transfection and the luciferase activity was measured using the Dual-Luciferase reporter assay system (Promega, Madison, WI, USA).

Cells were grown to log phase and lysed in Dulbecco’s PBS (150 mM NaCl, 10 mM Na₂HPO₄ and 10 mM NaH₂PO₄, pH 7.4) with 1% SDS and the protease inhibitor cocktail for mammalian cells (Sigma-Aldrich). Total protein concentration was determined by the Bradford method using a commercial reagent from Bio-Rad. Cell extracts or purified proteins were electrophoresed in 10% or 15% SDS-polyacrylamide gel electrophoresis (PAGE) gels, transferred to polyvinyl difluoride (PVDF) membrane, and incubated with specific primary antibodies. Monoclonal antibodies (mAbs) LN2B (anti-Uev1) and 4E11 (anti-Ubc13) were from the laboratory stock [10, 39]. Primary antibodies against HA (sc-7392), MMP1 (sc-30069), NF-κB (sc-372), Lamin B (sc-166729), β-tublin (sc-6216), and secondary goat anti-mouse antibody IgG-horseradish peroxidase (HRP) (sc-2005) and goat anti-rabbit IgG-HRP (sc-2004) antibody were from Santa Cruz. The P-IκBα antibody (#2859S) was from Cell Signaling Technology (Whitby, ON, Canada), while an anti-MMP9 antibody (ab38898) was from Abcam (Toronto, ON, Canada) and β-actin antibody (BM0627) was from Boster (Wunah, China).

We immunoprecipitated 1 mg of protein samples in a total volume of 1 ml with 2 μg of antibody and 20 μl of Protein-A beads (for rabbit polyclonal antibodies) or Protein-G beads (for mouse monoclonal antibodies). The samples were rotated at 4°C overnight. The beads were washed 4 times with 1 ml of cold NP40 lysis buffer containing protease inhibitors. The beads were then boiled for 10 minutes in the presence of 25 μl 2 × sample buffer and the released proteins fractionated by SDS-PAGE in 12% or 15% gels. Proteins were detected by immunoblotting as described above.

In vitro invasion assays were conducted using Transwells (Costar, Cambridge, MA, USA) with 8-μm-pore-size polycarbonate membrane filters in 24-well culture plates. The upper surface of the filter was coated with Matrigel (Becton Dickinson, Bedford, MA, USA) in a volume of 12.5 μl per filter. The Matrigel was dried and reconstituted at 37°C into a solid gel on the filter surface. The lower surface of the filter was coated in fibronectin (20 μg/ml), vitronectin (10 μg/ml), collagen IV (50 μg/ml), or 10% (BSA)-DMEM as chemoattractants. After starving in BSA-free DMEM overnight, 2 × 10⁴ cells were seeded in the upper chamber. The cells were allowed to invade for 48 hr. Cells that invaded the lower surface of the filter were counted in five random fields under a light-microscope at high magnification. Experiments were conducted at least in triplicate. The in vitro cell migration ability was detected by wound healing assay or transwell assay without Matrigel coating. The wound healing assay was performed by seeding 2 × 10⁵ cells onto 96-well plates. Confluent monolayers were wounded using a pipette tip. After 48 hr the cell migration distance was measured under a light-microscope at high magnification in at least three random fields. The transwell assay without Matrigel was conducted using Transwells with 8-μm-pore-size polycarbonate membrane filters in 24-well culture plates. The lower surface of the filter was coated with 10% BSA-DMEM as chemoattractants. After starving in a BSA-free DMEM medium overnight, 5 × 10⁴ cells were seeded in the upper chamber. Cells were allowed to invade for 6 to 36 hr and those migrated to the lower surface of the filter were counted in five random fields under a light-microscope at high magnification. Experiments were conducted at least in triplicate.

The experimental mouse work followed the animal care protocol CNUAREB-2012002 approved by the Capital Normal University Animal Research Ethics Board and was conducted at the Peking University Health Science Center, China. For the tumorigenesis assays, 5 × 10⁶ breast cancer cells were injected subcutaneously into the lateral flanks of 4- to 5-week-old BALB/c female nude mice. The palpable tumor diameters were measured once per week. Tumor length (L) and width (W) were measured with a caliper, and the volume (V) was calculated by the following equation:

The mice were sacrificed 6 weeks after cell injection. For experimental metastasis assays by intravenous (i.v.) injection, 2 × 10⁵ breast cancer cells were injected into the tail veins of the 4- to 5-week-old female BALB/c nude mice. Endpoint assays were conducted 5 weeks after injection. Metastatic lung nodules 0.5 mm in diameter were counted. Analysis of variance (ANOVA) was used for statistical analyses. To ensure representative sampling of lung tumor nodules, four sections were made per lung at various depths along the coronal plane of the lung. The nodules per lung (four sections) were counted under a light-microscope.

Formalin-fixed lungs were paraffin-embedded, and tissue sections derived from tumor nodules or other tissues were stained with H&E to evaluate the morphology and invasiveness of breast cancer cells. Metastatic tumor nodules were counted throughout the entire lung section at all three depths under a light-microscope. Anti-Uev1A (LN1), anti-MMP1 (sc-30069) and NF-κB p65 (sc-372) primary antibodies from Santa Cruz were used for immunohistochemistry (IHC). TissueFocus™ breast cancer tissue microarrays (CT565863) for IHC were obtained from Origene (Beijing, China). Microscopic images were captured by a SPOT digital camera mounted in a light-microscope.

HeLa cells were treated with 40 ng/ml TNF-α for 2 hr. Cells were washed, scraped with PBS, and centrifuged at 3,000 rpm at 4°C. The pellet was suspended in 10 mM Tris (pH 8.0) with 1.5 mM MgCl₂, 1 mM dithiothreitol, and 0.1% NP-40, and incubated on ice for 15 minutes. Nuclei were separated from cytosol by centrifugation at 12,000 rpm at 4°C for 15 minutes. The cytosolic supernatants were removed and the precipitated pellets were suspended in 10 mM Tris (pH 8.0) containing 100 mM NaCl and stored on ice for 30 minutes. After agitation for 30 minutes at 4°C, the lysate was centrifuged at 12,000 rpm for 15 minutes at 4°C, and the supernatant was collected.

The statistical significance of differential findings between the experimental and control groups was determined by Student’s t-test as implemented by Microsoft Excel 2010 (*P <0.05, **P <0.01 and ***P <0.001).

Two major human UEV1 transcripts (UEV1A and UEV1B) were previously reported [1]. It was determined that only Uev1A, but not Uev1B, is able to physically interact with Ubc13 and promote K63-linked polyubiquitination [10]. It turns out that Uev1B is excluded from the nucleus and involved in endosomal trafficking [40]. Interestingly, database analyses also indicate another splicing variant that would encode a 147-amino acid Uev1 core domain, which we name Uev1C (Figure 1A), the cellular function of which is currently unknown.

Western blot analysis of endogenous Uev1s using a Uev1-specific monoclonal antibody LN2B [39] could only detect Uev1C in several breast cancer cell lines including MDA-MB-231 and MCF7, and MCF10A, an immortalized normal mammary epithelial cell line (Figure 1B). qRT-PCT analysis revealed that the relative transcript level of UEV1C is approximately 100-fold higher than that of UEV1A in the above cell lines (Figure 1C), consistent with observations in other cell lines including HeLa and U2OS (data not shown).

We next examined relative transcript levels of UEV1A and UEV1C in breast cancer lines using MCF10A as a reference. Interestingly, the UEV1A transcript level is elevated in all breast cancer cell lines examined (Figure 1D), with no significant upregulation of UEV1C (Figure 1E) or MMS2 (Additional file 2: Figure S1A) in these lines. It has been previously reported that the UEV1A level may be elevated when normal cells undergo immortalization [2]. To further assess relative UEV1A expression in normal versus breast tumor tissues, we measured the UEV1A transcript level in five normal human breast samples and 43 breast cancer samples from TissueScan microarrays. Compared with the five normal human breast samples, 33/43 or 77% of breast cancer samples display UEV1A expression above the highest UEV1A level in normal samples, or at least 1.7-fold higher than the average level among the five normal samples (Additional file 2: Figure S1B), suggesting that Uev1A may play a role in promoting breast tumorigenesis.

To ask whether an elevated UEV1A level is indeed sufficient to promote breast cancer, UEV1A, UEV1C or MMS2 genes were cloned into a pcDNA4.0/TO/HA(+) vector and then transfected into MDA-MB-231-TR cells to construct stable cell lines, and the level of ecotopic gene expression after 10 μg/ml doxycycline (Dox) treatment was monitored by qRT-PCR (Additional file 2: Figure S2A-C) and western blot against the HA tag (Figure 2A), Uev1 (LN2B, Additional file 2: Figure S2D) or Uev1 plus Mms2 (LN3, Additional file 2: Figure S2E).

The cell growth and cell cycle progression of stable MDA-MB-231 transfectants were first measured, with no obvious alterations among each group (Additional file 2: Figure S3). The effects of ecotopic genes on breast cancer cell migration and invasion were then measured. The wound-healing experimental data show that the overexpression of UEV1A doubles the mobility compared with the same cells without ecotopic UEV1A expression or vector-transfected MDA-MB-231-TR cells with Dox treatment, while overexpression of UEV1C or MMS2 does not affect cell mobility compared with uninduced cells (Figure 2B, C and Additional file 2: Figure S4). In a transwell assay, the invasiveness of UEV1A transfectants after induction was approximately 2.3-fold higher than the control, UEV1C or MMS2 transfectants, whereas there was no significant difference among control, UEV1C and MMS2 transfectants regardless of Dox treatment (Figure 2D and E). These results suggest that UEV1A regulates breast cancer cell migration and invasion in vitro.

As an increased ability of cancer cells to migrate and invade in vitro is a faithful indicator of cell metastasis, to further confirm the correlation between UEV1A expression and breast cancer metastasis, we assessed the effects of UEV1A on metastasis using an in vivo xenograft mouse model. Stably-transfected MDA-MB-231-TR cells were injected into the lateral flanks of 4- to 5-week-old BALB/c female nude mice and Dox (625 mg/kg) was added in feed as soon as the cells were injected. Tumor growth and metastasis were then monitored. All mice (10/10) injected with the UEV1A-expressing cells had massively enlarged lymph nodes containing invasive breast cancer cells (Figure 2F). In contrast, there were no tumor metastasis foci in lymph nodes of mice injected with vector control or UEV1C-expressing cells, although some lymph nodes were enlarged (data not shown). Furthermore, overexpression of UEV1A but not UEV1C accelerated tumor growth compared to vector-transfected cells (Figure 2G). In the tail-vein injection groups, compared with UEV1C or vector control, overexpression of UEV1A significantly promoted lung metastasis colony fomation (Figure 2H, and I). These observations collectively demonstrate that elevated expression of UEV1A alone is sufficient to promote tumor growth and metastasis.

Compared with MCF-10A cells, there was a 2.7-fold increase in the UEV1A expression in MDA-MB-231 cells (Figure 1D). To ask whether this moderate overexpression of UEV1A contributes to breast cancer metastasis, the endogenous UEV1A expression in MDA-MB-231 cells was suppressed using an shRNA (shUEV1) delivered by lentiviral particles. It was found that two independent shUEV1 constructs, shUEV1-1 and shUEV1-2, reduced UEV1A expression to 40% and 55% of control shRNA-treated cells, respectively (Figure 3A). As expected, the cellular UEV1C mRNA and protein levels were also reduced but MMS2 remained unaffected (Additional file 2: Figure S5A-C). Moreover, partial depletion of Uev1 reduced cell migration (Figure 3B and Additional file 2: Figure S5D) and invasion (Figure 3C and Additional file 2: Figure S5E). The above findings were further extended by using a xenograft lung metastasis model, in which depletion of Uev1 limited tumor growth to the extent that no tumor was found in nude mice injected with MDA-MB-231 cells in which the Uev1 level was reduced by shUEV1-2 (Figure 3D and Additional file 2: Figure S5F). Furthermore, depletion of Uev1 in MDA-MB-231 cells significantly reduced the number of lung nodules formed in mice and completely abolished metastasis in lung, even with moderate depletion of Uev1 (Figure 3E, shUEV1-1, middle panel). These results clearly indicate that the elevated UEV1A expression in MDA-MB-231 cells plays a critical role in breast tumorigenesis and metastasis.

To understand the mechanism by which Uev1A promotes metastasis in breast cancer cells, we took into account that Uev1A has been reported to activate NF-κB in HepG2 [19], and that NF-κB regulates the expression of a large number of genes critical for tumorigenesis, inflammation and metastasis [41]. As a hallmark of NF-κB activation is its translocation from the cytoplasm to the nucleus, we transfected MDA-MB-231-TR cells with a variety of constructs, induced the target gene expression by adding Dox, fractionated cells and then measured the subcellular distribution of the p65 subunit of NF-κB. As seen in Figure 4A, only overexpression of UEV1A, but not UEV1C or MMS2, was able to increase the phosphorylation of the NF-κB inhibitor IκBα (presumably in the cytoplasm) and enrich p65 in the nucleus. Consistently, depletion of Uev1 by shRNA reduced IκBα phosphorylation and p65 nuclear translocation (Figure 4B).

It has been previously reported that Uev1A is a cofactor of Ubc13 in the NF-κB signaling pathway [10, 17]. To ask whether the above Uev1A function is indeed dependent on Ubc13, we created a Uev1A-F38E mutation as the corresponding Mms2-F12E mutation (Figure 1A) absolutely abolishes its interaction with Ubc13 and its ability to promote Ubc13-mediated K63 polyubiquitination [9]. Indeed we confirmed that the Uev1A-F38E substitution does not affect its expression (Additional file 2: Figure S2F) but abolishes its interaction with Ubc13 in vivo (Figure 4C). As expected, overexpression of UEV1A-F38E failed to activate NF-κB, as judged by a lack of IκBα phosphorylation and p65 nucleaar translocation (Figure 4D).

As NF-κB is a transcriptional factor that regulates the expression of a large number of genes including those involved in metastasis, we measured the transcription of many established or putative NF-κB target genes thought to be involved in metastasis such as COX2[42], VEGF[43] and MMP family genes, among which MMP1 and MMP9 transcripts were elevated by 4.6- and 3.9-fold, respectively, in UEV1A-overexpression cells, but not in UEV1C or MMS2 overexpression cells (Figure 5A,B), with corresponding elevation at proteins levels (Figure 5C). The observed increase is completely dependent on Ubc13, as overexpression of UEV1A-F38E failed to induce MMP1 or MMP9 (Figure 5B). Similarly, depletion of UEV1 in MDA-MB-231 cells significantly reduced MMP1 and MMP9 transcript (Figure 5D) and protein (Figure 5E) levels, and more efficient depletion of UEV1 (shUEV1-1 versus shUEV1-2) resulted in stronger repression of MMP1 and MMP9 expression (Figure 5D and E), indicating that MMP1 and MMP9 are tightly regulated by Uev1A-Ubc13.

To ask whether MMP1 and/or MMP9 are critical effectors for Uev1A-induced metastasis, we depleted MMP1 or MMP9 by siRNA in MDA-MB-231 cells (Figure 6A). The above treatment did not affect the UEV1A expression (Figure 6A), but the depletion of MMP1 significantly decreased the invasiveness of MDA-MB-231 cells as determined by a transwell assay, while MMP9 depletion had much less effect (Figure 6B and Additional file 2: Figure S6A). Similarly, MMP1 depletion in UEV1A-overexpressed MDA-MB-231 cells (Additional file 2: Figure S6B) also decreased invasiveness (Additional file 2: Figure S6C and D). To ask whether overexpression of MMP1 is indeed sufficient to dictate MDA-MB-231 cell invasion, we constructed an MMP1-expression plasmid and transfected it to MDA-MB-231 cells, which resulted in a 2.6-fold increase in the MMP1 transcript level (Additional file 2: Figure S6E) and a similar increase at the protein level (Additional file 2: Figure S6F), as well as a 2.3-fold increase in invasion (Figure 6C and Additional file 2: Figure S6G). We then restored MMP1 level in UEV1-depleted cells to that of control cells (Figure 6D and E), which was sufficient to rescue the invasiveness in both UEV1-depleted MDA-MB-231 cell lines (Figure 6F). As MMP1 is an important cancer cell metastasis factor [44‐46], the above findings allow us to conclude that UEV1A regulates metastasis through tightly controlling MMP1 expression.

To ask whether Uev1A regulates MMP1 through NF-κB, we cloned the 1.8-kb human MMP1 promoter sequence (Figure 7A) into pGL4.2 and co-transfected it with plasmids expressing UEV1A, UEV1A-F38E or an empty vector into MDA-MB-231 cells. A luciferase assay showed that UEV1A expression activates the MMP1 promoter and that this activation relies on its interaction with Ubc13, as the Uev1A-F38E substitution completely abolished the activation (Figure 7B). The predicted NF-κB binding site (−1133 to approximately −1125) in the MMP1 promoter was then mutated (Figure 7A) and the mutated reporter was used to co-transfect with plasmids expressing UEV1A, UEV1C, MMS2 or the empty vector into MDA-MB-231 cells. Overexpression of only UEV1A, but not UEV1C or MMS2, activated the wild-type P _MMP1 -Luc reporter and this activation absolutely required the intact NF-κB target site (Figure 7C). To gain direct evidence that the predicted NF-κB binding site indeed interacts with NF-κB, an EMSA was performed using nuclear lysates from TNFα-treated and untreated cells (Figure 7D). A biotin-labeled synthetic MMP1 promoter probe containing the putative NF-κB-binding sequence was able to interact with the nuclear lysate and this interaction was enhanced when cells were pretreated with TNF-α, which induced nuclear translocation of p65 (Figure 7E, lanes 4 and 5). This interaction was abolished when the NF-κB binding sequence was mutated (lane 6) or out-competed by adding excess unlabeled probe (lane 7), indicating that the interaction is sequence-specific. We conclude from the above observations that NF-κB directly binds to the MMP1 promoter at the predicted binding site.