The E3 ligase HECTD4 regulates COX-2-dependent tumor progression and metastasis

Abstract

E3 ubiquitin ligases mediating turnover of proteins engaged in cancer progression point to key regulatory nodes. To uncover modifiers of metastatic competency, we conducted an in vivo genome-wide CRISPR-inactivation screen using cultured breast circulating tumor cells, following intravascular seeding and lung colonization. We identified HECTD4, a previously uncharacterized gene encoding a conserved potential homologous to E6AP C-terminus domain-containing ubiquitin transferase, as a potent tumor and metastasis suppressor. We show that purified HECTD4 mediates ubiquitin conjugation in vitro, and proteomic studies combined with ubiquitin remnant profiling identify a major degradation target as the prostaglandin synthetic enzyme cyclooxygenase-2 (COX-2; PTGS2). In addition to COX-2 itself, HECTD4 targets its regulatory kinase MKK7. In breast cancer models, HECTD4 expression is induced as cells lose adherence to the matrix, and its depletion massively increases COX-2 expression, enhancing anchorage-independent proliferation and tumorigenesis. Genetic or pharmacologic suppression of COX-2 reverses the protumorigenic and prometastatic phenotype of HECTD4-depleted cells. Thus, HECTD4 encodes an E3 ubiquitin ligase that downregulates COX-2 suppressing anchorage independence in epithelial cancer cells.

Keywords: anchorage independence; breast cancer; metastasis suppressor genes; tumor suppressor gene; ubiquitination.

Fig. 1.

Suppression of CTC-mediated metastasis by HECTD4 in a CRISPR-i screen. (A) Schematic diagram of the in vivo CRISPR inactivation screen in CTCs. (B) Classification of known functions of the top 250 genes identified in the in vivo CRISPRi screen. (C) Distribution of ranked screen scores, according to the fold change compared to the input (log2 FC). The top genes classified as E3 ubiquitin ligases are indicated with HECTD4 highlighted in red. (D) The evolutionary tree showing the 13 members of the HECT subfamily. The distances are in the units of the number of amino acid substitutions per site. (E) Proteomics data from breast tumor tissues showing a reduction in HECTD4 protein in tumors compared with normal breast tissue. Significance was calculated using the unpaired Student t test. (F) Kaplan–Meier plot shows that breast cancer patients with high HECTD4 expression (n = 1,433) in their tumors have improved progression-free survival compared with those with low HECTD4 expression. Significance determined by log-rank (Mantel–Cox) test. (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).

Fig. 2.

Anchorage-independent proliferation and increased tumorigenesis following HECTD4 depletion. (A) The initial in vivo validation experiment comparing HECTD4-KD (sh #1) cells with scrambled control cells. MDA-MB-231 cells with HECTD4-KD (sh #1) or scrambled control were injected separately into the mammary fat pad of NSG mice (n = 4 mice for the control group, n = 4 mice in HECTD4-KD group). Primary tumors were harvested after 34 d and assessed for tumor weight and tumor volume. The bar graph shows tumor weights (Upper) and volumes (Lower) measured on day 34. Error bars represent mean ± SD. Significance was calculated using the unpaired Student t test. The mRNA levels of HECTD4 in the cells used for this experiment are shown in SI Appendix, Fig. S2A. (B) Schematic representation of the tumor cells mixing experiment to interrogate the effects of HECTD4 on tumorigenesis. shControl cells (GFP-shControl; HECTD4-WT, shown in gray) were mixed with either GFP-tagged and mCherry labeled HECTD4-knockdown cells [GFP-mCherry-HECTD4-KD, shown in red] (experimental cohort) or with GFP-tagged and mCherry labeled shControl cells [GFP-mCherry-shControl; HECTD4-WT, shown in red] (control cohort). A 1:1 mixture of GFP-shControl and GFP-mCherry-HECTD4-KD or GFP-shControl and GFP-mCherry-shControl was inoculated into the mammary fat pad of immunodeficient NSG mice (n = 4 mice for control mixture; n = 5 mice for KD mixture, see Materials and Methods for more detail). The mixtures were analyzed by flow cytometry right before inoculation to ensure that each of the populations was equally represented in the mixture inoculated into the mammary fat pad on day 0. The primary tumors were resected after 36 d via survival surgery of the mice and the colonized organs—the lungs and livers—were harvested after 62 d. The ratio of the GFP: mCherry populations was analyzed by flow cytometry in the control and experimental cohorts (Upper). The bars represent the percentage of green (GFP-shControl, shown in gray) and red (GFP-mCherry-shControl or GFP-mCherry-HECTD4-KD, shown in red) cells in the primary and metastatic tumors. Dots represent the ratio of m-Cherry-tagged (manipulated) cells compared with m-Cherry-negative (control) cells within each tumor. Significance was calculated with unpaired the Student t test (Lower). The mRNA levels of HECTD4 in the cells used for this experiment are shown in SI Appendix, Fig. S2B. (C) In vitro growth of HECTD4-depleted MDA-MB-231 cells compared to scrambled control cells under adherent 2D-culture on day 7. Error bars represent mean ± SD. Significance was calculated using two-way ANOVA, Tukey’s multiple comparison test (day 7) (Left). In vitro growth of HECTD4-depleted MDA-MB-231 cells compared to scrambled control cells under anchorage-independent suspension conditions (ultralow-adherent plates) on day 14. Error bars represent mean ± SD. Significance was calculated using the two-way ANOVA, Tukey’s multiple comparison test (day 14) (Middle). Representative photomicrographs of MDA-MB-231 cells grown in anchorage-independent suspension conditions (ultralow-adherent plates) on day 22 (Right). The mRNA levels of HECTD4 in the cells used for this experiment are shown in SI Appendix, Fig. S2A. (D) Single MDA-MB-231 cells with either HECTD4 depletion or scrambled control were seeded on soft-agar coated 6-well plates (10,000 cells/well). After 2 wk, the colonies were stained with crystal violet, imaged, and the number of colonies was counted using the HALO software. Error bars represent mean ± SD. Significance was calculated using the one-way ANOVA test, with Dunnett’s multiple comparison test (Left). Representative images of these colonies (Right). The mRNA levels of HECTD4 in the cells used for this experiment are shown in SI Appendix, Fig. S2F. (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).

Fig. 3.

HECTD4 modulates the stability of COX-2 and its regulatory kinase MKK7. (A) HECTD4 catalyzes ubiquitin chain formation in vitro. Active HECTD4 was immunoprecipitated from freshly lysed cells expressing the full-length alfa-tagged HECTD4 using a nanobody against the alfa-tag. Addition of E1, UBE2D, ATP, and ubiquitin results in the formation of polyubiquitin conjugates. The addition of the deubiquitinating enzyme USP2 leads to the loss of the polyubiquitin chain. The potential reaction products were detected by western blotting against HECTD4, ubiquitin, and UBE2D. (B) Schematic representation of the cloning strategy of the C-terminal end of HECTD4. The short active product contains the reference sequence of HECT and 380 nucleotides upstream, including the putative catalytic cysteine, C4396, in green. In the short mutant form, this cysteine was mutated into an alanine (C4396A). To facilitate the immunoprecipitation and increase its specificity, both products were tagged at the N terminus with an ALFA tag. The resultant protein has an expected molecular weight of 55 kDa (Upper). Lysates from cells transfected with the short-active and short-mutant forms of alfa-HECTD4 were flowed through an ALFA-tag activated column. The input lysate, flow-through, and the eluate were blotted and probed with antibodies against HECTD4 and ubiquitin. Untransfected parental cells are shown as control. The elute is shown here; the input lysate and flow-through are shown in SI Appendix, Fig. S4D. Ubiquitin blotting detects signal from the short active form of HECTD4 but not from the short mutant in which C4396 is converted to Alanine (Lower). (C) Schematic representation of the quantitative proteomics experiment to identify HECTD4 substrates. HECTD4-KD and control cells were cultured under anchorage-independent conditions for 4 d. Cells were lysed in a urea-based buffer, followed by protease digestion of the proteins. A fraction of the complete lysate was conserved for analysis by full proteome mapping. The remaining peptides were purified by reversed-phase, solid-phase extraction. Using an antibody targeting the diGly ubiquitin remnant motif (K-ε-GG), ubiquitinated peptides were then captured by immunoprecipitation and eluted to concentrate them for LC-MS/MS analysis. (D) Volcano plot of the complete proteome of HECTD4-KD MDA-MB-231 cells versus control cells. The x-axis shows the log2 FC of each identified protein (6266), and the y-axis shows the corresponding –log10 P adjusted value. The HECTD4 protein is marked in blue; enriched proteins in the HECTD4-KD cells are on the right side of the graph with a positive log2FC. 25 proteins highlighted in red were found to accumulate in the HECTD4-KD cells (FC > 1.5 and FDR < 0.25, thresholds in dotted lines) and to have their ubiquitinated forms less abundant in the same cells (FC < −1.5 and FDR < 0.25, thresholds in dotted lines) in the diGly screen. COX-2 ranks as the second most enriched protein (>12-fold increase). (E) Volcano plot of the ubiquitinated peptides captured by immunoprecipitation from HECTD4-KD cells versus control cells. The x-axis shows the log2 FC of each identified peptide (3311) aligning to 1,453 proteins, and the y-axis shows the corresponding –log10 P adjusted value. HECTD4 substrates are expected to be less ubiquitinated in the HECTD4-KD cells, thus exhibiting peptides on the left side of the graph with a negative log2FC. The 25 proteins described in Fig. 3D. are also highlighted in red. COX-2 is the protein with the most reduced ubiquitination (>threefold reduction compared to normal cells). (F) Western blot of MDA-MB-231 cultured in regular adherent conditions (adherent) and placed in suspension (suspension) showing the weak expression of COX-2 in adherent cells compared to suspension cells. HECTD4-KD cells express high levels of COX-2 compared with control. HECTD4 protein expression is also upregulated in suspension. (G) HECTD4 ubiquitinates COX-2 in vitro. Potential reaction products were detected by western blotting against HECTD4 and HA-tag. The smear demonstrates the presence of a polyubiquitinated HA-COX-2. (H) qRT-PCR of PP1A, HECTD4, MKK7, and COX-2 mRNA upon depletion of HECTD4 only (mix of short hairpins targeting HECTD4), of MKK7 only (mix of 4 siRNA targeting MKK7, shown in blue), and of a combined KD (mix of shHECTD4 and mix of siMKK7, shown in red and blue stripes). mRNA collected 3 d after siRNA transfection from cells growing in suspension. COX-2 mRNA levels increase upon HECTD4 depletion, and MKK7-KD reverses it. Error bars represent mean ± SD. Significance was calculated using the one-way ANOVA test, with Dunnett’s multiple comparison test. (I) Western blot of a similar experiment described in Fig. 3H, with protein lysate harvested after 4 d. (J) HECTD4 also ubiquitinates MKK7 in vitro. Potential reaction products were detected by western blotting against HECTD4 and HA-tag. The smear demonstrates the presence of a polyubiquitinated HA-MKK7. Lower exposure of the same gel shown below. (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).

Fig. 4.

HECTD4 modulation of COX-2-dependent tumorigenesis and metastasis. (A) Depletion of HECTD4 increases cell proliferation under anchorage-independent, suspension conditions (ultralow-adherent culture dish) compared to scrambled control. Depletion of COX-2 in the HECTD4-depleted cells reverts this phenotype. Cell viability and proliferation measured by CellTiter Glo luminescence. Error bars represent mean ± SD. Significance was calculated using with two-way ANOVA test, with Tukey’s multiple comparison test (Day 10). The mRNA levels of PP1A, HECTD4, and COX-2 in the cells used for this experiment are shown in SI Appendix, Fig. S6A. Only the cells harboring the control and shHECTD4 #3 shRNAs are shown in this figure. Cells with the control and shHECTD4 #1 shRNAs are shown in SI Appendix, Fig. S6B. (B) HECTD4-KD cell proliferation is increased under anchorage-independent, suspension conditions (ultralow-adherent culture dish) compared to scrambled control which remain stable after 9 d without treatment. Addition of celecoxib (80 umol/L) reverses the proliferation advantage of HECTD4-KD cells. Cell viability measured by CellTiter Glo luminescence upon celecoxib treatment. Error bars represent mean ± SD. Significance was calculated using the one-way ANOVA test, with Dunnett’s multiple comparison test. The baseline mRNA levels of PP1A, HECTD4, and COX-2 in the cells used for this experiment are shown in SI Appendix, Fig. S5C. The same cells were used in the experiments in SI Appendix, Fig. S5D. (C) Four different groups (scramble control, COX-2-KD, HECTD4-KD, and double KD) of single MDA-MB-231 cells were seeded on 6-well plates coated with soft agar (10,000 cells/well). After 2 wk, the colonies were stained with crystal violet and counted using the HALO software. Error bars represent mean ± SD. Significance was calculated using the one-way ANOVA test, with Dunnett’s multiple comparison test (Left). Representative images of colonies from this experiment (Right). The mRNA levels of HECTD4 and COX-2 in the cells used for this experiment are shown in SI Appendix, Fig. S6D. The same cells were used in the in vivo mixing experiment in Fig. 4D. (D) Fraction of the four GFP+/mCherry- cell mixing conditions: i. GFP-shControl (M1-M6), ii. GFP-COX-2-KD (M7-M12), iii. GFP-HECTD4-KD (M13-M18), and iv. GFP-Double-KD (M19-M24) [%] (n = 6 mice per group). Primary tumors were bilaterally injected into the mammary fat pads of each mice (see Materials and Methods for more detail) and the fraction of tagged cells arising within each mixed tumor cell populations was calculated using flow cytometry, as depicted in the schematic SI Appendix, Fig. S7A. Bars represent the percentage of red (GFP-mCherry-shControl, shown in red) and green (i. GFP-shControl, ii. GFP-COX-2-KD, iii. GFP-HECTD4-KD, and iv. GFP-Double-KD, shown in gray) cells in the primary (Left) and metastatic tumors (Middle-Lungs, Right-Liver). Dots represent the ratio of mCherry-negative cells (manipulated) compared with m-Cherry-tagged (control) within each tumor. Significance was calculated using the one-way ANOVA test, with Dunnett’s multiple comparison test. The mRNA levels of HECTD4 and COX-2 in the cells used for this experiment are shown in SI Appendix, Fig. S6D. The same cells were used in the soft agar experiment in Fig. 4C. The day 0 (in vitro) injection ratios of the cells used in this experiment are shown in SI Appendix, Fig. S7B. Representative images of the primary tumors, livers, and the lungs are presented in SI Appendix, Fig. S7 C–E, respectively. The volumes of the primary tumors are shown in SI Appendix, Fig. S7F. (E) Schematic model demonstrating the proposed function of HECTD4 as a suppressor of tumorigenesis and metastasis, through its regulation of COX-2. HECTD4 and COX-2 levels are undetectable under adherent cell culture. The levels of both proteins rise under anchorage-independent conditions, and suppression of HECTD4 results in massive COX-2 protein accumulation. High COX-2 levels in the setting of HECTD4 repression are associated with increased proliferation in vitro, as well as increased tumorigenesis and metastasis in vivo. (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).