The F-Box Domain-Dependent Activity of EMI1 Regulates PARPi Sensitivity in Triple-Negative Breast Cancers

Abstract

The BRCA1-BRCA2-RAD51 axis is essential for homologous recombination repair (HRR) and is frequently disrupted in breast cancers. PARP inhibitors (PARPis) are used clinically to treat BRCA-mutated breast tumors. Using a genetic screen, we identified EMI1 as a modulator of PARPi sensitivity in triple-negative breast cancer (TNBC) cells. This function requires the F-box domain of EMI1, through which EMI1 assembles a canonical SCF ubiquitin ligase complex that constitutively targets RAD51 for degradation. In response to genotoxic stress, CHK1-mediated phosphorylation of RAD51 counteracts EMI1-dependent degradation by enhancing RAD51's affinity for BRCA2, leading to RAD51 accumulation. Inhibition of RAD51 degradation restores HRR in BRCA1-depleted cells. Human breast cancer samples display an inverse correlation between EMI1 and RAD51 protein levels. A subset of BRCA1-deficient TNBC cells develop resistance to PARPi by downregulating EMI1 and restoring RAD51-dependent HRR. Notably, reconstitution of EMI1 expression reestablishes PARPi sensitivity both in cellular systems and in an orthotopic mouse model.

Keywords: BRCA1; CHK1; DNA damage; EMI1; HRR; PARPi resistance; RAD51; breast cancer; proteolysis; ubiquitin.

Figure 1.. The F-Box Domain of EMI1 Promotes Sensitivity to PARP Inhibition

(A) Schematic of the PARPi sensitivity screen. The siRNA smart pool library targeting individually the 69 human F-box proteins was transfected into SUM149PT cells in 96-well plates. 48 hr post-transfection, cells were treated with DMSO or 1 μM olaparib. Seventy-two hr after the addition of olaparib, cell viability was measured using alamarBlue. (B) Relative cell viability of cells transfected with siRNAs to F-box proteins compared to non-targeting siRNA. Dashed line indicates cells transfected with non-targeting siRNA (CTRL) normalized to 1. These cells were used as cutoff values to determine pools scoring with increased or decreased sensitivity to olaparib. The screening experiment was performed three times, and data are presented as mean ± SEM. (C) The indicated TNBC cell lines were infected with lentiviruses expressing either EMI1 or EMI1(ΔF-box) under the control of a doxycycline-inducible promoter and then transfected with either a non-targeting (NT) siRNA or an siRNA to the 3′ UTR of EMI1. Twenty-four hr post-transfection, cells were treated with 0.2 μg/μL doxycycline and, after another 24 hr, with the indicated concentrations of olaparib. Seven days after the addition of olaparib, cell viability was measured using alamarBlue. Cell viability was set as 100% for cells not treated with olaparib. Each experiment was performed at least three times, each with triplicate measurements (±SEM). *p ≤ 0.03, ***p ≤ 0.0002 calculated by ANOVA. (D) The cancer cells used in (C) were collected, lysed, and immunoblotted. 3X FLAG- 2X STREP-tagged exogenous (exo.) EMI1 was separated from endogenous (endo.) EMI1 based on their differential mobility on SDS-PAGE. (E) U2OS cells infected with lentiviruses expressing either EMI1 or EMI1(ΔF-box) under the control of a doxycycline-inducible promoter were transfected with an siRNA to the 3′ UTR of EMI1 and treated with doxycycline for 48 hr. Cells were pulsed with BrdU for 30 min, collected, and analyzed by flow cytometry.

Figure 2.. EMI1 Targets RAD51 for Ubiquitin-Mediated Degradation

(A) HEK293T cells were transfected with either an empty vector (EV) or the indicated FLAG-tagged F-box proteins (FBPs). Twenty-four hr post-transfection, cells were treated with MG132 for 3 hr and collected for immunoprecipitation (IP) and immunoblotting. Nedd. CUL1, neddylated CUL1; WCE, whole-cell extract. (B) HEK293T cells were transfected with either an EV or FLAG-tagged RAD51. The experiment was performed as described in (A), except that, where indicated, IPs were treated with Benzonase. Nedd. CUL1, neddylated CUL1. (C) HEK293T cells were transfected with either an EV or increasing amount of FLAG-tagged EMI1. Twenty-four hr post-transfection, cells were collected and immunoblotted. Where indicated, MG132 was added 3 hr prior to collection. (D) U2OS cells infected with lentiviruses expressing either EMI1 or EMI1(ΔF-box) under the control of a doxycycline-inducible promoter were transfected with an siRNA to the 3′ UTR of EMI1 and treated with 0.2 μg/μL doxycycline for 48 hr. Cells were then treated with cycloheximide (CHX), collected at the indicated times, lysed, and immunoblotted. Asterisks denote lanes loaded with molecular weight markers. (E) The experiment was performed as in (D), except that after doxycycline treatment, cells were fixed and stained with DAPI and an anti-RAD51 antibody. Percentage of cells with spontaneous RAD51 foci was calculated by counting the number of cells displaying more than 15 foci per cell. A minimum of 500 cells were counted for each condition in 3 independent experiments. Data are presented as mean ± SEM. ns, not significant; *p ≤ 0.03 calculated by ANOVA. (F) HEK293T cells were transfected with FLAG-tagged RAD51, either hemagglutinin (HA)-tagged EMI1 or HA-tagged EMI1(ΔF-box), and either MYC-tagged wild-type ubiquitin or MYC-tagged ubiquitin(K0). 24 hr post-transfection, cells were treated with MG132 for 3 hr before collection for IP under denaturing condition and immunoblotting. The bracket indicates a ladder of bands with a relative molecular mass of >37,000 corresponding to polyubiquitylated RAD51. WCE, whole-cell extract.

Figure 3.. Defects in the EMI1-Dependent Degradation of RAD51 Overcome BRCA1 Deficiency and Promote HRR

(A) U2OS cells were transfected with the indicated siRNAs. Forty-eight hr post-transfection, cells were treated with neocarzinostatin (NCS), collected at the indicated times, fractionated into soluble and chromatin fractions, and immunoblotted. (B) HEK293T cells were transfected with either an empty vector (EV) or the indicated FLAG-tagged proteins. Twenty-four hr post-transfection, cells were treated with MG132 for 3 hr before collection for IP and immunoblotting. Nedd. CUL1, Neddylated CUL1; WCE, whole-cell extract. (C) U2OS cells infected with lentiviruses expressing either wild-type RAD51 or RAD51(F129A) under the control of a doxycycline-inducible promoter were transfected with an siRNA to the 3′ UTR of RAD51 and treated with doxycycline for 48 hr. Cells were then treated with cycloheximide (CHX), collected at the indicated times, fractionated into soluble and chromatin fractions, and immunoblotted. (D) The experiment was performed and in (C), except that after doxycycline treatment, cells were incubated in the presence or absence of 2 mM hydroxyurea for the indicated times, fixed, and stained with DAPI and an anti-RAD51 antibody. A minimum of 1,000 cells were counted for each condition in 3 independent experiments (each with 4 technical repeats). Data are presented as mean ± SEM. (E) DR-GFP U2OS cells infected with either an empty lentivirus (EV) or lentiviruses expressing either wild-type RAD51 or RAD51(F129A) under the control of a doxycycline-inducible promoter were transfected with an siRNA to the 3′ UTR of RAD51 with or without an siRNA to BRCA1 and treated with doxycycline for 48 hr. Data are presented as fold change in frequency of repair of DR-GFP, resulting in GFP-positive cells after expression of I-SceI, relative to the samples transfected with a non-targeting (NT) siRNA. Each experiment was performed at least three times, each with triplicate measurements (±SEM). Bottom panels show immunoblotting analysis of a representative experiment. endo., endogenous; exo., exogenous. (F) U2OS cells infected with either an empty lentivirus (EV) or lentiviruses expressing either wild-type RAD51 or RAD51(F129A) under the control of a doxycycline-inducible promoter were transfected with an siRNA to the 3′ UTR of RAD51 with or without an siRNA to BRCA1 and treated with doxycycline for 48 hr. Cells were then incubated in the presence or absence of NCS, fixed, and stained with DAPI and an anti-RAD51 antibody. Slides were co-stained with an anti-BRCA1 antibody (see Figure S4B). UT, untreated cells. A minimum of 500 cells were counted for each condition in 3 independent experiments (each with 4 technical repeats). Data are presented as mean ± SEM. Representative immunofluorescence staining images are shown in Figure S4A. (G) The experiment was performed as in (C), except that after NCS treatment, cells were collected at the indicated times, fractionated into soluble and chromatin fractions, and immunoblotted. endo., endogenous; exo., exogenous.

Figure 4.. In Response to Genotoxic Stress, the EMI1-Dependent Degradation of RAD51 Is Inhibited by the CHK1-Dependent Phosphorylation of RAD51 on Thr309

(A) U2OS cells were transfected with FLAG-tagged EMI1. Twenty-four hr post-transfection, cells were treated with NCS for the indicated times before collection for IP with an anti-RAD51 antibody followed by immunoblotting. Where indicated, cells were treated with DMSO or ATR inhibitor for 1 hr prior to NCS addition. l.ex., lower exposure; s.ex., short exposure; WCE, whole-cell extract. (B) U2OS cells were transfected with FLAG-tagged EMI1 and the indicated siRNAs. Forty-eight hr post-transfection, cells were treated with NCS for the indicated times before collection for IP with an anti-RAD51 antibody followed by immunoblotting. The immunoblotting of the IPs was performed after normalizing the amount of the bait among all samples. UT, untreated cells; WCE, whole-cell extract. (C) U2OS cells were transfected with either an EV (−) or FLAG-tagged BRC4 and the indicated siRNAs. 48 hr post-transfection, cells were collected and immunoblotted. (D) U2OS cells were transfected with FLAG-tagged EMI1. 24 hr post-transfection, cells were pretreated with DMSO or the indicated inhibitors for 1 hr. Where indicated, NCS was added 3 hr prior collection. Cells were collected for IP with an anti-RAD51 antibody followed by immunoblotting. UT, untreated cells; WCE, whole-cell extract. (E) U2OS cells were pretreated with DMSO, KU60018 (to inhibit ATM), AZ20 (to inhibit ATR), or SB218078 (to inhibit CHK1) for 1 hr. Cells were then treated with NCS, collected at the indicated times, fractionated into soluble and chromatin fractions, and immunoblotted. (F) HEK293T cells were transfected with either an EV or the indicated FLAG-tagged proteins. Twenty-four hr post-transfection, cells were collected for IP and immunoblotting. Where indicated, NCS was added 3 hr prior collection. UT, untreated cells; WCE, whole-cell extract. (G) U2OS cells infected with either an empty lentivirus(EV) or lentiviruses expressing either wild-type RAD51 or RAD51(T309A) under the control of a doxycycline-inducible promoter were transfected with an siRNA to the 3′ UTR of RAD51 and treated with doxycycline for 48 hr. Cells were then treated with NCS, collected at the indicated times, fractionated into soluble and chromatin fractions, and immunoblotted. endo., endogenous; exo., exogenous.

Figure 5.. Expression of EMI1 and RAD51 Is Inversely Correlated in Human Breast Cancer, and TNBC Cells with High EMI1 Levels Do Not Form RAD51 Foci

(A) The graph shows the percentage of human breast cancer samples (n = 146) with absent (0), weak (1), medium (2), or strong (3) levels of EMI1 that have absent (0), weak (1), medium (2), or strong (3) RAD51 protein expression. Bottom: representative immunohistochemistry staining images with weak, medium and strong levels of EMI1 and RAD51 in consecutive tissue slides are shown. Linear regression was determined using the χ² test, p = 0.006892. (B) The indicated TNBC cell lines were collected, lysed, and immunoblotted. (C) The indicated TNBC cell lines were transfected with either a non-targeting (NT) siRNA or an siRNA to the 3′ UTR of EMI1. After 48 hr, cells were treated with DMSO or NCS for 3 hr, fixed, and stained with DAPI and an anti-RAD51 antibody. A minimum of 100 cells were counted for each condition in 3 independent experiments (each with 3 technical repeats). Data are presented as mean ± SEM. Representative immunofluorescence staining images are shown in Figure S6D.

Figure 6.. Selective Loss of EMI1 Promotes Resistance to Olaparib Both In Vitro and In Vivo

(A) The indicated cell lines (either parental SUM149PT cells or derived olaparib-resistant [OR] clones) were infected with lentiviruses expressing either an EV or EMI1 under the control of a doxycycline-inducible promoter. Cells were treated with 0.2 μg/μL doxycycline and, after 48 hr, with NCS for an additional 3 hr. Next, cells were fixed and stained with DAPI and an anti-RAD51 antibody. A minimum of 1,000 cells were counted for each condition in 3 independent experiments. Data are presented as mean ± SEM. (B) The experiment was performed as in (A), except that 24 hr after doxycycline treatment, cells were treated with the indicated concentrations of olaparib. 72 hr after the addition of olaparib, cell viability was measured using alamarBlue. Cell viability was set as 100% for cells not treated with olaparib. Each experiment was performed at least three times, each with triplicate measurements (±SEM). *p ≤ 0.03, **p ≤ 0.0021 calculated by ANOVA. (C) 1.0 × 10⁶ of the indicated cell lines were implanted into the mammary fat pad of 6- to 8-week-old female NOD/SCID mice that, 3 weeks after implantation, were divided into two groups, one treated with vehicle control and the other with olaparib for a total of 5 weeks. Tumor size was monitored weekly. Mice were sacrificed by the end of the treatment. Data represent mean ± SEM; n = 10. ns, not significant; **p ≤ 0.021, ***p ≤ 0.0002 calculated by Student’s t test. (D) Tumor weight was measured and recorded by the end of experiment for each group. Data represent mean ± SEM; n = 10. ns, not significant; **p ≤ 0.021 calculated by Student’s t test. (E) Western blot analysis showing EMI1 and RAD51 expression in tumor masses formed in xenotransplanted mice described in (C) and (D).

Figure 7.. A Model of EMI-Dependent Regulation of RAD51 Levels in Both BRCA1-Positive and BRCA1-Defective Cells

(A) Model of how DNA damage inhibits the EMI1-mediated degradation of RAD51, inducing its accumulation and promoting HRR in BRCA1-positive cells. See text for details. (B) Model of how low EMI1 levels promote resistance to PARP inhibition in BRCA1-defective cells. BRCA1-defective breast cancer cells display inefficient HRR and sensitivity to PARPis. Our findings indicate that one way to develop resistance (either primary or acquired) to PARPis is by decreasing EMI1 levels. Specifically, low EMI1 expression leads to the stabilization of RAD51, bypassing BRCA1 defects, allowing HRR, and promoting PARPi resistance.