Loss of the Nuclear Pool of Ubiquitin Ligase CHIP/STUB1 in Breast Cancer Unleashes the MZF1-Cathepsin Pro-oncogenic Program

Abstract

CHIP/STUB1 ubiquitin ligase is a negative co-chaperone for HSP90/HSC70, and its expression is reduced or lost in several cancers, including breast cancer. Using an extensive and well-annotated breast cancer tissue collection, we identified the loss of nuclear but not cytoplasmic CHIP to predict more aggressive tumorigenesis and shorter patient survival, with loss of CHIP in two thirds of ErbB2⁺ and triple-negative breast cancers (TNBC) and in one third of ER⁺ breast cancers. Reduced CHIP expression was seen in breast cancer patient-derived xenograft tumors and in ErbB2⁺ and TNBC cell lines. Ectopic CHIP expression in ErbB2⁺ lines suppressed in vitro oncogenic traits and in vivo xenograft tumor growth. An unbiased screen for CHIP-regulated nuclear transcription factors identified many candidates whose DNA-binding activity was up- or downregulated by CHIP. We characterized myeloid zinc finger 1 (MZF1) as a CHIP target, given its recently identified role as a positive regulator of cathepsin B/L (CTSB/L)-mediated tumor cell invasion downstream of ErbB2. We show that CHIP negatively regulates CTSB/L expression in ErbB2⁺ and other breast cancer cell lines. CTSB inhibition abrogates invasion and matrix degradation in vitro and halts ErbB2⁺ breast cancer cell line xenograft growth. We conclude that loss of CHIP remodels the cellular transcriptome to unleash critical pro-oncogenic pathways, such as the matrix-degrading enzymes of the cathepsin family, whose components can provide new therapeutic opportunities in breast and other cancers with loss of CHIP expression.Significance: These findings reveal a novel targetable pathway of breast oncogenesis unleashed by the loss of tumor suppressor ubiquitin ligase CHIP/STUB1. Cancer Res; 78(10); 2524-35. ©2018 AACR.

Figure 1. Reduced nuclear CHIP staining in primary breast cancer tissue microarray (TMA) specimens

(A) Anti-CHIP IHC staining of a representative breast cancer specimen (top) vs. normal breast tissue (bottom) to show cytoplasmic and nuclear staining. (B) Representative pictures depicting low or high nuclear vs. cytoplasmic CHIP staining patterns in breast cancer TMAs. (C) Kaplan-Meier analysis correlating breast cancer-specific survival (BCSS) with moderate/high vs. negative/low nuclear (left) or cytoplasmic (right) CHIP staining in TMA samples. Number of patients in each group are shown in parentheses next to captions. P values are indicated inside each box.

Figure 2. Suppression of in vitro oncogenic attributes upon CHIP overexpression in CHIP-lo ErbB2+ breast cancer cell lines

(A) Cumulative proliferation of CHIP-lo control (MSCV-puro) vs. CHIP-hi (MSCV-CHIP) 21MT1, BT474 and SKBR3 cell lines. Data points represent cumulative cell numbers with each serial passage. (B) Anchorage-independent cell growth in soft agar after 3 weeks of culture. Y-axis, number of colonies per 2500 seeded cells. (C) Transwell cell migration assay. 2000 cells were added in top chamber and migration scored after 24 hrs. Y-axis, number of migrated cells per high power field (HPF). (D) Matrigel-coated Transwell invasion assay. 2000 cells were added in top chamber and migration scored after 24 hrs. Y-axis, number of migrated cells per HPF. Data in each figure represent mean +/- SEM of 3 experiments, each in triplicates.

Figure 3. Reduced xenograft growth of CHIP-overexpressing BT474 cells in nude mice

(A) Tumor volume of CHIP-lo (MSCV-puro) control and CHIP-hi (MSCV-CHIP) BT474 xenografts over time. Mean+/- SEM, N = 6 per group., * = p<0.05, ** = p<0.01, *** = p<0.001. (B) Photograph of resected CHIP-lo (upper) and CHIP-hi (lower) BT474 cell xenografts. (C) Western blotting to visualize the endogenous and overexpressed (Myc-tagged) CHIP in resected BT474 xenografts; HSC70, loading control. (D) Representative H&E staining and mitotic events (red circles) of CHIP-lo (left) and CHIP-hi (right) BT474 xenograft tumor sections. (E) Quantification of mitotic events of sections depicted in D. Mean +/- SEM, n=3. (F) Ki67 (brown) and cleaved caspase 3 (CC3; red) co-staining of CHIP-lo (left) and CHIP-hi (right) BT474 xenograft tumor sections. (G, H) Quantification of Ki67 (G) and CC3 (H) positive cells depicted in F. Mean +/- SD, n=4.

Figure 4. Identification of MZF1 as a CHIP-regulated transcription factor in breast cancer cells

(A) DNA-binding activities of 345 transcription factors were analyzed in nuclear extracts of CHIP-lo (MSCV-puro control) vs. CHIP-hi (MSCV-CHIP) BT474 and MDA-MB-231 cell lines. Y-axis, log_e-fold binding in CHIP-hi over CHIP-lo cells. The 3-fold increase (upper) or decrease (lower) in binding (dotted lines) was used as cut-off. MZF1 (Open symbol). (B, C) Validation of CHIP-dependent downregulation of protein levels of selected transcription factors identified in A by western blotting of CHIP-lo vs. CHIP-hi BT474 (B) or MDA-MB-231 (C) cell lysates. (D) Real-time qPCR analysis of MZF1 mRNA levels in CHIP-hi vs. CHIP-lo BT474 cells. (E) A biotin-labeled double-stranded oligonucleotide with MZF1-specific DNA sequence was used to probe nuclear extracts of CHIP-hi (MSCV-CHIP, C) vs. CHIP-lo (MSCV-puro, P) BT474 cells. 200-fold excess of the non-labeled oligonucleotide served as a competitor. (F) Immunoblot analysis of MZF1 protein levels in CHIP-hi vs. CHIP-lo BT474 cells. (G) CHIP-dependent ubiquitination and degradation of MZF1. Anti-MZF1 immunoprecipitations from lysates of HEK-293T cells transfected with GFP-MZF1 (1μg) +/- Myc-CHIP (1 or 2 μg) were immunoblotted for ubiquitin or MZF1 (upper panel), and whole cell lysates (lower panel) were blotted for MZF1, CHIP or HSC70 (loading control). (H, I) HEK-293T cells were transfected with MZF1-GFP (250 ng) +/- Myc-CHIP (1 μg). Cells were treated with vehicle (-) or proteasomal inhibitor MG132 (50 μM) or lysosomal inhibitor bafilomycin A1 (100 nM) for 6 or 16 hr., respectively. Lysates were immunoblotted for MZF1 or CHIP. (J, K) Quantitation of the MZF1 levels in CHIP-co-transfected cells in the presence of MG132 or Bafilomycin A1, relative to vehicle controls assigned a value of 1. Data shown represent mean +/- S.D. of n=3; * = p<0.05, ** = p<0.01, *** = p<0.001. (L) The tetratricopeptide repeat (TPR) domain of CHIP is required for binding to MZF1. HEK-293T cells transfected with MZF1-GFP and their lysates used for pulldown with GST or the indicated GST-CHIP fusion proteins. Bound MZF1-GFP was visualized by immunoblotting for GFP (upper panel) or MZF1 (middle panel). Ponceau Red staining of the membrane shows the relative amounts of fusion proteins used in pulldowns (lower panel).

Figure 5. CHIP levels control the expression of MZF1 target genes cathepsins B and L

(A) EMSA analysis with the CTSB promoter sequence in CHIP-hi vs. CHIP-lo BT474 cells was done as in Figure 4B. (B, C) CTSB/L mRNA expression in CHIP-hi vs. CHIP-lo BT474 (B) and MDA-MB-231 (C) cells was analyzed by real-time qPCR; n=3. (D, E) Immunoblotting for CTSB (D) and CTSL (E) protein levels in CHIP-hi vs. CHIP-lo BT474 cells. (F) Immuno-histochemical analysis of MZF1 (left panel), CTSB (middle panel), and CTSL (right panel) in CHIP-hi vs. CHIP-lo MDA-MB-231 cells. (G,H) CTSB/L activities in CHIP-hi vs. CHIP-lo 21MT1 (G) and MDA-MB-231 (H) cells analyzed by the production of a red fluorescent cleavage product. (I,J) Quantification of CTSB (top) and CTSL (lower) activities presented in G and H panels in CHIP-hi vs CHIP-lo 21MT1 (I) and MDA-MB-231 (J) cells. Each square represents one replicate. Mean +/- S.D. shown; * = p<0.05; ** = p<0.01; *** = p<0.005.

Figure 6. The CHIP-MZF1-CTSB axis influences tumor cell invasiveness

(A,B) Degradation of FITC-labeled gelatin matrix (seen as patchy holes in the uniform green matrix) by CHIP-lo vs. CHIP-hi 21MT1 (A) and MDA-MB-231 (B) cells seeded for 48 hours and stained for actin-containing invadopodia with phalloidin (red) and for nuclei with DAPI (blue). (C, D) Quantification of gelatin degradation presented in A (C) and B (D). Mean +/- SD., n=4. (E) Restoration of FITC-labeled gelatin matrix degradation by MZF-1 overexpression in CHIP-hi MDA-MB-231 cells. Stably MZF-1 overexpressing CHHIP-lo or CHIP-hi MDA-MB231 cells were analyzed as in A and B. (F) Quantification of gelatin degradation shown in E. Each square represents a replicate. Mean +/- SD., shown, * = p<0.05, ** = p<0.01, *** = p<0.005. (G) Western blotting of MZF1 overexpression in CHIP-lo vs. CHIP-hi MDA-MB-231 cells. (H, n=3) Quantification of MZF1 overexpression shown in G; n=3.

Figure 7. CTSB inhibition reduces ErbB2+ breast cancer cell tumorigenesis

(A) 21MT1 cells cultured on FITC-gelatin were treated with DMSO vehicle or CTSB inhibitor (CA074; 25 μg/ml) for 48h and degradation analyzed as in Figure 5G. (B) Quantification of FITC-gelatin degradation shown in 7A. Mean ± S.D., n=4. (C) 21MT1 cells seeded on Matrigel-coated Transwell invasion membranes were treated with DMSO (control) or CA074 (25 μg/ml) for 24h, and cells that had invaded through Matrigel to the bottom surface were stained with CyQuant GR fluorescent dye and quantified using a fluorescence reader. Mean ± S.D., n=3. (D) FITC-gelatin matrix degradation by MDA-MB-231 cells was analyzed after 48h culture in DMSO (control) or CTSB inhibitor (CA074; 25μg/ml) as in Figure 5G. (E) Quantification of FITC-gelatin degradation shown in 7D. Mean ± S.D., n=4. (F) Invasion of MDA-MB-231 cells incubated with DMSO (control) or CA074 (25 m/ml) as in Fig. 6C. Mean ± S.D., n=8. (G) Groups of 9 nude mice carrying BT474 xenografts (average 0.5 cm³ size) received Trastuzumab (via tail vein; 4 mg/kg every 4 days), CA074 (i.p., 25 mg/kg in saline daily) or saline (control), and tumor volumes were monitored every other day. Mean +/- SEM., n=9; * = p<0.05, ** = p<0.01, *** = p<0.001.