Receptor for hyaluronan-mediated motility (RHAMM) defines an invasive niche associated with tumor progression and predicts poor outcomes in breast cancer patients

Abstract

Breast cancer invasion and metastasis result from a complex interplay between tumor cells and the tumor microenvironment (TME). Key oncogenic changes in the TME include aberrant synthesis, processing, and signaling of hyaluronan (HA). Hyaluronan-mediated motility receptor (RHAMM, CD168; HMMR) is an HA receptor enabling tumor cells to sense and respond to this aberrant TME during breast cancer progression. Previous studies have associated RHAMM expression with breast tumor progression; however, cause and effect mechanisms are incompletely established. Focused gene expression analysis of an internal breast cancer patient cohort confirmed that increased RHAMM expression correlates with aggressive clinicopathological features. To probe mechanisms, we developed a novel 27-gene RHAMM-related signature (RRS) by intersecting differentially expressed genes in lymph node (LN)-positive patient cases with the transcriptome of a RHAMM-dependent model of cell transformation, which we validated in an independent cohort. We demonstrate that the RRS predicts for poor survival and is enriched for cell cycle and TME-interaction pathways. Further analyses using CRISPR/Cas9-generated RHAMM^-/- breast cancer cells provided direct evidence that RHAMM promotes invasion in vitro and in vivo. Immunohistochemistry studies highlighted heterogeneous RHAMM protein expression, and spatial transcriptomics associated the RRS with RHAMM-high microanatomic foci. We conclude that RHAMM upregulation leads to the formation of 'invasive niches', which are enriched in RRS-related pathways that drive invasion and could be targeted to limit invasive progression and improve patient outcomes. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Keywords: breast cancer; cell division; extracellular matrix; hyaluronan; hyaluronan receptors; neoplasm invasiveness; tumor microenvironment; xenograft.

Figure 1.. RHAMM expression is correlated with clinicopathological features of aggressive breast cancers.

Human breast cancer tissues from the UMN breast cancer cohort were analyzed by Nanostring gene expression. Levels of HMMR mRNA were increased in association with: (A) lymph node (LN) positive status, (B) increased Nottingham tumor grade, (C) basal, HER2-enriched, and Luminal B subtypes by PAM50, and (D) triple negative and HER2+ disease by clinical hormone receptor expression. (E) Unsupervised hierarchical clustering of 94 human breast cancer cases with a 27-gene signature of RHAMM biologic activity enriches for more aggressive disease features. (F) Logistic regression analysis of T-stage of RHAMM high vs low patients. (G) Logistic regression analysis of T-stage of RRS high vs low patients. * = p<0.05, ** = p<0.01, *** = p<0.005. l

Figure 2.. RHAMM protein and RRS expression is heterogeneous across patient samples but correlates with aggressive breast cancer s.

Representative RHAMM high (A) and low (B) expressing tissues, scale bars 50 μm. (C-F) RHAMM IHC were analyzed with high RHAMM levels correlating with: (C) lymph node (LN) positive status, (D) increased Nottingham tumor grade, (E) basal, HER2-enriched, and Luminal B subtypes by PAM50, and F. triple negative and HER2+ disease by clinical hormone receptor expression. (G) Overview image of a TNBC tumor stained for pan-cytokeratin (green), CD45 (red), and dsDNA (blue). Regions of interest (ROIs) are numbered 1–23 with those in the tumor margin or tumor core highlighted. Serial sections of tumor in panel (G) stained for RHAMM at the tumor core (H) or the tumor margin, scale bar 500 μm (I). (J) Modified RRS using ROIs located at the tumor core or tumor margin. * = p<0.05, ** = p<0.01, *** = p<0.005.

Figure 3.. Characterization of RHAMM KO breast cancer cells.

(A) Representative immunoblot of MCF10DCIS.com parental, Ctrl, and RHAMM KO cell lines for RHAMM. (B) Quantitative-PCR expression of CD44 and RHAMM in MCF10DCIS.com parental, control and RHAMM KO cell lines. (C) Representative RHAMM immunofluorescence images of MCF10DCIS.com parental, control and RHAMM KO cell lines, in permeabilized and unpermeabilized conditions, insets highlighting mitotic spindle RHAMM staining. * = p<0.05, ** = p<0.01.

Figure 4.. RHAMM drives phenotypes associated with invasive breast cancer in MCF10DCIS (A-D) and MDA-MB-231 cell lines (E-H).

MTT proliferation assay of (A) MCF10DCIS.com or (F) MDA-MB-231 parental, control and RHAMM KO cell lines. Colony counts (size >50μm) of (B) MCF10DCIS.com or (G) MDA-MB-231 parental, control and RHAMM KO cell lines in soft agar. (C) MCF10DCIS.com parental, control and RHAMM KO cell lines embedded in Matrigel + 20% collagen + 10 ug/mL LMW-HA and scored for invasion, representative images of tumor spheres below. (D) Representative immunoblot of vimentin in MCF10DCIS.com parental, control and RHAMM KO cell lines. (E) Representative immunoblot of MDA-MB-231 Ctrl and RHAMM KO cell lines for RHAMM. (H) Quantification of transwell migration assay of MDA-MB-231 control (Ctrl) and RHAMM KO cells. * = p<0.05, ** = p<0.01, *** = p<0.005, **** = p<0.001.

Figure 5.. RHAMM drives breast tumor invasion.

(A) Tumor volumes for mice injected with MCF10DCIS.com parental, control (Ctrl), and RHAMM KO cells, * T-test Parental vs KO, # T-tests Ctrl vs KO. (B) Tumors from MCF10DCIS.com parental, control and RHAMM KO groups scored for proportion of invasive tumor area; 0 = intraductal tumor only; 1 = <20% invasive tumor area; 2 = 20–80% invasive tumor area; 3 = >80% invasive tumor area. (C) Size of invasive tumors from parental, control, and RHAMM KO cell lines. (D) MCF10DCIS.com parental tumors scored for invasion and then stained by IHC and scored for RHAMM expression (E) Representative images of invasive tumors stained for H&E and IHC for RHAMM and HA, scale bars 100 μm. ** = p<0.01, *** = p<0.005, **** = p<0.001.

Figure 6.. RRS correlates with poor outcomes in breast cancer patients

(A) Unsupervised hierarchical clustering of human breast cancer cases from the University of Nebraska breast cancer Collaborative Registry (BCCR). Kaplan-Meier plot of BCCR cases by RRS (B) or RHAMM (C) high or low for overall survival at 10 years. Kaplan-Meier plot of TCGA breast cancer cases by RRS (D) or RHAMM (E) high or low for overall survival at 10 years.