RHAMM marks proliferative subpopulation of human colorectal cancer stem cells

Abstract The cancer stem cell (CSC) theory features typically rare self‐renewing subpopulations that reconstitute the heterogeneous tumor. Identification of molecules that characterize the features of CSCs is a key imperative for further understanding tumor heterogeneity and for the development of novel therapeutic strategies. However, the use of conventional markers of CSCs is still insufficient for the isolation of bona fide CSCs. We investigated organoids that are miniature forms of tumor tissues by reconstructing cellular diversity to identify specific markers to characterize CSCs in heterogeneous tumors. Here, we report that the receptor for hyaluronan‐mediated motility (RHAMM) expresses in a subpopulation of CD44+ conventional human colorectal CSC fraction. Single‐cell transcriptomics of organoids highlighted RHAMM‐positive proliferative cells that revealed distinct characteristics among the various cell types. Prospectively isolated RHAMM+CD44+ cells from the human colorectal cancer tissues showed highly proliferative characteristics with a self‐renewal ability in comparison with the other cancer cells. Furthermore, inhibition of RHAMM strongly suppressed organoid formation in vitro and inhibited tumor growth in vivo. Our findings suggest that RHAMM is a potential therapeutic target because it is a specific marker of the proliferative subpopulation within the conventional CSC fraction.


| INTRODUC TI ON
Heterogeneous tumor tissues are composed of various cancer cell subpopulations including cancer stem cells (CSCs), the phenotypically distinct subpopulation of cancer cells. [1][2][3] CSCs possess the self-renewal ability and multilineage differentiation ability that contribute to tumor heterogeneity. [4][5][6] Analysis of CSCs in clinical tissues necessitates their prospective isolation based on their specific cell surface markers. 7,8 In human colorectal cancer (CRC), cell surface markers such as CD44, 9,10 CD133, 11 LGR5, 12 and CD166 9,13 are reported as representative CSC markers. However, the use of these markers is still insufficient for the efficient enrichment of cells with CSC properties that show a self-renewal ability in vivo or in vitro.
In xenograft assays, the frequency of tumor-initiating cells among CD44+ colorectal cancer cells was shown to range from 1 out of 62-223 cells. 14 Regarding in vitro models based on clinical colorectal cancer tissues, the organoid-formation assay is an ideal experimental tool for phenotypic characterization of the self-renewal ability with recapitulating clinical specimens. [15][16][17] In our previous study, only 1.5% of CD44+ cells exhibited a self-renewal ability in an in vitro organoid model. 10 These findings suggest that conventional CSC markers of colorectal cancer are inadequate for the enrichment of bona fide CSCs. Therefore, the identification of robust surface molecules that specifically mark CSCs is warranted to understand tumor heterogeneity and develop novel therapeutic strategies.
There is a growing interest in the characterization of adult stem cells with proliferative character, 18,19 although they have long been regarded as slow-cycling cells capable of self-renewal and multilineage differentiation. 20 LGR5+ adult intestinal stem cells, the normal counterpart of the CSCs of the colon and rectum, can change their phenotype plastically from a proliferative state to the quiescent state. [21][22][23][24] Although the phenotypic character of isolated CSCs based on several surface markers has been fully elucidated, further investigations are still required to understand the cell cycle status of CSCs in colorectal cancer.
Accumulated evidence suggests that the molecular function of CSC markers contributes to their specific character. 25,26 Transmembrane glycoprotein CD44 is a widely accepted cell surface marker of CSCs in colon 9,10,14 gastric, 27 prostate, 28 and other solid tumors. 29 CD44 was shown to act as a ligand-binding receptor by interacting with extracellular matrix (ECM) elements such as hyaluronan; this was shown to enhance the CSC properties by participating in signal transduction through the formation of transmembrane complexes. 30 The receptor for hyaluronan-mediated motility (RHAMM) is expressed extracellularly and intracellularly in various tumor types; it was shown to form a complex with CD44 upon hyaluronan binding. 31 Extracellular RHAMM-CD44 partnering sustains CD44 surface display and enhances CD44-mediated signaling through ERK1/2 and SRC, which promote the invasion, cell motility, and proliferation of cancer cells. [31][32][33][34] These data suggest that the co-existence of RHAMM and CD44 on the cell surface is a major determinant of the properties of cancer cells.
Here, we sought to characterize the proliferative state of colorectal CSCs using transcriptomics of organoids and colorectal cancer tissues. We identified RHAMM as a cell surface marker that allows separation of the conventional CD44+ CSC fraction. The proliferative character of RHAMM+CD44+ cells suggests that these represent a phenotypically distinct subpopulation of the conventional CSCs that facilitates a further understanding of tumor biology and the development of novel therapeutic strategies.

| Xenotransplantation into immunodeficient mice
NOD/SCID/IL2Rγ null (NSG) mice 35 were purchased from Charles River Laboratories Japan. C57BL/6. Rag2 null IL2Rγ null mice with NODtype SIRPA (BRGS) mice were developed in our laboratory. 36 Both types of mice were bred and housed at Kyushu University. BRGS or NSG mice aged 4-20 weeks old were administered subcutaneous injection of cells resuspended in 50 μL of PBS and 50 μL of Matrigel.
Tumor growth was followed up to 6 months. Estimated tumor volume was calculated using the formula: tumor volume = 1/2 × length × width × width.

| Tissue processing and organoid culture
We obtained surgically resected colorectal cancer tissues from 48 patients. Table S1 summarizes the clinical characteristics of these

| Limiting dilution assay
For the limiting dilution assay, the cells were dissociated into single cells, and the sorted cells were diluted to the desired cell number.
Cells were injected subcutaneously into the flanks of immunodeficient mice, and the number of tumors formed out of the number of sites injected was scored to determine the frequency of cancerinitiating cells using ELDA software (http://bioinf.wehi.edu.au/softw are/elda/).

| Viral transduction of organoids
Cells were dissociated and filtered through a cell strainer to eliminate cell clumps. Cells were diluted in the medium mentioned above in 250-μL volumes and incubated with viral particles in a 24-well dish. After 24 h of transfection, cells were washed and resuspended in culture medium for organoid culture or injected into immunodeficient mice as described above. The vectors showed expression of GFP, which made it possible to determine the proportion of successfully transduced cells. Transduction efficiency of more than 95% was observed by fluorescence-activated cell sorting (FACS), excluding dead cells. Transduction efficiency was determined after culture and after the sacrifice of mice.

| Plasmids and reagents
HMMR-specific shRNA lentiviral vector was purchased from ORIGENE (TL312389). Lentiviruses containing shHMMR (RHAMM) were generated by transient transduction of the lentiviral vector together with the packaging of HEK293T cells with PAX2 and VSVG helper plasmids, as described. 10 The medium for the 293T cells was changed at 18 h after transfection, and the viral supernatant was collected, filtered, and stored at −80°C.

| RNA isolation, RT-PCR assay, and microarray analysis
Total RNA was extracted using TRIzol (Thermo Fisher Scientific; catalog # 15596026), in accordance with the manufacturer's instructions, and cDNA was generated using reverse transcriptase SuperScript Vilo (Thermo Fisher Scientific; catalog # 11755050).
Quantitative real-time PCR (qPCR) was carried out using the StepOnePlus Real-time PCR System. Reactions were run in triplicate in three independent experiments. The geometric mean of the housekeeping gene GAPDH was used as an internal control.
All reagents and instruments including probes, were obtained from Applied Biosystems, unless specified otherwise.
For microarray analysis, target populations ranging from 5000 to 10,000 cells were sorted directly into TRIzol. Total RNA was extracted and purified using the RNeasy Micro Kit (Qiagen; catalog # 74004). cRNA (1.5 μg) from each sample was hybridized to the Illumina BeadChip. Gene expression data were imported and estimated with quantile normalization using GeneSpring GX software (Agilent Technologies). The microarray data were deposited in the Gene Expression Omnibus (GEO) with the codes GSE100433 10 and GSE137919.

| Single-cell RNA-sequencing of organoids
Organoids were grown from a single cell in a 24-well plate. The

| Gene Set Enrichment Analysis (GSEA)
The Broad Institute provides a Java implementation of the GSEA method on its website. The gene sets were provided by MSigDB (http://softw are.broad insti tute.org/gsea/msigd b/index.jsp). GSEA was performed to analyze the enrichment of the gene sets following the developer's protocol. 37 Enrichment scores of pathways were calculated in GSEA and were listed in bar plots.

| Statistical analysis
Unless mentioned otherwise, the results are representative of replicated experiments and are presented as mean ± standard deviation from triplicate samples or randomly chosen cells within a field. All statistical analyses were performed using the JMP software (version 11.0; SAS institute). Unpaired two-tailed Student's t-test was used to compare two groups and one-way ANOVA was used to compare three or more groups. p-values <0.05 were considered indicative of statistical significance.

| Characterization of proliferative CD44+ human colorectal cancer stem cells
Cancer stem cells are phenotypically characterized by high tumor-initiating capacity and a high potential for organoid formation. 9,10,14,38 To further investigate the properties of CSCs, we evaluated our previously reported transcriptomic data pertaining to human colorectal CD44+ CSCs. 10 Gene expression of CD44+ CSCs and CD44− non-CSCs from primary colorectal cancer tissues (n = 4) and organoids (n = 4) purified by FACS were analyzed.
We performed GSEA to identify the biological characteristics and key molecules that define CSCs. Compared with CD44− cells, gene sets associated with cell proliferation were significantly enriched in CD44+ CSCs of primary tissues and organoids ( Figure 1A). To further identify the transcriptomic character of CSCs that across the systems, we focused on differentially expressed genes (DEGs) that were universally expressed in CD44+ CSCs from both primary tissues and organoids with a fold change (FC) cut-off ≥2.0. In particular, the representative cell proliferation marker (MKI67), proliferative cell nuclear antigen (PCNA), 39

| Single-cell RNA-sequencing identified RHAMM+ proliferative cell population in organoid
To elucidate the proliferative nature of CD44+ CSCs, we used organoid as a model in which intratumoral heterogeneity is preserved including CSCs. [43][44][45] We isolated individual cells from patient-derived organoids and generated a single-cell full-length transcriptome using SMART-seq. We profiled transcriptome of 568 cells from four patients  Figure 2F). Taken together, we identified an HMMR (RHAMM)+CD44+ proliferative subpopulation using organoids that recapitulated the intratumoral heterogeneity of human colorectal cancer.

| RHAMM+CD44+ cells represent proliferative character of CSCs
First, the cell surface expression of RHAMM was examined by FACS in 24 patients with surgically resected colorectal cancer tissues ( Figure 3A). The expression of RHAMM was detected in all cancer tissue specimens; the median frequency in these specimens (25.7%) was significantly greater than that in normal tissues (10.3%) ( Figure 3B). Interestingly, we were able to divide the CD44+ CSC fraction into CD44+RHAMM+ and CD44+RHAMM− fractions; the median frequency of these fractions was 7.1% (CD44+RHAMM+) and 10.3% (CD44+RHAMM−), respectively ( Figure 3C). To characterize the gene expression patterns of RHAMM+CD44+ CSCs, we purified 5000 numbers of RHAMM+CD44+ and RHAMM−CD44+ cells by FACS and subjected these to microarray analysis. GSEA revealed that the enrichment of cell cycle-related genes in the RHAMM+CD44+ fraction compared with the RHAMM−CD44+ fraction ( Figure 3D).

| RHAMM expression enriched the functional CSCs within CD44+ CRC cells
In our previous study, ~1%-2% of CD44+ cells among freshly isolated human colorectal cancer tissues were found to form organoids 10  for RHAMM−CD44+ cells ( Figure 4C; Table S3). Furthermore, representative CSC surface markers, such as CD133 and CD166, were highly enriched in RHAMM+CD44+ cells compared to the other fractions including RHAMM−CD44+ cells ( Figure 4D). These results collectively suggest that RHAMM is a marker that can enrich the functional CSCs among the CD44+ conventional CSC population.

| RHAMM is a potential therapeutic target in colorectal cancer
We next tested whether RHAMM is a functional molecule in CSCs.
To this end, we transduced two lentiviral short hairpin RNAs (shRNA) into organoids. We confirmed the sufficient reduction of HMMR expression ( Figure S2). Subsequently, we evaluated the effects of HMMR silencing on organoid formation. Significant inhibition of organoid-forming ability was observed in seven clinical colorectal cancer samples, which suggests a close involvement of RHAMM in maintaining the organoid-forming ability of CSCs, at least in vitro ( Figure 5A,B). RHAMM inhibition does not affect CD44 expression ( Figure S3A,B). Next, we transplanted the 5000 cells derived from organoids after 24 h transfection of shRNA and scrambled shRNA into immunodeficient mice. Knockdown of HMMR significantly reduced the tumor formation, which suggests that RHAMM inhibition impaired the tumor-initiating potential of CSCs in vivo ( Figure 5C).
To evaluate the chemosensitivity of RHAMM+CD44+ cells, we eval-  15,18,21,23,24 hair follicle, 19 and liver 22  in broad cancer types is imperative. TGFβ signaling is another pathway that potentially attenuates CSC properties, however we did not observe any effect on RHAMM expression ( Figure S5).
Given that intestinal stem cells dynamically change their phenotype between the proliferative state and the quiescent state, [21][22][23][24] further studies should investigate the mechanism by which the