Upregulation of YKL-40 Promotes Metastatic Phenotype and Correlates with Poor Prognosis and Therapy Response in Patients with Colorectal Cancer

Abstract

YKL-40 is a heparin- and chitin-binding glycoprotein that belongs to the family of glycosyl hydrolases but lacks enzymatic properties. It affects different (patho)physiological processes, including cancer. In different tumors, YKL-40 gene overexpression has been linked to higher cell proliferation, angiogenesis, and vasculogenic mimicry, migration, and invasion. Because, in colorectal cancer (CRC), the serological YKL-40 level may serve as a risk predictor and prognostic biomarker, we investigated the underlying mechanisms by which it may contribute to tumor progression and the clinical significance of its tissue expression in metastatic CRC. We demonstrated that high-YKL-40-expressing HCT116 and Caco2 cells showed increased motility, invasion, and proliferation. YKL-40 upregulation was associated with EMT signaling activation. In the AOM/DSS mouse model, as well as in tumors and sera from CRC patients, elevated YKL-40 levels correlated with high-grade tumors. In retrospective analyses of six independent cohorts of CRC patients, elevated YKL-40 expression correlated with shorter survival in patients with advanced CRC. Strikingly, high YKL-40 tissue levels showed a predictive value for a better response to cetuximab, even in patients with stage IV CRC and mutant KRAS, and worse sensitivity to oxaliplatin. Taken together, our findings establish that tissue YKL-40 overexpression enhances CRC metastatic potential, highlighting this gene as a novel prognostic candidate, a predictive biomarker for therapy response, and an attractive target for future therapy in CRC.

Keywords: YKL-40; biomarkers; colorectal cancer; metastasis.

Figure 1

Knockdown and re-/overexpression of YKL-40 in HCT116 and Caco2 cells. (A) Knockdown via CRISPR/Cas9 approach is described. Human YKL-40 is located on chromosome 1. The exon 3 sequence is reported, the gRNA complement sequence is underlined, and the PAM sequence is indicated with red nucleotides. (B) Gel image of Genomic Cleavage Detection Assay using DNA extracted from transfected HCT116 and Caco2 cells. Control template and primers were included as a PCR technical control. Cleavage efficiency is indicated for each sample. (C) Map of pCMV-CHI3L1-GFPSpark plasmid used for YKL-40 gene expression. Validation of YKL-40 knockdown and re-/overexpression at (D) mRNA and (E) protein levels in both cell lines. qPCR data and densitometric data are shown as mean ± SD. Statistically significant differences were calculated using Student’s t test: *, p < 0.001; **, p < 0.01; ***, p < 0.0001. Data represent three independent experiments.

Figure 2

YKL-40 expression correlates with migration (A), invasion (B), and proliferation (C) in HCT116 and Caco2 colon cancer cell lines. Representative images of migratory (2× magnification; scale bar 500 µm) and invasive cells (10× magnification; scale bar 50 µm) are shown. Cell migration is expressed as wounding closure percentage at 0 h (T0), 24 h (T24), and 48 h (T48) after wounding. For both migration and invasion assays, pictures represent one of three independent experiments. Cell proliferation of HCT116 and Caco2 cell lines following YKL-40 gene expression modulation was assessed based on the results of the Resazurin assay. Data represent three independent experiments, performed in triplicate. Data are shown as mean ± SD. Statistically significant differences were calculated using Student’s t test: *, normalization vs. YKL-40.WT; *, p < 0.01; **, p < 0.001; ***, p < 0.0001; #, normalization vs. YKL-40.KD; #, p < 0.01; ##, p < 0.001; ###, p < 0.0001.

Figure 3

EMT marker expression regulation in HCT116 and Caco2 cells with modulated YKL-40 gene expression. qPCR data are shown as mean ± SD. Statistically significant differences were calculated using Student’s t test: *, normalization vs. YKL-40.WT; *, p < 0.01; **, p < 0.001; ***, p < 0.0001; #, normalization vs. YKL-40.KD; #, p < 0.01; ##, p < 0.001; ###, p < 0.0001. Data represent three independent experiments.

Figure 4

YKL-40 is overexpressed in CRC mouse model and can be used for differentiation of cancerous from normal mucosa. (A) qPCR analysis of YKL-40 mRNA levels in murine colon normal mucosa (N) (n = 5) and only-AOM- (Ctrl AOM) (n = 5) or only-DSS-treated (Ctrl DSS) (n = 5) mucosa and adenocarcinoma (AC) (n = 5). (B) IHC analysis of YKL-40 protein in murine dysplastic ACF, adenoma, and adenocarcinoma (AC) and its quantification. (C) qPCR analysis of YKL-40 mRNA levels in tumoral and matched non-tumoral tissues of CRC patient samples (n = 41) and (D) in samples stratified for CRC stages. (E) IHC analysis of YKL-40 protein in human colorectal tumoral and normal tissues and its quantification. (F) Measurement of YKL-40 protein levels both in normal individuals’ (n = 43) and in CRC patients’ (n = 43) sera and (G) in samples stratified for CRC stages. (A–G) Data are shown as mean ± SD. Statistically significant differences were calculated using Student’s t test: *** p < 0.0001; ** p < 0.001; * p < 0.01; § p = 0.1. (B,E) 20× and 40× magnification. Scale bars, 50 µm.

Figure 5

Diagnostic performance of YKL-40 mRNA levels in CRC tissues (H) and YKL-40 protein levels in CRC serum. Receiver operating characteristic (ROC) curve analysis in (A) CRC tissue samples (n = 19) and colon mucosa derived from cancer-free subjects (n = 19); area under the curve (AUC) = 85.6 (95% CI: 73.7–97.4%, p = 0.0002); (B) tumor serum samples (n = 43) and serum samples derived from cancer-free subjects (n = 43); area under the curve (AUC) = 78.7 (95% CI: 69.0–88.4%, p = 0.000004).

Figure 6

Kaplan–Meier survival curves of high YKL-40 (dashed line) versus low YKL-40 (solid line) for (A) all patients of cohorts 1–4 and cohort 6 and for (B) patients with advanced CRC stages from the same cohorts. Expression value thresholds for determining high and low groups were determined through maxstat R package, except for the analysis of all patients of TCGA, where it was determined with YKL-40 median expression threshold. p-values were calculated using log-rank tests. Tick marks represent censored data.

Figure 7

(A) Kaplan–Meier survival curves of YKL-40_high (solid line) versus YKL-40_low (dashed line) and EGFR_high (solid line) versus EGFR_low (dashed line) patients of cohort 5 (GSE5851). Survival curves of YKL-40_high (solid line) versus YKL-40_low (dashed line) patients belonging to the EGFR_high group. (B) Survival curves of YKL-40 for patients of cohort 5 with WT-KRAS and mutant KRAS. (C) Survival curves of YKL-40 for patients of cohort 4 (GSE40967) with WT-KRAS and mutant KRAS. (D) Survival curves of YKL-40_high (solid line) versus YKL-40_low (dashed line) patients of cohort 6 (TCGA-COAD). For all survival curves, p-values were calculated using log-rank tests, and expression value thresholds for determining high and low groups were determined through maxstat R package. Abbreviations, CTX, cetuximab; OX, oxaliplatin.