GABRD promotes progression and predicts poor prognosis in colorectal cancer

2.1 Human CRC samples and cell lines

This study was approved by the Shanghai Fifth People’s Hospital Institutional Ethics Committee (Ethical Approval Form Number: 2017-097) and adhered to the principles listed in the Declaration of Helsinki. Informed consent was obtained before the collection of tissues. Sixteen paired tumor and NTs were collected from CRC patients at the Shanghai Fifth People’s Hospital (Shanghai, China) between 2016 and 2018. The samples were snap-frozen in liquid nitrogen and stored at −80°C. The corresponding formalin-fixed and paraffin-embedded tissues were retrieved, and 4-µm tissue sections were prepared by the Department of Pathology at the same hospital. Tissue microarrays were prepared by Shanghai Outdo Biotech (Shanghai, China) and contained CRCs from 100 patients with a median follow-up of 30 months. Detailed information about these samples is summarized in Table S1.

Five human CRC cell lines COLO205, COLO320DM, HCT116, HT15, and HT29, as well as a colon epithelial cell line FHC were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in McCoy’s 5A or DMEM supplemented with 10% FBS, 100 µg/mL of penicillin, and 100 mg/mL of streptomycin at 37°C with 5% CO₂ in a humidified incubator (Thermo Fisher Scientific, Waltham, MA).

2.2 IHC

Sections were stained with a polyclonal antibody against GABRD (1:50 dilution; rabbit anti-human, mouse, and rat; abs141150; Absin Bioscience Inc., Shanghai, China) by IHC based on a standard protocol in the Department of Pathology at our institution [13]. Briefly, formalin-fixed and paraffin-embedded microarray sections were deparaffinized, dehydrated, and immersed in sodium citrate buffer (pH 6.0) for antigen retrieval, blocked with 3% hydrogen peroxide for 10 min at room temperature to inactivate endogenous peroxidase, rinsed with phosphate-buffered saline for 10 min, and pretreated in a microwave oven for 10 min. Then the slides were incubated at 4°C overnight with the primary antibody. After staining with a two-step plus Poly-HRP anti-Rabbit IgG Detection System (Elabscience Biotechnology Co. Ltd, Wuhan, China), the slides were visualized by reacting with the DAB chromogen (Biocare Medical LLC, Pacheco, CA, USA) counterstained with hematoxylin and covered with glycerin gel, and then photographed under a microscope. A modified H score system was used to semi-quantify GABRD expression, as previously described [14]. Briefly, the maximal intensity of staining (0, negative; 1, weak; 2, moderate; and 3, strong) was multiplied by the percentage of positive tumor cells (0–100%) to generate the modified H score (range: 0–300). The GABRD expression was classified into high or low by the median H score. Data interpretations were made independently by two pathologists who had been blinded to each other’s findings and to the original pathology reports.

2.3 Access to public datasets

We comprehensively searched transcriptomic datasets of CRCs in the GEO database [15]. Only those that compared gene transcription between NTs and CRCs, or between primary and metastatic CRCs were selected and further screened. Datasets that contained less than 50 tissue samples or provided incomplete information were ruled out. For repetitive datasets, after each data matrix was checked and compared with regard to sample ID and data type, only the one with most samples was kept. Ten datasets were finally included to compare the GABRD expression between NTs and CRCs, including GSE3629 [16], GSE6988 [17], GSE21510 [18], GSE28000 [19], GSE31279 [20], GSE37182 [21], GSE41258 [22], GSE44861 [23], GSE87221 [24], and GSE106582 [25]. In addition, the Colon Adenocarcinoma and Rectum Adenocarcinoma Projects of The Cancer Genome Atlas (TCGA-COAD and READ) were retrieved from UCSC Xena (https://xenabrowser.net/heatmap/) and combined into one CRC dataset.^[1] These 11 datasets include 707 NTs and 1358 CRCs. Besides, 15 datasets containing primary and metastatic CRCs, including GSE6988 [17], GSE18105 [26], GSE21510 [18], GSE27854 [27], GSE28722 [28], GSE29623 [29], GSE38832 [30], GSE40967 [31], GSE41258 [22], GSE41568 [32], GSE62322 [33], GSE64258 [34], GSE71222 [34], GSE81582 [35], and GSE81986 [36], were retrieved from GEO. These 15 datasets include 1607 primary and 581 metastatic CRCs.

2.4 Ectopic expression or silencing of GABRD, and transfection

Lentiviral plasmids expressing GABRD (using GV144 vector), short hairpin RNA (shRNA) oligos of GABRD (using GV248 vector), or respective controls were constructed by Shanghai Genechem Co., Ltd (Shanghai, China). The target sequences were CAGACACCATTGACATTTA (shGABRD-1), CTCATTTCAACGCCGACTA (shGABRD-2), TGACGATGACCACGCTCAT (shGABRD-3), GTTACTCATCGGAGGACAT (shGABRD-4), and TTCTCCGAACGTGTCACGT (scramble control). Transient transfection of cells was performed using Lipofectamine 3000 from Invitrogen (L3000015, San Diego, CA, USA) according to the manufacturer’s instruction. To select stable transfectants, puromycin with a concentration of 0.4 µg/mL was used (abs42025969, Absin Bioscience Inc., Shanghai, China). The medium containing puromycin was changed every two days to remove dead cells. Once establishment of stable transfectants was confirmed, puromycin treatment was reduced to 0.2 µg/mL.

2.5 RNA extraction and the quantitative polymerase chain reaction (q-PCR)

Tissue and cellular RNA extraction and q-PCR were performed, as previously described [37]. The sequences for q-PCR primers were: GABRD forward primer, 5′-GCATCCGAATCACCTCCACTG-3′; and GABRD reverse primer, 5′-GATGAGTAACCGTAGCTCTCCA-3′. The specificity of primers was validated by electrophoresis and sequencing. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal reference. Experiments were performed three times in duplicate.

2.6 Western blotting

Total cellular protein extraction and western blotting were performed, as previously described [37]. The following antibodies were used: a rabbit anti-GABRD polyclonal antibody (1:1,000 dilution; abs141150; Absin Bioscience Inc., Shanghai, China) and a rabbit anti-GAPDH polyclonal antibody (1:2,000 dilution; Beyotime Biotechnology, Shanghai, China) as the loading control.

2.7 Cell proliferation assay

Stably transfected HCT15 and HCT116 cells (5 × 10³ cells/well) were seeded in 96-well plates and cultivated overnight. Then proliferation assays were performed, as previously described [37].

2.8 Scratch wound healing assay

Stably transfected HCT15 and HCT116 cells (4 × 10⁵ cells/well) were seeded in 12-well plates and cultivated until 100% confluence. Then monolayer scratch wound healing assays were performed, as previously described [37].

2.9 Statistical analysis

Paired or unpaired Student’s t tests were used for continuous variables. The Fisher exact test and chi-square tests were utilized for categorical comparisons. Survival analyses were evaluated by the Kaplan–Meier method. Univariate and multivariate survival analyses were conducted with the Cox proportional hazards regression model. All tests and reported p values were two-sided, and p < 0.05 was defined as statistically significant. Statistical analyses were performed with the Microsoft Excel 2010 (Microsoft, Redmond, WA, USA), GraphPad Prism7 (GraphPad, San Diego, CA, USA), and SPSS software version 22 for Windows (SPSS Inc., Chicago, Ill, USA).

3.1 Expression of GABRD was increased in CRCs

As GABRD had not been investigated in CRCs previously, we first explored the expression pattern of GABRD in CRCs compared to that in NTs using 11 public transcriptomic datasets. As shown in Figure 1, GABRD was significantly upregulated in seven datasets (Figure 1a–g) and decreased in two datasets (Figure 1h–i), and was similar in two datasets (Figure 1j–k), in CRCs compared to that in NTs. We further compared the mRNA expression of GABRD between CRCs and NTs using clinically resected samples. The result confirmed that GABRD was increased in CRCs than in NTs (Figure 1l). We then wanted to make sure whether this was also the case in metastatic CRCs. In a panel of 15 transcriptomic datasets, no significant difference was observed with regard to GABRD expression in 14 of them in metastatic CRCs compared to that in primary CRCs (Figure S1a–n). While in one dataset that included different disease stages, GABRD expression was significantly increased in polyps and primary tumors compared to that in mucosae, and slightly but significantly decreased in metastatic CRCs compared to primary tumors, but was not significantly different between primary tumors and polyps (Figure 1o). Taken together, these results demonstrated that GABRD was upregulated in CRCs and might be involved in early tumorigenesis.

Figure 1

GABRD transcription was increased in CRCs. Expression of GABRD was compared between CRCs and adjacent NTs using transcriptomic data from one combined TCGA-COAD and READ dataset and 10 datasets in GEO. The results demonstrated that the GABRD expression was significantly increased in seven datasets (a–g), decreased in two datasets (h–i), and similar in two datasets (j–k) in CRCs compared to that in NTs. Sixteen paired CRCs and NTs from clinically resected samples were used to validate results from the transcriptomic datasets using q-PCR (l). Abbreviations: CRC, colorectal cancer; GABRD, gamma-aminobutyric acid type A receptor subunit delta; GEO, gene expression omnibus; NS, not significant; NT, adjacent normal tissue; q-PCR, quantitative polymerase chain reaction; TCGA-COAD and READ, the cancer genome atlas colon adenocarcinoma and rectal adenocarcinoma. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.2 Overexpression of GABRD predicted unfavorable patient prognoses in CRC patients

We next determined the relationship between the overexpression of GABRD and patient outcome. We first performed IHC on 16 paired CRCs and NTs from clinically resected samples. As shown in Figure 2a, GABRD staining was weak or negative in NTs and localized to the cytoplasm and membrane (Figure 2a and b). By contrast, GABRD expression was stronger and localized to the nuclei, cytoplasm, and membrane in CRCs (Figure 2c and d), consistent with those observed from the Human Protein Atlas database (http://www.proteinatlas.org/ENSG00000187730-GABRD/pathology). An evaluation of the sections using H score revealed significant difference between CRCs and NTs with regard to GABRD staining (p < 0.000, Figure 2b). We then evaluated the correlation between GABRD expression and clinicopathological variables. As shown in Table 1, overexpression of GABRD was significantly correlated with later pTNM stages. We next determined the prognostic role of GABRD in CRC patients using a TMA. As shown in Figure 2c, a high expression of GABRD was associated with unfavorable overall survival (OS) in these patients (estimated mean OS 44.9 [95% CI, 36.2–53.7] months vs 63.7 [95% CI, 54.5–73.0] months, log-rank p = 0.001). In the multivariate analysis using a Cox proportional hazards model, GABRD overexpression was significantly and independently associated with shorter OS, after adjustment by age, tumor size, and tumor stage (Table 2). Besides, GABRD overexpression was significantly associated with shorter OS and recurrence-free survival (RFS) in the combined TCGA-COAD and READ cohort, which supported the observation in the TMA cohort (Figure 2d and e). Collectively, these results demonstrated that the overexpression of GABRD was predictive of unfavorable prognoses in CRC patients.

Figure 2

Overexpression of GABRD was associated with poor patient prognosis. The protein expression and localization of GABRD were characterized by IHC in 16 paired NTs and CRCs. GABRD expression was weak or negative in NTs and mainly localized to the cytoplasm and membrane (a, a and b). By contrast, GABRD expression was stronger and localized to the nuclei, cytoplasm, and membrane in CRCs (a, c and d; b). The prognostic value of GABRD in CRCs was explored with a TMA and validated using the combined TCGA-COAD and READ dataset. Kaplan–Meier plots demonstrated that the overexpression of GABRD was associated with unfavorable patient OS in the TMA cohort (c) and TCGA cohort (d), and unfavorable patient RFS in the TCGA cohort (e). P values were obtained by using the log-rank test. Censored data are indicated by the + symbol. Patients were stratified into GABRD low and high expression according to the median H score or GABRD mRNA expression (<median vs ≥median). Abbreviations: CRC, colorectal cancer; GABRD, gamma-aminobutyric acid type A receptor subunit delta; IHC, immunohistochemistry; NT, normal tissue; OS, overall survival; RFS, recurrence-free survival; TCGA-COAD and READ, the cancer genome atlas colon adenocarcinoma and rectal adenocarcinoma; TMA, tissue microarray. ****p < 0.0001.

3.3 Overexpression of GABRD promoted the proliferation and migration of CRC cells

To further examine the role of GABRD in CRCs, we first examined the endogenous expression of GABRD mRNA and protein in a colon epithelial cell line FHC and five CRC cell lines using q-PCR and western blotting. The results demonstrated that the GABRD expression was dramatically increased in four of the CRC cell lines compared to that in FHC cells (Figure 3a and b). Two CRC cell lines were selected and transduced with vector, GABRD, scramble shRNA, or GABRD shRNAs, then underwent q-PCR and western blotting to detect the GABRD expression in these cells (Figure 3c and d). We next investigated the in vitro activities of GABRD using CCK-8 cell proliferation assay and monolayer scratch wound healing assay. As shown in Figure 4, overexpression of GABRD promoted while knock-down of GABRD impeded cell proliferation (Figure 4a and b) and migration (Figure 4c and d).

Figure 3

GABRD expression was upregulated in cultured CRC cell lines. The mRNA and protein expression of GABRD was investigated in a colon epithelial cell line FHC and five CRC cell lines using q-PCR and western blotting. The results demonstrated that the GABRD expression was dramatically increased in four of the CRC cell lines compared to that in the colon epithelial cell line FHC (a and b). Two CRC cell lines were selected and transduced with vector, GABRD, scramble shRNA, or GABRD shRNAs, then underwent q-PCR and western blotting to detect GABRD expression in these cells (c and d). Abbreviations: CRC, colorectal cancer; GABRD, gamma-aminobutyric acid type A receptor subunit delta; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; NS, not significant; q-PCR, quantitative polymerase chain reaction. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 4

GABRD promoted CRC cell proliferation and migration. HCT15 and HCT116 cells were transduced with vector, GABRD, scramble shRNA, or shGABRDs, and underwent CCK-8 cell proliferation assay and monolayer wound healing assay. Overexpression of GABRD promoted while knock-down of GABRD impeded cell proliferation (a and b) and migration (c and d) of these cells. Abbreviations: CCK-8, cell counting kit-8; CRC, colorectal cancer; GABRD, gamma-aminobutyric acid type A receptor subunit delta; NS, not significant. *p < 0.05, **p < 0.01, ***p < 0.001; #, ##, and ### represent p < 0.05, p < 0.01, and p < 0.001, respectively, for comparisons between scramble and shGABRD-transduced cells in (a and b). Data representative of three replicated experiments are given.

We next wanted to explore the possible mechanisms underlying GABRD’s role in progression of CRCs. A GSEA was performed using the TCGA-COAD dataset. As shown in Figure 5, a high expression of GABRD was positively correlated with hallmark gene sets defining epithelial–mesenchymal transition (EMT), angiogenesis, and hedgehog signaling (Figure 5a–c), as well as KRAS signaling, Wnt-β-catenin signaling, UV response, etc. (data not shown). We then conducted gene–gene correlation analyses using the same dataset. Figure 5d–l demonstrates that the expression of GABRD was correlated with those of EMT markers (except that of CDH1/E-cadherin) and VEGFA and its two receptors, but not with those of the key markers of the hedgehog signaling pathway (data not shown).

Figure 5

Potential mechanisms underlying the role of GABRD in CRCs. A GSEA was performed using the TCGA-COAD dataset to explore possible mechanisms underlying GABRD’s role in carcinogenesis and progression of CRCs. The results indicated that a high expression of GABRD was positively correlated with hallmark gene sets defining epithelial–mesenchymal transition, angiogenesis, and hedgehog signaling. Gene–gene correlation analyses demonstrated that GABRD was significantly correlated with several key genes associated with EMT and angiogenesis. Abbreviations: CDH1, cadherin 1; CDH2, cadherin 2; CRC, colorectal cancer; FDR, false discovery rate; GABRD, gamma-aminobutyric acid type A receptor subunit delta; GSEA, gene sets enrichment analysis; NES, normalized enrichment score; SNAI1, snail family transcriptional repressor 1; TCGA-COAD, the cancer genome atlas colon adenocarcinoma; TWIST1, twist family BHLH transcription factor 1; VEGFA, vascular endothelial growth factor A; VEGFR, vascular endothelial growth factor receptor; VIM, vimentin; ZEB1, zinc finger E-box binding homeobox 1.

Taken together, these observations suggested that GABRD promoted CRC progression by enhancing cell proliferation and migration, and interacting with crucial pathways involved in tumor progression.