https://orcid.org/0000-0002-3662-8815
Abstract
The pathogenesis of gallbladder cancer is complex, involving environmental and genetic risk factors. The matrix metallopeptidase 14 (MMP14) alters the tumor microenvironment and promotes tumorigenesis. In this study, we have characterized the role of the MMP14 promoter variants rs1004030 and rs1003049 in gallbladder cancer pathogenesis. Previously, we have shown the association of rs1004030 and rs1003049 with GBC and allele-specific differential expression of MMP14 in GBC patients. These variants reside within the cis-regulatory element (CRE) with high DNase and H3K4me3 signals, suggesting an active regulatory role in MMP14 expression. The luciferase-based reporter assay showed the role of promoter variants on expression levels in two GBC cell lines. Deleting the 119 bp promoter region surrounding the variants rs1004030 and rs1003049 by CRISPR-Cas9 genome editing resulted in reduced MMP14 expression in G415 cells. Electrophoretic mobility shift assay shows the presence of risk allele ‘C’/‘G’ at rs1004030 and rs1003049 and create binding sites for transcription factors SOX10 and MYB, respectively. Further, stable knockdown of these transcription factors in G415 and TGBC1TKB cells showed reduced expression of MMP14. However, in both GBC cells, ectopic expression of these transcription factors increased the expression of MMP14. Rescue of MYB and SOX10 expression levels showed a significant increase in luciferase activity only in risk allele-carrying constructs. In conclusion, our study unveils a mechanistic role of the MMP14 promoter variants rs1004030 and rs1003049 in gallbladder cancer.
Graphical abstract explaining the effect of SNP-dependent MMP14 gene expression in GBC, by creating respective binding site for SOX10 and MYB at rs1004030 and rs1003349.
Introduction
Gallbladder cancer (GBC) is the most common malignancy of all tumors originating from the biliary tract, with a very poor survival rate of six months in an advanced stage (1). More than 50% of GBC cases are found metastasized to lymph nodes at the time of diagnosis, which reduces the chance of complete surgical removal of the gallbladder (2). Mechanistically, in addition to the environmental risk factors, high frequency in certain geographic regions and ethnic groups suggests genetic predisposition in GBC (3). In fact, genome-wide association studies (GWAS) in high-risk populations, including India (4), Chile and Latin American (5) and Japan (6) have unraveled genetic contribution in GBC pathogenesis. Several candidate gene-based studies have suggested the role of genes related to lipoprotein metabolism (7–9), MMPs (10,11), DNA repair (12,13) and SERPINs (14,15). Study in the Indian population reported association of SNPs present in MMP2 (rs2285053, rs243865), MMP7 (rs11568818) and MMP9 (rs17577) with GBC (11). Most of these studies have focused on the genetic association, and there is limited understanding of the molecular mechanism. The high metastatic potential of GBC could be owing to its molecular makeup, which is also supported by the anatomical position of the gallbladder. Genetic variants can not only predispose for the development of GBC but also contributes to its metastatic potential. Unraveling these genetic variants and associated mechanisms might help develop targeted therapy and improve the overall prognosis of GBC patients.
The matrix metalloproteinases (MMPs) are the principal moderators of the tumor microenvironment during initiation, progression and metastasis (16). The availability of MMPs in the tumor microenvironment correlates with tumor aggressiveness and poor patient prognosis (17). Recent studies have shown specific upregulation of membrane type 1-matrix metalloproteinase (MMP14) in various cancer, viz. breast (18), oral (19), prostate (20), cervical (21), gallbladder (10) and lung (22). Invasion through the extracellular collagen basement primarily depends on MMP14, which is localized to invadopodia or leading edge of the cell membrane (23,24). The hemopexin domain (PEX) of MMP14 has been a critical region involved in the homodimerization and heterodimerization with cell surface adhesion molecule CD44. This cross-talk results in the phosphorylation of the EGF receptor and subsequent activation of downstream MAPK and PI3K signaling pathways involved in invasion, proliferation, inflammation and apoptosis (25). Although the role of MMP14 in tumorigenesis and metastasis is well established in various cancers, very little is known about GBC at the molecular level, which warrants further investigation.
We have previously reported the genetic variants present in the MMP14 as a risk factor for GBC (10). The MMP14 promoter characterization found binding elements for positive and negative regulators, including Sp-1 and PROX1 (26,27). Alternative alleles of SNPs may determine the presence or absence of transcription factor binding sites in the promoter. Previously off-target effect has been observed with small molecule-based inhibitors, which bind to the conserved catalytic domain across MMPs (28). Identification of transcriptional regulators of MMP14 can lead to the development of not only the biomarkers but also the specific therapeutic targets. The variants rs1004030 and rs1003349 present in the MMP14 promoter region are of significant importance as they are reported to modulate transcription factor binding sites (29). We have shown that patients homozygous for risk alleles ‘C’ and ‘G’ at rs1004030 and rs1003349, respectively, had a significant increase in MMP14 expression than those with allele ‘T’ at both the loci (10). These observations suggest the significant role of rs1004030 and rs1003349 in the expression of MMP14, which led us to investigate the functional implications in GBC progression further.
The current study focuses on understanding the molecular mechanism behind the regulation of MMP14 expression by the promoter variants rs1004030 and rs1003349. The luciferase reporter assay validated that the genomic sequence, encompassing the variants rs1004030 and rs1003349, has regulatory elements. Further, CRISPR-Cas9-mediated deletion of 119 bp genomic region flanking the variants rs1004030 and rs1003349 led to downregulation of MMP14 expression in G415 cells, suggesting these loci harbor regulatory elements. The transition from allele ‘T’ to ‘C’ at rs1004030 created a transcription factor binding site for SOX10 (SRY-Box Transcription Factor 10). Similarly, at rs1003349, the transversion from allele ‘T’ to ‘G’ formed a transcription factor binding site for MYB (proto-oncogene or c-Myb belongs to the myeloblastosis family of transcription factors). The binding of the transcription factors SOX10 and MYB to the MMP14 promoter elevated the expression of MMP14. Overall, our study emphasizes the regulatory role of MMP14 promoter genetic variants, which promote SOX10- and MYB-mediated expression of MMP14 in GBC pathogenesis. The dual advantage of this study is helping in understanding transcriptional regulation and defining the molecular basis for the genetic predisposition.
Results
Risk variants rs1004030 and rs1003349 may increase the expression of MMP14
To understand the regulatory role of loci consisting of the SNPs rs1004030 and rs1003349, we scanned the genomic region surrounding the variants in the UCSC genome browser containing Encyclopaedia of DNA Elements (ENCODE) data. Both the variants reside within the cis-regulatory element (CRE) with high DNase and H3K4me3 signals, suggesting an active promoter signature within the region (Fig. 1A).
Evaluation of the functional potential of promoter variants rs1004030 and rs1003349. The allelic effect of promoter variants rs1004030 and rs1003349 on reporter luciferase activity (firefly/renilla) was measured by cloning the 63 bp genomic region surrounding the variants into pGL4.23 vector. (A) ENCODE data representing the presence of variants rs1004030 and rs1003349 (indicated with lane) within cis-regulatory element (CRE) with high DNase and H3K4me3 region. (B) TGBC1TKB cells show a significant increase in luciferase activity with risk allele ‘C’/‘G’ (8.36 ± 0.43) compare to wild-type alleles ‘T’/‘T’ (3.51 ± 0.76) (P = 4.3E-06). (C) The G415 cells show a significant increase in luciferase activity in risk allelic combination ‘C’/‘G’ (9.73 ± 0.72) compared with wild-type alleles’ ‘T’/‘T’ (4.49 ± 0.25) (P = 3.55E-12). The experiments were performed in triplicate, and values are represented as mean ± SD. Student’s t-test was used to calculate statistical significance of difference between the groups, nsP > 0.05, **P ≤ 0.01, ****P ≤ 0.0001; ns, not significant.
Our initial study reported that GBC patients with risk alleles ‘C’ (30.1%) and ‘G’ (34.6%) at rs1004030 and rs1003349, respectively, had significantly increased MMP14 expression than patients with the ‘T’ allele at both the loci (10). In this study, we found rs1004030 ‘C’ allele in 51.9% and rs1003349 ‘G’ allele in 32.7% GBC patients. We hypothesized that these SNPs may alter the transcription factor binding site, affecting transcript levels. Previous studies in glomerulosclerosis (GS) have found the RR1-binding site and suggested enrichment of transcription factor binding sites in the genomic region surrounding variants rs1004030 and rs1003349, supporting our hypothesis (29–31).
To validate the same in the GBC cell lines, we cloned the promoter region surrounding rs1004030 and rs1003349 into luciferase minimal promoter vector pGL4.23 and transfected it into GBC lines G415 and TGBC1TKB. In both the cell lines, pGL4.23_MMP14_promoter fragment (wild type allele containing) showed a significant increase [G415 (P = 0.0001) and TGBC1TKB (P = 0.0011)] in luciferase activity than the minimal promoter vector pGL4.23, which highlights the regulatory role of the region (Fig. 1B and C).
To further validate the effect of individual SNP on transcription, we carried out luciferase assays with a combination of risk alleles. In the TGBC1TKB cell line, the combination of ‘C and G’ risk alleles showed a significant change (P = 0.0024) in luciferase activity compared with single risk allele combination ‘T and G’ at rs1004030 and rs1003349, respectively (Fig. 1B). However, in the G415 cell line, although the combination of alleles’ C and T’ (P = 3.80E-12) and ‘T and G’ (P = 2.26E-15) showed a significant difference in luciferase activity compared with ‘T and T’ yet, there was no additive effect of the two risk alleles (C and G) on luciferase activity (Fig. 1C). These observations also suggest that the risk allele at variant rs1004030 might affect MMP14 expression more than the rs1003349 but in a cell line-specific manner. This led us to explore further the role of any possible putative transcription factors which may affect the expression of MMP14 in an allele-dependent manner.
Risk alleles ‘G’ (rs1004030) and ‘C’ (rs1003349) create binding sites for transcription factors SOX10 and MYB
We scanned the surrounding region of rs1004030 and rs1003349 SNPs to explore putative transcription factors by in silico analysis of promoter using the JASPAR transcription factor database. In silico results showed that MYB and SOX10 bind to the region spanning rs1003349 and rs1004030 with alleles ‘G’ and ‘C’, respectively, but not with allele ‘T’. To validate the same, we performed electrophoretic mobility shift assays (EMSAs) using nuclear extracts from TGBC1TKB cells and 39 bp oligonucleotide identical to the genomic region spanning rs1003349 and rs1004030 with allele ‘G’ and ‘C’, respectively. We observed binding of a specific protein complex to the labeled oligos with allele ‘C’ (rs1004030) (Lane 2, Fig. 2A), but not with the labeled oligos with allele ‘T’ (Lane 5, Fig. 2A). The addition of unlabeled oligos comprising allele ‘C’ made the shift disappear, suggesting the binding complex is specific for allele ‘C’ (Lane 3, Fig. 2A). Further, we observed a supershift in the protein–DNA complex mobility in the presence of SOX10-specific antibody (Lane 4, Fig. 2A), confirming the specificity of the protein, bound to allele ‘C’ at rs1004030.
The risk allele ‘C’ (rs1004030) and ‘G’ (rs1003349) create binding sites for transcription factors SOX10 and MYB. EMSAs were carried out using allele-specific 39 bp biotin labeled or unlabeled oligonucleotides for rs1004030 and rs1003349 SNPs using nuclear extracts from TGBC1TKB cells. (A) The arrowhead indicates the shift in mobility (protein-DNA complex) with the biotinylated ‘C’ allele (lane 2). There is no shift with the ‘T’ allele (lane 5) at rs1004030. Lane 3 shows the competition between labeled and unlabeled probes (allele ‘C’) for protein binding, which leads to shift disappearance indicating the allele-specific protein–DNA interaction. The incubation of this complex with SOX10-specific antibodies leads to supershift (indicated with star, lane 4), indicating the allele-specific binding of SOX10 protein. (B) Bar graph for ChIP assay analysis of the genomic region surrounding rs1004030 shows chromatin enrichment for SOX10-binding site in wild-type G415 cells, G415 with SOX10 overexpression, G415 cells pulled with IgG negative control, and in 1% input. (C) Biotinylated probe with allele ‘G’ at rs1003049 shows shift in mobility (protein–DNA complex) indicated with arrowhead (lane 2), whereas the labeled probe with allele ‘T’ shows no shift (lane 5), indicating specific interaction of protein-DNA complex with allele ‘G’. The shift disappeared when competing with unlabeled probes with allele ‘G’ (lane 3). The supershift was observed at lane 4 following the incubation of this protein–DNA complex with antibody specific to MYB. Indicating allele–specific binding of transcription factor MYB. (D) Bar graph for ChIP assay analysis of the genomic region surrounding rs1003349 shows chromatin enrichment for MYB-binding site in wild type G415 cells, G415 with MYB overexpression, G415 cells pulled with IgG negative control, and in 1% input.
Similarly, the shift in electrophoretic mobility of labeled oligonucleotide showed binding of a protein with allele ‘G’ (rs1003349) (Lane 2, Fig. 2B) but not with allele ‘T’ (Lane 5, Fig. 2B). The competitive binding with an excess of unlabeled oligos comprising allele ‘G’ prompted the shift to disappear (Lane 3, Fig. 2B), suggesting the binding of the protein is specific to allele ‘G’. The bound protein complex is further challenged with MYB-specific antibodies (Lane 4, Fig. 2B), leading to supershift, suggesting the bound protein complex at rs1003349 comprises the MYB transcription factor. These findings confirm that the allele ‘C’ and ‘G’ at loci rs1004030 and rs1003349 facilitate the binding of transcription factors SOX10 and MYB, respectively.
To further confirm the allele-specific direct binding of SOX10 and MYB to rs1004030 and rs1003349 respectively, ChIP assays were carried out in G415 cells. As shown in Fig. 2B and D, there is significant enrichment of genomic region surrounding rs1004030 and rs1003349 following immunoprecipitation with antibodies specific to SOX10 and MYB, respectively, when compared with IgG control. This enrichment increased with the overexpression of SOX10 and MYB, which confirms the specificity of binding. Moreover, enrichment of MYB-binding site is more prominent compared with SOX10-binding site. Overall, this study further emphasizes in vivo binding of SOX10 and MYB to locus rs1004030 and rs1003349, respectively, and its putative role in expression of MMP14.
Perturbation in SOX10 and MYB expression changes MMP14 expression
Our understanding of the regulatory active promoter signature and binding of transcription factors within the regulatory region led us to investigate the SOX10- and MYB-mediated direct regulation of MMP14 expression. We perturbed SOX10 and MYB expression and checked the effect on MMP14 expression level, using G415 and HEK293T as these two cell lines have binding sites for SOX10 and MYB but not the TGBC1TKB (Supplementary Material, Fig. S1).
The stable knockdown of SOX10 and MYB significantly reduced the MMP14 expression in the HEK293T and G415 cells (Fig. 3A and B). Similarly, the ectopic expression of SOX10 and MYB in HEK293T and G415 led to a significant increase in the MMP14 expression in both the cell lines (Fig. 3C and D). These results further support the regulatory role of SOX10 and MYB expression levels on the expression of MMP14.
Expression levels of MYB and SOX10 regulate MMP14 expression. Knockdown and ectopic expression of MYB and SOX10 were carried out in HEK293T and G415 cells, and the effect on MMP14 expression level was determined by Western blot. (A and B) In the upper image, the knockdown of MYB (MYB_KD) in (A) HEK293T (P = 0.0002) and (B) G415 (P = 0.0054) cells show a significant reduction in MMP14 expression compared with control (SC). In the lower image, HEK293T (P = 0.0219) and G415 (P = 0.0006) cells with knockdown of SOX10 (SOX10_KD) show a significant reduction in MMP14 expression compared with control (SC). (C and D) In the upper image, the ectopic expression of MYB (MYB_Ex) in (C) HEK293T (P = 0.0052) and (D) G415 (P = 0.0283) cells shows a significant increase in the MMP14 level compared with control (EV) cells. In the lower image, HEK293T (P = 3.23E-05) and G415 (P = 0.0007) cells with increased expression of SOX10 (SOX10_Ex) show significant increase in MMP14 expression than control (EV). The experiments were performed in triplicate, and values are represented as mean ± SD. Student’s t-test was used to calculate statistical significance of difference between the groups, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.
Knockout of rs1004030 and rs1003349 containing promoter region reduces MMP14 expression and abolishes the effect of SOX10 and MYB on MMP14 expression
For further validation, we tested the notion that if SOX10 and MYB bind to the loci at rs1004030 and rs1003349, then the knockout of this region will not show any effect of SOX10 and MYB level perturbation on the MMP14 expression. We have generated a knockout cell line G415119−/− with 119 bp deletion surrounding the loci rs1004030 and rs1003349 using the CRISPR/Cas9 system (Fig. 4A and B). The single cell-derived colonies with homozygous deletion showed almost negligible expression of MMP14 protein (Fig. 4C). Transient overexpression of transcription factor SOX10 in the G415119−/− and corresponding control cells showed that the MMP14 protein expression remained drastically reduced in the knockout cells (G415119−/−) but not in the control cells (Fig. 4D). Similarly, the transient overexpression of MYB did not increase MMP14 protein expression in G415119−/− cells compared with G415 nondeleted cells (Fig. 4E).
Effect of SOX10 and MYB on MMP14 expression in promoter knockout cells. The CRISPR/Cas9-mediated deletion of 119 bp genomic sequence surrounding the variants rs1004030 and rs1003349 were done to validate SOX10- and MYB-binding sites. (A) The graphic view of the genomic region shows the positions of SNPs (allelic variation is in bold), the binding position of sgRNA1 and sgRNA2 (sequence in bold) and the arrowheads indicating the cleavage site for CRISPR/Cas9. (B) The agarose gel image shows PCR amplification of the target region in CRISPR/Cas9 deleted (del) and wild-type (WT) cells. (C) Western blot data showing expression of MMP14 in wild type (WT) and in cells with homozygous deletion of promoter fragment (MMP14_P Del-119bp). (D and E) The Western blots and corresponding densitometric graphs show MMP14 expression after perturbing the expression of SOX10 and MYB in cells with homozygous deletion of promoter fragment (MMP14_P Del-119bp) and wild-type cells. The experiments were performed in triplicate, and values are represented as mean ± SD. Student’s t-test was used to calculate statistical significance of the difference between the groups, **** P ≤ 0.0001.
These results further emphasize that the variants that reside within cis-regulatory active promoter signature facilitate the binding of transcription factors SOX10 and MYB to regulate the expression of MMP14.
SOX10 and MYB expression levels affect reporter gene expression in allele-specific manner
To further validate the specificity of the binding of SOX10 and MYB in an allele-specific manner and its effect on transcription regulation, we rescued the MYB/SOX10 expression depletion and checked the effect on reporter gene expression. We transfected HEK293T cells with altered MYB/SOX10 expression, using the luciferase constructs carrying the risk or wild-type alleles at SOX10- and MYB-binding sites. The HEK293T cells with stable MYB knockdown had no significant difference in the luciferase activity between the allele ‘G’ and allele ‘T’ of rs1003349. However, the HEK293T MYB knockdown cells with allele ‘G’ showed a significant increase in luciferase activity when rescued with transient overexpression of MYB, compared with the ‘C’ allele of rs1003349 (Fig. 5A).
Allele-specific effect of SOX10 and MYB expression levels on reporter gene expression. Allele specific binding of SOX10 and MYB to MMP14 promoter was further validated by luciferase assay. Luciferase activity was determined using allele-specific pGL4.23 constructs carrying 63 bp genomic region surrounding the variants rs1004030 and rs1003349 in HEK293T cells with SOX10 depletion (KD) and MYB depletion (KD) and with the rescue of SOX10 and MYB expression. (A) No significant change in relative luciferase activity (firefly/renilla) was observed in HEK293T cells with MYB depletion (MYB_KD) between the G allele (rs49_G) and T allele (rs49_T) groups. The MYB expression rescue (MYB_Ex) shows significantly increased relative luciferase activity in the allele ‘G’ (23 114 ± 4189) group compared with the allele ‘T’ group (10 859 ± 2921) (P = 0.0023). (B) No significant change in relative luciferase activity (firefly/renilla) was observed in HEK293T cells with SOX10 depletion (SOX10_KD) between the G allele (rs30_G) and T allele (rs30_T) groups. SOX10 expression rescue (SOX10_Ex) in HEK293T_SOX10_KD cells showed significant increase in relative luciferase activity (firefly/renilla) with risk allele ‘G’ (13 368 ± 129) compared with allele ‘T’ (10 136 ± 928) (P = 0.0517). The experiments were performed in triplicate and values are represented as mean ± SD. Student’s t-test was used to calculate statistical significance of difference between groups, nsP > 0.05, *P ≤ 0.05, **P ≤ 0.01; ns, not significant.
Similarly, we validated the binding of SOX10 by transfecting luciferase constructs in HEK293T cells with SOX10 knockdown. We observed no significant change in luciferase activity between risk allele ‘C’ and wild-type allele ‘T’ of rs1004030. However, the rescue with overexpression of SOX10 led to a significant increase in the luciferase activity between the two groups (Fig. 5B). These results suggest that SOX10 and MYB bind to the MMP14 promoter in an allele-specific manner and regulate the expression.
MMP14 increases tumorigenic properties of GBC cell lines
To evaluate the role of increased MMP14 expression on tumorigenic properties, we made GBC cell lines with stable overexpression of MMP14 to carry cell-based assays (Supplementary Material, Fig. S2). We examined the effect on cell proliferation using MTS and colony formation assays. Higher MMP14 expression significantly increased the proliferation ability of G415 and TGBC1TKB cells (Fig. 6A and B). Similarly, G415 and TGBC1TKB cell line colony-forming ability increased with MMP14 overexpression (Fig. 6C and D). To evaluate the effect on the metastatic potential of cells via migratory property, we did wound healing and transwell-invasion assays in Boyden chambers. We observed a significant increase in wound healing of both the cell lines with ectopic MMP14 expression compared with the controls (Fig. 6E and F). Similarly, an increase in expression of MMP14 led to a significant increase in transwell invasion of both the cell lines (Fig. 6G and H).
Effect of MMP14 expression on tumorigenic properties of GBC cells. The role of MMP14 on cellular properties was studied by cell-based assays in two GBC cell lines, G415 and TGBC1TKB. (A and B) The MTS proliferation assay in G415 (P = 8.05E-10) and TGBC1TKB (P = 6.15E-05) cells shows increased proliferation in the MMP14 ectopic expression (Ex) group compared with the control (EV) group. (C and D) The colony formation assay in G415 (P = 0.0001) and TGBC1TKB (P = 0.0009) cells; the colony-forming ability significantly increased in cells with MMP14 ectopic expression (Ex) compared with control (EV) cells. (E and F) The wound healing assay in G415 (P = 0.0290) and TGBC1TKB (P = 0.0093) cells show fast healing at 24 hours compared with the control EV group, and cells show a significant increase in wound healing in MMP14 Ex cells compared with control EV cells. (G and H) The transwell invasion assay in G415 (P = 3.33E-05) and TGBC1TKB (P = 1.93E-05) cells; the MMP14 ectopic expression (Ex) cells show increased invasion compared with control (EV) cells. The experiments were performed in triplicate, and values are represented as mean ± SD. Student’s t-test was used to calculate statistical significance of difference between groups, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.
We studied the MMP14 expression of 29 GBC patients by immunohistochemistry. The MMP14 expression was high in tumor tissue compared with adjacent normal tissue (Fig. 7A). Interestingly, a gradual increase in MMP14 expression was observed in patients with low to higher tumor grades, which emphasized the significance of MMP14 in advanced GBC (Fig. 7B and D). Tumor or tumor stroma tissue of more GBC patients had high or moderate expression than uninvolved (nonaffected tissue) (Fig. 7C).
The expression and distribution of MMP14 protein in gallbladder tumor tissue and adjacent normal tissue. The immunohistochemistry analysis of gallbladder tissue and adjacent normal tissue. (A) The patient samples show increased expression of MMP14 in gallbladder tumor tissue compared with adjacent normal tissue samples (P = 0.0022). (B) The grade-wise distribution of MMP14 expression shows a gradual increase with advanced tumor grade (P = 0.0054). (C and D) Percentage distribution of Allred scores in tumor, tumor stroma and normal tissue and representative images of gallbladder cancer tissue immunohistochemistry staining with different degrees of staining from weak or negative, intermediate and strong. The scale bar represents 50 μm (magnification 10×, objective 10×). (E and F) Graphs show expression correlation (Pearson) of MMP14 with MYB (E) and SOX10 (F) in gallbladder cancer from GEO datasets (GSE138772, GSE132223 and GSE139682). Statistical test used was unpaired nonparametric two-sided Mann–Whitney test, unpaired nonparametric one-way Kruskal–Wallis test and Pearson correlation analysis. nsP > 0.05, **P ≤ 0.01, ****P ≤ 0.0001.
Further, we used GBC expression combined data from three GEO datasets (GSE138772, GSE132223 and GSE139682) to study correlation between SOX10, MYB and MMP14 expression. The patients showed a positive correlation of MYB and SOX10 expression levels with MMP14 levels (Fig. 7E and F). Collectively, these results indicate the role of MMP14 expression in promoting tumorigenic cellular properties in G415 and TGBC1TKB cells.
MYB inhibitor treatment reduces the tumorigenic properties in GBC cells
To evaluate the potential therapeutic strategies for patients carrying risk alleles for genetic variants rs1003349 and rs1004030, we carried out cell-based assays in G415 cells by treating MYB inhibitor Monensin sodium (1 μM, PubChem SID: 24896992) (32,33). MYB inhibitor treatment significantly reduced the expression of MMP14 in a dose-dependent manner (Supplementary Material, Fig. S3). The inhibition of transcription factor MYB significantly reduced the cell proliferation and colony formation ability of G415 cells (Fig. 8A and B). We found significant reduction in migration and invasion efficiency of cells with MYB inhibitor treatment compared with untreated cells (Fig. 8C and D), which shows that it reduced metastatic potential of G415 cells.
Effect of MYB inhibition and CRISPR-Cas9-mediated deletion of transcription factor binding site on tumorigenic properties of G415 cells. (A) The graph shows MTS assay based OD values for G415 cells treated with MYB inhibitor (MYBi Treated, 1μM, Monensin sodium) and the control (DMSO) group (P = 2.45E-05). (B) The image and corresponding bar graph show the number of colonies formed in MYBi Treated and control (DMSO) groups (P = 2.33E-05). (C) The images and corresponding graph show wound closure in MYBi Treated and control (DMSO) G415 cells (P = 3.61E-05). (D) The images and corresponding bar graph show transwell invasion ability of MYBi Treated and control (DMSO) G415 cells (P = 0.0009). (E) The graph shows MTS assay based OD values for G415 cells with promoter deletion (MMP14_P Del-199bp) and wild-type cells (P = 0.0001). (F) The image and corresponding bar graph show number of colonies formed in G415 cells with promoter deletion (MMP14_P Del-199bp) and wild-type cells (P = 4.52E-04). (G) The images and corresponding graph show wound closure in G415 cells with promoter deletion (MMP14_P Del-199bp) and wild-type cells (P = 0.0022). (H) The images and corresponding bar graph show transwell invasion ability in G415 cells with promoter deletion (MMP14_P Del-199bp) and wild-type cells (P = 0.0001). The experiments were performed in triplicate, and values are represented as mean ± SD. Student’s t-test was used to calculate statistical significance of the difference between the groups. **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.
Deletion of transcription factor binding site reduces the tumorigenic properties in GBC cells
We modified the promoter to remove the binding site for transcription factors and evaluated the tumorigenic and metastatic potential of G415 cells. The cells with promotor deletion showed significant reduction in cell proliferation and colony formation ability compared with wild type cells (Fig. 8E and F). Similarly, cells with promotor deletion showed significantly reduced migration and invasion property compared with wild-type cells (Fig. 8G and H).
Discussion
MMPs have diverse functions in physiological processes, including inflammation, wound healing and organogenesis and in diseases like cancer, affecting tumor proliferation, invasion and apoptosis (34). Aggressive metastatic tumor cells modulate nearby extracellular matrix to facilitate invasion into stromal cells, tissues and lymph nodes (35). The primary sites for matrix remodeling are actin-rich invadopodia at the cell base (36,37). These structures are composed of various cytoskeleton proteins, proteases, kinases, GTPases and MMPs (36–43). MMP14 is overexpressed and colocalized to invadopodia in many cancers and is a key to stromal remodeling (44), which emphasizes the role of MMP14 during cancer initiation and progression. In our earlier study, we learned that the expression of MMP14 is high in GBC patients, and the same was validated in the TCGA database. Further to evaluate the potential role of increased MMP14 expression on GBC cellular properties, we carried out various cell-based assays and found increased tumor cellular properties such as wound healing, colony formation, proliferation and invasion in two GBC lines, G415 and TGBC1TKB.
The decades of experimental and clinical research have emphasized the role of MMP14 in tumor invasion and metastasis along with normal functions such as wound healing, inflammatory response and angiogenesis (45,46). So far, many studies have targeted specific MMPs by using function-blocking antibodies, cation zinc chelating hydroxylamine small molecules and other small molecule inhibitors (47–49). NSC405020 (3,4-dichloro-N-(1-methylbutyl)benzamide), a small molecule MMP14 inhibitor bound explicitly to the hemopexin domain, shows a significant reduction in the breast cancer tumor xenograft size (50). DX-2400, a human MT1-MMP antibody, represses metastasis in a breast cancer xenograft mouse model and sensitizes tumor response to radiation therapy (51,52). Despite the strong preclinical evidence, many of the trials were unsuccessful in controlling tumor burden and elevating patient survival (46,53–55). One reason is off-target effects owing to high structural similarities within each class of metalloproteinases and acquired resistance during treatment (55,56). Most targeted therapies directly block or inhibit specific MMP’s activity (57). However, focusing on regulation at the transcription level and targeting upstream signaling molecules may overcome small molecule inhibitors’ resistance and off-target effects. Defining the role of transcription factors that modulate metalloprotease expression is likely to provide new prospects to understand its regulation and function and provide more insight into targeted therapies.
Genetic predisposition is well studied in GBC owing to its high frequency in certain geographic regions and ethnic groups (58). Multiple genetic variants in matrix metalloproteinases (MMP’s), DNA repair genes and lipid metabolism-related genes have been shown to alter the risk for GBC, yet most of these studies are association-based, and the mechanism of genetic predisposition is not understood (59). The genetic predisposition and increased expression of MMP14 have been reported in various chronic inflammation-induced cancers (60,61), such as gastric adenocarcinoma (62), hepatocellular carcinoma (63), colorectal carcinoma (64), cholangiocarcinoma (65) and gallbladder cancer, ovarian cancer and (66) oral cancer (67).
Our initial study on MMP14 promoter variants reported a significant association of rs1004030 and rs1003349 minor alleles with increased risk alleles for GBC. The ENCODE data suggests both the variants reside within the cis-regulatory element (CRE) with high DNase and H3K4me3 signals, emphasizing the presence of an active promoter signature within the region. The earlier report suggests two regulatory-binding sites, for Sp1 and RR1, flanking SNPs rs1003349 and rs1004030, which show allele-dependent regulation of transcription of MMP14 in rat mesangial cells (29). This implies the presence of functional SNPs within the promoter region, which may have a direct or indirect regulatory role in gene expression by facilitating the binding of enhancer or repressor elements.
This study characterized the functional role of genetic variants rs1003040 and rs1003349 within the MMP14 promoter region. An in silico promoter analysis using the JASPAR database found MYB and SOX10 as putative transcription factors at loci rs1003349 and rs1004030, respectively. We validated the same by EMSAs, luciferase reporter assays and CRISPR-cas9-based deletion of 119 bp genomic region surrounding the locus and in which loci rs1004030 and rs1003349 harboring allele ‘C’ and ‘G’ found to facilitate binding of transcription factor SOX10 and MYB, respectively.
MYB (proto-oncogene or c-Myb) is overexpressed in leukemias, breast, colon, adenoid cystic carcinoma and osteosarcoma (68,69). The transcriptomic gene enrichment functional analysis also reveals genes involved in regulating the stress response, cell adhesion and cell differentiation or morphogenesis (70,71). Patients with gallbladder disease also show oxidative stress in gallbladder mucosa (72). The cells under stress regulate the MYB protein level and elevate transcriptional activity through post-translational SUMOylation (small ubiquitin-like modifier protein) (73). MYB expression protects colorectal carcinomas (CRC) cells from cisplatin and doxorubicin-induced apoptosis and promotes NOX1-mediated p38 MAPK pathway (74).
SOX-10 overexpression was reported in triple-negative breast cancer (TNBC) (75), bladder cancer (76) and nasopharyngeal carcinoma (NPC) as a differential diagnostic marker for metastasis and survival outcomes (77). The reduced expression of SOX10 in the bile duct TCGA cohort might be owing to frequent mutation and low sample size.
Moreover, SOX10 transcription factors play a critical role in the differentiation and function of neural crest cell (NCC) lineage, a highly migratory stem cell population (78). Given the similarity of cancer metastasis and NCC migration, it is predicted that MMP14 plays a significant role in cancer metastasis and neural crest stem cells’ epithelial to mesenchymal transition (79,80). SOX10 facilitates the binding of TCF4 to β-catenin and forms a stable SOX10/TCF4/β-catenin complex and promotes Wnt/β-catenin signaling in HCC (81,82). SOX10 expression levels in melanoma are regulated in a β-catenin-dependent manner via the canonical Wnt signaling pathway (83). β-Catenin in noncancer cells prevents membrane localization of MMP14 and thus its proteolytic activation of pro-MMP2. However, β-catenin increases the MMP14 expression and promotes Wnt-3a-mediated signaling via transcription factors Tcf-4/Lef in cancer cells (84). c-Myb interaction with β-catenin promotes invasion and metastasis of breast cancer by activating Wnt/β-catenin/Axin2 signaling (85). MYB also mediated activation of Wnt signaling and increased the MYC expression in colorectal cancers (CRC) (86), which depicts the complex regulation of MMP14 activity in cancer and noncancer cells mediated by transcription factors MYB and SOX10 in cooperation with Wnt signaling. However, transgenic or xenograft animal models might help understand the molecular mechanism in an in vivo system. The patient’s prognosis and expression studies in larger patient samples will help identify a biomarker for invasive advanced gallbladder tumors.
In conclusion, our study emphasizes the regulatory role of MMP14 promoter variants, which promote SOX10- and MYB-mediated expression of MMP14 in GBC pathogenesis.
Materials and Methods
Study participants
The study approval was taken by the institutional ethical committee of the National Institute of Science Education and Research (NISER) (NISER/IBSC/2017/32) and from the Acharya Harihar Regional Cancer Centre (AHRCC) (062/IEC/AHRCC), Cuttack. The cases recruited were the individuals with GBC enrolled for diagnosis and treatment.
The inclusion criteria for patients consist of all ages and gender should be residents of Odisha. We included 29 GBC (17 had adjacent nonaffected gallbladder tissue) patients diagnosed with GBC and confirmed by one of the diagnostic methods, such as histopathology or any image-based diagnosis (CT, ultrasound and MRI) of GBC were included. Patients with any other disease and cancer were excluded. From all the study subjects, informed written consent was collected. The study adhered to the Declaration of Helsinki.
Cell culture
The Human embryonic kidney cells (HEK293) were obtained from NCCS, India. The two GBC cell lines, G415 and TGBC1TKB, were obtained from Riken BRC cell bank, Japan. Both HEK293 and TGBC1TKB cells were cultured in Dulbecco’s Modified Eagle Medium (HiMedia) with 10% FBS (HiMedia). G-415 cells were cultured in RPMI-1640 medium (HiMedia), supplemented with 15% FBS (HiMedia). All cell lines were maintained at 37°C with 5% CO2 in a cell-culture incubator (Eppendorf).
Western blot
The whole-cell protein lysates were prepared using ice-cold RIPA buffer (Thermo Scientific) and protease and phosphatase inhibitor (Sigma). The total protein concentration was estimated by BCA Protein Assay Kit (Thermo Scientific). The lysates were separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred on to polyvinylidene fluoride (PVDF) membrane (Millipore). The PVDF membrane was blocked using 5% non-fat skim milk for an hour and subsequently incubated with rabbit-raised primary antibodies against MMP14 (dilution 1:1000, Abcam), MYB (dilution 1:1000, Cell Signaling) and SOX10 (dilution 1:1000, Cell Signaling) at 40C overnight. Blots were incubated for 1 hour in HRP conjugated anti-rabbit IgG secondary antibody (dilution 1:20000) (Abgenex) at room temperature. The probed protein was detected by a SuperSignal West Femto kit (Thermo Scientific), and signals were recorded by the ChemiDoc XRS+ system (Bio-Rad). The intensity of bands was analyzed by normalizing each sample with the GAPDH signal (dilution 1:5000) (Abgenex) using ImageJ software.
Luciferase reporter assay
The pGL4.23 (Promega) firefly luciferase reporter vector with minimal promoter was used for the luciferase reporter assay. pGL4.74 (Promega) renilla luciferase was used as an internal control to normalize luminescence signals. We commercially procured 63 bp long MMP14 promoter sequence (GRCh38.p12, chr-14, NC_000014.9, g.22836454G > T and g.22836440 T > C) with SNPs at the center, and the KpnI and XhoI restriction sites at the ends (Supplementary Material, Table S1). The promoter inserts with each allelic combination were cloned into the KpnI-XhoI (NEB) site of the pGL4.23 vector. The cotransfection of cloned pGL-4.23 (1 μg) and renilla vector (pGL4.74, 10 ng) was carried out in HEK293T, TGBC1TKB and G415 cells with MYB and SOX10 expression perturbation backgrounds. The transfection was carried out using lipofectamine 3000 (Invitrogen) in a 24-well plate at 50% cell confluency. The lysate was prepared according to the manufacturer’s instructions (Dual-Luciferase Reporter Assay System, Promega). The luciferase signal was recorded using a Varioskan Flash Multimode reader (Thermo Scientific). All experiments were independently repeated thrice.
Genome editing and preparation of CRISPR/Cas9 constructs
The deletion of 119 bp MMP14 promoter region flanking variants rs1004030 and rs1003349 in G415 cells was carried out by the CRISPR/Cas9 system as described earlier (87). The set of sgRNA was designed (https://chopchop.cbu.uib.no/; last accessed on June 12, 2019) targeting 119 bp, spanning promoter variants, with minimum off-target effect (Supplementary Material, Table S2). The sgRNA oligonucleotides were procured commercially (Integrated DNA Technologies [IDT]) and cloned into the BbsI restriction site of pSpCas9(BB)-2A-Puro (PX459) (Gift from Feng Zhang, Addgene #62988).
The G415 cells were cotransfected with sgRNA1 and sgRNA2-containing CRISPR construct at 70% confluency using lipofectamine 3000 (Invitrogen). 24-hour post-transfection, the cells were subjected to puromycin (1 μg/μl in complete media) based selection. Single-cell colonies were isolated and grown further. The genomic DNA was isolated from single-cell derived colonies using DNeasy Blood & Tissue Kits (Qiagen). The homozygous deletion was confirmed by PCR and Western blot. The confirmed clones with homozygous deletion were used for subsequent experiments.
Electrophoretic mobility shift assay
We scanned the MMP14 promoter region using JASPAR database for putative transcription factor-binding elements around rs1004030 and rs1003349 variants, using the default relative profile score threshold at 80% (https://jaspar.genereg.net/, last accessed on March 24, 2019). We commercially procured 39-mer oligonucleotides centered at rs1004030 and rs1003349 corresponding to all allelic combinations, with and without 5′-end biotin labeling (IDT). A list of probe sequences is provided in Supplementary Material, Table S3. The probes were annealed by incubating at 95°C for 5 minutes and gradual overnight cooling. The nuclear protein was isolated from TGBC1TKB cells using the NE-PER kit (Invitrogen). The concentration was estimated by BCA protein assay kit (Invitrogen). LightShift Chemiluminescent EMSA Kit (Invitrogen) was used to carry out EMSA as per the manufacturer’s instructions. The protein–DNA-binding reactions were performed using 2 μg of nuclear protein extract and 20 fmol of 5′ biotinylated annealed double-strand oligonucleotides. The binding specificity was confirmed by challenging it with the 200-fold excess (4 pmol) of unlabeled double-stranded oligonucleotides. The supershift assay was carried out by incubating the final binding reaction mix with 2 μg of MYB and SOX10 antibodies (Cell Signaling) for 20 minutes. The DNA–protein complex was run on 8% native polyacrylamide gel in TBE (0.5X) buffer. The complex was then transferred to a nylon membrane and post-UV cross-linking, the membrane was detected using a chemiluminescence detection kit (Invitrogen).
Chromatin immunoprecipitation assay
The chromatin immunoprecipitation (ChIP) assay was carried out using ChIP Kit as per manufacturer’s instructions (Abcam). In brief, cells were collected and fixed with 4% buffered formalin. Cell lysis was carried out by using buffer A and buffer B solutions provided in the kit. The cells were centrifuged and resuspended in buffer C followed by addition of buffer D/PI mix. The resulting DNA was sheared with a ultra sonicator (Cole-Parmer) to produce optimal DNA fragments size of 100–1000 bp. The sheared DNA was incubated overnight with 5 μg of ChIP grade antibodies including SOX10 (Cell Signaling Technology), MYB (Cell Signaling Technology) and rabbit IgG (Abgenex) as negative isotype control. Thereafter, we mixed Protein A Sepharose beads for immunoprecipitation. Following immunoprecipitation, DNA was purified and the fold enrichment of target region was determined by real-time PCR using specific primer sets (Supplementary Material, Table S3).
Generation of stable knockdown and overexpression cells
The oligonucleotide sequence corresponding to MYB shRNA (TRCN0000288659) and SOX10 shRNA (TRCN0000018985) was commercially procured (IDT) (Supplementary Material, Table S4) and cloned into pLKO.1-TRC cloning vector (Gift from David Root, Addgene #10878) (88). The pmiRA1-MYB (Gift from Jianjun Chen, Addgene #141120) (89), FUW-Sox10 (Gift from Bob Weinberg, Addgene #36978) (90) and pLPCX2-MT1-MMP-eGFP (Gift from Shengyu Yang, Addgene #89819) (91) expression vectors were procured from the Addgene repository. The cells were transfected with the constructs at 70% confluency using lipofectamine 3000 (Invitrogen) and subjected to puromycin (Sigma) selection 24-hour post-transfection. The single-cell colony clones were propagated, and the overexpression and knockdown levels were confirmed by Western blot. The stable knockdown of MYB, SOX10 and corresponding controls are called MYB_KD, SOX10_KD and Control_SC, respectively. The transient expression of MMP14 in G415 and TGBC1TKB cells and corresponding controls are referred to as G415_MMP14_Ex, TGBC1TKB_MMP14_Ex and Control_EV, respectively.
Cell proliferation assay
The 3000 cells per well were plated in 96-well plates and grown in complete media for 12, 36 and 60 hours. To assess the cell proliferation, each well was treated with 20 μl MTS reagent CellTiter 96 Aqueous One Solution (Promega) and incubated at 37°C for 1 hour. Following the incubation, the absorbance was recorded at 490 nm using a Varioskan Flash multimode reader (Thermo Scientific).
Wound healing assay
0.3 × 106 cells from G415_MMP14_Ex, TGBC1TKB_MMP14_Ex and the corresponding control cell lines were seeded in a 12-well plate. At 90% confluency a fine uniform scratch was made at the center of each well using a 200 μl pipette tip. The scratch images were captured at 0-, 12- and 24-hour time points using an inverted microscope (Nikon).
Transwell invasion assay
Millicell cell culture inserts (Millipore, 8-μm pore size) were coated with growth factor reduced matrigel (1 mg/ml, Corning). 0.3 × 106 cells from G415_MMP14_Ex, TGBC1TKB_MMP14_Ex and the corresponding control cell lines were seeded in the top chamber of the insert and subsequently placed in 12-well plates with 1.3 ml of complete media. After 24 hours, the media was decanted, and the inserts were washed twice with 1× PBS. Cells were fixed by 4% paraformaldehyde and stained with 1% crystal violet solution. From the upper chamber, nonmigratory cells were wiped off using a cotton swab. The imaging of migrated cells was carried out using an upright brightfield microscope (Olympus), in 5 fields per chamber at 4× magnification, and the average of five fields was used for further analysis.
Colony formation assay
G415_MMP14_Ex, TGBC1TKB_MMP14_Ex and the corresponding control cell lines were seeded into a 6-well plate at 2000 cells/well density. After forming visual colonies, the cells were washed, fixed with 4% paraformaldehyde for 20 min and stained with 1% crystal violet (MP Biomedicals). Colony imaging was carried out in the ChemiDoc XRS+ system (Bio-Rad).
Immunohistochemistry
The GBC tissues collected were fixed in 10% formalin and subsequently embedded in paraffin blocks. Tissue sections of 4 μm thickness were fixed on poly-L lysin-coated slides. The sections were subjected to gradual deparaffinization and rehydration. A microwave oven was used for heat-induced epitope retrieval in a high pH target retrieval solution (Dako). Immunostaining of tissue sections was carried out according to the manufacturer’s instructions using Envision Flex mini kit (Dako). MMP14 primary antibody was used at dilution 1:150 (Abcam). HeLa cell blocks were used as a positive staining control. The staining intensity scores were graded as ‘0—negative’, ‘1- weak’, ‘2- intermediate’ and ‘3- strong’. The percentage of positively stained cells was categorized as ‘0 for 0%’, ‘1 for 1%’, ‘2 for 2–10%’, ‘3 for 11–33%’, ‘4 for 34–66%’ and ‘5 for ≥67%’. The Allred score is the sum of staining intensity and percentage of positive cells, categorized as ‘0–1’ for negative, ‘2–3’ for weak, ‘4–6’ for moderate and ‘7–8’ for strong (92). Scoring was carried out by a pathologist using an upright light microscope (CX31, Olympus).
Statistical analysis
All continuous data are reported as mean ± SD. A 2-tailed unpaired Student’s t-test compared the means in group-wise data. The IHC Allred score, a ranked data, was analyzed by unpaired nonparametric two-sided Mann–Whitney test and Kruskal–Wallis one-way analysis of variance. The GBC genome wide expression profile data was downloaded from Gene Expression Omnibus (GEO) and Pearson correlation analysis was carried out using RStudio (93). The statistical analysis and graphs were generated using Microsoft Excel and GraphPad Prism 6.01. All the cell-based assay images were analyzed by NIH ImageJ software. P-value ≤ 0.05 was considered statistically significant in all the tests.
Acknowledgements
The authors thank the AHRCC Hospital staff and study participants for their contribution and consent.
Conflict of interest statement. We declare no conflict of interest.
Funding
Department of Atomic Energy, Government of India (SBS/PLAN/NISER/DAE/Govt. of India); Fellowship from NISER/DAE/Govt. of India to V.J. and A.P.
Data Availability
The data underlying this article will be shared on reasonable request to the corresponding author.
Authors’ contributions
Vinay J: Major contributing author performed majority of the in vivo and in vitro experiments, data curation, formal analysis, methodology, writing original draft, review and editing manuscript. Ananya Palo: Performed ChIP assay (Fig. 2B and D). Kusumbati Besra: Analysis and scoring of IHC samples. Manjusha Dixit: Conceptualization of the whole project, planned and guided the project, formal analysis, supervision, funding acquisition, resources, review, editing and project administration.
