Forkhead Box Q1 Expression Is Associated With Tumor Location of Right-Sided Colon, But Not With Acquisition of Oxaliplatin Resistance in Colorectal Cancer

Oxaliplatin (OHP) is a reagent for the standard treatment of advanced and recurrent colorectal cancer (CRC), although OHP resistance mechanisms are not fully elucidated. We found that OHP-resistant clones derived from HCT116, but not DLD1 were also resistant against the other drugs used for CRC treatment (5-uorouracil, OHP, and triuorothymidine) and their xenograft tumors were resistant against OHP treatment. Among the candidate genes derived from microarray analysis using the samples of OHP-resistant cells and their xenografts derived from HCT116, Forkhead box Q1 (FOXQ1) was further assessed for validation of OHP resistance and its association with clinicopathological features. Modication of FOXQ1 via siRNA knockdown and expression vector could not conrm the involvement of FOXQ1 in OHP resistance. In 173 CRC patients, FOXQ1 was upregulated in most CRC tumors compared to normal colonic mucosa. FOXQ1 expression was signicantly different by tumor location of the right-sided colon cancer compared with left-sided and rectal cancer. Moreover, expression level was signicantly associated with prognosis in advanced and recurrent patients. TCGA data also showed signicant association of FOXQ1 expression with tumor location. Our results indicated that FOXQ1 expression is associated with tumor location of right-sided colon, but not with acquisition of OHP resistance in colorectal cancer. with tumor location of RCC, which was considered prognostic marker in advanced and recurrent CRC patients 6–9 . In advanced and recurrent CRC patients of our cohort, FOXQ1 expression was associated with overall survival of RCC, which was not shown in stage IV patients of TCGA samples. This difference In TCGA samples, but not in our cohort, FOXQ1 expression was signicantly lower in MSI-H patients compared to MSS patients. In patients with KRAS or BRAF mutations, FOXQ1 expression was signicantly lower than those without mutations in both our cohort and TCGA samples. The differences between our cohort and TCGA samples include more frequency of stage IV and recurrent patients in our cohort. (Dojindo according to the manufacturer’s instructions. Half maximal inhibitory concentration (IC 50 ) values were calculated as the concentrations that corresponded to a 50% reduction in cellular proliferation compared with untreated cells. The proliferation rate was calculated as the doubling time of the cell number as measured by the cell counting kit. Experiments were performed independently at least three times, and data are shown as means ± standard deviations.


Introduction
Colorectal cancer (CRC) is one of the most common malignancies worldwide 1 .
Combination of anticancer drugs (oxaliplatin [OHP], irinotecan [CPT], and 5-uorouracil ) and antibodies targeting vascular endothelial growth factor and epidermal growth factor receptor (EGFR) has become the standard therapy, improving the prognosis of advanced CRC 2,3 .
However, advanced CRC is still a fatal disease because resistance against these anticancer drugs is common, despite the initial effectiveness of these drugs. Therefore, elucidating the mechanisms involved in drug resistance is indispensable for further improvements in the prognosis of CRC patients.
The mechanisms of resistance against OHP for advanced CRC treatment have been assessed to improve the prognosis of CRC 4 . However, no molecular marker is available for clinical use. Differing from molecular targeted therapy, cytotoxic reagents do not have target genes or proteins. Moreover, multidrug resistance after standard therapy for CRC is also a critical problem, although no molecular marker for multidrug resistance has yet to be identi ed for clinical use 5 .
Tumor location in the right-sided colon was reported as a prognostic marker before the era of molecular targeted therapy, although the background was unclear 6 . Recently, EGFR-targeted antibodies have been shown to be ineffective for right-sided colon cancer (RCC) patients regardless of the existence of RAS mutations [7][8][9] . However, the genetic background of patients with advanced RCC suffering from poor prognosis has not yet been elucidated 10,11 .
In this study, we established OHP-resistant clones from CRC cell lines (DLD1 and HCT116) to elucidate the mechanism of OHP resistance. Then, we assessed molecular markers associated with OHP resistance in HCT116-derived clones by microarray using both in vitro (cells) and in vivo (xenografts) samples, because the clones derived from HCT116 but not DLD1 also became resistant to the other drugs (5-FU, CPT and tri uorothymidine (TFT)). Among the identi ed genes, we selected Forkhead box Q1 (FOXQ1) as a candidate gene, because it was reported to be upregulated in CRC tissues and is associated with CRC progression and drug resistance [12][13][14][15][16][17][18][19] . Then, we performed validation of FOXQ1 involvement in OHP resistance and assessed the association of FOXQ1 with the clinical features of CRC patients treated at our department and TCGA samples.

Drug sensitivity and proliferation rate
The drug sensitivity of DLD1, HCT116, and OHP-resistant clones is shown in Table 1. IC 50 against the other drugs in   OHP-resistant clones derived from DLD1 were less than two times compared to those in DLD1, although there were  statistically signi cant differences between DLD1 and OHP resistant clones in limited cases. On the other side, IC 50 against the other drugs in OHP-resistant clones derived from HCT116 were more than two times compared to those in HCT116 except for HCT/OHP3 against 5-FU. There was statistically signi cant difference between HCT116 and OHP resistant clones except for HCT/OHP5 against TFT. Thus, OHP-resistant clones derived from HCT116 but not DLD1 had become multidrug resistant. This different pattern of drug sensitivity other than OHP suggested that DLD1 and HCT116 might use different mechanisms for acquisition of OHP resistance. This was also supported by principal component analysis of microarray data of these OHP resistant cells ( Figure S1). Therefore, the mechanism used for OHP resistance in HCT116 seems more critical than in DLD1, because this mechanism may induce multidrug resistance during treatments for the patients with advanced CRC.

Xenograft model of OHP-resistant clones
Unlike cell cultures, tumor growth was signi cantly different in tumors derived from HCT/OHP1 (P < 0.0001) and HCT/OHP3 (P < 0.0001) compared with those derived from HCT116, although tumor growth was similar between tumors derived from HCT11 and HCT/OHP5 (P = 0.63; Fig. 1A). These experiments have been performed twice and showed the same results. Thus, drug resistance induced slower tumor progression, although drug resistance was believed to enhance tumor progression. OHP treatment induced signi cant growth inhibition (P = 0.003) in HCT116derived tumors with a relatively low tumor growth inhibition (TGI) value of 0.37 (Fig. 1B). OHP treatment was ineffective in both HCT/OHP1-derived and HCT/OHP5-derived tumors with TGI values of 0.12 and 0.07, respectively.
CPT treatment induced signi cant growth inhibition in HCT116-derived tumors (P < 0.0001) with a TGI value of 0.8 (Fig. 1B). CPT treatment was statistically effective in HCT/OHP5-derived tumors (P < 0.0001) with a TGI value of 0.76. However, CPT treatment was not effective in HCT/OHP1-derived tumors (P = 0.17) with a TGI value of 0.53 ( Fig. 1C and   1D). These different sensitivities in vivo should be associated with the differences in CPT sensitivity in vitro (more resistance in HCT/OHP1 than in HCT/OHP5; Table 1).

Candidate genes responsible for OHP resistance
Eighty-eight genes showed more than two times higher expression in both three OHP resistant clones compared to HCT116 and two tumors derived from OHP resistant clones compared to tumor derived from HCT116 (Table 2). Twenty-nine genes showed less than half of expression in both three OHP resistant clones compared to HCT116 and two tumors derived from OHP resistant clones compared to tumor derived from HCT116 (Table S1). Among these genes, we considered FOXQ1 was the most suitable candidate gene because of its association with drug resistance and tumor progression 12,19 . Upregulation of FOXQ1 in OHP-resistant clones derived from HCT116 and tumors derived from these clones was con rmed by qRT-PCR (Table 3). However, upregulated FOXQ1 expression in tumors derived from OHP resistant clones was not associated with enhanced tumor growth ( Fig. 1C and 1D). In DLD1 and its OHPresistant clones, FOXQ1 expression has not changed by OHP resistance. (Table 3).  C6orf15  CTSD  FOXQ1  IGFL1  LGALS9  PCDH1  SNAR-G1   AKR1C1  CALB2  CXCL1  GATSL3  IKZF2  LGALS9C  PMEPA1  SOAT2   AKR1C3  CCL28  CXXC4  HERC6  IL15  LMO7  PPAP2B  SRPX2   ALCAM  CD274  CYP2J2  HOXA3  IL7  LOC100507165  PTHLH  STRA6   APOBEC3C  CFH  DMBT1  HOXB3  INHBB  LOC643072  RAB27B  TMEM164   APOBEC3D  CLIP4  DNER  HOXB8  ITGB2  MALT1  RPS6KA2  TMX4 APOBEC3F   (Table 5). However, tumor location was signi cantly associated with FOXQ1 expression (Table 5). FOXQ1 expression in RCC was signi cantly lower than left-sided colon (P = 0.006) and rectal cancer (P = 0.01). Microsatellite instability was not associated with FOXQ1 expression. KRAS mutation or BRAF mutation alone was not associated with FOXQ1 expression, although the combined status of KRAS mutation and BRAF mutation was associated with FOXQ1 expression (P = 0.02). In recurrent or stage IV CRC patients, overall survival (OS) was signi cantly better in those with a relative FOXQ1 expression of 20 ≤ than in those with < 20 (P = 0.0007; Fig. 2A). This signi cance was also shown when cases were limited to the right-sided colon (P = 0.014, Fig. 2B), but not left-sided colon (Fig. 2C) nor rectum (Fig. 2D). Then, the extent of FOXQ1 upregulation was associated with tumor location and poor prognosis in advanced cancer.

FOXQ1 expression in TCGA samples
We assessed FOXQ1 expression in colorectal samples to con rm our ndings in TCGA samples. Advanced clinical stage was not associated with FOXQ1 expression, although there was signi cant difference in FOXQ1 expression between stage II and stage IV. Tumor location was signi cantly associated with FOXQ1 expression (Table 6). FOXQ1 expression in RCC was signi cantly lower than left-sided colon (P < 0.0001) and rectal cancer (P = 0.0001).
Microsatellite instability was signi cantly associated with FOXQ1 expression (P < 0.0001). KRAS mutation, BRAF mutation, and the combined status of KRAS mutation and BRAF mutation were signi cantly associated with FOXQ1 expression. There was no signi cant difference in OS by extent of FOXQ1 expression. Then, signi cant association of FOXQ1 expression was shown with tumor location and combined status of KRAS mutation and BRAF mutation both in our department samples and TCGA samples.

Discussion
Here we assessed the mechanisms associated with OHP resistance in CRC cells. We found that OHP resistant clones derived from DLD1 and HCT116 showed different patterns of drug sensitivity against 5-FU, CPT, and TFT, which were used for CRC treatment after OHP resistance. We found that OHP resistant clones derived from HCT116 were also resistant against 5-FU, CPT, and TFT, although OHP resistant clones derived from DLD1 showed the same sensitivity as DLD1.Then, the mechanisms used in HCT116 seemed more critical than DLD1, because further treatments after OHP resistance maybe ineffective. These data indicated that there might be several mechanisms for OHP resistance in CRC.
Then, we further assessed the genes which have changed signi cantly in three OHP resistant clones compared to parental HCT116 cells both in vitro and in vivo using microarray analysis. Among the genes signi cantly changed in three OHP resistant clones compared to HCT116, we selected FOXQ1 as the candidate gene associated with acquisition of OHP resistance. This is because upregulation of FOXQ1 in CRC tissue and its association with tumor progression and drug sensitivity has been reported [12][13][14][15][16][17][18] . However, expression and in uence of FOXQ1 on CRC cell lines were not consistent in these reports.
In this study, inhibition of FOXQ1 expression did not change IC 50 of OHP in HCT116 and its OHP resistant clone, which was not consistent with previous report using SW480 cells 15 . Enhancement of FOXQ1 by expression vector also could not change IC 50 of OHP. Thus, we considered that upregulation of FOXQ1 in OHP resistant clones derived from HCT116 might result from OHP treatment but was not the cause of OHP resistance. These results also showed the limitations of exploring drug resistance mechanisms using drug-resistant cells.
Regarding in uence of FOXQ1 expression on tumor progression, tumor growth derived from OHP resistant clones was similar or downregulated compared to those derived from HCT116. These data were consistent 14,16 and inconsistent 12,18 with the previous reports, although upregulation of FOXQ1 was acquired by drug resistance in this study and by genetic modi cation in the previous reports. In clinical samples of our department and TCGA, FOXQ1 expression was not associated with clinical stage. These data are not inconsistent with the previous report 15 .
On the other hand, upregulation of FOXQ1 in CRC tissue compared with normal tissue was consistent with previous studies 12-15, 17,18 . Moreover, we found that FOXQ1 expression differed by tumor location both in our cohort and in TCGA samples. This is the rst report indicating the association of FOXQ1 with tumor location of RCC, which was considered prognostic marker in advanced and recurrent CRC patients [6][7][8][9] . In advanced and recurrent CRC patients of our cohort, FOXQ1 expression was associated with overall survival of RCC, which was not shown in stage IV patients of TCGA samples. This difference In TCGA samples, but not in our cohort, FOXQ1 expression was signi cantly lower in MSI-H patients compared to MSS patients. In patients with KRAS or BRAF mutations, FOXQ1 expression was signi cantly lower than those without mutations in both our cohort and TCGA samples. The differences between our cohort and TCGA samples include more frequency of stage IV and recurrent patients in our cohort.
Our study has several limitations. First, we have not elucidated the mechanisms of OHP resistance, although this study started this purpose at rst. FOXQ1 was not associated with acquisition of OHP resistance in DLD1 and HCT116 cells, and genes other than FOXQ1 have not yet been assessed. Second, we have not assessed the mechanisms associated with the differential expression of FOXQ1 by tumor location, although these may elucidate the reason of tumor location of RCC for the prognostic marker of CRC.
In conclusion, our study shows that several mechanisms are associated with acquisition of OHP resistance and FOXQ1 expression is associated with tumor location of RCC. Further study is necessary to elucidate these mechanisms.

Methods
Drugs and chemicals OHP, CPT, and 5-FU were purchased from NIPRO (Osaka, Japan), TOWA Pharmaceutical Co., (Kadoma, Japan), and KYOWA KIRIN (Tokyo, Japan), respectively. TFT was purchased from Tokyo Chemical Industry Co. (Tokyo, Japan) and dissolved in DMSO at a concentration of 20 mM. All drugs were diluted in culture medium immediately before use.

Cell lines and cloning of drug-resistant cells
Human CRC cell line DLD1 was purchased from Japan Health Sciences Foundation (Osaka, Japan). Human CRC cell line HCT116 was a kind gift from Dr. Yamamoto (Department of Surgery and Clinical Oncology, Osaka University Graduate School of Medicine, Osaka, Japan). These cells were authenticated by American Type Culture Collection using DNA pro ling (Manassas, Virginia, USA). Cells were maintained in DMEM supplemented with 10% fetal bovine serum, 10,000 units penicillin, 10 mg/ml streptomycin, and 25 µg/ml amphotericin B. Culture media and fetal bovine serum were obtained from Life Technologies Japan (Tokyo, Japan). All cells were grown at 37°C in a humidi ed incubator with 5% CO 2 . DLD1 and HCT116 cells were co-cultured with 20 µM OHP or 10 µM OHP, respectively. Then, To assess the tumor growth of xenografts derived from HCT116 and OHP-resistant clones, a total of 5 ⋅ 10 6 HCT116 cells and OHP-resistant clones (HCT/OHP1, HCT/OHP3, and HCT/OHP5 cells) were subcutaneously inoculated in the right and left anks of seven mice, respectively. Tumor size was measured twice a week. Tumor volume was calculated as a × b 2 (where a represents the tumor length and b represents its width).
To assess OHP resistance in xenografts, a total of 5 ⋅ 10 6 HCT116 cells and OHP-resistant clone HCT/OHP5 cells were subcutaneously inoculated in both anks of 12 mice (24 sites). Xenograft tumors derived from HCT/OHP1 cells were minced and inoculated into both anks of the 12 mice because the growth of HCT/OHP1-derived xenograft tumors was so slow compared with that of HCT116 and HCT/OHP5. Tumor size was measured twice a week and tumor volume was calculated as described above. When the tumor diameter was > 5 mm, the mice were randomized into no treatment (control) or drug treatment groups (5 mg/kg OHP or 10 mg/kg CPT injected intraperitoneally twice a week for a total of ve weeks, four mice [eight tumors] per each group) after adjusting the mean tumor volume among the groups. The rate of TGI was calculated as follows: 1 − (increase in tumor volume in the drug treatment group)/(increase in tumor volume in the control group). TGI was assessed after the completion of drug treatment.
Twenty-four hours later, total RNA was extracted using an RNeasy Mini kit (Qiagen K.K., Tokyo, Japan) according to the manufacturer's instructions. Total RNA was also extracted from human tumor xenografts derived from HCT116, HCT/OHP1, and HCT/OHP5 without drug treatment.
Gene expression pro les were analyzed by Agilent SurePrint G3 Human GE 8x60K v2 Microarray kit (Agilent). The data set is available at Gene Expression Omnibus under accession number GSE77932 for cell experiments and GSE124808 for tumor experiments.
Signal data were imported into GeneSpring (Agilent) for analysis. Signal evaluation was performed depending on signal uniformity and the signi cant difference between signal and background. Signal data were normalized by the 75th percentile among arrays. Genes for which signal data were ≥ 2 in OHP resistance clones compared to HCT116 both in vitro and in vivo were selected as upregulated genes. Genes for which signal data were 0.5 ≥ in OHP resistance clones compared to HCT116 both in vitro and in vivo were selected as downregulated genes.

Validation ofFOXQ1involvement in OHP resistance
Among the upregulated and downregulated genes, FOXQ1 was selected as a candidate gene because of its association with drug resistance and tumor progression [12][13][14][15][16][17][18][19] . FOXQ1 expression in cells and tumors was evaluated by quantitative real-time reverse transcription PCR (qRT-PCR), as previously described 17,20 . The primer sequences were FOXQ1-forward: CTTCCCTCCCCCCTAAGTACAT and FOXQ1-reverse: ATGCCACATACGTACACGGATG. GAPDH was used as an internal control. ΔCT was calculated using the CT (Threshold cycle) value of FOXQ1 and that of GAPDH in each sample. Data was calculated from triplicate wells.

Association of FOXQ1 expression with clinical features in CRC patients
Specimens were collected from 173 CRC patients who underwent surgery at our department. All protocols were approved by the ethics committee of Hyogo College of Medicine and all patients provided written informed consent (No. 0120 by the Institutional Review Board of Hyogo College of Medicine). All experiments were performed depending on the Declaration of Helsinki, the guidelines and the associated laws in Japan. The CRC specimens consisted of 155 tumors without chemotherapy, 13 tumors after chemotherapy including OHP, and ve after chemoradiotherapy. The CRC specimens were obtained with adjacent normal mucosal tissues for comparison and stored at − 80°C in RNAlater before use (Qiagen K.K.). Relative FOXQ1 expression was assessed by ΔΔCT generated from difference of ΔCT values in CRC tumors and normal mucosal tissue.

Association of FOXQ1 expression with clinical features in CRC patients of TCGA data
We further assessed FOXQ1 expression in CRC samples of TCGA. RNA-seq data of colorectal adenocarcinoma were collected from TCGA PanCancer Atlas in cBioPortal for Cancer Genomics (https://www.cbioportal.org/study/summary?id=coadread_tcga_ pan_can_atlas_2018). Differences between mRNA expression z-scores of FOXQ1 relative to normal samples and those of GAPDH relative to normal scores were used to evaluate association of FOXQ1 expression with clinical features as in CRC patients of our department.

Data analysis
Differences in IC 50 values and doubling times between OHP-resistant clones and parental cells were assessed by t-test.
Tumor volumes of xenografts at seven weeks after inoculation were assessed between each clone and HCT116 cells by t-test. The tumor volumes by OHP treatment were also compared between the control and treatment groups using a t-test. Association of FOXQ1 expression with clinical features was assessed by t-test in case of two categories and by analysis of variance in case of more than three categories. In uence of FOXQ1 expression on overall survival was assessed by Kaplan-Meier curve and evaluated by Log-rank test. A P value of < 0.05 was considered signi cant for all analyses. Declarations