Decreased expression of ubiquitin-specific protease 10 correlates unfavorable prognosis in colorectal cancer

Background: Ubiquitin-specific proteases (USPs) play an important role in fundamental cellular processes. Among these, USP10 is known for its association with tumor development and progression of multiple cancers. Here, we found a potential link between USP10 and p14ARF in colorectal cancer. Methods: USP10 and p14ARF protein expression was assessed via immunohistochemistry (IHC) on a tissue microarray from 280 colorectal cancer cases. IHC scores were evaluated by digital image analysis and compared with patients’ outcomes. In addition, we examined DNA hypermethylation in colorectal cancer cell lines and tissues, which were matched with adjacent normal colon samples. Results: USP10 expression (USP10 loss ) was lost in 18.6% of samples (52/280 cases), which was linked to lymphovascular invasion ( p =0.019) and distant metastases ( p <0.001). Similarly, loss of p14ARF expression (p14ARF loss ) was associated with more advanced tumors. USP10 expression correlated positively with p14ARF expression ( r =0.617, p <0.001). USP10 loss , p14ARF loss , and loss of both USP10 and p14ARF (USP10 loss /p14ARF loss ) were significantly associated with shorter disease-free survival and overall survival in comparison to USP10 intact , p14ARF intact , and USP10 intact /p14ARF intact , respectively. Multivariate analysis revealed that USP10 loss (Hazard ratio=2.07, p =0.046) and USP10 loss /p14ARF loss (Hazard ratio=1.41, p =0.010) are independent prognostic factors for disease-free survival in colorectal cancer patients. Furthermore, aberrant hypermethylation of the USP10 promoter region was found in colorectal cancer cell lines and tissues. Conclusions: The present results suggest that USP10 loss is a potential prognostic marker for colorectal cancer.


Background
Colorectal cancer is the fourth cause of cancer-related death worldwide [1]. The 5-year survival rate of metastatic colorectal cancer remains at a disappointing 10-20% despite significant improvements in metastatic colorectal cancer management via targeted therapies in combination with chemotherapy [2,3]. In the past two decades, many molecular markers and clinicopathological factors of colorectal cancer have been suggested as predictors of the postoperative prognosis of colorectal cancer patients to help clinicians determine appropriate treatments [4]. However, personalized therapy era requires that the heterogeneous group of colorectal cancer patients be classified in more detail. Thus, further research into new molecular biomarkers and clinical prognostic factors is essential to developing novel therapeutic strategies.
Ubiquitin-specific proteases (USPs) degrade substrates and modulate transcriptional regulation, which plays a key role in tumorigenesis [5,6]. Among the USPs, ubiquitinspecific proteases 10 (USP 10) is a primary cytoplasmic deubiquitinase that acts as a tumor suppressor gene or oncogene as it regulates different substrates, including p53 [7], AMP-activated protein kinase (AMPK) [8], fms-like tyrosine kinase 3 (FLT3) [9], and phosphatase and tensin homolog (PTEN) [10]. Furthermore, the elevation or loss of USP10 expression has already been linked to poor prognosis in various cancers, including stomach [11], breast [12], prostate [13], glioblastoma [14], hepatocellular [15], and nonsmall cell lung cancer [16]. Ko et al. recently reported that USP10 plays a crucial role in oncogene-induced senescence by regulating the stability of p14ARF and that dual loss of USP10 and p14ARF is a factor indicating an negative outcome in patients with non-small cell lung cancer [17]. We have also recently observed that the dual loss of USP10 and p14ARF expression predicts a worse prognosis for small intestinal adenocarcinoma patients [18]. The evaluation of oncogene-induced senescence-related proteins in clinical samples could lead to the development of novel therapeutic strategies that could improve patient outcomes. Currently, the clinical and prognostic value of USP10 and p14ARF expression in patients with colorectal cancer is largely unknown. In this study, we examined the clinicopathological features of USP10 and p14ARF in colorectal cancer patients. Moreover, we examined the hypermethylation of the USP10 promoter region, in colorectal cancer cell lines and primary tumors, for its down-regulating epigenetic mechanism in protein expression.

Patients and tumor specimens
We initially included 300 colorectal cancer specimens resected at Kangbuk Samsung

Pathologic diagnosis and generation of tissue microarray
The resection specimens were fixed in neutrally buffered 10% formalin solution and grossly examined by pathologists. Representative sections of tumor were embedded in paraffin blocks and the glass slides were used for pathologic diagnosis. During the microscopic diagnosis, tumor differentiation, pT stage, pN stage, lymphovascular invasion, and peritumoral dysplasia were recorded. After microscopic review, the tumor area was marked on each glass slide. Representative tumor cores 2 mm in diameter were obtained from the 280 colorectal cancer tissues, and 5 tissue microarray (TMA) blocks were generated.
The stained TMA slides were made into images using a NanoZoomer 2.0 HT scanner (Hamamatsu Photonics K.K., Hamamatsu, Japan). The immunohistochemical staining images were analyzed using Visiopharm software v6.9.1 (Visiopharm, Hørsholm, Denmark) to determine the percentage of positive cells.

DNA methylation analysis
For the methylation analyses, we used genomic DNA from colon cancer cell lines (HCT116, RKO, HT29, SW480, DLD1, COLO 320, SW48, Lovo, Caco-2, and SW620) from a previous study [19]. Other genomic DNA was isolated from normal colon tissues (n = 13) and colorectal cancer primary tissues (n = 13) (Supplementary Table S1) using a standard phenol-chloroform method. A methylation-specific polymerase chain reaction was used for to analyze methylation, as previously described [18]. Briefly, the EZ DNA Methylation Kit™ (Zymo Research, Orange, CA) was used to modify 2ug of DNA. Polymerase chain reaction (PCR) amplification of the bisulfite-converted DNA was performed with previously reported primer pairs specific for either the unmethylated or methylated sequence [18]. PCR amplicons (262-bp) were separated on 2% agarose gels and visualized by ethidium bromide staining. For the bisulfite sequence analysis, PCR amplicons were cloned using the TOPO TA cloning kit (Invitrogen, Carlsbad, CA). Plasmid DNA was isolated and purified from positive clones (10-15 from each sample), using a NucleoSpin Plasmid Isolation Kit Cox regression analyses were utilized to calculate the hazard ratios (HRs) of risk factors in univariate and multivariate conditions. A modified random survival forest (RFS) analysis was conducted as described previously [20,21]. Briefly, we built a clinical RFS model using lymph node (LN) metastasis and pT stage, and then combined it with molecular-level features (USP10 and p14ARF). The RSF models were built using R software ver. 3.1.3 (R Foundation, Vienna, Austria; http://www.R-project.org). The concordance-index (C-index) was calculated using the "surv-comp" R package. A C-index of 0.5 is as good as a random guess. p < 0.05 was considered significant.

USP10 and p14ARF protein expression
In the immunohistochemical staining, the normal colonic crypt showed cytoplasmic expression of USP10 and cytoplasmic and nuclear expression of p14ARF (Fig. 1).
Colorectal cancer tissue revealed the same expression pattern as normal colonic tissue.
Interestingly, the methylation level of the USP10 promoter region has not been actively investigated in human cancers. Accordingly, we examined USP10 promoter methylation levels in colon cancer cell lines, normal tissues, and primary colorectal cancer tissues using methylation-specific PCR and bisulfite sequencing. MSP analysis showed that USP10 was frequently hypermethylated in 10 colon cancer cell lines and primary colorectal cancer tumor tissues but not in normal colon tissues (Fig. 3A). These data suggest that promoter hypermethylation of USP10 is cancer-specific. We also confirmed the methylation status of USP10 with bisulfite sequencing analysis. The USP10 promoter region exhibited dense hypermethylation in both colon cancer cell lines and colorectal cancer tissues (77-83% of all CpG sites), but normal tissues exhibited relatively low methylation levels (30-35% of total CpG sites) (Fig. 3B).  (Table 3).

Discussion
We found that loss of USP10 and p14ARF protein expression correlates with worse prognosis in colorectal cancer patients, and we confirmed a positive correlation between the molecules. Interestingly, USP10 loss and USP10 loss /p14ARF loss were significant markers of poor prognosis in patients without LN metastasis at the time of surgery, suggesting that USP10 loss /p14ARF loss could predict disease progression in node-negative colorectal cancer. In addition, the methylation analysis revealed that the promoter region for USP10 is frequently hypermethylated in colorectal cancer cell lines and colorectal cancer primary tumors but not in normal colon tissues.
USP10 deubiquitinates p53 via MDM2, which regulates cellular p53 via reverse translocation and degradation of the protein [7]. Also involved in the deubiquitination and stabilization of PTEN in lung cancer cells, USP10 has been reported to be a tumor suppressor [10]. In addition, is an important mediator in the c-Myc-USP10-p14ARF axis because it deubiquitinates and stabilizes p14ARF [17]. Lu et al. also suggested that USP10 behaves as a tumor suppressor in hepatocellular carcinoma [15]. They demonstrated that USP10 inhibits the mTOR signaling pathway to deter cell growth in hepatocellular carcinoma. Further evidence that corroborates our results indicates that USP10 is a tumor suppressor in various cancer types, including gastric cancer [11], hepatocellular carcinoma [15], non-small cell lung cancer [17], small intestinal adenocarcinoma [18], and ovary cancer [25]. These studies suggest that the loss of USP10 expression is significantly correlated with poor patient outcomes. On the contrary, Ouyang et al. proposed that USP10 plays an oncogenic role in colon cancer [26]. They reported that USP10 promotes the expression of the oncogenic factor Musashi-2 (MSI2) by preventing its proteasomedependent degradation. In addition, they demonstrated that USP10 promotes colon cell proliferation by deubiquitinating MSI2. Other evidence supports USP10 as an oncogenic factor in other cancers, including prostate cancer [13], glioblastoma multiforme [14], and breast cancer [12]. Thus, the role of USP10 in cancer progression is controversial. One of the main reasons for this inconsistency in the data could be that cell conditions, such as the genetic mutation of tumor protein 53 (TP53), dictate the role of USP10 as a tumor suppressor or oncogenic factor [7]. It is well-known that TP53 encodes a 53-kDa phosphoprotein and is commonly inactivated in a wide range of tumors, including colorectal cancer. Although we did not assess it in this study, the mutation rate of TP53 is reported to be about 50% in colorectal cancer. Notably, the prevalence of the TP53 mutation in colorectal patients depends on multiple factors, including tumor stage, location, and the status of hypermutation [27]. Therefore, further studies are needed to clarify the association between the functional role of USP10 and the TP53 genotype in colorectal cancer.
Although several previous studies reported that USP10 is associated with tumor suppression in colon cancer, it remains unclear why USP10 expression is often downregulated in human cancers. In terms of the tumor suppressive role of USP10 in cancer, USP10 could contribute indirectly to regulating cell proliferation or tumor formation by interacting with other tumor suppressor proteins such as p53 or SIRT6 [28,29]. Recently, overwhelming evidence suggests that promoter methylation is a key epigenetic mechanism that regulates gene expression, and that tumor-suppressor gene silencing in most cancers is caused by aberrant hypermethylated gene promoter regions.
We had previously studied whether the CpG islands of the USP10 promoter region were hypermethylated in small intestinal adenocarcinoma [18], and our results imply that USP10 methylation occurs in an early stage of colorectal cancer development. Thus, the results of our previous and present studies strongly support that the downregulation of USP10 is epigenetically regulated in CRC.
The alternative reading frame (ARF) protein is frequently mutated in human cancer, and the CDKN2a locus encodes two different proteins (p14ARF and p16INK4a) [30]. ARF is involved in regulating cell cycle arrest and apoptosis through p53-dependent andindependent pathways and is a potent tumor suppressor [31]. The promoter region of p14ARF is known to be hypermethylated in a wide spectrum of human cancers, including colorectal cancer [32]. According to one recent meta-analysis studying the prognostic value of p14ARF, the hypermethylation of p14ARF was more frequently observed in right side colon cancer and microsatellite instability (MSI)-associated cancer than in left side colon cancer and non-MSI associated cancer [33]. In addition, methylation was not associated with tumor differentiation or colorectal cancer stage [33]. Although a CpG island within the promoter region of p14ARF has been widely studied, p14ARF protein expression has not been thoroughly assessed via immunohistochemistry in colorectal cancers. A previous study reported that intact p14ARF expression was found in 36.9% (17/46) of colorectal cancers in Korean patients [34]. On the other hand, we observed that p14ARF expression was not decreased in 71.4% (200/280) of colorectal cancer cases (Table 1). This discrepancy might be explained by the lack of well-defined p14ARF cut-off values for immunohistochemical scoring methods. Recently, Ko et al. demonstrated a positive correlation between USP10 and p14ARF expression in non-small cell lung cancer and found that the loss of both molecules correlated with poor prognosis. The role of USP10 and p14ARF as tumor suppressors has also been shown in small intestinal adenocarcinoma [18] and epithelial ovarian cancer [25]. Corroborating previous studies, we we found that downregulation of both molecules was associated with poor patient prognosis. Interestingly, the prognostic significance of USP10 loss /p14ARF loss was especially prominent in the group of patients that was LN-negative at the time of surgery ( Supplementary Fig. S1). Therefore, to predict patients' prognoses, immunohistochemical staining for both USP10 and p14ARF is recommended if no LN metastasis is identified during surgery.
Our study has a few limitations. First, we used combined conventional immunohistochemistry and digital image analysis to quantitate the USP10 and p14ARF markers. Although the conventional manual scoring of immunohistochemical staining provides reasonable intra-and inter-observer reproducibility, this methodology has issues that make optimal scoring and deciding standard cut-off values for positivity difficult. We previously demonstrated that continuous immunohistochemical scoring via digital image analysis could improve the identification of optimal cut-off points compared with manual visual scoring [35], but that methodology needs to be standardized. For instance, a previously validated, commercially available antibody clone and a well-defined algorithm and cut-off value are essential for clinical utility. Second, although drug resistance in neoadjuvant therapy is a major factor contributing to patients' negative outcomes, this single-institution study considered patients who did not undergo chemotherapy or radiation therapy prior to surgery. Interestingly, a previous study reported that the loss of USP10 correlated significantly with chemoresistance in epithelial ovarian cancer [25].
Therefore, further multiple-institution studies considering a broad spectrum of diseases, including chemosensitivity response and disease recurrence cases, is warranted to validate optimal cut-offs and prognostic values. Third, our study could not include the results of various molecular tests, including MSI. The molecular characterization of patients with colorectal cancer has become a routine examination in clinical practice because genetic information about oncogenes and tumor suppressor genes provides insight into disease progression and response to therapy. However, we could not evaluate the association between molecular variation and USP10 or p14ARF protein expression.
Further investigation with a larger number of cases and more molecular profiling data could be necessary to maximize the clinical value of USP10 or p14ARF information in colorectal cancer.

Conclusions
In conclusion, we found that the USP10 loss correlates with worse patient outcomes. In

Consent for publication
Not applicable

Availability of data and materials
Data is available in the supporting files.

Competing interests
The authors declare that there is no conflict of interest.    Supplementary Files