The positive correlation of TIPRL with LC3 and CD133 contributes to cancer aggressiveness: potential biomarkers for early liver cancer

Studies have reported dysregulation of TIPRL, LC3 and CD133 in liver cancer tissues. However, their respective relationships to liver cancer and roles as biomarkers for prognosis and diagnosis of liver cancer have never been studied. Here we report that the level of TIPRL is significantly correlated with levels of LC3 (Spearman r = 0.9) and CD133 (r = 0.7) in liver cancer tissues. We observed significant upregulations of TIPRL, LC3 and CD133 in hepatocellular carcinomas (HCCs) compared with adjacent normal tissues. Importantly, TIPRL, tested among additional variables, showed a significant impact on the prognosis of HCC patients. TIPRL knockdown significantly reduced expressions of LC3, CD133, stemness-related genes, as well as viability and stemness of liver cancer cell-lines, which were promoted by ectopic TIPRL expression. Either alone or as a combination, TIPRL, LC3 and CD133 showed significant values of area under the curve (AUC) and sensitivity/specificity in early liver cancer tissues. Furthermore, the statistical association and the diagnostic efficacies of TIPRL, LC3 and CD133 in HCC tissues were confirmed in a different IHC cohort. This data demonstrates that the complex involvement of TIPRL/LC3/CD133 in liver cancer aggressiveness can together or individually serve as potential biomarkers for the early detection of liver cancer.

As the second leading cause of death worldwide, cancer causes about 1 in 6 deaths. There were an estimated 782,000 reports of liver cancer related deaths in 2018, according to the World Health Organization (WHO) 1 . Despite rapid advances in liver cancer diagnosis and therapy in recent decades, liver cancer death rates have increased 3% per year since 2000. Public health agencies, such as WHO, have reported that this increase is mainly caused by late-stage presentation and inaccessible diagnosis and treatment: contrary to 90% of high-income countries, only 26% of low-income countries have general pathology services in the public sector 2 . Thus, there is an urgent demand to develop diagnostic resources for monitoring in high risk populations, which could provide information to clinicians for potential curative intervention, many of which require early tumor identification.
The human TOR signaling regulator (hereafter TIPRL) protein is the mammalian ortholog of the yeast TIP41 protein, which was originally reported as a binding partner for type 2A-associated protein of 42 kilodaltons (Tap42). TIPRL interacts with protein phosphatase type2Ac (PP2Ac), thereby inhibiting its activity to oppose ataxia telangiectasia mutated (ATM) and ATM and Rad3-related (ATR) dependent phosphorylation events; while not interacting with α4, which is the mammalian homolog of Tap42, unlike the yeast TIP41 3 . Previously, our team reported the TIPRL protein contributing to TNF-related apoptosis inducing ligand (TRAIL) resistance of liver cancers 4 . TIPRL is highly upregulated in hepatocellular carcinomas (HCCs) compared with the adjacent liver tissues, and negatively regulates the MKK7/JNK axis, thereby preventing prolonged activation of MKK7 and JNK and, as a consequence, TRAIL-induced apoptosis.

TIPRL is upregulated in HCCs and predicts a poor prognosis of HCC patients. Given our previous
report that TIPRL confers a TRAIL resistant property to hepatocellular carcinomas (HCCs) 4 , we examined levels of TIPRL as well as LC3 11 and CD133 12 , which are markers for autophagy and cancer stem cells (CSCs), respectively: all contributing to chemo-and radio-resistance in liver cancers. For these, we stained human liver tissues, including HCCs (Supplementary Table 1) with the indicated antibodies and the levels of TIPRL, LC3 and CD133 were determined following normalization of raw data obtained using confocal observation and the ZEN program (Methods; Supplementary Table 2). Consistent with our previous report 4 , we observed significant upregulation of TIPRL in HCCs compared with the adjacent liver tissues (Fig. 1a,b and Supplementary Fig. 1a-d). LC3 and CD133 were strikingly upregulated in HCCs in contrast to normal tissues. We confirmed the upregulations of TIPRL and LC3 mRNA in public datasets (www.oncomine.org, Supplementary Figs. 1e,f).
To further support this analysis, we used the public database (www.kmplot.com) 13 and studied the relationships between levels of TIPRL, LC3 and CD133 and the overall survival (OS) of HCC patients. In keeping with the multivariate Cox model, TIPRL significantly influenced the OS of patients in both a whole population of HCC (HR 1.42, logrank p = 0.05) and sorafenib-treated patients (HR 3, logrank p = 0.06), which is currently the only systemic agent approved for use in HCCs; however, MAP1LC3A (HR 0.93, logrank p = 0.7; HR = 1.1, logrank p = 0.9) and PROM1 (CD133; HR 0.8, logrank p = 0.3; HR 1, logrank p = 0.9) did not show significant influence ( Fig. 2b-d). Moreover, TIPRL exhibited a more enhanced HR ratio in the sorafenib-treated group than a whole population group of HCCs suggesting that TIPRL as an independent risk factor has significant prognostic influence on HCCs related to drug resistance.  Tables 1 and 2). p-values ( * P < 0.05; ** P < 0.01) were determined by a paired t-test, and % differences are shown. (n) is the number of samples. (b) The representative images were selected from normal and HCCs groups, respectively. DAPI was used for nucleus staining, and scar bars, 20 (inserted) and 100 μm. TIPRL is required for liver cancer cell survival and stemness. Next, we studied the roles of TIPRL in an HCC incidence. MTT assays show that TIPRL knockdown reduced cell proliferation in an attached condition and, significantly, the viability of Huh7 and SK-Hep-1 cells in an anoikis (Fig. 3a,b). TIPRL was originally determined as a negative regulator of the catalytic subunits of Type 2A phosphatases (PP2Ac) 3 , and the complex relationship between TIPRL, PP2Ac and mTOR has been recently reported 14 . Considering that mTOR is a master regulator of autophagy contributing to cancer cells' survival via promoting stemness 15 , we examined possible associations of TIPRL, LC3 and CD133 levels in HCC tissues. We found statistically significant associations of TIPRL with LC3 and CD133; when the level of TIPRL was increased, both expressions of LC3 and CD133 were correspondingly augmented, as shown by significant values of Spearman r (Fig. 3c-e). In addition, the LC3 level was statistically correlated with the CD133 level.
Given this significant correlation between levels of TIPRL, LC3 and CD133 in HCC tissues, we investigated and observed significant reductions in TIPRL, LC3, ATG7, CD133 and CD46 mRNA levels in HCC/liver cancer cell-lines transfected with two different small interfering RNA against TIPRL (siTIPRL) (Fig. 3f,h and Supplementary Fig. 2a,c). In agreement with their respective relationships to liver cancer, TIPRL knockdown strikingly decreased mRNA expressions of Oct-4, Nanog, SOX-2 and LIN-28, stemness related genes 16 (Fig. 3g,i and Supplementary Fig. 2b,d). This was further supported by the observations that TIPRL knockdown decreased LC3 and CD133 expression while CD133 knockdown failed to reduce TIPRL and LC3. Moreover, ectopic TIPRL promoted expressions of LC3 and CD133 as well as viability of HCC/liver cancer cell-lines, which were reduced in siTIPRL/siCD133-cells ( Supplementary Fig. 2e-j). Furthermore, these reductions were in keeping with significantly reduced activity of aldehyde dehydrogenase (ALDH) used to discriminate the CD133 liver cancer stem cell population, in siTIPRL-Huh7 cells, compared with the activity in siCont-cells (Fig. 3j,k). Overall, this data indicates that, as an upstream modulator of LC3 and CD133, which are involved in tumor aggressiveness, TIPRL is a critical player in HCC/liver cancer cell proliferation, viability and stemness, which are key events for HCC incidence and progression.
Significant association of TIRPL, LC3 and CD133 levels in liver cancer tissues. To extend our observation that the expression of TIPRL is associated with levels of LC3 and CD133 in HCC tissues (Fig. 3c-e), www.nature.com/scientificreports www.nature.com/scientificreports/ we analyzed the statistical relationship between their levels in the following liver cancer tissues according to cancer grade: hepatocellular carcinoma, intrahepatic cholangiocarcinoma, adenosquamous carcinoma, carcinoid and mixed carcinoma, comprised of intrahepatic cholangiocarcinoma and hepatocellular carcinoma (Supplementary Table 2). We determined significant correlations between levels of TIPRL, LC3 and CD133, as calculated by significant values of Spearman r, in each grade of liver cancers. Moreover, we observed that each correlation line moves from left to right in the progression of liver cancers (Fig. 4a-c). Confocal observation (Fig. 4d), as well as statistical analysis (Fig. 4e) consistently confirm gradual increases of TIPRL, LC3 and CD133 levels according to liver cancer progression, suggesting critical roles of TIPRL, LC3 and CD133 association in the progression of liver cancers.
Consistent with the data (Fig. 1a,b), TIPRL, LC3 and CD133 were significantly upregulated in HCCs compared with the adjacent normal tissues, while CD46 was downregulated (Fig. 6a). Next, we confirmed the significant correlation of TIPRL and LC3, of TIPRL and CD133, as well as of LC3 and CD133 levels in HCC tissues, as exemplified by both significant Spearman r and corresponding p values (Fig. 6b-d). On the other hand, the correlations of TIPRL and CD46 as well as of LC3 and CD46 were not significant (Fig. 6e,f). Furthermore, TIPRL, LC3 and CD133 showed consistently higher values of AUC and sensitivity with reasonable ranges of 95% CI compared to CD46's (Fig. 6g). This indicates more significant diagnostic efficacies of TIPRL, LC3 and CD133 than the efficacy of CD46, previously reported in liver carcinogenesis, for an HCC incidence (Fig. 6h).
Next, we performed a multivariate Cox proportional hazard analysis for TIPRL, LC3, CD46, CD133 and sex (male vs female) to determine the independent effects of risk factors in HCC patients (Supplementary Fig. 4). The regression coefficients show that all covariates except LC3 and sex, failed to gain significant p-values in the Cox analysis. However, LC3 did not present reasonable ranges of 95% CI. On the other hand, the sex and the TIPRL/LC3/ CD46/CD133/sex models gained lower p-values than α-0.05 in a proportionality test, suggesting that these factors failed to independently explain survivability of HCC patients. Although it presented 0.058 as a p-value in the Cox regression analysis, TIPRL presents the most significant hazard ratio (HR 14.2). It also has the smallest values of rho and Chi-square (0.0061795 and 0.9961495, respectively), with 0.94 as a p-value in the proportionality test, together (h) Our study shows that the combination of TIPRL contributing to drug resistance with the cancer stem marker CD133 and autophagy marker LC3 produces novel HCC biomarkers. These biomarkers show more significant diagnostic efficacies than the CD46 currently being used as a liver cancer marker.

Discussion
The WHO predicts that there will be 1,341,344 liver cancer-related deaths in 2034 worldwide. Studies have reported that patients who were diagnosed early and received immediate treatment show higher survival than patients who were not 18 . However, liver cancers show no significant symptoms in early stages, and the patients usually visit physicians either at late stages of cancer or when comorbidities, such as liver cirrhosis, occur. Therefore, the discovery of biomarkers 19 that can be used alone or with ultrasound scanning for the determination of early liver cancers are urgently required. Moreover, alpha-fetoprotein (AFP), which is a representable liver cancer biomarker, has reported poor diagnostic performance 20 ; thus, the reliability of current liver cancer biomarkers are being questioned in clinical settings. Discovering novel biomarkers for liver cancer is a necessity; especially ones that are not influenced by other clinical parameters with both through determination of their roles as well as prognostic and diagnostic efficacies.
As an evolutionarily conserved pathway, autophagy is responsible for the recycling of bio-energetic components, such as damaged-or aged-cellular organelles and/or macromolecules. Given controversial reports 21,22 in liver cancers, it is now accepted that autophagy has a dual role in liver cancer progression 23 . As an inhibitory role in the initial stages of liver tumorigenesis, autophagy decreases inflammation and/or preserves genome stability via limiting p62 accumulation 24 . On the other hand, in the later stages of liver tumorigenesis, autophagy exerts a pro-survival role to protect liver cancer cells against cell death caused by hypoxia 25 . This study determines the upregulation of TIPRL in HCCs, consistent with our previous report 4 , as well as an increase of LC3 and CD133 in tissues of HCCs, compared with the adjacent normal tissues. Furthermore, our study demonstrates the prominent prognostic effect of TIPRL on the HCC incidence and the patients' survivability among other variables, including LC3, CD133, even CD46, and sex, which have reported their contributions in the HCCs incidence. This is supported by the observations that TIPRL knockdown inhibited survival and stemness efficacies of HCC/liver cancer cells via the reduction of autophagy, as well as the strong associations between levels of TIPRL, LC3 and CD133 in HCC and liver cancer tissues. Conversely, CD133 knockdown failed to reduce TIPRL and LC3. Furthermore, ectopic TIPRL promoted the LC3 and CD133 expression and viability of HCC/liver cancer cells. Overall, these results indicate TIPRL as a key modulator of LC3 and CD133 expressions in progress of liver cancer aggressiveness, and suggest novel links between autophagy and hepatocarcinogenesis.
Kimhofer et al. 26 , summarized several key selection criteria for an ideal biomarker as the following: targets should be measurable, have excellent prognostic and/or diagnostic abilities for conditions of interest (i.e., high sensitivity and specificity) and be amenable to measurement techniques that are reliable, reproducible and cheap, which provides accessibility to all populations. In addition, the techniques ideally take a simple kit-based format. Furthermore, biomarker candidates should be validated across a broad range of populations. Considering these criteria, we determined the strong correlation of TIPRL, LC3 and CD133 in a broad range of liver cancer tissues including HCCs. In addition, we observed statistically concurrent argumentation of TIPRL, LC3 and CD133 according to the progression of liver cancers, together indicating the complex interrelationships between levels of the biomarker candidates, TIPRL, LC3 and CD133 in liver cancer aggressiveness. Additionally, when their diagnostic potentials were analyzed in different subtypes of liver cancers, these candidates showed excellent diagnostic efficacies in tissues of grade1 liver cancers with significant sensitivity/specificity. Furthermore, we were able to determine the reliability and reproducibility of the early diagnostic efficacies of TIPRL, LC3 and CD133 candidates, compared to CD46 showing low ratios of AUC as well as sensitivity/specificity, in the different IHC cohort. Together, our study provides TIPRL, LC3 and CD133 as biomarkers for early liver cancers through determination of their roles as well as diagnostic efficacies in liver cancers.
The statistically significant association between the quantified results of the immunohistochemistry and the protein level of the targets has been reported in several studies using various methods, such as Western blotting and immune-enzymatic assays 27 . Compared to semi-quantitative methods based on visual scores, the computer-assisted image analyses 28 have been reported to be superior in their quantification accuracy in multiple types of biomarkers. They thus complement the reproducibility and applicability of the semi-quantitative analysis, because they produce themselves to the desired quantitative results. In this study, human tissues were stained with the indicated antibodies followed by confocal observation, and then the levels of TIPRL, LC3, CD133 and CD46 were quantified using the ZEN program, provided by Carl Zeiss. Furthermore, we confirmed the reproducibility of the excellent diagnostic efficacies of TIPRL, LC3 and CD133 levels in HCCs at the different IHC cohort, thus providing reliability of TIPRL, LC3 and CD133 as novel biomarkers for HCCs and early liver cancers.
In conclusion, our study shows the statistical interrelationships between levels of TIPRL, LC3 and CD133 to HCC/liver cancer aggressiveness, and that prominent diagnostic efficacies of these proteins, either alone or a combination of TIPRL, LC3 and CD133, can be used as biomarkers for early diagnosis of liver cancer. Importantly, a Cox regression hazard model determined that TIPRL has a significant prognostic effect on the survivability of HCC patients, tested among additional variables, LC3, CD133 and sex. In keeping with this finding, TIPRL knockdown reduced survival and stemness efficacies of HCC/liver cancer cells via decreases in expressions of autophagy related-and of stemness related-genes. Ectopic TIPRL promoted the LC3 and CD133 expressions and viability of HCC/liver cancer cells. Furthermore, these significant associations and diagnostic efficacies of TIPRL, LC3 and CD133, especially in HCCs, were confirmed in the different IHC cohort. This study was based on the retrospective cohort study and a low number of samples. Thus, to be adapted to routine analysis to be used by clinicians to address prognostic of patients, our findings need to be further validated by prospective studies and clinical trials. (2019) 9:16802 | https://doi.org/10.1038/s41598-019-53191-5 www.nature.com/scientificreports www.nature.com/scientificreports/ determined for all target proteins provided in the Histo tool. After that, each IDVs were divided by the mean IDV value. Each value was given a log 2 for global normalization.

Statistical analysis.
A multivariate Cox proportional hazard model by XLSTAT version 2017.11.16., was used to determine the impact of TIPRL, LC3, CD133 and CD46 with sex as a covariate on HCCs patients' survival time and to evaluate the violation of the proportional risk assumption. The public database (www.kmplot.com) was used to compare patient subgroups stratified by mRNA expression for survival outcomes. GraphPad Prism 7 (GraphPad Software) was used to plot the ROC curves and to calculate the corresponding AUC and cut-off as well as sensitivity/specificity.
For in vitro study, data was analyzed using GraphPad Prism version 7. The parameters indicating statistical significance with the p values are noted within each figure. Error bars in the graphical data represent mean ± standard error of the mean (S.E.M). The data in the figure panels represents independently performed experiments several times on different days. Sample sizes were determined, given previous experience with the assays. Although it was not examined whether the data converges with the assumption of the statistical approach, the assays used in this study were usually mentioned in the literature we used in this study.
Ethics approval and consent. We gained review exemption from Korea Institutional Review Board; that is, no formal ethics approval was required in this particular case. The study was performed in accordance with the Declaration of Helsinki.