NLRC and NLRX gene family mRNA expression and prognostic value in hepatocellular carcinoma

Abstract Nucleotide‐binding oligomerization domain (NOD)‐like receptor (NLR)C and NLRX family proteins play a key role in the innate immune response. The relationship between these proteins and hepatocellular carcinoma (HCC) remains unclear. This study investigated the prognostic significance of NLRC and NLRX family protein levels in HCC patients. Data from 360 HCC patients in The Cancer Genome Atlas database and 231 patients in the Gene Expression Omnibus database were analyzed. Kaplan–Meier analysis and a Cox regression model were used to determine median survival time (MST) and overall and recurrence‐free survival by calculating the hazard ratio (HR) and 95% confidence interval (CI). High NOD2 and low NLRX1 expression in tumor tissue was associated with short MST (P = 0.012 and 0.014, respectively). A joint‐effects analysis of NOD2 and NLRX1 combined revealed that groups III and IV had reduced risk of death from HCC as compared to group I (adjusted P = 0.001, adjusted HR = 0.31, 95% CI = 0.16–0.61 and adjusted P = 0.043, adjusted HR = 0.63, 95%CI = 0.41–0.99, respectively). NOD2 and NLRX1 expression levels are potential prognostic markers in HCC following hepatectomy.

Introduction literature search revealed that only few have been identified [11,12]. Nucleotide-binding oligomerization domain (NOD)-like receptors (NLRs) are cystosolic pattern recognition receptors (PRRs) and include five subfamiliesthat is, NLRA, NLRB, NLRC, NLRP, and NLRX. These receptors play an important role in monitoring the intracellular microenvironment and mediating inflammation and pathogen clearance [13]. The NLRC family has five members-that is, NOD1, NOD2, NLRC3, NLRC4, and NLRC5 [13]. NOD1 and NOD2 are important components of the innate immune system that protects organisms from Helicobacter pylori infection [14] and function as pattern-recognition molecules that initiate intracellular signaling pathways in response to pathogen-associated molecular patterns [15]. NLRC3 was identified as a negative regulator of type I interferon and proinflammatory cytokine production [16]. In contrast, the functions of NLRC4 are not well understood [17]. NLRC5 is negative regulator of nuclear factor κB and type I interferon pathways, and is thus important for innate immune system homeostasis [18]. NLRX1, the only NLR localized in mitochondria and the sole member of the NLRX family, was found to stimulate reactive oxygen species production following Shigella flexneri infection [19].
Abnormal inflammation is considered as an indicator of tumorigenesis and malignancy. Four major families of PRR-that is, toll-like receptors (TLRs), C-type lectin receptors, RIG-I-like receptors, and NLRs-have been implicated in cell proliferation, angiogenesis, tissue remodeling and repair, and tumorigenesis [20]. Most studies of PRR signaling in malignancies to date have focused on TLR family members. However, recent studies indicate that NLR family members play a direct or indirect role in cancer cell death, angiogenesis, invasion, and metastasis [21,22]. The present study investigated the prognostic value of NLRC and NLRX family proteins in HCC.

Functional enrichment analysis of NLRC and NLRX families
The Database for Annotation, Visualization, and Integrated Discovery (DAVID) v.6.7 (https://david-d.ncifcrf.gov/, accessed February 25, 2017) [27,28] was used for functional enrichment analyses, including gene ontology (GO) functional analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. The former included biological process (BP) and molecular function (MF) terms; in the latter, no results were returned for NLRC and NLRX families.

Survival analysis
In TCGA database, mRNA expression levels in 360 HCC patients were divided into two groups at a cutoff value of 75%; low and high expression groups comprised 270 and 90 patients, respectively. The same cutoff value was applied to the GEO database in order to ensure a reasonable comparison between the two databases. Median survival time (MST) was used to evaluate the prognosis of HCC patients in TCGA database, whereas overall survival (OS) and recurrence-free survival (RFS) were used to assess that of patients in the GEO database. Sex, age, and TNM stage were adjusted in the Cox proportional hazards regression model in TCGA database, whereas gender, age, HBV infection status, alanine aminotransferase (ALT) status, main tumor size, multinodule status, cirrhosis, alphafetoprotein (AFP) level, and Barcelona Clinic Liver Cancer (BCLC) stage were adjusted in the Cox proportional hazards regression model in the GEO database.
X. Wang et al.

Joint-effects analysis
Only NOD2 and NLRX1 were statistically significant in TCGA database. We carried out a joint-effects analysis of the combination of NOD2 and NLRX1.
The combination of NOD2 and NLRX1 included group I (high NOD2 and low NLRX1 expression), group II (high NOD2 and high NLRX1 expression), group III (low NOD2 and high NLRX1 expression), and group IV (low NOD2 and low NLRX1 expression).
Sex, age, and TNM stage were adjusted in the Cox proportional hazards regression model according to the combination of genes in TCGA database.

Statistical analysis
Pearson correlation coefficients were used to assess correlations among NOD1, NOD2, NLRC3, NLRC4, NLRC5, and NLRX1 genes. Kaplan-Meier survival analysis and the logrank test were used to calculate MSTs and P

Characteristics of patients in TCGA and GEO databases
Detailed characteristics of the 360 patients in TCGA are shown in Table 1. Race, gender, age, BMI, were not associated with MST. On the other hand, TNM stage, NOD2 and NLRX1 levels showed significant associations with MST (P <0.001; adjusted P= 0.014 and 0.011, respectively).

Correlation analysis of NLRC and NLRX family mRNA expression levels in TCGA and GEO databases
We calculated Pearson correlation coefficients between NLRC and NLRX families. In TCGA database, NOD1 was correlated with other NLRC family members (all P < 0.001) but not with the NLRX family member (P = 0.541), except for NRLC4 (P < 0.001, r = −0.09) (Fig. 1A). Only NOD1, NOD2, and NLRX1 expression data were available in the GEO database. NOD1 was correlated with NOD2 (P = 0.001) but not with the NLRX family member (P = 0.164); there was also no correlation between NOD2 and the NLRX family member (P = 0.341) (Fig. 1B).

GO functional annotation analysis of NLRC and NLRX families
To investigate biological functions of the NLRC and NLRX families, BP and MF were evaluated in the GO analysis ( Fig. 1C  and D). In the KEGG pathway analysis, DAVID did not identify any associations between NLRC and NLRX families.

Survival analysis of NLRC and NLRX family mRNA expression levels in TCGA and GEO databases
The characteristics of patients in TCGA database related to prognosis including age, gender, and TNM stage were analyzed with a multivariate Cox proportional hazards regression model. NOD2 and NLRX1 showed significant associations with MST (adjusted P = 0.014, adjusted HR = 1.64, 95% CI = 1.11-2.44; adjusted P = 0.011, adjusted HR = 0.53, 95% CI = 0.33-0.86, respectively) ( Table 3). For patients in the GEO database, characteristics such as gender, age, HBV viral infection status, ALT status, main tumor size,   Table 4).

Analysis of mRNA expression levels in TCGA and GEO databases
Box plots of the expression levels of six genes were downloaded from an online website ( Fig. 2A-F). NLRC3, NLRC5, and NLRX1 were highly expressed in normal liver tissue whereas the expression in primary liver tumors was low. Scatter plots of NOD1, NOD2, and NLRX1 mRNA expression level in the GEO database revealed that only NOD1 expression differed significantly between tumor and nontumor tissue (P = 0.007; Fig. 2G).
Kaplan-Meier curves of mRNA expression levels in TCGA database at a cutoff of 75% are shown in Figure 3. NOD2, NLRC3, and NLRX1 all had significant P values at this cutoff value (P = 0.011, 0.043, and 0.014, respectively).
Kaplan-Meier curves of mRNA expression levels in the GEO database at 75% cutoff are shown in Figure 4. NOD1, NOD2, and NLRX1 did not have significant P values for OS and RFS (all P > 0.05). Scatter plots of the expression levels of six genes in the TCGA and GEO databases at a 75% cutoff are shown in Figure 5A and B.

Joint-effects analysis of NLRC and NLRX family mRNA expression levels in TCGA database
We carried out a joint-effects analysis for the combination of NOD2 and NLRX1. In the joint-effects analysis of the combination of NOD2 and NLRX1, group I had the shortest MST of 38 months (adjusted P = 0.007), whereas group III had the longest MST of 85 months (adjusted P = 0.001, adjusted HR = 0.31, 95% CI = 0.16-0.61) ( Table 5). Interaction networks among NOD1, NOD2, NLRC4, NLRC5, and NLRX1 are shown in Figure 5C. Kaplan-Meier survival curves of the analyses of two genes are shown in Figures 5D.

Discussion
In this study, we investigated the association between NLRC and NLRX family genes and HCC. We determined that the mRNA expression levels of these two NLR families are associated with distinct prognoses. Thus, the mRNA expression levels of NLRC and NLRX family genes alone or in combination-especially NOD2, and NLRX1 combined-can predict HCC prognosis.
NLR family genes are known to regulate the formation of the inflammasome and pro-inflammatory chemokines and cytokines that are involved in the host response to pathogens [29,30]. However, there is little known about the relationship between these gene families and cancer, especially HCC. NOD1 is an important factor in the defense against Pseudomonas aeruginosa [31], Listeria monocytogenes [32], and H. pylori [33] infection and has been linked to Crohn's disease [34,35], inflammatory bowel disease [36], and Behcet's disease [36]. NOD2 was found to be associated with Crohn's disease [37], ischemic cardiovascular disease [38], Blau syndrome [39], allergic rhinitis [40], and artherosclerosis [41]. NLRC3 is a biomarker for colorectal cancer [42]; NLRC4 was related to enterocolitis [43], recurrent macrophage activation syndrome [44], and familiar cold autoinflammatory syndrome [45]; and NLRC5 has been implicated in chronic periodontitis [46]. NLRX1 was found to be associated with risk of gastric cancer in the Chinese population [47]. Interestingly, the other four genes in the NLRC and NLRX gene families did not show any direct or indirect associations with HCC, with the exception of NOD1/NOD2 pathway, which acted synergistically with NLRP3.
In this study, we found that NOD2 was highly expressed in primary liver tumors, which was associated with shorter MST. In contrast, NLRX1 was expressed at low levels in primary liver tumors, which was also linked to short MST. In the joint-effects analyses, groups I, had the shortest MST. In theory, the opposite trend in expression level for each gene should be associated with the best prognosis. Strikingly, this was only observed in group III.
Mitochondria release molecules such as cytochrome c and apoptosis-inducing factor into the cytosol [63] and are associated with autophagy [64]. Exogenous substances applied to HCC cell lines can affect the release of these molecules and thereby alter caspase-independent apoptosis signaling (i.e., the mitochondrial pathway) [65]. Mitochondrial NLRX1 expression is altered in liver tissue in HCC, suggesting that it could affect apoptosis in HCC, although the detailed mechanisms remain to be determined.
There were some limitations to our study that need to be recognized. Firstly, larger sample sizes are needed in order to increase the reliability of the findings. Secondly, more clinical data concerning tumor progression and prognosis such as smoking and drinking status, Child-Pugh scoring, presence of cirrhosis, transarterial chemoembolization, antitherapy status, radical resection status, pathological differentiation diagnosis, main tumor size, numbers of tumors, status of tumor capsules, regional invasion, intrahepatic metastasis, and vascular invasion should be included to better evaluate the relationship between the two NLR gene families and HCC. Thirdly, the more commonly used indices of OS and RFS should be applied to the evaluation of HCC prognosis. Fourth, further investigations focusing on functional part needs to be well explored in multi-center, multi-racial countries. And functional validation in a well-designed clinical trial will be further studied in our future researches.

Conclusion
Our study demonstrates that NOD2, and NLRX1 may be potential prognostic biomarkers of HCC and their combination showed a strong interaction and better predictive value for HCC prognosis. Due to the small sample size and incomplete clinical information in this study, further well-designed and larger sample size studies are necessary to validate our results. Adjusted P*, adjustment for gender, age, TNM stage. Bold value in all the tables were statistically significant (P ≤ 0.05).