DHX37 Impacts Prognosis of Hepatocellular Carcinoma and Lung Adenocarcinoma through Immune Infiltration

Background RNA helicases have various essential functions in basically all aspects of RNA metabolism, not only unwinding RNA but also disturbing the interaction of RNA with proteins. Recently, RNA helicases have been considered potential targets in cancers. So far, there has been no detailed investigation of the biological functions of RNA helicase DHX37 in cancers. Objective We aim to identify the prognostic value of DHX37 associated with tumor microenvironments in cancers. Methods DHX37 expression was examined via the Oncomine database and Tumor Immune Estimation Resource (TIMER). We explored the prognostic role of DHX37 in cancers across various databases. Coexpression genes, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG), and fundamental regulators were performed via LinkedOmics. Confirming the prognostic value of DHX37 in liver hepatocellular carcinoma (LIHC) and lung adenocarcinoma (LUAD), we explored the role of DHX37 in infiltrated lymphocytes in cancers using the Gene Expression Profiling Interactive Analysis (GEPIA) and TIMER databases. Results Through GO and KEGG analyses, expression of DHX37 was also correlated with complex function-specific networks involving the ribosome and RNA metabolic signaling pathways. In LIHC and LUAD, DHX37 expression showed significant positive correlations with markers of Tregs, myeloid-derived suppressor cells (MDSCs), and T cell exhaustion, contributing to immune tolerance. Conclusion These results indicate that DHX37 can serve as a prognostic biomarker in LIHC and LUAD while having an important role in immune tolerance by activating the function of Tregs, MDSC, and T cell exhaustion.


Introduction
Cancer is now known to be a disease which involves multiple players, including its relationship with micro-and macroenvironments. Tumor microenvironment (TME) is crucial in cancer progression and therapeutic responses [1,2] and consists of complex components, among which inflammatory cells make up for the majority proportion and are valuable for diagnostic and prognostic assessment of tumors. The widely used therapy of immune checkpoint modulators, including programmed cell death protein 1 (PD-1), cytotoxic T lymphocyte-associated antigen-4 (CTLA4), lymphocyte activation gene-3 (LAG-3), and T cell immunoglobulin mucin 3 (TIM-3) [3,4], has become a promising anticancer therapy by reversing the suppressed immune status in tumors [5,6]. However, immunotherapy failed to have satisfactory effects in that only 20% of cancer patients benefited with significantly increased survival rates [7,8], due to a combination of different clinical and biological behaviors [9].
In recent years, cumulative evidence has revealed that RNA helicases can modulate physiological processes like innate immune reactions, carcinogenic disorders, and inflammatory disorders [10][11][12][13]. For example, RNA helicase A/DHX9, as a potential therapeutic target, is associated with cancer risk and inflammation [14]. The abnormally high expression of RNA helicases has been tested in a variety of cancers [15]. DHX15 is significantly upregulated in HCC, and its high expression was correlated with poor prognosis. DDX3X drives posttranscriptional programming that dictates melanoma phenotype and poor disease prognosis. RNA helicases p68 and p72 show increased expression during colon carcinogenesis.
DHX37 is a highly conserved DEAH box RNA helicase essential for development of the ribosome [16]. It conjugates to the U3 small nucleolar RNA and is significant for remodeling the U3 snoRNA-pre-18S rRNA structure during 40S maturation [17]; this ensures the formation of the central pseudoknot structure. The production of eukaryotic ribosomes is a massively complicated and energy-draining intracellular activity responsible for the translation of mRNAs into proteins [18].
Previous immunological characterization of CD8 T cells indicated that DHX37 suppressed effector functions, cytokine production, and T cell activation by modulating NF-κB [19], which first showed that DHX37 was related to the progression of cancers. This suggests that DHX37 may be a potential biomarker for cancers and is involved in the immune response.
However, the intrinsic mechanisms of effects of DHX37 on malignant tumor development and immune regulation have not been investigated. In this research, we conducted a comprehensive bioinformatic analysis of DHX37 expression profiles in multiple cancers. This is the first finding that reveals the association between DHX37 and Pan-Cancer in multidimensional biological functions. Our study has identified DHX37 as a potential marker of lung and liver cancers, which may guide the development of novel anticancer therapies.

Oncomine Database Analysis.
We examined the gene expression of DHX37 in Pan-Cancer using the Oncomine database [20]. The gene was assessed for differential expression with t-statistics using Total Access Statistics 2002 (FMS Inc., Vienna, VA). t-tests were conducted both as two-sided for differential expression analysis and one-sided for specific overexpression analysis. A P value of 0.001 and a fold change of 1.5 were set as significance thresholds.

PrognoScan Database Analysis.
We utilized PrognoScan database analysis [21] to assess the correlations between DHX37 expression and survival time in multiple cancers. Survival analysis in PrognoScan employs the minimum P value approach to find the cut point in continuous gene expression measurement for grouping patients. An adjusted Cox P value < 0.05 was considered statistically significant.

Kaplan-Meier
Plotter Database Analysis. The correlation between the gene level of DHX37 and overall survival as well as relapse-free survival in 21 types of tumors was performed by Kaplan-Meier Plotter [22]. The Kaplan-Meier Plotter was set up by searching the GEO, EGA, TCGA, and PubMed repositories to identify datasets with published gene expression and clinical data. The hazard ratio (HR) with 95% confidence intervals (CI) was analyzed. The correlation was significant at the 0.05 level.
2.4. UALCAN Analysis. UALCAN [23] was used to obtain the relative expression of DHX37 both in normal and neoplastic tissues and in several clinicopathological subgroups. UAL-CAN analysis use the TCGA-Assembler to download the TCGA level 3 RNA-seq data related to 31 cancer types and used TPM as the measure of expression. The P value cutoff was 0.05.
2.5. GEPIA. Gene Expression Profiling Interactive Analysis (GEPIA) was applied to confirm the correlational analyses from TIMER [24]. GEPIA analyzed the RNA sequencing expression data of both tumor samples and normal samples from TCGA and the Genotype Tissue Expression projects. To solve the imbalance between the tumor and normal data which can cause inefficiency in various differential analyses, GEPIA downloaded the TCGA and GTEx gene expression data that are recomputed from raw RNA-seq data by the UCSC Xena project based on a uniform pipeline. The P value cutoff was 0.05.
2.6. TIMER Database Analysis. The Tumor Immune Estimation Resource aims to evaluate relative proportions of various immune cell subsets based on data from TCGA with a deconvolution approach [25]. We assessed DHX37 expression and the correlation of DHX37 expression with the density of 6 types of infiltrating immune cells in diverse cancers by the TIMER algorithm database. In addition, the "differential expression module" was used to evaluate clinical prognosis affected by overexpressed DHX37.
2.7. TISIDB Database Analysis. Tumor and immune system interactions [26] were performed to explore the abundance of 28 types of tumor-infiltrating lymphocytes (TILs) and MHC markers by precalculating for 30 TCGA cancer types. The relational coefficient between DHX37 and TILs as well as MHC markers was measured by Spearman's test.

LinkedOmics
Analysis. The LinkedOmics database is a publicly available web server for analyzing multidimensional datasets based on TCGA [27]. DHX37 coexpression was presented in heat maps with analysis of Pearson's correlation coefficient. GO (CC (cellular component), BP (biological process), and MF (molecular function)) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways and enrichment of cellular regulators including kinase targets and miRNA targets and transcription factor target were analyzed by gene set enrichment analysis (GSEA). The rank criterion was an FDR ðfalse discovery rateÞ < 0:05, a minimum number of genes of 3, and a simulation of 500, using the LIHC and LUAD compared datasets.
Unlike the findings from PrognoScan, we found a high expression of DHX37 reduced survival in BRCA (breast invasive carcinoma) (Figures 2(k) and 2(l)) by using the Kaplan-Meier Plotter database. Similarly, the poor outcome [28] Figure S1O, P), DHX37 plays a protective role in their OS but not RFS. For esophageal adenocarcinoma, DHX37 was found to have a favorable effect on relapse-free survival while it worsened overall survival ( Figure S1A, B). In addition, DHX37 only had significant correlation with RFS for pancreatic ductal adenocarcinoma and ovarian cancers ( Figure S1G, H, S, T).
To further verify the value of differentially expressed DHX37 in the progression of different cancers, the RNAseq data in TCGA were also exploited to confirm the prognostic implications of DHX37 in each cancer type via GEPIA (Gene Expression Profiling Interactive Analysis). We assessed correlations between DHX37 expression and clinical outcomes in 33 types of cancer ( Figure S2). DHX37 overexpression was related to poor outcomes of OS (overall survival) and DFS (disease-free survival) in ACC (adrenocortical carcinoma), LGG (brain lower grade glioma), and LIHC (liver hepatocellular carcinoma); OS in LUAD (lung adenocarcinoma), MESO (mesothelioma), and THCA (thyroid carcinoma); and DFS (disease-free survival) in SKCM (Skin Cutaneous Melanoma). These results validated the predictive value of DHX37 in particular types of cancer, such as LIHC and LUAD.

DHX37 Expression Is Associated with Advanced
Clinicopathological Characteristics in LIHC and LUAD. To further reveal the potential relevance of DHX37 expression in cancers, we explored the relationship between DHX37 expression and several clinical features of LIHC and LUAD patients in TCGA cohorts by UALCAN. Subgroup analysis of several clinicopathological characteristics of 421 LIHC samples and 574 LUAD samples consistently showed significantly elevated DHX37 mRNA expression. As shown in Figure 3, the transcription level of DHX37 was significantly upregulated in LIHC and LUAD patients compared to the healthy group (with subgroup analysis based on gender, disease stages, pathological grade, cancer status, and TP53 mutation). Upregulated expression of DHX37 correlates with advanced stage and poor differentiation, particularly in older men and smokers.
Overall, these data analyses indicate that patients of LIHC and LUAD with high levels of DHX37 expression tend to have tumors with advanced clinicopathological parameters.

DHX37 Coexpression Networks in Patients with LIHC
and LUAD. Given the above prognostic findings from multiple databases, we chose LIHC and LUAD as representative cancers for further research. To elucidate the biological function of DHX37, LinkedOmics was applied to   Journal of Immunology Research analyze DHX37 coexpression genes by comparing LIHC and LUAD cohorts. We found that 3682 overlap genes were positively correlated with DHX37, whereas 2002 overlap genes were negatively correlated (Table S4). This result suggests an extensive influence of DHX37 on the transcriptome. A heat map (Figures 4(a) and 4(b)) provides details about the top 50 significant (positively and negatively correlated) genes with DHX37. DHX37 expression had a strong positive association with expression of DDX54, GCN1L1, and PUS1, which reflect changes in mRNA modifications, transcriptional regulation, and DNA repair [29][30][31]. In line with the fact that DHX37 suppresses the immune system by modulating NF-κB, DHX37 expression is positively correlated with the expression of PDCD11, which is known as the NF-κB binding protein (Figure 4(a)). Notably, 43/50 and 18/50 genes in LIHC and LUAD, respectively, had high HR (P < 0:05) in the top 50 significantly positive genes. Instead, there were 12/50 and 13/50 genes with low HR (P < 0:05) in negatively significant ones (Figure 4(c)). The functions of DHX37 were predicted by analyzing GO and KEGG by GSEA. The most highly enriched signaling pathway was determined by their normalized enrichment score (NES). As illustrated in Table 1 and Figure S3B, the biological processes, cellular components, and molecular functions strongly associated with DHX37 were cell cycle regulation, DNA replication, mRNA processing, and respiratory activity. Interestingly, GO analysis also uncovered that MHC (major histocompatibility complex), which plays a crucial role in antigen presenting in cancers, was one of the negatively correlated categories. KEGG analysis defined enrichment in DNA replication, cell cycle, ribosome biogenesis in eukaryotes, homologous recombination, Fanconi anemia pathway, notch signaling pathway, and microRNAs in cancers, while the activities like fatty acid degradation, drug metabolism, metabolism of  DHX37  DHX54  GCN1L1  PUS1  RBm19  NFKBIL2  RRP12  LMNB2  PGAM5  RAB35  TOOF1  SETD8  HCFC1  NOC4L  CAD  POLD1  RECOL4  SART3  TRIM28  DDX55  MYBBP1A  PKMYT1  PTBP1  INCENP  EP400  U2AF2  ANAPC7  PLK1  CDT1  DENR  KIF188  SCRIB  SRRT  PDCD11  CCDC137  TMEM120B  C15or142  GOLGA3  DDX51  FOXM1  MYBL2  ANKRD52  DBF4B  CCDC86  IPO4  NCAPD2  UTP20  FBXL19  ANKLE2  NAA25  DHX34  VPS33A  HMGA1  TACC3  EFTUD2  SFRSB  TIMELESS  UBE20  CHFR  RBM28  CDCA5  DAZAP1  SMARCD1  POLE  NUPB2  SMG5  BRDBP  BOP1  RNF34  C10or12  UBE2S  RAN  VARS  WDR4  MCM7  C16or159  EIF2B1  ZNF598  EIF3B  GTSE1  SFRS9  POLQ  NOP2  WDR62  TAF6  ATAD3A  ESPL1  MKI87  KIF2C  CDC25A  PTPN23  TOMM40  AACS  CCNF  RBM14  BUB19  CD3EAP  TROAP   LUAD   -  Journal of Immunology Research  Journal of Immunology Research xenobiotics by cytochrome, chemical carcinogenesis, complement and coagulation cascades, and activity of amino acid including histidine, arginine, and tyrosine were inhibited ( Figure S3A). These results reveal that the functions involving cell circle modulation, amino acid metabolism, and immune activity were highly correlated with DHX37 expression.

Regulators of DHX37 in LIHC and LUAD.
To delve further into the regulators of DHX37 in LIHC and LUAD, we analyzed the kinase, miRNA, and transcription factor derived from positively correlated gene sets. The top 5 most important target networks were the kinase-target ones related mainly to the polo-like kinase 1, checkpoint kinase 2, cyclin dependent kinase 2, ATR serine/threonine kinase, and cyclin dependent kinase 1 ( Table 2). The miRNA-target network was associated with (TCCGTCC) MIR-184, (TGCACGA) MIR-517, (GTGGTGA) MIR-197, (CCAGGGG) MIR-331, and (CAGCAGG) MIR-370. The transcription factor-target network was related primarily to the E2F Transcription Factor family, including E2F1DP2_01, E2F1_Q6, E2F1DP1_01, E2F1DP2_01, and E2F4DP2_01. The gene set enriched for kinase is responsible mainly for regulating stability and integrality of the genome.

DHX37 Expression Impacts Immune Infiltration Level.
DHX37 is expressed in immune cells and TILs (tumor infiltrating lymphocytes). These cells serve as independent prognostic factors of clinicopathological parameters and outcome in cancers [32]. Hence, we analyzed the correlation between DHX37 expression and immunophenotypic characteristics in Pan-Cancer from TIMER and TISIDB (tumor and immune system interaction). The comprehensive analysis indicated that the DHX37 expression had significant  Figure S4). Furthermore, we also found that DHX37 expression weakly to moderately negatively correlated with 28 types of TILs and MHC expression across all human heterogeneous cancers, except in KIRP (kidney renal papillary cell carcinoma), THCA (thyroid carcinoma), and LGG (brain lower grade glioma) (Figures 5(a) and 5(b)).
On the basis of the findings in immune infiltration landscapes, we further found that DHX37 expression level correlates with poorer clinical outcomes and specific immune cell infiltration in LIHC and LUAD (  (Table 3). We also analyzed the different functional T cells, including Th1 cells, Th2 cells, T regs , and exhausted T cells. In TIMER, after adjustments for tumor purity, the DHX37 expression level was significantly correlated with 59 out of 72 immune cell markers in LIHC and 38 out of 72 in LUAD. In LIHC, B cells, macrophages, and various types of T cells were strongly correlated with DHX37 expression (Table 3) and less significant in LUAD. Moreover, we found that expression of DHX37 positively correlates with markers of CAFs including ACTA2 (r = 0:13, P = 0:0098), FAP (r = 0:32, P = 4:5e − 10), PDGFR (r = 0:22, P = 1:5e − 05), and S100A4 (r = 0:21, P = 3:4e − 05) in LIHC.
The expression levels of most marker set of T regs and exhausted T cells, such as forkhead box P3 (FOXP3), C-C chemokine receptor type 8 (CCR8), signal transducer and activator of transcription 5B (STAT5B), PD-1, CTLA4, and LAG3, have strong positive correlations with DHX37 expression in LIHC and LUAD ( Figure 6). FOXP3 regulates the immune suppression and is a strong prognostic factor for distant metastases [33]. PD-1, a widely known marker related to T lymphocyte function, showed strong positive correlation with DHX37 expression, indicating that high DHX37 expression itself may be a novel predictor for immunotherapy response. In addition, DHX37 expression showed strong negative correlations with complement and strong positive correlations with markers of myeloid-derived suppressor cells. We further evaluated the correlation between DHX37 expression and the above strong significant markers in GEPIA (Table 3). Correlation results between DHX37 and the above markers are similar to this in TIMER. These findings  BRCA  CESC  CHOL  COAD  COAD  ESCA  GBM  HNSC  KICH  KIRC  KIRP  LGG  LIHC  LUAD  LUSC  MESO  OV  PAAD  PCPG  PRAD  READ  SARC  SKCM  STAD  TGCT  THCA  UCEC  UCS

Discussion
DHX37, a member of the DEAH box family of RNA helicases [34], plays indispensable roles in many aspects of gene expression [35]. Differential expression and dysfunction of RNA helicases have been reported in various cancers [36][37][38][39]. Although DHX37 functions have not been extensively elucidated, it was once discovered that it suppressed T cell activation in breast cancer [19]. As elucidated in other research, LAYN was identified as a prognostic biomarker and is highly correlated with immune infiltrates in gastric and colon cancers [40]. Therefore, it might reasonably be assumed that DHX37 expression may influence patients' clinical outcomes through immune infiltration. Here, we report that aberrant expression level of DHX37 correlated to prognosis in multiple cancers. Overexpression of DHX37 predicted higher rates of recurrence and shorter survival times in LIHC and LUAD. Interestingly, increased levels of DHX37 expression were related to advanced clinicopathological characteristics, indicating that DHX37 can be used as a prognostic indicator of cancer stages and metastasis. In addition, this study found that in liver hepatocellular carcinoma and lung adenocarcinoma, immune infiltration levels and various immune marker sets were correlated with the level of DHX37 expression. Therefore, this study revealed    the potential usage of DHX37 expression as a novel predictive biomarker.
Notably, DHX37 expression is related to various immune infiltration levels in cancers, especially in liver cancer and lung adenocarcinoma (Tables 3 and 4 and Figures 5 and 6). Surprisingly, the expression level of DHX37 in LIHC and LUAD is not correlated with tumor purity, indicating that the expression of DHX37 in cancer cells and the tumor microenvironment is equally important. Our results suggest that there is weak to moderate positive correlations between DHX37 expression and infiltration levels of CD4+ T cells and neutrophils in LIHC and LUAD (Figures 5 and 6). Moreover, gene markers of T regs , T cell exhaustion, and MDSCs (myeloid-derived suppressor cells) showed significantly strong positive correlation with DHX37 expression, which gives clues about DHX37 function in modulating tumor immunology in LIHC and LUAD. Furthermore, significantly negative correlations can be found between DHX37 expression and several markers of complement. Immunotherapy based on checkpoint inhibitors yields significant clinical benefit for multiple cancers. However, PD-1 inhibition meets resistance partially resulting from the accumulation of MDSCs [41] or T regs [42]. MDSCs, comprising macrophages, granulocytes, dendritic cells, and immature myeloid cells [43], were initially identified leading to tumor persistence and metastasis [44]. A set of microRNAs was associated with MDSCs and resistance to immunotherapy [45], which was consistent with our results that show expression of DHX37 0.21 * * * * 0.35 * -0.15 * * * -0.33 * * P < 0:05, * * P < 0:01, * * * P < 0:001, * * * * P < 0:0001.