TCGA-LUAD cohort data and the corresponding clinical information of 535 LUAD patients were downloaded from TCGA database (https://portal.gdc.cancer.gov/repository). LUAD patients were classified into low and high HMMR expression groups according to the median HMMR expression value. The gene expression profiles were normalized using the scale method provided in the “limma” R package. Data analysis was performed with R (version 3.6.3) and the ggplot2 (3.3.3) package. The expression data were normalized to transcripts per kilobase million (TPM) values before further analysis. In addition, a receiver operating characteristic (ROC) curve was used to evaluate the diagnostic value of HMMR using the R packages pROC and ggplot2.

GEPIA (http://gepia.cancer-pku.cn/index.html) is a user-friendly web portal for gene expression analysis based on TCGA and GTEx data (16). In the current study, the CCLE (https://sites.broadinstitute.org/ccle) and GEPIA (http://gepia.cancer-pku.cn/) databases were used to analyze the expression and prognostic value of HMMR in pan-cancer. Furthermore, we used GEPIA to study the correlation between HMMR expression and pathological stage.

We used KM plotter (http://kmplot.com), an online database that contains gene expression data and the survival information of 3,452 clinical lung cancer patients, to analyze the prognostic value of HMMR in lung cancer (17). The patient samples were divided into two groups based on the median expression (high expression and low expression) to analyze the OS, with hazard ratios (HRs), 95% confidence intervals (95% CIs), and log-rank p-values.

TIMER (https://cistrome.shinyapps.io/timer/) is a comprehensive resource for systematic analysis of immune infiltrates across diverse cancer types. In this study, TIMER was used to examine the correlation between the somatic copy number alterations of HMMR and the immune cell infiltration levels of B cells, CD4⁺ T cells, CD8⁺ T cells, neutrophils, macrophages, and dendritic cells (DCs). We also used the R package GSVA to quantify the LUAD immune infiltration of 24 tumor-infiltrating immune cells in tumor samples through single-sample gene set enrichment analysis (ssGSEA). According to the 509 gene signatures of 24 tumor-infiltrating lymphocytes (TILs) (18), comprising natural killer (NK) cells, T follicular helper (Tfh) cells, CD56^bright NK cells, CD56^dim NK cells, central memory CD4⁺ T cells, macrophages, cytotoxic cells, DCs, CD8⁺ B cells, effector memory T (Tem) cells, eosinophils, gamma delta T cells, activated DCs (aDCs), immature DCs (iDCs), mast cells, neutrophils, plasmacytoid DCs (pDCs), T helper cells, regulatory T cells (Tregs), type 1 T helper cells (Thp1), Th2, and Th17, the relative enrichment score of every immunocyte was quantified. The correlation between HMMR and the infiltration levels of immune cells was analyzed with Spearman’s correlation, and immune cells with different expression groups of HMMR were analyzed using the Wilcoxon rank-sum test.

Univariate and multivariate Cox regression analyses were performed to examine the prognostic value of HMMR in LUAD.

In the present study, we utilized the LinkedOmics database (http://www.linkedomics.org/login.php) to study the co-expression genes of HMMR in LUAD. The gene set “kegg.v6.2.symbols.gmt,” which served as a reference gene set, was downloaded from the Molecular Signatures Database (MSigDB) (http://software.broadinstitute.org/gsea/msigdb). We utilized the GSEA software and clusterProfiler package to perform the GO and KEGG enrichment analyses of the signaling pathways of HMMR in LUAD (19–21).

The GeneMANIA database (http://www.genemania.org) was utilized to construct the HMMR interaction network (22), and the STRING online database (https://string-db.org/) was employed to construct the protein–protein interaction (PPI) network of HMMR (23).

We used starBase (http://starbase.sysu.edu.cn/) to predict the potential upstream miRNAs of HMMR and to examine the expression, prognosis, and correlation between let-7b-5p and lncRNA. starBase was also used to predict the binding sites among the miRNAs, mRNAs, and lncRNAs (24). The lncLocator (www.csbio.sjtu.edu.cn/bioinf/lncLocator) is a subcellular localization predictor for long non-coding RNAs based on a stacked ensemble classifier, and CPC2 (http://cpc2.cbi.pku.edu.cn) is a fast and accurate coding potential calculator based on sequence intrinsic features (25, 26). In this study, lncLocator and CPC2 were used to explore the subcellular localization and the protein-coding ability of TMPO-AS1, respectively.

The human bronchial epithelial cell line (BEAS2B) and the LUAD cell lines were purchased from the Cell Bank of Kunming Institute of Zoology and cultured in bronchial epithelial cell growth medium (BEGM) (CC-3170; Lonza, Basel, Switzerland). The HEK-293T cell line was obtained from the American Type Culture Collection (ATCC). The lung cancer cell lines A549, H1299, and H1975 were purchased from Cobioer (Nanjing, China) with short tandem repeat (STR) document. A549, H1299, and H1975 cells were all cultured in RPMI 1640 medium (Corning, Corning, NY, USA) supplemented with 10% fetal bovine serum (cat. no. 10099141C; Gibco, Waltham, MA, USA) and 1% penicillin/streptomycin. HEK-293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Corning). The short hairpin RNA (shRNA) for HMMR was constructed using pLKO.1 vector. The shRNA for the HMMR primer sequences are as follows: HMMR shRNA#1: AAACAGCTGGAAGATGAAGAAGGAA; HMMR shRNA#2: CAGCTGGAAGATGAAGAAGGAAGAA.

The qRT-PCR assay was performed as described (27). For real-time reverse transcription PCR (RT-PCR) assay, the indicated cells were lysed with RNAiso Plus (cat. no. 108-95-2; Takara Bio, Beijing, China). Total RNA was extracted according to the manufacturer’s protocol and then reverse transcribed using the RT reagent kit. Real-time PCR was performed with the FastStart Universal SYBR Green Master Mix (cat. no. 04194194001; Roche, Basel, Switzerland; cat. no. FP411-02; TIANGEN Biotech, Beijing, China) using an Applied Biosystems 7500 machine. The primer sequences are as follows: HMMR-F: ATGATGGCTAAGCAAGAAGGC, HMMR-R: TTTTCCCTTGAGACTCTTCGAGA; β-actin-F: CTTCGCGGGCGACGAT, β-actin-R: CCATAGGAATCCTTCTGACC. Expression quantification was calculated with the 2^−ΔΔCt method.

The cancer cell migration and invasion abilities were assessed with the Transwell assay (28). For the cell proliferation assay, the indicated cells were plated into 12-well plates at a density of 2 × 10⁴. The cell numbers were subsequently counted each day using the automatic cell analyzer Countstar (IC1000; Shanghai RuiYu Biotech Co., Shanghai, China). For the colony formation assay, 500 cells were seeded in a six-well plate at 500 cells/well supplemented with 2 ml cell culture medium. The cell culture medium was changed every 3 days for 2–3 weeks. Indicated cells were fixed with 4% paraformaldehyde (PFA) and stained with 0.5% crystal violet.

The migration ability of the indicated cells was evaluated by the wound healing and Transwell assays. For the wound healing assay, cells were plated in six-well plates at 1 × 10⁶ cells/well in 2 ml culture medium. After 24 h, a wound was scratched on the adherent cell monolayer with an Eppendorf tip. Wounds were imaged at 5–10 positions along each well. For the Transwell assay, 2 × 10⁴ cells in serum-free medium were plated on uncoated insets and incubated using 24-well chemotaxis chambers (Corning cell culture inserts, 8-μm pore size). Fetal bovine serum (10%) was added into the bottom wells of the chambers as a chemoattractant. The cells were allowed to migrate through the membrane for an indicated time. Non-migrating cells were removed and cells migrating to the lower face were stained with cresyl violet (Sigma, St. Louis, MO, USA). Stained cells in the entire field were counted under an inverted microscope.

Western blotting and the immunohistochemistry staining assay were performed as described (29). Briefly, the cell lysates were collected, Western blot was performed, and the primary antibody (HMMR) was incubated overnight. The secondary antibody was also incubated. Finally, western blot develop using protein developing instrument conducted. The detail information of antibodies employ in our study are as follows: HMMR antibody (CST Group (HMMR, Rabbit mAb #87129, 1:1,000) and β-actin.

For the datasets from TCGA, statistical analyses were performed using R (v.3.6.3). The Wilcoxon rank-sum test and the chi-square test were used to estimate the association between HMMR and the clinicopathological characteristics. The Kaplan–Meier method was used to calculate the survival rates of LUAD patients. Cox univariate and multivariate analyses were performed to assess the correlation between clinical features and OS, disease-specific survival (DSS), and progression-free survival (PFS). GraphPad Prism 7.0 was used for statistical analysis of data regarding the function of HMMR. Student’s t-test evaluated the statistical significance between experimental groups, and multiple group comparisons were analyzed using one-way ANOVA. Values of *p < 0.05, **p < 0.01, and ***p < 0.001 were considered significant.

TIMER tools were utilized to examine the expression of HMMR in multifarious cancer. The results indicated that HMMR was elevated in pan-cancer, including bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), LUAD, lung squamous cell carcinoma (LUSC), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), thyroid carcinoma (THCA), and uterine corpus endometrial carcinoma (UCEC) ( Figure 1A ). To further verify the results, we used GEPIA database analysis and found that HMMR was significantly overexpressed in adrenocortical carcinoma (ACC), BLCA, BRCA, cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), COAD, lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), ESCA, glioblastoma multiforme (GBM), HNSC, LIHC, LUAD, LUSC, ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), READ, skin cutaneous melanoma (SKCM), STAD, thymoma (THYM), UCEC, and uveal melanoma (UVM) compared to matched normal tissues. HMMR showed low expression in acute myeloid leukemia (LAML) and TGCT ( Figure 1B ). Finally, we found that HMMR was highly expressed in diverse cancer cell lines, including LUAD cell lines ( Figure 1C ). Primarily, these results confirm that HMMR may play a crucial role in human cancer progression.

Given that HMMR was overexpressed in pan-cancer, we further analyzed its prognosis in pan-cancer. The results showed that a high HMMR expression was correlated with poor OS in ACC, COAD, chromophobe renal cell carcinoma (KICH), KIRC, KIRP, LGG, LIHC, LUAD, MESO, PAAD, PCPG, PRAD, THYM, and UVM ( Supplementary Figure S1A ), related to poor DSS in KIRP, LIHC, MESO, SARC, and THCA ( Supplementary Figure S1B ), and was associated with poor disease-free survival (DFS) in ACC, KICH, KIRC, KIRP, LGG, LIHC, LUAD, MESO, PRAD, THCA, and UVM ( Supplementary Figure S1C ). It was also linked to poor PFS in ACC, KICH, KIRC, KIRP, LGG, LIHC, LUAD, MESO, PCPG, and UVM ( Supplementary Figure S1D ). We also utilized GEPIA tools to examine the correlation between HMMR expression and the pathological stage of diverse cancer. Interestingly, we found that the expression of HMMR was markedly positive with the pathological stage of ACC, BRCA, ESCA, NHCS, KICH, KIRC, KIRP, LAML, LAML, LIHC, LUAD, and LUSC ( Figures 2A–L ). Taken together, these results suggest that HMMR may serve as an oncogene in human cancers.

Previous reports have shown that cancer could accord to the molecular characteristics divided into different immune and molecular subtypes. Thus, we utilized the Tumor Immune System Interactions Database (TISIDB) to analyze the expression of HMMR in diverse immune and molecular subtypes of human cancer. Concerning the immune subtypes, the analysis results showed HMMR to have differential expression patterns in cancers ( Supplementary Figure S2A ). A quintessential example is LUAD, where HMMR was elevated in C1 and C2, while its expression was observed to be low in C3. With regard to the molecular subtypes, HMMR also displayed a distinctive expression pattern ( Supplementary Figure S2B ), in which it showed a high expression in C1 of LUAD, but decreased in C3. To summarize, our results indicate that the expression pattern of HMMR has tissue-dependent specificity.

We initially examined the expression of HMMR between tumor and normal tissues in LUAD using TCGA database. In paired samples, we found that HMMR expression was significantly higher in tissues than in adjacent normal groups ( Figures 3A, B ). This result was validated by the Gene Expression Omnibus (GEO) dataset ( Supplementary Figure S3A ). A further study found that a higher HMMR expression was correlated with adverse clinicopathological features in LUAD, including pathological stage (p < 0.001), TNM stage (p < 0.001), residual tumor (p < 0.001), primary therapy outcome (p < 0.001), and smoking status (p < 0.001). Additionally, the ROC analysis showed that the expression levels of HMMR in LUAD, LUSC, and LUAD were 0.970, 0.989, and 0.975, respectively ( Figures 3C–J and Table 1 ). The ROC curve results were validated in the GEO dataset ( Supplementary Figure S3B ). Owing to HMMR overexpression being correlated with poor clinicopathological features, we then explored its prognostic value in LUAD. The results demonstrated that an elevated HMMR expression was associated with poor OS, DSS, and PFS ( Figure 3K ). This result was validated in the GEO dataset ( Supplementary Figure S3C ). The different subgroups, including TNM stage, age, pathological stage, primary therapy outcome, residual tumor, and smoking status, also correlated with poor prognosis ( Figures 4A–H ). Cox univariate and multivariate analyses indicated that TNM stage and HMMR expression were independent risk factors for LUAD patients resulting in adverse outcomes ( Table 2 ).

To explore the gene mutation information of HMMR in LUAD, cBioportal was employed to perform comprehensive analysis of HMMR. The results showed that the mutation rate of HMMR reached 2.9% in LUAD ( Supplementary Figure S4A ). We also examined the mutation type and base mutation in LUAD and found that missense substitution and base G>A reached the highest mutation rate in LUAD ( Supplementary Figures S4B, C ). The results also displayed the mutation of HMMR in different LUAD molecular subtypes ( Supplementary Figure S4D ). At the DNA level, gain and diploid were the main drivers of the high expression of HMMR in LUAD ( Supplementary Figure S4E ). Overall, these results emphasize that the gene mutation of HMME may contribute to HMMR being elevated in LUAD.

To elucidate the biological functions of HMMR in LUAD, the LinkedOmics database was used to examine positive genes with HMMR. As shown in Figures 5A–C , the most positive gene (r > 0.7) is displayed in a heatmap. Subsequently, we performed GO and KEGG analyses using the top 100 co-expression genes. Biological processes mainly involved DNA replication, chromosome segregation, cell division, and protein localization ( Figures 5D ). For the KEGG enrichment results, the pathways mainly included cell cycle, P53 signaling pathway, non-small cell lung cancer, and FOXO signal pathways ( Figures 5E ). In addition, we analyzed the most relevant gene of HMMR using GeneMANIA. The results indicated that 20 genes were most relevant, with gene functions mainly involved in cell cycle ( Figure 5F ). We also used the STRING database to construct a PPI network. The PPI network of HMMR mainly included PLK4, CD44, AURKA, NEK2, CDK1, and FAM83D ( Figure 5G ).

To uncover the signaling pathways of HMMR in the progression of LUAD, we utilized the GSEA software to perform KEGG pathway enrichment. The analysis results showed that upregulation of HMMR expression was mainly involved in the IL2/STAT5 signaling pathway, IL6/JAK/STAT3 signaling pathway, interferon-γ response, and TNF-α signaling pathway ( Figure 5H ). These results suggest that HMMR plays a pivotal role in immune response regulation in LUAD.

The TIMER database was utilized to examine the relationship between HMMR expression and immune infiltration in LUAD. The results showed that the gene copy number change of HMMR significantly affects the immune infiltration levels of B cells, CD4⁺ T cells, CD8⁺ T cells, macrophage cells, neutrophils, and DCs in LUAD ( Figure 6A ). Furthermore, the TIMER database was used to examine the relationship between the HMMR level and immune infiltration levels in LUAD. The results showed that HMMR expression was markedly positively associated with B cells (r = 0.47, p = 7.69e−30), CD4⁺ T cells (r = 0.71, p = 1.71e−78), CD8⁺ T cells (r = 0.51, p = 1.22e−35), neutrophils (r = 0.77, p = 3.54e−103), macrophage cells (r = 0.44, p = 8.88e−26), and DCs (r = 0.84, p = 5.94e−139) in LUAD ( Figure 6B ). We also found that the expression of HMMR was markedly positively correlated with immune modulators, including CD274 (r = 0.34, p = 9.3e−15), CTLA4 (r = 0.556, p = 5), HAVCR2 (r = 0.71, p = 0), LAG3 (r = 0.39, p = 0), PDCD1 (r = 0.56, p = 0), TIGIT (r = 0.71, p = 0), and PDCD1LG2 (r = 0.67, p = 0) in LUAD ( Figure 6C ). These results demonstrate that HMMR plays crucial roles in the regulation of tumor immune infiltration in LUAD.

Moreover, ssGSEA with Spearman’s rank correlation was employed to measure the correlation between the expression of HMMR and the infiltration levels of 24 immune cell types. The results revealed that the expression of HMMR was markedly positively correlated with the infiltrating level of Th2 cells, gamma delta T cells, T helper cells, NK CD56^dim cells, and aDCs and negatively correlated with the infiltrating levels of NK CD56^bright cells, Th17 cells, pDCs, CD8 T cells, NK cells, iDCs, Tfh cells, eosinophils, and mast cells ( Figure 6D ).

Considering the significance of HMMR in immune regulation, we next explored the relationship between HMMR expression and diverse immune modulators, including tumor-infiltrating lymphocytes, immune stimulators, immune inhibitors, chemokines, receptors, and major histocompatibility complexes (MHCs) in LUAD. The analysis revealed that HMMR expression was positively correlated with the 28 tumor-infiltrating lymphocytes ( Supplementary Figure S5A ), 24 immune inhibitors ( Supplementary Figure S5B ), 45 immune stimulators ( Supplementary Figure S5C ), 21 MHCs ( Supplementary Figure S5D ), 41 chemokines ( Supplementary Figure S5E ), and 18 receptors ( Supplementary Figure S5F ) in LUAD. These findings indicate that HMMR plays an indispensable role in the regulation of immune response in LUAD.

The KM plotter database was utilized to explore the prognosis of HMMR based on the different immune cells in LUAD. It was found that the cohort with an elevated HMMR expression and decreased B cells, CD4⁺ T cells, CD8⁺ T cells, macrophages, NK T cells, and Tregs had poor prognosis ( Figures 7A–L ). These data indicate that diverse immune cell infiltration could significantly affect the prognosis of HMMR in LUAD.

MiRNAs play an important role in the modulation of gene expression. We further examined the upstream miRNAs that regulate HMMR expression in the progression of lung cancer. We utilized starBase and TargetScan to predict the potential miRNAs of HMMR. The results showed four miRNAs that bind with the 3′-UTR of HMMR, namely, let-7b-5p, hsa-miR-18a-5p, hsa-miR-33a-5p, and hsa-miR-369-3p. In the analysis of the correlation between these miRNAs and HMMR, only let-7b-5p was markedly negative with the HMMR expression in LUAD ( Figures 8A, C ). Subsequently, starBase was used to examine the target sites between HMMR and let-7b-5p ( Figure 8B ). A further study found that let-7b-5p was downregulated in LUAD ( Figure 8D ), and a low expression of let-7b-5p was related to poor prognosis and clinicopathological features. The ROC analysis showed that let-7b-5p expression in LUAD was 0.754 ( Figures 8E–J ). Additionally, the qRT-PCR assay examined the expression of let-7b-5p in LUAD cell lines, with the data indicating that it was decreased in LUAD cells compared to control cells ( Figure 8K ). To determine whether let-7b-5p affects the expression of HMMR, we overexpressed let-7b-5p in the A549 cell line and found that the mRNA and protein levels of HMMR were significantly reduced after the overexpression of let-7b-5p ( Figures 8L, M ). Collectively, these data imply that let-7b-5p may participate in the regulation of HMMR expression in LUAD.

It has been shown that lncRNAs play crucial roles in the regulation of the expressions of miRNAs and mRNAs. The above findings showed that let-7b-5p may modulate the expression of HMMR via binding with its 3′-UTR. We next explored the upstream lncRNAs of let-7b-5p. starBase and lncBase were utilized to predict the potential lncRNAs that act as a miRNA sponge and control mRNA expression. By performing the related analysis, we obtained three lncRNAs—SNHG12, LINC02242, and TMPO-AS1—and further analyzed their correlation with hsa-let-7b-5p. The data indicated that SNHG12 and LINC02242 were positively correlated with let-7b-5p. The expression of TMPO-AS1 was not only negative with hsa-let-7b-5p but also positive with HMMR as a target gene of hsa-let-7b-5p ( Figures 9A–C ). In addition, we found that TMPO-AS1 was highly expressed in LUAD ( Figures 9D–G ). We performed localization and coding potential analysis using diverse public databases. The subcellular localization of TMPO-AS1 was determined employing lncLocator. The results indicated that TMPO-AS1 was mainly located in the cytoplasm ( Figure 9H ). We also examined the coding potential of TMPO-AS1 using the coding potential calculator and found that TMPO-AS1 does not possess a protein coding ability ( Figure 9I ). Finally, it was shown that depletion of TMPO-AS1 markedly reduced the expression of HMMR and upregulated the expression of hsa-let-7b-5p in A549 cells ( Figure 9J ).

Upon exploration of the potential mechanism of TMPO-AS1 overexpression in LUAD, we further found that the DNA methylation of TMPO-AS1 was decreased and negatively correlated with its expression in LUAD. Treatment with 5-azacytidine, an inhibitor of DNA methyltransferases, resulted in the increased level of TMPO-AS1 in LUAD cells ( Figures 9K–M ). These results suggest that hypomethylation in the TMPO-AS1 promoter DNA results in the increased expression of this gene in LUAD.

To further determine the function of HMMR in LUAD progression, IHC and qRT-PCR assays were performed to examine the expression of HMMR in different LUAD tissues and cell lines. The results showed that HMMR was significantly elevated in lung cancer and in LUAD cells ( Figures 10A, B ), especially in A549 and H1299 cells. Subsequently, we knocked down HMMR in A549 and H1299 cells and used qRT-PCR and Western blot to examine the knockdown efficiency ( Figures 10C, D ). The growth curve and colony formation assay showed that depletion of HMMR inhibited the cell growth of LUAD cells ( Figures 10E, F ). The Transwell and wound healing assays also demonstrated that the knock down of HMMR inhibited the cell growth of LUAD cells ( Figures 10G, H ). These findings suggest that HMMR promotes the cell growth and migration of LUAD cells.