3.1 Assessment of HMMR expression in different cancer and normal tissues
We first assessed the expression of HMMR in multiple tumour and normal tissue types using the Oncomine database, revealing that expression of this gene was elevated relative to normal tissue controls for bladder, brain, breast, cervical, rectal, colorectal, oesophageal, gastric, head and neck, renal, liver, lung, lymphoma, ovarian, pancreatic, sarcoma and prostate cancers. We also found that relative to normal tissue controls, HMMR expression was lower in leukemia and other cancer tissues (Figure 1A). Detailed findings in particular tumour types are compiled in Table S1. We also used the TCGA and TIMER databases to assess how HMMR expression differs in particular tumour types. We found that the expression of HMMR was significantly elevated relative to normal controls in bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), thyroid carcinoma (THCA) and uterine corpus endometrial carcinoma (UCEC). Differences between the expression of HMMR in tumours and normal adjacent tissue samples in the TCGA data set are shown in Figure 1B.
To further evaluated the expression difference of HMMR between the normal tissues and tumor tissues, we use the GTEx dataset as controls, which shown that high HMMR expression in tumor than the normal tissue in adrenocortical carcinoma (ACC), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD) (Figure 1C, P < 0.01), and sarcoma (SARC), thymoma (THYM), uterine carcinosarcoma (UCS), (Figure S1A-C, P < 0.01). However, we did not obtain a significant difference in kidney chromophobe (KICH), brain lower grade glioma (LGG) and pheochromocytoma and paraganglioma (PCPG), as shown in Figure S1D-F. What more, we found the expression of HMMR was high in normal tissues than tumor tissues in acute myeloid leukemia (LAML) and testicular germ cell tumors (TGCT) are shown in Figure S1G, H.
3.2 The association between HMMR expression and cancer patient prognosis
We next explored the link between the expression of HMMR and cancer patient outcome using the PrognoScan database (Table1 and Tables S2–S4). We found that multiple cancer types exhibited a significant association between patient prognosis and HMMR expression including bladder, brain, blood, breast, colon, ovarian skin and lung cancer (Figure 2A–H and Figure S2). We additionally employed the Kaplan-Meier plotter database in order to assess how HMMR expression relates to prognosis in a range of cancer types, revealing its elevation to be significantly linked with a poorer prognosis in ovarian cancer (OS HR = 1.32, 95% CI = 1.15–1.5, P = 4e-5; PFS HR = 1.3, 95% CI = 1.14–1.48, P = 7e-5), lung cancer (OS HR = 1.7, 95% CI = 1.49–1.93, P = 4e-16; PFS HR = 1.7, 95% CI = 1.67–2.47, P = 6.5e-13) and pancreatic ductal adenocarcinoma (PAAD) (OS HR = 2.31, 95% CI = 1.51–3.55, P = 8e-5; RFS HR = 3.62, 95% CI = 1.4–9.41, P = 0.0048) (Figure 2I–L and Table 2). However, we found reduced HMMR expression to be correlated with poorer patient prognosis in gastric cancer (OS HR = 0.59, 95% CI = 0.48–0.71, P = 4.7e-8; PFS HR = 0.63, 95% CI = 0.5–0.8, P = 8.2e-5) (Figure 2M–N). There was not any significant relationship between the expression of HMMR and the OS prognosis of breast cancer patients (Figure 2O), but the high HMMR expression level have a significant relationship with the RFS prognosis of breast cancer patients (RFS HR = 1.32, 95% CI = 1.19–1.46, P = 8.1e-8) (Figure 2P). In addition, we found HMMR expression to be linked with poor pancreatic ductal adenocarcinoma patients prognosis, by use the Kaplan-Meier plotter database to assess the relationship between HMMR expression and patient clinicopathological findings. We found that HMMR expression correlated significantly with OS, DFS and patient gender, stage, grade of pancreatic ductal adenocarcinoma patients (Table 2). We further used the GEPIA2 database to assess how HMMR expression relates to patient prognosis, analysing 33 TCGA cancer types and revealing that HMMR expression correlated both with OS and DFS in ACC, BLCA, KIRP, LGG, LIHC, LUAD and PAAD (Figure S3 and Figure S4).
We also conducted a analyses by Sangerbox tool to show the correlation between high expression HMMR and poor OS for different tumors, as shown in (Figure S5), high HMMR expression was associated with poor OS for LUAD, UCEC, BLCA, PAAD, KIRP, LIHC, MESO, KIRC, HNSC, LGG, KICH, ACC and UVM (all P < 0.05). The above results clearly demonstrate that HMMR expression significantly correlated with poorer outcome in multiple tumour types.
3.3 Genetic alteration analysis of HMMR
We observed the genetic alteration status of HMMR in different tumor samples of the TCGA cohorts. As shown in (Figure 3A), the highest alteration frequency of HMMR (> 6%) appears for patients with UCEC with “mutation” as the primary type. The “amplification” type of CNA was the primary type in the KIRC and CHOL cases, which show an alteration frequency of > 6% and ~3%, respectively (Figure 3A). The types, sites and case number of the HMMR genetic alteration are further presented in (Figure 3B). We found that missense mutation of HMMR in the Q52L site was detected in 1 cases of PAAD (Figure 3B), is able to induce a frame shift mutation of the HMMR gene. We can observe the Q52L site in the 3D structure of HMMR protein (Figure 3C). In addition, we analyzed the correlation between HMMR expression and TMB (Tumor mutational burden)/MSI (Microsatellite instability) across all tumors of TCGA. As shown in (Figure S6), we observed a positive correlation for GBM (P = 1.7e-06), LUAD (P = 0.00092), PRAD (P = 3.4e-29), UCEC (P = 0.0074), COAD (P = 0.013), STAD (P = 2e-04), SKCM (P = 4.5e-07), KIRC (P = 0.003), HNSC (P = 0.042), LAML (P = 0.029), KICH (P = 4.3e-08), and ACC (P = 0.0039). HMMR expression is negatively correlated with MSI of DLBC (P = 4.4e-05) but is positively correlated with that of GBM (P = 5.9e-05), PRAD (P = 0.032), UCEC (P = 8.8e-05), SARC (Sarcoma) (P = 1e-07), COAD (P = 0.00091), and STAD (P = 2.2e-05) (Figure S7). This result deserves more in-depth research.
3.4 Gene function analysis
We select human pancreatic ductal adenocarcinoma (PDAC) cell lines for functional analysis of HMMR. The cell lines HPDE6-C7, PANC-1, Capan-2, SW1990 and BxPC-3 were cultured in RPMI 1640 medium containing 10% fetal bovine serum, which was placed in incubator at 37 ℃ in a humidified atmosphere with 5% CO2. After 2 to 3 days of cell passage, the cells in the logarithmic growth phase were selected for further experiment.
To study expression of HMMR in PDAC cells, real-time polymerase chain reaction (PCR) and western blotting were performed. Compared with HPDE6-C7 cells, which are immortalized human normal pancreatic ductal epithelium cells, HMMR messenger RNA (mRNA) and protein were highly expressed in PDAC cells (Figure 4A and Figure 4B). Notably, HMMR expression in high-metastasis potential cell lines, such as Capan-2 (Figure 4A and Figure 4B).
To understand the function of HMMR in PDAC cells, we manipulated HMMR expression in Caoan-2 by short hairpin RNA (shRNA) knockdown. Three shRNA (shRNA1, shRNA2, and shRNA3) were designed to silence HMMR expression in Capan-2 cells named as Capan-2-shHMMR subsequently. Expression level of HMMR was identified by real-time PCR and western blotting; shRNA2 was the most effective one and was chosen for further study (Figure S8A, B). Compared to Capan-2, Capan-2-shHMMR cells had a lower absorbance in methyl thiazol tetrazolium assay, which indicated a lower proliferation rate (Figure S8C). Consistently, Capan-2 cells also formed more colonies compared to Capan-2-shHMMR in colony formation assay (Figure 4C). .
The wound-healing and transwell assays were used to investigate migration and invasion capacity. Results showed that Capan-2-control cells had a faster wound closure rate and more invasion cells than Capan-2-shHMMR cells (Figure 4D, 4E). It suggests that HMMR promotes PDAC cells proliferation, migration, and invasion capacity in vitro.
3.5 Enrichment analysis of HMMR-related partners
To further investigate the molecular mechanism of the HMMR gene in tumorigenesis and development, we attempted to screen out the targeting HMMR-binding proteins and the HMMR expression-correlated genes for a series of pathway enrichment analyses. Based on the STRING tool, we obtained a total of 29 HMMR-binding proteins, which were supported by experimental evidence. Figure 5A shows the interaction network of these proteins. We used the GEPIA2 tool to combine all tumor expression data of TCGA and obtained the top 100 genes that correlated with HMMR expression. As shown in Figure 5B, the HMMR expression level was positively correlated with that of KIF11 (Kinesin Family Member 11) (R = 0.77), BUB1 (Budding Uninhibited By Benzimidazoles 1) (R = 0.77), CCNA2 (Cyclin A2) (R = 0.77), AURKA (Aurora Kinase A) (R = 0.70), BRCA1 (Breast Cancer 1) (R = 0.63), and TPX2 (Targeting Protein For Xklp2) (R = 0.73) genes (all P < 0.001). The corresponding heatmap data also showed a positive correlation between HMMR and the above six genes in the majority of detailed cancer types (Figure 5C). An intersection analysis of the above two groups showed three common member, namely, AURKA, BRCA1 and TPX2 (Figure 5D).
We combined the two datasets to perform KEGG and GO enrichment analyses. The KEGG data of Figure 6A suggest that “Oocyte meiosis”, “Cell cycle”, “Progesterone-mediated oocyte maturation”, “FoxO signaling pathway” and “p53 signaling pathway” might be involved in the effect of HMMR on tumor pathogenesis. The GO enrichment analysis data further indicated that most of these genes are linked to the pathways or cellular biology of microtubule motor activity, such as microtubule binging, microtubule cytoskeleton organization, and protein kinase regulator activity, such as protein serine/threinine kinase activator activity, protein serine/threinine kinase activity and cyclin-dependent protein kinase holoenzyme complex, and others. (Figure 6B and Figure S9).