3.1 Expression and correlation of JMJD5 and PKM2 in the STAD tissues
We first determined the expressions of JMJD5 and PKM2 in the STAD tissues by western blot and RT-qPCR. We identified that the number of JMJD5 was remarkably lower in STAD in comparison with the normal gastric tissue (Fig. 1A, C). However, the number of PKM2 was greatly higher in STAD than in normal gastric tissue (Fig. 1B, D).
We further identified the relationship between the expressions of JMJD5 and PKM2. In the group of low JMJD5 expression (n = 60), 33 patients had a high expression of PKM2 (55%) and 27 had a low expression of PKM2 (45%). Contrarily, in the group of JMJD5 high expression (n = 33), 25 patients had a high expression of PKM2 (75.76%) and 8 had a low expression of PKM2 (24.24%). The expression of JMJD5 in the STAD tissue had a significant positive correlation with PKM2 expression (P = 0.048) (Table 1). Similarly, the result retrieved from GEPIA showed that JMJD5 had a positive correlation with PKM2 expression in STAD (P = 0.014) (Supplementary Fig. 1).
3.2 Correlation between the expressions of JMJD5 and PKM2 and the clinicopathological characteristics of the STAD tissues
The relationship between JMJD5 and the different clinicopathological features is shown in Table 2. Low expression of JMJD5 was related to poor differentiation (P = 0.002) and large tumor size (P = 0.044). There were no statistical associations of JMJD5 with gender, age, pathological tumor node metastasis (pTNM) stage, primary tumor (pT) classification, lymph node metastasis, CEA, and CA199. Furthermore, PKM2 expression had no significant correlations with gender, age, differentiation, pTNM stage, pT classification, tumor size, lymph node metastasis, CEA, and CA199 (Table 2).
3.3 Correlation of JMJD5 and PKM2 expressions with survival in STAD tissues
Kaplan–Meier survival analysis curves are displayed in Fig. 3. Patients with high expression of JMJD5 had remarkably higher OS (43.46 vs. 29.18 months, P = 0.008) and disease-free survival (DFS) (29.07 vs. 19.29 months, P = 0.065) than those with low expression of JMJD5 (Fig.3A, C). On the contrary, the results suggested that the prognosis of STAD patients with high expression of PKM2 was poorer than that of patients with low expression of PKM2 with regard to OS (27.69 vs. 42.22 months, P = 0.018) and DFS (16.87 vs. 29.83 months, P = 0.046) (Fig. 3B, D). Furthermore, STAD patients with both high expression of JMJD5 and low expression of PKM2 had obviously higher OS (P = 0.001) and DFS (P = 0.024) than the others (Fig. 3E, F).
Moreover, we assessed the relationship among differentiation, stage, pT size, lymph node metastasis, OS, and DFS (Fig.4). Our study found that patients with well-differentiated (Fig.4A), stage I (Fig.4B), T1+T2 pT staging (Fig. 4C), smaller tumor (Fig.4D), and no lymph node metastasis (Fig. 4E) had higher OS and DFS than the others.
3.4 Univariate and multivariate analyses of prognostic factors in STAD tissues
Univariate analysis of OS showed that low expression of JMJD5
(P = 0.010), high expression of PKM2 (P = 0.020), poor differentiation (P < 0.001), high pT (P = 0.001), positive lymph node metastasis (P = 0.001), advanced pTNM stage (P < 0.001), and larger tumors (P = 0.001) were unfavorable prognostic predictors. However, age, gender, CEA, and CA199 had no prognostic value. Multivariate analysis indicated that differentiation (HR 2.082, 95% Cl 1.326–3.269, P = 0.001) and pTNM stage (HR 2.811, 95% Cl 1.174–6.730, P = 0.020) were the independent poor prognostic factors (Fig. 5A, B).
Furthermore, the results of univariate analysis of DFS proved that high expression of PKM2 (P = 0.048), poor differentiation (P = 0.004), high pT (P = 0.002), positive lymph node metastasis (P < 0.001), advanced pTNM stage (P < 0.001), larger tumor (P = 0.001), and CA199 (P = 0.009) were the key factors leading to poorer DFS in patients with STAD. Multivariate analysis revealed that differentiation (HR 2.045, 95% Cl 1.308–3.199, P = 0.002) and pTNM stage (HR 2.714, 95% Cl 1.126–6.544, P = 0.026) were the independent prognostic factors of DFS, as shown in Fig. 5C, D.
3.5 Transcriptional levels of members of the JMJD family and PKM2 in STAD tissues
The alternative names, chromosomal locations, and amino acid sequences of members of the JMJD family and PKM2 are depicted in Table 3. We used UALCAN (http://ualcan.path.uab.edu/) to compare the mRNA expressions of JMJD family members and PKM2 between 415 STAD and 34 normal gastric tissues. The results signified that the expressions of JMJD1B, JMJD1C, JMJD2D, JMJD4, JMJD5, JARID2, HSPBAP1, TYW5, and PKM2 were higher in STAD when compared with the normal stomach tissues (Fig. 6). Furthermore, we contrasted the expressions of JMJD family members and PKM2 using GEPIA (http://gepia.cancer-pku.cn/). The expressions of JMJD1B, JMJD1C, JMJD2D, JMJD4, JARID2, HSPBAP1, TYW5, and PKM2 mRNAs were higher in the tumor tissues than in the normal stomach tissues (Supplementary Fig. 2A). In addition, JMJD1B had the highest expression in STAD, followed by JMJD1C and JMJD4. However, JMJD5 had the lowest expression (Supplementary Fig. 2B). Moreover, we obtained immunofluorescence images in The Human Protein Atlas (https://www.Fproteinatlas.org/) to verify the protein localization. We discovered that JMJD1B/JMJD1C/JMJD2/JMJD5 were all located in the nucleoplasm, JMJD4 in the plasma membrane, JARID2 in the nucleoplasm and mitochondria, and HSPBAP1/TYW5/PKM2 in the mitochondria, nuclear bodies, and cytosol, respectively (Supplementary Fig. 3).
3.6 Relationship of the clinicopathological parameters with JMJD family members and PKM2 in STAD tissues
After determining the expressions of members of the JMJD family and PKM2 in STAD, the relationship of these genes with cancer stage and grade was examined in the UALCAN database (Fig. 7,8). We found an obvious correlation between the mRNA expressions of these genes and the clinicopathological parameters. Importantly, the mRNA expressions of JMJD1B, JMJD1C, JMJD2D, JMJD4, JARID2, HSPBAP1, TYW5, and PKM2 were evidently higher in the advanced stage tumors than in the early stages (P < 0.05). The expression level of JMJD5 was also higher in advanced-stage tumors, but there was no significant difference between the normal tissue and advanced-stage tumors (Fig. 7).
In addition, we statistically found that the mRNA expressions of JMJD1B, JMJD1C, JMJD2D, and JMJD5 were the highest in grade III. However, the mRNA expressions of JMJD4, JARID2, TYW5, and PKM2 were the highest in grade II and that of HSPBAP1 was the highest in grade I (Fig. 8).
3.7 Prognostic value of members of the JMJD family and PKM2
We employed the Kaplan–Meier plotter (http://kmplot.com/analysis/) to determine the prognostic value of JMJD family members and PKM2 in STAD patients, including OS, FP, and PPS. The results showed that the members of the JMJD family and PKM2 were remarkably correlated with the prognosis (Fig. 9). In each group, the patients were divided into two sub-groups of low and high expression according to the cut-off value. In the STAD tissues, patients with high expressions of JMJD1C, JMJD2D, JMJD4, JARID2, HSPBAP1, TYW5, and PKM2 had significantly poorer OS/FP/PPS than those with a low expression (P < 0.05) (Fig. 9B, C, D, F, G, H, I). On the contrary, patients with high expressions of JMJD5 and JMJD1B had longer OS/FP/PPS (P < 0.05) (Fig. 9A, E).
3.8 Genetic mutations in members of the JMJD family and PKM2
We used the cBioPortal online tool (www.cbioportal.org) to explore the alterations in the JMJD family and PKM2 in STAD (TCGA, Firehose Legacy). According to the obtained results, these genes varied in 393 samples out of the 478 patients with STAD (82.22%). Among the STAD tissues, JARID2 had the highest mutation rate of 17%, followed by JMJD1C and JMJD4, which were both 13%. HSPBAP1 had the lowest mutation rate of 0.5% (Fig. 10A, B). We then retrieved the 3D structures of the JMJD family and PKM2, and the common mutations in the STAD sites were color-coded in the latter figures (Supplementary Fig. 4). We then constructed the network for the JMJD family, PKM2, and 90 co-expressed genes. The co-expressed genes comprised the glycolysis-related and cellular growth-related genes, including PGK1, PGAM1, and ANXA2 (Fig. 10C).
3.9 Predicting the functions and pathways of members of the JMJD family and PKM2 and similar genes in the STAD patients
Next, we applied Metascape (https://metascape.org) for GO, KEGG, and PPI enrichment analyses. The potential functions and pathways of the members of the JMJD family, PKM2, and similar genes are shown in Fig. 11 A–F. The participation of these genes was in the range of BP (20 terms), CC (20 terms), and MF (20 terms). The genes were enriched in several BP terms: covalent chromatin modification, DNA repair, regulation of DNA metabolic process, mRNA processing, NADH processing, protein acylation, positive regulation of GTPase activity, and DNA replication (Fig. 11A). Moreover, cellular components, including transferase complex, chromosomal region, nuclear periphery, nuclear speck, heterochromatin, and nuclear chromosome were significantly associated with members of the JMJD family, PKM2, and similar genes (Fig. 11B). We discovered that molecular functions such as transcription coregulator activity, chromatin binding, helicase activity, nucleoside-triphosphatase regulator activity, histone binding, transcription factor binding, N-acetyltransferase activity, and histone demethylase activity were remarkably regulated by members of the JMJD family, PKM2, and similar genes (Fig. 11C). The genes were enriched in the 12 KEGG pathways, including biosynthesis of amino acids, hypoxia-inducible factor-1 (HIF-1) signaling pathway, central carbon metabolism in cancer, cell cycle, and cysteine and methionine metabolism in cancer (Fig. 11D). Fig. 11E and F demonstrate that the genes were enriched in the PPI network. The genes were mainly associated with DNA repair, transcription elongation from RNA polymerase II promoter, DNA-templated transcription, elongation, and metabolism.