MAPK14 over-expression is a transcriptomic feature of polycythemia vera and correlates with adverse clinical outcomes

Abstract

Background: The transcriptomic signature has not been fully elucidated in PV, as well as mRNA markers for clinical variables (thrombosis, leukemic transformation, survival, etc.). We attempted to reveal and validate crucial co-expression modules and marker mRNAs correlating with polycythemia vera (PV) by weighted gene co-expression network analysis (WGCNA).

Material and methods: The GSE57793/26014/61629 datasets were downloaded from Gene Expression Omnibus (GEO) database and integrated into one fused dataset. By R software and 'WGCNA' package, the PV-specific co-expression module was identified, the pathway enrichment profile of which was obtained by over-representation analysis (ORA). Protein-protein interaction (PPI) network and hub gene analysis identified MAPK14 as our target gene. Then the distribution of MAPK14 expression in different disease/mutation types, were depicted based on external independent datasets. Genome-scale correlation analysis revealed the association of MAPK14 and JAK/STAT family genes. Then gene set enrichment analysis (GSEA) was performed to detect the activated and suppressed pathways associating with MAPK14 expression. Moreover, GSE47018 dataset was utilized to compare clinical variables (thrombosis, leukemic transformation, survival, etc.) between MAPK14-high and MAPK14-low groups.

Results: An integrated dataset including 177 samples (83 PV, 35 ET, 17 PMF and 42 normal donors) were inputted into WGCNA. The 'tan' module was identified as the PV-specific module (R² = 0.56, p = 8e-16), the genes of which were dominantly enriched in pro-inflammatory pathways (Toll-like receptor (TLR)/TNF signaling, etc.). MAPK14 is identified as the top hub gene in PV-related PPI network with the highest betweenness. External datasets validated that the MAPK14 expression was significantly higher in PV than that of essential thrombocytosis (ET)/primary myelofibrosis (PMF) patients and normal donors. JAK2 homozygous mutation carriers have higher level of MAPK14 than that of other mutation types. The expression of JAK/STAT family genes significantly correlated with MAPK14, which also contributed to the activation of oxidated phosphorylation, interferon-alpha (IFNα) response and PI3K-Akt-mTOR signaling, etc. Moreover, MAPK14-high group have more adverse clinical outcomes (splenectomy, thrombosis, disease aggressiveness) and inferior survival than MAPK14-low group.

Conclusion: MAPK14 over-expression was identified as a transcriptomic feature of PV, which was also related to inferior clinical outcomes. The results provided novel insights for biomarkers and therapeutic targets for PV.

Keywords: Expression; MAPK14; Polycythemia vera; WGCNA.

Fig. 1

The relationship of gene co-expression clusters and disease subtypes. All modules (colors) are displayed on the X axis, while all disease subtypes are displayed on the Y axis. Each box contains corresponding Pearson’s coefficients (by gradient of color, red = 1, blue =  − 1) and p value

Fig. 2

The correlation of MM and GS for the ‘tan’ module, in which individual dots stand for one specific gene

Fig. 3

The results of ORA for PV-specific gene module. A The dotplot of enriched pathways. The size of dots represented the count of genes involved in the pathway. While the color of dots correlated with the -log10(q value). B The network of interaction between signaling pathways and hub genes within the PV-specific module, in which the circles represented hub genes and red rhombuses stand for enriched pathways. The color gradient of circles correlated with the connectivity degrees of individual genes (red for high degrees, blue for low degrees)

Fig. 4

The PPI network for the PV-specific module. The color gradient of node indicated the betweenness (red for high betweenness, blue for low betweenness)

Fig. 5

The normalized expression value of MAPK14 in different disease types of MPN and normal donors for the fused dataset (A), GSE54644 (B) and GSE103237 (C). *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001

Fig. 6

The normalized expression value of MAPK14 in different mutation types of MPN and normal donors for GSE103237 (A) and GSE54644 (B). *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001

Fig. 7

The transcriptomic correlation of MAPK14 and JAK/STAT family genes by the fused dataset

Fig. 8

The curves of running enrichment score for MAPK14-related pathways