Aberrant expression of CXC chemokines in GBM vs normal tissue.
Sixteen CXC chemokines except CXCL15 in GMB were conducted on the UALCAN database. We screened out nine increased mRNA expression levels for CXCL3, CXCL5, CXCL6, CXCL9, CXCL10, CXCL11, CXCL12, CXCL13, CXCL16 in GMB patients compared to normal tissues. The results are presented in Fig. 1. Meanwhile, based on the database of GEPIA, the transcriptional levels of CXCL2, CXCL3, CXCL8, CXCL9, CXCL10, CXCL11, and CXCL16 in GBM tissues were significantly elevated compared to normal tissue (Fig. 2). All the differential expression gene were elevated. Combining the data from two databases of UALCAN and GEPIA, we gained eleven differentially expressed CXC chemokines in GBM totally (CXCL2, CXCL3, CXCL5, CXCL6, CXCL8, CXCL9, CXCL10, CXCL11, CXCL12, CXCL13, and CXCL16). CXC chemokines of CXCL1, CXCL4, CXCL7, and CXCL14 were excluded because of similar levels in GBM compared to normal tissue.
The prognostic value of CXC chemokines in GBM
We conducted Kaplan–Meier analysis of differentially expressed CXC chemokines and clinical outcome using GEPIA to assess the value of CXC chemokines in the prognosis of GMB. GBM patients with higher transcriptional levels of CXCL5 (p = 0.0086) and CXCL6 (p = 0.017) were significantly associated with worse disease-free survival (Fig. 3). Meanwhile, we found that GBM patients with higher transcriptional levels of CXCL3 (p = 0.021), CXCL5 (p = 0.0046), and CXCL8 (p = 0.049) were significantly associated with worse overall survival (Fig. 4).
The molecular characteristics, similar Gene, and PPI network analysis of CXC chemokines in patients with GBM
We conducted an overall analysis of the molecular characteristics of differentially expressed CXC chemokines, including genetic alteration and gene coexpression. We gained the genetic alterations of differentially expressed CXC chemokines using cBioPortal, and found CXCL2, CXCL3, CXCL5, CXCL6, CXCL8, CXCL9, CXCL10, CXCL11, CXCL12, CXCL13, and CXCL16 were altered in 13%, 6%, 4%, 6%, 33%, 7%, 21%, 8%, 6%, 5% and 0% in patients with GBM, respectively(Fig. 5A), which suggested high mRNA expression was the most common alteration. And then we explored the potential coexpression of the differentially expressed CXC chemokines with STRING (Fig. 5B). According to the coexpression scores, we revealed high correlation among CXCL9/10, CXCL9/10, CXCL8/2, and CXCL9/11, followed by CXCL3/8 and CXCL9/13. With the “similar gene detectioin” module of GEPIA, we identified additional genes with expression features similar to the differential expression CXC chemokines. The top five similar genes of each CXC chemokines were selected, which including ZC3H12A,TACSTD2༌IGFL3༌RP11-154H12.2༌LSM1P1༌BIRC3༌BIRC4༌BIRC5༌BIRC6༌BIRC7༌CXCL6༌OR8G3P༌SERPINB2༌RN7SKP65༌RP1-209B9.2༌CXCL5༌OR8G3P༌PF4V1༌CTB-158D10.3༌RPS27P23༌BCL2A1༌RP11-84C10.2༌RNASE2༌CXCL3༌CCL20༌ZBED2༌CCL17༌RP13-461N9.2༌RN7SKP266༌USP9YP8༌CXCL9༌TRAV12-3༌PLA2G2D༌TRBV20-1༌CCL17༌IDO1༌CXCL10༌RP11-44K6.4༌CXCL9༌DUTP8༌TMEM150B༌TIMD4༌TRIM50༌AQP9༌SLAMF8༌MMP13༌CYCSP20༌RP11-723G8.1༌SSXP3༌FAM218BP༌ZMYND15༌CD37༌DOCK2༌SPI1༌and VAV1. Using STRING, we conducted a PPI network analysis of differentially expressed CXC chemokines, which contained 11 nodes and 55 edges (Fig. 5C). The function of these differentially expressed CXC chemokines enriched in chemokine signaling pathway and inflammatory and immune response pathway.
Functional enrichment analysis of the differentially expressed CXC chemokines and their similar genes in patients with GBM
GO and KEGG enrichment analysis were used to explore the potential function of the differentially expressed CXC chemokines and their similar genes with R software (Fig. 6). GO analysis demonstrated that the differently expressed CXC chemokines and their similar genes were mostly enriched in BP (biological processes) category (response to chemokine, cellular response to chemokine, neutrophil migration, and et.al.) and MF (molecular function) category (chemokine activity, chemokine receptor binding, cytokine activity, and et.al.). CC category (cellular components) was absent. In addition, KEGG pathway analyses demonstrate that the top ten most significant pathways contained Chemokine signaling pathway, Viral protein interaction with cytokine and cytokine receptor, Cytokine-cytokine receptor interaction, IL-17 signaling pathway, Rheumatoid arthritis, TNF signaling pathway, NF-kappa B signaling pathway, Apoptosis - multiple species, Toll-like receptor signaling pathway, and Legionellosis.
Transcription factor targets and kinase targets of CXC chemokines in patients with GBM
To understand the context of CXC chemokines, we revealed possible transcription factor targets and kinase targets of the differentially expressed CXC chemokines with the resource of TRRUST and LinkedOmics respectively.
Query genes of CXCL2, CXCL5, CXCL8, CXCL10, CXCL12 included in TRRUST. Our data revealed two transcription factors (RELA and NFKB1) were the key regulators for gene targets of CXCL2, CXCL5, CXCL8, CXCL10, and CXCL12 (Table 1). We selected the top two kinase targets of CXC chemokines except CXC8/16 with LinkedOmics database. Kinase-target network of CXCL2 is related to PRKCD and MAPK14, while network of CXCL3 and CXCL9 is correlated to CDK2 and ATM. CDK2 and CDK1 were suggested as targets for CXCL5, CXCL6, CXCL11, and CXCL13 kinase-target network. CXCL10 kinase-target network were generally associated with LCK and MTOR, and CXCL12 is related to PRKCD and GRK3 (Table 2).
Table 1
Results of candidate key regulators from TRRUST.
Key TF | description | Regulated targets | P value | FDR |
RELA | v-rel reticuloendotheliosis viral oncogene homolog A (avian) | CXCL2/5/8/10/12 | 4.26e-07 | 4.4e-07 |
NFKB1 | nuclear factor of kappa light polypeptide gene enhancer in B-cells 1 | CXCL2/5/8/10/12 | 4.4e-07 | 4.4e-07 |
TF: transcription factor |
Table 2
Enriched kinase targets of CXC chemokines from LinkerOmics.
CXC chemokines | Enriched kinase target | Description | Leading Edge Number | P Value |
CXCL2 | Kinase_PRKCD | protein kinase C delta | 31 | 0 |
| Kinase_MAPK14 | mitogen-activated protein kinase 14 | 27 | 0.004 |
CXCL3 | Kinase_CDK2 | cyclin dependent kinase 2 | 75 | 0 |
| Kinase_ATM | ATM serine/threonine kinase | 60 | 0 |
CXCL5 | Kinase_CDK2 | cyclin dependent kinase 2 | 118 | 0 |
| Kinase_CDK1 | cyclin dependent kinase 1 | 104 | 0 |
CXCL6 | Kinase_CDK1 | cyclin dependent kinase 1 | 123 | 0 |
| Kinase_CDK2 | cyclin dependent kinase 2 | 107 | 0 |
CXCL9 | Kinase_CDK2 | cyclin dependent kinase 2 | 89 | 0 |
| Kinase_ATM | ATM serine/threonine kinase | 48 | 0 |
CXCL10 | Kinase_LCK | LCK proto-oncogene, Src family tyrosine kinase | 19 | 0 |
| Kinase_MTOR | mechanistic target of rapamycin | 19 | 0 |
CXCL11 | Kinase_CDK2 | cyclin dependent kinase 2 | 83 | 0 |
| Kinase_CDK1 | cyclin dependent kinase 1 | 79 | 0 |
CXCL12 | Kinase_PRKCD | protein kinase C delta | 28 | 0 |
| Kinase_GRK3 | G protein-coupled receptor kinase 3 | 24 | 0 |
CXCL13 | Kinase_CDK1 | cyclin dependent kinase 1 | 104 | 0 |
| Kinase_CDK2 | cyclin dependent kinase 2 | 99 | 0 |
Immune cell infiltration of CXC chemokines in patients with GBM
To systematical analysis of immune infiltrates across GBM, we use “Gene module” module of TIMER to explore the correlation between differentially expressed CXC chemokine (not including CXCL8) and abundance of immune infiltrates (B cells, CD4 + T cells, CD8 + T cells, Neutrophils, Macrophages, and Dendritic cells). CXCL2 expression was negatively associated with the infiltration of CD8 + T cells (Cor = -0.148, p = 0.002), and positively associated with the infiltration of neutrophils (Cor = 0.203, p = 2.76E-05) and dendritic cell (Cor = 0.400, p = 1.67E-17) (Fig. 7A). CXCL3 expression was negatively associated with the infiltration of B cells (Cor = -0.102, p = 0.036) and macrophage (Cor = -0.130, p = 0.007), and positively associated with the infiltration of neutrophils (Cor = 0.122, p = 0.012) and dendritic cell (Cor = 0.389, p = 1.31E-16) (Fig. 7B). CXCL5 expression was negatively associated with the infiltration of B cells (Cor = -0.109, p = 0.024) and CD8 + T Cell (Cor = -0.154, p = 0.001), and positively associated with the infiltration of dendritic cell (Cor = 0.320, p = 2.04E-11) (Fig. 7C). CXCL6 expression was negatively associated with the infiltration of B cells (Cor = -0.149, p = 0.002) and CD8 + T Cell (Cor = -0.099, p = 0.042), and positively associated with the infiltration of dendritic cell (Cor = 0.332, p = 2.82E-12) (Fig. 7D). CXCL9 expression was negatively associated with the infiltration of CD8 + T Cell (Cor = -0.220, p = 5.50E-06) and CD4 + T Cell (Cor = -0.151, p = 0.0018), and positively associated with the infiltration of B cells (Cor = 0.302, p = 2.84E-10) dendritic cell (Cor = 0.175, p = 3.15E-4) (Fig. 7E). CXCL10 expression was negatively associated with the infiltration of CD8 + T Cell (Cor = -0.172, p = 4.05E-04) and CD4 + T Cell (Cor = -0.131, p = 0.007), and positively associated with the infiltration of B cells (Cor = 0.309, p = 1.03E-10), neutrophil (Cor = 0.123, p = 0.011), and dendritic cell (Cor = 0.216, p = 7.60E-06) (Fig. 7F). CXCL11 expression was negatively associated with the infiltration of CD4 + T Cell (Cor = -0.120, p = 0.013), and positively associated with the infiltration of B cells (Cor = 0.302, p = 2.86E-10) and dendritic cell (Cor = 0.109, p = 0.025) (Fig. 7G). CXCL12 expression was negatively associated with the infiltration of CD8 + T Cell (Cor = -0.185, p = 1.38E-4), and positively associated with the infiltration of neutrophil (Cor = 0.111, p = 0.023) and dendritic cell (Cor = 0.287, p = 2.18E-09) (Fig. 7H). CXCL13 expression was negatively associated with the infiltration of CD8 + T Cell (Cor = -0.236, p = 1.01E-06) and CD4 + T Cell (Cor = -0.127, p = 0.009), and positively associated with dendritic cell (Cor = 0.101, p = 0.037) (Fig. 7I). CXCL16 expression was negatively associated with the infiltration of CD8 + T Cell (Cor = -0.281, p = 0.001), and positively associated with CD4 + T Cell (Cor = 0.349, p = 3.92E-05), macrophage (Cor = 0.186, p = 0.032), and dendritic cell (Cor = 0.340, p = 6.43E-05) (Fig. 7J). Our data revealed all the differentially expressed CXC chemokines are positively with infiltration of dendritic cell. In addition, we evaluated correlation of prognosis and immune cell infiltration in GBM using the Cox proportional hazard model. We found lower dendritic cell (p = 0.025) were significantly associated with longer cumulative survival of GMB (Fig. 7K). Previous study reported the autologous dendritic cell vaccine may extend survival in GBM patients15, suggesting CXC chemokines might be the potential therapeutic targets with modulating infiltration of dendritic cell.