Genome-wide Analysis of the Alternative Splicing Profiles Revealed Novel Prognostic Index for Kidney Renal Cell Clear Cell Carcinoma

Abstract

Alternative splicing (AS) is a major mechanism that greatly enhanced the diversity of proteome. Mounting evidence demonstrated that aberration of AS are important steps for the initiation and progression of human cancers. Here, we comprehensively investigated the association between whole landscape of AS profiles and the survival outcome of renal cell carcinoma (RCC) patients using RNA-seq data from TCGA SpliceSeq. Because of the limited number size of deaths in kidney chromophobe renal cell carcinoma (KICH) and papillary renal cell carcinoma (KIRP) TCGA cohorts, we only conducted survival analysis in kidney clear renal cell carcinoma (KIRC). We further constructed prognostic index (PI) based on prognosis-related AS events and built correlation network for splicing factors and prognosis-related AS events. According to the results, a total of 5351 AS events in 3522 genes were significantly correlated with the overall survival (OS) of kidney clear cell renal cell carcinoma (KIRC) patients. Seven of the PI models exhibited preferable prognosis-predicting capacity for KIRC with PI-ALL reaching the highest area under curve value of 0.875. The splicing regulatory network between splicing factors and prognosis-related AS events depicted a tangled web of relationships between them. One of the splicing factors: KHDRBS3 was validated by immunohistochemistry to be down-regulated in KIRC tissues. In conclusion, the powerful efficiency of risk stratification of PI models indicated the potential of AS signature as promising prognostic markers for KIRC and the splicing regulation network provided possible genetic mechanism of KIRC.

Keywords: alternative splicing; kidney clear renal cell carcinoma; prognostic index; splicing factor; the cancer genome atlas..

Figure 1

The distribution of all AS events and prognosis-related AS events in KIRC. A: Blue bars and orange bars indicated the number of AS events and the corresponding genes in each splicing type, respectively. B: The distribution of prognosis-related AS events in KIRC. Blue bars and orange bars indicated the number of prognosis-related AS events and the corresponding genes in each splicing type, respectively. AA: alternate acceptor site; AD: alternate donor site; AP: alternate promoter; AT: alternate terminator; ES: exon skip; ME: mutually exclusive exons; RI: retained intron.

Figure 2

UpSet plot of survival-associated AS events. The UpSet plot illustrates the interactions among seven types of survival-associated alternative splicing (AS) events in KIRC. A single gene could have up to four types of prognosis-related AS events.

Figure 3

PPI network for genes of top 1000 prognosis-related AS events. The PPI network, comprised of 220 nodes and 466 links, described the interactions between genes corresponding to top 1000 prognosis-related AS events. Nodes with different colors represented different proteins and the edges between nodes represented known interactions or predicted interactions between these proteins.

Figure 4

Circle map and dot plot of KEGG enrichment for genes of top 1000 prognosis-related AS events. (A): Circle map. Ribbons with different colors in the right half circle represented top 12 significant KEGG pathways. The 12 top pathways were enriched by genes listed in the left half circle. (B): Dot plot. The color spectrum ranging from green to red indicated an increasing P value and the size of dot represented the number of genes assembled in specific pathway. Each dot in the plot represented specific pathway terms listed in y axis.

Figure 5

Kaplan-Meier survival analyses for eight prognostic indexes (PIs) in KIRC. Patients in different risk groups of KIRC was divided based on the median value of PSI value of AS events in PI-AA (A), PI-AD (B), PI-AP (C), PI-AT (D), PI-ES (E), PI-ME (F), PI-RI (G) and PI-ALL (H). Numbers of patients in low or high risk groups and censoring data at different time were displayed beneath the Kaplan-Meier survival curves.

Figure 6

Time-dependent ROC curves for assessing the predicting efficiency of eight PIs in KIRC. A: PI-AA; B: PI-AD; C: PI-AP; D: PI-AT; E: PI-ES; F: PI-ME; G: PI-RI; H: PI-ALL.

Figure 7

Evaluation of the prognostic value of PI-ALL in early or advanced clinical stage of KIRC patients from Kaplan-Meier survival analysis. A: Differential survival outcome between KIRC patients (stage I-II) in low and high risk groups. B: Differential survival outcome between KIRC patients (stage III-IV) in low and high risk groups

Figure 8

Survival curves of significant prognosis-related splicing factors for KIRC. A: PCBP1; B: hnRNPK; C: KHDRBS1; D: hnRNPA0; E: RBMX; F: KHDRBS3; G: hnRNPU; H: hnRNPM; I: NOVA2; J: HTRA2; K: SF1; L: hnRNPF. Median expression value was set as cutoff value. The survival curves reflected the increase in the percentage of deaths over time.

Figure 9

HR values of the 12 significant prognosis-related splicing factors and splicing factor-based PI. A: HRs and corresponding 95%CI were illustrated as nodes and strings for each of the 12 significant prognosis-related splicing factors. B: Kaplan-Meier survival curves for splicing factor-based PI. C: Time-dependent ROC curves for splicing factor-based PI.

Figure 10

Splicing regulation network in KIRC. Blue and red nots in the network represented prognosis-related splicing factors and AS events, respectively. The gradually changing colors from blue to red indicated transition of correlations between splicing factors and AS events from negative to positive.

Figure 11

Representative images of immunostaining for KHDRBS3 in KIRC and adjacent normal tissues. A: Immunostaining of KIRC tissues (x100); B: Immunostaining of KIRC tissues (x200); C: Immunostaining of KIRC tissues (x400); D: Immunostaining of normal tissues (x100); E: Immunostaining of normal tissues (x200); F: Immunostaining of normal tissues (x400). The brown area in Figure 11D-F marked the positive staining of KHDRBS3 in adjacent normal tissues.