Submitted 22 May 2019Accepted 19 August 2019Published 17 September 2019
Corresponding authorsShuaibin Lian,[email protected],[email protected] Wang, [email protected]
Academic editorGiulia Piaggio
Additional Information andDeclarations can be found onpage 17
DOI 10.7717/peerj.7696
Copyright2019 Lian et al.
Distributed underCreative Commons CC-BY 4.0
OPEN ACCESS
The co-expression networks ofdifferentially expressed RBPs with TFsand LncRNAs related to clinical TNMstages of cancersShuaibin Lian1,*, Liansheng Li2,*, Yongjie Zhou1, Zixiao Liu1 and Lei Wang2
1College of Physics and Electronic Engineering, XinYang Normal University, Xinyang, HeNan, China2College of Life Sciences, XinYang Normal University, Xinyang, HeNan, China*These authors contributed equally to this work.
ABSTRACTBackground. RNA-binding proteins (RBPs) play important roles in cellular homeosta-sis by regulating the expression of thousands of transcripts, which have been reportedto be involved in human tumorigenesis. Despite previous reports of the dysregulationof RBPs in cancers, the degree of dysregulation of RBPs in cancers and the intrinsicrelevance between dysregulatedRBPs and clinical TNM information remains unknown.Furthermore, the co-expressed networks of dysregulated RBPs with transcriptionalfactors and lncRNAs also require further investigation.Results. Here, we firstly analyzed the deviations of expression levels of 1,542 RBPs from20 cancer types and found that (1) RBPs are dysregulated in almost all 20 cancer types,especially in BLCA, COAD, READ, STAD, LUAD, LUSC and GBM with proportionof deviation larger than 300% compared with non-RBPs in normal tissues. (2) Up-and down-regulated RBPs also show opposed patterns of differential expression incancers and normal tissues. In addition, down-regulated RBPs show a greater degree ofdysregulated expression than up-regulated RBPs do. Secondly, we analyzed the intrinsicrelevance between dysregulated RBPs and clinical TNM information and found that(3) Clinical TNM information for two cancer types—CHOL and KICH—is shownto be closely related to patterns of differentially expressed RBPs (DE RBPs) by co-expression cluster analysis. Thirdly, we identified ten key RBPs (seven down-regulatedand three up-regulated) in CHOL and seven key RBPs (five down-regulated and twoup-regulated) in KICH by analyzing co-expression correlation networks. Fourthly, weconstructed the co-expression networks of key RBPs between 1,570 TFs and 4,147lncRNAs for CHOL and KICH, respectively.Conclusions. These results may provide an insight into the understanding of thefunctions of RBPs in human carcinogenesis. Furthermore, key RBPs and the co-expressed networks offer useful information for potential prognostic biomarkers andtherapeutic targets for patients with cancers at the N andM stages in two cancer typesCHOL and KICH.
Subjects Bioinformatics, GenomicsKeywords RNA-binding proteins (RBPs), Co-expression networks, Clinical TNM system,Differential expression, Transcriptional factors, lncRNAs
How to cite this article Lian S, Li L, Zhou Y, Liu Z, Wang L. 2019. The co-expression networks of differentially expressed RBPs with TFsand LncRNAs related to clinical TNM stages of cancers. PeerJ 7:e7696 http://doi.org/10.7717/peerj.7696
INTRODUCTIONRecent research has highlighted the importance of changes in RNA metabolism in themechanisms of carcinogenesis, including long non-coding RNAs (lncRNAs). RNAtranscription, maturation, transportation, stabilization, degradation, and translationare molecular processes that regulate the cell cycle, as well as cell survival. The keysto regulating RNA metabolism are a group of proteins called RNA-binding proteins(RBPs), which participate in many steps at the post-transcriptional regulation level, andthereby determine the fate and function of each transcriptional transcript in the cell (Fu& Ares, 2014; Stefanie, Markus & Thomas, 2014; Moore & Proudfoot, 2009). Furthermore,dysregulated expression of some RBPs can lead to disease, including neurological disordersand cancers (Wang et al., 2018; Kechavarzi & Janga, 2014). For instance, PAIP1 is proposedas a novel prognostic biomarker by affecting breast cancer cell growth (Piao et al., 2018).TRNAU1AP has been confirmed to play an important role in the regulation of cellproliferation and migration via the PI3K/Akt signaling pathway (Hu et al., 2018). SRPRis reported to regulate keratinocyte proliferation by affecting cell cycle progression andtend to show high expression in epidermal keratinocytes (Kim et al., 2016). RBMS3 hasbeen found to inhibit breast cancer cell proliferation and tumorigenesis by inactivating theWnt/β-catenin signaling pathway (Yang, Quan & Ling, 2018). In addition, overexpressionof RPL34 is suggested to promote malignant proliferation of non-small cell lung cancer(NSCLC) (Yang et al., 2016). Silencing RPL34 plays a blocking role in cell proliferationand metastasis, but promoting cell apoptosis of oral squamous cell carcinomas (OSCCs)(Dai & Wei, 2017; Liu et al., 2015). Moreover, the splicing regulator PTBP2 is suggestedto control a network of genes involved in germ cell adhesion, migration, and polarityand is also very essential for neuronal maturation (Qin et al., 2014; Molly, Leah & Donny,2017; Leah, Sarah & Thomas, 2015). However, some RBPs can act as tumor suppressors.For example, ZNFX1-AS1 is reported to suppress HCC progression via regulating themethylation of miR-9 (Wang et al., 2016). In addition, the silencing of SRSF7 affects theexpression of osteopontin splice variants and decreases the proliferation rate of renal cancercells (Boguslawska et al., 2016). Finally, RPL22/eL22, as a cancer-mutated RBP, tend to beanti-cancer via regulation of the MDM2-p53 feedback loop (Cao et al., 2017a; Cao et al.,2017b).
Consequently, improving our understanding of the characteristics of RBPs and non-RNA-binding proteins (non-RBPs) is an essential step for understanding their rolesin tumorigenesis. Even though recent studies have shown that RBPs are predominantlydysregulated in cancers relative to normal tissues (Wang et al., 2018; Bobak & Sarath, 2014).But, the intensity of dysregulation of RBPs in different cancers is still need to be investigated.Furthermore, cancer is a complex genetic disease. The different developmental stages ofcancer indicate different degrees of severity. The early detection of cancer is linked toimproved survivorship. The tumor-node-metastases (TNM) system, formed by the Unionfor International Cancer Control (UICC) and the American Joint Committee on Cancer(AJCC), is the most widely used cancer staging system (Sobin, Gospodarowicz & Wittekind,1992; American Joint Committee on Cancer, 2010). The TNM system provides information
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 2/22
about the prognosis of the disease of patient based on pathologist evaluations of resectedspecimens. This information is also generally used to plan cancer treatment regimens. Dueto the dysregulation of RBPs in many cancers, it is also interesting to determine whetherthere is an intrinsic relevance between RBP dysregulation and the developmental stages ofcancers.
Transcription factors (TFs) perform the first step in interpreting the genome byrecognizing specific DNA sequences to control transcription and gene expression (Lambertet al., 2018). As the unique gene class, TFs represent the proteins whose binding sites areaffected by various regulatory variants in DNA. An accumulating genome-wide associationstudy (GWAS) shows that the mutations of TFs or TF-binding sites are closely related tomany human cancers (reviewed in Deplancke, Alpern & Gardeux, 2016), such as gastriccancer (Yin et al., 2017), liver cancer (Cao et al., 2017a; Cao et al., 2017b), prostate cancer,colorectal cancer (Saijo et al., 2016) and breast (Humphries et al., 2017) cancer. Severalstudies have emerged to identified the regulatory mechanisms and interactions (Drissiet al., 2015; Zhang, Shen & Cui, 2019). Yet, in most cases, we still do not know how tointerpret the regulatory interactions between RBPs and TFs. Furthermore, long non-codingRNAs (lncRNAs)-transcripts of greater than 200 nucleotides are of vital importance intranscriptional and post-transcriptional levels (Ishizuka et al., 2014). Several evidences havedemonstrated that the expression of lncRNAs is closely related to human diseases (Yuanet al., 2017; Vergara et al., 2012), such as viral infections, neurological disease and cancers(Gibb, Brown & Lam, 2011). In particular, the expression levels of lncRNAs in tumortissues are significantly different compared with the normal tissues (Gloss & Dinger, 2015).Moreover, lncRNAs have merged as the novel biomarkers in several diseases diagnosis andtargets for therapeutics (Zhou et al., 2013). It is well established that transcription factorsand long non-coding RNA have played a central role in the genetics of human diseases(St. Laurent, Wahlestedt & Kapranov, 2015). Yet, up to now, we are far from being able toknow the regulatory interactions and mechanisms of RNA-binding proteins (RBPs) withTFs and lncRNAs. For example, whether there is an interaction between RBPs and TFs inthe process of carcinomatosis and what types of TFs affect the regulatory interaction most?In which stage of cancers do RBPs affect the expression of lncRNA the most?
To comprehensively characterize intensity of dysregulated RBPs from many humancancers (Wang et al., 2018) and construct their interactive networks with TFs and lncRNAsrelated to TNM stages, we first assessed the deviations of gene expression levels of RBPs andnon-RBPs in 20 types of cancerous and normal (control) tissues, respectively (Chang etal., 2013) and analyzed the biological and molecular functions of dysregulated RBPs.Second, we analyzed the relationship between RBP dysregulation and TNM systemclinical data of 5,093 patients across 13 types of cancers. We found out two types ofcancers—cholangiocarcinoma (CHOL) and kidney chromophobe (KICH)—show asignificant relationship between RBP dysregulation and TNM stage information. Third,we constructed the interaction networks for dysregulated RBPs related to TNM stageinformation and TFs of CHOL and KICH, respectively. Fourth, we constructed theinteraction networks for key RBPs related to metastases (M) stage and lncRNAs for CHOLand KICH, respectively (see Fig. 1, which outlines the computational workflow, and
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 3/22
TCGA HTSeq-Counts
RBPs
Cancer-related Risk Modules
RBP information
WGCNA
GO Functional Annotation
TCGA HTSeq-Clinical
TCGA HTSeq-FPKM
Network heatmap plot, all module genes
TCGA TNM information
Networks of key RBPs and lncRNAs
Networks of key RBPs and TFs
DESeq2 and edgeR, LogFC and P_value
DEG=DESeq2∩edgeR
RBP expression heatmap
nonRBPs
Networks of RBPs and nonRBPs
Figure 1 Workflow chart showing the different steps presented in this study. The flow chart shows theacquisition and preparation of data (Pink box), differentially expressed genes analysis and module detec-tion (light yellow box), networks construction and function analysis (pale green box). RBP, RNA-bindingprotein; TCGA, The Cancer Genome Atlas; TF, transcription factor; lncRNA, long non-coding RNA.
Full-size DOI: 10.7717/peerj.7696/fig-1
Materials and Methods). This enabled us to identify the key regulatory RBPs for bothCHOL and KICH cancers.
MATERIALS & METHODSDifferential expression analysisWe downloaded the original data including count matrix, expression FPKM values andclinical informationof 20 cancer types andpairednormal tissues fromTCGAdata base usingSangerBox tool (SangerBox, http://sangerbox.com/). Then, we applied edgeR (Robinson,Mccarthy & Smyth, 2010) and DESeq2 (Anders, 2009) to select differentially expressedgenes (DEGs) from count matrix for each cancer type with parameter padj < 0.05, and|log2FC|> 1. Thirdly, according 1,542RBP genes information (Stefanie, Markus & Thomas,2014), we divided the DEGs into two types for each cancer types: RBPs and non-RBPs. Theexpression FPKM values were used to compute the degree of dysregulation of DEGs andconstruct the co-expression networks. Clinical TNM information for 5,093 patients wereused to co-expressed gene module detection. Our datasets contained 1,542 RBPs, 1,570 TFsand 4,147 lncRNAs; data acquisition information was presented in the Data Availabilitysection. The ‘pheatmap’ package in R was used to generate a heatmap of differentiallyexpressed RBPs shared by 16 cancer types in six systems based on log2-fold change values.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 4/22
The Pearson coefficient (R) was used to judge the similarity of RBPs genes in differentcancers occurring at related tissues have similar expression patterns. The gene cluster withR > 0.5 was considered as having the similar expression pattern. The gene ontologycategorization analysis tool DAVID (Huang, Sherman & Lempicki, 2009a; Huang, Sherman& Lempicki, 2009b) was used to determine biological processes and molecular functions ofDE RBPs.
Standard deviation and proportion of dysregulationIn order to investigate the degree of dysregulation of RBPs, we computed the standarddeviation of differentially expressed genes listed in Table 1 using the ‘std’ function in‘MATLAB’ version R2015b, respectively. In detail, we firstly computed the standarddeviation of differentially expressed RBPs in 20 cancers and paired normal tissues (Fig. 2A).Similarly, we next computed the standard deviation of differentially expressed non-RBPsin 20 cancers and paired normal tissues (Fig. 2B).
The definition of the standard deviation is as follows:
σ =1N
√√√√ N∑i=1
(xi−µ)2.
Here, N is the number of genes, xi isthe expression FPKM value of the ith gene, µ isthe mean FPKM value of all N genes. Larger σ values represent gene expression valuesthat deviate from the mean value to a greater degree, which is indicative of greater genedysregulation. The standard deviation values for RBP expression data is presented inTable S1.
In order to more clearly see the expression differences of RBPs and non-RBPs in cancersand normal tissues, we further computed their relative deviations. In detail, we firstcomputed the relative deviation of RBPs in cancers and paired normal tissues, and then wecomputed the relative deviation of non-RBPs in cancers and normal tissues (Fig. 2C). Therelative proportion of deviation is computed using the following function : p= σc
σn. Here,
σc and σn are the standard deviations of the gene expressions in cancers and normal tissuesrespectively.
Similarly, in order to investigate the expression differences of up- and down-regulatedRBPs, we also computed the standard deviation of them in cancers and normal tissues,respectively (Figs. 2D, 2E). Finally, the mean expression values of up- and down-regulatedRBPs in 20 cancers and paired normal tissues were computed and presented in Figs. 2F &2G.
Weighted co-expression network analysisWeighted gene co-expression network analysis (WGCNA) (Peter & Steve, 2008) is acomprehensive R package that summarizes and standardizes methods and functionsfor co-expression network analysis. Module detection function of WGCNA was usedto detect the correlations between co-expression gene modules and the clinical TNMinformation for 13 types of cancers with default settings one by one. Then, the thresholdof correlation coefficient R> 0.5 and statistical significance P < 0.05 was used to select
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 5/22
Table 1 Differentially expressed RBPs and non-RBPs across 20 types of cancers in seven systems.
System Cancer Cancer type TotalDEG genes
Totalnon-RBPs
TotalRBPs
Up-regulatedRBPs
Down-regulatedRBPs
Landular system PAAD Pancreatic adenocarcinoma 260 259 1 1 0THCA Thyroid carcinoma 4,804 4,765 39 14 25PRAD Prostate adenocarcinoma 4,777 4,722 55 34 21HNSC Head and neck squamous cell carcinoma 7,421 7,335 86 32 54
Respiratory system LUSC Lung squamous cell carcinoma 12,235 11,972 263 54 209LUAD Lung adenocarcinoma 9,004 8,859 145 35 110
Alimentary system READ Rectum adenocarcinoma 8,620 8,430 190 60 130COAD Colon adenocarcinoma 8,954 8,773 181 44 137STAD Stomach adenocarcinoma 9,037 8,932 105 39 66
Urinary system KICH Kidney chromophobe 9,926 9,758 168 76 92KIRC Kidney renal clear cell carcinoma 11,478 11,370 108 37 71KIRP Kidney renal papillary cell carcinoma 8,293 8,189 104 33 71BLCA Bladder urothelial carcinoma 7,146 7,025 121 53 68
Reproductive system CESC Cervical squamous cell carcinoma and endocervicaladenocarcinoma
4,302 4,154 148 57 91
UCEC Uterine corpus endometrial carcinoma 9,046 8,869 177 64 113BRCA Breast invasive carcinoma 7,426 7,298 128 36 92
Nervous system GBM Glioblastoma multiforme 11,946 11,628 318 118 200PCPG Pheochromocytoma and paraganglioma 4,717 4,560 157 101 56
Liver and gall system LIHC Liver hepatocellular carcinoma 6,677 6,576 101 25 76CHOL Cholangiocarcinoma 10,219 10,030 189 80 109
Lianetal.(2019),PeerJ,D
OI10.7717/peerj.7696
6/22
(A).
CancerNormal
Valu
e
RBP
non-RBP
0
5
10
0
10
20
30
0
10
20
30
down_RBPup_RBP
BLC
A
CE
SC
GB
M
HN
SC
KIR
CK
IRP
LUA
D
LUS
C
PR
AD
STA
D
THC
A
CH
OL
LIH
C
BR
CA
CO
AD
KIC
H
RE
AD
UC
EC
PC
PG
PAA
D
Log2
(FP
KM
+1)
Log2
(FP
KM
+1)
0
1
2
3
4
50
1
2
3
4
5
5 10 15 20 25 30 350
BLCA
CESC
GBM
HNSC
KIRC
KIRP
LUAD
LUSC
PRAD
STAD
THCA
CHOLLIHC
BRCA
COAD
KICH
READUCEC
PCPGPAAD
RBPnon-RBP
(D). Cancer
Normal
Valu
eVa
lue
(F).
0
5
10
Valu
e
BLC
A
CE
SC
GB
MH
NS
C
KIR
CK
IRP
LUA
DLU
SC
PR
AD
STA
D
THC
A
CH
OL
LIH
C
BR
CA
CO
AD
KIC
H
RE
AD
UC
EC
PC
PG
PAA
D up_RBP
down_RBP
BLC
A
CE
SC
GB
MH
NS
C
KIR
CK
IRP
LUA
DLU
SC
PR
AD
STA
D
THC
A
CH
OL
LIH
C
BR
CA
CO
AD
KIC
H
RE
AD
UC
EC
PC
PG
PAA
D
CancerNormal
p=1.41E-05
p=0.0011
p=0.0031
p=0.0962
p=0.0179
p=0.0041
(B).
(C).
(E). (G).
Figure 2 The different expression patterns of RBPs and non-RBPs in cancers and normal tissues. (A)The standard expression deviation of RBPs in 20 types of cancers and normal tissues. (B) The standard ex-pression deviation of non-RBPs in 20 types of cancers and normal tissues. (C) The relative expression de-viation of RBPs and non-RBPs. (D) The standard expression deviation of up- regulated RBPs in cancersand normal tissues. (E) The standard expression deviation of down- regulated RBPs in cancers and normaltissues. (F) The expression FPKM values of up- regulated RBPs in cancers and normal tissues. (G) The ex-pression FPKM values of down- regulated RBPs in cancers and normal tissues. Significance values calcu-lated from the Mann–Whitney U test are shown.
Full-size DOI: 10.7717/peerj.7696/fig-2
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 7/22
the cancer type whose TNM information related to co-expressed gene module. Genes withsame expression pattern were clustered into one module and marked with same color. Andthen, the network construction module was used to construct co-expression networks forRBPs with TFs and lncRNAs in the CHOL andKICH cancer types, respectively. The detailedsteps of constructing networks are as follows. Firstly, WGCAN network construction toolwas used to generate the nodes and edges of genes by computing correlations of expressionvalues. The nodes corresponded to genes, and the edges were determined by the pairwisecorrelations between the expression levels of genes. The corresponding called function inthe R package was ‘blockwiseModules’ and the parameters were set as follows: ‘powers =10, minModuleSize = 30, mergeCutHeight = 0.25’, other parameters we set to the defaultsetting. Secondly, nodes with correlation r < 0.5 and edges with weighted threshold <0.3were removed. Finally, the Cytoscape (https://cytoscape.org/) tool was used to plot theinteractions using the nodes and edges of conserved genes.
Clinical TNM information processingClinical TNM information for 5,093 patients exhibiting 13 types of cancers downloadedfrom the TCGA database presented in Table S4. Generally, in the TNM system, ‘T ’ refersto a primary tumor, ‘T1∼T4’ represents the severity of primary cancer according tothe increase in tumor volume and the extent of involvement of adjacent tissues, and‘T0’ indicates no primary tumor. ‘N’ represents the tumor spreading to regional lymphnodes. ‘N1∼N3’ represents the degree of spreading according to the extent of lymphnode involvement. ‘M ’ refers to tumor metastasis. No distant metastasis is expressed by‘M0’, and distant metastasis is expressed by ‘M1’. To investigate the correlation betweenTNM information and gene expression values using WGCNA, we converted the TNMinformation into a weighted matrix. For example, if TNM information for a patient was‘T1-N3-M1’, the corresponding weighted array is [1 3 1].
Statistical methodsWe used the Mann–Whitney U -test (function ‘ranksum’ in software‘MATLAB’ versionR2015b) to examine whether there is statistical significance between given two samples,the default significance level is 0.05 (Lian et al., 2018).
RESULTSThe deviations of expression levels of RBPs between cancerous andnormal tissuesTo investigate the degrees of dysregulation of RBPs in many human cancers, we computedthe standard deviations of gene expression levels of RBPs and non-RBPs in 20 types ofcancerous and normal (control) tissues, respectively. These results indicate that, relativeto normal tissue, RBPs show greater variation in expression in almost all types of cancer(P < 0.05, Mann–Whitney U test, Fig. 2A). In contrast, non-RBPs show considerably lessvariation in expression in cancers relative to normal tissues (P < 0.05, Mann–WhitneyU test, Fig. 2B). These results indicate that RBPs show a greater degree of dysregulatedexpression than non-RBPs in almost 20 types of cancers. Furthermore, the degrees of
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 8/22
dysregulation of RBPs in different cancer types are significantly different. In particular,RBPs show severely dysregulated expression in eight types of cancers, including BLCA,LUAD, STAD, READ, LUSC, GBM and COAD. For these cancers, the correspondingrelative proportions of dysregulation relative to normal tissues are 365.5%, 415.8%, 446.4%,456.3%, 494.5%, 540.9% and 759.1%, respectively (Fig. 2C, Table S1). Interestingly, up-regulated RBPs show a smaller standard deviation in expression in cancer tissue thanin normal tissue (P < 0.05, Mann–Whitney U test, Fig. 2D); However, down-regulatedRBPs show considerable differences in the standard deviation of their expression in cancertissue relative to normal tissues, especially in KICH, KIRC, KIRP, COAD, LUSC, CESC,UCEC, and GBM (P < 0.05, Mann–Whitney U test, Fig. 2E). This may indicate thatdown-regulated RBPs are more dysregulated in cancers than are up-regulated RPBs.Finally, biological process enrichment analyzing indicates both up- and down-regulatedRBPs were highly enriched in the processes, such as rRNAmetabolism, nuclear-transcribedmRNA catabolism, and ncRNA processing (Fig. S1). In addition, up-regulated RBPs werealso enriched in mitochondrial gene expression and in the regulation of mRNA metabolicprocesses, while down-regulated RBPs were enriched in ribosome biogenesis and rRNAprocessing.
Up- and down-regulated RBPs show opposite expression patterns incancer and normal tissueTo investigate cancer-specific differences in RBP expression, we analyzed the standarddeviations and mean expression values of up-regulated and down-regulated RBPs in 20types of cancers and in normal tissues (see methods and materials). Our results for all20 cancer types suggests that, compared to normal tissues, up- and down-regulated RBPsshow opposite patterns of expression in almost all cancers; what’s more, down-regulatedRBPs tend to show the larger expression deviations in cancers than up-regulated RBPs(P < 0.05,Mann–WhitneyU test, Figs. 2D, 2D). In particular, in almost all types of cancers,down-regulated RBPs show larger expression values than up-regulated RBPs. Furthermore,the expression deviations of up-regulated RBPs in cancers are lower than in normal tissues.However, the expression deviations of down-regulated RBPs are considerably greaterin cancer tissue than in normal tissue; this is especially true for BLCA, GBM, HNSC,LUAD, LUSC, STAD, BRCA, KICH, READ, and UCSC (P < 0.05, Mann–Whitney U test,Figs. 2E, 2G). These results suggest that down-regulated RBPs show a severer dysregulationin cancers than up-regulated RBPs. Furthermore, the expression pattern of up- anddown-regulated RBPs in 20 types of cancers are the opposite of those found in normaltissues; what’s more, molecular functions enrichment analyzing indicates that both up-and down-regulated RBPs showed enrichment in: catalytic activity acting on RNA, mRNAand mRNA 3′-UTR binding, nuclease and ribonuclease activity, and single-stranded RNAbinding. In addition, we found enrichment in translation factor activity and RNA bindingfor up-regulated RBPs, and in catalytic activity acting on tRNAs for down-regulated RBPs(Fig. S1). In addition, we also were able to identify which specific biological processes andfunctions were regulated by up- and down-regulated RBPs, which may reveal how thenormal processes of cells can be altered in a way that leads to cell carcinomatosis. These
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 9/22
PCPG GBM KICH KIRC KIRP BLCACHOL LIHC STAD READ COAD BRCA UCEC CESCLUAD LUSC
Nervous system Urinary systemLiver and gall Alimentary system Reproductive systemRespiratory system
C1
C2
C3
C4
C1 C1C1C1
C1
C2C2
C2C2
C2
C3
C3
C3
C3
C3
C4
C4C4
C4
C4
C5
C6
Log (Fold Change)2
4
20
-2
-4
-1
-0.5
0
0.5
1Pearson
(A) (B) (C) (D) (E) (F)
Figure 3 An expression heat map of 801 differentially expressed RBPs shared by 16 cancer types in sixsystems. (A) Respiratory system. (B) Liver and gall. (C)Nervous system. (D) Alimentary system. (E) Re-productive system. (F) Urinary system. Red and blue represent high and low expression. The right columnof each system is the Pearson coefficient R of the corresponding cluster. Gene cluster with R > 0.5 wasmarked with dark yellow, which represents the similar expression pattern.
Full-size DOI: 10.7717/peerj.7696/fig-3
results provide a new insight into understanding the roles of up- and down-regulated RBPsin the process of cell carcinogenesis.
RBPs of different cancers in same system have a similar expressionprofileTo gain clearer insight into RBP expression in cancerous tissues, we analyzed expressionheatmaps of 16 types of cancers, except four types of glandular cancer (PAAD, THCA,PRAD, and HNSC). Because the number of DE RBPs in these four types of cancers aretoo small (Table 1). We divided DE RBPs into six scale systems according to type (i.e.,organization of canceration) and analyzed the expression profiles of 801 DE RPBs sharedby all 16 cancers by system. In each system, we divided RBPs into different types accordingto their expression values and Pearson coefficient R. Gene clusters with R> 0.5 wereconsidered as having the similar expression pattern, which was shown in heatmaps. Thiswas true for all cancers except two cancers of the nervous system. The corresponding genelists of RBPs relevant for each system are presented in Table S2.
In the respiratory and liver and gall systems (Figs. 3A, 3B), two pairs of cancers in samesystems, LUAD and LUSC, CHOL and LIHC, show very similar RBP expression profiles.Among these two cancer systems, C1 and C4 genes showed the co-expression patterns ofRBP for high expression and low expression respectively (R > 0.5). The proportions ofco-expressed RBPs in the respiratory and liver and gall systems were 88.8% and 76.8%,
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 10/22
respectively. However, RBP expression in two cancers of the nervous system (PCPGand GBM) showed opposite expression patterns (Fig. 3C). The proportions of highlyexpressed RBPs in PCPG and GBM cancer tissue were 60.5% and 36.1%, respectively.Highly expressed RBPs in PCPG showed a lower degree of expression in GBM, whilehighly expressed RBPs in GBM showed lower expression in PCPG. The proportion ofRBPs showing opposing patterns of expression in PCPG and GBM was 68.2% (R < 0.5). Insystems containing three types of cancers (i.e., the alimentary and reproductive systems)(Figs. 3D, 3E), the proportion of co-expressed RBPs decreased slightly, reaching 70.4%and 71.5% (R > 0.5), respectively. In the urinary system, which was affected by four typesof cancers, the proportion of co-expressed RBPs reached its minimum value of 35.3%(Fig. 3F). Of the cancers of the urinary system, two (KIRC and KIRP) showed the strongestdegree of RBP co-expression reaching 85.6% (R >0.5). Taken together, our results suggestthat RBP expression in different cancers in similar tissues have similar expression profiles.
Co-expressed gene regulatory networks correlate with clinical TNMstageNext, we investigated whether expression patterns are closely related to developmentalstages of different cancers and constructed the co-expression networks for DE RBPs andnon-RBPs using module detection and network construction tools of WGCNA (Peter &Steve, 2008). Results showed that for two types of cancers—CHOL and KICH—clinicalTNM stage information was closely related to patterns of gene expression (Fig. 4). Theresults for the other 11 cancer types did not satisfy the threshold (Methods and Material,Figs. S2–S7).
In terms of module detection, we identified three modules (orange, green and red)related to the cancer metastasis stage (M -stage), as well as two modules (royal blueand red modules) closely related to the regional lymph node stage (N -stage) for CHOL(Fig. 4A). The redmodule was consistently in both theM andN stages. The gene list for eachcorresponding module is presented in Table S3. Furthermore, we found 10 differentiallyexpressed RBPs in these three modules. These included six RBPs (ACO1, PPARGC1A,PUS7, KHDC1, ELAVL3, and BICC1) in the green module, two RBPs (PANBP17 andHNRNPA1) in the royal blue module, and two RBPs (DCPS and C2orf15) in the redmodule. For KICH, we found that the royal blue module was most closely related tothe M -stage, with a correlation coefficient of 0.96 (P < 4e−15), while the red and greenmodules were closely related to the N -stage (Fig. 4B). The gene list for each correspondingmodule is presented in Table S3. Furthermore, we found seven DE RBPs related to differentdevelopmental stages of cancer. These included four RBPs (TNRC6A,MECP2, ZCCHC14,and POLR2F) in the green module and three RBPs (TDRD1, TDRD9, and CELF4) inthe red module. The regulatory networks of CHOL (Fig. 4C) revealed that (1) in eachsub-network, one RBP interacts with almost all non-RBPs, suggesting that these RBPs arekey regulators for each module; (2) the RBPs in different sub-networks interact with eachother, indicating that they work together to regulate the corresponding developmental stageof the cancer. (3) we found that most key RBPs are down-regulated, the proportion is 80%.In addition, in the red module, two RBPs—including one down-regulated RBP (C2orf15)
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 11/22
0.96 (4e−15)
0.0055 (1)
0.027 (0.9)
−0.087 (0.7)
−0.025 (0.9)
−0.041 (0.8)
−0.028 (0.9)
−0.038 (0.8)
−0.0098 (1)
−0.056 (0.8)
−0.077 (0.7)
0.26 (0.2)
−0.23 (0.2)
0.24 (0.2)
0.22 (0.3)
−0.22 (0.3)
0.27 (0.2)
0.23 (0.2)
0.12 (0.5)
0.27 (0.2)
0.27 (0.2)
0.3 (0.1)
0.22 (0.3)
−0.12 (0.6)
0.39 (0.05)
0.04 (0.8)
0.45 (0.02)
0.23 (0.2)
0.2 (0.3)
0.49 (0.01)
0.22 (0.3)
0.33 (0.1)
0.18 (0.4)
Module−trait relationships
greenoyellowmidnightblue
grey60
purple
redroyalblue
darkredlightgreen
cyan
green
grey
0.61 (0.001)
0.058 (0.8)
0.63 (5e−04)
0.15 (0.5)
0.1 (0.6)
−0.059 (0.8)
0.11 (0.6)
0.014 (0.9)
−0.014 (0.9)
0.32 (0.1)
MNT
0.53 (0.005)
−0.0068 (1)
0.47 (0.01)
0.18 (0.4)
0.52 (0.007)
0.42 (0.03)
0.007 (1)
0.058 (0.8)
−0.0056 (1)
0.45 (0.02)
Module−trait relationships
−1
−0.5
0
0.5
1orangebrown
blackgreen
cyan
redsaddlebrown
royalbluepaleturquoise
salmonblue
0.33 (0.09)
−0.017 (0.9)
−0.022 (0.9)
−0.091 (0.7)
0.31 (0.1)
0.025 (0.9)
0.7 (4e−05)
0.69 (6e−05)
−0.056 (0.8)
−0.05 (0.8)
−0.23 (0.2)
CHOL KICH
−0.045 (0.8) −0.13 (0.5)
MNT
ENSG00000280195 ENSG00000132681
ENSG00000119547
ZCCHC14
MECP2
TDRD9TNRC6A
POLR2FTDRD1
CELF4
ENSG00000238178
ENSG00000109424
ENSG00000180730
ENSG00000182798
ENSG00000231621
ENSG00000164082
ENSG00000070601ENSG00000260519
ENSG00000162891
ENSG00000183396 ENSG00000054938
ENSG00000237973
ENSG00000258551ENSG00000005249
ENSG00000235023
ENSG00000133636
ENSG00000237989
ENSG00000259763
ENSG00000228549
ENSG00000150471
ENSG00000233421
ENSG00000227218
ENSG00000228016
ENSG00000245651
ENSG00000224614
ENSG00000207138
ENSG00000258425
ENSG00000198223
ENSG00000259158
ENSG00000275772
ENSG00000274735
ENSG00000224161
ENSG00000270697
ENSG00000258896
ENSG00000178233
ENSG00000223715
ENSG00000166793
ENSG00000271547
ENSG00000232360
ENSG00000213171
ENSG00000230316
ENSG00000263862ENSG00000119715
ENSG00000164520ENSG00000114854
ENSG00000154252ENSG00000266968
ENSG00000251187ENSG00000253524
ENSG00000247624ENSG00000265944
ENSG00000122691
ENSG00000126010
ENSG00000163485
ENSG00000262884
ENSG00000254042
ENSG00000259518
ENSG00000198846
ENSG00000231794
ENSG00000152104
ENSG00000165309
ENSG00000114349
ENSG00000198626
ENSG00000187720
ENSG00000259417
ENSG00000261273
ENSG00000265046
ENSG00000258603
ENSG00000198756
ENSG00000205312
ENSG00000249274
ENSG00000125430
ENSG00000255750
ENSG00000177994
ENSG00000133640
ACO1
ENSG00000230121
ENSG00000253476
ENSG00000279353
ENSG00000125931
ENSG00000274150
DCPS
ENSG00000176907
ENSG00000253525
ENSG00000172817
ENSG00000273259
ENSG00000094755ENSG00000080709
ENSG00000142609
ENSG00000161249
ENSG00000234215
ENSG00000147509
ENSG00000226926
ENSG00000253671
ENSG00000152592
ENSG00000152527
ENSG00000107518
ENSG00000177138
ENSG00000176595
ENSG00000147614
ENSG00000146166
ENSG00000233532
ENSG00000248757
ENSG00000238283
ENSG00000243137
ENSG00000235897
ENSG00000178568
ENSG00000086570
ENSG00000182389
ENSG00000223774
ENSG00000224739
ENSG00000225362ENSG00000254656 ENSG00000186073
ENSG00000259732
ENSG00000214510ENSG00000137090
ENSG00000223949
ENSG00000253250
ENSG00000253213
ENSG00000246130ENSG00000232480
ENSG00000118997ENSG00000228940
ENSG00000214914ENSG00000214650
PUS7
PPARGC1A
ELAVL3
BICC1 KHDC1
C2orf15
HNRNPA1RANBP17
(A) (B)
(C) (D)
Pearson
Figure 4 The co-expressionmodules detection of CHOL and KICH, respectively. (A) and (B) Co-expression modules of CHOL and KICH correlated with their clinical TNM stage information, respec-tively. Different color represents the gene modules with different expression pattern. The first, middle,and last column are detected gene modules related toM, N, and T stages, respectively. The two num-bers in each module ‘‘a(b)’’ represent coefficients of co-expression and statistical significance, a is the co-expression coefficient and b is the corresponding P-value. The modules with coefficient larger than 0.5and P-value smaller than 0.05 were considered as the related modules. (C) and (D) Co-expression net-works of RBPs and non-RBPs in gene modules (green, red and royal blue) related withM - and N -stagesfor CHOL and KICH, respectively. Inner circle are key RBPs, outer circle are the non-RBPs. Triangularrepresents the up-regulated genes, block represents the down-regulated genes.
Full-size DOI: 10.7717/peerj.7696/fig-4
and one up-regulated RBP (CDPS) also play such a regulatory role. For the regulatorynetworks of KICH, we also identified three sub-networks that corresponded to the M -and N -stages (Fig. 4D). First, we found that RBPs interacted with almost all non-RBPsin each module, indicating that RBPs are key regulatory factors of the genes in thesemodules. Second, we identified seven key RBPs in the three modules, of which five weredown-regulated and two were up-regulated. In the green module, we identified four keyRBPs—TNRC6A, MECP2, ZCCHC14, and POLR2F—all of which were down-regulated.In the red module, we identified two key up-regulated RBPs (TDRD1 and TDRD9) andone key down-regulated RBP, CELF4. These results suggest that dysregulated RBPs play akey role in the regulation of the development of the CHOL and KICHM -stage, which mayprovide a new perspective for potential prognostic biomarkers and therapeutic targets forpatients with cancers atM stages in two cancer types CHOL and KICH.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 12/22
TMEM72-AS1
RP11-227H15.5
TTC39A-AS1
CTC-490G23.4
RP11-323F24.4
RP11-64C12.8 CLDN10-AS1
AC004540.5
LINC00323RP11-728F11.4SBF2-AS1
AC007365.3
RP11-540A21.2
ROR1-AS1
RP11-881M11.4
HIF1A-AS1
RP11-315D16.4
RP11-59H7.3
RP11-49I11.4LINC02115
LINC00665NEXN-AS1
RP11-96H17.1
ALDH1L1-AS2
RP11-143N13.2
RP11-247L20.3 RP11-548L20.1
LINC00460
RP11-368L12.1RP11-701H24.3
LINC00113RP11-268J15.5
CTB-178M22.2
RBPMS-AS1
C12orf80RP4-564F22.6
LINC01594
AC006547.13RP11-128A17.2
LINC01973
SERPINB9P1
RP11-92G12.3
GACAT2
RP11-359E10.1
TM4SF19-AS1
RP11-540O11.1
NDUFB2-AS1 RP11-20J15.2
SAPCD1-AS1LINC01730
AC144833.1
RP13-49I15.6
LINC00261
RP11-701H24.4
RASAL2-AS1
RP11-210M15.2
AC092667.2
RP11-148B18.4
LINC01117
LINC00322
LINC00957
LINC01426
SH3RF3-AS1
RP11-521O16.1
AFAP1-AS1
FAM212B-AS1LINC01956
RP5-942I16.1RP11-1081L13.4
U47924.32RP11-59H7.4
LINC02198
RP5-999L4.2
LINC01852
CHRM3-AS2AC002480.3
AC013275.2LINC01320
LINC01127
LINC00346
IFNG-AS1
LINC01697
TTC21B-AS1
RP11-713N11.6
RP11-272L14.2
BX470102.3LINC01444
CADM3-AS1PAX8-AS1
AP002954.3
RP5-965F6.2
DNM3OS
WT1-AS
LINC01353
RP11-66B24.4
CMAHP
LINC02182
RP11-527L4.6
RP11-613D13.8
CTD-3247F14.2
RP11-222K16.1RP11-276H7.2
CTC-498J12.1
SPATA13
AC006129.1
LINC01268 RP11-379B8.1
RP11-326C3.13
RP13-580F15.2
RP1-41C23.4
LINC00511
AP000355.2
CTD-2033D15.2
AC008753.6
RP11-443B7.3
AC007365.3
CTC-455F18.1
RP11-61G19.1
RP5-1050E16.2U62631.5
AC007950.1
LINC00589
AC013472.3
CH17-437K3.1
CTD-2566J3.1RP11-597D13.9
RAPGEF4-AS1
RP11-431J24.2
RP11-289F5.1
LINC00524
CTA-796E4.4
RP11-663P9.1
LINC00445
RP11-65M17.1
CTA-363E6.6
RP11-85G21.2
RP11-520D19.2
LINC01531
PKIA-AS1
AC011298.2
LINC01213
RP11-17M24.3
RP5-1050E16.1
RP11-793A3.1
RP11-737O24.1RP11-17E2.2
RP11-356J5.12
LINC01055
RP11-569G13.2
RP11-348F1.2
USP12-AS2
AC005392.13
RP11-107M16.2
AC016907.3
LINC00652
FLJ27354
RP11-413E1.4LINC00211
AC002401.1RP11-13N12.2
XXbac-BPG308K3.5
CTD-2147F2.1
RP11-575F12.2RP11-48G14.1
LINC01983
XXYLT1-AS2
RP11-284F21.10
RP11-347E10.1
RP1-122P22.4
LINC00443
AC129492.6
RP11-881M11.2
RP11-284F21.9
RP11-317M11.1
RP11-90P13.1
LINC01426
RP11-348F1.3
RP1-10C16.1
TNKS2-AS1
AF064858.10
RP1-78O14.1LINC00857
LINC01451 LINC00310RP11-169K16.4 RP11-131L23.2
RP4-583P15.16
WDR11-AS1
TMEM246-AS1
LINC00908
RP11-492E3.2
RP11-526F3.1
RP11-61E11.2
LIPC-AS1
RP11-213H15.1
LINC01136
HSD52
ZNF503-AS1
RP11-116O18.1
RP11-13N12.1
RP4-598P13.1
RP11-706O15.3
RP6-65G23.5
RNF157-AS1
CTD-2382E5.6
RP11-264E20.2
LINC00265
RP11-547D24.1
RP11-21A7A.4
TNK2-AS1
CRAT37
LINC01679
RP11-396O20.1
AC013264.2
FAM87A
LINC01387
RP11-713M15.2
CTD-2269F5.1
ADORA2A-AS1
RP11-193M21.1
LINC02147RP1-60O19.1
RP11-923I11.6
LINC02188
RP11-173C1.1
AC092614.2LINC00706
CTD-2506J14.1
RP11-311F12.1
AC061992.2RP11-1399P15.1
RP11-54A9.1
CALML3-AS1 C1RL-AS1AC002480.4
RP11-359N11.1
RP11-667K14.3ADAMTS9-AS1
LINC02038RP1-118J21.25
RP5-899B16.1EGOT
LINC01936
RP11-680C21.1
AC114730.2
LINC00840
CTD-3252C9.4
CTC-296K1.4
RP1-150O5.3
SLIT2-IT1
CTD-2377O17.1AC074289.1
RP11-1069G10.1ITPKB-IT1
XXbac-BPG249D20.9
LINC00924
RP11-276H7.3
CTD-2033D15.3
RP11-286H15.1
LINC02091
CTD-3094K11.1
(A)CHOL (B)KICH
C1QTNF1-AS
RP5-1009E24.9
RP11-806O11.1
RP11-563P16.1
cluster 1
cluster 2 cluster 3
cluster 1
cluster 3
cluster 4
cluster 2
C2orf15DCPS
TDRD1
TDRD9
CELF4
Figure 5 The co-expression networks of key RBPs and lncRNAs. (A) The co-expression network of twokey RBPs related toM stage (C2orf15 and DCPS) and lncRNAs for CHOL. (B) The co-expression net-work of three key RBPs (TDRD1, TDRD9 and CELF4) related toM stage and lncRNAs for KICH. Red andgreen represent up- and down-regulated DE genes.
Full-size DOI: 10.7717/peerj.7696/fig-5
The networks of key RBPs and lncRNAs for CHOL and KICHTo infer the potential regulatory mechanisms of lncRNAs with key RBPs related to Mstages, we constructed the co-expression networks of key RBPs and differentially expressed(DE) lncRNAs for CHOL and KICH respectively and performed Gene Ontology andfunctional enrichment analyses.
Ten key RBPs and 2,943 DE lncRNAs were used to construct the co-expression networkfor CHOL. The resulting co-expression network consisted of two key RBPs (down-regulation C2orf15 and up-regulation DCPS) and 75 lncRNAs, which were grouped intothree clusters (Fig. 5A). There were 63 up-regulated lncRNAs, and the proportion ofup-regulated lncRNAs in three clusters was 86%, 75%, 83.3%, which probably suggest thatup-regulated lncRNAs have a greater interaction with RBPs in the process of metastasisof CHOL cells. Functional enrichment analyzing demonstrated that cluster 1 and cluster3 had similar functions and mainly enriched in functional categories involved in genesilencing and negative regulation of translation, such as post transcriptional gene silencing,negative regulation of translation and cellular amide metabolic process, cellular responseto dsRNA, miRNA metabolic process. Cluster 2 had some special functions and mainlyenriched in such as positive regulation of mRNA catabolic process, cellular response tointerleukin-1 and calcium ion, RNA destabilization.
Seven key RBPs and 1,204 DE lncRNAs were used to construct the co-expressionnetwork for KICH. The resulting co-expression network consisting of 3 key RBPs (up-regulated TDRD1 and TDRD9, down-regulated CELF4) and 227 lncRNAs. There are 177down-regulated lncRNAs in the network and were grouped into four clusters (Fig. 5B).The corresponding proportion of down-regulated lncRNAs was 46.7%, 65.3%, 75% and95%, which probably suggest that down-regulated lncRNAs play more important roles in
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 13/22
interacting with key RBPs TDRD1, TDRD9 andCELF4 in the process of metastasis of KICHcells. Functional enrichment analysis demonstrated that cluster 1 shared by three key RBPsmainly enriched in dsRNA fragmentation and production of miRNAs involved in genesilencing by miRNA. Cluster 2 regulated by TDRD9 mainly enriched in endoribonucleaseand exon-exon junction complex, cluster 3 regulated by TDRD1mainly enriched in mRNAcatabolic process and regulation of mRNAmetabolic process, cluster 4 regulated by CELF4mainly enriched in transporting of RNA, mRNA and nucleic acids. Besides, cluster 1 andcluster 2 have some similar functions, such as telomere maintenance, histone mRNA andmiRNAmetabolic process, dosage compensation. Cluster 3 and cluster 4 have some similarfunctions, such as regulation of RNA stability, RNA localization and regulation of mRNAcatabolic process.
These results provide a new insight into the understanding of the interactions of keyRBPs with lncRNAs in the metastasis stage (M stage) of cancer cells.
The co-expression networks of DEG RBPs and TFs for CHOL andKICHTo investigate the interactions of RBPs and TFs, we constructed the co-expression networksof DEG RBPs and TFs for CHOL and KICH, respectively. The key regulatory RBPs werethose (ten for CHOL and seven for KICH) detected in above section.
The co-expression network of CHOL revealed two important insights. First, we found thefive largest transcription factor families, they areC2H2-ZF,Homeodomain,Nuclear receptor,bHLH and bZIP, and the corresponding proportion is 37%, 17%, 9%, 9% and 9% (Figs. 6A,6B), which interacted with almost all differentially expressed RBPs. This result indicatesthat these transcription factors tend to show a co-expression pattern with DEG RBPs,which further suggest that they play a major regulatory role in RBP post-regulatory levelsfor CHOL. Second,we also identified several special TFs related to up- or down-regulatedRBPs for CHOL. For example, Grainyhead, MADF, HMG-Sox and SAND are specifictranscription factors associated with down-regulated RBPs and GTF2I-like, Myb-SANT,MADS box and CENPB are specific transcription factors associated with up-regulated RBPsfor CHOL. The co-expression network for KICH also revealed the following insights. First,we found that the four largest transcription factor families, they are bHLH, bZIP, C2H2-ZFand Nuclear receptor, the proportion is 54%, 9%, 7% and 7% (Figs. 6C, 6D). Notably,transcription factor bHLH interact with all RBPs and it accounts for more than half of allinteracted transcription factors, which indicate that bHLH transcription factor probablyinvolved in regulation of all differentially expressed RBPs for KICH. Second, the proportionof up- and down-regulated RBPs co-expressed with TFs is 57% and 43% respectively. But,the up-regulated RBPs tend to show more interactions with TFs (Fig. 6C). These resultsprovide insights into understanding the mechanism of interaction between transcriptionfactors and RBPs.
DISCUSSIONRNA-binding proteins have been shown to be the key units to regulating RNA metabolism(Fu & Ares, 2014; Stefanie, Markus & Thomas, 2014; Moore & Proudfoot, 2009) and
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 14/22
bHLH
Myb-SANTT-boxSANDP53Nuclear recepotor
POUMADS boxMADF
Homeodomain
HMG-SoxGTF2I-likeGrainyhead
Forkhead
DM
CUT
CENPB
C2H2 ZF
bZIP
AT hook
ARID-BRIGHI
AP-2
Prospero
Homeodomain
Nuclear receptor
CUT
C2H2 ZF
bZIP
BED ZF
bHLH
AT hook
Ets
54%
9%
7%7% 5%
5%4%
4%
37%
17%
9%
9% 9% 2% 2%
(C) KICH-TFs
(A) CHOL-TFs
APOBEC2
SRRM4
APOBEC3G
ISG20
PURG
HMG-Sox
Grainyhead
POU
C2orf15
PUS7 KHDC1ELAVL3
BICC1MADF
SAND
SMG9
CELF4
TERT
CELF3
ENOX1
DM
PTGES3L-AARSD1POP1
IGF2BP3GTPBP2
PIWIL2
PAIP2B
Myb-SANT
LARS2
MADS box
TIPARP
CPEB4
CENPB
GTF2I-like
ZGPAT
PSTK
CPEB3
KHDRBS3
AUH
MOV10L1
AZGP1
DQX1DDX26BRBMXL2
RPP14
P53
HNRNPA1DCPS
RANBP17
ACO1
PPARGC1A
HomeodomainC2H2 ZFNuclear receptor
Forkhead
AT hook
EtsbZIP
bHLHCUT
T-box
RBMS3
MAEL
NXF3
HRSP12
DZIP1L
KHDRBS2
ZNF385A
ENOX1 MOV10L1
DDX58
CELF6
PEG10
CELF3
ELAVL4
SPATS2L ZFP36ZC3H12BANGSECISBP2L
Prospero
Homeodomain
CUT
ARID-BRIGHT
BED ZF
YBX2
XPOT
RBPMS2
UTP20
R3HCC1L
GARS
TDRD9
MECP2
CELF4
TNRC6A
ESRP1
ZCCHC14
TDRD1 POLR2F
TRNAU1AP
C2H2 ZF
IFIT1
CSDC2
ZCCHC24
Nuclear receptor
bZIP
bHLHAT hook
AP-2
(B) CHOL-Pie chart
(D) KICH-Pie chart
Figure 6 The co-expression networks of DE RBPs and TFs. (A) and (C) The co-expression networksof DE RBPs and TFs for CHOL and KICH respectively. (B) and (D) The corresponding pie chart of TFsco-expressed with DE RBPs for CHOL and KICH, respectively. Inner circle are key TFs, outer circle areDE RBPs. Red and green in outer circle represent up- and down-regulated RBPs. Block represents the keyRBPs identified by module detection, circle dot represents other DE RBPs.
Full-size DOI: 10.7717/peerj.7696/fig-6
dynamically interactwith both coding andnoncodingRNA (Kim et al., 2016). Furthermore,recent studies have shown that RBPs are down-regulated in cancers (Wang et al., 2018),but the study of 16 tissues from 80 healthy individuals indicated that RBPs show thehigher expression than non-RBPs (Bobak & Sarath, 2014). Consequently, we investigatedthe comprehensive expression differences of RBPs and non-RBPs simultaneously incancers and normal tissues. Results indicate that RBPs are significantly dysregulated incancers. In particular, recent studies have confirmed that RBPs show severely dysregulatedexpression in BLCA (Kato et al., 2012), LUAD (Dong et al., 2018), STAD (Hapkova etal., 2013), READ, LUSC (Shi et al., 2017), GBM (Pavlyukov et al., 2018) and COAD(Saki et al., 2016). Furthermore, up- and down-regulated RBPs tend to show oppositepatterns of differential expression in cancers and normal tissues (Figs. 2D, 2E). Up-regulated RBPs show higher expression in normal tissues than down-regulated RBPs,
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 15/22
which is consistent with the results in Bobak & Sarath (2014), but down-regulated RBPsshow the significantly higher expression in cancers than up-regulated RBPs (Figs. 2F, 2G),which is consistent with results in (Wang et al., 2018). These results probably suggest amechanism of RBPs in the process of carcinomatosis by which the up-regulated RBPstend to show lower expression but down-regulated RBPs tend to show higher expression.Carcinogenesis is probably caused by the combined actions of the low expression ofup-regulated RBPs and the high expression of down-regulated RBPs. This mechanism maybe useful for understanding the roles of RPBs and the design of targeted drugs for cancertherapy.
We found 10 key regulated RBPs for CHOL (Seven down-regulated RBPs, HNRNPA1,PANBP17, PUS7, KHDC1, ELAVL3, BICC1 and C2orf15; Three up-regulated RBPs, ACO1PPARGC1A and DCPS) and seven key regulated RBPs for KICH (Six down-regulatedRBPs, TNRC6A, MECP2, ZCCHC14, CELF4 and POLR2F; Two up-regulated RBPs,TDRD1 and TDRD9), respectively. Notably, recent studies have shown the importanceof these key RBPs. For instance, multiple PPARGC1A transcripts are more abundant andCNS-specific in Parkinson’s disease (PD) (Soyal et al., 2019). KHDC1A is highly expressedin oocytes and induces endoplasmic reticulum apoptosis (Cai et al., 2012). Elavl3 is closelyrelated to neurodegenerative diseases and play an important role in maintaining theaxonal homeostasis of neurons (Ogawa et al., 2018).HNRNPA1, regulated by miR-503 andmiR-424, is associated with breast cancer cell proliferation (Otsuka, Yamamoto & Ochiya,2018). DCPS is very essential for acute myeloid leukemia cell survival by interacting withpre-mRNA (Yamauchi et al., 2018). CELF4 plays an important role in brain development,the haploinsufficiency of CELF4 is associated with autism disorders (Barone et al., 2017).POLR2F is significantly high expression in colorectal carcinomas (Antonacopoulou et al.,2008) and potential molecule in carcinogenesis.TDRD1 is over-expressed inmajority of 131primary prostate tumors patients (Xiao et al., 2016). In all, these results have demonstratedthat the key RBPs have played the important roles in other types of cell carcinomatosis andprovide a new perspective for potential prognostic biomarkers and therapeutic targets forpatients with cancers at the N andM stages in two cancer types CHOL and KICH.
CONCLUSIONSIn this study, we analyzed detailed differences in the expression of RBPs and non-RBPsacross 20 types of cancers and constructed the co-expression networks of dysregulated RBPswith TFs and lncRNAs for CHOL and KICH, respectively. Our results indicate that: (1)RBPs are dysregulated in almost all 20 cancer types comparedwith normal tissues, especiallyin BLCA, COAD, READ, STAD, LUAD, LUSC and GBM with proportion of deviationlarger than 300% compared with non-RBPs in normal tissues. (2) Up- and down-regulatedRBPs also show opposed patterns of differential expression in cancers and normal tissues.In addition, down-regulated RBPs show a greater degree of dysregulated expression thanup-regulated RBPs do. (3) Clinical TNM information for two cancer types—CHOL andKICH—is shown to be closely related to patterns of differentially expressed RBPs (DERBPs). (4) We constructed the co-expression networks of key RBPs between 1,570 TFs
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 16/22
and 4,147 lncRNAs for CHOL and KICH, respectively. By analyzing these networks, weidentified ten key RBPs (of which seven were down-regulated and three up-regulated)in CHOL and seven RBPs (of which five were down-regulated and two up-regulated) inKICH. These key RBPs—and especially down-regulated RBPs—likely play important rolesin cell carcinomatosis. This study lays the foundation for further efforts to understand theroles played by RBPs in human carcinogenesis and provides a new insight into identifyingthe potential prognostic biomarkers and therapeutic targets for patients.
ACKNOWLEDGEMENTSThe authors thank three anonymous reviewers for their comments on the manuscript. Thelinguistic editing and proofreading provided by TopEdit LLC during the preparation ofthis manuscript are acknowledged.
ADDITIONAL INFORMATION AND DECLARATIONS
FundingThis work was supported by the National Natural Science Foundation of China (Grant.61501392) and the Nanhu Scholars Program for Young Scholars of XYNU (Xin YangNormal University). The funders had no role in study design, data collection and analysis,decision to publish, or preparation of the manuscript.
Grant DisclosuresThe following grant information was disclosed by the authors:National Natural Science Foundation of China: 61501392.Nanhu Scholars Program for Young Scholars of XYNU (Xin Yang Normal University).
Competing InterestsThe authors declare there are no competing interests.
Author Contributions• Shuaibin Lian conceived and designed the experiments, contributed reagents/material-s/analysis tools, authored or reviewed drafts of the paper, approved the final draft.• Liansheng Li performed the experiments, analyzed the data, contributed reagents/mate-rials/analysis tools, prepared figures and/or tables, approved the final draft.• Yongjie Zhou analyzed the data, contributed reagents/materials/analysis tools, preparedfigures and/or tables, approved the final draft.• Zixiao Liu analyzed the data, contributed reagents/materials/analysis tools, approvedthe final draft.• Lei Wang conceived and designed the experiments, approved the final draft.
Data AvailabilityThe following information was supplied regarding data availability:
Raw data is available in the Supplemental Files.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 17/22
Supplemental InformationSupplemental information for this article can be found online at http://dx.doi.org/10.7717/peerj.7696#supplemental-information.
REFERENCESAmerican Joint Committee on Cancer. 2010. Cancer staging manual. 7 edition. Chicago:
Springer, p649.Anders S. 2009. Differential gene expression analysis based on the negative binomial
distribution. Journal of Marine Technology & Environment 2:Article 2.Antonacopoulou AG, Grivas PD, Skarlas L, Kalofonos M, Scopa CD, Kalofonos
HP. 2008. POLR2F, ATP6V0A1 and PRNP expression in colorectal cancer: newmolecules with prognostic significance. Anticancer Research 28(2B):1221–1227DOI 10.1245/s10434-007-9682-3.
Barone R, Fichera M, De Grandi M, Battaglia M, Lo Faro V, Mattina T, Rizzo R. 2017.Familial 18q12.2 deletion supports the role of RNA-binding protein CELF4 in autismspectrum disorders. American Journal of Medical Genetics A 173(6):1649–1655DOI 10.1002/ajmg.a.38205.
Bobak K, Sarath CJ. 2014. Dissecting the expression landscape of RNA-binding proteinsin human cancers. Genome Biology 15:R14 DOI 10.1186/gb-2014-15-1-r14.
Boguslawska J, Sokol E, Rybicka B, Czubaty A, Rodzik K, Piekielko-WitkowskaA. 2016.MicroRNAs target SRSF7 splicing factor to modulate the expres-sion of osteopontin splice variants in renal cancer cells. Gene 595(2):142–149DOI 10.1016/j.gene.2016.09.031.
Cai C, Liu J, Wang C, Shen J. 2012. KHDC1A, a novel translational repressor, in-duces endoplasmic reticulum-dependent apoptosis. DNA and Cell Biology31(9):1447–1457 DOI 10.1089/dna.2012.1682.
Cao B, Fang Z, Liao P, Zhou X, Xiong J, Zeng S, Lu H. 2017b. Cancer-mutated ribosomeprotein L22 (RPL22/eL22) suppresses cancer cell survival by blocking p53-MDM2circuit. Oncotarget 8(53):90651–90661 DOI 10.18632/oncotarget.21544.
Cao TT, Lin SH, Fu L, Tang Z, Che CM, Zhang LY, Ming XY, Liu TF, Tang XM, Tan BB,Xiang D, Li F, Chan OY, Xie D, Cai Z, Guan XY. 2017a. Eukaryotic translation ini-tiation factor 5A2 promotes metabolic reprogramming in hepatocellular carcinomacells. Carcinogenesis 38(1):94–104 DOI 10.1093/carcin/bgw119.
Chang K, Creighton CJ, Davis C, Donehower L, Drummond J, Wheeler D. 2013. Thecancer genome atlas pan-cancer analysis project. Nature Genetics 45(10):1113–1120DOI 10.1038/ng.2764.
Dai J, WeiW. 2017. Influence of the RPL34 gene on the growth and metastasisof oral squamous cell carcinoma cells. Archives of Oral Biology 83:40–46DOI 10.1016/j.archoralbio.2017.06.035.
Deplancke B, Alpern D, Gardeux V. 2016. The genetics of transcription factor DNAbinding variation. Cell 166(3):538–554 DOI 10.1016/j.cell.2016.07.012.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 18/22
Dong F, Li C,Wang P, Deng X, Luo Q, Tang X, Xu L. 2018. The RNA binding proteintristetraprolin down-regulates autophagy in lung adenocarcinoma cells. ExperimentalCell Research 367(1):89–96 DOI 10.1016/j.yexcr.2018.03.028.
Drissi R, Dubois M-L, DouziechM, Boisvert F-M. 2015. Quantitative proteomics revealsdynamic interactions of the minichromosome maintenance complex (mcm) in thecellular response to etoposide induced dna damage.Molecular & Cellular Proteomics14(7):2002–2013 DOI 10.1074/mcp.M115.048991.
Fu XD, Ares M. 2014. Context-dependent control of alternative splicing by RNA-bindingproteins. Nature Reviews Genetics 15:689–701 DOI 10.1038/nrg3778.
Gibb EA, Brown CJ, LamWL. 2011. The functional role of long non-coding RNA in hu-man carcinomas.Molecular Cancer 10(1):Article 38 DOI 10.1186/1476-4598-10-38.
Gloss BS, Dinger ME. 2015. The specificity of long noncoding RNA expression. Biochim-ica et Biophysica Acta 1859(1):16–22 DOI 10.1016/j.bbagrm.2015.08.005.
Hapkova I, Skarda J, Rouleau C, Thys A, Notarnicola C, JanikovaM, Bernex F, RypkaM, Vanderwinden JM, Faure S, Vesely J, De Santa Barbara P. 2013.High expressionof the RNA-binding protein RBPMS2 in gastrointestinal stromal tumors. Experimen-tal and Molecular Pathology 94(2):314–321 DOI 10.1016/j.yexmp.2012.12.004.
HuX, Luo J, Lai H, Li M, Zheng X, Nie T, Li F, Li H. 2018. Knockdown of Trnau1apinhibits the proliferation and migration of NIH3T3, JEG-3 and Bewo cells via thePI3K/Akt signaling pathway. Biochemical and Biophysical Research Communications503(2):521–527 DOI 10.1016/j.bbrc.2018.05.065.
Huang DW, Sherman BT, Lempicki RA. 2009a. Bioinformatics enrichment tools: pathstoward the comprehensive functional analysis of large gene lists. Nucleic AcidsResearch 37(1):1–13 DOI 10.1093/nar/gkn923.
Huang DW, Sherman BT, Lempicki RA. 2009b. Systematic and integrative analysis oflarge gene lists using DAVID Bioinformatics Resources. Nature Protoc 4(1):44–57DOI 10.1038/nprot.2008.211.
Humphries MP, Sundara Rajan S, Droop A, Suleman CAB, Carbone C, Nilsson C,Honarpisheh H, Cserni G, Dent J, Fulford L, Jordan LB, Jones JL, Kanthan R,LitwiniukM, Di Benedetto A, Mottolese M, Provenzano E, Shousha S, StephensM,Walker RA, Kulka J, Ellis IO, Jeffery M, Thygesen HH, Cappelletti V, DaidoneMG, Hedenfalk IA, FjällskogML, Melisi D, Stead LF, Shaaban AM, Speirs V. 2017.A case-matched gender comparison transcriptomic screen identifies eif4e and eif5as potential prognostic markers in male breast cancer. Clinical Cancer Research23(10):2575–2583 DOI 10.1158/1078-0432.CCR-16-1952.
Ishizuka A, Hasegawa Y, Ishida K, Yanaka K, Nakagawa S. 2014. Formation of nuclearbodies by the lncRNA Gomafu-associating proteins Celf3 and SF1. Genes to Cells19(9):704–721 DOI 10.1111/gtc.12169.
KatoM,Wei M, Yamano S, Kakehashi A, Tamada S, Nakatani T,Wanibuchi H. 2012.DDX 39 acts as a suppressor of invasion for bladder cancer. Cancer Science 103(7)DOI 10.1111/j.1349-7006.2012.02298.x.
Kechavarzi B, Janga S. 2014. Dissecting the expression landscape of RNA-binding pro-teins in human cancers. Genome Biology 15:R14, 1–6 DOI 10.1186/gb-2014-15-1-r14.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 19/22
Kim BK, Yoo HI, Choi K, Lee AR, Yoon SK. 2016. Regulation of Srpr Expression bymiR-330-5p controls proliferation of mouse epidermal keratinocyte. PLOS ONE11(10):e0164896 DOI 10.1371/journal.pone.0164896.
Lambert SA, Jolma A, Campitelli LF, Das PK, Yin Y, AlbuM, Chen X, Taipale J, HughesTR,WeirauchMT. 2018. The human transcription factors. Cell 172(4):650–665DOI 10.1016/j.cell.2018.01.029.
Leah LZ, Sarah EG, Thomas J. 2015. Sweet, Molly M. Hannigan, R. Michael Sramkoski,Qin Li, Donny D. Licatalosi. RNA Binding Protein Ptbp2 is essential for malegerm cell development.Molecular and Cellular Biology 35(23):4030–4042DOI 10.1128/MCB.00676-15.
Lian S, Liu T, Jing S, Yuan H, Zhang Z, Cheng L. 2018. Intrachromosomal colocalizationstrengthens co-expression, co-modification and evolutionary conservation ofneighboring genes. BMC Genomics 19(1):455 DOI 10.1186/s12864-018-4844-1.
Liu H, Liang S, Yang X, Ji Z, ZhaoW, Ye X, Rui J. 2015. RNAi-mediated RPL34knockdown suppresses the growth of human gastric cancer cells. Oncology Reports34(5):2267–2272 DOI 10.3892/or.2015.4219.
Molly MH, Leah LZ, Donny DL. 2017. Ptbp2 controls an alternative splicing network re-quired for cell communication during spermatogenesis. Cell Rep 19(12):2598–2612DOI 10.1016/j.celrep.2017.05.089.
MooreMJ, Proudfoot NJ. 2009. Pre-mRNA processing reaches back to transcription andahead to translation. Cell 136:688–700 DOI 10.1016/j.cell.2009.02.001.
Ogawa Y, Kakumoto K, Yoshida T, Kuwako KI, Miyazaki T, Yamaguchi J, Konno A,Hata J, Uchiyama Y, Hirai H,WatanabeM, Darnell RB, Okano H, Okano HJ. 2018.Elavl3 is essential for the maintenance of Purkinje neuron axons. Scientific Reports8(1):2722 DOI 10.1038/s41598-018-21130-5.
Otsuka K, Yamamoto Y, Ochiya T. 2018. Regulatory role of resveratrol, a microRNA-controlling compound, in HNRNPA1 expression, which is associated with poorprognosis in breast cancer. Oncotarget 9(37):24718–24730DOI 10.18632/oncotarget.25339.
PavlyukovMS, Yu H, Bastola S, Minata M, Shender VO, Lee Y, Zhang S, Wang J,Komarova S, Wang J, Yamaguchi S, Alsheikh HA, Shi J, Chen D, MohyeldinA, Kim SH, Shin YJ, Anufrieva K, Evtushenko EG, Antipova NV, Arapidi GP,Govorun V, Pestov NB, ShakhparonovMI, Lee LJ, NamDH, Nakano I. 2018.Apoptotic cell-derived extracellular vesicles promote malignancy of glioblas-toma via intercellular transfer of splicing factors. Cancer Cell 34(1):119–135DOI 10.1016/j.ccell.2018.05.012.
Peter L, Steve H. 2008.WGCNA: an R package for weighted correlation network analysis.BMC Bioinformatics 9:559 DOI 10.1186/1471-2105-9-559.
Piao J, Chen L, Jin T, XuM, Quan C, Lin Z. 2018. Paip1 affects breast cancer cellgrowth and represents a novel prognostic biomarker. Human Pathology 73:33–40DOI 10.1016/j.humpath.2017.10.037.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 20/22
Qin L, Zheng S, Han A, Lin C-H, Stoilov P, Fu X-D, Black DL. 2014. The splicingregulator PTBP2 controls a program of embryonic splicing required for neuronalmaturation. eLife 3:e01201 DOI 10.7554/eLife.01201.
RobinsonM,Mccarthy D, Smyth GK. 2010. edgeR: differential expression analysis ofdigital gene expression data. Journal of Hospice & Palliative Nursing 4(4):206–207DOI 10.1093/bioinformatics/btp616.
Saijo S, Kuwano Y, Masuda K, Nishikawa T, Rokutan K, Nishida K. 2016. Serine/arginine-rich splicing factor 7 regulates p21-dependent growth arrest in colon cancer cells.The Journal of Medical Investigation 63(3.4):219–226 DOI 10.2152/jmi.63.219.
Saki S, Yuki K, Kiyoshi M, Tatsuya N, Kazuhito R, Kensei N. 2016. Serine/arginine-richsplicing factor 7 regulates p21-dependent growth arrest in colon cancer cells. TheJournal of Medical Investigation 63(3.4):219–226 DOI 10.2152/jmi.63.219.
Shi R, Yu X,Wang Y, Sun J, Sun Q, XiaW. 2017. Expression profile, clinical signif-icance, and biological function of insulin-like growth factor 2 messenger RNA-binding proteins in non–small cell lung cancer. Tumor Biology 39(4):Article 4DOI 10.1177/1010428317695928.
Sobin LH, Gospodarowicz MK,Wittekind C. 1992. TNM Classification of MalignantTumours, 7th Edition// TNM classification of malignant tumours. New York:Springer-Verlag DOI 10.1057/9780230271357_47.
Soyal SM, Zara G, Ferger B, Felder TK, KwikM, Nofziger C, Dossena S, Schwien-bacher C, Hicks AA, Pramstaller PP, Paulmichl M,Weis S, PatschW. 2019.The PPARGC1A locus and CNS-specific PGC-1α isoforms are associated withParkinson’s Disease Neurobiology of Disease. Neurobiology of Disease 121:34–46DOI 10.1016/j.nbd.2018.09.016.
St. Laurent G,Wahlestedt C, Kapranov P. 2015. The Landscape of long noncoding RNAclassification. Trends in Genetics 31(5):239–251 DOI 10.1016/j.tig.2015.03.007.
Stefanie G, Markus H, Thomas T. 2014. A census of human RNA-binding Proteins.Nature Reviews Genetics 15:829–845 DOI 10.1038/nrg3813.
Vergara IA, Erho N, Triche TJ, Ghadessi M, Crisan A, Sierocinski T, Black PC,Buerki C, Davicioni E. 2012. Genomic ‘‘Dark Matter’’ in prostate cancer: ex-ploring the clinical utility of ncRNA as biomarkers. Frontiers in Genetics 3DOI 10.3389/fgene.2012.00023.
Wang T, Ma S, Qi X, Tang X, Cui D,Wang Z, Chi J, Li P, Zhai B. 2016. Long non-coding RNA ZNFX1-AS1 suppresses growth of hepatocellular carcinoma cellsby regulating the methylation of miR-9. OncoTargets and Therapy 9:5005–5014DOI 10.2147/OTT.S103329.
Wang Z, Li B, Luo Y, Lin Q, Liu SR, Zhang XQ, Zhou H, Yang JH, Qu LH. 2018.Comprehensive Genomic Characterization of RNA-Binding Proteins across HumanCancers. Cell Reports 2:286–298 DOI 10.1016/j.celrep.2017.12.035.
Xiao L, Lanz RB, Frolov A, Castro PD, Zhang Z, Dong B, XueW, Jung SY, Lydon JP,Edwards DP, Mancini MA, Feng Q, IttmannMM, He B. 2016. The germ cell geneTDRD1 as an ERG target gene and a novel prostate cancer biomarker. Prostate Oct76(14):1271–1284 DOI 10.1002/pros.23213.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 21/22
Yamauchi T, Masuda T, Canver MC, Seiler M, Semba Y, Shboul M, Al-RaqadM,MaedaM, Schoonenberg VAC, Cole MA, Macias-Trevino C, Ishikawa Y, Yao Q, NakanoM, Arai F, Orkin SH, Reversade B, Buonamici S, Pinello L, Akashi K, Bauer DE,Maeda T. 2018. Genome-wide CRISPR-Cas9 screen identifies leukemia-specificdependence on a Pre-mRNA metabolic pathway regulated by DCPS. Cancer Cell33(3):386–400 DOI 10.1016/j.ccell.2018.01.012.
Yang S, Cui J, Yang Y, Liu Z, Yan H, Tang C,Wang H, Qin H, Li X, Li J, WangW,Huang Y, Gao H. 2016. Over-expressed RPL34 promotes malignant proliferationof non-small cell lung cancer cells. Gene 576(1):421–428DOI 10.1016/j.gene.2015.10.053.
Yang Y, Quan L, Ling Y. 2018. RBMS3 inhibits the proliferation and metastasis of breastcancer cells. Oncology Research 26(1):9–15DOI 10.3727/096504017X14871200709504.
Yin ZH, Jiang XW, ShiWB, Gui QL, Yu DF. 2017. Expression and clinical significance ofILF2 in gastric cancer. Disease Markers 2017(1):1–9 DOI 10.1155/2017/4387081.
Yuan JH, Liu XN,Wang TT, PanW, Tao QF, ZhouW. 2017. The MBNL3 splicingfactor promotes hepatocellular carcinoma by increasing PXN expression through thealternative splicing of lncRNA-PXN-AS1. Nature Cell Biology DOI 10.1038/ncb3538.
Zhang X, Shen B, Cui Y. 2019. Ago HITS-CLIP expands microRNA-mRNA interactionsin nucleus and cytoplasm of gastric cancer cells. BMC Cancer 19(1):Article 29DOI 10.1186/s12885-018-5246-0.
Zhou D, Teng F, Roel GWV, Zhen S, Yong Z, Myles B, Yiwen C, Liu XS. 2013. Integra-tive genomic analyses reveal clinically relevant long noncoding RNAs in human can-cer. Nature Structural & Molecular Biology 20(7):908–913 DOI 10.1038/nsmb.2591.
Lian et al. (2019), PeerJ, DOI 10.7717/peerj.7696 22/22