CD74, MIF and Breast Tumorigenesis: Insights from Recent Large-Scale Tumour Genomics and Proteomics Studies

Abstract

MIF and other cytokines are frequently detected at elevated levels of abundance in solid tumours. Their involvement in tumour biology has been studied for many years, and, with the advent of postgenomic tools such as next-generation DNA and RNA sequencing, and mass spectrometry-driven protein profiling, the underlying mechanisms can be studied in a systematic and quantitative way. This chapter discusses recent studies by our group that have shown that MIF and CD74 are mechanistically involved in breast cancer progression. Analysis of recently released data from the Cancer Genome Atlas (TCGA) as well as our proteomics data is presented and discussed. TCGA data show that MIF and CD74 are rarely mutated in cancer but are consistently overexpressed at the level of mRNA. Furthermore, using high-resolution mass spectrometry to analyse tumour protein abundance, we have identified MIF and CD74 among the proteins that are overexpressed in metastatic triple-negative breast tumours. A cell-based model showed that when CD74 is overexpressed, it interferes with the function of a known tumour suppressor, Scribble, leading to enhanced invasion, possibly because the functions of Scribble in maintaining cell polarity are compromised. The underlying mechanism, yet to be fully elucidated, involves deregulation of Scribble phosphorylation on specific sites in its C-terminal proline-rich domain.

Appendix

R code to merge individual RNA-Seq data files, cluster by CD74 and MIF expression, perform survival analysis and find genes which are co-regulated with MIF and CD74 in breast tumours. RNA-Seq data files are downloaded along with clinical information from TCGA (https://tcga-data.nci.nih.gov). The individual files are then merged and transposed to create a data frame containing patient data in rows and genes in columns. This is merged with the clinical information table using patients’ barcodes. Survival analysis is then performed using the R package “Survival”. Co-regulated genes are identified by calculating the Spearman rank-based correlation coefficient.

#RNA-seq import and merging #Download data files from TCGA site. Unpack into working directory temp = list.files(pattern="*.genes.normalized_results") myfiles = lapply(temp, read.delim) data<- NULL data<- myfiles[1] for (i in 2:length(myfiles)){ data<-merge(data, myfiles[i], by="gene_id") } names(data)<- c("gene_id", temp)) write.csv(data, "RNAseq_breast.csv") rownames(data)<-data$gene_id data$gene_id<-NULL data.breast<-t(data) #Download clinical data and import into R. Merge with the RNA-seq data using patients barcodes to produce a #dataframe called “data.tumors.only #Then get MIF and CD74 expression levels: cd74<- data.tumors.only[,3483] #CD74 is column 3483 in the table mif<- data.tumors.only[,11053] #MIF is column 11053 in the table #Cluster patients by expression of CD74 and MIF dataCD74MIF<-data.frame(cd74,mif) hc<- hclust(dist(scale(log(dataCD74MIF))), "ward.D2") plot(hc) #Look at the clustering and save as graphics file cl<- cutree(hc, 2) #Use clustering to do survival analysis library(survival) survival<- as.numeric(ifelse(data.tumors.only$days_to_last_followup!="[Not Available]", data.tumors.only$days_to_last_followup,data.tumors.only$days_to_death)) vital<- ifelse(data.tumors.only$vital_status!="Dead", 0,1) surv<-Surv(survival,vital) survdiff(surv~cl)#calculate p-value fit<- survfit(surv~cl) plot(fit, col=c("grey", "black"), xlim=c(0,5000), ylim=c(0.2,1), cex=0.5, xlab="Days to event", ylab="Survival") #Analysis of co-expressed proteins: calculate Spearman rho for all proteins against CD74 and MIF corMIF<- apply(data.tumors.only[,-c(1:110)], 2, function(x) cor(x, mif, method="spearman")) corCD74<- apply(data.tumors.only[,-c(1:110)], 2, function(x) cor(x, cd74, method="spearman")) write.csv(cbind(names, corMIF), "corMIF.csv", row.names=F) write.csv(cbind(names, corCD74), "corCD74.csv", row.names=F)