The complex transcriptional landscape of the anucleate human platelet

Abstract

Background: Human blood platelets are essential to maintaining normal hemostasis, and platelet dysfunction often causes bleeding or thrombosis. Estimates of genome-wide platelet RNA expression using microarrays have provided insights to the platelet transcriptome but were limited by the number of known transcripts. The goal of this effort was to deep-sequence RNA from leukocyte-depleted platelets to capture the complex profile of all expressed transcripts.

Results: From each of four healthy individuals we generated long RNA (≥40 nucleotides) profiles from total and ribosomal-RNA depleted RNA preparations, as well as short RNA (<40 nucleotides) profiles. Analysis of ~1 billion reads revealed that coding and non-coding platelet transcripts span a very wide dynamic range (≥16 PCR cycles beyond β-actin), a result we validated through qRT-PCR on many dozens of platelet messenger RNAs. Surprisingly, ribosomal-RNA depletion significantly and adversely affected estimates of the relative abundance of transcripts. Of the known protein-coding loci, ~9,500 are present in human platelets. We observed a strong correlation between mRNAs identified by RNA-seq and microarray for well-expressed mRNAs, but RNASeq identified many more transcripts of lower abundance and permitted discovery of novel transcripts.

Conclusions: Our analyses revealed diverse classes of non-coding RNAs, including: pervasive antisense transcripts to protein-coding loci; numerous, previously unreported and abundant microRNAs; retrotransposons; and thousands of novel un-annotated long and short intronic transcripts, an intriguing finding considering the anucleate nature of platelets. The data are available through a local mirror of the UCSC genome browser and can be accessed at: http://cm.jefferson.edu/platelets_2012/.

Figure 1

Estimates of platelet expressed mRNAs. A) Total RNA. B) rRNA-depleted RNA. The x-axis shows the RNA-seq read number in log2 ratios normalized to β-actin; the y-axis shows the number of expressed mRNAs. The result for total RNA from donor 2 N2 is an outlier with lower abundances.

Figure 2

Correlation of platelet mRNA levels assessed by RNA-seq and qRT-PCR. ΔC_t values obtained by qRT-PCR (y-axis) were plotted against the log₂-normalized transcript determined by RNA-seq (x-axis). Both methods normalized to β-actin expression. Transcripts were considered “present” in the qRT-PCR with a cycle threshold of ≤35. RNA-seq transcripts were considered that were no lower than a normalized log2 expression value of -15 (i.e., 15 PCR cycles [≥ 1/32,768^th] of β-actin expression). The transcript in the right lower quadrant represents β₂-microblobulin, which is expressed at a higher level than β-actin. Black points derive from microarray; red points were selected as known, representative platelet genes.

Figure 3

Correlation heatmap matrix for RNA-seq vs. microarray analysis of the platelet transcriptome. A) To compare the protein-coding transcripts as deduced from RNA-seq and previous microarray analyses (Affymetrix GeneChip and Illumina BeadChip) and also microarrays with one another, we used a Spearman correlation computed from the union of protein-encoding genes (13,691 in all) that were represented on at least one of the platforms. B) To compare the RNA-seq datasets with one another, we computed Pearson’s correlation between the genomic transcript profiles obtained by each dataset. In both A) and B), each square lists the correlation coefficient value between the corresponding profiles; also, the color-coding convention is the same in order to facilitate comparisons.

Figure 4

Gene Ontology (GO) analysis of the platelet transcriptome by RNA-seq. Top-ranking biological processes by gene number (panels A and C) and by categories (panels B and D) that emerge from a GORILLA analysis of those protein-coding transcripts common to the four sequenced individuals. GO terms and p-values were computed and are shown separately for the total RNA (panels A and B) and the rRNA-depleted preparations (panels C and D). Note that the GO category of "coagulation" includes all aspects of platelet biology. See also Figure S1.