Prospective, international, multisite comparison of platelet isolation techniques for genome-wide transcriptomics: communication from the SSC of the ISTH

Abstract

Genome-wide platelet transcriptomics is increasingly used to uncover new aspects of platelet biology and as a diagnostic and prognostic tool. Nevertheless, platelet isolation methods for transcriptomic studies are not standardized, introducing challenges for cross-study comparisons, data integration, and replication. In this prospective multicenter study, called "Standardizing Platelet Transcriptomics for Discovery, Diagnostics, and Therapeutics in the Thrombosis and Hemostasis Community (STRIDE)" by the International Society on Thrombosis and Haemostasis Scientific and Standardization Committees, we assessed how 3 of the most commonly used platelet isolation protocols influence metrics from next-generation bulk RNA sequencing and functional assays. Compared with washing alone, more stringent removal of leukocytes by anti-CD45 beads or PALL filters resulted in a sufficient quantity of RNA for next-generation sequencing and similar quality of RNA sequencing metrics. Importantly, stringent removal of leukocytes resulted in the lower relative expression of known leukocyte-specific genes and the higher relative expression of known platelet-specific genes. The results were consistent across enrolling sites, suggesting that the techniques are transferrable and reproducible. Moreover, all 3 isolation techniques did not influence basal platelet reactivity, but agonist-induced integrin αIIbβ₃ activation is reduced by anti-CD45 bead isolation compared with washing alone. In conclusion, the isolation technique chosen influences genome-wide transcriptional and functional assays in platelets. These results should help the research community make informed choices about platelet isolation techniques in their own platelet studies.

Keywords: leukocytes; next-generation RNA-seq; platelet transcriptomics; platelets.

Figure 1.. Schematic of the overall study design.

Blood was drawn, and platelets were isolated from 53 healthy participants recruited from three European Union (EU) sites and one site in the United States of America (USA). Once isolated, platelets were lysed in TRIzol^™, an RNA-preserving buffer, and platelet lysates were frozen at −80°C. Following the completion of participant recruitment, samples were sent on dry ice to the University of Utah, where platelet RNA was isolated and sequenced on an Illumina platform.

Figure 2.. Platelet isolation technique influences RNA yield.

(A) Percentage of platelets recovered from whole blood and each isolation technique, (B) platelet RNA concentration, (C) total platelet RNA in platelet isolations prepared by washing, LDPs, or PALL^™ filtration (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, ns = not significant). Statistical significance was determined using one-way ANOVA using Tukey’s multiple comparisons test. Violin plots show median plus interquartile range (IQR).

Figure 3.. RNA sequencing metrics were similar across platelet isolation techniques.

(A) Histogram of GC content, (B) RNA read mapping, (C) base assignment (UTR = untranslated region), and (D) histogram of Phred scores in platelet isolations prepared by washing, LDPs, or PALL filtration.

Figure 4.. PTPRC expression is associated with the highest variance in the platelet transcriptome.

(A) Principal component (PC) analyses of all 114 platelet RNA samples from 4 enrollment sites based on isolation technique. The left PC plot is before adjustment for biological sex, while the right PC plot is after adjustment for biological sex. (B) PC analyses were separated by individual enrolling site after adjustment for biological sex. Note that Site 2 also isolated platelets by PALL filtration (green). (C) Multivariate analysis with violin plots shows the fraction of mean-variance within each sample attributable to each variable (RIN = RNA integrity score). The thick horizontal line indicates the median with the interquartile range (IQR). Each dot represents an RNA-seq dataset that lies outside the IQR.

Figure 5.. The number of residual leukocytes and the relative expression of the leukocyte transcript PTPRC and the platelet transcript ITGA2B are influenced by platelet isolation technique.

Residual leukocyte counts, plotted as mean ± SEM, were measured by (A) Sysmex or (B) flow cytometry in washed platelets, LDPs, and PALL^™ filtered platelets. (C-D) Normalized RNA expression was plotted as a median with interquartile range (C) PTPRC and (D) ITGA2B in washed platelets, LDPs, and PALL^™ filtered platelets (E) Correlation between PTPRC and ITGA2B expression in platelet preparations (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, ns = not significant). Statistical significance was determined using one-way ANOVA using Tukey’s multiple comparisons test and Pearson’s correlation).

Figure 6.. Platelet isolation technique influences integrin αIIbβ3 activation.

(A) Basal integrin αIIbβ₃ activation as assessed by PAC-1 binding measured by flow cytometry on washed platelets, LDPs, or PALL^™ filtered platelets at Site 2. (B) Integrin αIIbβ3 activation as assessed by PAC-1 binding in unstimulated (resting), ADP-stimulated, or PAR1-activating peptide (SFLLRN, also called TRAP) stimulated washed platelets and LDPs at Site 3 (****p<0.0001, ns = not significant). Statistical significance was determined using one-way ANOVA using Tukey’s multiple comparisons test. Violin plots show median plus IQR.