Advertisement

Comparative genomics: fishing nets hemostatic catch

Andrew S. Weyrich and Guy A. Zimmerman

In this issue of Blood, Watkins and colleagues report a detailed analysis of gene expression profiles in human blood cells and precursors that identifies genes involved in lineage commitment and cellular responses as well as new candidates with uncharacterized functions. In our new online series, e-Blood, O'Connor and colleagues report that a functional screen in zebrafish yields evidence that a subset of candidate gene products—each a platelet membrane protein—have previously unrecognized activities in thrombus formation. Together, these studies provide additional support for genomic surveys as tools for cataloging patterns of expressed genes in human hematopoietic cells and for employing zebrafish as a surrogate system for initial examination of proteins of unknown functions as potential modulators of hemostasis.

Several groups have used transcript profiling to characterize mRNA expression patterns in murine and human blood cells. Although interesting data have emerged, variations in sample representation, purity of starting cell populations, sensitivity, scope of surveyed genes, and rigor of post hoc analysis limits the findings in some cases. Watkins et al1 performed replicate hybridizations starting with mRNA isolated from positively selected circulating human leukocyte subtypes (granulocytes, monocytes, cytotoxic and helper T cells, B lymphocytes, natural killer cells) separated from the blood of individual donors, and interrogated the samples using whole genome microarrays; model erythroblasts and megakaryocytes differentiated from cord blood hematopoietic progenitor cells2 were also included (see figure). The resulting individual cell profiles were compiled into an expression atlas and examined for key features and comparisons using bioinformatics approaches. Transcription factors, immunoglobulin superfamily members, and subsets of gene products that may have novel roles in hemostasis and thrombosis were of particular interest, as were comparisons between expressed transcripts in individual human hematopoietic cell types. An additional feature of interest was comparison of the human data sets and expression profiles from analogous murine blood cells, using findings reported by others in a separate study. The percent of total transcripts expressed in both human and mouse cell subtypes ranged from 63% to 75%, and the overlap in transcription factor expression was 49% to 58%. While these comparisons indicate significant similarities in ortholog expression in the 2 species, they also suggest substantial differences that may account for some of the limitations of mouse models intended to reproduce features of human immune and inflammatory diseases.3

Fishing for hemostatic proteins using comparative genomics and zebrafish models. Watkins et al compared gene expression patterns in primary human myeloid leukocytes, lymphocytes, and natural killer cells, and in model erythroblastic and megakaryocytic cells cultured from hematopoietic precursors by using whole genome chip microarray technology. Bioinformatics analysis of the resulting atlas of data identified lineage-specific genes, co-expression patterns, similarities and differences in the patterns of expressed genes in analogous murine blood cells, and genes whose protein products are candidates for new functional roles. O'Connor et al used this database to examine a subset of human platelet membrane proteins without established hemostatic functions— identified as candidates by the presence of their mRNA transcripts in megakaryocytic cells in the study by Watkins et al—for activities in experimental thrombosis. The strategy involved identification of orthologs corresponding to human platelet proteins in zebrafish cells, “knockdown” of the zebrafish orthologs using antisense morpholino oligonucleotides, and examination of responses of the morpholino-treated fish in a laser-induced thrombosis model. Reduced expression of 4 candidate proteins resulted in altered thrombus formation. Functional roles of the corresponding human proteins now merit evaluation in relevant models based on this screening strategy.

In a parallel study, O'Connor et al used information from the multiple cell, whole genome expression atlas,1 and from a previous examination of expression patterns in human megakaryocytic cells differentiated in vitro,3 focusing on a subset of genes encoding putative transmembrane proteins in megakaryocytes and platelets that might have uncharted roles in thrombus formation.4 They documented expression of transcripts and protein products of candidate genes in primary human platelets from blood, adding evidence for use of examination of the platelet transcriptome and proteome,5,6 and also identified orthologs of the candidate genes in zebrafish (Dario rerio).4 The strategy here was to attempt to predict functions of the human proteins using a “reverse genetic” screen in which expression of the fish orthologs was “knocked down” and the phenotype, or lack of it, was determined in an induced thrombus model.4 Zebrafish have been touted as unique surrogate systems for reverse genetic analysis, and their potential use in specific studies of hemostasis is supported by similarities between blood cells and humoral coagulation pathways in humans and this fish species.4,7,8

The fishing expedition looking for new functional proteins was a success, and O'Connor et al netted 4 factors with roles in laser-induced thrombosis (2 promoters, 2 modulators) in the surrogate zebrafish system. The phenotypes presumably indicate activities of the fish orthologs when they are expressed on circulating thrombocytes, but a caveat is that knockdown by morpholino antisense technology, which was the approach used, is not cell specific. Thus, one or more of the proteins could also have activities on endothelial cells or other cell types.4,8 Nevertheless, the results of the reverse genetic screen are enticing enough to merit evaluation of the candidate proteins in human platelets—where their functions remain in question—and, again with appropriate caveats, in mouse knockout models.

Footnotes

  • Conflict-of-interest disclosure: The authors declare no competing financial interests. ■

REFERENCES