Search

Next generation sequencing (NGS) technologies have revolutionized the genomics field and are becoming more commonplace for identification of human infectious diseases. However, due to the low abundance of viral nucleic acids (NAs) in relation to host, viral identification using direct NGS technologies often lacks sufficient sensitivity. Here, we describe an approach based on two complementary enrichment strategies that significantly improves the sensitivity of NGS-based virus identification. To start, we developed two sets of DNA probes to enrich virus NAs associated with respiratory diseases. The first set of probes spans the genomes, allowing for identification of known viruses and full genome sequencing, while the second set targets regions conserved among viral families or genera, providing the ability to detect both known and potentially novel members of those virus groups. Efficiency of enrichment was assessed by NGS testing reference virus and clinical samples with known infection. We show significant improvement in viral identification using enriched NGS compared to unenriched NGS. Without enrichment, we observed an average of 0.3% targeted viral reads per sample. However, after enrichment, 50%–99% of the reads per sample were the targeted viral reads for both the reference isolates and clinical specimens using both probe sets. Importantly, dramatic improvements on genome coverage were also observed following virus-specific probe enrichment. The methods described here provide improved sensitivity for virus identification by NGS, allowing for a more comprehensive analysis of disease etiology.

Both diffusible factors acting in trans and chromatin components acting in cis are implicated in gene regulation, but the extent to which either process causally determines a cell's transcriptional identity is unclear. We recently used cell fusion to define a class of silent genes termed “cis-silenced” (or “occluded”) genes, which remain silent even in the presence of trans-acting transcriptional activators. We further showed that occlusion of lineage-inappropriate genes plays a critical role in maintaining the transcriptional identities of somatic cells. Here, we present, for the first time, a comprehensive map of occluded genes in somatic cells. Specifically, we mapped occluded genes in mouse fibroblasts via fusion to a dozen different rat cell types followed by whole-transcriptome profiling. We found that occluded genes are highly prevalent and stable in somatic cells, representing a sizeable fraction of silent genes. Occluded genes are also highly enriched for important developmental regulators of alternative lineages, consistent with the role of occlusion in safeguarding cell identities. Alongside this map, we also present whole-genome maps of DNA methylation and eight other chromatin marks. These maps uncover a complex relationship between chromatin state and occlusion. Furthermore, we found that DNA methylation functions as the memory of occlusion in a subset of occluded genes, while histone deacetylation contributes to the implementation but not memory of occlusion. Our data suggest that the identities of individual cell types are defined largely by the occlusion status of their genomes. The comprehensive reference maps reported here provide the foundation for future studies aimed at understanding the role of occlusion in development and disease.

The epigenome is a complex assortment of proteins and chemical modifications that are associated with DNA, control its transcription (Brink 1960; Bernstein et al. 2007), and mediate stable phenotypic states as exemplified by cell differentiation. The genome and the epigenome are associated in the chromosomes and are inherited together, but the degree to which the epigenome is encoded by the genome is not known. Furthermore, it is not clear to what extent the epigenome is maintained in the germline and transmitted between generations (Feng et al. 2010). It might be reset in each generation using genetically encoded information, but persistence of epigenetic states in the germline creates the potential for semi-independent inheritance of epigenetic information (Rakyan and Beck 2006; Richards 2006). Finally, it is not clear if epigenetic states that are present in the germline influence somatic epigenotypes, or conversely, if somatic epigenetic states are generated during cell differentiation using genetically encoded information. We have explored the use of comparative epigenomic analysis (“phyloepigenomics”) to obtain insights into changes in the epigenome in human evolution. Epigenetic differences provide a means to modulate the regulatory activity of noncoding regions. Functionally significant changes may be more readily identified in the epigenome than in the genome: Sequence change is not always associated with functional change (Boffelli et al. 2004), but the epigenome mediates genome function by controlling transcription (Brink 1960; Bernstein et al. 2007), and so changes in it are more likely to reflect functional changes. Epigenomic comparison might thus complement the evidence of potentially adaptive genomic changes identified with multiple sequence-based approaches (Pollard et al. 2006; Prabhakar et al. 2006; Grossman et al. 2010; Yi et al. 2010). This study focuses on one component of the epigenome, cytosine methylation, a covalent modification of DNA that acts as a focal point in mechanisms that suppress transcription initiation (Klose and Bird 2006). We have compared the methylomes of human and chimpanzee in a homogeneous somatic cell type, the neutrophil, using the orangutan as an outgroup. Our comparison uses a “methyltyping” assay based on digestion with the methylation-sensitive restriction enzyme HpaII and deep sequencing (MethylSeq) (Ball et al. 2009; Brunner et al. 2009). MethylSeq data was analyzed with MetMap (Singer et al. 2010), a computational method that we developed to correct bias in MethylSeq data and infer the true methylation status of all HpaII sites within the scope of the experiment. The combination of MethylSeq and MetMap obtains very deep sequence data focused on regions of the genome that are not methylated and can reliably detect methylation changes in these regions (Singer et al. 2010). Compared to methods that use methylation-independent restriction enzymes, MethylSeq retrieves information spanning more of the genome because of a more favorable profile of fragment sizes (Singer et al. 2010). We find that, while the methylomes of human and chimp are similar, a set of ∼1500 stable differences in CpG island-like regions distinguishes human from chimp; these differences identify regions that may have diverged in gene regulatory function. The methylation states can be used to build a tree that recapitulates the phylogenetic relationship of the three species. Analysis of CG substitution patterns in CpG island-like regions that have conserved their methylated state in human, chimp, and orang indicates that methylation in the neutrophil reflects germline methylation. Our findings demonstrate that methyltyping can identify differences that distinguish human from chimp and that these differences exist in the germline. This raises the question of whether a germline epigenome is transmitted along with the genome, and if so, whether it is completely determined by genome sequence.

MicroRNAs (miRNAs) are small regulatory RNAs that derive from distinctive hairpin transcripts. To learn more about the miRNAs of mammals, we sequenced 60 million small RNAs from mouse brain, ovary, testes, embryonic stem cells, three embryonic stages, and whole newborns. Analysis of these sequences confirmed 398 annotated miRNA genes and identified 108 novel miRNA genes. More than 150 previously annotated miRNAs and hundreds of candidates failed to yield sequenced RNAs with miRNA-like features. Ectopically expressing these previously proposed miRNA hairpins also did not yield small RNAs, whereas ectopically expressing the confirmed and newly identified hairpins usually did yield small RNAs with the classical miRNA features, including dependence on the Drosha endonuclease for processing. These experiments, which suggest that previous estimates of conserved mammalian miRNAs were inflated, provide a substantially revised list of confidently identified murine miRNAs from which to infer the general features of mammalian miRNAs. Our analyses also revealed new aspects of miRNA biogenesis and modification, including tissue-specific strand preferences, sequential Dicer cleavage of a metazoan precursor miRNA (pre-miRNA), consequential 5′ heterogeneity, newly identified instances of miRNA editing, and evidence for widespread pre-miRNA uridylation reminiscent of miRNA regulation by Lin28.