Next-Generation Sequencing Reveal Proviral Genome and Transcriptome in Adult T-Cell Leukemia/Lymphoma

Background:

Adult T-cell leukemia/lymphoma (ATL) is a peripheral T-cell neoplasm caused by human T-cell leukemia virus type-1 (HTLV-1) retrovirus infection. As for its pathogenesis, viral products, such as Tax and HBZ, play indispensable roles and their oncogenic mechanisms have been extensively studied. Recently, we have performed an integrated genetic study of a large number of ATL cases and revealed the entire landscape of somatic mutations, copy number alterations, and gene fusions in ATL. However, the detailed analysis of HLTV-1 integration using next-generation sequencing has not been performed so far. In this study, combining whole-genome and RNA sequencing data, we delineated the effect of HTLV-I integration on viral and cellular transcription.

Patients and Methods:

We performed WGS and RNA-seq for 48 and 57 ATL cases, respectively. All the analyses of the sequencing data were performed using our in-house pipelines. We analyzed HTLV-1 proviral genomic structure and the effect of HTLV-1 integration on viral and cellular transcription.

Results:

A cardinal feature of ATL genome is HTLV-1 integration, which was precisely located in all the cases analyzed by WGS. Multiple proviral integration sites were detected in 12 cases (total, 62 HTLV-1 integrations sites). The provirus integration was clonal in the architecture inferred from somatic mutations, and apparently randomly integrated into the host genome as previously reported. Within the HTLV-1 genome, frequent 5' proviral segment (gag/pol/env loci) deletions and/or sense gene (gag/pol/env/tax/rex/p13/p30) mutations were observed, which seem to cause defective viral replication/production, whereas HBZ gene was maintained in all the cases. RNA-seq revealed that HTLV-1 integration in ATL cells was associated with aberrant transcription. In general, viral transcripts were predominantly derived from the antisense strand, whereas sense transcription was largely suppressed, leading to global silencing of the sense genes. Especially, in contrast to the ubiquitous HBZ expression (antisense strand), tax expression (sense strand) was almost completely lost in all but one case, which exceptionally exhibited high expression of both tax and HBZ. Strikingly, in most cases, the antisense transcripts were not terminated in 5'-long terminal repeat (LTR), but read through into the juxtaposed cellular genome, extending into up to 50 kb downstream therefrom (read-through transcript). Moreover, in 11 sites of intragenic proviral integration, aberrantly spliced fusion transcripts were observed between LTR and the affected gene, and more commonly associated with antisense (n = 9) than sense (n = 2) integration, accompanied by upregulated cellular gene expression. In other cases (n = 3), fusion transcripts were also generated between HBZ and an exon of highly expressed cellular gene adjacent to the integration site. These results indicate the potential significance of antisense transcription and aberrant fusion transcripts with host genome sequences during ATL development. Although the precise role of these novel aberrant antisense transcripts remains unknown, antisense transcripts containing the LTR region has been implicated in NF-κB activation, which is a hallmark of ATL pathogenesis.

Conclusion:

In summary, combining WGS and RNA-seq data, we demonstrated the global silencing of sense-oriented viral transcripts (including Tax) and the predominance of aberrant antisense-directed transcription, which often involved cellular gene expression, including aberrant fusion transcripts between host and viral genomes (read-through and aberrantly spliced fusion transcripts). These results suggest that antisense transcription and abnormal virus-host fusion transcripts play pivotal roles in the pathogenesis of ATL.