Characterization of novel Bovine Leukemia Virus (BLV) antisense transcripts by deep sequencing reveals constitutive expression in tumors and transcriptional interaction with viral microRNAs
© Durkin et al. 2016
Received: 25 March 2016
Accepted: 28 April 2016
Published: 3 May 2016
Bovine Leukemia Virus (BLV) is a deltaretrovirus closely related to the Human T cell leukemia virus-1 (HTLV-1). Cattle are the natural host of BLV where it integrates into B-cells, producing a lifelong infection. Most infected animals remain asymptomatic but following a protracted latency period about 5 % develop an aggressive leukemia/lymphoma, mirroring the disease trajectory of HTLV-1. The mechanisms by which these viruses provoke cellular transformation remain opaque. In both viruses little or no transcription is observed from the 5′LTR in tumors, however the proviruses are not transcriptionally silent. In the case of BLV a cluster of RNA polymerase III transcribed microRNAs are highly expressed, while the HTLV-1 antisense transcript HBZ is consistently found in all tumors examined.
Here, using RNA-seq, we demonstrate that the BLV provirus also constitutively expresses antisense transcripts in all leukemic and asymptomatic samples examined. The first transcript (AS1) can be alternately polyadenylated, generating a transcript of ~600 bp (AS1-S) and a less abundant transcript of ~2200 bp (AS1-L). Alternative splicing creates a second transcript of ~400 bp (AS2). The coding potential of AS1-S/L is ambiguous, with a small open reading frame of 264 bp, however the transcripts are primarily retained in the nucleus, hinting at a lncRNA-like role. The AS1-L transcript overlaps the BLV microRNAs and using high throughput sequencing of RNA-ligase-mediated (RLM) 5′RACE, we show that the RNA-induced silencing complex (RISC) cleaves AS1-L. Furthermore, experiments using altered BLV proviruses with the microRNAs either deleted or inverted point to additional transcriptional interference between the two viral RNA species.
The identification of novel viral antisense transcripts shows the BLV provirus to be far from silent in tumors. Furthermore, the consistent expression of these transcripts in both leukemic and nonmalignant clones points to a vital role in the life cycle of the virus and its tumorigenic potential. Additionally, the cleavage of the AS1-L transcript by the BLV encoded microRNAs and the transcriptional interference between the two viral RNA species suggest a shared role in the regulation of BLV.
The deltaretrovirus Bovine Leukemia Virus (BLV), like its close relative Human T-cell leukemia virus (HTLV-1), produces a chronic infection in its natural host (cattle and water buffalo), with most infected individuals remaining asymptomatic and about 5 % going on to develop leukemia/lymphoma . In humans the time between infection and the onset of Adult T cell leukemia/lymphoma (ATL) generally spans several decades , while in cattle several years separate infection from progression to B-cell leukemia/lymphoma . BLV infections are common in cattle throughout the world (with the exception of most European countries, the result of eradication programs), imposing significant economic costs on the livestock industry . While not a natural host, it is also possible to experimentally infect sheep with BLV and in contrast to the situation in cattle, all infected sheep generally develop B cell leukemia/lymphoma, about 20 months post infection .
BLV infects B-cells (in both cattle and sheep) and following an early and transient phase of horizontal replicative dissemination, primarily proliferates via polyclonal expansion, producing many long lived clones that can be tracked via their proviral integration site . In a subset of infected individuals, for unknown reasons, one of these clones eventually begins to expand, producing an aggressive neoplasm that accumulates in the blood (B-cell leukemia) and/or organs (B-cell lymphoma). In both BLV and HTLV-1 the Tax protein has been seen as the principal actor in oncogenesis, especially as it is capable of driving cellular transformation . However, the lack of Tax expression in the majority of ATLs  and BLV induced tumors , points to a more diverse cast of actors beyond Tax. In the case of HTLV-1 it has been evident for the last decade that the antisense product HTLV-1 basic leucine zipper factor (HBZ) plays a central role in the process of oncogenesis as it is found in 100 % of ATL cases examined [7, 9]. A large number of roles have been attributed to HBZ, with its inhibition of Tax mediated viral transcription postulated as a vital part of the viruses immune evasion strategy [7, 10]. Intriguingly, in addition to the protein encoded by HBZ, the HBZ mRNA has been implicated in supporting proliferation of ATL cells [11, 12]. In the case of BLV, no equivalent antisense transcript has been described, although it has recently been reported by us and others that the provirus encodes a cluster of non-canonical RNA polymerase III transcribed microRNAs on the positive strand. These viral microRNAs represent ~40 % of the microRNA pool in transformed B cells [13, 14]. Similar to HBZ in HTLV-1, the viral miRNAs are expressed in all BLV induced tumors examined to date, pointing to a vital role in the life cycle of the virus and cellular transformation .
The close relationship between BLV and HTLV-1 in genome organization and pathogenesis makes BLV an attractive model for investigating deltaretrovirus induced cancer. To further our understanding of the transcriptional landscape of BLV infected cells and to build on our previous work describing small noncoding RNAs we carried out high throughput RNA sequencing (RNA-seq) of total RNA from ovine and bovine BLV infected cells. Surprisingly, we observed a large number of reads mapping back to the BLV genome and found that these reads were the products of previously unidentified BLV antisense transcripts originating in the BLV 3′LTR. We present evidence that, like the case of HBZ, BLV antisense transcription appears to be a consistent feature of BLV infections. We also find that one of the transcripts is cleaved by the BLV microRNAs and discuss the possible role these transcripts play in the life cycle of BLV.
Identification of two antisense transcripts encoded by BLV via deep sequencing
The number of viral antisense reads counted per million reads (CPM) from RNA sequencing of YR2/L267 and ovine/bovine tumors was broadly similar, ranging from 0.45 to 124.5 (mean = 15, median = 4.4). The CPM of viral sense reads ranged from 0 to 19.3 (mean = 1.7, median = 0.5) (Additional file 1: Table S1). The YR2LTaxSN and L267LTaxSN cell lines , which constitutively express 5′LTR dependent viral transcripts and proteins, in addition to the RNA pol III dependent microRNAs had antisense CPM of 29 and 4 respectively (Additional file 1: Table S1), suggesting that transcription from the 5′ and 3′LTRs of BLV, as well as from the RNA pol III promoters, are not mutually exclusive. When compared to other mRNAs the antisense transcripts have relatively modest expression levels, with ~73 and ~76 % of host genes having higher expression levels than AS1 and AS2 respectively. We also carried out RNA-seq on PBMCs isolated from 6 asymptomatic BLV infected sheep at 17 months post-inoculation. The proviral load in these animals ranged from 5.3 to 40.6 % (mean = 14.4 %; median = 9.3 %). High throughput sequencing (HTS) mapping of proviral integration sites showed that in five of these animals the largest clone (defined by percentage of reads observed at each insertion point) did not exceed 5 %. Antisense reads were detected in all 6 samples and at substantially higher numbers than sense reads. Additionally, the number of antisense reads mirrored the proviral load (Additional file 1: Table S1). We also observed consistent production of BLV microRNAs in these early stage samples (Additional file 1: Table S1). Collectively, these data support the conclusion that BLV antisense transcription, like BLV microRNA production, is a feature of both the asymptomatic and leukemic stages of BLV infection.
Identification of 5′/3′ ends
To determine the 5′ and 3′ ends of both antisense transcripts rapid amplification of cDNA ends (RACE) was completed in combination with high throughput sequencing. For 5′RACE, total RNA from the cell line YR2 was used as template and mapping of the resultant reads showed that 42 % of AS1 and 46 % of AS2 transcripts started in U5 close to the R boundary at position 8650 (Fig. 1). A second site (8591) within the R region was also frequently utilized by both AS1 (27 %) and AS2 (21 %) (Fig. 1). Other less frequently used start sites are also listed (Additional file 1: Table S1).
In the case of AS2, the majority of the samples showed no evidence of a poly A tail, rather the AS2 transcript ended abruptly at position 2704. Closer examination of the sequences showed the transcript had also utilized canonical splice donor acceptor sites in the viral and host genomes to form a virus-host fusion transcript. In each tumor examined, the sequence forming the fusion with AS2 corresponded to the region of the host genome where the provirus was inserted (Fig. 2c). In the case of L267, in addition to BLV-host fusion transcripts, a small number of AS2 reads were observed with appetent poly A tails (Fig. 2c). No PAS was observed in the vicinity and the position of the poly A varied over a ~220 bp region with the majority slightly upstream of the splice site utilized in the formation of the BLV-host fusion transcript.
BLV 3′LTR promoter activity and regulatory sequences affecting antisense transcription
Having confirmed the promoter activity of the 3′LTR and estimated the approximate scope of the BLV antisense transcripts, we next examined them for coding potential using the Coding Potential Assessment Tool (CPAT) . The limited number of annotated bovine long non-coding RNAs (lncRNAs) prevented development of a robust bovine specific model to calculate coding probability. As a result, human and mouse CPAT models were utilized. In addition to the BLV antisense transcripts, a number of known BLV and HTLV-1 protein coding genes and 220 bovine lncRNAs were analyzed. The known protein coding genes (including the HTLV-1 antisense transcript HBZ) were all assigned a high coding probability, while in contrast, >90 % of the bovine lncRNAs were assigned a coding probability less than 0.25 in both models (Additional file 1: Table S2). AS2 coding probability was low (0.003 human model, 0.022 mouse model), allowing us to classify it as an lncRNA. AS1-S/L were assigned an ambiguous coding probability of ~0.46 in the human and ~0.39 in the mouse model, with a potential open reading frame (ORF) of 264 bp (shared by both AS1-S and AS1-L). The small putative peptide sequence did not show any significant matches when used to search the non-redundant protein sequence (nr) database with BLASTP and the protein families database .
We next examined patterns of nucleotide variation in the region containing the putative peptide. Using our RNA-seq data, we extracted the sequence of the putative AS1-S/L ORF from each of our bovine and ovine tumors and supplemented these with six publicly available whole genome BLV sequences from NCBI. AS1-S/L ORF was characterized by a nucleotide diversity (average difference per nucleotide site for all pairwise comparisons of available sequences) of 0.0149. In comparison Tax, known to be essential, had a nucleotide diversity of 0.0106. Throughout the putative ORF, the majority of the nucleotide diversity of AS1-S/L, like that of Tax, was concentrated in the less constrained 3rd codon position of the negative strand (0.034 and 0.0247 for AS1-S/L and Tax respectively) (Additional file 1: Fig. S3; Additional file 1: Table S3). It should be noted though, that for just over 2/3 of the sequence, the potential AS1-S/L ORF overlaps with parts of the R3 and G4 ORFs on the sense strand. Additionally, the 3rd codon position of G4 on the sense corresponds to the 3rd codon position of the potential AS1-S/L ORF on the negative strand. Nevertheless, this pattern of nucleotide diversity was not confined to the overlapping sequences and extends into the non overlapping region (no overlap 0.0236 vs no overlap 3rd 0.0521; Additional file 1: Fig. S3; Additional file 1: Table S3). Consequently, the evidence for coding potential in AS1-S/L is suggestive, but not definitive, making it best to classify AS1-S/L as transcripts of unknown coding potential (TUCP) .
Sub-cellular localization of BLV antisense transcripts
BLV microRNA guided cleavage of AS1-L
Altering the microRNA region in the BLV provirus reveals transcriptional interactions between the antisense transcripts and viral microRNAs
The increased fraction of AS1-L transcripts could be explained in two ways. Firstly, the RNA pol III positioned at the BLV microRNAs may inhibit RNA pol II progression allowing the PAS adjacent to BLV miR-B5-3p microRNA to mediate polyadenylation of the antisense transcripts, thereby forming AS1-S. By removing these elements, AS1-L levels increase. Secondly, by removing the microRNAs we prevent the cleavage of AS1-L in the cytoplasm, increasing the apparent number of AS1-L reads and causing their accumulation in the cytoplasm. In an attempt to distinguish between these two possibilities, we fractionated transfected cells in into nuclear and cytoplasm enriched RNA pools. Real-time PCR with primers spanning the splice sites for AS1-S/L, AS2 and BLV sense transcripts did not show differences in the sub-cellular distribution of transcripts for the three proviruses (Fig. 6b). As a result, it does not appear that the increased fraction of AS1-L transcripts observed is a consequence of eliminating BLV microRNA mediated slicing of AS1-L, rather it suggests the first scenario involving RNA pol II and pol III interference.
Knock down of AS1 with locked nucleic acids (LNAs)
A mix of three LNAs, targeting different parts of AS1 were used to treat both the YR2 and L267cell lines, with a biological replicate for each cell line, at concentrations of 5 and 10 μM. In addition, the cell lines were treated with 10 μM of control LNAs and with a mock treatment of H2O. Dunn’s multiple comparisons test was then used to examine differences between the individual samples and found that at the 10 μM concentration, the drop (>50 %) between the control and AS specific LNAs was significant (p value = 0.0347). In the case of AS2 a significant difference was observed between the treatments (Kruskal–Wallis test, p value = 0.0208), however Dunn’s multiple comparisons test failed to show any difference between individual samples (Additional file 1: Fig. S4A). The real time assay for the BLV sense transcript Tax failed to show amplification or was indistinguishable from the background in all the YR2 and L267 treatments (Tax expressing cells showed robust amplification with an average Ct of 21.5). As a result, it appears that knock down of AS1 does not cause a detectable reactivation of transcription from the 5′LTR. Finally, there was no apparent difference in the viable cell counts in the treated and control cells at the end of the experiment (Additional file 1: Fig. S4B).
It has been known for over a decade that BLV’s close relative HTLV-1 encodes an antisense transcript  and like HTLV-1, the BLV 3′LTR remains unmethylated in tumors , begging the question of a BLV antisense transcript. Prior to the adoption of RNA-seq, identifying these transcripts required a certain amount of luck in the placement of primers/probes for Rt-PCR/Northern blotting. However, as can be seen from Fig. 1 the use of RNA-seq, especially in combination with stranded libraries, made the presence of antisense reads in BLV tumors obvious. The consistent pattern of antisense transcription across all the BLV samples sequenced to date strongly suggests that these transcripts fulfill an essential function in the biology of BLV and leukemogenesis. Furthermore, in the majority of samples little or no evidence of sense transcription was observed, mirroring results recently obtained in HTLV-1 . While we saw evidence of antisense without sense transcription, we did not see the inverse. RNA-seq of YR2LTaxSN/L267LTaxSN and the 3′RACE library from the FLK cell line showed that antisense expression levels and splicing patterns were comparable to that seen in ovine and bovine tumors (Additional file 1: Table S1; Additional file 1: Fig. S2), despite robust sense transcription. A related question was the temporal pattern of expression of the BLV antisense transcripts. By combining stranded RNA-seq, high-throughput mapping of proviral integration sites, 3′RACE and small RNA-seq of PBMCs from asymptomatic BLV infected sheep (Additional file 1: Table S1) we could be confident that antisense/microRNA transcription is a feature of BLV infected B cell clones present at the polyclonal stage and not confined to the tumor clone. Given the tractability of the BLV sheep model, in comparison to the difficulty of studying primary HTLV-1 infections  it will be interesting to extend this analysis back to the earliest stages of infection to examine the dynamics of sense, antisense and microRNA expression from BLV. The Tax protein has been seen as the primary actor in both HTLV-1 and BLV, however over the course of infection it is now apparent that its expression is the exception rather than the rule, while antisense transcription appears the rule. Probing the initial infection in experimentally infected sheep may elucidate if and how Tax sets the stage for tumor development and if antisense transcription displays more dynamism in comparison to its consistent expression in both the established nonmalignant and expanded tumor clones.
The presence of BLV antisense transcription implies that the BLV LTR (like that of HTLV-1) is capable of driving transcription in both directions. Extensive work over the years has revealed a rich set of regulatory motifs bound by cellular transcription factors and Tax [18, 24], regulating the BLV 5′LTR. As expected given the evidence of antisense transcription, we found that the BLV LTR was able to drive transcription in the orientation of the 3′LTR and with a significantly higher activity compared to the weak basal activity typical of the 5′LTR. Our data are consistent with observations reported by Van Driessche et al.  demonstrating RNA Pol II recruitment to the 3′LTR-host boundary and its association with positive epigenetic marks. Removal of the majority of the regulatory motifs in U3, taken with the results from Tax co-transfection suggest that the motifs in U3 do not have a role in driving, but rather obstruct BLV antisense transcription. In contrast, the motifs in the R and U5 regions appear to have a dual role in driving expression from both the 5′ and 3′LTRs. Taken in conjunction with our sequencing results it appears that consistent antisense expression from the 3′LTR is a feature of the BLV provirus in both nonmalignant and tumor clones. In contrast, the 5′LTR is tightly regulated and does not show evidence of consistent expression. As a consequence, it is apparent that Tax is not required for maintaining malignancy and the absence of this immuno-dominant viral protein in transformed cells plays a vital role in BLVs immune evasion strategy. In contrast, the permanent production of both antisense transcripts and microRNAs, besides pointing to an essential function in tumorigenesis, indicates that these transcripts are incapable of eliciting an effective host immune response.
Having established the presence of BLV antisense transcripts we attempted to determine their approximate size with a combination of short and long reads. It was somewhat surprising to see that the majority of AS1 is only ~600 bp in length, a much shorter transcript than HBZ (>2 kb) in HTLV-1 , while AS2 exon 2 was small (~125 bp) and generally forms fusion transcripts with the host genome. Our 3′RACE and long read sequencing showed that the longer version of the AS1 transcript (AS1-L > 2 kb) could extend to a poly A tail just beyond the canonical AAUAAA (PAS) at 5257–5262. However the balance between the use of this poly A tail and the formation of fusions with the host genome remains unknown and will likely depend on the location of the provirus in the genome. The use of long read technologies such as the MinION will be particularly useful in answering this question, as full-length sequences can unambiguously determine the structure of fusion transcripts.
RNA-seq data indicated that the majority of the AS1 reads take the form of the AS1-S variant, which terminates next to the end of miR-B5-3p and is considerably smaller than HBZ. In the case of the AS1-L, there is an overlap with the BLV microRNAs leading to the cleavage of AS1-L via action of the RISC complex . Our modified 5′RACE experiments confirmed this and showed that RISC mediated cleavage predominated in the cytoplasm (the fraction also enriched for BLV microRNAs in YR2). The limited number of microRNAs involved in cleavage is intriguing as previous work indicates that all the BLV microRNAs tested associate with the RISC complex . This may point to secondary structures in BLV AS1-L that inhibit cutting at certain positions, however the significance of this observation remains to be determined and more samples are required to see if this pattern is consistent. Another question requiring future work is the fraction of reads that are subjected to cleavage. The majority of the AS1 transcripts (AS1-S) do not contain the region complementary to the BLV microRNAs. Additionally, a majority of the antisense transcripts are retained in the nucleus while the BLV microRNAs are predominantly found in the cytoplasm. How the nuclear versus cytoplasmic localization of AS1-L and the balance between AS1-S versus AS1-L evolves over the entire life cycle of the virus, in different cell types and in times of cellular stress remains an open question. Exploring these dynamics could be important to understand the significance of microRNA mediated cleavage of AS1-L.
Other microRNA producing viruses such as SV40  and Epstein–Barr virus  possess viral microRNAs that cleave products from the complementary strand. It is thought that this microRNA mediated regulation manages the transition from latency (low stress) to the lytic stage (high stress). With stress mediated suppression of microRNA production, cleavage of the antisense product is relaxed, allowing it to signal activation of the latent virus . It is conceivable that a similar mechanism is involved in regulating the relative abundance of the BLV AS1-S/L transcripts as RNA pol III transcription is known to be regulated by cell stress and cell cycle stage . In addition to reducing the amount of microRNAs available for cleaving the AS1-L, changes in RNA pol III promoter occupancy and transcription could increase the fraction of AS1-L produced by removing RNA pol III complexes on the sense strand potentially impeding RNA Pol II and encouraging production of the larger transcript (AS1-L). It is apparent in Fig. 1 and Additional file 1: Fig. S2 that antisense coverage diminishes as we enter the region containing the BLV microRNAs. RACE experiments with RNA from HeLa transfections showed that deleting/inverting the microRNAs/PAS in the BLV provirus results in a higher fraction of AS1-L (Fig. 6a) without altering the fraction of AS transcripts localized in the nucleus or cytoplasm. This suggests the removal of an insulator that prevents progression of RNA pol II rather than reduced microRNA mediated cleavage of AS1-L. Our data are in agreement with recent Chip-seq work by Van Driessche et al.  showing evidence of a high RNA pol III occupancy peak in the BLV microRNA region with a large RNA pol II occupancy peak just downstream of this position, consistent with RNA pol III–RNA pol II collision.
The significant size difference between AS1-S and AS1-L is intriguing as both contain the same potential ORF. Assuming that the ORF does produce a functional protein, its potential role remains opaque as database searches did not reveal any significant homology with known protein domains. In HTLV-1, a number of roles have been assigned to the HBZ protein, including the suppression of HTLV-1 5′LTR transcription and repression of the host immune response . In HTLV-2 the antisense protein APH-2 also plays a role in suppressing viral transcription . However, the potential protein in BLV AS1-S/L (87 amino acids) is considerably smaller than that reported for HTLV-1 (206 amino acids spliced/209 amino acids unspliced) and HTLV-2 (183 amino acids) which argues against an exactly analogous protein (although the final effect may be equivalent). In addition to the roles ascribed to the HBZ protein, it has been reported that the HBZ mRNA is capable of supporting proliferation  and is primarily retained in the nucleus , suggesting an lncRNA like function. Only a small portion of AS1-L is occupied by the potential ORF and the AS1-L\S transcript is primarily retained in the nucleus, again hinting at a potential lncRNA like role. Recently, there has been a rapid expansion of our understanding of the numerous roles lncRNAs play in the cell . Many of these lncRNAs also show a predominately nuclear localization and are involved in transcriptional regulation . Additionally, many transcripts previously described as lncRNAs possess small ORFs (<100 amino acids) that do not resemble those found in well characterized mRNAs . The situation for the BLV antisense transcripts appears similar. Thus, it is reasonable to propose that one role of the BLV AS1-S/L may be silencing of the 5′LTR. Future work using in vitro infection with altered proviruses will be needed to firstly establish the precise role AS1-S/L plays in the life cycle of BLV and secondarily if these different roles are carried out via the RNA or potential protein product. With this information in hand it will be interesting to compare and contrast the roles AS1-S/L and HBZ transcript play and given the tractability of the BLV model, make targeted changes to the transcript and examine the effect on tumor development.
Looking beyond AS1-S/L there remains the enigma of the AS2 transcript, where at the moment the potential functions are even more obscure than for AS1-S/L. No comparable antisense transcript has been reported for HTLV-1 and no potential ORF is present in the transcript. Furthermore, the expression levels of AS2 are lower than that observed for AS1 (Fig. 1; Additional file 1: Fig. S2). Additionally, the second exon of AS2 is small (125 bp) creating a transcript containing just over ~400 bp of sequence from the BLV genome. However, the consistent splicing of AS2 with the host genome results in the production of a fusion transcript containing ~400 bp from BLV and additional sequences from the host genome. As a result the AS2 transcript size and composition will differ from cell to cell depending on provirus location in the host genome. It has recently been reported for HTLV-1 that antisense transcripts can also form fusion transcripts with the host genome . Additionally we also see extensive splicing between the common AS exon 1 and the host genome (using canonical splice acceptor sites) and find evidence that these fusions may play a role in leukemogenesis. These observations are outlined in detail in Rosewick et al. submitted and point to an important role for antisense dependent viral/host transcripts in the biology of deltaretroviruses.
A striking feature of many HTLV-1 and BLV proviruses is the large deletions frequently observed in tumors, that can remove over half the proviral DNA, often including the entre 5′LTR and many vital ORFs, while always retaining the 3′LTR [9, 13]. These observations strongly indicate that HBZ plays an important role in leukemogenesis and a similar pattern of deletions in BLV may point to an analogous role for the BLV AS1-S/L. The case of the BLV provirus T1345 is intriguing for this reason as the large deletion gives us a clue about which BLV antisense transcripts are important in the tumor. In this provirus a ~4.3 kb deletion removes many essential viral genes but retains the BLV microRNAs  and continues to display robust antisense expression (Additional file 1: Fig. S5). This deletion also removes the second exon of AS2 and the last ~100 bp of AS1-L indicating that these transcripts/regions are dispensable in the tumor.
Our attempt to investigate the function of the AS1-S/L transcript via knock down was hampered as the YR2 and L267 B cell lines are refractory to transfection and require high levels of LNAs for unassisted uptake, making optimization and long-time courses cost prohibitive. Consequently, we carried out a limited number of assays and designed our experiment to examine if knock down of AS1-S/L could lead to the re-activation of viral transcription from the 5′LTR. It has been known for some time that the lack of 5′LTR based expression is imposed via histone repressive marks and DNA methylation at the 5′LTR [15, 16] and we hypothesized that the AS1-S/L has a role in maintaining this state. While we did achieve a significant knock down of AS1-S/L we did not see any evidence of transcription from the 5′LTR and based on final cell counts there was no apparent effect on cell viability (Additional file 1: Fig. S4). More prolonged or complete knock down may be required to produce an effect, however given the large quantities of LNAs required for unassisted uptake this was not practical. Previous work by Colin et al.  showed that following experimentally induced chromatin remodeling in the YR2 cell line there is a rapid return to a closed chromatin conformation. The AS-1S/L transcripts may play a role in this process and in order to observe any effect on 5′LTR based expression, AS1-S/L knock down must be coupled with prior relaxing of chromatin. It will be interesting to examine if combining LNA knock down of AS1 and treatment to open chromatin changes the patterns of epigenetic marks observed at the BLV 5′LTR.
The identification of BLV antisense transcripts taken in conjunction with the recent identification of the BLV microRNAs [13, 14] represent a major shift in our understanding of BLV pathogenesis. In contrast to the prevailing paradigm of a silent provirus, our work reveals that the BLV provirus is a prodigious producer of viral microRNAs and constitutively expresses antisense transcripts in all tumors examined. The consistent expression of these transcripts in both leukemic and nonmalignant clones points to a vital role in the life cycle of the virus and its tumorigenic potential. Additionally, the cleavage of the AS1-L transcript by the BLV encoded microRNAs and the transcriptional interference between the two viral RNA species suggest a shared role in the regulation of BLV. While additional work is required to elucidate the precise roles of the BLV antisense transcripts, it is hoped that the greater tractability of the BLV model will provide key insights into HTLV-1 induced leukemia and provide a test bed for exploring potential treatment options.
See Additional file 1: Text S1 for full details of materials and methods used.
Ovine/bovine samples and cell lines
Primary ovine and bovine tumors were part of an existing BLV induced tumor collection maintained at −80 °C and described previously . Asymptomatic sheep samples came from animals infected with the molecular clone pBLV344 following experimental procedures approved by the University of Saskatchewan Animal Care Committee, following Canadian Council on Animal Care Guidelines (Protocol #19940212). The cell lines YR2 and L267 were originally derived from ovine B-cell tumors, do not express viral mRNAs originating in the 5′LTR , L267LtaxSN and YR2LtaxSN were produced via transduction with the pLTaxSn retroviral vector expressing Tax .
The Illumina TruSeq Total RNA stranded kit was used to prepare ribosomal RNA depleted libraries followed by sequencing on an Illumina HiSeq 2000 (2 × 100 bp paired-end reads), producing approximately 60 million raw paired-end reads per library. For small RNAs the Illumina TruSeq Small RNA Library Preparation Kit was used followed by sequenced on an Illumina NextSeq 500 (75 bp single reads). Reads were mapped to appropriate bovine UMD3.1 or ovine OAR3.1 genomes as well as BLV YR2 provirus genome (NCBI Accession: KT122858) using STAR (v2.3.1.u) and BWA (See Additional file 1: Text S1).
End point PCR
SuperScript III Reverse Transcriptase (Life technologies) was used to produce cDNA. PCR primers used are listed in Additional file 1: Table S4.
High-throughput sequencing of BLV integration sites
The Bioruptor Pico (Diagenode) was used to shear 5ug of DNA, followed by indexing with Nextera XT indexes and sequencing on a Illumina MiSeq instrument. Reads were mapped to the host-provirus hybrid genome with BWA and numbers were determined for each proviral integration site using in-house R and Perl scripts.
Proviral load quantification
Proviral load was assessed with PrimeTime qPCR Assays (IDT) using diluted DNA from the YR2 cell line to produce a standard curve. Primers used are listed in Additional file 1: Table S4.
Identification of 5′ and 3′ ends with RACE
The GeneRacer Kit (Life Technologies) was used for 5′ and 3′RACE followed by high throughput sequencing of the resultant PCR products. Near full-length cDNA transcripts were also produced by PCR followed by either shearing with a Bioruptor Pico (Diagenode) and sequencing on a MiSeq instrument or sequencing of the full-length transcripts on a MinION instrument (Oxford Nanopore Technologies).
The LTR from BLV was derived from the pBLV344 plasmid , inserted into the pGL3 basic luciferase reporter plasmid. Constructs were transfected in triplicate with Lipofectamine 2000 (Life technologies) and processed using the Dual-Glo Luciferase Assay System (Promega). Statistical significance was assessed with Tukey’s multiple comparisons test, a p value <0.05 considered statistically significant.
Protein coding potential and nucleotide diversity
The Coding-Potential Assessment Tool (CPAT)  was used to assess the coding potential of the BLV antisense transcripts. Sequence conservation was examined using sequences extracted from the bovine and ovine tumors sequenced as well as six genomes from NCBI (http://www.ncbi.nlm.nih.gov/nuccore), using the same approach outlined in Rosewick et al. .
Sub-cellular location of BLV antisense transcripts
Cytoplasmic and nuclear fractions were produced from YR2 following the method outlined in Weil et al. . RNA sequencing was carried out as outlined above. Real time PCR based absolute quantification was carried out on a ABI Prism 7900HT Sequence Detector System (Applied Biosystems) using ABsolute Blue QPCR SYBR Green ROX Mix (Thermo Scientific) and the appropriate PCR product in a plasmid to produce a standard curve.
Identifying products of RISC mediated cleavage
The products of cleavage were identified with the protocol of Davis et al.  with the addition of high throughput sequencing of the resultant PCR products on a Illumina MiSeq. The precise position of cleavage was determined via mapping to the BLV provirus genome with BWA and a custom BASH script.
Deleting and inverting the BLV microRNAs
The altered proviruses used the wild type molecular clone pBLV344  as the starting material. In the deleted provirus 554 bp containing the BLV microRNAs was removed and replaced with the MluI restriction site. In the inverted provirus the BLV microRNAs were reinserted into this MluI restriction site in an inverted orientation. The constructs were transfected into HeLa cells with Lipofectamine 2000 (Life technologies) and real time PCR and 3′RACE carried out as described above.
Antisense knock down with locked nucleic acid antisense oligos (LNAs)
The antisense transcript was targeted with a mix of three locked nucleic acid antisense oligos (Exiqon) introduced via unassisted uptake  in duplicate for a half million YR2 and L267 cells. Final concentrations of 5 and 10 μM were tested, as well as the LNA longRNA GapmeR Negative control A (Exiqon), at a final concentration of 10 μM and a mock treatment of H2O. Real time PCR was carried out as outlined above and statistical significance assessed via a Kruskal–Wallis test and Dunn’s multiple comparisons test, p values of ≤0.05 were considered to be statistically significant.
KD performed the experiments, analyzed the data, and generated the text and the figures. NR performed bioinformatics analyses. MA contributed to HTS integration mapping, VH contributed to real-time RT-PCR and PVL quantifications. PG and NA collected and provided animal samples. KD, AV, MG and AB contributed to drafting of the manuscript, AV and MG supervised the study. All authors read and approved the final manuscript.
We thank the GIGA Genomics Platform and more particularly Wouter Coppieters, Benoit Hennuy and Latifa Karim for their help with high-throughput sequencing.
The authors declare that they have no competing interests.
This work was supported by the Fonds de la Recherche Scientifique—FNRS (Grant No. 3.4587.12), Télévie, les Amis de l’Institut Bordet, the International Brachet Stiftung (IBS), the Fondation Lambeau-Marteaux, and a Télévie Grant to VH. KD is Postdoctoral Researcher of the FNRS, NR and MA are Scientific Research Worker of Télévie and the FNRS respectively.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Gillet N, Florins A, Boxus M, Burteau C, Nigro A, Vandermeers F, Balon H, Bouzar A-B, Defoiche J, Burny A, Reichert M, Kettmann R, Willems L. Mechanisms of leukemogenesis induced by bovine leukemia virus: prospects for novel anti-retroviral therapies in human. Retrovirology. 2007;4:18.View ArticlePubMedPubMed CentralGoogle Scholar
- Ishitsuka K, Tamura K. Human T-cell leukaemia virus type I and adult T-cell leukaemia-lymphoma. Lancet Oncol. 2014;15:e517–26.View ArticlePubMedGoogle Scholar
- Burny A, Willems L, Callebaut I, Adam E, Cludts I, Dequiedt F, Droogmans L, Grimonpont C, Kerkhofs P, Mammerickx M. Bovine leukemia virus: biology and mode of transformation. In: Minson AC, Neil JC, McRae, MA, editors. Viruses and cancer. Cambridge: Cambridge University Press; 1994. p. 213–34.Google Scholar
- Rodriguez SM, Florins A, Gillet N, de Brogniez A, Sánchez-Alcaraz MT, Boxus M, Boulanger F, Gutiérrez G, Trono K, Alvarez I, Vagnoni L, Willems L. Preventive and therapeutic strategies for bovine leukemia virus: lessons for HTLV. Viruses. 2011;3:1210–48.View ArticlePubMedPubMed CentralGoogle Scholar
- Gillet NA, Gutiérrez G, Rodriguez SM, de Brogniez A, Renotte N, Alvarez I, Trono K, Willems L. Massive depletion of bovine leukemia virus proviral clones located in genomic transcriptionally active sites during primary infection. PLoS Pathog. 2013;9:e1003687.View ArticlePubMedPubMed CentralGoogle Scholar
- Grassmann R, Aboud M, Jeang K-T. Molecular mechanisms of cellular transformation by HTLV-1 Tax. Oncogene. 2005;24:5976–85.View ArticlePubMedGoogle Scholar
- Cook LB, Elemans M, Rowan AG, Asquith B. HTLV-1: persistence and pathogenesis. Virology. 2013;435:131–40.View ArticlePubMedGoogle Scholar
- Merimi M, Klener P, Szynal M, Cleuter Y, Kerkhofs P, Burny A, Martiat P, Van den Broeke A. Suppression of viral gene expression in bovine leukemia virus-associated B-cell malignancy: interplay of epigenetic modifications leading to chromatin with a repressive histone code. J Virol. 2007;81:5929–39.View ArticlePubMedPubMed CentralGoogle Scholar
- Kataoka K, Nagata Y, Kitanaka A, Shiraishi Y, Shimamura T, Yasunaga J-I, Totoki Y, Chiba K, Sato-Otsubo A, Nagae G, Ishii R, Muto S, Kotani S, Watatani Y, Takeda J, Sanada M, Tanaka H, Suzuki H, Sato Y, Shiozawa Y, Yoshizato T, Yoshida K, Makishima H, Iwanaga M, Ma G, Nosaka K, Hishizawa M, Itonaga H, Imaizumi Y, Munakata W, et al. Integrated molecular analysis of adult T cell leukemia/lymphoma. Nat Genet. 2015;47:1304–15.View ArticlePubMedGoogle Scholar
- Gaudray G, Gachon F, Basbous J, Biard-Piechaczyk M, Devaux C, Mesnard JM. The complementary strand of the human T-cell leukemia virus type 1 RNA genome encodes a bZIP transcription factor that down-regulates viral transcription. J Virol. 2002;76:12813–22.View ArticlePubMedPubMed CentralGoogle Scholar
- Satou Y, Yasunaga J-I, Yoshida M, Matsuoka M. HTLV-I basic leucine zipper factor gene mRNA supports proliferation of adult T cell leukemia cells. Proc Natl Acad Sci. 2006;103:720–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Mitobe Y, Yasunaga JI, Furuta R, Matsuoka M. HTLV-1 bZIP factor RNA and protein impart distinct functions on T-cell proliferation and survival. Cancer Res. 2015;75:1–11.Google Scholar
- Rosewick N, Momont M, Durkin K, Takeda H, Caiment F, Cleuter Y, Vernin C, Mortreux F, Wattel É, Burny A, Georges M, Van den Broeke A. Deep sequencing reveals abundant noncanonical retroviral microRNAs in B-cell leukemia/lymphoma. Proc Natl Acad Sci. 2013;110:2306–11.View ArticlePubMedPubMed CentralGoogle Scholar
- Kincaid RP, Burke JM, Sullivan CS. RNA virus microRNA that mimics a B-cell oncomiR. Proc Natl Acad Sci. 2012;109:3077–82.View ArticlePubMedPubMed CentralGoogle Scholar
- Colin L, Dekoninck A, Reichert M, Calao M, Merimi M, Van den Broeke A, Vierendeel V, Cleuter Y, Burny A, Rohr O, Van Lint C. Chromatin disruption in the promoter of bovine leukemia virus during transcriptional activation. Nucleic Acids Res. 2011;39:9559–73.View ArticlePubMedPubMed CentralGoogle Scholar
- Pierard V, Guiguen A, Colin L, Wijmeersch G, Vanhulle C, Van Driessche B, Dekoninck A, Blazkova J, Cardona C, Merimi M, Vierendeel V, Calomme C, Nguyên TL-A, Nuttinck M, Twizere J-C, Kettmann R, Portetelle D, Burny A, Hirsch I, Rohr O, Van Lint C. DNA cytosine methylation in the bovine leukemia virus promoter is associated with latency in a lymphoma-derived B-cell line: potential involvement of direct inhibition of cAMP-responsive element (CRE)-binding protein/CRE modulator/activation transcription factor binding. J Biol Chem. 2010;285:19434–49.View ArticlePubMedPubMed CentralGoogle Scholar
- Kiermer V, Van Lint C, Briclet D, Vanhulle C, Kettmann R, Verdin E, Burny A, Droogmans L. An interferon regulatory factor binding site in the U5 region of the bovine leukemia virus long terminal repeat stimulates tax-independent gene expression. J Virol. 1998;72:5526–34.PubMedPubMed CentralGoogle Scholar
- Calomme C, Nguyên TL-A, de Launoit Y, Kiermer V, Droogmans L, Burny A, Van Lint C. Upstream stimulatory factors binding to an E box motif in the R region of the bovine leukemia virus long terminal repeat stimulates viral gene expression. Journal of Biological Chemistry. 2002;277:8775–89.View ArticlePubMedGoogle Scholar
- Wang L, Park HJ, Dasari S, Wang S, Kocher JP, Li W. CPAT: Coding-Potential Assessment Tool using an alignment-free logistic regression model. Nucleic Acids Res. 2013;41:e74.View ArticlePubMedPubMed CentralGoogle Scholar
- Finn RD, Bateman A, Clements J, Coggill P, Eberhardt RY, Eddy SR, Heger A, Hetherington K, Holm L, Mistry J, Sonnhammer ELL, Tate J, Punta M. Pfam: the protein families database. Nucleic Acids Res. 2014;42(database issue):D222–30.Google Scholar
- Mattick JS, Rinn JL. Discovery and annotation of long noncoding RNAs. Nat Struct Mol Biol. 2015;22:5–7.View ArticlePubMedGoogle Scholar
- Davis E, Caiment F, Tordoir X, Cavaillé J, Ferguson-Smith A, Cockett N, Georges M, Charlier C. RNAi-Mediated Allelic trans-Interaction at the Imprinted Rtl1/Peg11 Locus. Curr Biol. 2005;15:743–9.View ArticlePubMedGoogle Scholar
- Cook LBM, Melamed A, Demontis MA, Laydon DJ, Fox JM, Tosswill JHC, Freitas D, Price AD, Medcalf JF, Martin F, Neuberger JM, Bangham CRM, Taylor GP. Rapid dissemination of human T-lymphotropic virus type 1 during primary infection in transplant recipients. Retrovirology. 2016;13:1–9.View ArticleGoogle Scholar
- Calomme C, Dekoninck A, Nizet S, Adam E, Nguyen TLA, Van den Broeke A, Willems L, Kettmann R, Burny A, Van Lint C. Overlapping CRE and E Box motifs in the enhancer sequences of the Bovine Leukemia Virus 5′ long terminal repeat are critical for basal and acetylation-dependent transcriptional activity of the viral promoter: implications for viral latency. J Virol. 2004;78:13848–64.View ArticlePubMedPubMed CentralGoogle Scholar
- Van Driessche B, Rodari A, Delacourt N, Fauquenoy S, Vanhulle C, Burny A, Rohr O, Van Lint C. Characterization of new RNA polymerase III and RNA polymerase II transcriptional promoters in the Bovine Leukemia Virus genome. bioRxiv. 2016. doi:10.1101/043034.
- Hutvagner G, Simard MJ. Argonaute proteins: key players in RNA silencing. Nat Rev Mol Cell Biol. 2008;9:22–32.View ArticlePubMedGoogle Scholar
- Sullivan CS, Grundhoff AT, Tevethia S, Pipas JM, Ganem D. SV40-encoded microRNAs regulate viral gene expression and reduce susceptibility to cytotoxic T cells. Nature. 2005;435:682–6.View ArticlePubMedGoogle Scholar
- Barth S, Pfuhl T, Mamiani A, Ehses C, Roemer K, Kremmer E, Jäker C, Höck J, Meister G, Grässer FA. Epstein-Barr virus-encoded microRNA miR-BART2 down-regulates the viral DNA polymerase BALF5. Nucleic Acids Res. 2008;36:666–75.View ArticlePubMedPubMed CentralGoogle Scholar
- Cox JE, Sullivan CS. Balance and Stealth: the role of noncoding RNAs in the regulation of virus gene expression. Ann Rev Virol. 2014;1:89–109.View ArticleGoogle Scholar
- Moir RD, Willis IM. Regulation of pol III transcription by nutrient and stress signaling pathways. BBA Gene Regul Mech. 2013;1829:361–75.Google Scholar
- Barbeau B, Péloponèse J-M, Mesnard J-M. Functional comparison of antisense proteins of HTLV-1 and HTLV-2 in viral pathogenesis. Front Microbiol. 2013;4:226.View ArticlePubMedPubMed CentralGoogle Scholar
- Rende F, Cavallari I, Corradin A, Silic-Benussi M, Toulza F, Toffolo GM, Tanaka Y, Jacobson S, Taylor GP, D’Agostino DM, Bangham CRM, Ciminale V. Kinetics and intracellular compartmentalization of HTLV-1 gene expression: nuclear retention of HBZ mRNAs. Blood. 2011;117:4855–9.View ArticlePubMedGoogle Scholar
- Iyer MK, Niknafs YS, Malik R, Singhal U, Sahu A, Hosono Y, Barrette TR, Prensner JR, Evans JR, Zhao S, Poliakov A, Cao X, Dhanasekaran SM, Wu Y-M, Robinson DR, Beer DG, Feng FY, Iyer HK, Chinnaiyan AM. The landscape of long noncoding RNAs in the human transcriptome. Nat Genet. 2015;47:199–208.View ArticlePubMedPubMed CentralGoogle Scholar
- Vance KW, Ponting CP. Transcriptional regulatory functions of nuclear long noncoding RNAs. Trends Genet. 2014;30:348–55.View ArticlePubMedPubMed CentralGoogle Scholar
- Pauli A, Valen E, Schier AF. Identifying (non-)coding RNAs and small peptides: challenges and opportunities. BioEssays. 2014;37:103–12.View ArticlePubMedPubMed CentralGoogle Scholar
- Willems L, Kettmann R, Dequiedt F, Portetelle D, Vonèche V, Cornil I, Kerkhofs P, Burny A, Mammerickx M. In vivo infection of sheep by bovine leukemia virus mutants. J Virol. 1993;67:4078–85.PubMedPubMed CentralGoogle Scholar
- Weil D, Boutain S, Audibert A, Dautry F. Mature mRNAs accumulated in the nucleus are neither the molecules in transit to the cytoplasm nor constitute a stockpile for gene expression. RNA. 2000;6:962–75.View ArticlePubMedPubMed CentralGoogle Scholar
- Stein CA, Hansen JB, Lai J, Wu S, Voskresenskiy A, Høg A, Worm J, Hedtjärn M, Souleimanian N, Miller P, Soifer HS, Castanotto D, Benimetskaya L, Ørum H, Koch T. Efficient gene silencing by delivery of locked nucleic acid antisense oligonucleotides, unassisted by transfection reagents. Nucleic Acids Res. 2010;38:e3.View ArticlePubMedPubMed CentralGoogle Scholar
- Van den Broeke A, Bagnis C, Ciesiolka M, Cleuter Y, Gelderblom H, Kerkhofs P, Griebel P, Mannoni P, Burny A. In vivo rescue of a silent tax-deficient bovine leukemia virus from a tumor-derived ovine B-cell line by recombination with a retrovirally transduced wild-type tax gene. J Virol. 1999;73:1054–65.Google Scholar