A mutation of the BRE decreases antisense transcriptional activity of the LTR promoter

To understand the biological function of BLV antisense transcription, a reverse genetics approach was used to selectively target the antisense promoter located in the 3’-LTR (Fig 1A). Therefore, a series of mutations were introduced by site-directed mutagenesis in the BRE and MTE/DPE core elements located in the R-U5 region of the LTR as indicated on Fig 1B. The mutations were targeted to the 2 transcription initiation sites of antisense RNAs (A, A1 and A2 in the BRE and B in the MTE/DPE). The GG>TA and GC>TA mutations were designed to disrupt the interaction of the promoter element to the Transcription factor II B (TFIIB). The mutated LTR sequences were inserted into a reporter plasmid containing the Renilla and Firefly luciferase genes cloned in opposite orientations (Fig 1C). The different mutants were introduced in bidirectional reporter vectors containing the full-length LTR or a minimal promoter devoid of U3 (R-U5). Upon transfection of human embryonic kidney cells (HEK293FT) with the R-U5 reporter, no Firefly-luciferase activity could be detected above background levels indicating that sense transcription required U3, as expected (Fig 1D). Transfection of the R-U5 reporter produced Renilla-luciferase activity that was not significantly affected by the A2/B1 mutations (Fig 1E). In contrast, the GG to TA mutation present in the A/A1 reporters reduced Renilla-luciferase activity. A similar effect of the mutations was observed in the presence of the Tax transcriptional activator (pSGTax [26]). In the context of a full-length LTR, the A1 mutation slightly increased expression of Firefly-luciferase in basal conditions (pSG5) and after co-transfection of pSGTax (Fig 1F). Compared to the WT, Renilla-luciferase activity was significantly reduced by the A1 mutation in absence and presence of Tax (Fig 1G). The A1 mutation also reduced antisense transcription in reporter assays performed with the Ovine Kidney (OVK) cell line (S1 Data).

These dual luciferase reporter assays thus revealed that a minimal two-nucleotide mutation (A1) in the BRE affects antisense transcriptional activity directed by the LTR promoter.

A provirus carrying the A1 mutation transcribes less antisense RNAs, does not affect envelope-mediated cell fusion and is slightly more infectious in cell cultures

To evaluate the effect of antisense transcription in the proviral context, the A1 mutation was introduced in an infectious BLV molecular clone (pBLV344, Fig 2A). After transfection into HEK293FT, viral transcription, transmission to ovine cells and envelope-mediated cell fusion were analyzed by RT-qPCR, co-culture with OVK-U3Luc reporter cells and syncytia formation, respectively [27–29] (Fig 2B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Compared to the WT, a provirus containing the A1 mutation transcribes less antisense RNAs but is infectious and forms syncytia.

(A) Schematic representation of the provirus containing the A1 mutation (in orange): genomic structure, viral microRNAs (miRs in yellow), antisense transcripts (as1-S, as1-L, as2 in purple) and the Tax/Rex mRNA (in blue). Note that other sense mRNAs encoding structural and regulatory proteins are not shown. (B) Experimental design. HEK293FT cells were transfected with either the WT provirus or the A1 mutant. At 48 hours post-transfection, HEK293FT cells were collected to determine viral transcription, infection of OVK-U3Luc and syncytia formation with CC81 cells. (C) Quantification of tax, as1-S/L and as2 mRNAs by RT-qPCR. Relative quantification was performed by the ΔΔCT method using ß-actin as reference [78]. Data resulting from three independent experiments were normalized to the mean of the WT samples, arbitrarily set to 1. p-values were calculated according to two-tailed Wilcoxon signed rank test (D) After 24 hours of coculture, luminescence activity was quantified in OVK-U3Luc reporter cells containing a stably integrated Firefly luciferase gene cloned downstream of the U3 promoter. Data in relative luminescence units (RLU) were normalized to the mean of the WT samples, arbitrarily set to 1. p-values were calculated according to Mann-Whitney tests. (E) Representative images of syncytia formed after five days of coculture between CC81 cells and HEK293FT transfected with WT/A1 proviruses. Cells were stained by May-Grünwald Giemsa and imaged with a Motic AE 2000 inverted light microscope. (F) The number of syncytia characterized by 10 or more nuclei in 9 images from 3 independent experiments were counted using the Motic 3.0 live software. p-values were calculated according to Mann-Whitney tests.

https://doi.org/10.1371/journal.ppat.1012659.g002

RT-qPCR quantification revealed that the levels of antisense RNAs were reduced in cells transfected with the A1 provirus compared to the WT control (p = 0.004; as1 by 5-fold and as2 by 11.9-fold; Fig 2C). In contrast, the WT and A1 viruses expressed similar amounts of Tax RNAs (p = 0.41), indicating that sense transcription remained mostly stable. To assess viral transmission, HEK293FT transfected with proviruses WT/A1 were then co-cultured with an ovine cell line transduced with a luciferase gene controlled by the U3 promoter (OVK-U3Luc). The luciferase activity increased by a factor of 1.7-fold (p = 0.03), suggesting that the A1 mutation may slightly improve viral infectivity (Fig 2D). However, the ability to promote cell fusion with CC81 indicator cells, as estimated by the number and the size of syncytia, was not statistically different (p = 0.55) (Fig 2E and 2F). The coculture of HEK293FT with CC81 cells did not significantly modify the transcriptional pattern of the WT and A1 proviruses (S2 Data).

These observations thus show that, compared to the WT, the A1 provirus transcribes significantly less antisense RNAs, is slightly more infectious and induces similar levels of cell fusion events in transient transfection experiments.

The A1 provirus is infectious in sheep but replicates at reduced proviral loads

To investigate the role of antisense transcription in vivo, sheep were inoculated with the WT proviral clone (n = 5) or the A1 mutant (n = 5). After 20 to 35 days, antibodies directed against the surface receptor-binding subunit of the viral envelope protein (gp51, SU) were detected in the plasma of all animals. The anti-gp51 reactivity was similar in sheep inoculated with the A1 and WT proviruses and persisted over time revealing the onset of an antiviral immune response (p = 0.64; Fig 3A). Quantitative PCR revealed that, compared to the WT, the proviral loads were reduced by 39-fold in the PBMCs of A1-infected sheep (p = 0.002; Fig 3B). The normalized RNA loads of as1-S/L and as2 evaluated by RT-qPCR were also significantly decreased in the A1 PBMCs (p = 0.004 and p = 0.04 respectively; Fig 3C). In contrast, the levels of the Tax transcripts in the two groups were variable but not statistically different (p = 0.91). To evaluate the expression of viral proteins, PBMCs were transiently cultured overnight and analyzed by confocal microscopy using a monoclonal antibody directed against the p24 capsid protein (Fig 3D). At the single cell level, the p24 mean fluorescence intensities (MFIs) were similar in PBMCs infected with WT and A1 viruses (p = 0.42; Fig 3E).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The A1 mutant provirus is infectious in sheep and modifies the host cell transcriptome.

(A) Ten sheep were inoculated with either the wild-type provirus (n = 5) or the A1 mutant (n = 5). Anti-gp51 antibody titers were determined at 46 days post-inoculation using a blocking ELISA (IDEXX). The y axis represents three-fold serial dilutions of the plasma of a BLV-infected sheep with high proviral load (#3684) used as reference. (B) qPCR quantification of the proviral loads in sheep PBMCs. The y axis represents the number of proviral copies in genomic DNA (1 μg) of sheep #2130 (WT) and #0522 (A1). The number of viral copies was normalized to a standard curve performed with the 18S cellular gene. p-values were calculated according to Mann-Whitney tests. (C) Quantification of Tax, as1 and as2 mRNAs by RT-qPCR. PMBCs were collected on day 811 post-inoculation. Relative quantification was performed by the ΔΔCT method using ß-actin as reference [78]. Data resulting from at least three independent experiments were normalized to the means of the WT sample, arbitrarily set to 1. p-values were calculated according to two-tailed Wilcoxon signed rank test. (D) Confocal microscopy images of PBMCs cultured overnight, stained with Draq5 (nuclei in red) and labelled with anti-p24 antibody (4’G9) combined with an AlexaFluor 488 conjugate (p24 in green). (E) Mean fluorescence intensity (MFI) of p24 labeling calculated using the ZEN Blue software (n = 5). p-values were calculated according to Mann-Whitney test. (F) RNA sequencing of PBMCs isolated from sheep infected with the WT or A1 proviruses. PBMCs (n = 3) were divided into B and non-B cell populations by MACS using the 1H4 anti-sIgM monoclonal antibody. After RNA extraction and reverse transcription, cDNAs were sequenced with an Illumina NovaSeq6000 device (100bp, paired-end, 30 million reads). (G) Volcano plot of most significant up- or downregulated genes in the B-lymphocytes of A1-infected sheep compared to WT controls. The volcano plot was generated using the EnhancedVolcano (version: 1.16.0) and ggplot2 (version 3.3.3) R packages. The x and y axes are log₂(FC) and -log₁₀(p-value), respectively. Cut-off values were defined as p-adj < 0.05 and |Log₂FC| > 1. The red arrow points to the Hmcn1 gene. (H) Unsupervised heatmap of the 15 most significant (based on Log₂ FC) up- and down-regulated genes in B-cells from non-infected or A1- and WT-infected sheep. After PCA analysis, the non-infected sheep (#5751) was removed due to aberrant variation compared to samples of the same group. Aberrant behavior was further confirmed through correlation analysis. (I) Chord diagram of the most significant pathways affected in B-lymphocytes from A1-infected sheep compared to the WT. The left side of the graph represents the genes that are ranked by their log₂(FC) and that are connected to the terms of the KEGG pathways on the right. Chord diagrams were generated using GOplot (version:1.0.2) R package.

https://doi.org/10.1371/journal.ppat.1012659.g003

To characterize the mechanisms ongoing in sheep infected with the WT provirus and the A1 mutant, the transcriptome of their B lymphocytes was determined by RNA sequencing. After isolation of B-cells by magnetic-activated cell sorting (MACS), RNAs were purified, converted to cDNA and sequenced with an Illumina NovaSeq6000 device (Fig 3F). After quality control and trimming, the sequencing reads (100bp, 30M, paired-end) were aligned on the Ovis aries reference genome (Oar_rambouillet_v1.0) using the HISAT2 (version 2.2.1) software. After alignment and quantification using featureCounts, differentially expressed genes were identified between cell populations using DESEq2 Bioconductor package. The volcano plot highlighted specific changes (DEG at p-adj < 0.05 and |Log₂FC| > 1) that correlated with infection with A1 and WT viruses (S3–S6 Data). Particularly, the Hmcn1 gene, coding for the hemicentin-1 protein, was highly upregulated (p-adj = 5.29*10⁻¹²⁷ and |Log₂FC| = 7.46; red arrow on Fig 3G). The heatmap of the 30 most significant DEG highlighted significant differences in the transcriptome of A1 and WT-infected B-cells (Fig 3H). To obtain a comprehensive view of the underlying mechanisms, an unsupervised analysis of the transcriptomic data set was performed. Gene Ontology (GO) comparison and pathway enrichment analysis revealed that KEGG pathways pertaining notably to proliferation (oas04110 “Cell cycle”, oas04110 “DNA replication”), signal transduction (oas04015, “Rap1 signaling pathway”, oas04060 “cytokine-cytokine receptor interaction”) and adhesion (oas04514 “cell adhesion molecule”, oas04510 “focal adhesion”) were the most significantly affected (Fig 3I). In contrast, the transcriptome was almost similar in the non-B cells from A1- and WT-infected sheep (S3, S7–S9 Data). Sashimi plots drawn by Integrative Genomics Viewer (IGV) confirmed that RNA loads were reduced in A1 PBMCs compared to WT but could not reveal any significant alteration of splicing events (S10 Data).

Altogether, these results reveal that the A1 mutant replicates at reduced proviral loads and deeply modifies the transcriptome of infected B-cells.

Hemicentin-1 is upregulated in PBMCs and secondary lymphoid organs of A1-infected sheep

Among the DEGs identified by RNA sequencing, the expression of Hmcn1 appeared to be particularly high in B-lymphocytes of A1-infected sheep compared to the WT and mock controls (NI) (Fig 3). To validate this observation independently, RT-qPCR showed that the normalized fold-change of Hmcn1 RNA in A1-infected sheep exceeded that in NI and WT by 538 and 810-fold, respectively (p<0.0001; Fig 4A). Confocal microscopy further demonstrated that the expression of the hemicentin-1 protein encoded by the Hmcn1 gene increased in A1-infected PBMCs and p24⁺ cells compared to the WT (Fig 4B, 4C and 4D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. HMCN-1/hemicentin-1 is upregulated in peripheral B-cells, lymph nodes and spleen of A1-infected sheep.

(A) RT-qPCR quantification of Hmcn-1 RNA in PBMCs collected at day 811 post-injection. Relative quantification was performed by the ΔΔCT method using ß-actin as reference [78]. Data resulting from three independent experiments were normalized to the means of WT samples, arbitrarily set to 1. p-values were calculated according to the two-tailed Wilcoxon signed rank test. (B) Expression of hemicentin-1 protein in ex-vivo cultures of BLV-infected PBMCs. Hemicentin-1 was labelled with the HPA051677 rabbit antiserum and an AlexaFluor 647 secondary antibody (in red). The p24 capsid protein was revealed by the 4’G9 monoclonal antibody and an AlexaFluor 488 conjugate (p24 in green). Scale bars indicate 5 μm. (C) Quantification of the mean fluorescence intensity (MFI) of hemicentin-1 in PBMCs from sheep #2130 (WT) and #0522 (A1) measured using the ZEN Blue software (n = 30 cells). (D) MFI of hemicentin-1 in cells expressing the p24 protein. (E) The OVK cell line was transduced with lentivectors encoding two Hmcn-1 shRNAs (sh1, sh2) or a scrambled control. Hmcn-1 RNA levels in transduced cell lines were quantified by RT-qPCR. Relative quantification was performed by the ΔΔCT method using ß-actin as reference [78]. Data resulting from five independent experiments were normalized to mock OVK cells, arbitrarily set to 1. p-values were calculated according to the Mann-Whitney test. (F) OVK cells transduced with lentivectors encoding scrambled/sh1/sh2 shRNAs were recorded for 48 hours using the Incucyte live-cell imaging system. Tracking plots of cell migration over time (in μm) of 20 individual cells identified manually by the Celltracker and modelized with the Matlab software. (G) Migration distances covered by 60 individual cells during the culture as measured with Celltracker. (H) Representative immunohistochemistry of hemicentin-1 in spleen biopsies collected from sheep #2130 (WT) and #0522 (A1). The numbers of hemicentin-1+ cells per mm² of spleen tissue. p-values were calculated according to Mann-Whitney tests. (I) Hemicentin-1 immunohistochemistry in lymph node biopsies isolated from sheep #2130 (WT) and #0522 (A1). Enumeration of hemicentin-1+ cells per mm² of spleen tissue. p-values were calculated according to Mann-Whitney tests. (J) Spleen sections from WT and A1-infected sheep were stained with DAPI (nuclei in blue) and labelled with anti-HMCN1 antibody combined with an AlexaFluor 647 conjugate (in red). Slides were imaged with a Stellaris confocal microscope (Leica).

https://doi.org/10.1371/journal.ppat.1012659.g004

Hemicentin-1 (Fibulin-6 or ARMD1) is a member of the immunoglobulin superfamily that is expressed in a wide range of tissues including lymphoid conduits [30,31]. Hemicentin-1 mediates a wide range of mechanisms such as cytoskeleton rearrangements, mitosis [32,33], basement membrane invasion [30,34–36] and cell migration [37,38]. To investigate the role of hemicentin-1, OVK cells were transduced with lentivectors encoding Hmcn1 shRNAs (sh1, sh2) or a scrambled (scrbl) control. Effective RNA interference of Hmcn1 transcription was demonstrated by RT-qPCR (2.5-fold by sh1 and 3-fold by sh2; Fig 4E). The motility of shRNA-transduced OVK cells was then recorded for 48 hours by Incucyte live-cell imaging. Migration of individual cells was measured using the QuPath software and graphed using MatLab (Fig 4F). As hypothesized, knock-down of Hmcn1 reduced the ability of OVK cells to undergo cell migration (p = 0.001 and p<0.001 for sh1 and sh2, respectively; Fig 4G).

To extend these observations, the expression of hemicentin-1 was determined by immunohistochemistry in lymphoid organs from WT- and A1-infected sheep. Hemicentin-1 was expressed in the spleen and lymph node of an A1-infected animal (Fig 4H and 4I). In contrast, hemicentin-1 could not be detected in a biopsy of a prescapular lymph node isolated from a WT-infected sheep (Fig 4I). Expression of hemicentin-1 was significantly higher in the spleen of A1-infected sheep compared to the wild-type (n = 7.21 and n = 1.68 hemicentin-1⁺ cells / mm² in A1 and WT biopsies, respectively). Confocal microscopy revealed that hemicentin-1 predominantly aggregated in the intercellular space within lymphoid organs (Fig 4J and S11 Data).

Extending the RNA sequencing data (Fig 3), it appears that a factor regulating cell migration, the hemicentin-1 protein, is overexpressed in PBMCs and in lymphoid organs from sheep infected by the A1 mutant.

The A1 mutation attenuates oncogenesis in sheep

The pathogenesis induced by BLV is tightly controlled by the immune response mediated by lymphoid organs. In particular, cell proliferation in the spleen has a central role in virus-driven leukemogenesis [39,40]. To evaluate proliferation, Ki67-based immunochemistry was performed on biopsies of lymphoid organs. Compared to WT, the number of Ki67⁺ cells was moderately reduced by 1.73-fold in the lymph nodes of sheep infected with the A1 mutant (p<0.0001; Fig 5A). In contrast, cell proliferation in the A1 spleen increased by a factor of 9 compared to WT (p<0.0001; Fig 5B). The number of proviral copies per μg of genomic DNA determined by qPCR dropped by 54.3- and 197.8-fold in the lymph node and the spleen of A1-infected sheep (p = 0.002; Fig 5C and 5D). When normalized to the total viral load, the spatial distribution of the A1 proviruses in the PBMCs and the lymphoid organs was affected. Compared to WT, the relative proportion of A1 proviral DNA was reduced by 4.3-fold in the spleen (Fig 5E). In contrast, the percentages of A1 and WT proviral DNA were proportionally similar in the lymph nodes (1.9% and 1.2%, respectively). In other terms, attenuated replication of the A1 mutant correlated with increased cell proliferation and reduced spatial distribution of the proviruses in the spleen.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The A1 provirus affects cell proliferation in lymphoid tissues and has reduced pathogenicity in sheep.

(A) Immunohistochemistry of Ki67 labelling (in red) in the prescapular lymph node. The numbers of Ki67+ cells per mm² were quantified using the QuPath 0.3.0 software. p-values were calculated according to Mann-Whitney tests. (B) Same experiment with spleen biopsies as described in panel A. (C) Quantification of the proviral loads in the prescapular lymph nodes. The number of proviral copies in genomic DNA (1 μg) of sheep #2130 (WT) and #0522 (A1) was measured by qPCR. The number of viral copies was normalized to a standard curve performed with the 18S cellular gene. p-values were calculated according to Mann-Whitney tests. (D) Same experiment as described in panel C performed with spleen biopsies. (E) Comparison of the viral spatial distribution between the lymph node and the spleen of WT- and A1-infected sheep. (F) At regular time intervals, plasma was collected and tested for the presence of anti-gp51 antibodies using a blocking ELISA (IDEXX). The dashed line represents the seropositivity threshold. (G) The proviral loads in PBMCs of infected sheep were quantified by qPCR. The y axis represents the number of viral copies per 100 PBMCs. p-values were calculated according to Wilcoxon matched-pairs signed rank test. Note that data result from 2 independent experiments. † means dead. (H) Genomic DNA was extracted from PBMCs of WT- or A1-infected sheep (#5667 and #0582, respectively). Representative sequencing profiles of the LTR sequence amplified by PCR.

https://doi.org/10.1371/journal.ppat.1012659.g005

The five sheep inoculated with either the WT provirus or the A1 mutant remained persistently infected as indicated by the continuous presence of plasmatic antiviral antibodies (Fig 5F). The PVLs were significantly lower in the A1-infected sheep than in the WTs (p = 0.0001; Fig 5G). While the WT provirus was pathogenic, none of the A1-infected sheep developed any clinical manifestation of leukemia or lymphoma (S12 Data). Quantitative PCR and sequencing revealed that the two-nucleotide mutation introduced in the BRE of the A1 proviruses did not revert to WT (illustrated for #0582 in Fig 5H).

Collectively, these data thus show that the A1 mutant replicates at reduced proviral loads and attenuates pathogenicity in sheep.

Ethics statement

Animal experimentation was conducted in accordance with national (Royal Decree on the Protection of Experimental Animals) and international (European Union) guidelines. Sheep were kept under L2-restricted conditions (registration number LA1900600), according to the ethical protocols n°1515/2067, approved by the Commission d’éthique de l’utilisation des animaux à l’université de Liège (Uliège), at the Centre d’Etude en Production Animales (CEPA) facility located in Gembloux (Belgium).

Plasmid constructions

The full-length LTR (531bp) and the R-U5 fragment (last 132bp) of the pBLV344 provirus [71] were inserted into the KpnI and PstI restriction sites of the dual luciferase reporter plasmid pAsLuc(Fire)-HTLV-Luc(Reni) [72] using standard molecular biology techniques. In these pAsLuc(Fire)-531bp-Luc(Reni) and pAsLuc(Fire)-132bp-Luc(Reni) constructs, the HTLV LTR was thus replaced by BLV sequences. Mutations (A, A1, A2 and B1) were introduced in these reporter vectors using the XL QuickChange II Site-Directed Mutagenesis kit (Agilent Technologies) and primers listed in S15 Data. Mutation A1 was also introduced in both LTRs of plasmid pBLV344 containing a full-length wild-type provirus [71], yielding pBLV344-A1.

The pU3BLV-Luc lentivector (Vector ID VB201221-1032xzh; VectorBuilder) contains the 79bp BLV enhancer element [73], cloned upstream of the pGL3Basic minimal promoter and the humanized Luc2 Firefly luciferase gene. Lentivectors coding for short hairpin RNA against Hmcn1 (shRNA1 5’-AGCGATAAAGGGACATATATT-3’ and shRNA2 5’-CAAAGATGGCAAGCCTTTATT-3’) were obtained from VectorBuilder (Vector ID VB220127-1038fqz and vector ID VB220127-1040mmq, respectively). The scrambled control (GVV_659 pLV shRNA NT) was provided by Emmanuel Di Valentin (GIGA viral vector platform). The packaging plasmids pSPAX2 (#12260) and VSV.G (#14888) were obtained from Addgene.

Cell culture and lentiviral transduction

Human embryonic kidney cells (HEK293FT; Cellosaurus CVCL_6911) and OVK [28] were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Biowest) supplemented with 10% decomplemented fetal bovine serum (FBS, Gibco) and 1% of penicillin-streptomycin solution 100x (VWR). The CC81 feline cell line [29] (Cellosaurus CVCL_7216) was cultured in Roswell Park Memorial Institute 1640 medium (RPMI 1640, Biowest) supplemented with 10% decomplemented FBS and 1% of penicillin-streptomycin solution. All cultures were maintained at 37°C, in a humidified 95–5% air/CO₂ incubator (Memmert ICO150).

The OVK-U3Luc, OVK-shRNA1, OVK-shRNA2 and OVK-Scramble cell lines were generated by lentiviral transduction of the OVK cell line. Lentivectors (429 ng) encoding scrambled/sh1/sh2 shRNAs and pU3BLV-Luc were co-transfected in HEK293FT (2.5x10⁵ cells/well of a 6-well plate) with VSV.G (857 ng) and pSPAX2 (1.714 μg) using Jetprime (Westburg). Supernatant (2 mL) of transfected HEK293FT cells was harvested 72 hours post-transfection, filtered through 0.45 μm cellulose membrane filters and added to OVK (5x10⁵) in presence of hexadimethrine bromide (3.66 μg/mL; Sigma-Aldrich). At 72 hours post-transduction, antibiotic selection was performed for two weeks with blasticidin (20 μg/mL; InvivoGen).

Dual-luciferase assays

The WT and mutated bidirectional reporters (50 ng pAsLuc(Fire)-531bp-Luc(Reni) and pAsLuc(Fire)-132bp-Luc(Reni) were cotransfected with 300 ng of a plasmid encoding Tax (pSGTax [26]) or a control (pSG5) into HEK293FT cells (5x10⁵ cells/well of a 6-well plate) using Jetprime reagent. Cells were lysed at 24 hours post-transfection and Renilla/Firefly luciferase activities were measured with a luminometer (Turner Designs TD-20/20) using the Dual-Luciferase Reporter Assay system (Promega), according to the manufacturer’s protocol.

Syncytia formation

The pBLV344 and pBLV344-A1 plasmids (20 μg) were mixed with 21 μg of polyethyleneimine (25 kD, Polysciences) in OPTI-Minimum Essential Medium (OPTI-MEM, Gibco) and added to HEK293FT cells (2x10⁶ cells in a 10 cm diameter culture plate) for 24 hours. After replacement of the medium, HEK293FT cells were cultured for 24 hours, washed with Dulbecco’s Phosphate Buffered Saline (DPBS, Biowest) and detached with trypsin-EDTA (Biowest). Transfected HEK293FT (3x10⁵) cells were added to CC81 cells (1.5x10⁶) in a 10 cm diameter culture plate. After 72 hours, cells were washed and stained in 5mL May-Grünwald buffer (Sigma-Aldrich) for 3 minutes at room temperature. After addition of 5mL of water for 1 minute, the stain was removed and replaced by 10 mL of 10% giemsa’s azue methylene blue solution (Merck) for 20 minutes. Finally, the stain was removed, and cells were washed with 10 mL of neutral water.

The number of syncytia with more than 10 nuclei were scored by two different operators using a inverted light microscope (Motic AE 2000). Representative pictures were taken with the Motic 3.0 live software.

In vitro infectivity assays

Plasmids pBLV344 and pBLV344-A1 (1 μg) were transfected using Jetprime into HEK293FT cells (5x10⁵ cells/well of a 6-well plate). After replacement of the medium 24 hours post-transfection, cells were cultured for one day and mixed with OVK-U3Luc (5x10⁵ cells/well of a 6-well plate). 24 hours later, Renilla and Firefly luciferase activities were measured with a luminometer using the Dual-Luciferase Reporter Assay system, according to the manufacturer’s protocol.

Live cell imaging

OVK-scramble, OVK-shRNA1 and OVK-shRNA2 cells (9x10³ cells/well of a 96-well plate) were cultured at 37°C in a humidified 95/5% air/CO₂ incubator of an Incucyte imaging S3 Live-cell system (Sartorius). Migration of individual cells was recorded every 15 minutes using the CellTracker software (v1.1) [74]. Each frame underwent vignetting correction by bicubic interpolation and frames were aligned prior to the analysis. Manual tracking of each individual cell was performed, with a linear interpolation between each frame. Motility graphs were generated using the MATLAB software (MathWorks).

Inoculation of proviruses into sheep

Plasmids pBLV344 or pBLV344-A1 (20 μg) were transfected into HEK293FT cells (9x10⁶ cells in a 15 cm diameter culture plate), using 60 μL of polyethyleneimine (1 μg/μL, 25 kD, Polysciences) in 1 mL of OPTI-MEM. After overnight culture, cells were washed with DPBS and cultured in fresh medium for 24 hours. Transfected HEK293FT cells were detached with trypsin-EDTA, pelleted by centrifugation and resuspended in 1 mL DPBS. After a freeze-thaw cycle at -80°C, the cell lysate was inoculated intravenously and subcutaneously into sheep.

At regular time intervals, whole blood was collected by jugular venipuncture using 2% ethylenediaminetetraacetic acid (v/v) (EDTA, VWR) as anticlotting agent. After collecting the plasma, PBMCs were separated by Percoll (Merck) gradient centrifugation and preserved at -80°C in FBS containing 10% DMSO. The detection of anti-gp51 antibodies in the plasma was performed by ELISA (BLV gp51 Antibody Test Kit; IDEXX), according to the manufacturer’s protocol. Optical density at 450nm (OD₄₅₀) was measured with a Multiskan GO Microplate Spectrophotometer (Thermo Fisher Scientific). The seropositivity score was calculated using the following formula: 100–100 multiplied by (OD₄₅₀ of the sample to be tested divided by OD₄₅₀ of the negative control). The sample was considered seropositive providing that the score was > 60%.

Proviral load quantification by qPCR

Genomic DNA was extracted from PBMCs using the Blood and Tissue kit (Qiagen), according to the manufacturer’s protocol. The PCR mix contained 2.5 μL of DNA at 20 ng/μL, 6.25 μL of TAKYON ROX master mix (Eurogentec), 0.25 μL of forward and reverse primers at 20 μM (S15 Data) and 3.25 μL of milliQ-grade water. Real-time PCR amplification was performed in a thermocycler (Stepone Real-Time PCR, Applied Biosystems). The thermal protocol was initiated by a 10 minutes denaturation step at 95°C, followed by 40 cycles (15 sec at 95°C, 60 sec at 60°C) and terminated by a melting curve (1 minute at 60°C, increasing 0.3°C steps every 15 sec up to 95°C). The number of viral copies was normalized to a standard curve performed with the 18S cellular gene (2*BLV/18S), as described previously [28,75]. The number of viral copies per cell was then obtained by multiplying the BLV/18S ratio by the average value of the number of 18S copies per cell. The spatial distribution of the virus was determined by comparing the calculated proviral copy number measured in 1μg of DNA extracted from PBMCs, spleen and lymph node to the total amount of proviral copies detected in these tissues.

RNA quantification by RT-qPCR

RNA was extracted from transfected cells and PBMCs using the NucleoSpin RNA kit (Macherey-Nagel) according to the manufacturer’s protocol. Total RNA (200 ng) was converted to cDNA with the Reverse Transcriptase core kit (Eurogentec). 2 μL of complementary DNA (cDNA) was amplified by qPCR, using the same thermocycling protocol as described above, using primers corresponding to hmcn1, actin and viral transcripts (as1, as2 and Tax; S15 Data). Relative quantification was performed using actin as housekeeping gene.

Sequencing of proviral amplicons

DNA was extracted from PBMCs using the DNeasy Blood and Tissue kit (Qiagen), according to the manufacturer’s protocol. Genomic DNA (500 ng) was amplified in a mix containing 1 μL forward and reverse primers (20 μM round 1; S15 Data), 0.25 μL Taq polymerase (500 U/μL, Thermo Scientific), 5 μL Taq buffer 10X (Thermo Scientific), 3 μL MgCl₂ (25mM, Thermo Scientific), 1 μL dNTP mix (2 mM, Thermo Scientific) and 36.75 μL miliQ-grade water. The protocol of the thermocycler (Veriti 96-Well Thermal Cycler, Applied Biosystems) was initiated by a 5 minutes denaturation step at 95°C, followed by 40 cycles (30 sec at 95°C, 30 sec at 67°C, 1 minutes at 68°C) and terminated by 5 minutes at 68°C. A 2 μL aliquot of the PCR mix was reamplified with round 2 primers at an annealing temperature of 58°C. DNA amplicons were migrated onto a 1% agarose gel, purified using the QIAquick gel extraction kit (Qiagen) and sequenced by the GIGA genotranscriptomic platform. Chromatograms were analyzed using the Chromas software (Technelysium).

Immunohistochemistry of lymphoid organs

Spleen and lymph node biopsies were fixed in 4% paraformaldehyde (PFA, Sigma-Aldrich) for 24 hours and conserved in 70% ethanol (VWR). Biopsies were then embedded in paraffin, sectioned with a motorized microtome (Microm HM355S, Thermofischer Scientific / RM2165, Leica) and deposited onto glass slides. Deparaffinization and rehydration were performed by successive washes: xylene (VWR) (n = 3; 5 minutes), 100% ethanol (n = 2; 10 minutes), 95% ethanol (n = 2; 10 minutes) and distilled water (n = 2; 5 minutes). Antigen was retrieved in a pressure cooker (Ultra Cell Conditioning Solution, 950–224, Roche) in either EDTA (11 minutes, Invitrogen) or citrate buffer (15 minutes, Invitrogen). After cooling for 30 minutes, slides were washed three times with Tris-buffered saline with Tween 20 (TBST) (pH 7.4) under gentle agitation and incubated one hour at room temperature with a blocking solution (TBST-FBS 10%). Slides were washed in distilled water and incubated with primary antibodies against Ki67 (Ventana 790–4286) or hemicentin-1 (18837-1-AP, ProteinTech) overnight in a humidified chamber at 4°C. Slides were washed 4x 5 minutes with TBST under gentle agitation and incubated with secondary antibodies (anti-mouse IgG AlexaFluor 488 (Invitrogen) or anti-rabbit IgG AlexaFluor 647 (Invitrogen), respectively). After 3 TBST washes for 5 minutes under gentle agitation, the nuclei were labelled with DAPI (2μM, 422801, Biolegend) for 15 minutes in the dark. Immunostaining was revealed using the HIGHDEF red IHC chromogen (AP) (Enzo Life Science), and imaged with a Microscope Slide Scanner (ZEISS Axioscan). For confocal microscopy, slides were washed 3x 5 minutes with TBST under gentle agitation, fixed with 8 μL of Prolong Glass Antifade Montant (ThermoFisher), and recorded with a Leica Stellaris device. All images were analyzed using the QuPath-0.3.0 software [76].

Confocal microscopy and flow cytometry analysis of PBMCs

After venipuncture, PBMCs were purified by Percoll gradient centrifugation and cultured in RPMI 1640 (Lonza) supplemented with 10% decomplemented FBS, 1% of penicillin-streptomycin solution, phorbol 12-myristate 13-acetate (PMA, 10ng/mL, Invivogen) and ionomycin (1 μg/mL, VWR) [25,77]. After fixation in 4% PFA and permeabilization with PBS-Triton X-100 0.5%, cells were labelled with antibodies directed against p24 (4’G9 monoclonal) and hemicentin-1. After staining with Draq5 (Invitrogen), samples were incubated with appropriate conjugates (anti-mouse IgG AlexaFluor 488 and anti-rabbit IgG AlexaFluor 647, respectively), mounted on slides using Fluoroshield (Sigma-Aldrich) and imaged with the Zeiss LSM 510 confocal microscope. Data was analyzed with the ZEN Blue software (Zeiss). For flow cytometry analyzes, cell events were recorded with a BD FACSAria (BD Biosciences) or a CytoFLEX (Beckman Coulter) cytometer. Data was analyzed using the BD FACS Diva (BD Biosciences), CytoFLEX (Beckman Coulter), FlowJow V10 and the CytExpert 2.5 softwares.

Transcriptomic analysis

After Percoll gradient centrifugation, 10⁶ PBMCs were labelled with anti-sIgM antibody (monoclonal 1H4) and loaded onto a LS MACS column (Miltenyi Biotec) fixed on a QuadroMACS magnetic separator (Miltenyi Biotec). After collecting the flowthrough fraction (i.e., non-B cells) and washing the column with DPBS, the B cells were recovered. Total RNA was extracted from the B and non-B populations using the RNA Nucleospin extraction kit (Macherey-Nagel) according to the manufacturer’s protocol. After verifying with a bioanalyzer that the RNA Integrity Number (RIN) exceeded 7.0, poly-A libraries were generated with the Illumina TruSeq stranded mRNA kit (Illumina) and sequenced with a NovaSeq6000 sequencing device (Macrogen). At least 30 million 100 bp paired-end, reads were recorded corresponding to a sequencing coverage of approximately 46 using the Lander/Waterman equation. After quality check with FastQC, reads were aligned on the ovine reference genome (Oar_rambouillet_v1.0) using HISAT2 (version 2.2.1). The number of sequences that mapped to known genes were quantified with the featureCounts method from subread package (version 2.0.3). A Principal Components Analysis (PCA) was performed to exclude sample outliers. Differentially expressed genes were identified using DESEq2 (version 1.32.0). Volcano plots of top significant genes were generated using the EnhancedVolcano (version: 1.16.0) and ggplot2 (version 3.3.3) R packages. Chord diagrams were generated using GOplot (version:1.0.2) R package.

Statistics

All statistical analyses were performed using the GraphPad Prism 8.0.1 and the Microsoft Office 365 Suite softwares. For all tests, the threshold significance α was set at 0.05. Normality of distributions was tested according to the D’Agostino & Pearson test, using the method of Royston. Luciferase luminescence, number of syncytia, proviral loads, mean fluorescence intensity and number of positive cells in immunohistochemistry were compared between groups according to Mann-Whitney tests or unpaired t tests. Fold changes of RNA quantities were calculated according to two-tailed Wilcoxon signed rank test. Proviral load kinetics between WT and A1-infected animals were evaluated with Wilcoxon matched-pairs signed rank test. Kaplan-Meier survival curves were compared using the Log-rank Mantel-Cox test.