A spliced latency-associated VZV transcript maps antisense to the viral transactivator gene 61

Abstract

Varicella-zoster virus (VZV), an alphaherpesvirus, establishes lifelong latent infection in the neurons of >90% humans worldwide, reactivating in one-third to cause shingles, debilitating pain and stroke. How VZV maintains latency remains unclear. Here, using ultra-deep virus-enriched RNA sequencing of latently infected human trigeminal ganglia (TG), we demonstrate the consistent expression of a spliced VZV mRNA, antisense to VZV open reading frame 61 (ORF61). The spliced VZV latency-associated transcript (VLT) is expressed in human TG neurons and encodes a protein with late kinetics in productively infected cells in vitro and in shingles skin lesions. Whereas multiple alternatively spliced VLT isoforms (VLT_ly) are expressed during lytic infection, a single unique VLT isoform, which specifically suppresses ORF61 gene expression in co-transfected cells, predominates in latently VZV-infected human TG. The discovery of VLT links VZV with the other better characterized human and animal neurotropic alphaherpesviruses and provides insights into VZV latency.

Fig. 1

HSV-1 transcriptome profile during lytic and latent infection. Strand-specific mRNA-seq of lytically HSV-1-infected ARPE-19 cells and seven latently HSV-1-infected human trigeminal ganglia (TG) (Supplementary Table 1). a Circos plots of the HSV-1 genome [purple band; sense and antisense open reading frames (ORFs) indicated as blue and red blocks, respectively], the latency associated transcripts (LATs) indicated as green blocks, with internal tracks revealing the lytic (left) and latent (right) transcriptomes using unenriched (grey tracks) and HSV-1-enriched (black tracks) libraries. Lytic transcriptomes were profiled using HSV-1-infected ARPE-19 cells. Latent HSV-1 transcriptomes are profiled from seven TGs, with each track depicting a single specimen. Peaks facing outward from the centre indicate reads mapping to the sense strand, while peaks facing inward originate from the antisense strand. The y axis is scaled to maximum read depth per library in all cases. b Linear representation of the HSV-1 LAT genomic region (green blocks in a), with blue and yellow tracks depicting HSV-1-enriched mRNA-Seq reads originating from the sense and antisense strand, respectively. Unenriched mRNA-Seq tracks for ARPE-19 cells, and TGs 1 and 2, are superimposed and shown in light blue (sense) or yellow (antisense), with overlapping regions in medium-blue and orange, respectively. HSV-1 genome coordinates are shown the HSV-1 reference strain 17 (NC_001806.2); Previously described HSV-1 ORFs within this locus (red boxes), miRNAs (orange blocks), LAT-encoded ORFs (green blocks), LAT-encoded small RNAs (dark red blocks) and LAT (blue boxes) are scaled representatively. . Paired-end read data sets were generated with read lengths of 2 × 34 bp (ARPE-19) or 2 × 76 bp (TG1 and TG2) or 2 × 151 bp (TG3–TG7)

Fig. 2

VZV transcriptome profile during lytic and latent infection. Strand-specific mRNA-seq of lytically VZV-infected ARPE-19 cells and seven latently VZV-infected human trigeminal ganglia (TG) (Supplementary Table 1). a Circos plots of the VZV genome [purple band; sense and antisense open reading frames (ORFs) indicated as blue and red blocks, respectively], with internal tracks revealing the lytic (left) and latent (right) transcriptomes using unenriched (grey track, left panel) and VZV-enriched (black tracks, left and right panels) libraries. Right panel: latent VZV transcriptome of seven TG, with each track depicting a single specimen. Peaks facing outward from the centre indicate reads mapping to the sense strand, while peaks facing inward originate from the antisense strand. The y axis is scaled to maximum read depth per library in all cases. b Linear representation of the varicella latency-associated transcript (VLT) genomic region (black lines in a), with blue and yellow tracks depicting VZV-enriched mRNA-Seq reads originating from the sense and antisense strands, respectively. Unenriched mRNA-Seq tracks for ARPE-19 cells, and TGs 1 and 2, are superimposed and shown in light blue (sense) or yellow (antisense), with overlapping regions in medium-blue and orange, respectively. No VZV-mapping reads were obtained from unenriched sequence data sets generated from TGs 1 and 2. VZV genome coordinates are shown the VZV reference strain Dumas (NC_001348.1); blue and red arrows indicate previously described VZV ORFs, and light blue boxes indicate the five VLT exons. Paired-end read data sets were generated with read lengths of 2 × 34 bp (ARPE-19) or 2 × 76 bp (TG1 and TG2) or 2 × 151 bp (TG3–TG7)

Fig. 3

The genomic locus encoding the VZV latency-associated transcript (VLT). a Schematic diagram showing the location and structure of the five VLT exons (blue blocks) and introns (blue lines) within the genomic region 101,000–106,000 (coordinates refer to VZV reference strain Dumas; NC_001348.1) (Supplementary Table 3). b The VLT mRNA sequence including the 5’ untranslated region; start and stop codons are highlighted in red italic, while the cleavage factor I (CFI)-binding motifs are highlighted in green italic and the canonical polyadenylation signal site (AATAAA) is underlined. Location and boundaries of VLT exons are indicated by vertical black lines. c The fully translated VLT protein (pVLT), with the sequence of peptide (red) used to produce rabbit polyclonal anti-pVLT antibody

Fig. 4

Prevalence of VLT and ORF63 transcript in human TG. a PCR amplification of cDNA (n = 5 TGs, Supplementary Table 1), synthesized in the presence (+) or absence (−) of reverse transcriptase. Sanger sequencing of all five purified PCR products yielded identical sequences corresponding to VLT (Fig. 3b). b Integrative Genomics Viewer (IGV) screenshot showing representative RNA-Seq data from a single TG. Paired-end reads are shown as fragments mapped across the VLT locus. Grey boxes (exons) connected by black lines (introns) indicate individual read pairs with distinct read-pairs separated by white space. Each fragment spans between 2 and 5 exons and there was no evidence of additional upstream exons. c In lytically VZV-infected MeWo cells, Sanger sequencing of amplicons generated through rapid amplification of cDNA ends revealed multiple VLT_ly isoforms, visualized using IGV. Sequencing of 29 clones identified the three most abundant VLT_ly isoform groups (red boxes) A (35%), B (19%) and C (15%). Note that the specific VLT isoform observed in latently infected TG was not observed among VLT_ly isoforms. d Quantification of ORF63 transcript and VLT isoforms in lytically VZV-infected ARPE-19 cells (red; n = 3) and latently VZV-infected TG (blue; n = 19 TG), using primers/probes spanning VLT splice junctions between exons 3→4, A→1, B→1 and C→1 (Fig. 4c and Supplementary Table 5). Data represent mean (±SEM) relative transcript levels normalized to β-actin RNA. nd not detected. e Levels of paired ORF63 transcript and VLT (primers/probe spanning exon 2→3) in the same VZV^POS TG (VZV^POS; n = 15) determined by RT-qPCR. ***p < 0.001; Wilcoxon signed rank test. f Correlations (Spearman) between relative ORF63 transcript and VLT levels in VZV^POS TG (n = 15), as determined by RT-qPCR. g–j In situ hybridization (ISH; red signal) analysis of ORF63 RNA and VLT in latently VZV-infected TG (n = 12). g Frequency of neurons positive for ORF63 RNA and VLT in consecutive TG sections of the same donor. *p < 0.05; paired Student’s t-test. h Nuclear and cytoplasmic ORF63 RNA and VLT expression in consecutive TG sections from individual donors. ns not significant; paired Student's t-test. i, j Representative ISH images of VZV^POS TG sections (i) and two VZV naive human fetal dorsal root ganglia (j). Nuclei were stained with haematoxylin. Magnification: ×400 (with 3× digital zoom for insets). Bars = 50 µm

Fig. 5

Expression of VLT protein in vitro and in situ. a Representative confocal microscopic image of ARPE-19 cells at 48 h post-transfection with C-terminal FLAG-tagged VLT (pVLT-FLAG) expression plasmid or empty control plasmid (empty). pVLT-FLAG was detected with antibodies directed to FLAG (red) and VLT protein (pVLT; green). Magnification: ×1000 and ×2 digital zoom. Bars = 5 µm. b Representative confocal microscopic images of uninfected and VZV strain EMC1-infected ARPE-19 cells at 2 days post-infection, stained for both pVLT (green) and VZV glycoprotein E (gE; red) (left panel) or pVLT (green) combined with VZV ORF62 protein (IE62; red) (right panel). Magnification: ×200, with area indicated by the white box shown at ×600. Bars = 20 µm (×200) and 10 µm (×600). In a, b, nuclei were stained with Hoechst 33342 (cyan) and images are representative of results from four independent experiments. c RT-qPCR quantitation of VZV ORF61, ORF29 and ORF49 transcript and VLT levels in VZV pOka-infected ARPE-19 cells cultured with (PFA+) and without phosphonoformic acid (PFA−) for 24 h. Data represent mean (±SEM) fold-change in gene expression, using the respective ‘PFA−' value as a calibrator, from four independent experiments. ***p < 0.001; Wilcoxon signed rank test. d Detection of VLT (red) in a human herpes zoster (HZ) skin lesion by in situ hybridization. Magnification: ×100. Inset: ×400 and 2× digital zoom. Bar = 200 µm. e Consecutive HZ skin sections stained immunohistochemically for ORF63 protein (IE63; brown) and pVLT (brown) (left and middle panels) and sections from unaffected control skin stained for pVLT (right panel). Magnification: ×200. Inset: ×400. Bars = 100 µm. Images in d, e are representative of five HZ skin biopsies stained

Fig. 6

Selective repression of VZV ORF61 gene expression by VLT in co-transfected cells. ARPE-19 cells were transfected with plasmids encoding FLAG-tagged VLT (VLT_ATG), mutated VLT in which the ATG start codon was replaced by ATA sequence (VLT_ATA) or empty control plasmid (empty) in combination with plasmids encoding ORF61, ORF62 and ORF63. a Analysis of VZV ORF61, ORF62 and ORF63 and human β-actin transcript by RT-qPCR. Data represent mean (±SEM) fold-change in gene expression using the empty vector to calibrate ORF61, ORF62 and ORF63 or using VLT_ATA to calibrate VLT_ATG. Data are from two independent experiments performed in duplicate. nd not detected. ***p < 0.001; one-way ANOVA with Bonferroni’s correction for multiple comparisons. b Western blot analysis using antibodies directed to proteins encoded by VZV ORF61 (IE61), ORF62 (IE62), ORF63 (IE63), VLT (pVLT) and FLAG-tagged pVLT and α-tubulin. None, untransfected ARPE-19 cells. Images are representative of four independent experiments