Ethics statement

Studies were performed at the Oregon National Primate research Center (ONPRC), Beaverton, Oregon. All animal work was conducted in accordance with the recommendations of the Weatherall report, "The use of non-human primates in research." Specifically, the research was regulated by the United States Department of Agriculture for compliance with the “Animal Welfare Act” and the National Institutes of Health’s Office of Laboratory Animal Welfare for compliance with the “Guide for the Care and use of Laboratory Animals”. The ONPRC Laboratory Animal Care and Use Program is fully accredited by the American Association for Accreditation of Laboratory Animal Care (AALAC), and it has an approved Assurance (#A3304-01) for the care and use of animals on file with the Office for Protection from Research Risks at NIH. The study’s experimental protocols were approved by the Oregon Health & Science University Institutional Animal Care and Use Committee.

Animals and animal care

Indian-origin rhesus macaques (Macaca mulatta) utilized in this study were bred and raised at the ONPRC. All animals were pair-housed in cages in environmentally controlled rooms and receive monkey chow twice daily supplemented with fresh fruit and water ad libitum. Environmental enrichment included food treats, manipulatable items in cages, human interactions with care takers and perches. All animals were observed twice daily by trained husbandry staff for activity, appetite and normal species-specific behavior. Ketamine HCl was used to induce anesthesia for all routine non-invasive clinical procedures associated with the study protocols such as blood sampling and agent administration. All surgical and invasive clinical procedures were conducted by trained personnel under the supervision of veterinarians in dedicated surgical facilities using aseptic techniques and comprehensive physiologic monitoring. Telazol^® was used to induce anesthesia for axillary and/or inguinal peripheral lymph node biopsies. Local analgesia at incision sites was provided by lidocaine and systemic analgesia was provided by buprenorphine. Post operatively all animals were evaluated daily for seven days to closely monitor recovery. For euthanasia macaques were sedated with ketamine HCl (15 mg/kg IM) and painlessly killed with greater than 50 mg/Kg body weight of sodium pentobarbital. Euthanasia was assured by exsanguination and bilateral pneumothorax. This method is consistent with the recommendations of the American Veterinary Medical Association’s Guidelines on Euthanasia.

Animal infections and lymph node biopsy sample collection

Lymph node (LN) biopsy samples were obtained from a cohort of eight RM that were described previously [18]. Briefly, animals in this cohort (Cohort IV) consisted of expanded specific pathogen free (ESPF) RM that were infected intravenously with 5x10⁶ PFU of purified RRV, with 4 RM (animals 29119, 27386, 29000, and 25617) receiving WT RRV BAC-derived RRV (WT BAC), and 4 RM (animals 26430, 25662, 28902, and 28834) receiving vCD200 non-sense mutant BAC-derived RRV (vCD200 N.S.). LN biopsy samples were obtained from all animals at day 0 (d0) and day 28 (d28) post-infection (pi), from axillary (AX) and inguinal (ING) lymph nodes, respectively. At the time of LN biopsy collection, total LN cells were isolated from homogenized whole LN tissue samples, frozen in RPMI containing 10% fetal bovine serum (FBS) and 10% dimethyl sulfoxide (DMSO), and stored in liquid nitrogen until time of RNA isolation.

After the establishment of a long term latent/persistent RRV infection, cohort IV RM were infected with 5ng of SIV_mac239 p27 antigen-containing cell-free supernatants to induce immunodeficiency. In addition, treatments were also later performed in an effort to further promote disease development in conjunction with SIV infection. Specifically, animals 27386 and 28902 were subjected to natural killer (NK) cell depletion through the administration of anti-rhesus IL-15 depleting antibody (M111R1, NIH Nonhuman Primate Reagent Resource) at days -42, -28, and -14 pre-SIV infection, the day of SIV infection, and at days 14, 28, and 42 post-SIV infection (10 mg/kg per dose), while animals 29119 and 26430 received anti-CD8α depleting antibody (cM-T807, NIH Nonhuman Primate Reagent Resource) on the day of SIV infection, and days 3 and 7 post-SIV infection (10 mg/kg per dose), to deplete both CD8+ T cells and NK cells. Finally, anti-rhesus CD8β depleting antibody (CD8beta255R1, NIH Nonhuman Primate Reagent Resource) was utilized to specifically deplete CD8+ T cells in animals 25617, 28834, 29000, and 25662, and was administered at day -71 pre-SIV infection (animals 25617 and 28834) or day 162 post-SIV infection (animals 29000 and 25662) at a dose of 50mg/kg.

Lymph node biopsy RNA isolation and RNA-seq library preparation

RNA was isolated from total cells derived from LN tissue samples collected at d0 and d28 pi using a Quick-RNA Miniprep kit and an RNA Clean and Concentrator Kit (Zymo Research, Irvine, CA). The number of frozen cells available for RNA isolation from each animal and timepoint was variable, ranging from approximately 4x10⁶ to 9x10⁶ cells.

RNA-seq libraries were prepared by the OHSU Massively Parallel Shared Sequencing Resource (OHSU MPSSR), using total RNA and a TruSeq RNA-seq kit (Illumina, San Diego, CA). Briefly, poly(A)+ RNA was isolated using oligo-dT coated magnetic beads, chemically fragmented, and converted to double stranded cDNA using random priming. The resultant double stranded DNA ends were blunted, a single “A” nucleotide was added to the 3’ end of each strand, and standard barcoded Illumina adaptors were ligated to the cDNA. The resulting libraries were cleaned using AMPure beads (Agencourt) and amplified by a limited number of rounds of PCR. The final concentration of each library was determined by real time PCR (Kapa Biosystems). Samples were combined for multiplexing and the concentration of the final mix determined by real-time PCR and diluted to the appropriate concentration. The flow cells were run on a HiSeq 2500 using a single read 100 cycle protocol, and base call files were converted to standard fastq format using Bcl2Fastq (Illumina).

RNA-seq data analysis

The quality of fastq files was assessed using FastQC (v0.11.3) [20] combined with MultiQC (http://multiqc.info). Fastq files were imported into PRIMe-Seq, and Trimmomatic [21] was used to trim the reads. Low quality bases at the beginning and end of each read were trimmed as well as any remaining Illumina adapters. Reads with less than 25 bases remaining were discarded. Version 7 of the RM genome (http://www.unmc.edu/rhesusgenechip/) was used along with its corresponding annotation (v7.6.8) [22], and the program STAR (v020201) [23] was used to align the reads to the genome. Counts for all samples were merged into one file, and low count genes were filtered out prior to normalization. Genes with at least 5 counts in at least 3 samples were retained for further analysis. The counts were imported into Bioconductor’s (v3.1) DESeq2 [24] for differential gene expression analysis using the DESeq2 rlog normalization approach. For RRV viral gene counts, the annotated RRV₁₇₅₇₇ genome (accession number AF083501.3) was used to align sequence reads.

Custom statistical modeling was performed by The Oregon National Primate Research Center (ONPRC) Bioinformatics & Biostatistics Core (BBC) to evaluate the effects of RRV infection and vCD200 expression over longitudinal changes. Mixed linear models (gene-by-gene repeated measures ANOVA) were used with mutation status (WT BAC versus vCD200 N.S.) as a between group factor and time point (d0 versus d28 pi) as a within group factor. Pairwise comparisons under repeated measures ANOVA were performed, and compound symmetry (CS) was used as correlation within each animal. False Discovery Rate (FDR) adjustment was used to control for multiple comparisons. DESeq2’s differential expression analysis method was also used to confirm results from custom modeling.

Volcano plots and lists of significantly altered genes were generated using the Shiny Transcriptome Analysis Resource Tool (START) (https://kcvi.shinyapps.io/START/). The online tool Venny (http://bioinfogp.cnb.csic.es/tools/venny/) was utilized to perform cross comparisons of infection groups and generate accompanying Venn diagrams. For cross comparison analyses, genes from WT BAC and vCD200 N.S. infection groups displaying a +/- 1.45-fold change or greater in expression from day 0 to day 28 post-infection (after rounding to 3 significant digits), and with an accompanying FDR p value of 0.05 or less, were utilized.

qRT-PCR analysis of cellular gene expression

Primer pairs for qRT-PCR analysis of cellular RM genes were designed using NCBI Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) with parameters that required each PCR product to span at least one intron, or span an exon-exon junction, such that only RNA transcripts from the associated genes were capable of being amplified from RNA samples. The RM genes analyzed by qRT-PCR, their corresponding NCBI reference sequence, and designed primer pairs utilized for each gene are as follows: AICDA (XM_015151031.1; Forward—GTCGGCATGAGACCTACCTG, Reverse—CACGTGACAGCCGGACTTAT), GPC1 (NM_001266539.1; Forward—GCCAGATCTACGGAGCCAAG, Reverse—GCAGATGCTCACCCGAGAT), CX3CR1 (NM_001265871.1; Forward—ACGCCAGGCTGTCACCAT, Reverse—ACAAATTTCCCACCAGGCCA), RasD1 (NM_001266563.1; Forward—TCCATCCTCACAGGAGACGTT, Reverse—TTGGTGTCGAGGATCTGCTG), TXNIP (NM_001257935.1; Forward—GAAGACCAGCCAACAGGTGA, Reverse—GGGAAGCTCAAAGCCGAACT), MMP-12 (XM_001098589.3; Forward—TTTTTGCACGTGGAGCTCAT, Reverse—GCTCATGAACAGCAACGAGG), AURKA (NC_027902.1; Forward—CAGGCATCATGGACCGATCT, Reverse—TTTGAAGGACACAGGACCCG), CCR5 (NC_027894.; Forward–GAGCTGAGACATCCGTTCCC, Reverse–ACACCAGTGAGTAGAGCGGA), CCL5 (XM_015119022.1; Forward–TGACTCACCTCTCCCACAGG, Reverse–TACTCCTTGATGTGGGCACG), LAG3 (NC_027903.1; Forward–CTGGGACCTACATCTGCCAT, Reverse—CTGGAGTCACCTCACAAAGCA), NUGGC (XM_015145044.1; Forward–GCATGTTTGAAAGGGCCCAG, Reverse–TTCACTCCCGACATCTGCAA).

LN RNA samples for each animal and timepoint were converted to cDNA using SuperScript III First-Strand Synthesis SuperMix (Invitrogen) and qPCR was performed with Power SYBR Green PCR Master Mix using a StepOnePlus Real-Time PCR system (Applied Biosystems). Individual samples were analyzed in duplicate reactions using primers specific for each gene of interest, as well as duplicate reactions utilizing rhesus macaque GAPDH primers, and copy numbers were determined using the relative standard curve method. Average values obtained for each gene were normalized to average GAPDH values, and fold change values of increased or decreased expression were determined by calculating the ratio of normalized expression levels at d28 pi versus d0, or d0 versus d28 pi, respectively.

Cell sorting for RNA isolation

4x10⁶ to12x10⁶ archived frozen LN biopsy cells were thawed, counted, and subjected to sequential sorting using magnetic beads specific for non-human primate CD20, CD3, and CD14 (Miltenyi Biotech, Bergisch Gladbach, Germany). Analysis of representative samples by flow cytometry indicates >97% purity of each population after sorting. Each sorted population, as well as remaining unsorted cell populations, were resuspended in RNA lysis solution and total RNA was isolated using a Quick-RNA Miniprep kit and an RNA Clean and Concentrator Kit (Zymo Research).

Tissue staining

LN biopsy tissue was collected from infected RM at d0 and d28 pi, and either placed in neutral-buffered formalin or neutral-buffered 4% paraformaldehyde for paraffin embedding. Sections from the LN were cut at 4 μm, deparaffinized and hydrated. After appropriate blocking with 5% normal goat serum and 5% bovine serum albumin and quenching, sections were incubated with for immunostaining with primary antibodies specific for AICDA (rabbit anti-AICDA-1; Novus biologicals; 1:1000), Glypican-1 (rabbit anti-Glypican-1; Novus biologicals 1:250) T cells (rabbit anti-human CD3, Dako, Carpinteria, CA, 1:200), B cells (mouse anti-CD20, clone L26, Dako, 1:200). For immunofluorescence detection, secondary biotinylated goat anti-mouse (Vector Laboratories) or goat anti-rabbit IgG (H + L) (Vector Laboratories) were used, followed by treatment with an Elite ABC kit (Vector Laboratories), and then further stained with streptavidin Alexa 488 (Molecular Probes, Eugene, OR) and streptavidin Alexa 594 (Molecular Probes), respectively, to visualize the antigens of interest. The sections were counterstained with 4′,6-diamino-2-phenylindole dihydrochloride (DAPI) (1:5000) and covered with Prolong gold anti-fade medium (Invitrogen, Carlsbad, CA) or Omnimount (National Diagnostics, Atlanta, Georgia). Sections were examined using a Zeiss Axio Imager M1 microscope (Carl Zeiss, Thornwood, NY) using Plan NeoFluar objective lenses (2.5 ×/0.5 NA and 40 ×/0.75 NA). Optical images were obtained with standard conditions of illumination and exposure by scanning on an Olympus VS110 slide scanner on a 20x objective (Olympus). Sections were routinely stained with isotype control monoclonal antibodies and the appropriate secondary antibodies, or secondary antibodies alone as negative controls.

Flow cytometry analysis

Total cells isolated from d0 and d28 pi LN biopsy samples and periorbital lymphoma tissue from animal 29119 were stained with antibodies directed against surface molecules CD20 (Biolegend, San Diego, CA), CD3 (BD Pharmigen, San Diego, CA), CD14 (Biolegend), and stained with Glypican 1 antibody (Novus Biologicals, Centennial, CO) conjugated to APC. Glypican 1 antibody was conjugated to APC using an APC conjugation kit (Abcam, Cambridge, UK). Cells were first stained with surface antibodies, fixed using FACS Lyse solution (Biolegend), washed with Permeabilization Wash Buffer (Biolegend), and then stained with conjugated Glypican 1 antibody. Cell were first gated on surface markers, then gated on Glypican 1, to determine percentage of Glypican 1+ cells in each subset. All samples were acquired using an LSRII instrument (BD Bioscience, San Jose CA), and data was analyzed using FlowJo software (TreeStar, Ashland, OR).

Statistical analysis

Unpaired two-tailed t-tests were performed using Prism software (version 8.1.2).

Animal samples and RNA-seq analysis

In a previous in vivo infection study, 4 independent cohorts of naïve RM were infected with a BAC-derived vCD200 mutant RRV (vCD200 N.S.) (n = 12), and were directly compared to control RM infected with WT BAC-derived RRV (WT BAC) (n = 11) [18]. As is typically observed after RRV infection, all RM displayed measurable viral loads in whole blood (WB) from days 14 to 42 pi, with peak viral loads occurring between d21 to d28 pi. However, on average, vCD200 N.S.-infected RM displayed higher average peak viral loads at d21 and d28 pi when compared to WT BAC-infected RM. Further, for one cohort in this study (cohort IV; WT BAC n = 4, vCD200 N.S. n = 4), LN biopsy samples were obtained at defined times pi, and examination of viral loads in LN biopsy samples from these animals revealed that vCD200 N.S.-infected RM displayed higher average peak viral loads in LN tissue at d21 and d28 pi, when compared to WT BAC-infected RM, with a statistically significant difference occurring between the two infection groups at d28 pi [18].

In an attempt to better understand the observed difference in viral load levels due to vCD200 expression, and to simultaneously assess the effects of de novo RRV infection on host cell gene expression patterns in LN tissue of infected RM, RNA-seq analysis was performed on RNA isolated from LN biopsy samples from all eight animals in cohort IV. Specifically, given that a statistical difference in viral DNA loads was detected in LN samples between infection groups at d28 pi, RNA was isolated from these samples, while RNA isolated from d0 LN biopsies collected just prior to infection served as a baseline control (Table 1). Due to limitations in LN biopsy tissue sample sizes, and the relatively low abundance of cells obtained from these samples, cells were not sorted into individual cell types prior to RNA purification and library preparation. Therefore, the resulting samples obtained largely represent RNA derived from total immune cells in LN, but also contain RNA from LN tissue stromal cells. After RNA purification and library preparation, RNA-seq analysis was performed on d0 and d28 pi samples from all eight animals. This analysis resulted in the identification of over 13,000 cellular genes with normalized counts, as well as the detection of transcripts from RRV genes.

Download:

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

Table 1. Animal infections and lymph node biopsies.

https://doi.org/10.1371/journal.pone.0228484.t001

Data from individual animals was compiled based on infection group and timepoints, and then subjected to analysis using custom statistical modeling. Initially, RRV gene expression patterns were assessed, confirming the presence of RRV transcripts only in d28 pi samples (Fig 1). Interestingly, slightly higher expression levels of viral genes were detected in LN biopsy samples of vCD200 N.S.-infected RM, a finding which is likely attributable to the fact that on average, more virus was present in LN tissue from these animals [18]. Despite this, comparison of viral gene expression levels between infection groups only identified one gene with a statistically significant difference in expression, ORFRU14-L, a gene with undescribed function, which displayed average normalized counts of 0.48 and 5.5 in d28 LN biopsy samples of WT BAC and vCD200 N.S.-infected RM, respectively (p = 0.0023, unpaired t-test). In general, the patterns of viral gene expression indicate that the majority of RRV genes expressed in d28 LN are those located near the right end of the viral genome, a region that contains the latency-associated genes ORF 71, 72, and 73, which encode the viral FLIP homolog (vFLIP), cyclin D homolog (vCyc), and latency-associated nuclear antigen (LANA), respectively (Fig 1). RRV ORF R15 encodes vCD200, and is transcribed as part of a lytic bi-cistronic message that also contains the downstream ORF74, which encodes the viral G protein-coupled receptor (vGPCR) [25]. Though transcripts from R15-ORF74 and other known lytic genes such as R3 (vMIP) remain detectable, the predominance of gene expression in the latency region suggests that viral gene expression patterns are trending towards latency in LN from infected RM by d28. This timepoint coincides closely with the onset of decreased peak viral loads in all infected RM, and the eventual establishment of a latent/persistent infection by approximately day 56 pi [18]. Overall, however, the general viral gene expression profiles are similar in both WT BAC and vCD200 N.S.-infected RM, indicating that the absence of vCD200 does not drastically alter RRV gene expression in LN tissue.

Download:

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

Fig 1. Viral gene expression in LN biopsy samples of RRV-infected RM.

RNA-seq data obtained from d28 pi LN biopsy samples of WT BAC and vCD200 N.S. infection groups was analyzed for RRV transcript expression. Values presented are the averages of normalized viral gene counts for all animals within each infection group (n = 4), with error bars indicating standard deviation. RRV gene names are listed on the X axis as organized from 5′ to 3′ in the RRV₁₇₅₇₇ genome. ORFRU14-L was the only viral gene found to display a significant difference in expression levels between infection groups (*, p = 0.0023, unpaired t-test).

https://doi.org/10.1371/journal.pone.0228484.g001

RRV infection induces changes in cellular gene expression profiles in LN tissue

In order to assess the overall effects of RRV infection on changes in cellular gene expression profiles in infected RM, analyses were performed to compare d0 and d28 pi expression profiles from all eight infected animals combined, regardless of the virus type utilized for infection. Using the selection criteria of a ≥ 2-fold change in expression levels from d0 to d28 pi, and an FDR adjusted p value of ≤ 0.05 as a cut-off value for the level of significance, genes either up- or down-regulated after RRV infection were identified. Based on these criteria, 60 genes displayed a ≥ 2-fold increase in expression from d0 to d28 pi, while only one gene displayed a >2-fold decrease in expression after RRV infection (Fig 2A). Genes of interest were next identified from this list based on their putative functional properties and potential relevance to both KSHV and RRV infection and disease development, and were then manually sorted into functional categories of immunity, oncogenesis, DNA replication and cellular proliferation, as well as other genes of interest not strictly associated with these categories (Fig 2B and 2C).

Download:

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

Fig 2. Effects of RRV infection on cellular gene expression profiles in LN tissue.

(A) Volcano plot displaying the log₂ fold-change in expression from d0 to d28 pi, and the associated -log₁₀ FDR p value, for all cellular genes detected by RNA-seq analysis in LN biopsy samples of all 8 RRV-infected RM combined. Only those genes that displayed a ≥ 2-fold change in expression from d0 to d28 pi, with an associated FDR p value of <0.05, were considered for further analysis. Using these criteria, 60 genes were found to be significantly upregulated, and only 1 gene was found to be significantly downregulated after RRV infection. Upregulated (B) and downregulated (C) genes of interest were sorted into categories based on putative functionality. Asterisks indicate the 3 genes displaying the highest changes in expression overall, and genes in bold font indicate those assessed by follow-up qRT-PCR analysis.

https://doi.org/10.1371/journal.pone.0228484.g002

A majority of genes up-regulated after RRV infection were found to be associated with DNA replication and cellular proliferation (e.g. cyclins and cell division cycle associated proteins), which is likely attributable to the generalized increase in cellular activation due to RRV infection, and the induction of immune cell proliferation due to the development of anti-RRV immune responses. For example, Ki67, a nuclear proliferation marker whose expression was previously found to be elevated at d28 pi by flow cytometry in CD4+ and CD8+ T cells in PBMC samples from these same animals [18], was also found to be up-regulated at d28 pi in LN biopsy samples by RNA-seq analysis, suggesting that gene expression alterations observed in vivo can also be associated with protein level changes in RRV-infected RM.

Although elevated proliferation and cell division markers likely correlate with a general increase in replication of immune cells within LN tissue, of more specific interest was the identification of genes associated with oncogenesis and immune regulation that display elevated or decreased expression after RRV infection, since these genes are most likely to be associated with alterations that affect anti-RRV immune responses, or affect cellular pathways relevant to RRV-associated disease development. Examination of these categories identified several genes of interest, three of which had the highest changes observed overall in our analysis, with the highest change occurring for GPC1 (fold change +3.0750), followed by AICDA (fold change +3.0197), and CX3CR1 (fold change +2.9575).

From this initial group of genes displaying a ≥ 2-fold change at d28 pi, several genes involved in chemokine signaling (CX3CR1, CCR5, CCL5), antibody diversification (AICDA, NUGGC), lymphocyte signaling (LAG3), B cell regulation (RasD1), and with connections to oncogenesis and cancer (GPC1, AURKA), were chosen for follow-up qRT-PCR analysis, to confirm observations made by RNA-seq. To achieve this, primers were designed to specifically amplify transcripts of these genes, and d0 and d28 pi LN RNA samples from individual RM were tested for transcript expression levels of each gene, as well as GAPDH as a control for normalization (Table 2). Next, a comparison of d28 pi versus d0 normalized expression levels (gene copies/GAPDH copies) for each gene of interest was made to determine the overall fold-increase in expression in individual animals after RRV infection. Of the 8 RRV up-regulated genes examined, only GPC1, AICDA, and CX3CR1 were successfully confirmed by qRT-PCR as displaying elevated expression levels in individual animals similar to those detected by RNA-seq analysis of RNA from these same tissue samples. Specifically, a ≥ 2-fold increase in transcript expression was detected in 6 of 8 RM analyzed for GPC1 and AICDA expression, and all 8 RM analyzed for CX3CR1 expression levels (Fig 3). The only gene identified that displayed a ≥ 2-fold reduction after RRV infection, RasD1, was found to be downregulated in all 8 RM by qRT-PCR analysis, further confirming the RNA-seq observations made for this gene (Fig 3D). Though a lack of confirmation by qRT-PCR of the remaining genes tested does not diminish observations made by RNA-seq analysis, those genes capable of being confirmed in this manner are further strengthened in their potential relevance to RRV-associated changes in cellular gene expression profiles in LN tissue of RRV-infected RM.

Download:

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

Fig 3. qRT-PCR analysis of LN biopsy samples from individual RM.

RNA samples from d0 and d28 post-infection LN biopsies of individual RM were analyzed by qRT-PCR for the expression of GPC1 (A), AICDA (B), CX3CR1 (C), and RasD1 (D). Data represents fold change in normalized gene expression values (gene copy number/GAPDH copy number) at d28 post-infection compared to d0 values. Error bars denote standard deviation, and a dashed line indicates 2-fold change value. Individual animal numbers and their associated infection groups are noted.

https://doi.org/10.1371/journal.pone.0228484.g003

Download:

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

Table 2. Gene-specific primers utilized for qRT-PCR analysis.

https://doi.org/10.1371/journal.pone.0228484.t002

Interestingly, the 4 genes identified by RNA-seq and confirmed by qRT-PCR as being the most highly up- or down-regulated after RRV infection all possess functional characteristics of genes with the potential to be involved in immune responses against RRV, or in the development of RRV-associated malignancies. For example, CX3CR1 is the receptor for fractalkine, and is involved in normal B cell function and immune regulation, but has also been found to be overexpressed in B cell lymphomas and B-CLL in humans [26–28], suggesting that a similar upregulation due to RRV infection may also be associated with disease development in RM. Alternatively, RasD1, the single gene identified as being downregulated > 2-fold after RRV infection, is a member of the Ras superfamily of G proteins, and has been implicated as a potential regulatory protein in B lymphocytes that may play a role in inhibiting B lymphocyte activity and proliferation [29]. RasD1 has also been found to be capable of inhibiting the growth of cancer cells [30], and thus, decreased expression of RasD1 in RM LN tissues after RRV infection could potentially promote increased B cell proliferation and the development of B cell abnormalities in RRV-infected RM.

Another gene confirmed to be upregulated after RRV infection, and of particular interest due to its role in B cell function, is activation-induced cytidine deaminase (AICDA). AICDA is induced in germinal center (GC) B cells upon antigen exposure, and is known as a major regulator of antibody diversification, where it is involved in both somatic hypermutation and class switching [31, 32]. Importantly, despite its normal role in development of antibody responses, the overexpression of AICDA also has the potential to promote off-target mutations, and has been implicated in the development of a variety of cancers, including lymphoma [33–35]. Normally, AICDA expression is tightly regulated, and though predominantly restricted to GC B cells, expression of this protein has also been found in some instances to occur in other cell types, where aberrant expression can lead to development of a variety of non-B cell cancers [31, 36]. In an attempt to identify cell types in which AICDA expression is induced in LN of RRV-infected RM, LN biopsy cells were subjected to sequential magnetic bead sorting to obtain purified CD20, CD14, and CD3 cell populations, and RNA was isolated from these subsets, as well as remaining unsorted cells, for use in qRT-PCR analysis. Due to limiting cell numbers, this analysis was only performed on sorted cells obtained from animals that displayed > 2-fold increase in the initial analysis of AICDA expression of total LN cells via qRT-PCR (Fig 3B), and included 3 WT RM (27386, 25617, 29000) and 2 vCD200 N.S. RM (26430 and 28834). As anticipated, the most consistent and robust changes in AICDA expression occurred in CD20+ B cells from these animals, with all 5 animals examined displaying a > 2-fold increase in expression, and animals 26430 and 29000 demonstrating the most dramatic changes, with 25.7-fold and 51.6-fold increases in expression from d0 to d28, respectively (Fig 4A). Interestingly, > 2-fold increases in AICDA expression were observed in the CD14+ subset from animals 25617 and 28834, the CD3+ subset of animal 29000, and in the unsorted subset of animal 25617, suggesting AICDA expression may also be induced in non-B cell populations after RRV infection in some instances. Overall, however, these findings demonstrate that AICDA gene expression is induced largely in B cells after RRV infection. In an attempt to examine changes AICDA protein expression in LN tissues, staining was performed on LN tissue sections of RM 29000, which displayed the highest level of induction in gene expression in total LN cells and CD20+ B cells by qRT-PCR, and for which tissue sections were available. Staining of this tissue revealed that AICDA protein expression was visibly elevated at d28 pi, when compared to d0 (Fig 4B), and although localized with B cells, AICDA was also associated with CD3+ T cells in these tissues (Fig 4C). These staining patterns parallel the RNA expression patterns observed in sorted cell subsets obtained from this same animal. Taken together, these findings confirm that RRV infection induces AICDA gene and protein expression in LN of infected RM in vivo, implicating the potential involvement of this gene in RRV associated disease development.

Download:

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

Fig 4. AICDA expression in sorted cell populations and tissue sections of LN biopsy samples.

(A) Cells from LN biopsy samples of three WT BAC and two vCD200 N.S.-infected RM were subjected to sorting using magnetic beads to obtain purified CD20+, CD14+, and CD3+ cell populations, and RNA was isolated from each population as well as remaining unsorted cells for use in qRT-PCR using AICDA-specific primers. Due to lack of sufficient RNA, unsorted samples of animals 29000 and 28834, and CD14 and CD3 samples of animal 26430, were not capable of being analyzed. Data represents fold change in normalized gene expression values (gene copy number/GAPDH copy number) at d28 pi compared to d0 values, and error bars indicate standard deviation. (B) Tissue staining of LN biopsy samples from animal 29000. Tissue sections were stained with anti-AICDA antibody (green), anti-CD20 or anti-CD3 antibody (red), and DAPI (blue) to stain nuclei. Original magnification X20. (C) Enlarged images of d28 pi tissue sections from animal 29000 demonstrating co-localization of AICDA with CD20+ B cells (left panel) and CD3+ T cells (right panel).

https://doi.org/10.1371/journal.pone.0228484.g004

Overall, the most highly upregulated gene identified by RNA-seq analysis of LN samples from RRV-infected RM was Glypican 1 (GPC1) (Fig 2). GPC1 is a member of a family of cell surface proteoglycans that promote formation of ligand-receptor complexes, enhancing signaling of cell surface receptors such as fibroblast growth factor (FGF), transforming growth factor-β (TGF-β), and vascular endothelial growth factor (VEGF), the effects of which can ultimately promote cell proliferation and oncogenesis [37]. Indeed, the ability of GPC1 to modulate cellular adhesion and motility has been implicated in promoting the growth and angiogenic and metastatic potential of cancer cells in a variety of models [38]. Of further importance, GPC1 has also been found to be enriched in cellular exosomes from cancer types ranging from pancreatic to colorectal cancer [37, 39, 40], and has been investigated as a potential biomarker for the early detection of pancreatic cancer [40]. Thus, a particularly intriguing finding was the fact that the highest induction of GPC1 gene expression was observed in animal 29119, the only animal in the entire cohort to develop lymphoma.

Increased Glypican-1 expression is associated with lymphoma development in vivo

As part of our standard protocol to assess RRV-associated disease development, cohort IV animals were infected with SIV_mac239 after establishing long-term latent/persistent RRV infections. Further, in an attempt to assess additional methods to enhance disease development, animals in this cohort were also differentially subjected to treatment with anti-IL-15 antibody to deplete NK cells (animals 27386 and 28902), anti-CD8α antibody to deplete both NK and CD8 T cells (animals 29119 and 26430), or anti-CD8β antibody to deplete only CD8 T cells (animals 25617, 28834, 25617, and 28834) (Table 3). By the end of the study period, RM 29199 was the only animal to present with RRV-associated disease, with the detection of a periorbital lymphoma at the time of necropsy. As only 1 of 8 RM in this cohort developed lymphoma, NK cell and/or CD8 T cell depletion did not specifically enhance viral-associated disease development, which typically occurs in 20–30% of RRV/SIV co-infected animals [2]. Of specific interest, however, was the fact that lymphoma development occurred in the same animal that also displayed the highest average level of induction of GPC1 gene expression in d28 pi LN biopsy samples.

Download:

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

Table 3. Secondary treatment strategies and lymphoma development in RRV-infected RM.

https://doi.org/10.1371/journal.pone.0228484.t003

To further explore the possible connection of GPC1 with lymphoma development, and to assess whether changes in GPC1 protein levels could be detected in both LN and lymphoma tissue from an RRV-infected RM, cells from LN biopsies and lymphoma tissue from animal 29119 were analyzed by flow cytometry. Specifically, cells isolated from d0 and d28 pi LN biopsy and lymphoma tissue were stained with anti-GPC1 antibody, in conjunction with antibodies directed against CD20, CD3, and CD14, to determine the levels of GPC1+ cell subsets in these tissues (Fig 5). In LN samples from RM 29119, 43.8% of d0 and 46.5% of d28 pi cells were CD3+ T cells (Fig 5A), while 49.1% of d0 and 45.8% of d28 pi cells were CD20+ B cells (Fig 5B). However, in lymphoma tissue from this same animal, only 4.23% of cells stained CD3+, while 51.7% of all cells stained CD20+, indicating the predominance of B cells in lymphoma tissue compared to normal LN. Less than 1% of all cells were CD14+ in lymphoma and LN samples. Analysis of GPC1 staining of LN and lymphoma cells indicated that total GPC1 protein levels are increased in CD20+ B cells when comparing d28 pi (12.1%) to d0 LN samples (6.39%), while a slight increase in GPC1 staining was also observed from d0 to d28 pi in CD3+ cells from LN samples (5.24% versus 2.42%). In general, this increase in GPC1 protein levels in CD20+ and CD3+ cells in d28 pi LN samples mirrors observations demonstrated by RNA-seq and qRT-PCR analysis, suggesting that increased GPC1 gene expression due to RRV infection also results in higher levels of GPC1 protein expression in LN cells. More interestingly, however, was the robust increase in GPC1 protein levels observed in lymphoma tissue B cells from this animal compared to levels detected in B cells from d0 and d28 pi LN samples, with 51.2% of lymphoma B cells staining positive for GPC1 expression, a 44.8% and 39.1% increase from levels detected in d0 and d28 pi LN B cells, respectively (Fig 5B). Although other factors are likely to have contributed to the eventual development of lymphoma in this animal, the elevation in GPC1 protein levels in CD20+ cells from LN after RRV infection at d28, and the high level detected in lymphoma-derived B cells, strongly suggests that GPC1 induction has the potential to be directly associated with development of cancers in RRV-infected RM.

Download:

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

Fig 5. Flow cytometry analysis of Glypican-1 expression in LN biopsy and lymphoma tissues.

Total cells derived from d0 and d28 pi LN biopsy and lymphoma tissue samples from animal 29119 were stained with antibodies directed against Glypican-1 (GPC1), CD20, CD3, and CD14, and analyzed by flow cytometry. Data was gated to determine the percentage of (A) CD3+ and CD3+/GPC1+ or (B) CD20+ and CD20+/GPC1+ cells in each sample. CD14+ cells represented < 1% of all cells in lymphoma and LN tissue samples.

https://doi.org/10.1371/journal.pone.0228484.g005

Effects of vCD200 on cellular gene expression profiles

Given that RRV vCD200 has inhibitory effects on host immune responses, and negatively affects RRV viral loads in vivo, this molecule is likely to alter host gene expression patterns during infection. In order to assess the contributions of RRV vCD200 to the modulation of host gene expression, and to determine if any of these changes may be associated with differences in viral loads observed between WT BAC and vCD200 N.S. viruses in LN tissue of RRV-infected RM, RNA-seq data from both infection groups was subjected to a cross comparison. Specifically, RNA-seq gene expression data for d0 and d28 pi samples was compiled separately for WT BAC (n = 4) and vCD200 N.S. (n = 4) infection groups, and a cross comparison between groups was then performed to identify genes differentially regulated during infection with each virus. When compiling data for individual infection groups, a less stringent cutoff value of ≥ 1.45-fold was utilized in order to include genes that might display fold change expression values slightly below 2 in initial cross comparisons, while the significance cut-off was maintained at an FDR p value of 0.05 (Fig 6A). Next, a cross comparison of all genes upregulated or downregulated ≥1.45-fold in each infection group was performed (Fig 6B), and the resulting lists of genes found to be altered only in specific infection groups were then manually examined for those displaying a ≥ 2-fold change in expression.

Download:

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

Fig 6. Cross comparison of cellular gene expression between WT BAC and vCD200 N.S. infection groups.

(A) Volcano plot displaying the log₂ fold-change in expression from d0 to d28 pi, and the associated -log₁₀ FDR p value, of all cellular genes identified by RNA-seq in WT BAC and vCD200 N.S. infection groups. (B) Venn diagrams comparing genes upregulated or downregulated in WT BAC or vCD200 N.S. infection groups. Only genes displaying a ≥1.45-fold change in expression with an accompanying FDR p value of < 0.05 were used for cross comparison analysis. The number of genes unique to each group displaying a ≥ 2-fold change in expression are indicated. Identified genes of interest from cross comparison analysis that display specific ≥ 2-fold upregulation in vCD200 N.S. infection group (C), ≥ 2-fold downregulation in WT BAC infection group (D), or ≥ 2-fold downregulation in vCD200 N.S. infection group (E). Functional categories of genes are indicated, and genes highlighted in bold were assessed by follow-up qRT-PCR analysis.

https://doi.org/10.1371/journal.pone.0228484.g006

Examination of genes that displayed ≥1.45-fold increase in expression in individual infection groups from d0 to d28 pi revealed that the largest number of changes occurred in vCD200 N.S.-infected RM (209 genes, 58.9%), while a smaller number of genes displayed increases in expression in RM infected with WT BAC (35 genes, 9.9%). None of the 35 genes specifically increased in WT BAC-infected animals displayed an induction ≥ 2-fold, and were not examined further, while 37 of the 209 vCD200 N.S.-specific genes were upregulated ≥ 2-fold at day 28. Further analysis revealed that 3 of the 37 genes meeting criteria for upregulation in the vCD200 N.S. infection group, specifically AICDA, aurora kinase B (AURKB), and NUGGC, were actually similarly elevated in the WT BAC infection group at d28 pi, having only been excluded from cross comparison analysis due to the possession of FDR p values in this group that were slightly above the cutoff utilized for level of significance (AICDA p = 0.118, AURKB p = 0.0772, NUGGC p = 0.0522). In addition, AICDA and NUGGC had previously been identified as displaying significant ≥ 2-fold increases in expression when comparing d28 pi to d0 LN samples from both infection groups combined (Fig 2). Thus, overall a total of 34 genes were confirmed as being specifically and significantly upregulated in the vCD200 N.S. infection group at d28 pi. Examination of these 34 genes revealed that none possess predicted functional properties with apparent connections to immune regulation or RRV pathogenesis, with the majority being associated with cellular proliferation and DNA replication (Fig 6C). While the increase in expression of cellular proliferation and DNA replication-associated genes in vCD200 N.S. infection group at d28 pi may simply be attributable to the elevated levels of viral replication observed in vCD200 N.S.-infected animals at this timepoint, the inhibition of their expression in WT BAC-infected RM due to vCD200 expression is also a possibility, although the specific implication of particular gene(s) from this list in directly regulating RRV replication is not readily apparent.

Genes downregulated in individual infection groups from d0 to d28 pi were less in number, with 24 displaying ≥1.45-fold decrease in the WT BAC infection group, and 56 displaying ≥1.45-fold decrease in the vCD200 N.S. infection group. Genes displaying a ≥ 2-fold decrease in either group were further limited, with only 2 genes in the WT BAC infection group, coiled-coil domain containing 152 (CCDC152) and matrix metallopeptidase 12 (MMP12) (Fig 6D), and one gene in the vCD200 N.S. infection group, thioredoxin interacting protein (TXNIP) (Fig 6E), demonstrating a ≥ 2-fold decrease in expression from d0 to d28 pi. The lack of variable expression of genes with ≥ 2-fold expression between infection groups is likely due to more subtle effects of vCD200-CD200R signaling on gene expression patterns overall, and though less stringent selection criteria could suggest alterations in a larger number of genes, we chose to focus only on those with more detectable and robust ≥ 2-fold changes in expression. Of the 3 differentially downregulated genes identified, only TXNIP appears to have potential functional properties that could directly influence viral replication, and result in the observed phenotypic difference in viral loads measured between vCD200 N.S. and WT BAC viruses in vivo.

TXNIP is a widely expressed cellular protein that is a negative regulator of thioredoxin (Trx), and directly interacts with and inhibits Trx function [41]. The Trx system is a cellular thiol-reducing antioxidant system that is involved in the regulation and maintenance of a reducing environment and inhibition of reactive oxygen species (ROS) production, protecting cells from damage due to oxidative stress. However, in addition to its role in Trx regulation, TXNIP has also been found to be a multifunctional protein that can affect a wide variety of cellular processes, and whose expression has been shown to promote apoptosis, NLRP3 inflammasome activation, and inhibit cellular proliferation [41–47]. In addition, TXNIP expression has also been found to be affected by infection with several viruses, including human T lymphotropic virus type-I (HTLV-I), hepatitis B virus (HBV), hepatitis C virus (HCV), as well as herpesviruses cytomegalovirus (CMV), Epstein-Barr virus (EBV), and KSHV [48–54]. In some instances, dysregulation of TXNIP has been suggested to play a direct role in viral oncogenesis [52, 53], and to affect virus replication [54].

TXNIP displayed a significant 2.07-fold decrease in expression levels at d28 pi by RNA-seq in vCD200 N.S. RM, but was not similarly downregulated in WT BAC-infected RM (1.27-fold decrease, FDR p = 0.437), excluding it from cross comparison analysis (Fig 6B). To further confirm the variance in downregulation of TXNIP between virus infection groups, RNA from individual RM was subjected to qRT-PCR analysis using primers specific for RM TXNIP (Fig 7A). Using this approach, a statistically significant difference in the average fold-decrease of TXNIP expression was not identified between infection groups, however, this analysis revealed that all four vCD200 N.S.-infected RM displayed 6.3-fold or higher downregulation of TXNIP expression at d28, while WT BAC-infected RM displayed much lower decreases in TXNIP expression, ranging from 1.3-fold to 4.9-fold. Given the roles of TXNIP in preventing cell proliferation, promoting apoptosis, and inducing inflammasome activation, it is plausible that elevated TXNIP levels in infected cells in LN tissue would negatively affect RRV replication. This finding suggests a possible mechanism that may help explain why vCD200 N.S.-infected RM, which display lower levels of TXNIP expression, possess higher viral loads in LN tissues than those observed in WT BAC-infected RM.

Download:

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

Fig 7. TXNIP gene expression in LN of individual RM and sorted cell populations from WT BAC and vCD200 N.S. infection groups.

(A) RNA samples from d0 and d28 pi LN biopsy cells of individual RM were analyzed for expression levels of TXNIP by qRT-PCR. Data represents the fold change decrease in normalized gene expression values (gene copy number/GAPDH copy number) from d0 to d28 pi, and individual animals and infection groups are noted. Error bars denote standard deviation, and a dashed line indicates 2-fold change value. (B) Cells from LN biopsy samples of individual animals from WT BAC (animals 29119, 27386, 29000, and 25617) and vCD200 N.S. (animals 26430, 25662, 28902, and 28834) infection groups were subjected to sorting using magnetic beads to obtain purified CD20+, CD14+, and CD3+ cell populations, and RNA was isolated from each population as well as remaining unsorted cells for use in qRT-PCR using TXNIP-specific primers. The average fold-change decrease in normalized gene expression values (gene copy number/GAPDH copy number) from d0 to d28 pi for each infection group is shown. Error bars denote standard deviation, and a dashed line indicates 2-fold change value. The difference in decreased TXNIP expression in CD20+ cells was found to be significantly different between WT BAC and vCD200 N.S. infection groups by unpaired t-test (*, p = 0.0369).

https://doi.org/10.1371/journal.pone.0228484.g007

To further investigate the potential role of TXNIP in regulating RRV infection, we also sought to determine whether variances in TXNIP expression were localized to specific immune cell subsets in LN cells. In particular, as RRV naturally infects B cells in vivo, serving as a site for RRV replication and latency, it was of interest to assess changes in TXNIP expression levels in LN B cells from WT BAC and vCD200 N.S.-infected RM. To achieve this, cells from d0 and d28 pi LN biopsies from each animal were subjected to sequential magnetic bead sorting to obtain purified CD20+, CD14+, and CD3+ cell populations, and RNA was then isolated from these and any remaining unsorted cells. qRT-PCR analysis of TXNIP expression levels in individual cell subsets revealed that the largest average decrease in TXNIP expression occurred in CD20+ B cells from RM infected with vCD200 N.S. (8.5-fold), which was a higher reduction than that observed in CD20+ B cells from WT BAC-infected RM (3.1-fold) (Fig 7B). This observed difference in B cell TXNIP expression levels was found to be statistically significant between infection groups. A slightly larger decrease in TXNIP expression in CD3+ T cells was also observed in vCD200 N.S.-infected RM when compared to WT BAC-infected RM, though this difference was not statistically significant, while no difference in reduction of TXNIP expression levels was observed in CD14+ cells or unsorted LN cells. Overall, this data demonstrates that in the absence of vCD200, TXNIP expression is reduced predominantly in CD20+ B cells in LN tissue of RRV-infected RM. Although the specific mechanism by which vCD200 may help promote TXNIP expression is currently unknown, in general, these findings suggest that decreased TXNIP expression in vCD200 N.S-infected RM may be associated with the development of higher viral loads in LN tissues of these animals.