Fig 1.
The schematic depicts the repair steps performed to generate FL-RhCMV. Unaltered ORFs and the unmodified viral genome are shown in orange while the terminal repeats are indicated in blue. ORFs containing known mutations that were repaired in this study are highlighted in red. Genes contained in the acquired inversion in the ULb’ region are shown in green, while genes lost in 68–1 but re-inserted into the genome during the repair are highlighted in purple. The transposon picked up during the generation of 68–1.2 RhCMV is highlighted in yellow and the 180.92 RhCMV PRC members used in the construction of 68–1.2 RhCMV are marked in grey. Repair 0: The frameshift resulting in a premature stop codon in Rh61/60 of 68–1 RhCMV was repaired and the two missing PRC members (Rh157.4 and Rh157.5) were inserted to generate 68–1.2 RhCMV as described previously [25]. Repair 1: Two DNA fragments combined spanning 6.9kb corresponding to the genomic sequence of the ULb’ homologous region in the circulating virus originally isolated from sample 68–1 [40] were synthesized. Three undefined bases in the published nucleotide sequence (KF011492) were taken from the consensus sequence of all sequenced low-passage RhCMV isolates. A synthetic DNA fragment spanning the region upstream of Rh157.5 (UL128) to Rh161 (UL146G) in its original orientation was used to replace the corresponding gene region in 68–1.2 RhCMV. The resulting construct maintains the repaired Rh61/60 gene while also containing the original isolate 68–1 genes Rh157.4 (UL128) and Rh157.5 (UL130) as well as the genes coding for the UL146 homologs Rh158.2, Rh158.3 and Rh161.1. Repair 2: Two previously described frameshift mutations in Rh13.1 [10] were repaired resulting in an intact Rh13.1 ORF. Repair 3: A premature stop codon in the viral Fcγ-Receptor homolog Rh152/151 [10] was repaired restoring the ORF to its original length. Repair 4: A nonsynonymous point mutations in Rh164 (UL141) initially predicted by us was confirmed by sequencing the original urine isolate. Hence, we restored the natural DNA sequence. Repair 5: Full genome sequencing of the 68–1.2 RhCMV BAC revealed that an E. coli derived transposon had inserted itself into the Rh167 ORF. The transposon was removed by en passant mutagenesis and the intact Rh167 ORF was restored. Repair 6: The US14 homolog Rh197 contained a premature stop codon which was repaired.
Fig 2.
Sequence relationship of FL-RhCMV with NHP CMVs.
A phylogenetic tree for FL-RhCMV and rodent and primate CMVs was constructed based on full genome alignments using the Geneious Prime Tree Builder application. Sequences previously published include RhCMV 180.92 [37] as well as the RhCMV isolates 19262, 19936 and 24514 and the Cynomolgus CMV isolates 31906, 31907, 31908 and 31909 [41]. We also included the published sequences for the CyCMV strains Ottawa [46] and Mauritius [47], the simian (African green monkey) CMV isolates Colburn [94], GR2715 [45] and stealth virus 1 [95] as well as the BaCMV strains OCOM4-37 [96] and OCOM4-52 [97] and the DrCMV strain OCOM6-2 [97]. For comparison we included the HCMV TR3 strain [34], the chimpanzee CMV strain Heberling [98] and the only two complete genome sequences of new world NHP CMVs, Aotine betaherpesvirus 1 (AoHV-1) strain S34E [99] and Saimiriine betaherpesvirus 4 (SaHV-4) strain SqSHV [100]. New genome sequences included in this alignment are as follows: the two RhCMV isolates 34844 and KF03, the CyCMV isolate 31709, the Japanese macaque CMV JaCMV 24655 and the two baboon CMVs BaCMV 31282 and 34826. These CMVs were isolated from fibroblast co-cultures of urine samples obtained from NHP housed either at the Oregon National Primate Research Center (ONPRC) or the Tulane National Primate Research Center (TNPRC). Also included in the alignment are the genomic sequences of the previously published RhCMV isolates UCD52 and UCD59 that originated at the UC Davis National Primate Research Center [29,30]. The rodent CMVs include the rat CMV (RCMV) isolates Maastricht [101], England [102] and Berlin [103], the guinea pig CMV (GPCMV) isolate 22122 [104] and the murine CMV (MCMV) strain Smith [105], which was used as an outgroup.
Fig 3.
Viral ORFs contained in FL-RhCMV compared to other NHP CMVs.
Full genome annotations of all listed old world NHP CMVs are shown. The leftmost column indicates the HCMV nomenclature for CMV encoded genes. Each ORFs that has a defined orthologue in HCMV and old world NHP CMVs is marked. If an orthologue cannot be clearly identified, the homologous gene family is given. The second column identifies the old world NHP CMV nomenclature. The same ORF nomenclature is used across all shown species, with the first or the first two letters corresponding to the host species (e.g. Rh for rhesus macaque). The virus strain analyzed is indicated. Green boxes indicate ORFs present in a particular strain, whereas red boxes indicate ORFs that are absent. Frameshifts or point mutations leading to shortened or elongated ORFs are highlighted in yellow or blue, respectively. Grey boxes indicate absent ORFs due to missing genome sequence information whereas ORFs with partial sequences are highlighted in purple. Orthologous ORFs that lack a conserved canonical start codon in some strains are highlighted in orange.
Fig 4.
Conditional expression of the RL13 homolog Rh13.1 results in reduced spreading and genomic rearrangements.
A) Deletion or reduced expression of Rh13.1 results in increased plaque size. Telomerized rhesus fibroblasts (TRF) or TRF expressing the tet-repressor (tetR) were transfected with the depicted BACs shown above. All recombinant BACs were engineered to express GFP from a P2A linker after UL36 [6]. 18 days later plaque sizes were visualized by GFP expression, and measured using ImageJ. Statistical significance was determined using an ordinary one-way ANOVA test with a p-value significance of <0.05. B) Representative images of the GFP positive plaques produced by the indicated constructs on either TRFs or TRFs expressing the tetR are shown. C) Genetic instability of the genomic region surrounding the Rh13.1. Top: The position and relative frequencies of single nucleotide changes were determined by NGS within the genomes of FL-RhCMV/Rh13.1apt after two passages in vitro in the presence or absence of tetracycline. The lower panel depicts the positions of deletions/insertions of multiple nucleotides. Frequencies for each deletion are given as percentages of all reads analyzed. Since the short reads generated by the Illumina platform do not cover the entire Rh13.1 locus it is not possible to determine which deletions co-occurred in individual viral genomes resulting in combined frequencies of >100%.
Fig 5.
A) Comparing in vitro growth kinetics of FL-RhCMV/Rh13.1/apt in fibroblasts to 68–1 and 68–1.2 RhCMV in a multistep growth curve. Primary rhesus fibroblasts were infected with 68–1, 68–1.2 and FL-RhCMV/Rh13.1/apt at an MOI of 0.01 on day 0. Cell culture supernatants were harvested on the indicated days and virus titers were determined by TCID50. Two biological repeats of each sample were titrated in duplicate and the arithmetic mean of the results were graphed. B) FL-RhCMVΔRh13.1gag, depicted above the figure, shows increased infection of epithelial cells compared to 68–1 RhCMV. Primary rhesus fibroblasts or rhesus retinal epithelial cells (RPE) were infected with MOIs of 0.3 or 10, respectively, and all experiments were performed in triplicates. After 48 hours post infection, cells were harvested, fixed, permeabilized, stained with a RhCMV specific antibody [62] and analyzed by flow cytometry. Statistical significance was shown using an unpaired t-test with a p-value significance threshold of <0.05. C) Synthetic N-terminal peptide (Peptide 1) of OR14I1 blocks PRC-positive RhCMV infection of RPE cells. Virus inoculum of 68–1, 68–1.2 or FL-RhCMVΔRh13.1gag was preincubated with peptide 1 (32.6uM) or DMSO before infection of RPE cells (MOI 4.0). On day 3 pi, cells were fixed, permeabilized, immunostained for IE2 to identify infected cells, stained for DNA (blue), and imaged (4×). Results were then quantified and plotted. Data represent the mean of n = 3 experiments ±SD. Statistical significance was determined using an unpaired t-test. D) Rh159, the RhCMV homolog of UL148, is upregulated in FL-RhCMV/Rh13.1/apt. Relative mRNA copy numbers of Rh159 (UL148) and Rh137 (UL99) were determined by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using specific probes. The data shown represent the mean of triplicate repeats (+/- SEM). Unpaired student t-tests with a p-value significance threshold of <0.05 were performed to show statistical significance in both graphs comparing FL-RhCMV/Rh13.1/apt to either 68–1 RhCMV or 68–1.2 RhCMV at 48 hpi.
Fig 6.
In vivo replication of FL-RhCMV is similar to low passage isolates in RhCMV negative animals.
A) Upper panel: Replication of RhCMV isolates UCD52, UCD59 and 180.92 in RhCMV seronegative RM. Plasma, saliva, and urine RhCMV DNA loads in three RhCMV-seronegative pregnant female RM inoculated i.v. with 2 x 106 TCID50 RhCMV 180.92, 1x106 PFU RhCMV UCD52, and 1x106 PFU RhCMV UCD59 are shown. The viral loads (VL) in each RM were determined at the indicated time points by qPCR targeting exon 1 of the immediate early gene (IE) region as described previously [11,84]. Middle panel: Replication of FL-RhCMV/Rh13.1apt in RhCMV-seronegative RM. 1.79x106 PFU of FL-RhCMV/Rh13.1apt were inoculated i.v. into three RhCMV-seronegative male RM and the VL was determined on the indicated days by qPCR using the same primer/probe set as described above. Lower panel: The VL for all animals included in the previous panels (low passage RhCMV isolates in red and FL-RhCMV/Rh13.1apt in black) are shown in direct comparison. The data indicate that FL-RhCMV/Rh13.1apt can induce a VL comparable to commonly used virulent RhCMV isolates. B) VL in plasma, saliva, and urine were determined in the same animals shown in the middle panel of A) using qPCR primer/probe sets specific for Rh13.1 (upper panel) or Rh157.5 (lower panel) shown in green and in blue, respectively. The data indicates the presence of both genes in FL-RhCMV/Rh13.1apt at 65 days post infection while both genes are rapidly selected against during in vitro tissue culture on fibroblasts.
Fig 7.
Spreading of FL-RhCMV in vivo.
A) Tissue genome copy numbers of FL-RhCMV. Three RhCMV-naïve RM (RM1-RM3) were inoculated with 107 PFU FL-RhCMVΔRh13.1/TB6Ag while another three RhCMV-naïve RM (RM4-RM6) were inoculated with 107 PFU of FL-RhCMVΔRh13.1/TB6AgΔRh157.4–5ΔRh158-161. The genome regions shown depict the alterations and deletions introduced into the FL-RhCMV backbone to create the constructs used in this experiment. All 6 RM were necropsied at day 14 post-infection and viral genome copy numbers per 107 cell equivalents were determined in the indicated tissues using ultra-sensitive nested qPCR specific for TB6Ag. Statistical analysis was performed using a two-sided Wilcoxon tests (unadjusted p values < 0.05) excluding all tissues at the injection site and the nearest draining lymph nodes to detect significant differences in dissemination. B) In situ immunofluorescence phenotyping of cells expressing RhCMV RNA was performed by multiplexing RNAscope in situ hybridization with antibody detection of cellular markers specific for myeloid/macrophage cells (CD68/CD163), endothelial cells (CD34), and mesenchymal cells (vimentin) in the spleen of macaques inoculated with either FL-RhCMVΔRh13.1TB6Ag (FL-RhCMV) or FL RhCMVΔRh13.1/TB6AgΔRh157.4–5ΔRh158-161. The majority of cells inoculated with the FL-RhCMV were vimentin+ CD34- CD68/CD163-, indicating they were of mesenchymal origin. C) To quantify differences in RhCMV infection and expression levels in macaques inoculated with either FL-RhCMV or FL RhCMVΔRh13.1/TB6AgΔRh157.4–5ΔRh158-161, we used three independent quantitative approaches in the HALO image analysis platform from Indica Labs: i) the percent area of the tissue occupied by infected cells, ii) the number of infected cells per 105 cells, and iii) an estimate of RhCMV viral RNA copy number per infected cell. Statistical significance was calculated using an unpaired t-test. D) Tissue distribution of TB6Ag insert–specific CD4+ and CD8+ T cell responses elicited by FL-RhCMVΔRh13.1TB6Ag versus FL RhCMVΔRh13.1/TB6AgΔRh157.4–5ΔRh158-161 vectors. Flow cytometric ICS (CD69, TNF-α and/or IFN-γ readout) was used to determine the magnitude of the CD4+ and CD8+ T cell responses to peptide mixes corresponding to the six Mtb antigens contained in the TB6Ag-fusion (Ag85A, ESAT-6, Rpf A, Rpf D, Rv2626, Rv3407). Mononuclear cells were isolated from the indicated tissues from three RhCMV-naïve RMs inoculated with 107 PFU FL-RhCMVΔRh13.1TB6Ag (blue bars) and three RMs inoculated with 107 PFU FL-RhCMVΔRh13.1/TB6AgΔRh157.4–5ΔRh158-161 (green bars) and all RM taken to necropsy at either 14 or 15 days post infection. Response comparisons per tissue are shown as the mean + SEM percentage of T cells specifically responding to the total of all peptide mixes (background subtracted) within the memory CD4+ or CD8+ T cell compartment for each tissue (n = 3 per tissue, unless otherwise noted by * n = 1 or † n = 2).
Fig 8.
Viral gene expression profile of FL-RhCMV in vitro and in vivo.
A) Comparison of in vitro and in vivo gene expression profiles by principal component analysis. In vitro: Rhesus fibroblasts were infected with an MOI = 5 of FLRhCMVΔRh13.1/TB6Ag and the cells were harvested at the indicated times. In vivo: RNA-was isolated from indicated tissues of RM1-RM3 described in Fig 7. Total RNA was isolated from all samples and RNAseq was performed on libraries build from polyA-fractionated RNA using an Illumina HiSeq-2500 next generation sequencer. PCA was done on the combined and quantile normalized expression matrix (see Materials and Methods). We observed that PC1 and PC2, shown herein, combined capture over 70% of total variance with distinct sets of co-regulated genes. B) In vitro and in vivo expression levels of each ORF. Expression levels were normalized between the in vitro and in vivo samples using quantile normalization (see Materials and Methods). C) For all samples analyzed in B) the ten viral ORFs showing the highest mRNA coverage after normalization for ORF size are shown. All values are given as percent of total viral reads mapping to all annotated ORFs normalized for size.
Fig 9.
Acquisition of genes and gene families in old world NHP CMVs during co-evolution with their primate hosts.
Full genome annotations of old world NHP CMVs across different species allowed for comparative genomics to identify single ORFs or entire gene families that were present in one or several CMV species but absent in others. These differences clustered in six independent loci across the genome. Examination of the phylogenetic relationship of the individual CMV species as well as their host species reveals at which point in time these gene acquisitions and gene duplications occurred. The phylogenetic tree depicted is based on the full genome alignment shown in Fig 1. The green circles indicate genetic events that took place during the evolution of each species. The blue circle represents the acquisition of RL11K, a gene duplication found in RhCMV and JaCMV but not in CyCMV. Since CyCMV appears to be more closely related to RhCMV than JaCMV by full genome alignment (see Fig 1) this appears counterintuitive. However, phylogenetic alignments of the corresponding macaque host species based on morphology [106], mitochondrial DNA data [107] or Alu elements [108] reveals that Japanese macaques (M. fuscata) and rhesus macaques (M. mulatta) speciated more recently compared to cynomolgus macaques (M. fascicularis).