Disease caused by CHV-1 is widespread in canine populations [1
] and affects both neonatal and adult animals. The virus is a significant cause of sudden death in neonates [4
], ocular disease [5
], respiratory disease [6
] and reproductive disease [8
]. Despite this, there has been minimal genomic assessment of the virus to date.
CHV-1 is a varicellovirus of the herpesviridae family, which establishes latency following an initial infection period [10
]. The virus has a worldwide distribution [12
] and a high seroprevalence among various canine populations [1
]. Diagnosis of disease is currently based on viral isolation, PCR or serology. Although symptomatic treatment of certain forms of CHV-1 (ocular disease) has progressed in recent years [15
], minimal options are available for effectively treating the disease in neonatal animals. Vaccines are available but do not offer complete protection against infection [16
]. Together, this means that disease surveillance and prevention are essential for control of this virus.
The structure of CHV-1 is similar to other herpesviruses, with unique long (UL
) and short (US
) sequences flanked by terminal (TRS
) and internal inverted (IRS
) repeats. The overall length is approximately 125 kbp with 76 open reading frames [18
]. Several complete/near-complete CHV-1 isolates have been sequenced and analyzed in the past [18
]. Phylogenetic analysis of complete CHV-1 genomes has been previously been performed using a very limited number of isolates [18
]. While most CHV-1 genomes were previously found to be mostly homogenous (0.005% distance in UK isolates), one isolate (BTU-1) obtained from a Brazilian host canid in 2012, was noted to be 0.34% distant from the others, mostly due to variants in the CHV-1 UL50
protein sequence region [20
Our objective was to characterize the genomes of circulating CHV-1 isolates in the USA and to relate their sequence characteristics to global isolates. We hypothesized that CHV-1 isolates obtained in the USA would be genomically homogenous but would separate into clades based on geography, as is the case for the related varicellovirus, feline herpesvirus 1 (FHV-1) [21
2. Materials and Methods
2.1. Cells and Viruses
Virus isolation was performed in a routine manner using diagnostic samples collected from naturally infected host canids (with informed owner consent). Sample collection was performed by vigorously swabbing the conjunctival fornices of host animals using a flocked swab, which was then immediately placed into universal viral transport media (BD, NJ) for transport. Tubes containing swabs and viral transport medium were vortexed. Approximately 1.5 cc of viral transport medium was inoculated onto cells and allowed to absorb for 1 h with periodic rocking. Isolations were performed with A-72 canine cells (ATCC, CRL-1542, VA), Madin-Darby Canine Kidney (MDCK) cells (ATCC, CRL-2935, VA), and a laboratory-developed canine skin cell line in minimum essential medium-E with 10% fetal bovine serum, 5% serum replacement solution, 2% penicillin-streptomycin solution, 1% amphotericin B solution and 1% gentamicin sulfate solution. Cultures were incubated at 37 °C (5% CO2
), checked at 24-h intervals for cytopathic effect (CPE), subcultured every 5–7 days, and held for up to 21 days. Cell cultures with CPE typical of CHV-1 were stained with anti-CHV-1 polyclonal antiserum conjugated to fluorescein isothiocyanate (CHV-1 direct fluorescent antibody conjugate, CJ-F-CHV-10ML, VMRD Inc, Pullman, WA, USA) [22
] to confirm viral identification. At this point, each flask was subjected to 3 freeze/thaw cycles before the contents were transferred to a 15 cc conical tube and centrifuged at 600× g
for 5 min at 4 °C. The supernatant was then removed and aliquoted for storage at −80 °C prior to bulk processing for sequencing.
Using a 500 µL volume of previously stored supernatant, a T25 flask (Thermo Fisher, Waltham, MA, USA) containing a monolayer of MDCK cells was subsequently infected in a similar manner for each isolate using Dulbecco’s Modified Eagle Media (DMEM) (Thermo Fisher, Waltham, MA, USA) containing 2% fetal bovine serum (Thermo Fisher, Waltham, MA, USA) and 1% penicillin/streptomycin (Thermo Fisher, Waltham, MA, USA). Following observation of 100% CPE, 3 freeze/thaw cycles and centrifugation (600× g for 5 min at 4°C), 200µL of the resulting supernatant was used for viral DNA extraction. Viral DNA was prepared using a commercial kit according to the manufacturer’s instructions (Purelink Viral RNA/DNA Mini Kit, Invitrogen, Carlsbad, CA, USA).
Extracted DNA was submitted to the Louisiana State University School of Veterinary Medicine GeneLab. DNA purity and concentration were assessed using a NanoDrop One Microvolume Spectrophotometer (Thermo Scientific, Waltham, MA, USA) and then processed using the Nextera DNA Flex Library Prep Kit (Illumina Inc., San Diego, CA, USA) with modifications specific for the 100–500 ng DNA input range. Quality and quantity of the finished libraries were assessed using a Fragment Analyzer Instrument (Advanced Analytical) and dsDNA HS Assay Kit, respectively. Libraries were pooled, standardized to 10 μM and paired end sequencing was performed using the Illumina MiSeq Sequencer and a MiSeq 500 bp (v2) sequencing kit (MS-102-2003).
2.3. Genome Assembly
Reference-based assembly was performed using Geneious Prime ver 2020.2.4. Paired end reads were trimmed using BBDuk adapter/quality trimming ver 38.84 (right end, Kmer length = 27, maximum substitutions = 1, minimum quality = 20, minimum overlap = 20, minimum length = 20). Trimmed paired end reads were then assembled to the reference sequence for CHV-1 (0194, Genbank accession NC_030117). A consensus sequence was extracted from the aligned reads with gaps filled with “N’s”. Genomes were annotated and submitted to Genbank using Geneious Prime ver 2020.2.4.
2.4. Viral Genome Alignments
Alignments were created using MAFFT ver 7.450 [23
] within Geneious Prime ver 2020.2.4. In all instances, the default parameters were used; a scoring matrix of 1 PAM/k = 2, gap penalty of 1.53 and offset value of 0.123. Alignments created included all whole CHV-1 genomes (USA + UK + Brazil + Australia), all whole CHV-1 genomes plus a Feline herpesvirus (FHV-1) outgroup (C-27, Genbank accession NC_013590.2) and the isolated CHV-1 V57
gene region from 3 CHV-1 genomes (0194, ELAL-1 and BTU-1). An additional alignment using the same 3 complete CHV-1 genomes (0194, ELAL-1 and BTU-1) was also created.
2.5. Variant Analysis
Variant analysis was performed using the Geneious variant finder (Geneious Prime ver 2020.2.4) in regions with a minimum coverage = 100, minimum variant frequency (proportion of reads matching variant) = 0.25, maximum variant p-value = 10−6 and minimum strand-bias (disagreement between the forward and reverse strand) p-value =10−5 when exceeding 65% bias. Variants were called by comparing each sequenced isolate to the reference genome (CHV-1 0194). Linear regression with viral gene length as the repressor and the total number of the variants per gene as the outcome (with 99% confidence limits) was constructed using JMP Pro 15.0.0 (SAS Institute Inc., Cary, NC, USA).
Regions of genomic distance between 3 CHV-1 isolates (0194, ELAL-1 and BTU-1) were visualized using RDP ver 4.100 [24
] using the manual distance plot function (window = 1200, step = 500, transition:transversion rate ratio = 2, coefficient of variation = 1, Jin and Nei model [25
2.6. Phylogenetic and Recombination Analysis
An alignment containing all whole CHV-1 genomes plus a Feline herpesvirus (FHV-1) outgroup was processed using ModelFinder [26
] via IQ-Tree 2 ver 1.6.12 [27
] to automatically select the best-fit model (TVM+F+G4). The resultant maximum likelihood tree was viewed using Splitstree ver 4.16.1 [28
]. Recombination analysis was performed using RDP ver 4.100 [24
] on an aligned set of CHV-1 genomes using a manual bootscan (window = 1200, step = 500, replicates = 100, 70% cutoff, Jin and Nei model [25
]), RDP, GENECONV, MaxChi and Chimaera.
Pairwise genomic distances were calculated using MEGAX [29
] with the gamma distribution model (5), partial deletion of gaps and 1000 bootstrap replicates.
2.7. Sequence Accession Numbers
To the authors’ knowledge, the present study represents the first known description of the genomics and phylogeny of CHV-1 isolates obtained from USA-based host canids. Overall, most of the isolate genomes studied were very similar to previously sequenced isolates from the UK and Australia, with a low overall genomic distance (0.09%). Two isolates were notably different from the others and formed a clear separate clade: BTU-1 (isolated in 2012 from a host canid in Brazil) and ELAL-1 (isolated in 2019 from a host canid in TX, USA). Herein, we have described evidence for trans-boundary transmission of this virus in canid populations; ELAL-1 is very likely to have originated from a recombination event involving BTU-1. This discovery provides critical information for our collective understanding of the transmission of the virus and may play a future role in surveillance and control as the availability of whole viral genome sequencing increases. Despite robust import controls in the USA for live dogs and canine semen, there are presently no steps in place to test for CHV-1. Using a simple phylogeny assessment, we were able to create a visualization of the relationship between the CHV-1 isolates (Figure 2
), which confirmed that isolates obtained from multiple host animals in the same outbreak in New York State (Figure 2
b) were extremely closely related. Future sequenced isolates can be considered within this framework to determine likely phylogeny. Aside from this example of geography influencing clade organization, this effect appears to be much less pronounced for CHV-1 than for comparable viruses such as FHV-1, where geography appears to be a strong determinant of clade organization [21
]. The high degree of similarity between the majority of CHV-1 genomes (Clade 1, Figure 2
a) from host animals housed in a variety of geographic locations provides additional evidence for trans-boundary spread of this pathogen.
In the present study, two CHV-1 clades were detected by phylogenetic analysis. This is in agreement with previous analysis of 4 CHV-1 isolates; BTU-1 (Clade 2), 0194 (Clade 1), V777 (Clade 1) and V1154 (Clade 1) [20
]. This previous work determined that the UL50
gene of BTU-1 was 12.2% distant from the remaining isolates, hypothesized to be the result of a recombination event between CHV-1 and an unknown virus [20
]. In the present study, the genome of ELAL-1 was found to be very similar to BTU-1 with one exception in the V57
gene region. For comparison, the V57
gene region of ELAL-1 was approximately 1.6% distant from both BTU-1 and 0194. BLAST searches (blast.ncbi.nlm.nih.gov
) (Table S3
) confirmed the closest identity of ELAL-1 V57
to various other CHV-1 isolates, suggesting that positive selection is likely to be the underlying cause of this variation.
As most of the isolates were found to have homogenous genomes, it is unsurprising that variant detection (compared to 0194, the reference genome) yielded a modest number of results for each isolate. The exception was for ELAL-1, which contained 188 synonymous variants and 154 non-synonymous variants. For comparison, our set of CHV-1 isolates contained more variants than FHV-1 [21
] but less than herpes simplex virus (HSV) [32
]. As expected, most variants were found in larger regions of the CHV-1 genome (Table 3
, Figure S2
). It can be seen from the regression analysis in Figure S2
that a single gene (UL50
, Deoxyuridine triphosphate) had a higher-than-expected total number of variants (53) relative to the gene size (918 bp). This gene had a relatively low number of variants in prior analyses of FHV-1 [21
] and HSV-1 [34
] and it is therefore unclear why this would be the case for CHV-1. Of possible clinical relevance is that one isolate (ELAL-1) contained variants in UL23
, all regions targeted by antiviral medications [30
]. Further investigation is necessary to determine if these variants have any impact on antiviral susceptibility. The host canid from which ELAL-1 was obtained had been treated by a veterinarian with multiple ocular medications for disease related to CHV-1 infection including topical ocular idoxuridine, without significant improvement of clinical signs. Following cessation of idoxuridine and initiation of topical ocular trifluridine [35
], the ocular clinical signs improved. Antiviral resistance has been previously documented for HSV [36
] but not for CHV-1.
The G-C content across the CHV-1 genomes was consistent at 31.6%. As has been previously noted [18
], this is the lowest for known varicelloviruses. For comparison, the G-C content of FHV-1 is 45.8% and 72.6% for bovine herpes virus type 1 [20
]. Sequencing coverage was excellent for CHV-1, and this is likely related to the consistently low G-C content across the genome for this virus [32
]. It has been previously suggested that ungulate varicelloviruses, which have a higher G-C content, have a greater degree of intraspecies distance than those from non-ungulates (such as CHV-1). Our results fit within this framework, with a low overall distance of 0.09% for the 20 CHV-1 genomes analyzed.
We detected evidence of recombination between multiple isolates, a process which is thought to be common in herpesviruses [20
]. In line with what has been previously reported, we did not detect any evidence of recombination between clades. By monitoring future occurrences of isolates from Clade 2, this should provide a useful method to monitor the spread of CHV-1 in the USA canid population. Recombination seems to be subjectively less prevalent in this sample of CHV-1 isolates compared to viruses such as FHV-1 [21
] and HSV [32
]. The reason for this is unknown.
All the CHV-1 isolates sequenced for the present study were obtained from the conjunctiva of animals with ocular disease (conjunctivitis and/or keratitis). Similar to other herpesviruses, CHV-1 establishes latency and is periodically excreted at various mucosal sites such as the conjunctiva, oral cavity and genitalia [40
]. It is possible that the body sampling site influenced characteristics of the viral genomes which we obtained. This is considered unlikely given that many of the genomes from ocular isolates appeared to be very similar to those obtained from other tissues such as kidney, liver and lung.