Twenty-two wild raccoons were captured by OMNR staff in southern Ontario from regions that had not been exposed to RRV (lack of prior exposure to rabies was confirmed through serology tests). These animals were transported to a biocontainment level 3 animal care facility at the CFIA, in Nepean Ontario where the challenge study was conducted. All animal experimentation was performed according to the guidelines of the Canadian Council on Animal Care with the approval of the institute’s animal care committee. Additional details of the challenge study are taken from the doctoral thesis of Szanto [32] .

RRV stock was propagated in murine neuroblastoma cells as previously described by Szanto [32] . Two raccoons were euthanized four days prior to the start of the experiment to serve as negative controls. Twenty raccoons were inoculated into the right hind leg muscle with 0.5 ml of virus stock (10^6.7 TCID₅₀). Raccoons were housed individually and monitored daily for clinical signs of rabies. Two raccoons on each of 2, 4, 8, 15 and 32 days post infection (dpi) were chosen at random and euthanized (test group of 10 animals). The remaining 10 animals were kept under observation until they exhibited clinical signs of rabies, at which point they were euthanized. Within this group six animals were euthanized at days 20, 21, and 23; the remaining four animals did not exhibit any clinical signs of the disease, but were euthanized at the end of the experiment on day 533 pi. Tissues collected from all animals included: muscle tissue from the inoculated right leg muscle, a section of lumbar spinal cord, cerebral cortex and salivary gland. Samples were immediately frozen and stored at −80˚C until use.

Taqman probes to target each individual RRV gene were designed using Custom TaqMan® Expression Assays using previously published RRV genome as the template (GenBank accession # EU311738; Applied Biosystems; Table 1).

Total RNA was extracted from all samples using TRIzol (Invitrogen) following the manufacturer’s standard protocol. Precipitated RNA was dissolved in 50 µl of DEPC-treated water and quantified using a NanoDrop^TM 8000 (Thermo Scientific) spectrophotometer. DNA was removed by treatment with Turbo DNase enzyme (Applied Biosystems) following the manufacturer’s protocol. RNA quality and quantity was assessed using gel electrophoresis and the NanoDrop^TM 8000 (Thermo Scientific) spectrophotometer.

DNA contamination was assessed through attempted amplification of mitochondrial DNA cytochrome b gene (4680F-TGACTTGAARAACCAYCGTTG; 4682R-CCCTCAGAATGATATTTGTCCTCA). PCR was prepared with components in the following concentrations: 1X PCR buffer (Invitrogen), 0.2 mM of each dNTP, 1.5 mM MgCl₂, 0.2 µM of forward primer, 0.2 µM of reverse primer, 0.6 U of Taq DNA polymerase (Invitrogen), 10 ng of RNA and ddH₂O as needed for a total volume of 12 µl. Amplification conditions were as follows: initial denaturation at 94˚C for 5 min; [denaturation at 94˚C for 30 sec, annealing at 55˚C for 1 min, extension at 72˚C for 1 min] × 35 cycles; final extension at 60˚C for 45 min; and 4˚C hold. 2 µl of the reaction was subjected to agarose gel electrophoresis to assess DNA contamination.

To determine the expression levels of the selected genes, RT-qPCR was performed on RNA extracts. 1 µl of purified RNA was used in a 20 µl reverse transcription reaction using High Capacity reverse transcriptase en-

^*Probes were labelled with FAM at 5’ and NFQ at 3’.

zyme (Applied Biosystems) following the manufacturer’s protocol. Gene specific primers listed in Table 1 were used to prime cDNA synthesis. TaqMan® PCR was performed on a Step One Plus™ instrument, using TaqMan® Fast Universal PCR Master Mix (Applied Biosystems) according to the manufacturer’s conditions. All experimental samples contained at least three technical replicates with no template controls. The transcripts of N gene were normalized against β-actin; whereas the other four viral genes were normalized against N gene and expression levels were calculated using the 2^−ΔΔCt method [36] . Statistical significance was determined by student’s t-test, with P-values of <0.05 considered statistically significant.

RNA from the brain of a raccoon that showed clinical symptoms after the challenge was utilized (sample 20 + 2; see Table 2) to prepare a cDNA library for clonal, next generation sequencing. Approximately, one µg of RNA was used to prepare cDNA using the Ovation RNA-Seq system (Nugen) following the manufacturer’s protocol.

^aViral RNA detected by qPCR < 45 CT.

Amplified cDNA was purified using a QIAquick PCR Purification Kit (Qiagen) as per the manufacturer’s instructions. The cDNA library was prepared using protocols described by a Roche Diagnostics application note published on their webpage (Roche Diagnostics). The resulting cDNA library was sequenced using 454 pyrosequencing on a Roche GS Junior platform following the manufacturer’s instructions. The bioinformatics were performed on GS De Novo Assembler and GS Reference Mapper (Roche) using previously published RRV genome (GenBank accession #EU311738) as reference.

The RRV nucleoprotein and glycoprotein were amplified in the samples that the virus was detected in using primers outlined in Table 1. PCR was prepared with components in the following concentrations: 1X PCR buffer (Invitrogen), 0.2 mM of each dNTP, 1.5 mM MgCl₂, 0.2 µM of forward primer, 0.2 µM of reverse primer, 0.6 U of Taq DNA polymerase (Invitrogen), 10 ng of cDNA and ddH₂O as needed for a total volume of 12 µl. Amplification conditions were as follows: initial denaturation at 94˚C for 5 min; [denaturation at 94˚C for 30 sec, annealing at temperatures listed in Table 2, for 1 min, extension at 72˚C for 1 min] × 35 cycles; final extension at 60˚C for 45 min; and 4˚C hold. 2 µl of the reaction was subjected to agarose gel electrophoresis to visualize the fragment amplification. PCR products were purified using ExoSAP (NEB) following the manufacturer’s instructions. Amplicons were sequenced in both directions using BigDye® Terminator v3.1 Cycle Sequencing Kit (ABI). Sequenced PCR products were ethanol precipitated, then electrophoresed and visualized on an ABI™ 3730 DNA Analyser. Sequences were edited and aligned to the RRV genome (GenBank accession #EU311738) using MEGA™ 4.1 [37] .

Structure of RV was obtained from the RCSB protein databank (2GTT). Structural predictions were performed on a SWISS-MODEL using automated mode for both the wildtype and mutant [38] [39] . The structural stability was assessed on SDM (a server for predicting effects of mutations on protein stability and malfunction: [40] ).

At the site of inoculation viral RNA was detected in two of raccoons, one that showed clinical symptoms, and another that was euthanized at 8 dpi (Table 2).Viral RNA was not present in the lymph nodes of any of the samples. RRV was detected in spinal cords of all the animals that showed clinical symptoms and in one of the raccoons that was euthanized at 8 dpi. Viral RNA was detected in the brains of eight raccoons after the peripheral inoculation (Table 2). Viral RNA was found in all the samples that showed clinical symptoms, it was also detected in a racoon that was euthanized at eight days and it was also present in a one raccoon which survived the experiment (euthanized 533 dpi. with no symptoms of RRV: Table 2). RRV was also detected in the salivary glands of two the samples that showed clinical symptoms (Table 2). All results described herein focus only on samples in which viral RNA was detected.

All five genes were detected in two samples at the site of inoculation, a sample euthanized at eight days and a sample that showed clinical symptoms at 22 dpi (Figure 1(a); not statistically significant). All five viral genes were expressed at the spinal cord in seven of the raccoons. Transcripts of P gene were higher than N gene in all the samples (Figure 1(b)), and transcripts of G and L genes were lower (8 and 66 fold, respectively) compared to N gene (Figure 1(b)). The levels of transcripts were similar between the animal that was euthanized at 8 dpi and the animals that showed clinical symptoms. At the cerebral cortex, N, P and M genes were expressed in a sequential manner in the samples (Figure 1(c)). Transcripts of G and L genes were only detected in the raccoon that was euthanized at 8 dpi and all the animals that exhibited clinical symptoms from 20 - 24 dpi; they were not detected in the animals that survived the challenge (533 + 2). Levels of transcripts were similar in the individual that was euthanized at 8 dpi and samples that showed clinical symptoms with the exception of M gene, which was expressed at higher levels in the samples that showed clinical symptoms (Figure 1(c)). Although N, P and M were present in the sample that survived the challenge (533 + 2), the transcripts of these genes were much lower when compared to other samples.

Expression patterns of viral genes were examined temporally within an individual 22 dpi (22 + 1). At the site of inoculation the N and P genes were barely detectable (2^?ΔΔCt < 0.05; Figure 2). Transcripts of all five genes increased at the spinal cord and cerebral cortex, however transcript levels of N, P and M were higher in the spinal cord compared to the cerebral cortex. In contrast, the level of expression of G and L genes increased in the cerebral cortex when compared to the spinal cord. From the salivary gland samples, only the N gene was detected at very low levels (data not shown).

Next generation, clonal sequencing was employed to assess viral variants present within a single rabid raccoon. From a single run on the 454 GS Junior (Roche) platform we recovered 15,800 reads that were mapped to the reference RRV sequence (GenBank accession #EU311738) providing 150 × coverage of the virus. When compared to the reference RRV sequence, 42 variable sites were found, with 17 synonymous substitutions and 26 non-synonymous substitutions (Table 3). The N gene had the greatest amount of diversity (0.51%), with 7 variable sites, 4 synonymous substitutions and 3 non-synonymous substitutions. Four variable sites were present at the P gene (0.44%) with 1 synonymous substitution and 3 non-synonymous substitutions. M gene had 2 variable sites (0.32%) with both of them being non-synonymous substitutions. The G gene had the least amount of diversity (0.12%) with 2 variable sites (both synonymous substitutions). The L gene had 26 variable sites (0.41%),

with 8 synonymous substitutions and 18 non-synonymous substitutions.

N and G genes were amplified in all the samples that RRV was detected in. In the N gene, there was one non-synonymous substitution at position 412 in the all the samples that showed clinical symptoms of RRV (Figure 3). This variation was not present in the raccoon euthanized at 8 dpi or the animal that showed no symptoms for 533 dpi. Although the G gene was amplified and sequenced in all the samples that RRV was detected in, there were no variable sites present.

We have shown that RRV genes are expressed in a sequential manner in the challenge study samples (Figure 1). Rabies is a slow replicating virus that aims to preserve host cell function and relies on other mechanisms, rather than speed alone, to overcome host defences [42] -[44] . Our results corroborate previous findings where expression of viral genes increased with progression of disease (Figure 1). Several studies have illustrated that glycoprotein is a major contributor to rabies pathogenicity, and one of the identified mechanisms using mouse models is the tight regulation of glycoprotein expression [45] -[47] . In line with these findings, our study found expression of glycoprotein was minimal until the virus reached the cerebral cortex. This low expression may prevent apoptosis by maintaining structural integrity of neurons and enabling the virus to reach the CNS [17] [46] .

Clinical outcome of viral infection depends on the balance between viral replication and host response. At-

^aIndicates the percentage of the sequences that contains the variant.

tenuation of rabies is usually correlated with increased replication and transcription, as well as activation of the immune response in the host [43] [45] -[47] . In one of the samples that that was euthanized at eight dpi, viral genes were expressed at similar levels to samples that showed clinical symptoms between 20 - 24 dpi. This suggests that the virus may have been attenuated in this particular sample, hence the increased replication and transcription that was observed. It is interesting to note that genes associated with the interferon pathway, as well as inflammatory cytokines, and viral receptors were also expressed in this particular sample [14] . When comparing results of this study, to Srithayakumar et al. [14] , it is evident that the presence of the virus was vital for the initiation of the immune response. In this study, we found viral genes to be expressed at low levels at the site of inoculation in two of the samples (Figure 1(a)), however in the same samples the earliest time point to detect the genes indicative of the innate immune response was in the spinal cord and brain. From the expression patterns noted in this study, and the immune response data obtained from Srithayakumar et al. [14] , there is a weak correlation between the expression of glycoprotein and the initiation of the immune response.

All raccoons in this study were infected with the same stock of RRV; however, variability was noted in both viral expression and immune response between samples. Although host genetic factors may have contributed to these differences, genetic properties of the virus will also have an impact. Given the short generation time and high mutation rates of RNA viruses, viral populations are not comprised of a single variant, but rather a complex of non-identical variants termed, quasispecies [48] -[50] . These variants provide an adaptive advantage in new environments and challenges during infection [48] -[50] . We examined the quasispecies present within an individual using next generation sequencing technology, which allowed for in depth sequencing of the variants present with a 150× coverage of the genome. A study by Khawplod et al. [51] , examined the diversity of rabies at two fragments of glycoprotein using cloning and Sanger sequencing (653 bp and 1056 bp, respectively). Khawplod et al. [51] found that from within a single rabid dog [51] , there were 57 base substitutions at the ~1700 bp region analyzed, whereas we only found 42 variable sites within the whole genome of RRV, with only two synonymous substitutions at the glycoprotein (Table 3). Although several factors such as high PCR cycles and artifacts inherent to clonal methods may have introduced artificial variants, the number of substitutions we found is still much lower in comparison. The amount of time the virus was replicating within an individual will also have an effect on the amount of variability. For the sample utilized in the present study time of replication was less than 22 days (given that the sample was euthanized 22 dpi), whereas the duration of rabies prior to death is not known for the sample utilized by Khawplod et al. [51] , suggesting that the duration of replicating virus may have contributed to the number of mutations.

Although the number of mutations we found was low compared to Khawplod et al. [51] , that may not have had an impact on the virulence of RRV. Vignuzzi et al. [52] examined quasispecies diversity in poliovirus and found that the greater amount of sequence diversity present within the viral subpopulation does not enhance virulence. Instead they suggest a more complex model where there is a positive interaction between the variants within the population, with different variants having advantages at different stages of infection. For example, certain variants may facilitate viral replication at local neurons, whereas another set of variants may be advantages in the axonal transport, another set of variants may have a role in evading the host’s immune system, and yet another subpopulation of the virus may facilitate entry to the CNS. Therefore, maintaining the complex quasispecies enables the virus to spread systematically by complementing functions of the different viral variants to successfully infect the host, suggesting quasispecies diversity, rather than individual mutations, enhances pathogenicity [52] . Having too many mutations in the genome can lead the population to viral extinction; whereas, having few variants create “bottlenecks”, rendering the virus unable to adapt to changing environments, such as replication in different tissues [52] [53] . These findings suggest that viral mutation is fine-tuned and optimized during the evolution of the virus, which facilitates adaptation to changing environments while maintaining the number of optimal variants.

In order to assess the functionality of the variants present within an individual we examined if any of the variable sites overlapped with any known catalytic domains of RV. Of the 42 variable sites, only one site on the nucleoprotein (Pos. 412) was located on a T helper cell (Th) epitope [54] -[56] . This mutation was present in 98% of the variants present suggesting that this mutation was being selected for in the brain of the sample that was showing clinical symptoms. At this position, histidine was replaced by asparagine (H412N), which may have a consequence on the protein structure as these two amino acids have different properties [57] . Examination of the nucleoprotein structure revealed that this position is exposed, suggesting this change will have an impact on the functionality of the protein. Further, structural prediction of the chain this position is located on reveals that this mutation destabilizes the protein (ΔΔG = −1.96), indicating that this mutation may have an effect on the pathogenicity of RRV. Evaluation of nucleoprotein sequences in all raccoons where viral RNA was detected, revealed that this mutation was only present in samples that showed clinical symptoms at 20 - 24 days post inoculation. As such, these data suggest this mutation provides the virus with an advantage to establish an infection in the brain. Although results of this study suggest that mutation at position 412 may be a molecular determinant for pathogenicity of RRV in raccoons, future studies should examine the x-ray crystal structure of the virus with the mutation to assess the stability of the virus and the functional consequence of this mutation.