Materials and Methods

In vitro experiments—Confluent monolayers of CRFK cells were cultured in Dulbecco modified Eagle medium^a with 10% fetal calf serum on 12-well tissue culture plates.^b The medium was then aspirated by vacuum, and a dry swab^c or cytology brush^d was brushed vigorously against all parts of the bottom of each well. Twelve replicates of this experiment were performed. Swab and brush samples were processed via 1 of 2 techniques. Half of the swab samples (n = 6) and half of the cytology brush samples (6) were placed individually into 1.5-mL microcentrifuge tubes with 400 ML of PBS solution at room temperature (approx 20° to 25°C). The swabs and cytology brushes were oriented with their tips downward; the protruding swab handles were broken off by bending, and the brush handles were cut with scissors. After each tube was pulse-vortexed for 15 seconds, swabs or brushes were inverted so that the tips pointed upward. Flame-sterilized tissue thumb forceps were used to manipulate the swabs or brushes. Tubes were centrifuged at 20,800 × g for 1 minute, and each swab or brush, along with the supernatant, was carefully removed and discarded without disturbing the pellet. The DNA was then extracted from the pellet by use of a commercial kit^e according to the manufacturer's guidelines.²² The remaining swab samples (n = 6) and cytology brush samples (6) were handled in an identical manner except that the microcentrifuge tubes did not contain PBS solution (dry-tube collection); initial DNA extraction was performed with the swab or cytology brush remaining in the tube without being inverted, and the sampling instrument was removed just prior to transfer of the entire lysate to the DNA elution column. The dsDNA concentration in each sample from all 4 sample groups was determined via absorption spectroscopy.^f A constant volume of buffer (60 ML) was used to elute the DNA from each column during the in vitro and the in vivo experiments so that for each sample, the resulting concentration was directly proportional to the total dsDNA yield.

In vivo experiments—All protocols used in this study were approved by the university animal care and use committee. Twenty cats (40 eyes) from a local humane animal shelter were examined by a single observer (HDW) and assigned a score for clinical signs of potential herpetic disease by use of a published scoring system.²³ Severity of conjunctivitis was assigned a score of 0 (none) through 3 (moderate to severe). Severity of blepharospasm was assigned a score of 0 (none) through 4 (eye completely closed). Severity of ocular discharge was assigned a score of 0 (none) through 3 (marked mucopurulent discharge). Sneezing was assigned a score of 0 (absent) or 1 (present). Severity of nasal discharge was assigned a score of 0 (none) through 3 (marked mucopurulent discharge). Clinical score for each eye was defined as the sum of all ocular (conjunctivitis, blepharospasm, and ocular discharge) and nonocular (sneezing and nasal discharge) scores. The minimum total clinical score possible was 0, and the maximum total clinical score possible was 14.

Following clinical scoring, 1 drop of 0.5% proparacaine was applied to each eye, and a dry swab and cytology brush were sequentially and vigorously rolled in the ventral conjunctival fornix of each eye of all cats. Both the eye from which a sample was collected first and the order of sampling instruments used were randomized for all cats. Swabs and brushes were collected into dry 1.5-mL microcentrifuge tubes and left in the tube during DNA extraction. Although 80 conjunctival samples were collected, 1 was lost before processing; therefore, 79 samples were assessed. After purification, the dsDNA concentration in each sample was measured in the same manner as in the in vitro experiments, and a single-phase, traditional-endpoint, 40-cycle PCR assay that targeted a 322-bp region of the FHV-1 thymidine kinase gene was performed as described¹⁰ with a DM or DV of DNA. This PCR assay reliably detects as few as 240 genomic copies of viral DNA²⁴ and is moderately sensitive for detection of this virus, compared with other FHV-1–specific PCR assays.¹⁰ The DM-PCR assay was performed first with 250 ng of dsDNA from each sample in which the extracted DNA was sufficiently concentrated (69 samples). For the remaining 10 samples, the concentration of eluted DNA was ≤ 10 ng/ML. Therefore, the volume required to deliver 250 ng of DNA exceeded the maximum volume possible for this assay. For these latter samples, the maximum possible DNA volume was used, which delivered 138 to 230 ng of DNA. Because of insufficient volume, the DV-PCR assay could not be performed in 22 samples. For each of the remaining 57 samples in which sufficient volume was available, the DV-PCR assay was performed with 23 ML of sample (the maximum possible sample volume for this assay) without regard to the mass of dsDNA that this volume contained. A commercially available PCR master mix^g and thermocycler^h were used for all PCR assays.

Statistical analysis—The concentration of dsDNA extracted during the in vitro experiments was compared among groups by use of a 1-way repeated-measures ANOVA followed by a Holm-Sidak all-pairwise multiple comparison procedure. For in vivo experiments, effect of sampling instrument on dsDNA yield was assessed by use of a Wilcoxon signed rank test. Eyes of the same cat were considered independently for all analyses. The relationship between dsDNA yield from swabs or cytology brushes and clinical signs was assessed by creating a scatterplot of the dsDNA concentrations versus clinical score for each eye. A Mann-Whitney rank sum test was used to assess the relationships between dsDNA yield and PCR result and between clinical score and PCR result for swabs and brushes, and to assess the differences in clinical score of the respective donor cats and the dsDNA concentration for samples that could be tested and those that could not be tested by use of the DV-PCR assay. A McNemar test was used to evaluate the effect of sampling instrument and type of PCR assay on FHV-1 DNA detection. Significance was set at a value of P < 0.05 for all tests.

Results

In vitro experiments—The mean concentration of dsDNA extracted from CRFK cells cultured in vitro differed significantly depending on processing method and sampling instrument. Mean dsDNA concentration for each group was calculated from 6 samples. By use of the dry-collection procedure, the mean ± SD concentrations of dsDNA extracted for swab and cytology brush samples were 197 ± 34 Mg/mL and 182 ± 18 Mg/mL, respectively; these values were not significantly (P = 0.397) different. However, by use of the PBS solution–collection procedure, the mean ± SD concentrations of dsDNA extracted for cytology brush samples (96 ± 32 Mg/mL) was significantly (P < 0.001) greater than that obtained from swab samples (17 ± 11 μg/mL). Regardless of whether a swab or brush was used, more dsDNA was extracted when the sampling instrument was collected into a dry microcentrifuge tube and left in the tube during initial DNA extraction steps (P < 0.001 for each sampling instrument). On the basis of these findings, swab and brush samples were collected into dry microcentrifuge tubes (with the swab or cytology brush remaining in the tube throughout the whole DNA extraction) in the subsequent in vivo experiments to maximize dsDNA yields.

In vivo experiments—Eighty conjunctival samples were collected from 20 cats (40 eyes); however, because 1 cytology brush sample was lost before processing, 79 conjunctival samples were assessed. Clinical scores for the 40 eyes examined differed greatly (range, 0 to 10; median, 2). Thirteen of 40 (33%) eyes were considered clinically normal (clinical score, 0), and 3 of 20 (15%) cats had unilateral ocular disease without nonocular (sneezing and nasal discharge) signs. The median concentration of extractable dsDNA obtained from swab samples (46.8 μg/mL) was significantly (P = 0.002) greater than that obtained from brush samples (28.5 μg/mL). There was no apparent correlation between clinical score and dsDNA yield, regardless of whether samples were collected by use of swabs or brushes (Figure 1).

Association between clinical disease score and concentration of total dsDNA extracted from samples collected by use of swabs (A) or cytology brushes (B) from the ventral conjunctival fornix of each eye of 20 cats. One sample collected by use of a brush was lost before processing. Severity of disease was scored as follows: conjunctivitis, 0 (none) through 3 (moderate to severe); blepharospasm, 0 (none) through 4 (eye completely closed); ocular discharge, 0 (none) through 3 (marked mucopurulent discharge); sneezing, 0 (absent) or 1 (present); and nasal discharge, 0 (none) through 3 (marked mucopurulent discharge). For each eye, the clinical score was defined as the sum of all scores; the minimum total clinical score possible was 0, and the maximum total clinical score possible was 14. No correlation between the cat's clinical score and dsDNA yield was apparent for either sampling instrument.

Citation: American Journal of Veterinary Research 69, 6; 10.2460/ajvr.69.6.811

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Association between clinical disease score and concentration of total dsDNA extracted from samples collected by use of swabs (A) or cytology brushes (B) from the ventral conjunctival fornix of each eye of 20 cats. One sample collected by use of a brush was lost before processing. Severity of disease was scored as follows: conjunctivitis, 0 (none) through 3 (moderate to severe); blepharospasm, 0 (none) through 4 (eye completely closed); ocular discharge, 0 (none) through 3 (marked mucopurulent discharge); sneezing, 0 (absent) or 1 (present); and nasal discharge, 0 (none) through 3 (marked mucopurulent discharge). For each eye, the clinical score was defined as the sum of all scores; the minimum total clinical score possible was 0, and the maximum total clinical score possible was 14. No correlation between the cat's clinical score and dsDNA yield was apparent for either sampling instrument.

Citation: American Journal of Veterinary Research 69, 6; 10.2460/ajvr.69.6.811

• Download Figure
• Download figure as PowerPoint slide

Association between clinical disease score and concentration of total dsDNA extracted from samples collected by use of swabs (A) or cytology brushes (B) from the ventral conjunctival fornix of each eye of 20 cats. One sample collected by use of a brush was lost before processing. Severity of disease was scored as follows: conjunctivitis, 0 (none) through 3 (moderate to severe); blepharospasm, 0 (none) through 4 (eye completely closed); ocular discharge, 0 (none) through 3 (marked mucopurulent discharge); sneezing, 0 (absent) or 1 (present); and nasal discharge, 0 (none) through 3 (marked mucopurulent discharge). For each eye, the clinical score was defined as the sum of all scores; the minimum total clinical score possible was 0, and the maximum total clinical score possible was 14. No correlation between the cat's clinical score and dsDNA yield was apparent for either sampling instrument.

Citation: American Journal of Veterinary Research 69, 6; 10.2460/ajvr.69.6.811

• Download Figure
• Download figure as PowerPoint slide

All 79 conjunctival samples (40 swab and 39 cytology brush samples) were assessed for the presence of FHV-1 DNA via the DM-PCR assay. However, only 57 samples (32 swab and 25 brush samples) provided sufficient volume to permit repeat assessment via the DV-PCR assay. To assess the potential bias that this may have created, the clinical disease scores of cats from which samples were collected and the sample dsDNA concentrations were each compared between those samples that were and were not assessed via the DV-PCR assay. No significant (P = 0.184) difference in median clinical score was detected between cats that yielded samples on which the DV-PCR assay could (median score, 2) or could not (median score, 1) be performed. Median dsDNA concentration (49.3 μg/mL) for samples assessed by use of the DV-PCR method was significantly (P < 0.001) greater than the value for those samples on which this assay could not be performed (13.3 μg/mL).

By use of the DM-PCR assay, there was agreement between swab and cytology brush FHV-1 DNA detection in 27 of 39 (69%) eyes. Feline herpesvirus-1 DNA was detected in only swab samples in 5 eyes and in only brush samples in 7 eyes. This difference was not significant (P = 0.77). For eyes in which FHV-1 DNA was detected only by use of 1 of the 2 sampling instruments, the first sample collected yielded positive results from 3 of 5 swabs and 4 of 7 brushes. By use of the DV-PCR assay, there was agreement between swab and brush FHV-1 DNA detection in 21 of 23 (91%) eyes. Feline herpesvirus-1 DNA was detected only by use of a swab in 1 eye and only by use of a brush in another eye. In both instances, the sample that yielded positive results was the first to be collected. There were 37 samples in which FHV-1 DNA was detected by use of both the DV- and DM-PCR assays. There were an additional 10 samples in which FHV-1 DNA was detected by use of the DV-PCR method only. There were no samples in which FHV-1 DNA was detected via the DM-PCR method but not via the DV-PCR method.

Irrespective of which PCR assay was used, samples in which FHV-1 DNA was detected were collected from cats with significantly higher median clinical scores than the cats from which samples in which FHV-1 DNA was not detected were collected (Table 1). When cytology brushes were used to collect the sample, median dsDNA concentration did not differ significantly between samples in which FHV-1 DNA was detected and those in which FHV-1 DNA was not detected, regardless of whether the DM- or DV-PCR assay was used. However, when swabs were used, median dsDNA yield from samples collected from cats in which FHV-1 DNA was detected by use of the DM-PCR assay was significantly less than the yield from samples collected from cats in which FHV-1 DNA was not detected by use of that assay. This difference was not significant when the DV-PCR assay was used. The FHV-1 DNA detection rate for the DV-PCR assay was significantly (P = 0.002) higher than the rate for the DM-PCR assay.

Discussion

Because FHV-1 is an obligate intracellular organism, we hypothesized that the number of host cells harvested during sample collection and the amount of dsDNA extracted during processing and subsequently undergoing thermocycling would influence FHV-1 PCR results. More specifically, because FHV-1 causes epithelial cytolysis and inflammatory cell infiltration,²¹ we hypothesized that more cells would be collected from cats with higher clinical disease scores. Taken together, these hypotheses suggested that there would be a correlation between clinical score, dsDNA yield, and FHV-1 PCR result. However, data generated in the present study suggest that this is not always the case. Specifically, the study results indicated no apparent correlation between clinical score and dsDNA yield. Moreover, when the DM-PCR assay was used, samples with a lower dsDNA yield were more likely to yield detectable FHV-1 DNA. Despite this, cats from which samples had detectable FHV-1 DNA had a higher median clinical score than that in cats from which samples had no detectable FHV-1 DNA.

The lack of correlation between clinical disease scores and dsDNA yield could be explained in several ways. The scoring system used in our study was not limited to signs of ocular disease, and yet only ocular samples were collected. It is possible that the clinical scores assigned to some cats reflected solely or largely nonocular (nasal) signs. Therefore, it would be possible for a cat to have a high clinical score yet have a completely healthy conjunctival epithelium. It is also possible that conjunctival swabbing does not provide an accurate sample of all cells present within the conjunctival fornix. Because of the inherent variability associated with collection of conjunctival swab samples from conscious cats, it is likely that the swabbing time, pressure, and motion varied among cats. This may have obscured important pathologic variation among samples. This is supported by the observation that dsDNA yield in our study varied widely even among samples obtained from cats without clinical signs of FHV-1–associated disease.

In the present study, detection of FHV-1 DNA via the DM-PCR assay was less likely in conjunctival samples with a higher dsDNA yield, compared with samples with a lower dsDNA yield. If this association is causal, then at a purely mechanistic level, several explanations are possible. First, the presence of FHV-1 in the conjunctival fornix of cats may be associated with impaired exfoliation of host cells. However, the fact that FHV-1 infection causes epithelial cytolysis and inflammatory cell infiltration²¹ suggests that the presence of FHV-1 would be expected to increase dsDNA yield. It is possible that data in that study²¹ differ from findings of our study because the cats in the former study were undergoing primary FHV-1 infection. Because all cats in the present study were adults, it is more likely they had recrudescent than primary herpetic disease. As described for the more chronic herpetic diseases of humans,²⁵ it is possible that cytolysis is less notable in patients with recrudescent disease (in whom immunopathogenesis predominates) than in patients with primary disease. It is also possible that maximal host cellular exfoliation or inflammatory cell infiltration occurs prior to or after peak FHV-1 replication, leading to variable relative amounts of FHV-1 and host DNA at different stages of disease. Regardless of whether the cats of the present study had primary or recrudescent disease, information regarding the duration of their clinical signs was not available. It is also possible that the increased concentration of inflammatory mediators expected within samples from cats with herpetic keratoconjunctivitis yielded false-negative PCR results²⁶ even though all samples in the present study underwent a DNA purification step that was intended to denature and eliminate such mediators. Finally, because the presence of other pathogens was not evaluated in the cats of our study, other infectious organisms may have been directly or indirectly responsible for greater host or microbial dsDNA yield in the absence of FHV-1.

It is possible that increased dsDNA concentration within a sample may reduce the chance of detecting FHV-1 DNA in conjunctival samples during processing. Because a low concentration of template DNA may decrease the sensitivity of a PCR assay and because excess template DNA can result in breakthrough bands on electrophoretic gels²⁷ or PCR primer depletion, it is common practice to vary the volume of sample in the PCR reaction in an attempt to deliver an optimum mass of DNA in relation to other reagents, especially DNA polymerase, primers, and deoxynucleoside triphosphates. Quantification of DNA via spectrophotometry cannot distinguish between viral dsDNA and host dsDNA. Thus, when viral DNA is the template, standardization of the template DNA concentration is based on the assumption that viral dsDNA concentration is proportional to the host dsDNA concentration in a sample. If that assumption is not true, dilution of the sample to provide the desired total dsDNA concentration may lead to dilution of the viral dsDNA below the assay's detection limit. To test this hypothesis, we repeated the PCR assay for FHV-1 DNA and included a DV instead of a DM of sample. If decreasing the volume of a sample to achieve a given genomic dsDNA concentration were a reason for lack of detection of FHV-1 DNA, the difference in median dsDNA concentration between samples in which FHV-1 DNA was detected and those in which it was not would accentuate that effect because, for use in the DM-PCR assay, the required volume of a sample with high dsDNA concentration is small, compared with the required volume of a sample with low dsDNA concentration. Interestingly, we did not find a significant difference in dsDNA concentration between samples in which FHV-1 DNA was detected via the DV-PCR assay and those in which it was not detected via that assay. Moreover, by use of the DV-PCR assay, FHV-1 DNA was detected in 10 additional samples, which led to a significant difference in detection rate between the 2 types of assay. This provides some evidence that viral DNA concentration is unrelated to genomic DNA concentration and that the sample for the PCR assay for FHV-1 DNA should not be standardized to a constant mass of genomic DNA, but rather to a constant volume of sample.

Although data from our study failed to reveal a difference in FHV-1 detection rate between samples collected with swabs and those collected with cytology brushes, other factors may affect the choice of sampling instrument. For example, data from our in vitro experiments suggested that more extractable dsDNA was obtained from brush samples than from swab samples when the sampling instrument was removed prior to the DNA extraction process but that there was no difference when the sampling instrument remained in the tube containing extraction medium during the extraction process. If we assume that all harvested cells are lysed and all DNA is extracted when the swab or cytology brush is left in the tube during the entire extraction process, the higher dsDNA yield obtained from brush samples, compared with the yield from swab samples, when the sampling instrument was removed prior to extraction suggests that harvested cells exfoliate more readily from brushes than from swabs. This is consistent with previous findings that the negatively charged bristles of a brush²⁸ release higher numbers of cells onto a glass slide, compared with swabs.²⁹ Therefore, if it is not possible to perform the entire extraction process with the collection instrument within the extraction medium, a cytology brush, rather than a swab, might be a better choice of sampling instrument.

By contrast, swab samples yielded a significantly greater amount of dsDNA than did cytology brush samples when the 2 sampling instruments were used to collect cells from the conjunctival fornix of cats. These differences between the in vivo and in vitro performances of the 2 sampling instruments may be attributable in part to the difference in cell types. Although CRFK cells are epithelial cells, they are highly modified as a result of multiple passages in vitro, and their adherence to the plastic substrate is almost certainly different from the complex electrostatic and physical interactions between adjacent conjunctival epithelial cells and between conjunctival epithelium and substantia propria in vivo. Additionally, sample collection technique differed between the in vitro and in vivo experiments. During the in vitro experiments, we were able to vigorously brush both sampling instruments across all parts of the bottom of each well and to ensure that the entire length of the cytology brush or swab came in contact with the CRFK cell monolayer. During the in vivo experiments, it was not possible to place an entire brush head into the fornix of a cat's eye. Moreover, it seemed that cats tolerated application of the swabs within their conjunctival fornix better than they did the application of the brushes. This hypothesis was also supported by the greater median dsDNA yield achieved by use of swabs and brushes in vitro, compared with in vivo findings. Taken together, these results suggest that sampling instrument selection should be based not only on the nature of the source of the sample, but also on the intended use of the sample.

Irrespective of instrument type, dsDNA yield was affected by extraction technique in the present study. Data from the in vitro experiments suggested that more extractable dsDNA was obtained when the swab or cytology brush sample was placed directly into a dry microcentrifuge tube and processed without manipulation. In addition to increasing dsDNA yield, this processing method offers further advantages. Because it does not require manipulation of the swab or brush during dsDNA extraction, it is associated with decreased processing time and fewer opportunities for cross-contamination among samples. Therefore, this technique was adopted for the subsequent in vivo experiments in our study.

On the basis of the findings of our study in cats, it is evident that variations in the technique used to collect and process a conjunctival sample can lead to significant variations in the final concentration of dsDNA. Although these variations did not affect the likelihood of detecting FHV-1 DNA via PCR assay, decreasing the sample volume to deliver a constant mass of genomic DNA may lead to a decrease in FHV-1 detection rate. Overall, results of the present study suggest that FHV-1 shedding from the conjunctiva of cats is best assessed by collecting a sample with a swab, placing the swab into a dry microcentrifuge tube, and not removing the swab prior to DNA extraction. A standard volume of the sample should then be evaluated via the FHV-1 PCR assay.