Animal experiments and samples

Four individual animal experiments using 100–150 kg Holstein Friesian calves were performed. For each experiment, two animals (referred to later as inoculates) were selected at random, and were needle challenged intradermolingually (1) with 1 × 10^5.7 TCID₅₀ of cattle adapted FMDV O UKG 34/2001. Forty eight hours after challenge naïve animals (“donors”, 2 animals for each of 4 experiments) were introduced to the inoculates and were challenged by direct contact exposure for 24 h. The inoculates were removed from the study and the animals exposed to infection (n = 8) were used to attempt transmission to further naive cattle (“recipients”) at two, four, six and eight (in experiments 3 and 4 only) days post infection for a period of 8 h each time. A total of 28 recipients were used in this study design.

Individual donors were examined daily for the presence of clinical signs (lesions in the mouth, tongue, snout, feet and the presence of nasal discharge). Rectal temperature was also recorded. Blood and nasal fluid samples were taken daily for the first 8 days following challenge and then every other day for up to 14 days after challenge. The samples were transferred immediately to the laboratory for processing; nasal fluid was stored at −80°C and heparinised blood aliquoted and stored for subsequent virus detection in bovine thyroid cells (BTY) culture as described earlier [19]. Aliquots of serum were stored at −80°C for subsequent total antibody (Ab) detection, nucleic acid extraction and analysis by qRT-PCR. Samples of oropharyngeal fluid (OPF) were collected by probang cup from all the animals before challenge and thereafter from the donors daily for the first week and fourteen days after challenge. All probang samples were stored at −80°C for subsequent virus detection using BTY cell culture and genome detection using real-time quantitative PCR.

Several air samples using multiple devices were collected simultaneously, each hour, during all but 2 of the 28 challenge periods. Air samples were collected using an all-glass Cyclone sampler (operated for 5 min at a flow rate of approximately 390l/min) and an all glass Porton impinger sampler (operated for 5 min at a flow rate of 11l/min). These sampling periods were determined as the optimal sampling configurations for the instruments [20–22]. The collecting media employed in these samplers was modified eagle’s medium (MEM) -HEPES with antibiotics and 0.1% (w/v) BSA [16, 23]. The concentration of virus per litre of air was determined by endpoint titration for each particular air-sample, which was multiplied by the volume of the collecting fluid and the flow of the sampler. The amount of infectivity recovered was expressed as the total amount (50% tissue culture infectious dose, TCID₅₀) of airborne FMDV per animal per challenge period (8 hrs).

Virus detection

Live virus was detected in the biological samples collected (heparinised blood, nasal swabs, OPF) and in the collecting media from the air samples using BTY primary cell cultures [16, 23, 24]. Given the large number of samples taken, they were first screened to determine the presence of virus, and then a tenfold dilution series of those verified to be virus positive were made and each dilution inoculated onto five BTY tubes. Titres were calculated by the Karber equation according to Lennette [25]. The specificity of any cytopathic effect was confirmed by an antigen capture ELISA [26–28].

Viral nucleic acid purification

RNA (200 μL of sample, mixed with 300 μL of MagNA Pure LC total nucleic acid Lysis/Binding Buffer) was extracted using the MagNA Pure LC total nucleic acid isolation kit (Roche, UK) and an automated nucleic acid robotic workstation according to the manufacturer’s instruction (MagNA Pure LC, Roche, UK). The samples were eluted in a volume of 50 μL and stored at −80°C for later analysis.

FMDV RNA standards and reverse transcription

As these experiments were performed with the O UKG 34/01 isolate, a homologous RNA standard was used (synthesized in vitro from a plasmid containing a 500 base pair insert of the internal ribosomal entry site (IRES) of FMDV O UKG 34/01) as described by Quan et al. [29]. The reverse transcription was performed as previously described [29–31].

Quantitative RT-PCR

To determine the amount of FMDV RNA in extracts of the total nucleic acid from blood, OPF, and nasal fluid a qRT-PCR was performed according to the methodology previously described [30, 32]. In the PCR reaction, primers SA-UK-IRES-248F (50-AAC CAC TGG TGA CAG GCT AAG G-30)/SA-UK-IRES-308R (50-CCG AGT GTC GCG TGT AC CT-30) and a UK-IRES- 271T (6-FAM-TGC CCT TTA GGT ACC C-MGB) TaqMan® Minor Groove Binding probe (Applied Biosystems) were used, as this primer/probe set was designed for optimum detection of the FMDV O UKG 2001 virus [29]. The PCR was performed on a Stratagene® MX3005P™ QPCR system using MXPro-MX3005 v 3.00 Build 311 software (Stratagene, UK), and fifty PCR cycles were carried out. Once obtained, results and amplification plots were analysed and standard curves constructed from cycle threshold values [32–34] of the RNA standard dilutions, to provide a measure of the number of FMDV genome copies.

Assay for FMDV specific antibodies

Serum samples were tested for the presence of antibodies to FMDV using a liquid phase blocking ELISA (LPBE) [35, 36].

Assay for interferon detection

Type-I interferon (IFN) biological activity was measured in serum samples from donor animals by using an Mx/chloramphenicol acetyltransferase (Mx/CAT) reporter gene assay [37].

Database management and statistical analysis

Identification of the predictors of the onset of clinical signs was performed using a Generalised Linear Mixed Model (GLMM). The onset of clinical signs (as opposed to transmission) was chosen as the response variable as it was previously shown that there is a significant association between transmission and the onset of clinical signs [19] (see Figure 1). Onset of clinical signs refers to whether or not the donor cow developed clinical signs on the given day. These clinical observations were confirmed later by detection of live virus from the recipient. No subclinical infections were observed.

Only data up to and including day 8 were used in the statistical analyses as after day 8 data was not collected daily. In addition, all donor animals had exhibited clinical signs by day 8. To account for the repeated measures structure in the data (i.e. daily sampling) analyses using multiple explanatory variables were performed using a GLMM with covariance pattern model (Proc Glimmix SAS version 9.3, SAS Institute Inc., Cary, NC). The GLMM was fitted with a binomial distribution and a logit link function. The response variable was the presence or absence of any clinical signs. Donor was fitted as the sole random effect, other variables were fitted as fixed effects. Given the small sample size (n=8 donors) only single factor models were generated. In total 63 variables were analysed.

Diagnostics were performed and plots of residuals were examined, confirming the goodness-of-fit of each model. Odds ratios (OR) and their associated 95% confidence intervals were estimated in the final models for factors statistically significantly associated with the onset of clinical signs. The generalised chi-square/DF (χ²/DF) was used to compare the fit of different models. Unfortunately no significance test of model fit is available, however, the χ²/DF can identify models that given their fixed and random effect specifications more (or less) closely meet the specified distribution (binomial logit). The closer the χ²/DF is to unity the better the model and data meet the assumed residual distribution.

Virus in the blood and OPF was measured using both the virus isolation method (from heparinised blood samples) and the qRT-PCR method (from serum samples). The relationship between the two methods was examined by looking at the temporal trends using PROC LOESS (SAS version 9.3). LOESS is a nonparametric method for estimating local regression surfaces. In addition, agreement between the two methods was examined using Cohen’s kappa (StatXact v.8, Cytel Software Corp, Cambridge, MA, USA).

Preclinical predictors were identified in the data by examining the virological and immunological variables in the time frame surrounding the onset of clinical signs. Data from ±4 days from the onset of clinical signs for each donor animal were included.

Prior to analysis, it was specified that results with p <0.05 would be reported as exhibiting formal statistical significance.

Associations with transmission

As previously described [19] there were 28 attempts to transmit the disease from donor to recipient animals over the 8h periods, 8 (29%) of which were successful in transmitting disease. Six of these transmissions occurred on either day 4 or day 6 since exposure of the donor to the virus. Only one transmission event occurred on each of day 2 and day 8. One donor transmitted the disease on two occasions, days 4 and 6 post exposure to the virus. One cow failed to transmit FMDV, even though FMDV was detected in the nasal fluid (NF) and oesophageal-pharyngeal fluid (OPF), although not the blood.

Transmission was significantly associated (Fisher’s exact = 6.16; p = 0.021) with the onset of clinical signs (Figure 1). The peak at time 0, illustrates that for most cows transmission occurs on the same day as clinical signs appear. Only one transmission event occurred prior to the onset of clinical signs. However, this animal showed overt signs the next morning, approximately 16 h after the end of the successful challenge period. The clinical signs observed were varied: nasal discharge and lesions in the mouth or tongue were the most frequently reported “first” signs (Figure 1).

Cut-off levels

Receiver Operating Characteristic (ROC) curve analysis was used to determine the best cut-off for the virological variables measured in this study (Figure 2). The cut-off chosen was one that maximized the sum of the sensitivity and specificity. The cut-off value for each variable is shown in Figure 2. The cut-off values for blood VI, blood qRT-PCR, OPF VI and OPF qRT-OCR were 2.4 log₁₀ TCID/mL, 4.6 log₁₀ copies/mL, 5.0 log₁₀ TCID₅₀/mL and 6.5 log₁₀ copies/ml respectively.

Predictors of transmission and the onset of clinical signs

VI vs qRT-PCR measurements

The pattern of virus detection with time (days) was similar for both the qRT-PCR and the virus isolation (VI) methods (Figure 3). Interestingly, virus in the OPF was always detected or not on a given day by both methods (Cohen’s kappa measure of agreement 1.00 (95% CIs 1.00-1.00)). However, agreement for detection of virus in the blood was lower (Cohen’s kappa 0.77 (95% CIs 0.62-0.93)). In general, in the blood low levels of virus were detected early post infection using the VI method when there was no virus detected using the qRT-PCR method. This may represent cell-associated virus. Viral genomes were detected using the qRT-PCR method late in study period, after there was no longer any virus detected using the VI method.

Pre-clinical predictors

Figure 4 is a violin plot representing the appearance of various predictors with reference to the time (days) since the onset of clinical signs. The thicker the line the more data within a given category are within the time frame specified. Very early in the infectious process OPF VI and OPF qRT-PCR were detectable whereas virus was not detected in the blood and nasal fluid until 2–3 days before the onset of clinical signs. Antibodies were not detected until 2–3 days after the onset of clinical signs.

In summary, virus is first detectable in the OPF, but detection of virus in the blood and nasal fluid may also be good candidates for preclinical indicators. Interestingly, the donor that did not transmit in this study never had any measurable amount of virus in the blood. However, virus was detectable in the OPF and nasal fluid.

Temperature

A change in temperature was significantly associated with the onset of clinical signs (GLMM, p < 0.05). On the day that clinical signs appear temperature increases by an average of 1°C (Figure 5). Average temperature on the day that clinical signs appear was 39.6°C (95% CIs, 38.9°C-40.2°C) whereas the day before it was 38.6°C (95% CIs, 38.3°C-38.8°C).