Effect of Spatial Separation of Pigs on Spread of Streptococcus suis Serotype 9

The spread of an infectious agent in a population can be reduced by interfering in the infectiousness or susceptibility of individuals, and/or in their contact structure. The aim of this study was to quantify the effect of prevention of direct contact between infectious and susceptible pigs on the transmission of Streptococcus suis (S. suis). In three replicate experiments, S. suis-free pigs were housed in boxes either in pairs (25 pairs) or alone (15 pigs). The distance between the boxes was ±1 m. At 7 weeks of age, one pig of each pair was inoculated intranasally with S. suis serotype 9; the other pigs were exposed to S. suis by either direct (pairs) or indirect contact (individually housed pigs). Tonsillar brush and saliva swab samples from all pigs were collected regularly for 4 weeks post inoculation to monitor colonization with S. suis. All inoculated pigs became infected, and their pen mates became colonized within 2 days. Thirteen indirectly exposed pigs became positive within 7–25 days after exposure. The rate of direct transmission βdir was estimated to be 3.58 per pig per day (95% CI: 2.29–5.60). The rate of indirect transmission increased in time, depending on the cumulative number of days pigs tested positive for the presence of S. suis. The estimate β’ind was 0.001 (95% CI: 0.0006–0.0017) new infections per pig per day for each day that an infected pig was tested positive for S. suis. We conclude that prevention of direct contact reduces the rate at which susceptible pigs become colonized. Simulation studies using these parameters showed, however, that such intervention measure would not limit S. suis serotype 9 spread in a commercial pig farm to a relevant extent, implying that spatial separation of groups op pigs within a compartment would not be effective on a farm.


Introduction
Infectious diseases are a large problem in animal husbandry and various control measures are implemented to reduce their impact, either by blocking transmission of the infection or by minimizing disease upon infection. Transmission of an infection depends on the infectiousness of infected individuals, the susceptibility of uninfected ones, and their mutual interaction [1][2][3][4]. Direct contact between infected and susceptible animals is considered to be the most important risk factor for spread of infections [5][6][7][8]. Consequently, interference in the contact structure between infected and susceptible animals, either by reducing the frequency or the intensity of the contact, might contribute to reduction of transmission of infections within a farm [2,5,6].
Separation of animals by adjustments of the housing or management systems has been applied on dairy and sheep farms to prevent transmission of pathogens from dams to offspring [5,9]. Implementation of this measure is also considered for pig farms to improve the health status [10][11][12][13]. Measures to reduce contacts between pigs, however, do not fit well in current pig management practices for various reasons. One is that it is current management practice to mix pigs to create homogeneous groups; another reason is that individual housing is not acceptable for welfare reasons. Therefore, before considering measures to reduce infection transmission by separation of pigs or groups of pigs, it should be demonstrated that such measures have a substantial effect.
Evaluation of such separation measures can be performed at different scales, e.g. at region scale, with herds as units, at herd scale, with compartments as units, or at compartment scale, with individual pigs as units. The dynamics of spread of a pathogen among individuals within a compartment highly determines the dynamics at a higher scale e.g. between compartments or farms. Consequently, the effectiveness of a measure with respect to reducing the spread of a pathogen at a higher scale depends on the effect at a lower scale [14]. As a first step, we therefore focus on the lowest scale, i.e. individual pigs in a compartment.
Many routes have been suggested for pig-to-pig transmission of S. suis. As S. suis is frequently isolated from the pigs upper respiratory tract, the direct oro-nasal route is generally assumed an important one [27,28]. Other, indirect routes are also possible (e.g. airborne) [35].
The aim of our study was to determine whether the spread of S. suis within a compartment could be reduced by prevention of direct contact between pigs. We quantified the transmission rate of S. suis serotype 9 in an experimental set up between pigs housed in pairs and between pigs housed in pens placed at a distance of approximately 1 m. Such a design has been considered as robust in quantifying transmission [36,37], and has been used for several other bacterial pathogens [38][39][40]. The parameter estimates derived from our experiments were used in a simulation model to study the effect of separation of pigs in a hypothetical conventional farm. Our experimental results show that the transmission rate is influenced by contact structure between pigs, but the simulation showed that separation most likely does not restrict spread of S. suis in a conventional stable compartment.

Ethics Statement
The experiments were approved by the Animal Care and Ethics Committee of Utrecht University, in accordance with the Dutch law on experimental animals (approval number DEC 2008.II.08.072). To reduce the number of animals, we carried out experiments to study the effect of vaccination on transmission of S. suis as well. The results of that research objective have been published elsewhere [41].

Inoculum
S. suis serotype 9 strain 7997 was used as inoculum (provided by H. Smith, Central Veterinary Institute, Wageningen UR, Lelystad, The Netherlands). The strain was isolated from a clinical case in a commercial farm in The Netherlands. The strain contains genes encoding for several (putative) virulence factors, like suilysine (SLY) and variant (higher molecule weight) muramidase released protein (MRP*) [42], and belongs to the clonal complex (CC) that includes the vast majority of invasive S. suis serotype 9 strains found in the Netherlands, i.e. CC16 [43]. After overnight culture from a 280uC stock on agar plates at 37uC and 5% CO 2 , one colony was suspended in 10 mL Todd-Hewitt broth (TH) (BioTrading, The Netherlands), and incubated for 3-4 h at 37uC until an optical density of 0.5-0.6 at 600 nm. After overnight storage at 4uC this suspension was diluted tenfold in TH and cultured for 2 h at 37uC resulting in an optical density of 0.5-0.6 at 600 nm. Ten mL of this suspension was washed twice and suspended in 10 mL physiologic saline solution. The bacterial concentration of the final suspension was 2-3610 8 colony forming units (CFU) per mL.

Pigs
We used Landrace6Yorkshire pigs from sows housed at the animal facilities of the department. Piglets were caesarean derived and colostrum deprived. The first 4 weeks of life the piglets were housed in isolators, and thereafter in ground floor pens (12-15 pigs/pen). The pigs were fed with milk replacers during the first 4 weeks, and then with gamma-irradiated pelleted concentrates (Sloten B.V., The Netherlands; Trouw Nutrition Nederland B.V., The Netherlands). The feed contained Enterococcus faecium (DSM 7134), Bacillus licheniformis (DSM 5749) and Bacillus subtilis (DSM 5750).

Experimental design
Three replicate experiments were conducted sequentially (for details: see Table 1 and Figure 1). Pigs were housed either in pairs or alone in boxes. Pair-wise housed pigs were used to measure direct transmission, the pigs housed alone to determine indirect transmission. The distance between boxes was 80-100 cm. This distance was sufficient to prevent direct contact between pigs in different pens, and might be applicable between groups in commercial farms. The boxes had multiplex walls (height: 80 cm), iron grid floor with rubber lying area, with a total area of 1.2 m 2 per box.
At the age of 7 weeks, one pig in each pair was inoculated intranasally with 1610 9 CFU S. suis in 5 mL saline while sedated. Inoculation was performed in another stable compartment to prevent infection of contacts due to spread of the inoculum. Two days later the inoculated pigs were placed back into their original boxes. The pen mates were considered to be directly exposed, the pigs housed alone to be indirectly exposed. In total 25 inoculated (referred to as I), 25 direct contact pigs (referred to as C dir ) and 15 individually housed, indirect contact pigs (referred to as C ind ) were  Figure 1). Rectal temperatures and clinical observations were recorded daily. Neurologic signs and signs caused by lesions of the locomotor apparatus were scored separately. Pigs that showed severe clinical signs were euthanized. The remaining pigs were euthanized at 26-28 days post inoculation (DPI).
Biosecurity protocols were applied to prevent contact infections by animal handling. Animal handlers wore sterile gloves, boots and coveralls which were changed before handling between each box. C ind pigs were sampled first, and C dir pigs were sampled before I pigs.
To study the effect of vaccination on transmission of S. suis 13 of the 25 pairs of pigs (see Table S1 for details) were vaccinated at 4 and 7 weeks of age with a formalin inactivated whole bacterin vaccine, containing the same S. suis strain as used for inoculation (homologous vaccination). The results showed that both level of S. suis colonization and within-pen transmission were not affected by vaccination [36].

Sampling
Saliva and tonsil brushing samples were taken at 5 days and 3 h before inoculation, and 2-7, 9,12,15,19,22 DPI, and at the end of the experiment. We used a sterile steel wedge to open the mouth; saliva was sampled by turning round a swab (CultiplastH) under the tongue for 5 s. Both palatine tonsillar areas were brushed for 3 s each with a sterile toothbrush. The brush and swab heads were put in separate sterile tubes containing 10 mL saline solution and transported to the laboratory. All pigs were necropsied and macroscopically affected organs were, if present, sampled for bacteriological examination. Palatine tonsillar tissue was collected.

Laboratory tests
Quantitative bacteriological examination was performed on all samples. Serial dilutions of swab or tonsil brushing samples (10 0 -10 24 ) were plated on selective agar plates, containing Columbia agar, 6% sheep blood, 0.2 mg/mL crystal violet and colistin/ oxolinic acid (BioTrading, The Netherlands) [41,44]. After incubation for 18-24 h at 37uC and 5% CO 2 , plates with 10-200 colonies were selected. Suspected S. suis colonies were counted and per plate two S. suis suspected colonies were subcultured and tested for amylase activity [45]. Isolates that showed amylase activity were stored at 220uC in 0.5 mL TE buffer pH 7.5 (10 mM) until further processing. DNA-isolation of these isolates was performed with InstaGene TM Matrix (Biorad, The Nether- Experimental design for evaluation of the effect of prevention of direct contact by spatial separation on transmission of S. suis serotype 9 among pigs. Inoculated pigs (I) were intranasally infected with S. suis serotype 9 two days before placing in these boxes. Stable compartments contained 2 to 4 boxes with pair-wise housed pigs (C dir +I) and 1 to 3 boxes with pigs housed individually (C ind ). C dir pigs had direct contact with I pigs; C ind pigs were only indirectly exposed to S. suis. The air inlet in the compartment is marked with ; the air outlet with . doi:10.1371/journal.pone.0061339.g001 lands), according to the manufacturers instructions. A real time polymerase chain reaction (RT-PCR) on the cps9H gene [41,46] was performed on suspected isolates from the first two and the last saliva and tonsil samples in which amylase positive colonies were detected. All colonies that were visually classified as S. suis tested amylase-positive, and were all positive in the cps9H-PCR. Tonsil tissue samples that were obtained at necropsy were submerged in boiling water (5 s), crunched in a StomacherH macerator (Laméris, The Netherlands) for 10 min, and diluted, plated and confirmed as described above for samples, except for the dilutions that were plated (10 21 to 10 25 ).

Statistical analysis of culture data and clinical data
Mean S. suis 10 logCFU levels of the first 3 samples taken after onset of colonization in individual pigs were compared among the 3 groups (C dir , C ind and I) using a Kruskal-Wallis one-way ANOVA. The same procedure was performed for mean levels over the whole period after pigs tested positive, and for S. suis counts (in 10 logCFU/g tissue) in tonsillar tissue. For samples where no S. suis was detected, the detection limit of the test procedure (i.e. 2610 3 CFU/sample) was used for the calculations. It did not affect the results if we had used a much lower value, i.e. 2 CFU/sample, instead for analysis.
Clinical signs were compared between the groups with a Kruskal-Wallis test applied on the percentage of days a pig showed signs. If the result of the Kruskal-Wallis test was significant, the Mann-Whitney U test was performed to determine differences between groups. The Bonferroni approach was used to correct for multiple comparisons. Differences in outcome variables were considered significant if P,0.05. Analysis of culture and clinical data was performed in SPSS version 16.0.2.

Statistical analysis of transmission
To analyze transmission of S. suis, a stochastic SI (susceptibleinfectious) model was applied. A pig was considered to be infected if at least one sample tested positive. The transmission rates b dir and b ind represent the rates at which infectious animals transmit the infection to other animals within and among pens, respectively. The parameter b ind includes the process of S. suis transmission without direct contact between hosts, and thus exclusively via the environment. The parameter b dir includes both transmission via direct contact and transmission via the environment, as these two routes cannot be distinguished in a pair wise setting.
According to the standard SI model, susceptible pigs become infectious at rate b dir * I w (or b ind * I b ) in which I w and I b are the numbers of infectious animals present within the same pen, and in the whole compartment, respectively. The standard model assumes that susceptible animals are infected at a constant rate if I w (or I b ) is constant, which was more or less constant during our experiment. For within-pen transmission the data allowed the standard model to be used. For between-pen transmission, however, we did not see new infections in C ind until one week after exposure to inoculated pigs, and counted 13 cases in the two weeks thereafter. This implies that for between-pen transmission the assumption of constant infection rate per unit of time was not met, and the standard SI model could not be used for analysis. For between-pen transmission, we therefore fitted an alternative model in which susceptible animals are infected at rate b' ind * SI, in which SI is the cumulative number of shedding days of all infectious animals in that stable compartment (at time t). In this model a shedding day was defined as a day that S. suis was found in the tonsil and/or nasal swab sample taken at that day. This model reflects a gradual build-up of infectivity in the environment. The alternative transmission rate b' ind (unit: 1/day 2 ) is now interpreted as the number of new infections an animal will cause each day in a susceptible population, for each day that it has already been infectious. In the analysis it was assumed that from the moment that pigs died or were euthanized, they did not add any infectivity to the cumulative infectivity anymore. The infectivity they had added before death was still included in the total infectivity in that compartment for the remaining experimental period.
Additionally, as the bacterial load in infected pigs might influence the transmission rate, models were tested which included cumulative colonization or shedding levels in tonsillar brushing or saliva samples instead of shedding days. The 'cumulative colonization or shedding level' at a particular day was calculated by adding the CFU/sample up to that day, using interpolation for all days on which no sample was taken. We used Akaike's Information Criterion (AIC) to test if this alternative model fitted the data better than the original model [47]. The transmission rate calculations were performed in statistical program R version 2.13.0.
As mentioned, the experiment was designed to serve a dual purpose to reduce the number of experimental animals. As vaccinated and non-vaccinated pigs were distributed over all but one compartments, it was not possible to distinguish between infectivity of vaccinated and unvaccinated pairs. In addition, the bacterial load in swab or brushing samples did not differ significantly between vaccinated and unvaccinated pigs. It was therefore assumed that all colonized pigs contributed equally to transmission of S. suis. Therefore, vaccination was not included as explanatory variable in the statistical analysis to determine the effect of spatial separation. Power analysis was based on the research question about vaccination [41].

Simulation study
The effect of separation of groups of pigs on the spread of S. suis in a conventional farm compartment was studied using a simulation model. A situation with separated litters (scenario A) was compared with one of mixing litters (scenario B). In scenario shown separately for pigs inoculated with (I), directly (C dir ) or indirectly contact exposed (C ind ) to S. suis. Time is expressed as days after a pig was found positive firstly. As sampling was not conducted daily, and because of the delay in S. suis transmission to mainly C ind pigs the numbers and time points of observations differ between the groups. No significant differences were observed in mean levels between C dir , C ind and I pigs for the first 3 positive samples and for mean levels over the whole sampling period. doi:10.1371/journal.pone.0061339.g003 A, the compartment contained 10 pens each consisting of 10 pigs. In scenario B, 10 litters of 10 pigs were mixed to form two groups of 50 pigs housed in two pens. The model was independent of distances between pens. Simulations were started with one infected pig. The estimate for b' ind was divided by two to compensate for the in comparison to a compartment extra space needed to house 100 pigs. Per situation 10.000 iterations were performed. Simulation was done with statistical program R version 2.13.0.

Colonization
No S. suis was detected in samples taken before inoculation. All inoculated pigs tested positive for S. suis at 2 DPI and all C dir pigs tested positive within two days after exposure. The first C ind pigs tested positive on day 7 post exposure and their number gradually increased to 13 (Figure 2). Once positive, pigs remained positive during the remaining experimental period.
Mean levels of S. suis are shown in Figure 3. No differences were observed between the animal categories (C dir , C ind and I) in mean levels in tonsil and saliva samples (P-values are 0.25 and 0.12, respectively, for the first 3 positive samples and 0.19 and 0.66, respectively, for the whole sampling period). No differences between these categories were observed in the load of S. suis in their tonsil tissue samples (P = 0.99). Based on these results, the assumption of homogeneity of infectivity was considered reasonable.

Estimation of transmission rates
The rate for within-pen transmission b dir was estimated at 3.58 (95% CI: 2.29-5.60) new infections an animal will cause in a susceptible population, and for between-pen transmission b' ind was 0.001 (95% CI: 0.0006-0.0017) new infections an animal will cause each day in a susceptible population, for each day that it has already been infectious. Spatial separation reduced the transmission on average 36 times (range 20-895). Addition of cumulative S. suis levels did not change the fit of the model for estimating indirect transmission.

Clinical signs and mortality
Lameness was the only clinical sign observed in C pigs; it occurred in 4 out of 25 C dir and 3 out of 15 C ind pigs (P = 0.99). Signs were observed more often in I than in C dir (P = 0.01) or C ind pigs (P , 0.001). Out of the 25 I pigs, in 20 pigs fever was recorded on one or more days, in 10 pigs lameness signs, in 3 pigs neurologic signs and 4 pigs died before the end of the experiment.

Discussion
The objective of this study was to quantify the effect of separation of pigs on the transmission rate of S. suis serotype 9. All directly and indirectly exposed pigs became colonized after exposure to inoculated ones, but the rate at which it occurred was approximately 36 times lower for indirectly compared to directly exposed pigs. This finding suggests that this intervention measure could contribute to the reduction of S. suis spread in a farm. However, the simulation study using these parameters showed that the cumulative incidence in piglets did not differ between 'separated' and 'mixed' groups to a relevant level. This implies that it is unlikely that separation of pigs as simulated in this simulation exercise will contribute to reduction of the prevalence of S. suis serotype 9 infections in a farm compartment, also taking into account that on conventional farms less stringent hygienic measures are implemented than in this experiment.
Separation sharply reduced the transmission rate in the experiment, but had only a small effect on colonization time in a farm compartment. This counterintuitive finding is caused by the large number of pigs present in such a compartment. Although each of the indirectly exposed pigs only has a small probability of getting infected, the probability that at least one of them will contract the infection is considerable and, moreover, strongly increases as the number of infectious pigs in the 'source' pen increases. Once indirect transmission has taken place, direct transmission further increases the infectious load and consequently increases exposure of the remaining uninfected pigs and the probability of further indirect transmission.
In total 15 C ind pigs were colonized, most likely via contaminated dust particles, as has been described for S. suis serotype 2 [35,48]. We consider other indirect routes of S. suis spread, e.g. via animal caretakers [20,21,49], less likely because of the stringent biosecurity measures that were applied.
A model assuming increasing infectivity fitted the data of indirect transmission better than a model assuming fixed infectivity. The increasing infectivity is likely to have occurred due to accumulation of S. suis in the environment [28], a phenomenon which has also been observed in studies with other pathogens [40,[50][51][52][53][54][55]. In studies with Campylobacter jejuni and Escherichia coli in broilers, for example, a similar pattern of increasing probability of infection was observed [40,50]. The authors of these studies concluded that the model fitted the data better when assuming increasing infectivity [40,50]. We also monitored the levels of the pathogen in the samples, as an association between these levels and an increasing infectivity could be suggested. The levels remained rather similar during the experiment, and addition of cumulative levels of S. suis in saliva or tonsil samples did not further improve the fit of the model.
Transmission of S. suis is not only dependent on the type of contacts between pigs, but also on the infectivity and susceptibility of individual pigs, which might be affected by (changes in) host behaviour. This may, for example, occur after mixing pigs from different litters leading to fighting [56][57][58], which would change the frequency and/or intensity of contacts between pigs [11,12,[56][57][58]. Moreover, the stress induced by mixing [11,12,[56][57][58] could further increase both susceptibility, and infectivity by increasing colonization and shedding of already infected hosts, as described for other bacterial pathogens [59][60][61]. If and to which extent the transmission changes after mixing is, however, not known. Moreover, even with adjustments the conclusions about the relevance of the effect of spatial separation would not change, as in the optimal 'separation' scenario the spread of S. suis serotype 9 was already so rapid that all pigs in a compartment were infected within about two weeks.
In the experiment stringent hygienic measures were applied between pens at compartment scale, and the same was assumed in the simulation study. In pig farms it is probably nearly impossible to perform such strict measures between pens in the same compartment, but they might be implemented at herd scale between compartments. As the rate of airborne spread is probably lower between compartments than between pens, spread of S. suis strains between different compartments, which are physically separated, might be reduced. Table S1. Loads of S. suis in saliva, tonsillar brushing and tonsillar tissue samples, and clinical signs in individual pigs. (DOCX)