Detection of African Swine Fever Virus in the Tissues of Asymptomatic Pigs

RESEARCH ARTICLE Abworo et al., Journal of General Virology 2017;98:1806–1814 DOI 10.1099/jgv.0.000848

Detection of African swine fever virus in the tissues of asymptomatic pigs in smallholder farming systems along the Kenya–Uganda border: implications for transmission in endemic areas and ASF surveillance in East Africa

Edward Okoth Abworo,1,* Cynthia Onzere,1 Joshua Oluoch Amimo,2 Victor Riitho,1 Waithaka Mwangi,3 Jocelyn Davies,4,5 Sandra Blome6 and Richard Peter Bishop1

Abstract The persistence of African swine fever virus (ASFV) in endemic areas, with small-scale but regular outbreaks in domestic pigs, is not well understood. ASFV has not been detected using conventional diagnosis in these pigs or adjacent populations of resistant African wild pigs, that could act as potential carriers during the outbreaks. However, such data are crucial for the design of evidence-based control strategies. We conducted cross-sectional (1107 pigs) and longitudinal (100 pigs) monitoring of ASFV prevalence in local pigs in Kenya and Uganda. The horizontal survey revealed no evidence of ASFV in the serum or blood using either conventional or real-time PCR. One pig consistently tested positive using ELISA, but negative using PCR assays on blood. Interestingly, the isotype of the antibodies from this animal were strongly IgA biased relative to control domestic pigs and warthogs, suggesting a role for mucosal immunity. The tissues from this pig were positive by PCR following post-mortem. Internal organ tissues of 44 healthy pigs (28 sentinel pigs and 16 pigs from slaughter slabs) were tested with four different PCR assays; 15.9 % were positive for ASFV suggesting that healthy pigs carrying ASFV exist in the swine population in the study area. P72 and p54 genotyping of ASFV revealed very limited diversity: all were classified in genotype IX at both loci, as were virtually all viruses causing recent ASF outbreaks in the region. Our study suggests that carrier pigs may play a role in ASF disease outbreaks, although the triggers for outbreaks remain unclear and require further investigation. This study significantly increases scientific knowledge of the epidemiology of ASF in the field in Africa, which will contribute to the design of effective surveillance and control strategies.

Household monthly income, current household monthly INTRODUCTION expenditure on meat, relative price of pork, preference for Pigs are increasingly contributing to improved nutrition value-added pork products, prices of substitutes and and household incomes in regions of Africa where pork response of households to improvements in pork quality consumption and pig keeping are culturally acceptable. In have been shown to be associated with increased consump- Uganda, for example, pork is second only to beef in terms of tion of pork in some parts of Africa [2]. However, African meat production and accounts for at least a third of the swine fever (ASF), an infectious and lethal disease of domes- À current 10 kg year 1 per capita meat consumption [1]. tic pigs, constrains the realization of economic benefits

Received 19 January 2017; Accepted 1 June 2017 Author affiliations: 1International Livestock Research Institute (ILRI), P.O. Box 30709, GPO 00100, Nairobi, Kenya; 2The University of Nairobi, Faculty of Veterinary Medicine, P.O Box 29053 00625, Nairobi, Kenya; 3Department of Diagnostic Medicine/Pathobiology, College of Veterinary Medicine, Kansas State University, Manhattan, KS 66506, USA; 4Land and Water, The Commonwealth Scientific and Industrial Research Organisation (CSIRO), Alice Springs, Australia; 5The Northern Institute, Charles Darwin University, Grevillea River, Alice Springs, Australia; 6Friedrich-Loeffler-Institut, Bundesforschungsinstitut für Tiergesundheit, Federal Research Institute for Animal Health, Südufer 10 , D-17493 Greifswald-Insel Riems, Germany. *Correspondence: Edward Okoth Abworo, [email protected] Keywords: African swine fever; domestic pigs; tissue sequestration; carrier pigs; prevalence. Abbreviations: ABI, Applied Biosystems; ASF, African swine fever; ASFV, African swine fever virus; AusAID, Australian Agency for International Devel- opment; CSIRO, Commonwealth Scientific and Industrial Research Organisation; Ct, cycle threshold; FAO, Food and Agriculture Organization of the United Nations; FLI, Friedrich-Loeffler-Institut; GIS, geographic information system; HRP, horseradish peroxidase; IgA, immunoglobulin A; OIE, World Organisation for Animal Health; PBS-M, phosphate-buffered saline–milk; PBS-T, phosphate-buffered saline–Tween; qPCR, quantitative polymerase chain reaction; UPL, Universal ProbeLibrary. The GenBank accession numbers of the sequences mentioned in this paper include KM00023, KM000146, KM000154–158, KM000160–162, KM000165, KM000167, KX776419, KX776420, KM000194, KM000196, KM000197, and KM000198. One supplementary table is available with the online Supplementary Material.

000848 Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111806 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814 along pig value chains and presents a major risk for farmers RESULTS who invest in pig production [3, 4]. The disease, first ASFV surveillance study site diagnostic results reported in Kenya in 1910 [5], has spread within the African continent over the past two decades, most recently to All pigs (1107) sampled during the cross-sectional study Ethiopia [6]. The disease currently has no effective chemo- were negative using PCR assay screening on blood, while therapeutic treatment or vaccine available for its control. only one was positive for ASFV using a commercial ELISA Development of innovations and interventions to effectively kit (Table 1). During the longitudinal study, five pigs were manage the constraint posed by ASF requires understanding positive for ASFV by both conventional PCR [9] and Uni- the epidemiology of the disease in smallholder farming versal ProbeLibrary (UPL) real-time PCR [10] diagnostic regions. Such information on constraints to pig productivity assays using blood, with three of them being positive for is rarely available for formulating livestock sector policy and ASFV in tissues, following post mortem (Table 1). ASFV institutional changes (http://www.africalivestockdata.org/). was successfully isolated from the PCR-positive tissue sam- The ASF virus (ASFV) is known to be potentially transmit- ples in vitro. ted to susceptible pigs though multiple pathways [3, 4]. This Detection of the ASFV genome in tissues can be by direct contact with infected pigs, meat and slaughter waste from infected pigs, and potentially also clothing or Four diagnostic assays were applied in an effort to investi- tools transported by people who have been in contact with gate the possibility of ASFV sequestration in tissues from infected pigs. It could potentially also be transmitted or the study area and slaughter slabs (Table 2). In vitro isola- maintained in a pig population by Ornithodoros soft ticks, tion of the virus was only successful for tissue samples that the presumptive ancestral arthropod reservoir and vector tested positive in both the conventional and UPL PCR for transmission of the virus from warthogs, specialized assays. African wild pigs which can carry the virus without signs of From the longitudinal study, 28 ear-tagged sentinel pigs disease, to domestic pigs. that tested negative for virus in blood by PCR and serologi- The mechanism of persistence of ASFV in endemic areas cal assays, with the exception of two pigs (UG64/2013 which was seropositive during the entire study period and Ken13/ where there are sporadic outbreaks of ASF in domestic pigs, busia.3 which was positive in blood by conventional and but no apparent virus reservoir in resistant African wild UPL PCR from the first longitudinal time point at pigs or argasid soft ticks, is currently unclear. The potential 3 months), were euthanized and tested for the presence of role of carrier domestic pigs as a source of infection has ASFV in the tissues. Three out of 28 samples (10.7 %) tested been documented in Kenya [7]. This study was designed to PCR-positive in tissues using both conventional and UPL confirm the hypothesis that carrier pigs may contribute to PCR assays; however, they were ASFV-negative by PCR virus transmission and thus the spread and maintenance of using blood samples (Table 2). Using a modified Taqman the disease, thereby complicating attempts at control [8]. qPCR assay that was initially developed by King et al. [11] We investigated the presence of long-term infected pigs in implemented at the Friedrich-Loeffler-Institute (FLI; the smallholder swine population that could be responsible Germany), 15 out of 28 pigs were positive for ASFV in the for the outbreaks that are regularly reported in the border tissues. Tissue and blood samples from 12 of the 15 pigs that had tested positive at the FLI were selected and tested region of Kenya and Uganda, as part of an in-depth study with a modified PCR assay initially developed by Zsak and of ASFV prevalence and ASF outbreaks in 640 households, colleagues at the Plum Island Animal Disease Center, New using a randomized cluster design. The overarching goal York, USA [12, 13]. Using this assay, 9 out of 12 pigs were was to increase scientific knowledge relating to the epidemi- found to be positive for ASFV in the tissue samples, while in ology of ASF in the region as a prerequisite for the design of the blood samples 7 out of 12 pigs were positive for ASFV. effective ASF surveillance and control strategies. Both the King et al. assay using an ABI 7500 platform and

Table 1. Detection of ASFV across sampling periods using PCR and serological diagnostic assays

ASFV positivity across sampling periods

Diagnostic assay Sample type Cross-sectional survey Longitudinal survey phase 1 (3 months) Longitudinal survey phase 2 (6 months)

Conventional PCR Blood 0/1107 1/232 4/142 Serum 0/1107 0/232 0/142 Tissues 0/0 2/232 1/142 UPL PCR Blood 0/1107 1/232 4/142 Serum 0/1107 0/232 0/142 Tissues 0/0 2/232 1/142 ELISA Serum 1/1107 1/232 1/142

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111807 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814

Table 2. Summary of ASFV diagnostic results from sentinel pigs sacrificed after the end of the longitudinal survey and additional samples from slaughter slabs within the study site

Sentinel pigs Slaughter slab

Tissues Blood Tissues Blood n n n n Diagnostic assay =28 =28 =16 =16

Conventional PCR 3 0 3 0 UPL real time PCR 3 0 4 0 Modified King et al. PCR assay 15 0 11 2 Modified Zsak et al. PCR assay* 9 (n=12) 7 (n=12) ––

NB, The specific organs that tested positive in the study site sacrificed and slaughter slab sentinel pigs in this study are comprehensively summarized in Table S1. *The slaughter slab pigs were not tested using the Zsak et al. assay, due to the lack of an ASFV-specific Taqman probe.

the modified Zsak et al. assay using a Smartcyler (Cepheid) KM000160–162, KM000165, KM000167 and KX776419. platform yielded additional ASFV positives, as compared to The full-length p54 gene sequence was also determined and conventional PCR and UPL assays; however, these had rela- clustered in p54 genotype IX. The p54 accession numbers tively high Ct values (>40) and none were validated by virus are: KM00023, KX776420, KM000198, KM000194, isolation. KM000197, KM000196 (see Table S1, available in the online Supplementary Material). Additionally, 16 pigs sold by farmers at different times at local slaughter slabs in the Busia study district (Kenya) over Isotype analysis of anti-ASFV immune response a period of 1 month in July 2014 were sampled using both UG64/2013 serum which was positive by diagnostic p62 blood and internal tissues in order to test the extent of competition ELISA (INgezim PPA Compac, Ingenasa, ASFV tissue sequestration in the pig population using an Madrid, Spain) was further characterized for isotype specific independent sampling strategy. The data revealed 25 % (4/ antibody responses against p72 antigen in comparison to 16) positive pigs in the tissue, but none were ASFV-positive serum from a pig vaccinated with an attenuated virus in blood using the UPL PCR assay. The percentage of posi- (E75CV1), two field warthog sera that were also positive by tivity in tissues increased to 68.75 % (11/16) using the modi- the INgezim ELISA and serum from an ASFV naïve Euro- fied Taqman qPCR assay developed at the FLI (Table 2). pean breed pig (Fig. 1). UG64/2013 serum had lower IgG Samples from reported outbreaks responses compared to the E75CV1 and warthog sera, and higher IgA and IgM responses compared to all the other Several ASFV outbreaks were reported to veterinary author- sera. A low IgG:IgM ratio was observed in the UG64/2013 ities in Busia district during the study period. In two of these outbreaks, samples were obtained. Conventional PCR and UPL qPCR assays conducted on the samples collected dur- UGA64 4 ing the outbreak (n=5) confirmed that in both cases the sus- E75_CV1 pected pigs were ASFV-positive. ) Warthog 1 3 Warthog 2 Virus isolation 450 nm Naive pig To confirm that the tissue positives detected by PCR con- 2 tained viable viral isolates, ground-up tissues were used to infect peripheral blood mononuclear cells (PBMCs) as

Mean OD (A 1 described in Methods. This resulted in the collective isolation of nine viral isolates in culture using all four sampling 0 strategies. IgG IgA IgM Control IgG:M ratio Genetic analyses Isotype Detailed genetic characterization of the ASFV detected during the study period (in blood and tissues) has been per- Fig. 1. ASFV p62 isotype-specific antibody responses. IgG, IgA, IgM formed (C. Onzere et al., in preparation). The analysis by specific anti ASFV p62 responses and responses to an irrelevant spe- ¢ cies isotype control are shown. The columns show sera from a field partial sequencing of the p72 gene (3 end) revealed that all seropositive animal (UG64/2013), a monkey CV1 cell line attenuated the ASFVs detected clustered with ASFV genotype IX and E75 virus vaccinated pig (E75CV1), Ol Pejeta warthogs (warthogs 1 the sequences are available in GenBank. The GenBank and 2) and an ASFV-naïve pig. The IgG:IgM ratio in the respective sera accession numbers of the p72 gene sequences for ASFV is also shown. detected in this study are: KM000146, KM000154–158,

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111808 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814 serum (<1), while the E75CV1 and warthog sera had an immediate area and two of these were confirmed by our IgG:IgM ratio >1. The IgG:IgM ratios are useful in distin- laboratory during the study period, as indicated in the guishing responses from a recent/acute infection and well- results. This result of a low frequency of long-term ASFV- established responses, with acute responses having a low infected pigs detectable by sampling in blood, is similar to (<1) ratio compared with well-established responses (>1). that reported from an ASFV endemic region in central The measurement of IgM antibodies as well as the IgG:IgM Uganda [14], although key differences in this study were, ratio may be a criterion in epidemiological studies to dis- first, that we were able to confirm suspected ASFV out- criminate acutely infected animals and animals with well- breaks by PCR and, second, that a percentage of animals established responses that are potentially protective. The were positive in the tissues and were confirmed by virus other notable feature of the UG64/2013 serum was the high isolation and genotyping in vitro. However, whether these level of the IgA response. tissue positive animals represent ‘carriers’ in the sense that they are capable of directly transmitting the disease to unin- DISCUSSION fected naïve pigs remains to be determined. The virus transmission dynamics and biology underpinning Limitations in the sensitivity of virus detection assays could geographically restricted clinical foci of ASF in districts explain non-detection of positive samples in blood [15]. exhibiting persistent outbreaks of disease in African domes- Differences in the sensitivity of the modified King et al. and tic pigs is not currently understood. ASFV prevalence in Zsak et al. qPCR assays relative to UPL and conventional pigs has not been systematically assessed in these systems. PCR support this contention. However, none of the addi- However, such data are crucial for the design of evidence- tional positives identified by the modified qPCR assays were based control strategies. We conducted both cross-sectional confirmed by viral isolation and all exhibited high Ct values. and longitudinal monitoring of ASFV prevalence in African Previous studies suggest that lack of success in detection of genotype domestic pigs in the border region of western anti-ASFV antibodies in the case of genotype IX is not Kenya and eastern Uganda. Internal organ tissues of 44 attributable to antigenic polymorphism in the target anti- healthy pigs (28 sentinel pigs and 16 pigs from slaughter gens but may, alternatively, be related to the specific charac- slabs) were tested with four different PCR assays; 15.9 % teristics of the phenotypes of African pigs which differ were positive for ASFV, using all assays, suggesting that genetically and likely result from different domestication healthy pigs carrying ASFV exist in the domestic pig popu- events [16]. PCR analysis confirmed outbreaks, and the fact lation in the study area. Genotyping of ASFV by sequencing that virus was isolated in the two outbreak cases rules out the C-terminus of p72 indicated very limited diversity; all the failure of the haemadsorption assay or a fundamental were within p72 and p54 genotype IX, which has been asso- failure of PCR assays to detect the virus in the pigs that were ciated with all ASFV outbreaks in the region analysed genet- sampled. ically over the past decade. Our study suggests that carrier pigs may be involved in causing in ASF disease outbreaks, Recent studies in south-western Kenya [7] have demon- although the specific proximal triggers require further strated that some p72 genotype X-infected pigs were posi- investigation. tive with ASFV from blood, but appeared asymptomatic. However, as in the case of the Busia pigs, it was unproven Both conventional and UPL PCR analyses of 1107 pig blood whether these infected pigs were ‘carriers’ that could trans- and sera samples included in the horizontal survey were mit the virus to naïve animals on the same farms. Tissues negative for ASFV. Similarly, whole blood and serological from these animals were not tested concurrently. analysis of sentinel pigs sampled longitudinally at three time points (a baseline sampling and two subsequent longi- To test the potential involvement of ‘carrier’ pigs in tudinal samplings at 3 month intervals) mostly exhibited ASFV transmission in the Busia/Teso production system, negative results for ASFV according to antibody and virus we further investigated the presence of detectable virus detection. Only six of these pigs tested positive on whole DNA in tissues from apparently healthy pigs. Earlier stud- blood and only three of these were also ASFV-positive in ies have shown high ASFV prevalence through screening tissues. Interestingly, there was one exception (an animal of blood and sera from pigs at slaughter slabs [17, 18]. A designated UG64/2013), which was consistently seroposi- high prevalence as detected by antibody and virus detec- tive, negative according to PCR applied to DNA extracted tion assays in slaughterhouse samples is expected during from whole blood, but positive in several tissues using a the acute phase, following outbreaks, that is associated PCR assay subsequent to post-mortem. The naïve interpre- with farmers disposing of pigs by slaughter to avoid losses tation of these results was that the population of pigs sam- through mortality. pled was minimally exposed to and free from ASFV in the blood according to the diagnostic assay methods used. The Virus characterization data were consistent with the hypothesis that these sentinel Viruses from all the PCR-positive tissues sampled from the pigs had limited exposure to ASFV infection during the sentinel pigs, slaughter slabs and outbreak cases were geno- period of the study in 2012–2013. This was a surprising typed. The molecular characterization of ASFVs from the result given that a number of suspected ASF outbreaks had study region revealed limited diversity with all viruses being been reported to the local veterinary authorities in the classified within p72 genotype IX (C. Onzere et al., in

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111809 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814 preparation), which is the major genotype associated with both previously published observations [12] and also the recent ASF outbreaks throughout both Kenya and Uganda current study. [19, 20] and different from genotype X viruses that have It seems possible that multiple modes of transmission may recently been associated with a carrier-like status in pigs in exist for ASFV: an initial one in which farmers dispose of south-west and western Kenya and also observed in ticks pigs incubating disease that are not yet clinically ill, through collected from warthog burrows [7, 20]. An earlier Kenyan sale or agistment [25, 26], a second one during active out- genotype X isolate from the 1950s whose complete genotype breaks when farmers rapidly sell-off animals, and which is has been determined [21] induced a lethal infection in difficult to measure in the field due to the very rapid death domestic pigs. However, in the south-western Kenya study, of infected animals resulting in rapid ‘burn out’ of infections the ASFV detected in blood by PCR was not associated with [24, 25], and a third one involving virus sequestrated in the clinical symptoms of ASF and the animals had no detectable tissues, that we describe in this study. It seems quite con- ASFV antibody response using the World Organisation for ceivable that tissues from slaughter slabs, or shedding of Animal Health (OIE) indirect-ELISA. Sequence analysis of body fluids (whose duration and frequency is unknown in a two genotype X isolates from a virulent pig isolate from the field context), may be key determinants in maintaining 1950s and a recent tick isolate plus a genotype IX isolate transmission in endemic areas. from a clinically sick pig demonstrate that the complete genomes of the Kenyan and Uganda isolates are very dis- Such a mechanism of persistence of carriers through seques- tinct from the other complete ASFV genomes sequenced to tration in tissues is similar to that seen in African wild suids, date [22]. Genotypes IX and X are, however, closely related particularly warthogs [27]. However, there are several dif- to one another and the p72 genes of genotypes IX and X ferences: first, unlike the African domestic pigs, adult wart- comprise the most similar pairing among the 23 currently hogs are almost universally seropositive, and second publicly available genotypes. Ornithodoros ticks infesting burrows are typically involved in the warthog ‘sylvatic cycle’. Antibody isotype analysis The current data demonstrate that there are potential car- Although there was only one seropositive pig observed in rier pigs, devoid of clinical signs of ASF or detectable ASFV the current study (UG64/2013), interestingly, the isotype of in blood in the study region that nonetheless have ASFV the antibodies from this animal were strongly IgA biased, sequestered in tissues. These data imply a requirement to relative to control domestic pigs and warthogs, suggesting a design control approaches that limit the potential impact of role for mucosal immunity. It should be noted that this was these carrier pigs on the transmission of ASFV. The result among a sample of >1000 individual pigs that were assayed. indicated the importance of further sampling to test the The low IgG:IgM ratio and high IgA level in UG64/2013 extent of ASFV tissue sequestration in the wider population may be attributable to the route of virus administration. of pigs in the eastern African region. Insight into factors ASFV was likely transmitted to UG64/2013 through pig-to- that could trigger latency and shedding of the virus by these pig contact, whereas the attenuated E75CV1 was delivered carrier pigs leading to outbreaks is also required. One possi- via the intramuscular route. The two warthogs were almost bility is that co-infections may play a role; for example, bur- certainly infected as neonates by Ornithodoros ticks in wart- dens of gastrointestinal helminths (E. Okoth and R. Bishop, hog burrows. The data on these responses is sufficiently dis- unpublished data) and enteric viruses [28, 29] were high in tinct to warrant further comprehensive investigation (with some animals in the current study. even larger pig numbers >10 000) or even an entire population, from birth. The data suggest that at least for genotype IX, non-invasive surveillance of key tissues using techniques such as lung Data collected in this study confirm the presence of ASFV lavage, or use of novel molecular markers of infection iden- in the tissues of apparently asymptomatic pigs. Analysis of tified by techniques such as high-throughput RNA sequen- virus transmission during the acute phase is difficult, due to cing or proteomics, may be required to monitor the extent rapid death of infected pigs combined with the lack of of potential ‘carrier pigs’ and their role in the ASF epidemi- timely reporting to veterinary authorities by farmers. This ology in free-range African pig production systems. illustrates the importance of understanding both biological parameters, including transmission timescales [23], and Conclusions anthropogenic factors, including the motivation underlying Cross-sectional and longitudinal surveys of >1000 and 100 decisions to sell or slaughter pigs and report ASF outbreaks smallholder pigs, respectively, using serology, combined [24] in obtaining a more in-depth understanding of ASFV with real-time and conventional PCR, revealed only two epidemiology. The analysis of the questionnaires combined putative carriers of the ASFV p72 genotype IX in the blood with social network studies indicated that farmers probably of African domestic pigs in a specific region on the border recognized ASF symptoms at an early stage and rapidly sold of Kenya and Uganda. However, sampling of other tissues animals either to local butchers or distant neighbours [24]. from animals that died in outbreaks, animals euthanized at This statement was supported by a considerably higher the end of the longitudinal survey, or samples obtained ASFV prevalence in apparently healthy animals sampled at from slaughter slabs during the study period, revealed a slaughter slabs, than in the population at large, based on minimum prevalence of 15.9 %, based on positivity in four

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111810 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814 different PCR assays, supported in nine cases by virus isola- weaner to finish or sale to other farmers (mixed), with some tion. This is consistent with the possibility that carrier pigs farmers using a combination of the three systems [30]. The may play a role in maintenance of ASFV in pig production study area along the Kenya–Uganda border was selected in systems in which persistent clinical ASF outbreaks regularly order to help understand the management of ASF as a occur, although the exact role of these animals in virus trans-boundary animal disease. Indications that outbreaks transmission requires further investigation. probably originating in this region had spread to central and eastern Kenya in the recent past [8, 19], and that the METHODS number of outbreaks was markedly under-reported were also factors in its selection. Data collection and analysis Cross-sectional and longitudinal studies were conducted to were designed to inform identification of key intervention determine prevalence and incidence of ASF, respectively. points for disease surveillance and control. Sampling of pigs taken to slaughter was also performed. Blood, sera and, during the longitudinal phase of the study, Cross-sectional survey internal tissues were sampled from apparently healthy pigs The target population in this study was pigs in the border and where suspected outbreaks were reported by farmers or region of Kenya and Uganda. Using an estimated incidence veterinary authorities. of ASF-positive (+ve) pigs (classified in ASF p72 genotype X) as 0.28 based on [7], 95 % CL and an error margin of Study areas 0.05, a sample size of 640 households was indicated after the The study was conducted in western Kenya and eastern design effect of cluster sampling was accounted for in the Uganda along the border (Fig. 2), a region characterized by sample size calculation. A stratified multi-stage sampling mixed rain-fed farm production systems in a humid/sub- approach was then applied to select the number of pigs. humid environment. Pig farmers in the area are classified Briefly, in the first sampling stage eight administrative loca- into three categories: farrow to weaner, farrow to finish, and tions/sub-counties per country were randomly selected by spatial random sampling executed using GIS and the 2008 Kenyan and 2010 Ugandan administrative boundaries, the most recent datasets that were available to the project. Administrative locations were selected from Busia county: four each from Teso and Busia districts in Kenya. On the Ugandan side, eight administrative sub-counties were selected: four each from Busia and Tororo districts. The next stage randomly selected two sub-locations/parishes from each location or sub-county in Kenya and Uganda, respectively, using computer-generated random numbers, making a total of 32 sub-locations/parishes selected along the border. The third stage was a random selection of two villages from each sub-location/parish making a total of 64 villages. A current list of villages in each selected sub- location/parish was obtained from veterinary officials and village leaders. The fourth stage was random selection of ten pig-keeping households from each selected village totalling 640 pig keeping households selected for a baseline cross- sectional study. Households sampled in village clusters are shown in Fig. 2. Finally, a maximum of four pigs (1–4 pigs) were selected for sampling from each household. A total of 1107 pigs were selected for the initial cross-sectional sampling. Longitudinal survey A random selection was used to sub-sample pigs within a target age range from the 1107 pigs selected for the initial cross-sectional sampling for inclusion in the longitudinal study. Of the 640 households from the cross-sectional survey, 114 households with pigs aged 3–4 months were randomly selected for longitudinal sampling (as ‘sentinel pigs’). The pigs were purchased by the project and left in the farm- er’s care. Three repeat samples were obtained from the sen- Fig. 2. Study region showing sampled households in village clusters. tinel pigs, the first taken during the cross-sectional survey, the second sampling taken 3.95 (+/–0.78) months after the

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111811 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814 initial cross-sectional survey sample and the third sampling described by Fernandez-Pinero et al. [10] to confirm the taken 3.8 (+/–0.31) months after the first longitudinal sur- gel-based PCR results. Suspected ASFV samples that gener- vey. The sentinel pigs were either right censored due to ated Ct values less than 40 were considered positive while mortalities, agistments or sales. A sub-set of the sentinel those that yielded Ct values greater than 40 were considered pigs that survived throughout the longitudinal survey period doubtful/negative. Positive and negative amplification con- were humanely sacrificed at the end of the study and blood trols that consisted of DNA obtained from known ASFV- and tissue samples collected. All other pigs in the sentinel positive tissue samples and nuclease-free sterile water, households were also sampled. respectively, were run alongside the test samples as positive and negative controls. Investigation of ASFV in tissue Twenty-eight sentinel pigs that survived till the end of lon- Additionally, two qPCR assays were applied to tissue sam- gitudinal study were selected for sacrifice from the study ples obtained from the study area and slaughter slabs to area after the second longitudinal survey. They were confirm our results. The first PCR assay was a modification et al. humanely euthanized and tissue, blood and serum samples of the Taqman qPCR assay developed by King at the were collected for tissue sequestration investigation. All Pirbright Laboratory, Institute for Animal Health, UK [11] slaughter slabs in the study area were also identified within that was conducted at the OIE reference laboratory at the the same period and an additional 16 domestic pigs were FLI. Briefly The second qPCR assay was performed on the sampled on specific days when the animals were taken for smartcycler (Cepheid) platform in our laboratory at the slaughter, usually coinciding with village market days. Blood BecA-ILRI hub using Taqman qPCR assay protocol devel- et al. and serum sampling of pigs at slaughter slabs were per- oped by Zsak in Plum Island Animal Disease Center, formed ante mortem and tissues were further sampled post New York, USA [13] with minor modifications, as previ- et al. mortem. Additionally, during the longitudinal study period, ously described by Thomas [12]. domestic pigs were also sampled from outbreaks reported Virus isolation by veterinary authorities around the study site. Virus isolation was performed on tissues derived from Animal sampling: live pigs were physically restrained prior domestic pigs that had tested positive for ASFV upon diag- to sampling. Blood was collected from the jugular veins nosis with conventional PCR. PBMC (macrophage) cultures using BD Vacutainer needles (gauge Â length: 21Â1–1/2 used for the isolation of the virus were derived from naïve inch) into 10 ml BD Vacutainer glass serum tube and 4.5 ml domestic pigs as previously described [32]. Briefly, cells 15 % EDTA tubes (Becton, Dickinson and Company, UK). were seeded into 96-well tissue culture grade microtitre Non-EDTA blood was allowed to clot, and the serum separ- plates (volume 200 µl; 300 000 cells per well) in autologous swine serum, and incubated in a humidified atmosphere ated. Both serum and EDTA blood aliquots were dispensed into 2 ml cryo-vials (Greiner bio-one, Germany) and stored containing 5 % CO2 at 37 C. Three-day cultures were infected at a multiplicity of infection (m.o.i.) of 1 : 10 with at À20 C. The tissues sampled post mortem included ton- 10 % suspensions of ground tissues supplemented with 5 µg sils, lymph nodes, heart, lungs, spleen, kidney and liver. À ml 1 gentamycin sulfate (Bio Whittaker) and incubated for ASF diagnosis 24 h at 37 C. After inoculation, a preparation of 1 % autolo- Antibody detection: the blocking enzymatic immunoassay gous red blood cells in buffered saline (pH 7.0) was added (INgezim PPA Compac – R.1.1.PPA.K3) developed by to each well. The plates were examined for haemadsorption Ingenasa (Spain) was used in the detection of ASF antibod- over a 6-day period. The samples were blind-passaged three ies following the manufacturer’s instructions. times. Virus detection ASFV isotype-specific antibody responses DNA was extracted directly from serum, blood or 10 % sus- ASFV isotype-specific antibody responses were assessed pensions of ground tissues using the Qiagen DNAeasy using an in-house direct p62 ELISA. Briefly, ELISA plates blood and tissue kit (Qiagen, CA) following the manufac- were coated overnight at 4 C with 100 ng recombinant p62 turer’s instructions. Positive and negative extraction con- antigen in 100 µl per well carbonate/bicarbonate coating trols that consisted of a known ASFV positive blood sample buffer, after which they were washed four times with 200 µl and sterile PBS (pH 7.0), respectively, were included in the per well PBS with 0.05 % Tween-20 (PBS-T) and blocked DNA extraction process to ensure that the protocol worked with PBS-T with 5 % skimmed milk (PBS-M) for 1 h at effectively and also to check for possible contamination. A 37 C. After a second wash step, 100 µl per well serum hot-start gel-based conventional PCR assay using the ASF diluted 1 : 100 in PBS-M was added to the plate in triplicate diagnosis primers PPA1/PPA2 that target the ASFV VP73 and incubated for 1 h at 37 C. Following a third wash step, (p72) coding region [31] was used for primary detection of horseradish peroxidase (HRP) conjugated anti-pig isotype the presence of ASFV in the extracted DNA. ASFV-positive specific antibodies [anti-Pig IgA, M and G (Thermo Scien- samples were determined by the amplification of a 257 bp tific, Rockford, IL, USA)], were added in 100 µl per well product that was visible on agarose gel using a UV transillu- PBS-M (1 : 5000) and incubated for 1 h at 37 C. An irrele- minator. A UPL real-time PCR assay, which is more specific vant species HRP conjugate was used as a negative control. and sensitive than the conventional PCR, was conducted as After a final wash, 50 µl of TMB was added to each well and

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111812 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814 incubated at room temperature for approximately 10 min, 13. Zsak L, Borca MV, Risatti GR, Zsak A, French RA et al. Preclinical after which the reaction was stopped with 3 N sulphuric diagnosis of African swine fever in contact-exposed swine by a real-time PCR assay. J Clin Microbiol 2005;43:112–119. acid. Optical density at 450 nm was determined on an 14. Muhangi D, Masembe C, Emanuelson U, Boqvist S, Mayega L ELISA plate reader (BioTek Synergy HT, BioTek Instru- et al. A longitudinal survey of African swine fever in Uganda ments, Winooski, Vermont, USA). reveals high apparent disease incidence rates in domestic pigs, but absence of detectable persistent virus infections in blood and serum. BMC Vet Res 2015;11:106. Funding information 15. Oura CA, Edwards L, Batten CA. Virological diagnosis of African We thank the Australian Agency for International Development swine fever–comparative study of available tests. Virus Res 2013; (AusAID) for funding this research (grant number 059/11) and the 173:150–158. partners who managed the research programme: CSIRO, the Bioscien- ces East and Central Africa (BecA) Hub and the Animal Biosciences 16. Gallardo C, Soler A, Nieto R, Carrascosa AL, de Mia GM et al. programme at the International Livestock Research Institute, Nairobi. Comparative evaluation of novel African swine fever virus (ASF) antibody detection techniques derived from specific ASF viral Acknowledgements genotypes with the OIE internationally prescribed serological We are grateful to members of pig-keeping households in Kenya and tests. Vet Microbiol 2013;162:32–43. Uganda for contributing information that has helped build our under- 17. Atuhaire DK, Afayoa M, Ochwo S, Mwesigwa S, Mwiine FN et al. standing of the epidemiology of ASFV in free-range systems, and to Prevalence of African swine fever virus in apparently healthy the many members of the ASF project team who helped to collect domestic pigs in Uganda. BMC Vet Res 2013;9:263. the data. 18. Owolodun OA, Obishakin ET, Ekong PS, Yakubu B. Investigation of Conflicts of interest African swine fever in slaughtered pigs, Plateau state, Nigeria, The authors declare that there are no conflicts of interest. 2004-2006. Trop Anim Health Prod 2010;42:1605–1610. 19. Gallardo C, Mwaengo DM, Macharia JM, Arias M, Taracha EA Ethical statement et al. Enhanced discrimination of African swine fever virus iso- All farmers participating in the study were required to sign a consent lates through nucleotide sequencing of the p54, p72, and pB602L form, based on those in use by both CSIRO and ILRI, which adhere to – internationally recognized guidelines. (CVR) genes. Virus Genes 2009;38:85 95. 20. Gallardo C, Okoth E, Pelayo V, Anchuelo R, Martín E et al. African References swine fever viruses with two different genotypes, both of which 1. FAOSTAT. Meat Consumption in Uganda (2010). Food and Agricul- occur in domestic pigs, are associated with ticks and adult wart- ture Organization of the United Nations; 2010. hogs, respectively, at a single geographical site. J Gen Virol 2011; – 2. Oyewumi OA, Jooste A. Measuring the determinants of pork con- 92:432 444. sumption in Bloemfontein, Central South Africa. Agrekon 2006;45: 21. de Villiers EP, Gallardo C, Arias M, da Silva M, Upton C et al. Phy- 185–197. logenomic analysis of 11 complete African swine fever virus 3. Penrith ML, Vosloo W. Review of African swine fever: transmis- genome sequences. Virology 2010;400:128–136. sion, spread and control. J S Afr Vet Assoc 2009;80:58–62. 22. Bishop RP, Fleischauer C, de Villiers EP, Okoth EA, Arias M et al. 4. Penrith ML, Vosloo W, Jori F, Bastos AD. African swine fever virus Comparative analysis of the complete genome sequences of Ken- eradication in Africa. Virus Res 2013;173:228–246. yan African swine fever virus isolates within p72 genotypes IX and X. Virus Genes 2015;50:303–309. 5. Eustace Montgomery R. On a form of swine fever occurring in British East Africa (Kenya Colony). J Comp Pathol Therap 1921;34: 23. Charleston B, Bankowski BM, Gubbins S, Chase-Topping ME, 159–191. Schley D et al. Relationship between clinical signs and transmis- 6. Achenbach JE, Gallardo C, Nieto-Pelegrín E, Rivera-Arroyo B, sion of an infectious disease and the implications for control. – Degefa-Negi T et al. Identification of a new genotype of African Science 2011;332:726 729. swine fever virus in domestic pigs from Ethiopia. Transbound 24. Lichoti JK, Davies J, Kitala PM, Githigia SM, Okoth E et al. Social Emerg Dis doi: 10.1111/tbed.12511. [Epub ahead of print]. network analysis provides insights into African swine fever epide- 7. Okoth E, Gallardo C, Macharia JM, Omore A, Pelayo V et al. Com- miology. Prev Vet Med 2016;126:1–10. parison of African swine fever virus prevalence and risk in two 25. Nantima N, Davies J, Dione M, Ocaido M, Okoth E et al. Enhancing contrasting pig-farming systems in South-west and Central knowledge and awareness of biosecurity practices for control of Kenya. Prev Vet Med 2013;110:198–205. African swine fever among smallholder pig farmers in four dis- 8. Gallardo MC, Reoyo AT, Fernandez-Pinero J, Iglesias I, Muñoz MJ tricts along the Kenya-Uganda border. Trop Anim Health Prod et al. African swine fever: a global view of the current challenge. 2016;48:727–734. Porcine Health Manag 2015;1:21. 26. Nantima N, Ocaido M, Ouma E, Davies J, Dione M et al. Risk fac- 9. Agüero M, Fernandez J, Romero L, Sanchez Mascaraque C, Arias tors associated with occurrence of African swine fever outbreaks M et al. Highly sensitive PCR assay for routine diagnosis of African in smallholder pig farms in four districts along the Uganda-Kenya swine fever virus in clinical samples. J Clin Microbiol 2003;41: border. Trop Anim Health Prod 2015;47:589–595. – 4431 4434. 27. Penrith ML, Thomson GR, Bastos ADS. African swine fever. In: 10. Fernandez-Pinero J, Gallardo C, Elizalde M, Robles A, Gómez C Coetzer JAW (editor). Infectious Diseases of Livestock. Oxford, UK: et al. Molecular diagnosis of African swine fever by a new real- Oxford University Press; 2004. pp. 1088–1119. time PCR using universal probe library. Transbound Emerg Dis 28. Amimo JO, El Zowalaty ME, Githae D, Wamalwa M, Djikeng A 2013;60:48–58. et al. Metagenomic analysis demonstrates the diversity of the 11. King DP, Reid SM, Hutchings GH, Grierson SS, Wilkinson PJ et al. fecal virome in asymptomatic pigs in East Africa. Arch Virol 2016; Development of a TaqMan PCR assay with internal amplification 161:887–897. control for the detection of African swine fever virus. J Virol Methods 2003;107:53–61. 29. Amimo JO, Junga JO, Ogara WO, Vlasova AN, Njahira MN et al. Detection and genetic characterization of porcine group A rotavi- 12. Thomas LF, Bishop RP, Onzere C, Mcintosh MT, Lemire KA et al. Evidence for the presence of African swine fever virus in an ruses in asymptomatic pigs in smallholder farms in East Africa: endemic region of western Kenya in the absence of any reported predominance of P[8] genotype resembling human strains. Vet outbreak. BMC Vet Res 2016;12:192. Microbiol 2015;175:195–210.

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111813 On: Sat, 08 Dec 2018 02:07:22 Abworo et al., Journal of General Virology 2017;98:1806–1814

30. Kagira JM, Kanyari PW, Maingi N, Githigia SM, Ng’ang’a JC et al. swine fever virus in clinical samples. J Clin Microbiol 2003;41: Characteristics of the smallholder free-range pig production sys- 4431–4434. – tem in western Kenya. Trop Anim Health Prod 2010;42:865 873. 32. Malmquist WA, Hay D. Hemadsorption and cytopathic effect pro- 31. Agüero M, Fernandez J, Romero L, Sanchez Mascaraque C, Arias duced by African swine fever virus in swine bone marrow and M et al. Highly sensitive PCR assay for routine diagnosis of African buffy coat cultures. Am J Vet Res 1960;21:104–108.

Five reasons to publish your next article with a Microbiology Society journal 1. The Microbiology Society is a not-for-profit organization. 2. We offer fast and rigorous peer review – average time to first decision is 4–6 weeks. 3. Our journals have a global readership with subscriptions held in research institutions around the world. 4. 80% of our authors rate our submission process as ‘excellent’ or ‘very good’. 5. Your article will be published on an interactive journal platform with advanced metrics.

Find out more and submit your article at microbiologyresearch.org.

Downloaded from www.microbiologyresearch.org by IP: 54.70.40.111814 On: Sat, 08 Dec 2018 02:07:22