Foot-and-mouth disease (FMD) is a highly contagious viral disease affecting a wide range of domestic and wild cloven-hoofed animals [1, 2]. It is known to cause substantial economic losses, directly from the effect of the virus on animal health and indirectly through control efforts including quarantines and trade restrictions [2].

The disease is caused by infection with a single stranded, positive sense, RNA virus (FMDV) belonging to the genus Aphthovirus within the family Picornaviridae. This virus occurs in seven distinct serotypes namely: O, A, C, Asia 1, SAT 1, SAT 2 and SAT 3, each having multiple strains [3, 4]. All serotypes, except Asia 1, have been reported in eastern Africa, however, serotype C was last detected in Kenya in 2004 and may be extinct [5, 6]. Serotypes O and A are considered endemic in most parts of Africa, while the three SAT serotypes have usually been restricted to sub-Saharan Africa [7], where they are commonly diagnosed [7–9]. However, incursions of SAT 1 into the Middle East (1961–1965 & 1970), and SAT 2 into Saudi Arabia (2000), Libya (2003) and more recently into Egypt and Libya (2012) have been reported [7, 10].

The maintenance of the three FMDV SAT serotypes in cattle in southern and eastern Africa has been associated with contact with wildlife, especially African buffalo (Syncerus caffer) [11]. FMDV has been demonstrated to persist in individual buffalo for up to 4–5 years and in isolated buffalo herds for up to 24 years [11]. In southern Africa, it has, in some cases, been concluded that buffalo have been the source of FMDV infections in cattle and impala (Aepyceros melampus) [8, 12, 13]. As a result, FMD control policies in southern Africa are centred on isolating the buffalo, in the National Parks and game reserves, away from the surrounding livestock populations, and vaccinating cattle kept in buffer zones around these areas against the FMDV serotypes that are carried by the nearby buffalo [14].

In eastern Africa, FMD was first reported in cattle in 1932 [15] and since then, the marketability of livestock and animal products within, and outside, the region has been negatively affected [16]. The maintenance of FMDV in the region has been attributed to various factors including: the presence of numerous wildlife reservoirs [7, 17–19], communal and pastoral grazing systems that enable contact between livestock and wildlife [20], poor diagnostic capacity in the region [9], emergence of new strains [21–23], transboundary animal movements and traditional cultural practices such as the exchange of live animals for dowry and gifts [24, 25]. East African countries have tried to control the disease mainly by quarantine and post outbreak vaccination. However, these measures have been difficult to enforce and suboptimal use of vaccines [26] has resulted in large parts of the FMDV susceptible animal populations remaining at risk [9, 27]. Recently, the countries in this region have established a network to combine their control efforts to move towards freedom from FMD by 2020 following the progressive control pathway (PCP) for FMD as described [28]. The PCP defines six steps in the process of achieving freedom from FMD including effective monitoring of circulating serotypes, vaccination and improved biosecurity measures [28]; this strategy has been adopted by all countries in the region, including Uganda [29].

In Uganda, SAT 1 FMDV was first reported in 1959, and since then, FMD outbreaks caused by serotypes O, A, SAT 1 and SAT 2 [7, 9, 24, 27, 30–33] have been regularly reported in livestock. In addition, for the first time, serotype SAT 3 FMDV has been recently isolated from a clinically healthy long horned Ankole calf shortly after its introduction into the area close to the Queen Elizabeth National Park (QENP) [34]. Ugandan buffalo are only infrequently sampled for FMDV, and thus the virus has only been isolated from them during two studies in QENP; these yielded serotypes SAT 1 and SAT 3 in 1997 [35] and SAT 1 and SAT 2 in 2007 [36]. Moreover, in a post-outbreak study conducted on adult cattle and goats in the same area, concurrently with the 2007 buffalo sampling, only serotype O FMDV was isolated. Furthermore, 11 herds of cattle had animals with higher antibody titres against serotype O than against the SAT serotypes, while only one had convincing serological evidence of infection with SAT FMDV [22, 31]. Thus, although wildlife within the Ugandan National Parks are often blamed for being the source of FMD outbreaks in their surroundings [37], this could not be proven for the major 2006 FMD outbreak in the QENP area [31, 36]. Moreover, although it is clear that FMDV can pass from buffalo to cattle [13], recent attempts in Uganda and Kenya to establish the role of the African buffalo in the epidemiology of FMD in domestic species in eastern Africa have so far not settled this [36, 38].

The objective of this study was to assess the presence of FMDV in young, unvaccinated cattle grazing in close proximity to African buffalo in the QENP (Fig. 1) and to establish whether these buffalo play a major role in the maintenance of FMD in livestock around the National Park.

This study was designed in compliance with REFLECT guidelines and was approved by the Ugandan Ministry of Agriculture, Animal Industries and Fisheries (MAAIF) (Reference LHE 199/01, Uganda) and Makerere University (MU) “School Higher Degrees and Research Committee”. The sampling was carried out by MAAIFs National Animal Disease Diagnostics and Epidemiology Centre (NADDEC) in collaboration with Kasese District Veterinary Office in accordance to the guidelines from MAAIF. The study included meetings for the herd owners where the purpose of the project was described and informed consent to the sampling, and to participation in the questionnaire survey, was obtained from the individual herd owners.

Clinical signs of FMD were not observed in any of the sampled animals or herds.

Characterisation of FMDV isolates

Sequencing revealed that the two 657 nucleotide (nt) long regions, from the 731 bp amplicons, corresponding to the VP1 coding sequences of UGA/116/13 and UGA/161/13 were identical. Alignment of these complete VP1 sequences with 18 other African FMDV isolates (Table 1) showed that they belonged to the SAT 1 topotype IV (EA-1) (see Tekleghiorghis et al. [52]), clustered together with three previous Ugandan buffalo isolates of 1970 and 2007 (UGA10/70, UGA/21/70 and UGA/1/07) and also the UGA/7/99 isolate whose species of origin was not identified (Fig. 2). The VP1 sequences of the 2013 isolates had 21 %, 20 % and 24 % nucleotide (nt) difference from these earlier buffalo isolate VP1 sequences, respectively, while their relationship with another Ugandan buffalo isolate from 1997 (UGA/1/97) and an earlier Ugandan SAT 1 cattle isolate (UGA/13/74) was even more distant (nt sequence divergences of 30 % and 31 %, respectively). However, except for the UGA/13/74 strain [topotype VII (EA-2)] and UGA/1/97 [topotype VIII (EA-3)], the various Ugandan isolates represent independent lineages within topotype IV (EA-1) (Fig. 2). The SAT 1 vaccine strain (TAN/155/71) used in Uganda belongs to a different topotype I (NWZ) and differs from the SAT 1 VP1 sequences obtained from this study by 26 % and 21 % at the nucleotide and amino acid level, respectively.

Since the VP1 sequences of the two isolates from H18 were identical, the 20 aligned SAT 1 VP1 coding sequences (Table 1) encoded 19 different amino acid sequences, each with 219 amino acid residues (Fig. 3), which were conserved at 51.3 % of the sites. Multiple amino acid substitutions were observed within the known hypervariable regions [53]. In the G-H loop, compared to the vaccine strain (TAN/155/71), residues 137 and 149–151 (the receptor binding RGD motif) remained conserved in all the 19 sequences, while multiple substitutions occurred in the section including residues 138–148 which is known to be a key antigenic region in FMDVs [53]. The seven Ugandan isolates were conserved at residues 146, 152 and 153 with each having a distinct amino acid (aa) substitution at positions 140–143. UGA/116/13 showed full aa conservation with UGA/7/99, UGA/1/07, UGA/13/74 and UGA/21/70 at positions 138, 139, 144 and 145, while the UGA/1/97 and UGA 10/70 strains had different substitutions (Fig. 3). UGA/116/13 had multiple substitutions at positions 140(N → A), 141(H → D), 142(E → Q), 143(T → A), 147(H → N) and 148(I → V) compared to the vaccine strain (TAN/155/71) and distinct substitutions in these positions were also observed in the other Ugandan isolates but without a pattern. In the H-I loop (residues 172–182), all the 19 sequences were fully conserved at positions 172, 176 and 181–182. However, there were distinct substitutions among the Ugandan isolates within this region. Near the C-terminus (residues 203–219), five positions (203, 208, 216, 218 and 219) were conserved in all the 19 isolates, while the Ugandan isolates also showed full conservation at positions 209, 210 and 213 plus almost complete conservation at positions 207, 211 and 217 (Fig. 3). The UGA/116/13 differed from the other 6 strains by having substitutions outside the G-H loop (139–153), the H-I loop (172–182) and the C-terminus (203–219) (see Fig. 3).

At the QENP livestock-wildlife interface, young unvaccinated cattle from 24 herds without reports of clinical signs of FMD had an overall seroprevalence of 15 % (37/247) for antibodies against FMDV NSPs. This is higher than the 5 % seroprevalence reported in cattle herds without clinical signs in the same area in 2006 [31] but compares well with the 12 % value reported in young calves in FMD-endemic Ethiopia [41].

Evidence for prior FMDV infection varied between herds, thus evidence for exposure to FMDV was absent in ten herds, scanty in eight herds and more substantial in six herds (25 %). The observed scattered presence of antibodies does not appear to fit well with the general appreciation of the epidemiology of FMD as an efficiently spreading disease with high morbidity in affected herds and high risk of transfer to other herds through physical contact and fomites [54]. Moreover, although the farmers were well acquainted with clinical signs of FMD, a number of them would not report outbreaks, but rather used traditional control measures like application of local herbs or walking the animals through the salt lakes known to contain common salt and sodium bicarbonate [55].

There is an undefined level of trade with cattle, including young stock for fattening, coming into the area from other districts; thus individual animals with antibodies against FMDV may be explained as animals infected elsewhere and brought into the herds as replacement stock or gifts to the owners as part of local customs [32]. However, the higher antibody prevalences in H7 and H19 and the isolation of SAT 1 FMDV from OPs from two cattle in H18 support spread within these herds (originating either from within the district or from other districts); moreover, it cannot be excluded that the infections causing antibodies in the other herds have taken place in Kasese district.

Another possibility is that some of the antibodies against FMDV observed in this study are residual maternally transferred antibodies. Since cattle in this area are likely to have their first calf at 3–4 years of age and to have several calves, it is likely that some of the dams to the sampled calf cohort (born between August 2009 and February 2011) were exposed to FMDV by infection and/or vaccination during the 2006–2007 outbreak [31, 56]. A fraction of these dams may have retained sufficient level of antibodies, especially against the NSPs, to transfer these to their offspring. Moreover, although FMD outbreaks had not been reported in the area since 2006, the District Veterinary Office noted that underreporting was suspected since quarantine restrictions due to FMD would paralyze all economic activities in the area.

Although it is not clear when, or how, some of the antibody positive animals in these herds were infected, SPBE assays showed that 11 of the 37 NSP positive samples had antibody titres ≥ 80 against one or more serotypes. Based on SPBE and VNT results from the few positive animals, it appeared that serotype O FMDV had infected some animals from H1, H14 and H19, while SAT 1 FMDV had infected some animals from H3 and possibly H18, in the latter case this was confirmed by isolation of SAT 1 FMDV from two animals. The situation was unclear in herds H19 and H23 with high SPBE antibody titres against serotypes O, SAT 1, SAT 2 and SAT 3, which could be due to the previously observed cross-reactivity between the SPBEs and also between the VNTs [31, 56]. Moreover, if animals were purchased from other districts, vaccine-derived antibodies cannot be excluded either. In accordance with previous findings in the area [31, 56], there was no conclusive serological evidence for previous infection by serotypes A, C, Asia 1 and SAT 3 in the sampled cattle, although very recently SAT 3 FMDV was isolated from an asymptomatic calf from the area [34].

The SAT 1 FMDV isolates obtained from the two healthy animals in H18 are the first reported SAT 1 isolates from Ugandan cattle since 1999 [57], while SAT 1 isolates have been obtained from buffalo in QENP in 1997 and 2007 [35, 36]. The relatively infrequent isolation of SAT 1 FMDV from outbreaks in Ugandan cattle could be the result of poor surveillance in the form of underreporting of FMD outbreaks and limited submission of samples of sufficient quality for virus isolation to the central diagnostic laboratory at the National Animal Disease Diagnostics and Epidemiology Centre (NADDEC) (Dr. Anna Rose Ademun, personal communication). Moreover, there have been limited options for typing FMDV locally in Uganda [9].

The combination of rather low C_T values (22 and 23) in the 5'UTR tests for the two H18 OP samples that yielded the SAT 1 FMDV isolates and SPBE antibody titres of 40 and 80 for antibodies against SAT 1 FMDV in sera from the same animals, indicate that these two animals were in the subacute stage of the infection. However, no clinical signs were observed during the sampling and there were no reports of clinical signs in this or in other cattle herds in the area. Moreover, the questionnaire responses indicated that the indigenous Ankole cattle from this area have less severe clinical signs of FMD than the pure exotic and crossbred cattle. Hence, these data support the notion that some of the Ugandan indigenous Long-horned Ankole cattle may go through FMDV infection with only minimal clinical signs of FMD. This is consistent with the very recent isolation of SAT 3 FMDV from an OP sample from a clinically healthy young calf in the QENP area [34] and has been previously reported for indigenous cattle in Botswana [58, 59]. If true, this notion has a bearing on the passive surveillance reporting system for FMD in Uganda that is based on clinical observations.

This study shows a considerable genetic difference between the current Ugandan SAT 1 cattle virus and the vaccine strain (TAN/155/71) that is currently used to control FMD outbreaks caused by SAT 1 FMDV in Uganda, indeed these belong to different topotypes. Thus, vaccine matching studies using the currently circulating SAT 1 strain are required. In addition, the observed differences from earlier buffalo and cattle isolates imply that different SAT 1 lineages have been in circulation in this area over the last 50 years. To verify this, more Ugandan FMDV SAT 1 isolates should be collected and characterised.

Despite the close interaction that occurs between cattle and buffalo at the QENP livestock-wildlife-interface, this study did not provide evidence to support the view that the FMD viruses circulating in cattle in this area originate from wildlife as has been observed in South Africa [60]. This could be attributed to the limited numbers of isolates from cattle and the long time-intervals between their isolations; but it also does not exclude the possibility that separate pools of viruses circulate in cattle and wildlife. The VP1 coding sequences of five of the six Ugandan SAT 1 isolates were different from those of isolates from other countries, indicating that the Ugandan SAT 1 strains may be geographically limited to Uganda as previously suggested [61].

Although FMD was last reported in Kasese District in 2006, evidence of infection with FMDV was found in six of 24 cattle herds from the QENP livestock-wildlife-interface. The SAT 1 FMDV isolated from one of these six herds was markedly different from the earlier buffalo isolates from the same area; thus transmission between buffalo and cattle could not be proven. The genetic difference (ca. 26 % within the VP1 coding region) between the current SAT 1 isolates and the vaccine strain used to control FMD outbreaks in the country calls for vaccine matching to ensure the effectiveness of vaccination in this region. Additional surveillance and characterization of field strains from cattle and buffalo at this livestock-wildlife interface is necessary for monitoring of emerging strains and to enhance the understanding of the role of wildlife in the epidemiology of FMD in cattle in East Africa.