Experimental screening studies on rabies virus transmission and oral rabies vaccination of the Greater Kudu (Tragelaphus strepsiceros)

Rabies in the Greater Kudu (Tragelaphus strepsiceros) in Namibia is unique and found in such magnitude as has not been reported elsewhere in southern Africa. Reasons as to why Kudus appear to be exceptionally susceptible to rabies still remain speculative at best. Because the current severe rabies endemic in Kudus continues to have an enormous negative impact on the Namibian agricultural sector, we set out to question existing dogmas regarding the epidemiology of the disease in a unique experimental setting. In addition, we explored effective measures to protect these antelopes. Although we were able to confirm high susceptibly of kudus for rabies and sporadic horizontal rabies virus transmission to contact animals, we contend that these observations cannot plausibly explain the rapid spread of the disease in Kudus over large territories. Since parenteral vaccination of free-roaming Kudus is virtually impossible, oral rabies vaccination using modified life virus vaccines with a high safety profile would be the ultimate solution to the problem. In a proof-of-concept study using a 3rd generation oral rabies virus vaccine construct (SPBN GASGAS) we found evidence that Kudus can be vaccinated by the oral route and protected against a subsequent rabies infection. In a second phase, more targeted studies need to be initiated by focusing on optimizing oral vaccine uptake and delivery.


Material and Methods
Animals and housing conditions. Adult free-living Kudus (n = 46) were caught by mass capture or by individual darting from a helicopter on game farms in three different Conservancies and transported to the experimental holding facility on the Okosongoro Safari Ranch located about 265 km northwest of Windhoek (Fig. 1). Blood samples were taken on the day of capture to assess the immunological status of the animals.
Separate groups of 4 to 6 animals of the same gender were housed in two attached pens ("boma") (7 m × 14 m). Each boma consisted of a covered -and an open area which could be separated from each other by a sliding door. Both areas could be entered separately. Also, a sliding door could be used to connect the pen with an adjacent pen for separating single animals. The height of the indoor part of the pen was 3 m and the outdoor section was surrounded by a 3 m high wall. Large doors opening in both directions connect the pens with the central passage. Overhead catwalks enabled the staff to observe animals and facilitated the separation and manipulation of animals, e.g. immobilization for vaccination, infection and blood sampling. For the purposes of this study, to prevent contact with free-roaming wildlife from adjacent areas, the entire experimental holding facility and surrounding area was secured by a single 3.2 m high wire mesh game proofed fence (Fig. 2). Furthermore, no other animals were held in the experimental facility during the entire study period. The site was guarded 24 h per day.
Animal feeding and prophylactic treatment. Upon arrival, the Kudus were treated with doramectin (Dectomax −1% solution, Pfizer, Sandton, South Africa) against internal and external parasites and received multivitamin preparations (Kyrovite B Co Super, Kyron, Johannesburg, South Africa) and metabolic stimulant injections (Kyrophos Metabolic-V, Kyron). During the period of confinement animals received a diet of lucerne straw (Medicago sativa) and game cubes (Boskos WESenterprises, Thabazimbi, South Africa) supplemented with pods of the camel thorn tree (Acacia erioloba) and umbrella thorn tree (Acacia tortilis). When the supply of Boskos was exhausted it was replaced by Standard Game Cubes alternated with Game Cubes plus Ivermectin (Feedmaster, Windhoek, Namibia). The provision of acacia pods was particularly reduced during the hot dry season when these fruit became unobtainable. Pens were cleaned daily by removing both droppings and uneaten feed Transmission studies. For transmission studies, in total 21 Kudus (Table 1) were held in one group of six animals (group A) and 3 groups of five animals each (groups B-D, Table 2). Prior to infection, animals were adapted to captivity for an 80 days period. One animal in each group was infected with a high (10 5.3 TCID 50 , N = 4) dose, while another animal was inoculated with a low (10 3.3 TCID 50 , N = 4) dose of a Kudu rabies virus isolate (see below). The remaining conspecifics served as contact animals (group A -N = 4; groups B-D -N = 3) (Tables 1, 2). The challenge virus was administered by intramuscular (IM) injection bilaterally in the masseter muscle at a volume of 0.5 ml per site. Upon infection, the animals were observed at least twice daily for a period  To prevent contact with free-roaming wildlife from adjacent areas, the entire experimental holding facility and surrounding area was secured by a single 3.2 m high wire mesh game proofed fence. Overhead catwalks enabled staff to better observe and manipulate animals (B). Individual pen ("boma") (7 m × 14 m) consisting of a covered -and an open area (C). The sliding door could be used to connect the pen with an adjacent pen for separating single animals. The height of the indoor part of the pen was 3 m and the outdoor section was surrounded by a 3 m high wall. of 261 days and the development of clinical signs was recorded (Fig. 3). According to the animal welfare protocol, immediately at the onset of the first clinical signs, the animals were to be sedated and humanely euthanized by a supervising veterinarian with an overdose of thiafentanil oxalate (10 mg/ml Thianil, Wildlife Pharmaceuticals, Windsor, USA). Clinical signs included bellowing, throwing head back, persistent swallowing movements, persistent licking, paresis and paralysis, hypersalivation or frothing at the mouth. Point survivors were ear-tagged, vaccinated parenterally against rabies and rehomed on the spot at the end of the observation period.
Vaccination studies. Twenty five Kudus were used for vaccination studies and held in 5 groups of four animals each and one group of five animals (Table 1). Prior to vaccination animals were allowed to adapt to captivity for 24 days. Ten animals (7 females and 3 males) in three groups received approx. 5.0 ml of the oral rabies virus vaccine construct SPBN GASGAS (10 8.1 FFU/ml) by direct oral administration (DOA). As positive controls, 9 female Kudus divided over 3 groups received 2.0 ml of a commercial inactivated rabies vaccine (Rabisin, Merial, France) by the IM route (Table 1). All vaccinated animals were challenged on the same day together with the animals from the transmission study (56 days post vaccination (p.v.)) with the high dose (10 5.3 TCID 50 ) Kudu RABV isolate using the same route of administration and following the same termination criteria as described above. Survival of vaccinated animals was followed over a study period of 183 days post infection (p.i.) (Fig. 3).
Sampling. Serum samples (B) were collected at different time points prior, p.v. and p.i. to investigate the development and kinetics of rabies induced antibodies (Fig. 3). Initial blood sampling was on the day of capture (B0); additional blood samples were taken on day 28 p.  20 . It is postulated that the overexpression of the RABV G increased not only its efficacy but also its safety profile by reducing potential risk of reversion to virulence and enhancement of apoptosis 21,22 . The vaccine virus was propagated on BHK21 BSR Cl13 cells and harvested after 48 hours. The cell-cultured vaccine material was subsequently 5 times ultrafiltrated and stabilized (140 ml antigen + 60 ml GS8) to reach a final titre of 10 8.1 FFU/ml. The challenge RABV virus used (Lab ID 23079) was initially isolated from the brain of a naturally infected Kudu (240K09, GenBank accession JX473841) and genetically characterized 15 . After three serial passages on mouse neuroblastoma cells (NA42/13), the inoculum had a titre of 10 3.3 MICLD 50 /ml and 10 5.3 TCID 50 /ml when assayed by intracerebral inoculation of mice and cell culture, respectively. Verification of the genetic identity of the RABV isolate after passaging in cell culture using next generation sequencing 23 revealed an additional insertion of three nucleotides at positions 2475-2477 (AAC) in the intergenic region between the phosphoprotein (P) and matrix protein (M) gene. This specific sequence variation was confirmed by conventional Sanger sequencing of two independent PCR amplicons (Supplementary material) and let to the extension of the 6 A transcription stop signal by 2 additional As, while the following intergenic region starts with an additional C comprising 7 nucleotides (nt).  Informed consent. The experiments did not involve human participants (including the use of human tissue samples). Therefore, informed consent is not needed.
Diagnostic Assays. The presence of RABV antigen in brain tissue of Kudus was detected by fluorescent antibody test (FAT) as described previously 24 using both polyclonal (OIE-RL Onderstepoort, South Africa) and commercial FITC-labeled monoclonal (SIFIN, Berlin, Germany) anti-rabies antibodies. Defined positive (PC, positive fox brain) and negative controls (NC, naïve cattle brain) were included in every test run. FAT indeterminate results were confirmed by realtime polymerase chain reaction (rt-qPCR) 25 and the rabies tissue culture infection test (RTCIT) 26 , the latter with three consecutive passages to confirm a negative result. Sera were tested for the presence of rabies specific antibodies using three different serological assays with modified cut-offs. Virus neutralizing antibodies (VNAs) were detected in a modified rapid fluorescent focus inhibition test (RFFIT) essentially as described 27 using the calibrated WHO international standard immunoglobulin (2nd human rabies immunoglobulin preparation, National Institute for Standards and Control, Potters Bar, UK) adjusted to 0.5 international units (IU) and a naive bovine serum as PC and NC, respectively. VNA titres were calculated using inverse interpolation as described and expressed in international units (IU/mL) 28 . Presence of rabies specific binding antibodies were tested using a commercial blocking ELISA (BioPro Rabies ELISA, Czech Republic) strictly following manufacturer's instructions. A study evaluating this ELISA kit found 100% specificity and 95.5% sensitivity with fox and raccoon dog sera; and 95% concordance with an assay measuring VNAs (the Fluorescent Antibody Neutralization Test) 29 . None of the serological assays employed in this study have been fully evaluated with Kudu sera, and a previous publication has demonstrated variable cut-off levels per species 27 . To account for lack of a unique identifiable cut-off in the assays for Kudu sera, besides sero-positive and sero-negative results indeterminate values were considered within a range of variation around the presumed  cut-off of positivity (0.5 IU/mL +/− 0.25 IU/mL for RFFIT and 40% + 4.41%/−5.28% for BioPro ELISA) as described previously 27 . As a versatile and practical method for measuring rabies specific antibodies p.v., additionally the RAPINA test 30 was applied as a 3 rd method by following instructions of the manufacturer. Final scoring if a serum sample was deemed antibody negative or positive was based on a weighted analysis of all results obtained by the 3 different assays without giving an absolute value to it. ELISA and RFFIT results were given more weight (++; −−; ++/−−) than the RAPINA test (+, −, +/−), for the latter is not considered a standard test for rabies serology yet; whereby ++ or + is sero-positive, −− or − is sero-negative and +/− or ++/−− indeterminate. Every + received a score of +1 and every − a score of −1. The total score was calculated by adding the values; a negative score (<0) was considered sero-negative and a positive score (>0) as sero-positive. FAT testing was done in parallel at the Central Veterinary Laboratory (CVL) Windhoek and the Friedrich-Loeffler-Institut (FLI), Greifswald-Insel Riems, while rt-qPCR, RTCIT and all serological assays were exclusively conducted at FLI after the end of the experimental study.

Statistical Analyses.
To infer statistical differences in the serological results, the Fisher's exact test was applied, whereas for survival rates the Mantel-Cox test (log-rank test) was used as implemented in GraphPad Prism version 7.00 (GraphPad Software, La Jolla California USA), with p-values < 0.05 considered significant.

Results
Because of illnesses and severe injuries (fractures) as a result of capture stress, 7 animals had to be removed at various times throughout the study; one animal (K06) in the transmission group and 6 animals (K26, K25, K30, K37, K44, K45) in the vaccination groups between 26 and 78 days prior to infection (Tables 1-3). As a release of those animals back into the wild could not be justified, for reasons of animal welfare they had to be humanely euthanized. None of the animals showed clinical signs suggestive of rabies.
Transmission studies. While all four animals (K02, K12, K13, K18) inoculated with the high dose Kudu RABV isolate (10 5.3 TCID 50 ) were seronegative at B0 and succumbed to infection between day 12 and 16 p.i., all but one low dose infected (10 3.3 TCID50) Kudus (K03, K15, K16) died or had to be euthanized between day 93 and 245 p.i. (Table 1, Fig. 4). Two of the latter animals (K15, K16) had rabies specific antibodies at B0 (Tables 2,  S2). The Kudu (K21) that survived the low dose challenge virus infection seroconverted at day 261 p.i. Clinical signs included swallowing movements associated with moderate salivation, paresis, and mild frothing at the mouth in two cases. The remaining animals in the group mostly tried to stay away from and avoid contact with those displaying clinical signs. Six animals that succumbed to the infection were FAT positive while two were regarded FAT inconclusive, with the latter two being positive in RT-qPCR. All but two of the contact animals survived the 261 days observation period. Two of 12 contact animals died 99 (K11) and 156 days p.i. (K05), however, only in the latter were viral antigen and RNA detected by FAT and PCR, respectively (Tables 1, 2, Fig. 4). Sequencing of the RABV isolate of the FAT positive contact animal revealed a 100% sequence identity with the challenge virus including the insertion.
There was a significant difference (Fisher's exact test, p = 0.0031) in the parenteral versus the DOA vaccinated group. In general, with < 31.4% the DOA vaccinated animals showed only a 1.3 fold increase in ELISA mean PB values at days 28 and 56 p.v. compared to day 0 and remained below the ELISA cut-off of positivity. Though there was a 1.7-2.0 fold increase compared to day 0 the GMT of VNAs of orally vaccinated animals remained below 0.5 IU/ml until challenge (Tables 3, S3, Fig. 5).
Only 3 of 10 Kudus (K40, K43, K46) vaccinated by the oral route survived the challenge (30%) ( Table 2, Fig. 4). While two of these animals remained seronegative prior to infection, one Kudu (K46) had detectable levels of antibodies at the time point of capture but turned seronegative at days 28 and 56 p.v.. After challenge infection all three survivors developed high levels of rabies specific antibodies; the GMT of these individual animals increased to 1.27 IU/ml at day 183 p.i., resembling a 5.8 and 2.9 fold increase compared to day 0 and day 56 p.v., respectively (Tables 3, S3, Fig. 5). Of the 7 initially seronegative (B0, time point of capture) DOA vaccinated Kudu that succumbed to infection, two animals (K35, K36) developed rabies specific antibodies p.v.. All animals died between 12 and 26 days p.i. (Table 1, Fig. 4). Six animals were FAT-positive, while one FAT-negative animal tested positive in rt-PCR. There was no (Log-rank/Mantel-Cox test, p = 0.0786) difference in the survival curves of orally vaccinated vs. high dose infected animals from the transmission study.

Discussion
There is no other country in the world with such a large population of Kudus than Namibia. The high population density as a result of game farming is thought to favor rabies epidemics among this species. Experiments involving large, highly stress-sensitive wild-caught species like Kudus are typically limited by space and cost. Also, the lack of suitable research facilities able to house these large antelopes under quarantine conditions restricted this study to a relatively remote holding site that is normally used for temporary housing of wild game for sale and auctioning, hence complicating observations, sampling, sampling storage and -transportation. The precautions and safety measures specifically developed for the purpose of the study were of high standard and have proven to be useful in preventing contact to free-roaming wildlife in adjacent areas (Fig. 2).
While the relatively short incubation periods of 12-26 days observed after high dose infection are within the lower bounds of the expected range and are comparable to those of carnivorous reservoir species, Kudus and cattle 12,19,[31][32][33][34] , the extremely long incubation periods of 93-245 days after low dose infection are remarkable, particularly against the background of the same route of infection and the fact that the difference in viral dose was just two log 10 steps (Fig. 4). However, long incubation periods using a similar dose have also been reported for skunks, golden jackals, raccoons and foxes [35][36][37] . In addition, the outcome of the transmission studies showed that depending on the viral dose Kudus are able to survive an infection with RABV and hence, confirm previous observations 12 . The animal (K21) that survived infection never showed clinical signs, indicating that the virus was recognized and cleared by the immune system either prior to movement into peripheral nerves or prior to infection of the central nervous system. This finding may also explain why several animals were sero-positive at the day of capture (Tables 2, 3, S2, S3).
The sheer magnitude of the epizootic and phylogenetic data are believed to support horizontal transmission and maintenance of a rabies cycle within this species 1,8 . However, despite anecdotal evidence of horizontal rabies transmission among Kudus, to this day this hypothesis has still not been definitively confirmed in a larger experimental setting. The hypothesis is based on the observation of hypersalivation and high viral titres in saliva of rabid Kudus, their grooming habits and in particular experimental studies 11,12 . Although artificial exposure to infection showed that 2 out of 4 Kudus died of rabies after experimental infection by instillation of infected saliva onto their buccal mucosae, intranasal (i.n.) infection cannot be excluded to have provoked the disease because the saliva was also instilled into the nasal cavity at the same time 12 . In fact, i.n. administration is a very effective way of delivering RABV directly to the brain via the olfactory nerve 38,39 . Our study seems to corroborate the possibility of horizontal transmission as the virus isolated from the contact animal possessed the unique marker of the challenge strain used. Since the length of the stop signals and intergenic regions affect polymerase termination and downstream re-initiation 40 potential effects of this three nt insertion in the intergenic region between P and M gene on virus replication cannot be excluded. However, since selection of the additional insertion in the course of limited cell culture passages is more likely to support virus replication and the glycoprotein has been identified as the major pathogenicity factor 41 , major effects on the virulence of the challenge virus are unlikely. Also, the observation that 3 of 8 contact animals (K04, K08, K10) considered seronegative at the time point of capture seroconverted 261 days p.i. (Tables 2, S2) could be considered evidence for exposure with infectious saliva from inoculated Kudus which succumbed to the disease. Unfortunately, the relationship between dose and strength of priming using street RABVs is not well characterized in both bats and other wildlife 42 . Although our experimental settings allowed natural social and feeding behavior to the greatest possible extent, Kudus had to be humanely euthanized early after showing clinical signs. This may have limited the likelihood of successful onward transmission leading to horizontally acquired rabies. The fact that only one of 12 contact animals (K05) died of rabies needs careful interpretation. If our observations resemble naturally occurring infections it would suggests that the extent of horizontal transmission under field conditions is less than is actually believed. Even if the transmission rate was underestimated and acquired seropositivity of contact animals (K04, K08, K10) (Tables 2, S2) was also considered as evidence of transmission, the latter being highly debatable, local horizontal transmission cannot plausibly explain the recent epidemics and rapid spread of the disease in Kudus over large territories. In this sense, long incubation periods might add to the situation explaining why rabies in Kudu is resurgent after months in areas where it is believed to have disappeared. Interestingly, high number of cattle rabies cases and other small ruminants in areas where Kudu rabies is endemic (unpublished) also bring into question the role of the Kudu as a sole reservoir. While the phylogenetic work of Scott and co-workers 15 seem to support evidence for independent horizontal transmission among Kudus, this is contradicted by an earlier study which suggest that jackal and Kudu may form part of the same epidemiological cycle of rabies in Namibian wildlife 43 . From an epidemiological point of view, a combination of spill-over events and perhaps locally restricted horizontal transmission cannot be excluded, however, has not been considered or discussed thus far. Also, other epidemiological aspects including management, translocation, natural movement behavior and other parameters would definitely need a more thorough investigation and consideration to unravel the mystery of Kudu rabies in Namibia.
Expanding the concept of oral vaccination against rabies as successfully developed for wild carnivores to Kudus seems to be a feasible approach considering that commercial products for oral immunization of ruminants already exist; e.g. for immunization of calves in controlling diarrhea caused by bovine rotavirus and -coronavirus (Calf-Guard ® , Zoetis). However, mucosal immunization through oral delivery is often compromised by antigen degradation in the stomach, especially considering the complex gastro-intestinal tracts of larger ruminants 44 . Therefore, to assure antigen uptake at Peyer's patches of the intestine additional substances are often added to oral vaccine formulations protecting the antigen against enzymatic and proteolytic degradation. However, the gut is not the only site with mucous-associated lymphoid tissue (MALT) containing lymphoid follicles and M-cells. MALT can also be found in nasal and oral cavities, whereby tonsils form a major component in the latter. Recent studies have indicated that for oral vaccination of meso-carnivores against rabies the palatine tonsils are a major site of vaccine uptake [45][46][47] . Palatine tonsils are also present in ruminants like cattle, sheep and goats 48 . Experimental studies in wild ruminants like white-tailed deer (Odocoileus virginianus) showed that DOA of BCG-vaccine targeting the pharyngeal lymphoid tissue was able to induce an immune response against bovine tuberculosis 49 . Also, oral administration of heat-inactivated Mycobacterium bovis to red deer (Cervus elaphus) induced some level of protection against bovine tuberculosis 50 . In our proof-of-principle experiment, 3 of 10 animals orally vaccinated with SPBN GSAGAS survived a severe rabies challenge infection with a high challenge dose (Table 1, Fig. 4). Even though no statistical significance in survival between unvaccinated and DOA vaccinated animals was found, the survival of Kudu suggest that oral vaccination may elicit a protective immune response in this species.
Interestingly, none of these 3 animals (K40, K43, K46) that survived had detectable levels of antibodies in any of the assays (ELISA, RAPINA, RFFIT) p.v., however, the strong immune response p.i. is clear evidence for a prime-boost effect (Tables 3, S3, Fig. 5). It must be mentioned that the vaccine virus was evenly distributed in the oral cavity of the Kudus using a needleless syringe simulating release of vaccine when chewing on a vaccine sachet and the palatine tonsillar crypts as potential uptake site were not specifically targeted. However, unfortunate spillage of vaccine virus could not be prevented as most of the anesthetics including thiafentanil oxalate induce salivation in wild herbivores and in combination with grinding often produce frothing around the mouth 51 . So this per se makes DOA in these species difficult. On the other hand, one can argue that 30% survival in orally vaccinated Kudus (Table 1, Fig. 4) after challenge does not appear very convincing. However, it is the first time that such a proof-of-concept study has ever been conducted in an herbivorous species such as Kudus. The vaccine strain SPBN GASGAS has shown to be efficacious in several animal species like red foxes (Vulpes vulpes), raccoon dogs (Nyctereutes procyonoides), raccoon (Procyon lotor) and small Indian mongoose (Herpestes auropunctatus) 19,52,53 . However, striped skunks (Mephitis mephitis) seem to be rather refractory to vaccination through the oral route in a vaccine virus titer dependent manner 47 . It cannot be excluded that this could also apply to Kudus. Performance of serological assays used and the interpretation of the serological results obtained presented problems. While the RFFIT is an OIE and WHO recommended test for detection of VNAs as for response to vaccination in humans and animals (particularly in connection with international travel of pets), the ELISA and RAPINA test have been developed for qualitative and semi-quantitative detection of rabies antibodies in foxes and raccoon dogs 29 , and humans and dogs 30 , respectively. None of these serological assays, however, have been validated for Kudus. Therefore, we modified the thresholds of positivity by introducing an additional 'grey zone' for inconclusive or indeterminate results based on analysis by Moore et al. 27 . The subsequent weighted scoring allowed a clear identification of sera as being seropositive and negative without giving an absolute value to it and by avoiding classifying sera as indeterminate. Of note, additional results from another ELISA (Platelia II, Bio-Rad, Tables S2, S3) 54,55 was excluded from the analysis as Protein A does not bind well to IgG from bovine and equine species 56 .
Using this approach, only the results obtained p.v. with the animals vaccinated by the parenteral route showed a perfect correlation between seroconversion and protection for both serologic assays. However, for the animals that received the vaccine by the oral route, the serology results are more difficult to interpret (Tables 3, S2, Fig. 5) (see above). While 3 DOA vaccinated but seronegative animals survived challenge, two Kudus with supposedly acquired or preexisting antibodies at the day of challenge (K35, K36) succumbed to infection indicating a non-protective immunity (Tables 3, S3). Perhaps, the measured humoral antibody response prior to challenge is not of sufficient specificity or not associated with cellular immunity to prevent a lethal infection. The results of this study somewhat corroborate results obtained in studies on rabies repeated challenge in bats 42 and suggest that a single exposure may not protect Kudus significantly against subsequent infections. These findings also underscore that diagnostic assays validated (specificity and sensitivity) for samples from certain species are not automatically suitable for samples from other species, as was recently shown for serology by Moore et al. 27 .

Conclusions
Kudus can be vaccinated by the oral route and protected against a subsequent rabies infection, although it seems that they are rather refractory to this route of vaccine administration. In any case, further studies need to be initiated to optimize oral vaccine uptake and delivery of this 3 rd generation attenuated oral rabies vaccine. Alternatively, recombinant rabies virus vaccines expressing the RABV glycoprotein could also be considered in future studies 57 . For the time being the minimum effective titer of both attenuated and recombinant vaccine viruses required to efficiently immunize the animals is not known yet. Hence, further research should investigate how vaccine uptake effectiveness can be improved, for example by increasing vaccine titre, vaccination intervals or adding muco-adhesive substances. Attractive baits for oral vaccination of Kudus have been developed already 58 , however, bait delivery systems need to be optimized in case vaccine potency can be enhanced in this species. Also, validation of serological assays for Kudus is required to make better informed decisions on the immune status.