Safety and immunogenicity of four-segmented Rift Valley fever virus in the common marmoset

Rift Valley fever virus (RVFV) is an emerging mosquito-borne bunyavirus that is highly pathogenic to wild and domesticated ruminants, camelids, and humans. While animals are exclusively infected via mosquito bites, humans can also be infected via contact with contaminated tissues or blood. No human vaccine is available and commercialized veterinary vaccines do not optimally combine efficacy with safety. We previously reported the development of two novel live-attenuated RVF vaccines, created by splitting the M genome segment and deleting the major virulence determinant NSs. The vaccine candidates, referred to as the veterinary vaccine vRVFV-4s and the human vaccine hRVFV-4s, were shown to induce protective immunity in multiple species after a single vaccination. Anticipating accidental exposure of humans to the veterinary vaccine and the application of hRVFV-4s to humans, the safety of each vaccine was evaluated in the most susceptible nonhuman primate model, the common marmoset (Callithrix jacchus). Marmosets were inoculated with high doses of each vaccine and were monitored for clinical signs as well as for vaccine virus dissemination, shedding, and spreading to the environment. To accurately assess the attenuation of both vaccine viruses, separate groups of marmosets were inoculated with the parent wild-type RVFV strains. Both wild-type strains induced high viremia and disseminated to primary target organs, associated with mild-to-severe morbidity. In contrast, both vaccines were well tolerated with no evidence of dissemination and shedding while inducing potent neutralizing antibody responses. The results of the studies support the unprecedented safety profile of both vaccines for animals and humans.

In addition to infecting ruminants, RVFV is also infectious to humans. Humans can be exposed to the virus 43 via contact with contaminated tissues or blood released during the slaughtering of RVFV-infected animals, or via 44 mosquito bites. Although humans generally develop a self-limiting febrile illness, a significant fraction develops 45 neurological disorders or haemorrhagic fever 4 , which is often fatal. Furthermore, increasing evidence suggests that 46 RVFV infection during human pregnancies may result in complications [5][6][7] . Presently there are no vaccines or 47 therapeutics available to prevent or treat human infections. 48 The RVFV genome is divided over three RNA segments of negative polarity that are named after their 49 size; small (S), medium (M) and large (L). The L segment encodes the viral RNA-dependent RNA polymerase. The 50 S segment encodes the nucleocapsid that protects the viral RNA from degradation and is involved in transcription 51 and replication, and a non-structural protein named NSs. NSs interferes with host innate immune signalling 52 pathways and is considered the major virulence determinant of the virus 8 . The M segment encodes a polyprotein 53 precursor that is co-translationally cleaved by host proteases into the structural glycoproteins Gn and Gc. Gn is 54 involved in attachment to target cells and Gc is required for fusion of the viral and endosomal membranes. The M 55 segment additionally encodes a small 14-kDa protein, named NSm, which counteract apoptosis 9 , and a large 78-kDa glycoprotein, named LGp, that comprises the NSm and Gn coding regions, shown to be important for 57 dissemination of the virus in mosquitoes 10-12 . 58 We previously reported the development of a novel live-attenuated RVF vaccine that was constructed by 59 splitting the M genome segment into two M-type segments, one encoding Gn and the accessory proteins NSm and 60 LGp, and one encoding Gc, resulting in a four-segmented RVFV (RVFV-4s) 13 . To optimize the safety profile, the 61 NSs gene was deleted from the S segment. The technology to create four-segmented RVF viruses was 62 subsequently used to create tailor-made vaccines for both animal and human application. The veterinary vaccine, 63 here referred to as "v"RVFV-4s, is based on strain 35/74 and optimally replicates in ruminant cells, whereas the 64 vaccine to be applied in humans, here referred to as "h"RVFV-4s is based on the naturally attenuated Clone 13 65 isolate 14,15 .

66
Both vRVFV-4s and hRVFV-4s were shown to be safe and efficacious in various rodent and ruminant 67 animal models 13,16-20 , however, safety-data obtained with a more appropriate model for humans was lacking.

68
Demonstration of safety in a nonhuman primate (NHP) model is not only relevant for the candidate human vaccine, 69 but also for the candidate veterinary vaccine, as veterinarians applying this vaccine could be exposed accidentally 70 to the vaccine virus. NHPs, considering their close phylogenetic relationship to humans, are one of the most 71 predictive models with regard to human vaccine safety and efficacy 21 . Historically, rhesus macaques were used for 72 the evaluation of candidate RVF vaccines and therapeutics, despite their relatively low susceptibility to the wild-type 73 virus 22,23 . More recently, common marmosets (Callithrix jacchus) were shown to provide a more convenient model, 74 as these animals develop higher morbidity, more consistent viremia and marked aberrations in hematological and 75 biochemistry values following wild-type RVFV infection compared to macaques 24,25 . 76 In this study, we evaluated safety and immunogenicity of a high dose of the vRVFV-4s vaccine candidate 77 and of 3 doses of the hRVFV-4s vaccine candidate in the common marmoset. We specifically assessed body 78 temperatures, virus dissemination and shedding, and induction of neutralizing antibody responses compared to 79 their parent wild-type RVFV strains.

81
Experimental design 82 To assess the safety and immunogenicity of both vRVFV-4s and hRVFV-4s in the marmoset model, we conducted 83 two independent experiments (Fig. 1). In experiment 1, eight marmosets were inoculated with a dose of 10^7 TCID50 84 of vRVFV-4s via a combined intramuscular (IM) and subcutaneous (SC) route. On day 14 and day 28, four 85 inoculated animals were euthanized to collect organ samples. Another 4 marmosets were inoculated with a 10^7 86 TCID50 dose of parent strain 35/74, and were necropsied on day 14. In experiment 2, three groups of six animals 87 were inoculated with either 10^5, 10^6 or 10^7 TCID50 of hRVFV-4s via IM route only as the IM route is the 88 anticipated route for human vaccination. In this experiment, a group of identical size was inoculated with 10^7    102 None of the animals inoculated with vRVFV-4s (experiment 1) or hRVFV-4s (experiment 2) presented abnormal 103 behaviour or manifested with clinical signs or weight loss during the course of the experiment (Fig. 2a, b). One out  Disease severity was more pronounced in marmosets inoculated with the 74HB59 strain compared to strain 35/74.

116
No behavioural changes, clinical signs or weight loss were observed in any of the vRVFV-4s or hRVFV-4s animals. with the wild-type virus did not show this early-phase temperature increase but instead had a more pronounced 130 temperature increase at 2-4 DPI that persisted on average for 1.5-2 days (Fig. 3).  virus gradually decreased to undetectable levels at 11 and 7 DPI, respectively (Fig 5a-d). Peak viremia in animals 175 inoculated with wild-type virus coincided with increases in abdominal body temperatures (Fig. 3). In the vRVFV-4s 176 and hRVFV-4s inoculated animals viral RNA was detected at DPI 1 and gradually declined to a level below the 177 detection limit around DPI 5. The total genome copy numbers detected at this time point (<10^7/ml) were 178 approximately three logs lower than the total number of genome copies present in the inoculum (>10^10/ml). No 179 increases in viral RNA levels were detected in these animals, suggesting that RVFV-4s inoculation does not result 180 in vaccine virus viremia. This is supported by the lack of detectable infectious vaccine virus in blood collected from 181 RVFV-4s inoculated animals (Fig 5c and d).  Supplementary Fig.1, panel A and B). Animals M16017 an M16040 showed a multifocal meningo-encephalitis 203 with immunostaining of neurons and neuronal processes throughout the brain (Supplementary Fig 1, panel C and 204 D).

205
In the vRVFV-4s inoculated animals no viral RNA was detected in liver, spleen, kidney, heart and adrenal gland 206 samples, but viral RNA was detected in various lymph node samples and occasionally in spleen samples.

207
Remarkably, the overall number of RVFV-4s RNA positive lymph node samples was higher in experiment 1 208 compared to experiment 2. inoculated animals ( Fig. 6c and d). These results suggest that hRVFV-4s does not disseminate to the environment.  As neutralizing antibodies are the only known correlate of protection for RVFV, we assessed those responses in 225 both experiments. Starting from DPI 7, neutralizing antibodies were detected in both vRVFV-4s and hRVFV-4s 226 inoculated animals, and neutralizing antibody levels increased up to 3 weeks post inoculation (Fig. 7). In experiment 227 2, in which several RVFV-4s doses were assessed, lower levels of neutralizing antibodies were observed in animals 228 inoculated with 10^5 TCID50, relative to the higher dose groups. No differences were observed between animals 229 vaccinated with 10^6 and 10^7 TCID50. As expected, all parent virus-inoculated animals developed high levels of 230 neutralizing antibodies. Remarkably, the overall levels of neutralizing antibodies of the parent virus-inoculated 231 animals were only slightly higher compared to those detected in the vRVFV-4s and hRVFV-4s high-dose groups.    To collect blood samples, animals were sedated with Alfaxan (12 mg/kg, IM route). Blood was drawn from the 351 femoral vein in the groin using aseptic techniques using a Vacutainer blood collection system (Becton Dickinson, 352 Vacutainer systems). EDTA tubes with collected blood were centrifuged for 10 min at 1,000 x g followed by collection 353 of plasma. A volume of 125 µl plasma was used for clinical chemistry measurements. The remaining material was stored at -80˚C for virus detection via PCR and for the detection of RVFV antibodies. Upon euthanasia, 2ml EDTA 355 blood and plain blood samples were collected. Serum samples were prepared by centrifugation of clotted blood for 356 10 min (1,000 -1,300 x g), serum was stored at -20˚C.   of each cheek was swabbed for a few seconds. The tip of each swab was placed into a 14 ml tube containing 2 ml 375 of transport medium (MEM, 0.5% BSA, 2.5μg/mL Fungizone, 100 U/mL Penicillin, 100μg/mL Streptomycin) and the 376 applicator stick was broken off. Vials with medium and swabs were vortexed for 20 s. Supernatants were centrifuged 377 for 5 min (750 x g) and stored in at least 2 aliquots at -80°C. Anal swabs were placed in transport medium and 378 handled as described above. Possible debris was pelleted by centrifugation (750 x g, 5 min) and supernatants 379 collected and stored in at least 2 aliquots at -80°C. Proteinase K (5 μg/ml, Sigma). Next, 200 μl AL buffer (Qiagen), supplemented with 2 μl polyadenylic acid A (5 388 mg/ml, Sigma) was added, after which the samples were thoroughly mixed and incubated at 56°C for 15 min. 389 Subsequently, 250 μl 99% ethanol was added and RNA was isolated using the Qiagen RNeasy kit according to the   Subsequently, 20,000 BHK-21 cells (in 50 μl) were added to each well. Plates were incubated for 2 days at 37 °C 411 and 5% CO2 and scored using an EVOS-FL microscope (Life Technologies). VNT50 titres were calculated using the 412 Spearman-Kärber algorithm. Presence of RVFV nucleoprotein-specific antibodies in sera was determined using