Genetic stability (in vivo) of the attenuated oral rabies virus vaccine SAD B19


Thomas Müller, Friedrich-Loeffler-Institute, Seestrasse 55, 16868 Wusterhausen, Germany.
Tel: +49 33979 80186; fax: +49 33979 80200; email:


The distribution of oral rabies vaccine baits containing replication-competent live viruses poses certain environmental safety risks; among others, the possibility of reversion to or an increase in virulence. Hence, the genetic stability of the complete genome of the most widely used oral rabies vaccine virus, SAD B19, was examined after four and 10 serial i.c. passages in foxes and mice, respectively. It was shown that the consensus strain of SAD B19 was extremely stable in vivo. After 10 consecutive passages in mice not a single mutation was observed. In foxes, seven single nucleotide exchanges were found between the first and fourth passage, of which only one resulted in an amino acid exchange at position 9240 of the L-gene. This mutation was not observed during the first three passages and, furthermore, it was shown that this mutation was not linked to enhanced virulence.

List of Abbreviations: 

focus-forming units




minimum essential medium


newborn calf serum

Rabies is one of the oldest and most feared infectious diseases known to humans and, therefore, great efforts have been made reducing the public health burden caused by this viral disease. For example, one of the most prominent milestones in vaccinology has been the development of a rabies vaccine by Pasteur. Another more recent breakthrough was the development of efficacious oral rabies vaccines. The distribution of these vaccine-loaded baits has been highly effective in controlling and eradicating fox-mediated rabies from large areas in Europe and North America. The initial successes did not only lead to an increasing number of countries implementing such oral rabies vaccination campaigns but also to an increasing number of animal species targeted. The first campaigns were solely aimed at the red fox (Vulpes vulpes), but nowadays baits are distributed to vaccinate a large number of different reservoir species; among others, raccoons (Procyon lotor), coyotes (Canis latrans), raccoon dogs (Nyctereutes procyonoides), grey foxes (Urocyon cinereoargenteus), golden jackals (Canis aureus), skunks (Mephitis mephitis) and domestic dogs (Canis familiaris). The method has also been considered for conservation purposes by protecting highly endangered canid species such as African dogs (Lycaon pictus) and Ethiopian wolves (Canis simensis) against spill-over rabies infections. The first oral rabies vaccines were live attenuated rabies viruses. Actually, these vaccines have shown to be remarkably efficient and are still the most widely used almost 30 years after the first oral rabies vaccination campaign in Switzerland in 1978 (1).

However, several genetically modified oral rabies vaccines are now also available or under development (2). All presently available commercial oral rabies vaccines are replication-competent live viruses that theoretically could induce disease. Several reports are available on oral vaccine-associated rabies cases or related incidents (3–6). Furthermore, it was speculated that as a result of the high mutation rates characteristic of RNA viruses, some of these oral vaccines based on rabies viruses could revert to virulence and, hence, the vaccine strain could become established in the field. To examine the possibility of reversion to or an increase in virulence, serial passages in target and non-target species have been carried out for most available oral rabies vaccines as required by the regulatory authorities (7). Unfortunately, the genetic stability of the vaccine virus after these serial passages has not been examined. At most, the genetic stability of certain small segments of the genome that can be used for genetic identification of the virus strain has been investigated in detail. Therefore, for the first time, the sequence of the complete genome of one of the most widely used oral rabies vaccines, SAD B19, was examined after four and 10 i.c. passages in foxes and mice, respectively. It was shown that the consensus strain of SAD B19 was extremely stable after serial passages in vivo, in contrast to some other attenuated rabies virus strains that can revert to more pathogenic variants by a few passages in mouse brain (8).



The sequences of the SAD B19 virus originated from material collected during a previous study in which the oral rabies vaccine virus SAD B19 was passaged i.c. four and 10 times in foxes and mice, respectively (9). This route was selected for because the regulatory authorities require that the initial administration and subsequent passages should be carried out using a natural or recommended route of administration that most likely leads to reversion to or an increase in virulence in the target species and result in recovery of the virus (7). Thus, for the attenuated SAD B19 vaccine virus, the only available route that would guarantee disease was by direct inoculation into the brain, although this does not represent the natural route of infection. Initially, four mice were inoculated with 0.025 mL pre-diluted 108 FFU/mL Working Seed Virus SAD B19 by the i.c. route. The animals were killed upon showing clinical signs of rabies. Brain material was collected from the individual mice and from each sample a 10% brain suspension was prepared using MEM +2% NCS. The brain suspension (1:2 pre-diluted) of one of the mice was inoculated i.c. (0.025 mL) in the next group of four mice. The procedure was repeated until the 10th passage was re-isolated. Additionally, two farm-bred foxes were inoculated i.c. with 1.0 mL SAD B19 (108 FFU/mL). After 2 days the animals were killed and a 10% brain suspension using MEM/SNT +2% NCS was prepared from each fox. The suspension was passaged once in baby hamster kidney cells (BHK21-BSR) and inoculated into the following group of two foxes by the i.c. route. A previous attempt to passage the virus directly by inoculation of the brain suspension in other foxes failed, therefore an intermediate passage in cell culture was introduced. The fourth passage re-isolated from a fox was subsequently injected i.c. in four foxes and four dogs; these animals were observed for 46 and 40 days, respectively. Brain suspension material from each passage was stored and kept frozen until further processing.

The handling and invasive procedures were conducted in compliance with the German Animal Welfare Act and recommendations of the GV-SOLAS (Society for Laboratory Animal Science). Furthermore, approval was obtained from the appropriate authorities.

RT-PCR and sequencing

RT-PCR and forward and reverse sequencing of the PCR products using direct cycle sequencing was carried out essentially as described elsewhere (10).

The first, fifth and 10th mouse passages, as well as the first and fourth fox passages, were sequenced as described, from nucleotide position 10 to 11904 of the SAD B19 genome (11). The second and third fox passages were analyzed in those gene segments that contained nucleotide changes.

Cloning of PCR fragments

Genome regions of individual samples with varying results for single nucleotides (positions 4664 to 4902 [first fox passage] and 3575 to 4017 [fourth fox passage]) were amplified by RT-PCR, directly cloned into the plasmid vector pCR2.1 and transformed in Escherichia coli TOP10 cells by using the TA cloning kit (Invitrogen, Carlsbad, CA, USA) following the manufacturer's instructions. The inserts of 11 clones (first fox passage) and 15 clones (fourth fox passage) were analyzed for each strain by direct cycle sequencing using IRD800 labeled primers (M13universal (−21) and M13reverse (−29)).

Nucleotide sequences generated in the present study have been submitted to GenBank and assigned the following accession numbers: first fox serial passage (EU877068), fourth fox serial passage (EU877067), first mouse serial passage (EU877069), fifth mouse serial passage (EU877070) and 10th mouse serial passage (EU877071).

Details on the primers used for long one-step RT-PCR and sequencing as well as the corresponding RT-PCR conditions applied can be provided upon request.


Based on the alignment of 65 obtained individual sequence files, the consensus sequence of the first (EU877069) mouse passage encompassing 11 894 bp was generated.

The consensus sequence of the fifth (EU877070) and 10th (EU877071) mouse serial passages was based on 57 and 59 obtained individual sequences files, respectively. No differences in nucleotides were found between these three mouse passages and also a 100% identity was observed with the sequence of SAD B19 (Genbank EF206709).

The consensus sequence of the first (EU877068) and fourth (EU877067) fox serial passages was generated using 70 and 65 sequence files, respectively. Individual sequence files of the first fox passage indicated non-conformity at position 4792 (G-gene) and also possible insertions between positions 4804 and 4805 (Fig. 1a). Consequently, a 238 bp covering this fragment was synthesized, cloned and sequenced. All 11 sequences obtained indicated, without exception, thymidine at position 4792 and insertions were no longer detectable (Fig. 1b). Also, the fourth passage showed variability after direct sequencing at positions 3846 and 3862 of the G-gene. This region of 442 bp in length was also synthesized and cloned. Subsequent sequencing of these clones identified unambiguously adenine at both positions 3846 and 3862.

Figure 1.

(a) Individual sequence files of the first fox passage indicated differences within the shown segment of the glycoprotein (Software: SeqMan): Sequence differences were located at position 4792 (the ambiguity is indicated by ‘Y’) and, additionally, possible insertions between positions 4804 and 4805. (b) Alignment of the individual sequences obtained from a 238 bp clone covering the fragment with the discrepancies shown in (a): All sequences indicated thymidine at position 4792 and insertions were no longer detectable.

The first fox passage and the complete sequence of SAD B19 vaccine virus (Genbank EF206709) showed 100% identity. However, between the first and fourth fox passages, several single nucleotide exchanges were found; and a mutation of the second base in codon CAA from A → G at nucleotide position 9240 of the L-gene resulted in an amino acid exchange at position 1276 (Gln → Arg) (Table 1, Fig. 2).

Table 1.  Differences in genome sequence between the first (EU877068) and fourth (EU877067) fox serial passages (i.c.) of the oral rabies vaccine virus SAD B19
Nucleotide positionGeneNucleotide exchangeAmino acid exchange
106NT → CNone
1723PC → TNone
3862GG → ANone
7642LA → GNone
9240LA → GGln → Arg
9766LA → CNone
11074LT → CNone
Figure 2.

Comparison of the sequence obtained with primer Tw-9623r-seq of the four fox passages: the fragment between positions 9228 and 9252 is shown and the nucleotide exchange of the second base in codon CAA (red frame) at position 9240 (A → G) is framed (purple).

In order to assess when those nucleotide exchanges occurred for the first time, the consensus sequence of the second and third fox passages was determined for those regions where the difference between the first and fourth passages had been observed. These regions were 100% identical with the first passage, indicating that the observed changes occurred during the last serial passage (Fig. 2).


The rabies virus genome comprises five genes, encoding nucleoprotein (N), phosphoprotein (P), matrix protein (M), glycoprotein (G) and polymerase (L) in the order 3′-N-P-M-G-L-5′. The M and G proteins are predominantly involved in the formation of the viral envelope. The M protein is also most likely involved in other host–cell interactions (12). The G-gene is responsible for cell attachment and fusion and is the main viral protein responsible for the induction of virus-neutralizing antibodies. It is also involved in the budding process (13). The N, P and L proteins are components of the viral ribonucleoprotein and each of them is essential for accomplishing RNA replication and transcription (14). The N protein is considered the most conserved followed by L, M and G proteins, with the P protein being the most variable (15).

In the present study, it was shown that, for the first time, a live attenuated oral rabies virus vaccine was extremely stable after serial passages in the target and non-target species. In the mouse model, no differences in the complete genome were observed after 10 serial i.c. passages. Also, in the fox, no mutations were seen in the consensus sequence during the first three i.c. passages. However, between the fourth and the previous passages, seven nucleotide exchanges were detected. Most of these changes at nucleotide level can be considered silent and are located at the third base position resulting in no amino acid substitution (15), except for the one at position 9240 (L-gene) (Fig. 2). Unfortunately, the function of this position is not known. However, it can be ruled out that this substitution increases pathogenicity because the subsequent inoculation of the fourth fox passage i.c. in dogs and foxes did not cause disease and all animals remained healthy during the entire observation period (9). It is not to be expected that the observed silent mutations would increase virulence. In contrast, recent studies with polio virus have shown that silent mutations can actually cause further attenuation, most likely, by slowing down protein production (16, 17).

Also, Kissi et al. demonstrated that the rabies virus was remarkably stable after in vivo and in vitro serial passages (18). This seems in contradiction with the high mutation rate of RNA viruses resulting from the lack of proofreading by RNA polymerases. These high mutation rates, together with rapid replication and often large population size, may lead to very heterogeneous virus populations; sometimes, also termed viral quasispecies (19, 20). Such virus populations normally consist of a widely dispersed mutant distribution where the obtained consensus sequence is the weighted average of all genotypes present rather than a homogeneous one formed by a single most-fit sequence. However, considering the small genome size and limited number of genes of the rabies virus in respect to the complexity of its life cycle, the vast majority of the single mutations may have a negative effect on fitness and are strongly selected against. Only under specific conditions, such as small population sizes (bottlenecks), could this strong selective pressure against these mutants be reduced to such an extent that the heterogeneity of the virus population is modified. This could explain the results obtained by others where it was shown that under certain conditions shifts in the consensus sequence of the passaged rabies virus strain were observed (21, 22).

Actually, the functions of only a few positions of the rabies genome have been well characterized; one of these is position 333 of the glycoprotein. Dietzschold et al. showed that an amino acid exchange at this position of the rabies glycoprotein could greatly alter the pathogenicity of the rabies virus (23). For example, when the codon at this position was altered from arginine to glutamate the rabies virus became less pathogenic (24, 25). The characterization of this antigenic determinant therefore offered great opportunities for vaccine development. Several commercially available and candidate oral rabies vaccines have a mutation in at least one of the three nucleotides of this amino acid at position 333 rendering the constructs less pathogenic or apathogenic for immunocompetent animals. However, during the characterization of one of these constructs, it was shown that after multiple passages in suckling mice a partial recovery of pathogenicity occurred (26). This pathogenic revertant still contained glutamate at position 333, but, at position 194 of the glycoprotein, an amino acid exchange was detected (Asn → Lys). Subsequent studies showed that this single amino acid exchange was responsible for the reversion to virulence (27). This example clearly shows that for the assessment of genetic stability it is not sufficient to determine only the sequence of selected regions that are known to play a role in infectivity and pathogenicity or have been altered purposely and act as some type of genetic marker. Also, Ito et al. identified another region in the glycoprotein as a possible determinant of pathogenicity (28). More recently, it was demonstrated that the N, P and M genes were also involved in viral pathogenicity (29).

The reduced genetic stability of this genetically altered rabies virus as described by Dietzschold et al. (26) could be an artefact of this particular construct, but, unless proven, might well be a general characteristic of such genetically modified constructs. Also, attenuated oral rabies vaccines obtained by monoclonal antibody selection can revert to the parenteral strain (24). As these specific monoclonal antibodies are not present in the animals after bait consumption; there is the risk that the vaccine virus reverts to its original strain in the vaccinated animal. By using techniques like reverse genetics and monoclonal antibodies, new vaccine strains can be developed relatively rapidly by discretional genetic manipulation and selective pressure, respectively (24, 30); this is in contrast with classically attenuated viral strains for live vaccines that have been developed by numerous passages of virus in vitro and/or in vivo. The relatively long attenuation process during the period of adaptation of the (whole) virus to the cells and/or tissues can lead to a highly adapted state that tends to be genetically stable. In contrast, the approach used by reverse genetics or monoclonal antibody selection is often targeted at a single position without an adaptation process and could be less resistant to change during subsequent passages. Hence, it is suggested that live viruses incorporated in oral rabies vaccine baits that are distributed in the environment will be investigated in more detail to assure genetic stability of the complete genome.