Lentivirus-induced diseases pose serious threats to both humans and other animals. Human immunodeficiency virus-1 (HIV-1), a lentivirus that causes acquired immunodeficiency syndrome (AIDS), is responsible for more than 5 million new infections annually and has resulted in more than 20 million deaths worldwide [3]. Equine infectious anemia virus (EIAV) is another member of the family Retroviridae, genus Lentivirus, that infects equines. This virus is present in many countries and causes significant economic losses, especially in developing countries such as China (Feng JL and Shen RX, personal communication and [13]).

Development of effective vaccines against lentiviruses remains a major challenge. In the process of HIV-1 vaccine development, researchers have employed a variety of strategies, including subunit vaccines, live vector-based envelope protein-expressing recombinant vaccines, and DNA prime/protein boost protocols as well as genetically engineered live attenuated simian immunodeficiency virus (SIV) and simian–human immunodeficiency virus (SHIV) vaccines (reviewed in refs. 14 and 22). Among the existing experimental vaccines, only live vaccines have been found to induce comparatively strong resistance to infection by SIV/SHIV in non-human primates [1, 9, 35, 36]. Other experimental vaccines induce certain levels of immune responses but do not provide sterile protection against infections by pathogenic strains [19].

Safety has been the major obstacle for developing live attenuated lentiviral vaccines, especially for HIV-1. Major concerns are the possibility of live virus vaccines having increased pathogenicity in individuals with dysfunctional immune systems and reversion to virulence during their long-term replication in hosts. Such events actually occurred in studies of attenuated SIV and SHIV vaccines, developed by deletion or mutation of the nef gene [11, 20, 29, 33]. While these safety issues have greatly limited research on live HIV-1 vaccines for humans, there has been a successful application of an attenuated lentiviral vaccine for horses. In the 1970s, a Chinese attenuated EIAV vaccine developed by Shen et al. [24, 27] successfully controlled the prevalence of equine infectious anemia (EIA) in China. During a 15-year period, over 70 million equines were safely immunized, and the safety of the vaccine was fundamentally established (Feng JL and Shen RX, personal communication). In addition, in vivo passages and vertical transmission tests of the vaccine also showed no reversion of virulence (Feng JL and Shen RX, personal communication and [23]). Our recent work on the Chinese attenuated EIAV vaccine, including characterization of the in vivo evolution of the gp90 gene and the interaction between the vaccine strain and its hosts under immunosuppression, further suggest that this attenuated EIAV vaccine is safe for practical immunizations.

Some published studies showed that particular sequences of the gp90 protein, i.e. the surface subunit of the EIAV envelope protein (SU), were closely related to the immunogenicity and virulence of viruses [2, 12]. In order to understand the replication and evolutionary features of the attenuated Chinese EIAVFDDV vaccine in vivo, we analyzed the gp90 gene of the virus from immunized hosts at different time points by reverse transcriptase-polymerase chain reaction (RT-PCR) and nucleotide sequencing. The analysis of virus diversity based on the gp90 gene was performed by random selection of multiple clones of PCR products for sequencing. Reverse transcription was performed with the primer p8, 5′-TGACCCCATGATTCATTCCA-3′ (nucleotide positions 6980–6999 of the genome of a Chinese attenuated EIAV vaccine strain, GenBank accession no. AF327878) and a reverse transcription kit (Invitrogen). Subsequently, PCR was performed with primers p7, 5′-CCACCAGAGTGTTGTGGAAAGGTG-3′ (5022–5045) and p8 to amplify a fragment of approximately 1.9 kb. Because the copy number of plasma genomic RNA of EIAVFDDV was low, highly sensitive nested PCR was utilized to amplify the gp90 gene from samples that were pre-ultracentrifuged to concentrate viral particles. The process of nested PCR for amplifying the EIAVFDDV gp90 gene was similar to that for reverse transcription. Nested PCR primers included the outer primers p7 and p8 and the inner primers p7-1, 5′-TTGTAAGGTTTGGTGTATGGG-3′ (5282–5302) and p8-1, 5′-ATGGCAGCTATTATAGCAGA-3′ (6680–6699). This fragment obtained by the nested PCR was about 1.41 kb, which was large enough to encompass the entire gp90 coding region.

DNA polymerase ExTaq (Takara) was used for PCR amplification. This DNA polymerase synthesizes DNA with higher fidelity than the common Taq DNA polymerase and results in fewer errors in amplified fragments. The amplified gene fragments were then inserted into the pMD18-T vector and introduced into E. coli DH5α competent cells by transformation. Twelve to fifteen positive clones were picked for each PCR product and sequenced. DNA Star, Mega 4.0 and GENEDOC were used to analyze sequencing data. Proper negative controls and a positive control of a molecular clone of EIAVFDDV3-8 gp90 were added to RT-PCR and nested PCR. Fifteen clones from EIAVFDDV3-8 gp90 were also sequenced to rule out potential contamination (data not shown).

Phylogenetic analysis of PCR clones of the EIAVFDDV gp90 gene showed that sequence diversity among these clones was high (

Fig. 1). The nucleotide variations among the different clones reached as high as 4.0%, and differences in deduced amino acids were up to 6.5% (data not shown). The high diversity among these gp90 gene clones is considered to be due to the in vitro attenuating process of the parental virulent strains in cells cultured for more than 100 generations. In the process, mutations in viral genomes were accumulated during replication of each generation due to the lack of immunological selective pressure in the cultivation environment.

Fig. 1
figure 1

Phylogenetic analysis of EIAVFDDV gp90 gene evolution in vaccinated ponies. EIAVFDDV gp90 gene evolution in one of the four vaccinated ponies is presented. Similar evolution patterns were detected for the other experimental ponies (data not shown). a Changes of gp90 diversity in vivo during the experimental period. Twelve to fifteen clones were sequenced at each time points. b Sequences of all of the gp90 clones shown in 1a were used for phylogentic analysis of in vivo evolution over the experimental period. The bar in figures represents 0.002 amino acid changes per site. DPI day post-injection

Full size image

Phylogenetic analysis on the gp90 gene of the attenuated vaccine in vaccinated ponies was also performed. The gp90 sequences at the time point of 15 days post-infection (DPI-15) were closely related to the gp90 population of the original in vitro-prepared EIAVFDDV in their genotypes and diversity (data not shown). However, the diversity among the gp90 clones at each time point of in vivo evolution of the vaccine strain was largely different (Fig. 1a). The diversity of the gp90 gene sequences from the 4 experimental ponies followed a declining trend for up to 18 months after vaccination, gradually decreasing from 1.45 ± 0.51% in the early stage after immunization (DPI-15) to 0.51 ± 0.09% and 0.29 ± 0.04% in the mature immune stages (DPI-180 and DPI-540, respectively). The values of gp90 gene diversity were calculated using Dnasp4.0 software.

To illustrate the genetic relationships between the vaccine strains evolving in vivo over time after vaccination, a phylogenetic tree for determining the genotypes of all of the gp90 sequences, shown in Fig. 1a, was made using Mega 4.0 (Fig. 1b). The distribution of each analyzed sequence in Fig. 1a, b suggested that (1) gp90 evolved over the time of infection, especially during the earlier stages of infection (before DPI-180), and (2) the diversity of gp90 had shifted from highly heterologous at time points DPI-15 and DPI-60 to homologous at DPI-180 and DPI-540. These results indicate a tendency of self-limitation of diversity of the EIAV vaccine strain EIAVFDDV in vivo accompanying the maturation of the immunity.

Comparison of the deduced SU amino acid sequences from a total of 52 clones of the vaccine strain EIAVFDDV and the relevant sequence of the parental virulent strain EIAVLiao indicated that there were 11 consensus mutation sites in all of the sequences of the vaccine clones (Table 1). Two of these 11 consensus mutations, S193 → N and N237 → K, removed two potential N-linked glycosylation sites in SU. Some studies have shown that removal of the N-linked glycosylation sites in the envelope protein of HIV-1 or SIV enhances the neutralizing antibody response [17, 21]. Therefore, the deletion of these two N-glycosylation sites of pathogenic EIAV SU is presumed to elicit higher levels of neutralizing antibody. It was reported that the reverse mutation of four amino acid residues in SU of an attenuated EIAV strain to the residues of its parental virulent strain resulted in an increased replication rate in vivo and a reversion of virulence [27]. These published studies suggest that the 11 consensus mutations in the EIAVFDDV SU are at least partially involved in the attenuated virulent and enhanced immunogenicity of the Chinese EIAV vaccine strain.

Table 1 Comparison of consensus mutation sites in deduced SU amino acid sequence of the vaccine strain EIAVFDDV and its derived viruses sequenced from plasma in vaccinated ponies at different time points
Full size table

Interestingly, 5 of the 11 consensus mutation sites had reverse-mutated into the corresponding residues of the parental virulent strain in some clones at DPI-60, i.e. residues E189 → K, K190 → E, K237 → N, Q410 → H, and an insertion of 236D (Table 1). However, the other parts of SU sequences retained the characteristics of the vaccine strain. At DPI-60, the plasma viral loads of the four ponies were 103–105 copies/ml, which was within normal levels for EIAV vaccine strains. In addition, no clinical signs of EIA were observed. To some extent, these results suggested that the partial reverse mutations at these amino acid residues did not alter the attenuated virulence of the vaccine strain EIAVFDDV. However, inoculation of these reverse-mutated clones into naïve animals is required to confirm that they are indeed completely avirulent.

In addition, reverse mutations of these 5 residues were not identified in most gp90 clones at DPI-180 and DPI-540, except one for reverse mutation that appeared at another consensus vaccine-specific site (E46 → A) in clones from one of the four ponies at DPI-180 (Table 1). At this time point, the predominant quasispecies of EIAVFDDV were formed and became stable in the vaccinated ponies. The percent identity of the gp90 sequences of DPI-180 clones to the corresponding sequences of some predominant EIAVFDDV clones was approximately 99% (data not shown), and the 10 specific consensus amino acid mutation sites of the vaccine strains were fully preserved. Furthermore, the reverse mutation E46 → A no longer existed in the gp90 clones of EIAVFDDV isolated at the later time point (DPI-540). Such highly homologous and EIAVFDDV reference-strain-like quasispecies were predominant even 18 months after the immunization (DPI-540). Our finding of certain highly homologous EIAVFDDV quasispecies coexisting with other quasispecies in the host during the early stage post-immunization was different from the finding of Craigo et al. [7]. Their study showed that the ancestral species in a long-term inapparent EIAV carrier were eliminated and were substituted by new populations in response to ongoing immune pressure.

Although the attenuated vaccine strain developed reverse mutations at certain amino acid residues in the SU at the DPI-60 time point, the above results indicated that these mutated viral species could be substituted by species with SU characteristics of the attenuated vaccine strain after the maturation of specific immunity (6 months after immunization, DPI-180). These vaccine species continuously existed in the vaccinated hosts for up to 18 months in this study. The exact mechanism for the transient reverse mutation is not clear. However, the combination of the genomic stability and the reduction in diversity of the vaccine strain in the hosts with matured immunity indicates that the Chinese attenuated EIAV vaccine strain had stabilized in vivo, which is an extremely important requirement for a safe attenuated vaccine.

These data indicate that even though the attenuated vaccine strain with variant gp90 sequences could retain their diversity at the early stage of infection, the genomic variation was obviously reduced with the maturation of protective immunity. Genetic variation could increase the rate of reverse mutations of the attenuated vaccine and therefore increase the danger of virulence reversion. However, the evolutionary trend, from diverse to homologous, of the Chinese EIAV vaccine strain in vaccinated ponies implied that vaccine-induced immunity may play a major role in restricting the diversity of the vaccine strain and driving the genotypes to singularity. If this is true, such immune-driven constraints on the virus would help in reducing the chance of virulence reversion of the attenuated vaccine during replication. Additionally, factors other than pressure from the immune system may have contributed to the observed reduction in SU complexity in vivo, such as selection pressure of re-adaptation from donkey cells to horse cells and from in vitro to in vivo [4].

The major reason for the rare acceptance of attenuated lentiviral vaccines is the risk of increased virulence due to viral mutations when viruses replicate in the host for a long period of time. This concern is especially prominent in the context of decreased immunity, which may result in a dramatic increase of viral load and result in enhancement of pathogenicity of attenuated vaccines [34]. In this study, the immune systems of the vaccinated ponies were transiently suppressed by two weeks of daily administration of dexamethasone. The replication levels and virulence of the vaccine strain EIAVFDDV were investigated by monitoring the change in plasma viral load and appearance of EIA clinical symptoms. Three ponies, #7, #8 and #9, which had been vaccinated with EIAVFDDV for 18 months, were used to study the change in virulence of the vaccine strain in immune-suppressed hosts. An inapparent EIAV carrier (#6), which was confirmed EIAV positive by sequencing of the gp90 gene and agar gel immunodiffusion test for the antibody, was used as a positive control. Rectal temperatures were monitored daily from 2 days before administration of dexamethasone until 6 months after the drug treatment. Blood cell count and classification, platelet count and plasma viral load were also regularly measured.

Immunosuppression was induced in ponies in this experiment by a procedure described previously [5, 31]. Dexamethasone was injected intramuscularly at 0.11 mg/kg of the body weight for each pony for 14 days. The effects of immunosuppression were evaluated using a delayed-type hypersensitivity (DTH) test [5] by intradermal injection of 50 μg/ml PHA-P (Sigma) diluted with saline. In addition, lymphocyte proliferation [30] in peripheral blood mononuclear cells (PBMCs) in response to PHA-P stimulation was monitored before and after treatment with dexamethasone. Results of the DTH test and lymphocyte proliferation indicated that the immunity of all the four ponies was significantly suppressed (Fig. 2 a, b).

Fig. 2
figure 2

Evaluation of the immunosuppressive effects and plasma viral loads before and after dexamethasone treatment. EIAVFDDV-vaccinated ponies (#7, #8 and #9) and an inapparent EIAVLiao carrier (#6) were injected intramuscularly with dexamethasone every day for 2 weeks. The proliferation of lymphocytes stimulated with PHA-P (a) was measured, and a delayed-type hypersensitivity test (DTH, b) was performed to evaluate the efficiency of immunosuppression. EIAV RNA copy numbers were measured 2 days before dexamethasone treatment, 10 days after the start of treatment, and 14 days after the end of treatment (c). *P < 0.5, **P < 0.01 as analyzed using SAS 8.1

Full size image

Plasma EIAV loads in the ponies before and after immunosuppression were measured by real-time RT-PCR. The results are shown in Fig. 2c. The viral copy number of the EIAV virulent strain carrier (pony #6) increased 25,000-fold in 10 days after administrating the drug, reaching 2.5 × 108 copies/ml, which was higher than the “pathogenic threshold” of 107 RNA copies/ml for clinical EIA. The increased EIAV plasma load corresponded to the appearance of other EIA clinical signs, such as fever and thrombocytopenia (data not shown). As a result, it can be determined that the unapparent EIAV carrier, pony #6, developed an acute episode of EIA when the immunity was suppressed. These EIA symptoms disappeared when the dexamethasone treatment ended. These results indicated that clinical EIA may be induced in long-term asymptomatic carriers of EIAV pathogenic strains when the immunity of the host was suppressed, due to the dramatic increase in plasma viral load.

In contrast, the viral loads of the immunized ponies, #7 and #9, did not show any EIA symptoms during the drug treatment and thereafter. The viral loads of EIAVFDDV were between 103–105 copies/ml. The viral load in pony #8 increased slightly 10 days after the start of dexamethasone treatment and 14 days after the end of drug treatment, but at both times it remained below 106 copies/ml, which was within the dynamic levels for the vaccine strain EIAVFDDV. Neither the vaccinated ponies nor the inapparent carrier showed any EIA clinical signs during the following 6 months of observation.

Viral load monitoring in these experiments revealed two important facts: First, the immunity, especially the T cell-mediated immunity, induced by attenuated EIAV vaccine was not the major reason for the maintenance of a consistently low level of replication of the vaccine strain in vivo. This low-level replication may be different from that of the virulent strain in inapparent carriers, which was apparently restricted by the immune pressure. This consistently low-level replication of EIAVFDDV in immune-suppressed hosts strongly demonstrated the safety of the attenuated vaccines. Secondly, no clinical EIA symptoms were observed in the vaccinated ponies within at least 6 months after the end of immune suppression. This fact indicated that replication of the attenuated EIAV vaccine in immunosuppressed hosts did not cause any visible health problems, which provided further evidence for the safety of EIAVFDDV.

The development of the Chinese attenuated EIAV vaccine was obviously different from that of EIAVD9, a live EIAV vaccine that was developed by Li et al. [15] and Craigo et al. [6] by introducing mutations into a defunct S2 gene. This vaccine strain completely protected immunized horses from infection with a parental virulent strain virus with homologous SU sequence. However, the effectiveness of EIAVD9 in protecting animals from infection and disease was significantly reduced when the immunized animals were challenged with viruses containing a relatively modest variation in SU [8]. Since the life cycle of attenuated vaccine virus in the body is far closer to the infected pathogenic lentiviruses than any other types of vaccine, this study provides valuable information for future studies on animal lentivirus and human AIDS vaccines.

The Chinese attenuated EIAV vaccine strain EIAVFDDV was developed by consistently passing the parental virulent strain EIAVLiao in cultured donkey monocyte-derived macrophages and fetal donkey dermal cells for 125 and 15 generations, respectively [25]. This long-term in vitro attenuation process resulted in high rates of mutation in multiple genes of the vaccine strain, mainly LTR, env and accessory genes S2 and rev [32]. Several published studies have demonstrated that mutations in these genes altered the pathogenicity of wild-type strains [10, 16, 18, 28, 37].

Accompanied by the variation of genotypes, the phenotype of EIAVFDDV is different from that of EIAVD9, including the effectiveness of its immune protection against heterogeneous virulent strains. Previous results showed that EIAVFDDV had successfully protected more than 70% of vaccinated animals from disease after infection by the heterogeneous virulent strain EIAVWyoming [26].

The attenuation resulting from multiple-site mutation is the major difference between the attenuated Chinese EIAV vaccine and most of the other attenuated lentiviral vaccines, which were developed by recombinant techniques. The development of attenuated SIV or SHIV vaccines has been based primarily on the deficiency or mutation of a single gene (such as nef). Because the number of modified sites was limited, and because they were located only in certain segments of the viral genome, the possibility of virulence reversion was high during long-term replication in vivo, as observed in some studies in non-human primates [11, 20, 29, 33, 34]. Since the Chinese EIAV vaccine strain EIAVFDDV was attenuated through multiple-gene and multiple-site mutations, it is logical to assume that the possibility of virulence reversion is substantially reduced. In addition, the unique in vivo evolution pattern and stable, moderate-level replication feature of the vaccine strain is considered favorable for the induction of protective immunity, which could further reduce the risk of virulence reversion of attenuated EIAV vaccines.

In conclusion, our data reveal that the in vivo evolution of the gp90 gene from EIAVFDDV shows an evolutionary trend going from polymorphism to singularity. There were also some occurrences of reverse mutations in the vaccine strain in hosts before the maturation of immunity. However, such reverse mutations were limited in number and were eliminated with the maturation of immunity. Combined with the consistently moderate level of replication, even in immunosupressed hosts, the Chinese attenuated EIAV vaccine strain appears to be a unique attenuated lentiviral vaccine. Further studies are required to compile the safety profile of this vaccine and to elucidate the mechanisms of attenuated virulence and enhanced immunogenicity.