Phylogenetic analysis to define feline immunodeficiency virus subtypes in 31 domestic cats in South Africa

INTRODUCTION Feline immunodeficiency virus (FIV) was first isolated in 1986 from a group of cats in California that were exhibiting signs of an immunodeficiency syndrome. It has subsequently been classified as a lentivirus in the Retroviridae family based on morphology and genome organisation. Human immunodeficiency virus (HIV) is also a member of the genus Lentivirus and both viruses behave similarly in their respective hosts. Therefore, FIV is becoming an increasingly useful small animal model in understanding HIV infection, particularly in the development of vaccines and therapeutic agents. In domestic cats, seroepidemiological surveys have revealed a worldwide distribution of FIV with variable prevalence. In some regions of the United States, FIV prevalence is less than 2 % of the healthy cat population whilst in Japan FIV is found in up to 12 % of healthy cats. The same surveys recorded prevalences in sick cats of 14 % and 44 %, respectively. South Africa could be considered to have a relatively high FIV seroprevalence with a recent study finding that 22.2 % (101/454) of sick domestic cats were FIV antibody positive. The variation in prevalence is likely to reflect the different systems of cat management throughout the world. Since FIV is spread via saliva during biting, the increased prevalence in certain countries is thought to be associated with cats being allowed to roam more freely between indoors and outdoors, with a consequent increase in territorial fighting. Lentiviruses are characteristically unstable viruses with genetic variants constantly evolving due to the high error rate of the viral polymerase enzyme during DNA synthesis. The complete genome sequence of FIV was first reported in 1989 and subsequent sequence comparisons have revealed high levels of heterogeneity of the envelope (env) gene sequence in particular. Nine distinct variable regions (V1–V9), interspersed by more conserved domains, have been identified within the FIV env gene. Based on sequence differences in the variable V3–V5 region of the env gene, FIV has been divided into 5 subtypes (designated A, B, C, D, and E). Nucleotide sequence divergence among subtypes is 17.8–38.0 %, whereas intra-subtype differences are 2.5–15.0 %. Worldwide, subtypes A and B are the most prevalent subtypes and have a relatively wide geographical distribution in comparison with subtypes C, D and E. Subtypes A and B have been found in the United States, Canada, Europe, Australia and Japan. Subtype C has been identified predominantly in Canada, Vietnam and Taiwan. Subtypes D and E have a more limited distribution, having been isolated only in Japan and Argentina, respectively. The extensive genetic diversity of FIV has been a major obstacle in the development of a successful vaccine. In 2002, a dual subtype FIV vaccine became commercially available (Fel-O-Vax FIV, Fort Dodge Animal Health). This vaccine contains inactivated infected cells and whole viruses from subtypes A and D. In addition to providing protection against homologous subtypes, this vaccine has been shown to be protective against subtype B challenge. Vaccinated cats developed broad-spectrum humoral and cellular immunity. Identifying the type and diversity of FIV strains is essential, not only to minimise the chance of vaccine breakdown in regions where vaccination is to be implemented but it will also assist in the development of molecular based diagnostic assays. Currently, vaccinated and infected cats cannot be distinguished because diagnostic tests rely on the detection of FIV antibodies. Therefore, new assays based on viral nucleic acid or antigen detection will become increasingly important as they directly detect components of the virus. However, the accuracy of some of these diagnostic tests may be affected by variability in the target sequence, and therefore an understanding of subtype variation is required. The prevailing FIV subtype(s) present in the South African domestic cat population is unknown. Preliminary data from a worldwide survey were based on only 2 naturally infected South African cats whose FIV clustered with subtype A isolates. This was based on an indirect measure of variability, a heteroduplex mobility assay on PCR fragments encompassing the V3–V4 region of the env gene. In this


INTRODUCTION
Feline immunodeficiency virus (FIV) was first isolated in 1986 from a group of cats in California that were exhibiting signs of an immunodeficiency syndrome 27 .It has subsequently been classified as a lentivirus in the Retroviridae family based on morphology and genome organisation 23,26 .Human immunodeficiency virus (HIV) is also a member of the genus Lentivirus and both viruses behave similarly in their respective hosts 2 .Therefore, FIV is becoming an increasingly useful small animal model in understanding HIV infection, particularly in the development of vaccines and therapeutic agents.
In domestic cats, seroepidemiological surveys have revealed a worldwide distribution of FIV with variable prevalence.In some regions of the United States, FIV prevalence is less than 2 % of the healthy cat population whilst in Japan FIV is found in up to 12 % of healthy cats 11,41 .The same surveys recorded prevalences in sick cats of 14 % and 44 %, respectively.South Africa could be considered to have a relatively high FIV seroprevalence with a recent study finding that 22.2 % (101/454) of sick domestic cats were FIV antibody positive 33 .The variation in prevalence is likely to reflect the different systems of cat management throughout the world.Since FIV is spread via saliva during biting, the increased prevalence in certain countries is thought to be associated with cats being allowed to roam more freely between indoors and outdoors, with a consequent increase in territorial fighting 28 .
Lentiviruses are characteristically unstable viruses with genetic variants constantly evolving due to the high error rate of the viral polymerase enzyme during DNA synthesis 2 .The complete genome sequence of FIV was first reported in 1989 23,36 and subsequent sequence comparisons have revealed high levels of heterogeneity of the envelope (env) gene sequence in particular.Nine distinct variable regions (V1-V9), interspersed by more conserved domains, have been identified within the FIV env gene 24 .Based on sequence differences in the variable V3-V5 region of the env gene, FIV has been divided into 5 subtypes (designated A, B, C, D, and E) 8,20 .Nucleotide sequence divergence among subtypes is 17.8-38.0%, whereas intra-subtype differences are 2.5-15.0% 4,34 .
Worldwide, subtypes A and B are the most prevalent subtypes and have a relatively wide geographical distribution in comparison with subtypes C, D and E. Subtypes A and B have been found in the United States, Canada, Europe, Australia and Japan 1,13,21,34 .Subtype C has been identified predominantly in Canada, Vietnam and Taiwan 1,19,39 .Subtypes D and E have a more limited distribution, having been isolated only in Japan and Argentina, respectively 20,25 .
The extensive genetic diversity of FIV has been a major obstacle in the development of a successful vaccine 5,40 .In 2002, a dual subtype FIV vaccine became commercially available (Fel-O-Vax FIV ® , Fort Dodge Animal Health).This vaccine contains inactivated infected cells and whole viruses from subtypes A and D. In addition to providing protection against homologous subtypes, this vaccine has been shown to be protective against subtype B challenge 16,29,30 .Vaccinated cats developed broad-spectrum humoral and cellular immunity 30 .
Identifying the type and diversity of FIV strains is essential, not only to minimise the chance of vaccine breakdown in regions where vaccination is to be implemented but it will also assist in the development of molecular based diagnostic assays.Currently, vaccinated and infected cats cannot be distinguished because diagnostic tests rely on the detection of FIV antibodies 18 .Therefore, new assays based on viral nucleic acid or antigen detection will become increasingly important as they directly detect components of the virus.However, the accuracy of some of these diagnostic tests may be affected by variability in the target sequence, and therefore an understanding of subtype variation is required.
The prevailing FIV subtype(s) present in the South African domestic cat population is unknown.Preliminary data from a worldwide survey were based on only 2 naturally infected South African cats whose FIV clustered with subtype A isolates 1 .This was based on an indirect measure of variability, a heteroduplex mobility assay on PCR fragments encompassing the V3-V4 region of the env gene.In this study, we present sequence diversity data of the V3-V5 region of the FIV env gene from domestic cats in South Africa.

Viral samples
Thirty-three blood samples from South Africa were sent to the Veterinary Virology Laboratory, University of Queensland, Australia.All samples were collected from FIV positive cats, based on antibody detection by the ELISA method (FeLV antigen/FIV antibody Snap Combo Plus, IDEXX Laboratories).The samples originated from cats presented to the Onderstepoort Veterinary Academic Hospital in Pretoria, private veterinary practices in Durban and Cape Town, and a cat rescue centre in Johannesburg.The cats were bled from either the jugular or cephalic veins and blood was drawn into heparin tubes.The samples required gamma irradiation (an Australian Quarantine and Inspection Service requirement) before handling in Australia.Before the samples were received, 2 Australian samples collected during a similar survey underwent a 'trial' irradiation to ensure that the PCR was not adversely affected by the irradiation procedure.The 2 samples were successfully amplified following the 50 k Gray irradiation.

DNA extraction
Genomic DNA was extracted from 200 µ of whole blood using the Wizard ® SV Genomic DNA Purification System (Promega), according to the manufacturer 's instructions.The genomic DNA was eluted in a total volume of 250 µ nuclease-free water and stored in aliquots at -20 °C until use.
The genomic DNA of 4 samples was extracted using the BioRobot ® EZ1 Workstation and EZ1 DNA blood cards (Qiagen) following manufacturer 's protocols.This system utilises magnetic beads to bind DNA.

Polymerase chain reaction (PCR)
The V3-V5 region of the env gene was amplified in 2 overlapping fragments using the primer pairs su3-su4 and su5-su6.Primers su3-su4 amplify the region between nucleotides 7314 and 7806 of the published Petaluma sequence (GenBank accession number M25381) and the primer pair su5-su6 amplify from nucleotides 7660 to 7942 23 .Primer sequences are as follows: su3 5' ATWCCAAAATGTGGATGGTGG 3' su4 5' AATAAGGTCATCTACCTTCAT 3' su5 5' AATCCTGTAGATTGTACCATG 3' su6 5' TCCTGCYACTGGRTTATACCA 3' PCR amplification was performed in a reaction mixture (12.5 µ ) containing 2 mM MgCl 2, 200 µM of each deoxynucleoside triphosphate, 0.8 µM of each primer and 1 unit Expand High Fidelity Enzyme mix (Roche®).Reaction conditions were an initial denaturation at 94 °C for 1 minute, followed by 40 cycles of denaturation at 94 °C for 1 minute, annealing at 55 °C for 1 minute and extension at 72 °C for 1 minute.This was followed by a final extension at 72 °C for 10 minutes.
For some products, 2 rounds of PCR were performed.The band of the appropriate size was excised from an agarose gel and gel purified followed by sodium acetate and isopropanol precipitation and used to reseed a PCR with conditions as described above, with the exception of 1.5 mM MgCl2 and an annealing temperature of 60 °C in a 30 cycle PCR program.

Sequencing
PCR amplified fragments were purified using the MinElute™ PCR Purification Kit (Qiagen) and directly sequenced using the Big Dye Terminator Cycle Sequencing Kit (v3.1) (Applied Biosystems) using both forward and reverse PCR primers or using 1 primer at least twice.Full length (V3-V5) sequences were deduced from alignment of the overlapping fragments.Owing to poor sample quality, 6 samples (Ca10, Jo6, Pr4, Pr5, Pr6, and Pr8) could be sequenced for the shorter su5-su6 fragment only and 1 sample (Ca6) could be sequenced for the su3-su4 fragment only.Two samples from Johannesburg were excluded from the analysis as they could not be amplified.

Sequence and phylogenetic analysis
Multiple sequence alignments were created with ClustalW 37 aligning the sequences in this study with up to 38 previously reported sequences outlined in the caption of Fig. 1.Amino acid sequences were compared to evaluate the conservation of amino acid sites such as cysteine residues and N linked glycosylation sites.Phylogenetic and molecular evolutionary analyses were conducted using MEGA version 3.1 15 .Genetic distances between pairs of sequences were determined using the Kimura 2-parameter model 14 .An unrooted neighbour joining tree 32 was constructed to determine the phylogenetic relationships.In each instance, bootstrap analysis 7 was performed with 1000 repetitions to evaluate clade consistency.Information on the direction of evolutionary change could not be deduced as there is no outgroup sequence.Phylogenetic trees were generated using full length sequences and sequences obtained from the individual fragments (5' su3-su4 and 3' su5-su6).

Sequence analysis
Up to 682 nucleotides, representing the V3-V5 region of the env gene were sequenced from 31 blood samples of FIV infected cats from 4 regions of South Africa (GenBank accession numbers DQ873700-DQ873733).Sequences of this whole region (in 2 overlapping fragments) were achieved for 24 samples while only the 3' su5-su6 region was able to be sequenced for 6 samples and for 1 sample only the 5' su3-su4 region was successfully amplified and sequenced.Comparisons of the predicted amino acid sequences with representatives from each of the subtypes confirmed that the targeted region was sequenced.In concordance with previous reports, cysteine residues and N-linked glycosylation sites were relatively well conserved, which reflects their functional importance to the envelope glycoprotein (data not shown) 12,20,25 .

Phylogenetic analysis
Analysis of pairwise genetic distances is indicative of the presence of 2 subtypes within South Africa with pairwise genetic distances of up to 38.1 % observed with the su5-su6 sequences, whereas the su3-su4 sequences showed a maximum pairwise genetic distance of 14.8 % 4,12 .
In order to classify the sequences from this study into the appropriate subtypes, neighbour joining phylogenetic trees were constructed with sequences representative of each of the subtypes.Phylogenetic analyses showed that the majority of South African sequences clustered within subtype A (Fig. 1).In Fig. 1 (a) samples were excluded from the phylogenetic analysis if only 1 fragment was amplified or if inconsistencies in subtype assignment between the 2 fragments were observed.For these sequences, separate phylogenetic analyses of the su3-su4 and su5-su6 fragments were performed.The su3-su4 fragments (25 samples) clustered within subtype A. The su5-su6 fragments clustered either within subtype A or subtype C (Fig. 1 b).
For the majority of samples, phylogenetic grouping of sequences of the individual fragments was consistent with respect to subtype classification when analysed separately.However, for 3 samples (Pr9, Ca11, Ca12), an inconsistency with subtype assignment was noted when the different fragments were subjected to phylogenetic analysis.The 5' fragments, amplified with the primers su3-su4, clustered within subtype A while the 3' fragments, amplified with the primers su5-su6, grouped with other subtype C viruses confirming the presence of 2 sub-types, A and C, in domestic cats in South Africa.
In this study, isolates within South Africa did not cluster by geographical region, with sequences from each region distributed throughout the tree.The South African A and C isolates did not cluster separately from the other A or C isolates from other regions of the world.However, bootstrap values are generally quite low within the subtypes, and therefore a definitive association between subtype and geographical location cannot be made.

Subtype relationship to clinical disease
The cats in this study showed clinical signs that are often associated with FIV infection and these are summarised in Table 1.Only the 4 feral cats sampled from Johannesburg were considered to be free of clinical disease.A wide variety of clinical signs was reported, the most common being anorexia and weight loss.Other common presenting signs were pyrexia, lethargy, inflammation of the mouth, skin disease, haematological disorders, neoplasia, hepatic and respiratory disease.No obvious relationship of subtype to clinical disease was apparent.

DISCUSSION
This preliminary study demonstrates the presence of FIV subtypes A and C in the domestic cat population of South Africa, with subtype A the most prevalent.The presence of subtype A had previously been suggested based on heteroduplex mobility analysis utilising the V3-V4 region of the env gene of the virus from only 2 cats from South Africa 1 .
Sequences were generated for 2 overlapping fragments encompassing the V3-V5 region of the env gene.For most samples, (21/24), the individual fragments grouped consistently within subtype A when phylogenetic analyses was performed.However, 3 isolates (Ca11, Ca12, Pr9) showed an inconsistency with subtype assignment.It was observed that the 5' region of these sequences (amplified by su3-su4) most closely resembled subtype A while the remaining downstream sequence appeared most closely related to subtype C. Dual infection and/or intersubtype recombination, either within an individual or as a PCR artefact, may explain some of these findings.However, to definitively confirm the presence of recombinants, analysis of full length genomes is necessary.
The genetic diversity of lentiviruses, including FIV, is a result of errors that occur during replication, in addition to recombination 1,2 .Superinfection and recombination in FIV infected cats have been demonstrated under experimental conditions 1,17,22 and as more sequence data become available from around the world, there is increasing evidence to support their occurrence during natural FIV infection, particularly in areas where different subtypes are known to exist 31 .Dual infection and recombination are, therefore, theoretical possibilities in South Africa and may explain the subtype discrepancies observed with the 3 discordant isolates.
Although many disease states are associated with FIV, the association between subtype and severity of disease is still uncertain.Potential correlations between FIV subtype and disease have been reported 1,3,21 .Cats infected with subtypes A or C are generally more likely to be symptomatic than cats infected with subtype B 1,20 .Ikeda et al. 10 also found that subtype A infected cats developed AIDSrelated clinical signs earlier than cats with subtype B infection 10 .In the present study, cats harboured subtype A or C viruses and except for the 4 feral cats from Johannesburg, all cats demonstrated some sign of disease.The sample popula-tion could be considered to be biased because, apart from the feral cats, all the sampled cats were visiting a veterinary clinic.Nonetheless, there appeared to be no difference in clinical status between the cats infected with subtype C or subtype A in this study.In humans, several studies investigating the relationship between HIV subtype and pathogenicity have been performed with varying conclusions 9.38 .It is possible that any observed differences could be due to confounding factors such as environmental factors and, therefore, more studies are required to determine whether any significant correlation exists between subtype and pathogenicity.
FIV isolates have been observed to cluster according to geographical location.Two types of geographical clustering have been noted.Firstly, subtypes may group by geographical location within a country.An example of this is seen among FIV infected cats in Japan where subtype B is found predominantly in eastern regions and subtype D is found mainly in western regions 20 .Secondly, the presence of geographically localised subclusters within a subtype has been observed.For example, Austrian and Portugese isolates form distinct clusters within subtype B 4,35 .In the phylogenetic trees constructed from sequences in this study, no clustering of isolates from similar geographical regions was observed, nor did the South African isolates form a unique cluster within subtypes A or C.However, it is premature to infer that this is not occurring due to the low bootstrap values observed within subtype A in particular.
The substantial and increasing genetic diversity of FIV has made the development of a successful vaccine difficult.However, the Fel-O-Vax FIV ® dual subtype vaccine has been shown experimentally to provide good protection against homologous and heterologous subtype challenge 16,29,30 .Its efficacy in the field had been uncertain but recent studies designed to mimic natural conditions have supported its efficacy against more diverse FIV 16,29 .It was found to afford good protection against a subtype B (Aomori2) infection when vaccinated, unvaccinated and infected cats were mingled.Additionally, it was found to be efficacious against a virulent in vivo derived subtype B challenge 29 .However, others have reported that the vaccine afforded no protection against a virulent subtype A challenge 6 .
It is difficult to speculate on the potential efficacy of this FIV vaccine in South Africa.The majority of evidence suggests that it would be beneficial in reducing infection with subtype A viruses (which predominate in South Africa) but its efficacy against subtype C infection is uncertain.The ability of the vaccine to protect against heterologous subtype challenge is suggestive of the development of broad spectrum immunity.This broad spectrum immunity may be the consequence of utilising 2 subtypes (subtypes A and D) within the vaccine 30 .It has been speculated that using 2 subtypes may direct the immune system to common epitopes which could enhance vaccine immunity across the subtypes 42 .Given that the vaccine provides protection against diverse subtype B infection, it is possible that protection will be achieved against other subtypes and more diverse viral strains.However, this cannot be confirmed until vaccine trials are performed with more diverse challenges.
In conclusion, this study has found FIV subtypes A and C in the South African domestic cat population, with subtype A predominant.Of interest is the finding of a high degree of FIV genetic diversity among infected cats in South Africa and the detection of diverse sequences within some individual cats.Further research is required to fully elucidate these findings.This knowledge assists in our understanding of FIV diversity which may have implications in preventative and diagnostic strategies.