Introduction

Feline immunodeficiency virus (FIV) is a retrovirus of the genus Lentivirus that infects members of the family Felidae and Hyaenidae. This widespread pathogen is closely related to human immunodeficiency virus (HIV), sharing a similar structure, life cycle and pathogenesis, but humans are not susceptible to FIV infection [1, 2]. Based on the significant similarities between FIV and HIV, FIV serves as a valuable model for prophylactic and therapeutic studies of HIV [3].

Since its discovery in 1986 in a cat with immunodeficiency-like syndrome [4], FIV has been isolated from a variety of non-domestic members of the family Felidae, including puma (Puma concolor), lion (Panthero leo), leopard (Panthera pardus), pallas cat (Otocolobus manul) and bobcat (Lynx rufus). Apparently, FIV is less pathogenic in these species than in domestic cats (Felis catus). The difference in the disease outcome suggests that FIV has persisted in non-domestic cats much longer than in domestic cats, and therefore, adaptation of the host has occurred [5–9].

FIV causes an AIDS-like disease in domestic cats that is characterized by a latent phase, followed by depletion of CD4+ T lymphocytes [10]. Most clinical signs that FIV-infected cats present are not directly caused by FIV itself but instead are caused by the secondary and opportunistic infections and immune stimulation resulting in immune-mediated disease. It appears that certain virus subtypes can cause neurological disease as well [11]. The majority of natural FIV infections are acquired by biting, presumably through the inoculation of the virus, or virus-infected cells from the saliva of persistently infected cats.

FIV is endemic in the domestic cat population worldwide. The seroprevalence is highly variable in different geographical areas, with estimates of 1 to 14 % in cats with no clinical signs and up to 44 % in sick cats [12]. In Italy, the seroprevalence of FIV in sick cats ranged from 24 to 30.9 % [13, 14], while antibodies against FIV were found in 10.7-14.1 % of clinically healthy cats [14, 15]. Similar results were obtained in Spain, where 8.3 % of healthy cats and 13.9 % of sick cats were FIV positive [16]. Other studies performed in Europe showed 8.4 % FIV-infected cats in Germany [17], 10.4 % in England [18] and 33.3 % in Slovenia [19]. In Croatia, FIV was serologically confirmed in 3.3 % of healthy and 10.3 % of sick cats [20]. Recently published data point out the seroprevalence of 20.88 % FIV-positive sick cats in the urban areas of the city of Zagreb [21].

As with human immunodeficiency virus (HIV), several genetically distinct subtypes have been reported. Based on the nucleotide sequence diversity of the V3-V5 region of env, FIV is currently classified into five subtypes (clades A-E) [22–24]. Subtype A isolates are common in Australia, New Zealand, the western part of the United States, South Africa and northwestern Europe [25–30]. Subtype B has most frequently been isolated in the central and eastern United States, central and southern Europe, Brazil and eastern Japan [22, 23, 26, 31–33]. Subtype C has been identified in Canada, New Zealand and Southeast Asia [34–37]. Subtypes D and E are quite infrequent but were found originally in Canada, Japan and Argentina [24, 32]. Due to relatively rapid rates of recombination, several sequences of unknown subtypes have been documented in isolates from Texas [38], Portugal [39] and New Zealand [27, 40].

The phylogenetic diversity of FIV sequences is explained by the occurrence of relatively frequent changes within the env gene. The heterogeneity of the env gene sequence poses a great problem for developing a broadly effective and protective vaccine. There have been several attempts to design a protective vaccine against FIV infection. Single-subtype vaccines have shown protection against challenge with homologous or slightly heterologous virus but failed to protect against more distantly related strains [41]. In contrast, the Fel-O-Vax (Fort Dodge) inactivated infected-cell vaccine, which is commercially available in a number of countries, including the US, Australia, Japan and New Zealand, provides combined protection against different subtypes of FIV [42]. These differences highlight the impact of genetic diversity on the efficacy of vaccines against FIV and the importance of assessing the genetic diversity of local subtypes for vaccine development. The recognition of predominant strains in a particular region is essential for developing an effective vaccine and suitable reagents for molecular diagnosis of FIV infection. So far, there have been no studies describing molecular and phylogenetic characterization of FIV in Croatia. For that reason, the aim of this study was to determine the phylogenetic relationships of FIV sequences identified in Croatia, as well as to asses and compare the specificity and sensitivity of the rapid ICA diagnostic assay to that of the PCR-based assay detection of the provirus.

Materials and methods

Research was performed on 29 domestic cats presented to the Department of Microbiology and Infectious Disease with Clinic at the Faculty of Veterinary Medicine, University of Zagreb. All of the cats had clinical signs that could be related to FIV. Common findings in the physical examination included lethargy, fever, generalized peripheral lymphadenopathy, chronic stomatitis, upper respiratory tract infection, diarrhoea and ocular abnormalities. Laboratory abnormalities included anaemia, lymphopenia and hyperproteinemia. All of the 29 tested domestic cats had no vaccination history.

During clinical treatment, peripheral blood samples were collected into EDTA-coated tubes for commercial rapid immunochromatographic testing for the qualitative detection of FIV antibodies (MEGACOR Diagnostik, Hoebranz, Austria) and for the performance of polymerase chain reaction (PCR).

Immunochromatographic testing was performed immediately after blood sampling according to the manufacturer’s instructions, and the rest of the blood was stored at -80 °C until the molecular assay was performed.

All whole blood samples tested were initially analysed by PCR. DNA was extracted from EDTA-treated cat whole blood using a QIAamp DNA Mini Kit and Blood Mini Kit (QIAGEN, Hilden, Germany). The proviral gag gene was amplified with primers FIV-1026F (5’ –GGC ATA TCC TAT TCA AAC AG- 3’) and FIV-1700R (5’ – AAG AGT TGC ATT TTA TAT CC- 3’) [43]. Agarose gel electrophoresis was performed to visualize PCR products. PCR analysis revealed 11 positive samples out of 29 tested. All 11 positive PCR product samples were sequenced in both directions by Macrogen Europe, Amsterdam, The Netherlands.

The nucleotide sequences were initially subjected to a BLAST search, using the blastn and blastx algorithms (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Sequences were assembled using Pregap4 and Gap4 implemented in Staden software [44]. A multiple sequence alignment of the 11 Croatian sequences and 44 FIV gag sequences retrieved from the GenBank database, was performed using the ClustalW codon algorithm implemented in the Mega 6.0.6. software package [45]. The compared gag sequence fragments consisted of 558 nucleotides, corresponding to nucleotides 1084-1641 of the reference strain (Petaluma; M25318), and were translated into 187 amino acids.

Sequence identity matrices for FIV nucleotide (nt) and amino acid (aa) gag sequences were calculated using BioEdit [46].

The GenBank accession numbers of the sequences used in this study are as follows: subtype A – M25381 (Petaluma), AY684181 (Bangston), M36968 (PPR), L06136 (Wo), GQ339818 (DENW032), GQ339841 (DENS072), GQ339863 (DERP115), X57002 (SwissZ1), GU055218 (UK8), D37820 (Sendai1), GQ339871 (GB127), DQ365591 (TN3), X68109 (DutchFIV-113), AF531075 (ATVId02), GQ339808 (BOc10), GQ339870 (LUCL126), GQ339869 (DEBA121), GQ339868 (L120), AF531070 (DEBAd58), GQ339821 (DENW038), GQ339822 (DEBR039) GQ339847 (DENW091), GQ339859 (DERP108), AF531069 (DEBAb91), GQ339850 (DEHA097); subtype B – U11820 (USIL2849_7B), AF361320 (Maryland), D37823 (Aomori1), D37824 (Aomori2), D37821 (Sendai2), Y13867 (M2), Y13866 (M3), D37819 (Yokohama), DQ365589 (TN1), DQ365595 (TN8), AY139105 (TX120), DQ365596 (FC1), AF531049 (ATESb20), AF531050 (ATESd03), AF531054 (ATSTb30), AF531055 (ATVIb97), AF531057 (ATVIa33), AF531057 (ATNOd16), AF531058 (ATSTc01), AF531066 (ATVId23), AF531060 (ATVIb31), AF531062 (ATVId05), GQ339806 (ATVIc09), AY196330 (ATNOc07), AF531060 (ATNOd01), AJ304959 (PP2), AJ304962 (RP1), (ITROd78), (ITROd76), AF531061 (ATVIa85), AF531059 (ATVIa90), AF531064 (ATVId20); subtype C – AF474246 (BM3070), AY600517 (C36); subtype D – D37822 (Fukuoka), AY679785 (Shizuoka); subtype E – AB027302 (LP3), AB027303 (LP20), AB027304 (LP24), EF413007 (RUS02), EF413009 (RUS04), EF413011 (RUS06), EF413014 (RUS09), EF413015 (RUS10), EF413016 (RUS11), EF413017 (RUS12), EF447304 (RUS14); subtype F – GQ406242 (NZ1).

The phylogenetic reconstruction of FIV gag sequences was based on p-distance analysis of nucleotide sequences, and maximum-likelihood inference of protein sequences. The best-fit model for phylogenetic analysis of FIV protein sequences was determined using the ProtTest program [47]. The JTT+G model was selected as the most appropriate evolutionary model according to both AIC and BIC. P-distance analysis was used as the best-fit model for phylogenetic analysis of nucleotide sequences, as estimated using jModelTest V.0.1.1 [45]. P-distance analysis was performed with the MEGA 6.0.6 software package [45], and a maximum-likelihood (ML) tree was constructed using PhyML-3.1, implemented in SeaView software version 4 [48].

The phylogenetic inference based on the p-distance (for nt sequences) and the ML method (for aa sequences) was consistent with molecular analysis of the FIV gag sequences.

Results

Of 29 blood samples tested, 20 were positive in the rapid immunochromatographic assay (ICA), seven were ICA negative, and two gave borderline results. Subsequently, all 29 blood samples were tested using molecular diagnosis. PCR assay revealed proviral DNA in 11 blood samples. Both borderline ICA samples were negative by PCR. The results showed that the sensitivity of rapid immunochromatographic testing for the qualitative detection of FIV antibodies in comparison to the PCR method was 100 %, and the specificity was 66.67 % (Table 1).

Table 1 Results of serological and molecular diagnosis of naturally infected cats
Full size table

The similarity between the Croatian sequences was higher at the amino acid level (95.6-100 %) than at the nucleotide level (91.2-98.9 %). Compared to aligned FIV gag sequences, sequence identities ranged from 81.3 to 96.7 % for nt sequences, and 85.5-100 % for aa sequences. Generally, the greatest sequence similarity among the Croatian sequences was observed with FIV isolates of the B subtype.

Three Croatian sequences (CRO-3, CRO-10, CRO-11) showed a unique change at nucleotide 1096 of the reference strain, suggesting the occurrence of a synonymous substitution that did not have any effect on amino acid composition.

Croatian FIV gag sequences were clustered within the B subtype, with the majority demonstrating a close phylogenetic relationship to Austrian isolates (Fig. 1).

Fig. 1
figure 1

ML tree based on a 187-amino-acid alignment of 83 FIV gag sequences. Croatian FIV sequences are shown in bold. ML distances were calculated based on the best-fit evolutionary model selected by ProtTest (JTT with gamma shape parameter = 0.67 according to the BIC criterion). The numbers near the branches indicate the percentage of probability calculated by 1000 bootstrap iterations (only values ≥70 are shown). The scale bar at the bottom of the tree represents the number of amino acid substitutions per site

Full size image

Discussion

Feline immunodeficiency virus (FIV) represents a great problem in veterinary medicine, causing severe illness and death in domestic cats. FIV is spread worldwide with prevalence estimates from 1 to 14 % in cats with no clinical signs and up to 44 % in sick cats [12]. The prevalence of infected cats in Croatia coincides with the data from other European countries in which 3.3 % of tested healthy cats were found FIV positive, and in a group of sick animals, 10.3 % cats were infected [20]. More-recent data indicated that 20.8 % of sick cats in the Zagreb area were FIV positive [21]. Due to their nonspecific clinical manifestations, relatively long incubation period, and possible long-term asymptomatic phase, FIV infections are often left undiagnosed.

There are currently a number of different commercial tests available to detect FIV infection based on detection of specific antibodies.

In clinical practice, commercial ICA tests designed to detect specific antibodies are widely used. The specificity and sensitivity of commercial ICA tests are inconsistent. One comparison of six in-house assays found positive predictive values that ranged from 83 % on the high end to only 51 % on the low end, while the negative predictive values ranged from 96 % to 99 % [49]. Such results emphasize the importance of confirmatory testing, particularly in cases of positive results. When compared with the gold-standard Western blot assay, the ICA test used in our research (MegaCor Diagnostic Fastest FIV) in a previous study showed a sensitivity of 96.4 % and a specificity of 99.2 % [50].

A variety of PCR assays have been developed for detection of FIV infection. The sensitivity and specificity vary among laboratories and methods from 41 % to 80.5 % and from 44 % up to 99.9 %, respectively [51, 52]. PCR can be insensitive because viral loads in healthy cats are often extremely low, and some strains may not be detected because of variability in the sequence of the viral genome among FIV isolates. False positive test results have the potential to occur as a result of laboratory contamination. The specific PCR primers and PCR protocol used in our study were able to amplify subtype B, which is predominant in central Europe, as well as subtype A, which is the predominant subtype in northwestern Europe [26, 53].

One of the goals of this work was to investigate the agreement between the commercial ICA and PCR assay results. For 18 out of 29 tested samples, the ICA and PCR assays showed consistent results. In our study, the sensitivity of ICA was 100 % while the specificity was 66.67 % when compared to PCR. The unusually low specificity of 66.67 % for the ICA test used in our study compared to PCR (9 ICA positive but PCR negative samples) could be repercussive and the result of a low proviral DNA load. On the other hand, variability in the viral genome of FIV always presents a great problem for amplification due to sequence variation. Despite using primers that were able to amplify non-B subtypes in previous studies, it is possible that other subtypes of FIV are circulating in Croatia that were not detected with these primers [26, 53]. With the introduction of the FIV vaccine and the associated problems of interpreting serologic test results in vaccinated animals, the molecular diagnostic technique found its application and important role. If appropriately inactivated, the application of the vaccine should not result in provirus production and thus would not interfere with PCR assays that detect proviral DNA.

Any ICA-positive result in a low-prevalence population (e.g., young, indoor, pure-bred cats) must therefore be confirmed by Western blot, which is considered a “gold standard” for FIV serology testing and is used to confirm questionable results.

Based on sequence diversity, several main subtypes of FIV have been found. Studies suggest regional differences in subtype distribution. Information about subtype distribution is crucial for appropriate diagnosis and eventually, introduction of an FIV vaccine. For that reason, the goal of this study was to investigate, for the first time, the phylogenetic relationships of feline immunodeficiency virus in Croatia. As pointed out previously, subtypes A and B are the most commonly occurring ones worldwide [33]. These subtypes are also predominant in Europe. It is interesting to note that the majority of the isolates sequenced in northwestern Europe (Germany, UK) were clustered in the A subtype, while in central and southeast Europe (Italy, Austria, Turkey, Croatia), the majority of isolates were of subtype B [26, 31, 54, 55]. Portuguese FIV sequences were clustered into A and B subtypes, but a few sequences might represent the prototype of a new subtype termed subtype F. This hypothesis supports the opinion that Texas isolates represent a unique cluster, possibly a new subtype, arising from subtype B [38, 39]. According to available GenBank data, Russian isolates are clustered within the E subtype. To our knowledge, there are no other reports of phylogenetic analysis of FIV sequences from Eastern European countries. The results show that Croatian isolates were clustered within the B subtype, showing close phylogenetic relationship with Austrian isolates, most likely due to geographical vicinity. Despite being closely related to Austrian sequences, a group of six Croatian sequences (CRO3, CRO6, CRO7, CRO8, CRO10, CRO11) branched separately, indicating the possible existence of multiple subgroups within the B subtype. Apart from this separate group of six isolates, phylogenetic analysis of Croatian FIV sequences indicated additional heterogeneity. A group of four Croatian sequences (CRO1, CRO4, CRO5, CRO9) are branched separately with Austrian isolates. Further study and a larger number of sequences originating from Eastern Europe are needed to obtain a deeper insight into the geographically based diversity of FIV. Sequence CRO2, although classified within the B subtype with all other Croatian isolates, is on a different subtype B branch. For this particular cat, we do not have a history of travel or origins outside Croatia, and there was no major clinical difference in clinical presentation compared to the rest of the examined cats.

Further investigation of FIV isolates in Europe, particularly Eastern Europe, is crucial for a better understanding of the epidemiology and circulation of different FIV strains and for developing more-reliable diagnostic tools for detection of FIV infection. In addition to improving diagnostics, future research and efforts are necessary in order to develop and introduce an effective vaccine against FIV in Europe.

Conclusions

This study revealed for the first time the phylogenetic characteristic of Croatian FIV strains. All Croatian sequences were clustered within subtype B, which is the predominant subtype in Central Europe. Croatian isolates show close similarities to Austrian isolates. Despite the fact that all Croatian isolates were clustered in subtype B, phylogenetic analysis suggested three separate clusters circulating in Croatia. To our knowledge, there are no reports of phylogenetic characteristics of FIV isolates in the majority of Southeast European countries. This confirmed heterogeneity of Croatian isolates emphasizes the importance of further analysis to optimize diagnostic and immunoprophylaxis of FIV. In addition to phylogenetic analysis of Croatian FIV, results of this study indicate the importance of confirmatory testing for FIV-positive ICA and FIV-negative PCR results in the low prevalence population using the gold-standard Western blot technique.