INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Feline immunodeficiency virus (FIV) is a useful model for investigating strategies for human immunodeficiency virus type 1 (HIV-1) vaccination because of important similarities between the two viruses in terms of immunobiology, pathogenesis, and disease induction (7, 20, 44, 46, 67). Like HIV-1 isolates, FIV isolates are highly heterogeneous. Five subtypes of FIV (designated A to E), which are differently distributed throughout the world, have been recognized, and even within a given subtype, genetic and antigenic heterogeneity is high (17, 29, 45, 49, 60). Thus, like anti-HIV-1 vaccines, anti-FIV vaccines should elicit broad-spectrum protective immune responses in order to defend against the wide variety of viral strains that circulate in nature.

Vaccine approaches tested so far with the FIV/cat model include inactivated whole viruses, fixed infected cells, recombinant proteins, peptides, and DNA plasmids (9, 15, 25-28, 33-35, 38, 39, 51-53, 59, 65, 69). While recombinant Gag and Env and env DNA have usually exerted marginal or no protective activity and, in some instances, appeared to facilitate subsequent challenge infection (33, 35, 59), fixed infected cell and inactivated cell-free virus vaccines have generally proved efficacious against homologous or closely related strains of FIV (26, 69) and also conferred short-lived protection against an ex vivo-derived strain (38, 39). However, even the latter vaccines have failed to generate significant protection against highly heterologous strains (25).

Previous studies have unequivocally demonstrated that certain neutralization antigens of FIV, such as those measured by assays performed in fibroblastoid CrFK cells, are shared among most, possibly all, FIV isolates (43, 64). Thus, one possible explanation for the failure of anti-FIV vaccines to protect against heterologous strains was that the forms of immunogens used so far did not trigger sufficiently powerful cellular and/or humoral immune responses to cross-protective epitopes or that these epitopes were lost, masked, or altered during preparation of the vaccines. In general, live attenuated virus vaccines produce longer-lasting, more effective, and broader protections than do inactivated or subunit vaccines (13). Thus, it was conceivable that immunization with an attenuated FIV vaccine might evoke protective immunity against heterologous challenges more effectively than the types of vaccines tested so far. Although live attenuated vaccines have been successfully developed in the simian immunodeficiency virus (SIV) model (16), this approach has yet to be tested with FIV.

Here we investigated whether preinfection with a strain of FIV partially attenuated as a result of prolonged growth in vitro could protect against subsequent infection with a highly heterologous in vivo-grown strain. The virus selected for preinfecting cats was FIV Petaluma (FIV-P), a subtype A virus widely used in vaccination experiments, which has been shown to lose a significant fraction of its virulence after prolonged propagation in vitro (4). The stock used, a high-passage virus obtained from chronically infected cells, although not specifically designed as a vaccine, is relatively avirulent in cats. The challenge virus was wild-type FIV-M2, a subtype B virus passaged only in cats, where it is highly virulent. The two viruses are 20% divergent at the amino acid level in the env gene (49). The results have shown that preinfection with subtype A FIV did not prevent superinfection by subtype B virus, in this respect confirming previous findings (42). However, preinfection prevented the increase of viral burden observed in naive cats starting from 2 years postchallenge (p.c.), thus suggesting that, in the long term, attenuated anti-FIV vaccines may exert beneficial effects also against highly heterologous virus strains. By evaluation of the contributions of the two viral strains to total viral burden, an inverse relationship between their replication dynamics, which might explain the beneficial effects, was also observed. The results have also suggested that the dose of attenuated virus used for preinfection can be critical for induction of optimal heterologous protection. These studies set the stage for experimental investigations of attenuated FIV vaccines.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Experimental design. Four groups of two specific-pathogen-free (SPF) cats each were injected with graded doses of a high-passage in vitro-derived FIV-P and monitored at intervals for infectious virus and proviral loads in the peripheral blood mononuclear cells (PBMC), plasma viremia, anti-FIV antibodies, and circulating CD4⁺ T-lymphocyte counts. Seven months later, the eight FIV-P-preinfected cats and two uninfected, age-matched cats were challenged with fully virulent FIV-M2 derived ex vivo. All animals were then monitored for an additional 3 years as described above, and in addition, the relative contributions of the two viral strains used for infection to total viral and proviral loads in the bloodstream were determined. At the end of follow-up, the animals were sacrificed and the proviral loads in solid tissues were measured and characterized.

Animals and infections. Female SPF cats were purchased from Iffa Credo (L'Asbrege, France), housed individually in our climatized animal facility under European Community law conditions, and used when they were 14 months old. They were evaluated for clinical symptoms weekly and bled under light anesthesia periodically. Prior to use, animals tested free of FIV and feline leukemia virus. The stock of FIV-P used was cell-free supernatant of the 181st passage of persistently infected FL4 cells (68), known to possess a neutralization-sensitive phenotype (3) and to produce low-grade infections in cats (8). For cat infection, FIV-P was 10-fold serially diluted to contain 300 to 0.3 50% cat infectious doses (CID₅₀) per ml, and 1 ml was injected intravenously. FIV-M2 was pooled cell-free plasma from three cats infected 2 weeks previously with virus never passaged in vitro. The dose of FIV-M2 used for challenge was 30 CID₅₀ administered intravenously in 1 ml and was known to produce florid infections and rapid falls of circulating CD4⁺ T cells. In vivo titers were determined as previously described (3).

FIV provirus detection and quantitation. Protocols for DNA extraction from buffy coat, testing for competence of amplification, and FIV gag-specific nested-PCR and competitive-PCR (cPCR) amplification have been described in detail elsewhere (49). Samples from solid tissues collected at necropsy were processed for DNA extraction as described elsewhere (36). The results are expressed as the number of proviral copies in 1 µg of cellular DNA.

Plasma viremia measurement. FIV RNA was detected and quantified by reverse transcriptase (RT) nested PCR and RT-cPCR, respectively. Briefly, RNA was extracted from 200 µl of cell-free plasma in modified Chomczynski and Sacchi lysis solution (Gene Dia, Lammari, Lucca, Italy) under PCR-clean conditions, pelleted, and resuspended in 40 µl of nuclease-free water. Ten microliters was amplified by RT-nested PCR for FIV gag sequences, and positive samples were then amplified by RT-cPCR in order to quantify the FIV genomes in plasma. The RNA competitor was produced by runoff in vitro RNA transcription of the HindIII-linearized pD117 plasmid containing a deletion of the FIV gag fragment and was used in the quantitative DNA assay. RNA transcripts were then purified from the DNA template by digestion with RNAse-free DNAse, phenol-chloroform extraction, and alcohol precipitation; resuspended in nuclease-free water; spectrophotometrically quantified at 260 nm; and analyzed on a 10% polyacrylamide gel to check for integrity and absence of plasmid. RT-cPCR was carried out by addition of 5 µl of plasma RNA to 5 µl of serial 10-fold dilutions (10⁶ to 10²) of RNA competitor and 10 µl of an RT mixture containing 40 pmol of FIV gag antisense primer 302M and 6 U of avian myeloblastosis virus RT (Promega, Madison, Wis.). Reaction mixtures were incubated at 42°C for 1 h followed by RT inactivation at 95°C for 5 min. PCR was performed by adding to each well a master mix (80 µl) containing 40 pmol of FIV gag sense primer 301M and 1 U of Taq DNA polymerase (Perkin-Elmer, Foster City, Calif.). Analysis of PCR products was performed as described previously (48). The method consistently detected as few as 100 copies of the RNA competitor.

Differentiation between FIV-M2 and FIV-P and determination of the relative proportions. Unless otherwise specified, discrimination between FIV-P and FIV-M2 was carried out with a recently described fluorescence-based restriction fragment length polymorphism (F-RFLP) method which exploits restriction site differences in nested gag PCR products. Amplicons were digested with the enzymes HindIII and SacII (New England Biolabs, Beverly, Mass.), selected because of the presence of unique restriction sites in the gag p25 region of FIV-P and FIV-M2, respectively. Briefly, 15 µl of PCR samples were diluted to 50 µl in an appropriate restriction buffer and digested with the two enzymes at 37°C for 2 h. The samples were then run on a 2% agarose gel and on an automatic laser fluorescence (ALF) DNA sequencer (Pharmacia Biotech, Uppsala, Sweden) as described previously (12). To determine the percentages of FIV-P and FIV-M2 in samples, restriction was applied to nested gag-specific PCR products as described previously (12). Percentages were then used to calculate the numbers of copies of each strain from the total number of FIV genomes as determined by cPCR. When indicated, the same approach was applied to env nested PCR products obtained with highly conserved primers encompassing the V3-to-V4 region. The resulting amplicons were digested with the enzymes AvaII and HinfI, selected because of the presence of unique restriction sites in the V4 region of FIV-M2 and FIV-P, respectively. Samples were then run on a 2% agarose gel and processed exactly as described above.

Virus reisolation from and quantitation of infectious units in the PBMC. FIV was isolated from the PBMC by cocultivating 10⁶ Ficoll-Hypaque-separated cells with MBM cells and testing the cultures for RT once per week as described elsewhere (37). Cultures regarded as negative showed no evidence of RT in any sample collected during the 5-week culture period. Infectious units in the PBMC were determined by limiting dilution (23).

Assays for anti-FIV antibodies. Total anti-FIV antibodies were measured by an in-house enzyme-linked immunosorbent assay (ELISA). Microwells were coated overnight with 100 µl of 2-µg/ml gradient-purified, disrupted whole FIV-P. Then the microwells were coated with skim milk, and serially diluted sera were added to the plates in duplicate. Bound immunoglobulin G (IgG) was revealed with a biotinylated mouse anti-cat IgG serum followed by an antibiotin-peroxidase conjugate. Absorbance was read at 450 nm. In order to minimize plate-to-plate variability, the results were normalized by including a control positive serum of known titer in each plate and correcting the titer of each sample based on the titer obtained for the control serum. The titers were expressed as the reciprocal of the highest dilution of serum that gave optical density readings higher than the average values obtained with 20 control FIV-negative serum samples plus threefold the standard deviation. Sera that proved unreactive at a 1:100 dilution, the lowest dilution tested, are indicated as having titers of <100. Levels of neutralizing antibodies (NA) were determined by using an assay based on inhibition of syncytium formation as described elsewhere (63). Briefly, twofold dilutions of heat-inactivated serum were mixed with 100 syncytium-forming units of FIV-P, incubated at room temperature for 1 h, and then added to 10⁴ CrFK cells/well in 24-well plates. Six days later, cultures were stained and syncytia were counted. NA titers were expressed as the reciprocal of the highest dilution of serum which completely prevented the formation of syncytia.

Lymphocyte subset composition analysis. The absolute counts of CD4⁺ and CD8⁺ T lymphocytes were obtained by flow cytometry as described elsewhere (38). CD8⁺ T-cell counts are not reported because they do not provide important information for this study.

Statistical analysis. The significance of the differences between means was evaluated by a two-sample Student's t test, assuming equal variances. This test is independent of sample size, which only affects the degrees of freedom.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Outcome of FIV-P injection. The outcome of FIV-P inoculation was monitored for 7 months. The cats given 3 to 300 CID₅₀ exhibited an infection course similar to what was observed in previous studies using the same virus stock (8, 34). They repeatedly, though not constantly, yielded positive virus cultures, were consistently low positive for proviral genomes in the PBMC, and developed moderate titers of anti-FIV antibodies. In contrast, the cats inoculated with 0.3 CID₅₀ showed no signs of infection (data not shown).

Outcome of FIV-M2 challenge. Seven months after FIV-P inoculation, the eight cats described above and two naive cats were challenged with a plasma preparation of FIV-M2 and then monitored systematically for 3 additional years, at which point they were sacrificed in order to determine the viral content in solid tissues. The main findings were as follows.

(i) Clinical conditions. With the exception of the cat that died of cryptogenetic renal failure, the animals showed no overt symptoms of disease throughout the experiment. This was not surprising, because the stock of FIV-M2 used was known to generally induce clinically inapparent infections for several years at least (6).

(ii) Proviral loads in the PBMC. Two weeks after FIV-M2 challenge, the PBMC of the two control cats contained more than 500 copies of FIV provirus per µg of DNA, and after 2 further weeks, the PBMC contained approximately 1,500 copies. Then proviral loads declined rapidly to what could be considered steady-state levels, since they remained essentially unchanged from 2 to 16 months p.c. Afterwards, however, the proviral loads of the control cats increased again, so that by 3 years p.c. they had reached peak values of 1,850 and 2,300 copies per µg of DNA (Fig. 1A).

View larger version (26K): [in this window] [in a new window]   FIG. 1.   Proviral loads in the PBMC of cats challenged with FIV-M2, as determined by cPCR and F-RFLP analyses of gag amplicons. (A) Total proviral loads in control cats. (B) Total proviral loads in cats preinfected with FIV-P. (C) FIV-M2-specific proviral loads in cats preinfected with FIV-P. The inset shows a magnified plot of the results at the last two sampling points. (D) FIV-P-specific proviral loads in cats from previous experiments that were preinfected with FIV-P. The shaded area shows the range of proviral loads found in cats singly infected with FIV-P at times ranging between 7 and 36 months postinfection. Results are expressed as numbers of proviruses per microgram of PBMC DNA. Cats were preinfected with 300 (squares), 30 (circles), 3 (stars), or 0.3 (inverted triangles) CID50. Asterisks indicate statistically significant differences between the mean values for preinfected and control cats (*, P <0.05; **, P <0.01).

(iii) Virus isolation from PBMC. At selected times p.c., the infectious virus recoverable from the PBMC was isolated in cocultures with MBM cells and then characterized as FIV-P or FIV-M2 by F-RFLP (Table 2). The cocultures performed at 2 weeks p.c. still yielded FIV-P alone, but at 1 month p.c., FIV-M2 was already isolated from both control cats and from three or four cats preinfected with 30 or 300 CID₅₀ of FIV-P. At subsequent samplings, FIV-M2 isolation was the rule and was often heralded by a reduction in the time needed for the cultures to become positive, whereas FIV-P was isolated much less frequently. This pattern, however, changed at the last two sampling points (30 and 36 months p.c.), when FIV-P was again recovered from all the cats injected with this virus and was the only virus detected in three samples. Thus, although the picture was less clear-cut, by and large virus isolation confirmed what was revealed by the proviral content of PBMC, i.e., opposite kinetics of FIV-P and FIV-M2 replications and a tendency of the former virus to reemerge late in the follow-up, after a period of relative undetectability. It is also worth noting that the cats that had been inoculated with 3 CID₅₀ of FIV-P were the only ones to yield cultures negative for virus at 36 months p.c., indicative of much-reduced virus expression at this time.

(iv) Plasma viremia. In general, levels of total viral RNA in plasma paralleled total provirus loads in the PBMC, albeit loosely. In the controls, viral genomes peaked at 28,340 and 42,450 RNA copies per ml of plasma at 1 month p.c., then decreased to reach approximately 2,000 copies at 12 months p.c., and eventually increased to values that at the end of the follow-up approximated those of the acute phase p.c. (Fig. 2A). Throughout the first 18 months p.c., FIV-P-preinfected cats showed viremia patterns similar to those of the controls and steady-state levels of 5,000 copies on average. At later times, however, these cats did not show the marked increase in plasma viremia that was seen in the controls. As a consequence, at 30 and 36 months p.c., the levels of viremia detected in preinfected cats were significantly lower than those in control cats (Fig. 2B).

View larger version (24K): [in this window] [in a new window]   FIG. 2.   Plasma viremia in cats challenged with FIV-M2, as determined by RT-cPCR and F-RFLP analyses of RT-nested PCR gag amplicons. (A) Total numbers of genomes in control cats. (B) Total numbers of genomes in cats preinfected with FIV-P. (C) FIV-M2 genomes in cats preinfected with FIV-P. The inset shows a magnified plot of the results at the last two sampling points. (D) FIV-P genomes in cats preinfected with FIV-P. The shaded area shows the range of viral genomes found in cats from previous experiments that were singly infected with FIV-P at times ranging between 7 and 36 months postinfection. Results are expressed as numbers of RNA genomes per milliliter of plasma. Symbols and asterisks are as explained for Fig. 1.

(v) Proviral loads in solid tissues at the end of the experiment. Quantification of viral DNA in tissues obtained at necropsy confirmed previous findings that FIV is disseminated throughout the bodies of infected cats and is generally more abundant in lymphoid than in nonlymphoid tissues (1, 5, 36). As shown in Fig. 3, the quantitative differences in total proviral loads of preinfected and control cats were much less evident in solid tissues than in the PBMC but reached statistical significance in several tissues. However, the relative proportions of FIV-P and FIV-M2 DNA were similar to those detected in contemporaneous PBMC samples. In particular, consistent with observations in PBMC, no FIV-M2 DNA was detected in the solid tissues of the two cats preinfected with 3 CID₅₀ of FIV-P. Although throughout the study virus characterization had been conducted by examining gag-derived amplicons alone, in these cats F-RFLP analysis was extended to amplicons from the env gene and the results were entirely substantiated (Fig. 4 and data not shown).

View larger version (21K): [in this window] [in a new window]   FIG. 3.   Proviral loads in solid tissues at the end of follow-up, 36 months after challenge with FIV-M2. Dashed lines indicate the lower limit of sensitivity of the cPCR method used for quantitation (~100 copies). DNA samples that were found provirus negative by cPCR but positive by nested PCR (lower limit of sensitivity, ~10 copies) are indicated by bars that do not reach the dashed line. The mean proviral loads of preinfected cats were significantly reduced relative to those for control cats in the bone marrow and frontal cortex (P < 0.05) and in the mesenteric lymph nodes, spleen, thymus, mesencephalon, and kidney (P < 0.01). ND, not determined.

View larger version (30K): [in this window] [in a new window]   FIG. 4.   Characterization in env of FIV-P and FIV-M2 proviruses present in solid tissues at the end of follow-up, 36 months after challenge with FIV-2. (A) Strategy. A 334-bp amplified fragment from the env gene was digested with the restriction nucleases AvaII and HinfI and analyzed by electrophoresis on an agarose gel and on an ALF DNA sequencer; the presence of unique restriction sites allows for the identification of the two viral strains. Numbers represent fragment sizes (in base pairs). (B) Representative results are shown for DNA from FL4 cells persistently infected with FIV-P (FIV-P/FL4 cells), DNA from MBM cells infected with FIV-M2 (FIV-M2/MBM cells), and DNA from mesenteric lymph nodes of cats preinfected with 0, 3, or 30 CID50 of FIV-P and challenged with FIV-M2 (four rightmost lanes). Numbers to the left of the DNA ladder represent fragment sizes (in base pairs).

(vi) Antibody responses. In control cats, FIV-M2 challenge induced anti-FIV ELISA antibodies which after 2 months were still relatively low in titer, although in subsequent samplings they increased steadily. The antibody titers found in superinfected animals at the same time were considerably higher, suggestive of an anamnestic response. Interestingly, this was true also for the cats preinfected with 0.3 CID₅₀ of FIV-P despite the absence of detectable antibodies at challenge, suggesting that they too had been immunologically primed by the first FIV exposure. Antibody measurements performed at later times were not very informative (Fig. 5). NA titers were also markedly increased by challenge, with no differences between preinfected and control animals (Table 3 and data not shown).

View larger version (19K): [in this window] [in a new window]   FIG. 5.   Effect of FIV-M2 challenge on ELISA antibody titers to whole FIV antigen in control cats (A) and cats preinfected with FIV-P (B). The titers are expressed as the reciprocal of the highest serum dilution that gave optical density readings at least threefold higher than the values obtained with 10 control FIV-negative serum samples. Symbols are as explained for Fig. 1.

(vii) Circulating CD4⁺ T-lymphocyte counts. In the controls, FIV-M2 challenge produced a rapid loss of circulating CD4⁺ T lymphocytes; after 2 months, these had approximately halved in number (Fig. 6A). Then there was a return to normal counts, which lasted up to 12 months p.c., when they started again to gradually decrease. Thus, in the control cats, the kinetics of CD4⁺ T-cell fluctuations were almost the opposite of those of viral loads (Fig. 1A and 2A), which might suggest a direct role of FIV-M2 replication in CD4⁺ T-lymphocyte depletion. As already mentioned, at challenge FIV-P-preinfected cats had CD4⁺ T-cell counts within normal values, except for the cat which died 5 months later. The effects of FIV-M2 on the CD4⁺ T-cell counts of these cats were similar to those in the controls. There were, however, three exceptions. One was again the cat which succumbed, which 1 month before death had a very low number of CD4⁺ T cells despite the short interval after challenge. The other two exceptions were the cats preinfected with 3 CID₅₀ of FIV-P; their CD4⁺ T-cell counts showed an initial decrease similar to that in the other animals but showed no further reductions at later times (Fig. 6B).

View larger version (25K): [in this window] [in a new window]   FIG. 6.   Effects of FIV-M2 challenge on circulating CD4+ T lymphocytes of control cats (A) and of cats preinfected with FIV-P (B). Symbols are as explained for Fig. 1.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

This study examined the effects of infection with a relatively attenuated FIV (FIV-P, subtype A) on subsequent challenge of cats with a fully pathogenic virus of a different subtype (FIV-M2, subtype B). During the early months p.c., the results were consistent with those of previously reported short-term superinfection experiments with FIV (31, 32, 42) in that establishment of the highly heterologous challenge virus was not even marginally inhibited. Our study would thus have provided little valuable new information had the cats not been monitored well beyond the acute phase p.c. In fact, significant differences indicating that preinfected cats controlled challenge infection better than naive animals started to become evident only after 2 years p.c.

In the 7 months between FIV-P and FIV-M2 inoculation, the cats displayed virological and immunological markers that varied with the dose of FIV-P received but were essentially in accord with the premises of the experiment. In general, the animals given 3 to 300 CID₅₀ of FIV-P exhibited markers compatible with infection by an attenuated virus, including low provirus loads in the PBMC, moderate titers of antiviral antibody, and normal CD4⁺ T-cell counts. On the other hand, of the two cats inoculated with 0.3 CID₅₀, one tested negative by all criteria and the other exhibited low numbers of provirus in the PBMC but remained seronegative and virus reisolation negative; these results resembled the observations of Barlough et al. (4) with a zidovudine-resistant mutant of FIV-P and of Sparger et al. (61) with molecularly cloned FIV-P. After FIV-M2 challenge, both these animals were found to harbor FIV-P as well as FIV-M2. It is therefore likely that they harbored very few copies of replication-competent FIV-P or had become carriers of defective FIV-P which, upon challenge, was activated (47) or complemented by the superinfecting virus.

In the early months following FIV-M2 challenge, all parameters were indicative of a much more active course of infection in both preinfected and naive animals than that observed in the FIV-P-infected cats prior to FIV-M2 challenge. The parameters monitored measured FIV replication directly as well as FIV-specific antibody responses and circulating CD4⁺ T-cell levels. Since virulent viruses in general replicate faster and at higher titers than avirulent viruses, the findings were already strong evidence that in the preinfected animals the challenge virus was actively replicating and producing pathological effects. Formal proof of this was obtained by quantifying the contributions of the two viruses to total proviral burden in PBMC and plasma viremia by means of a previously developed F-RFLP (12). Interestingly, during the early months p.c., virtually all of the peripheral virus burden in dually infected cats was of FIV-M2 origin, and the concentrations of FIV-P in the bloodstream dropped rapidly relative to prechallenge values. Although we lacked a contemporary cohort of cats infected with FIV-P alone, in historical controls we had never seen FIV-P proviral burdens as low as in the FIV-M2-superinfected cats during this early phase (Fig. 1D) or fluctuations of proviral burdens of similar magnitude in a comparatively short time (data not shown). Thus, we believe that the absence of an FIV-P-only control group does not confound the interpretation of results. This diminution of FIV-P was not investigated further, but possible explanations include accelerated killing of FIV-P-infected cells by the superinfecting FIV-M2 and/or occupation of FIV target cells (18). In cultured lymphoid cells FIV-M2 replicates more rapidly and is more cytopathic than the stock of FIV-P used for preinfection (data not shown); thus, it seems feasible that the latter virus was at a replicative disadvantage in dually infected hosts. Alternatively, it is possible that an anamnestic immune response that was mainly directed against FIV-P was induced by FIV-M2 infection due to "original antigenic sin" phenomena (22, 24, 30) and that it led to a preferential, rapid clearance of FIV-P and FIV-P-infected cells. In any case, both in the early acute phase p.c. and in the postacute stage, characterized by stable levels of peripheral viral loads, none of the FIV-P-preinfected cats showed indications of even partially increased resistance to FIV-M2, and there were actually suggestions that in the cats preinfected with large doses of FIV-P, early replication of the challenge virus was facilitated.

As mentioned, important differences in the course of infection between preinfected and control cats started to emerge 2 years p.c. From this sampling onwards, PBMC proviral loads and plasma viremia underwent a pronounced progressive increase in the control cats, so that at sacrifice, 3 years p.c., their values were severalfold higher than during the postacute steady-state period. In contrast, in all the FIV-P-preinfected cats, these markers remained essentially stable, thus indicating that preinfection, although incapable of blocking establishment and acute-phase replication of the challenge virus had reduced the rate of infection progression at later times. This conclusion was corroborated by examination of the proviral content of solid tissues at the end of the experiment, albeit the differences between preinfected and control cats in these specimens were less marked than in the bloodstream.

The mechanisms by which the delayed protective effect exerted by FIV-P preinfection was achieved are not clear. At present, it is not clear whether superinfection resistance in lentiviruses is an immunological effect (56) and if so, to what antigens (19). As discussed in the introduction, FIV-P and FIV-M2 share neutralization epitopes demonstrable in certain cell substrates, and it is conceivable that cross-protective immune effectors elicited by these antigens are slow to develop or to enter into action. Alternatively, it is possible that with time the less-virulent FIV-P had resumed a leading role in infection and dictated its course. In fact, it might not be a mere coincidence that, also starting at 2 years p.c., the contribution of FIV-P to total virus burden, which soon after FIV-M2 challenge had become minimal, grew considerably and in some cats apparently eclipsed that of FIV-M2. Another possibility is that recombinants formed between the two viral strains (2, 31, 32) and, due to mixed phenotypic traits, somehow contributed to the observed effects. Recombination between different genotypes of HIV-1 has been clearly documented in coinfected humans (11, 54) and chimpanzees (21). Although we did not investigate the occurrence of such recombinants in our dually infected cats, the results of F-RFLP analysis performed on gag and env sequences of viral DNA from necropsy samples coincided, suggesting that recombinants were not a major component of total virus burden in such animals.

How FIV-P managed to eventually become predominant in dually infected cats is unclear. If in vivo replication of FIV-P is more compatible with cell survival than that of FIV-M2 (because it is less cytopathic or simply replicates less efficiently), FIV-P-preinfected cells could survive longer than FIV-M2-infected cells, develop a state of relative resistance or interference to superinfection (62, 66), and slowly became more abundant. Okada and colleagues (42) described the superinfection in vitro of cells chronically infected with FIV-P by a heterologous virus, but the situation in the intact host might be very different. A similar mechanism has been proposed to explain the disappearance of highly cytopathic, syncytium-inducing variants of HIV in favor of non-syncytium-inducing variants (55), and Bonhoeffer and Nowak (10) developed a mathematical model in which a slowly replicating attenuated HIV-1, even if inoculated into already infected individuals, could successfully compete with and progressively replace a more pathogenic fast-replicating strain.

In summary, this study has shown that, in agreement with observations in previous short-term experiments using wild-type viruses (31, 32, 42), preinfection with a partially attenuated strain of FIV did not protect cats from acute infection with a second, highly heterogeneous strain. However, from 2 years p.c. until 3 years p.c., when the experiment was interrupted, preinfected cats exhibited reduced total viral burdens, and some also exhibited a diminished decline of circulating CD4⁺ T cells. Thus, these results are in agreement with those of studies in which attenuated SIV strains were found to confer different degrees of protection against heterologous challenges (14, 50, 57, 58). Interestingly, in our study the challenge virus, which soon after inoculation had largely dominated the scene, at later times, in concomitance with reduced total viral burdens, was almost completely replaced by the preinfecting attenuated virus. Further experiments with larger groups of animals will be needed to investigate whether this event and protection were linked, what mechanisms are involved, and whether the observed late containment of virus replication would be sufficient to prevent disease. In lentiviral infections, reduced viral loads have often been seen to correlate with mild or postponed disease development (40, 41, 57). Collectively, these findings give impulse to the development of properly attenuated anti-FIV vaccines. Their use in the field might provide answers to at least some of the safety concerns raised by proposals to use live attenuated vaccines for AIDS prophylaxis in humans.