VOL. 70, 1996 EIAV GENOMIC QUASISPECIES 3347 EIAVPV was derived from the tissue culture-adapted strain of EIAV (19) by three serial back-passages in Shetland ponies followed by biological cloning of neutralization escape mutants that were selected in the presence of autologous neutralizing serum (27, 37).
Although the EIAVPV stock has been widely used in experimental infections, the genetic heterogeneity of this biological clone has never been determined.
Moreover, there is no information on the EIAVPV quasispecies that establish the infection in ponies and cause the initial acute episode of disease.
Because of the rapid variation that occurs during EIAV infection, it was anticipated that genetic characterization of EIAVPV during the initial stages of infection and disease could direct the production of a genetically deﬁned pathogenic molecular clone of EIAV, a goal that has thus far eluded researchers who have described only avirulent infectious molecular clones.
To examine the diversity present in the EIAVPV stock and to investigate which viral sequences are associated with the ﬁrst disease episode in Shetland ponies infected with EIAVPV, the variable region of gp90 and the U3 region of the LTR (U3-LTR) were sequenced.
In addition, the predominant LTR and gp90 variable region sequences were used to create chimeric recombinant viruses in an attempt to deﬁne segments of the EIAV genome that may be involved in pathogenesis. MATERIALS AND METHODS Cells.
Equine mononuclear cells were enriched from whole blood as described previously (18) and were seeded into 24-well plates at 4 ϫ 106 cells per well in modiﬁed Eagle’s medium alpha medium (GibcoBRL, Grand Island, N.Y.) supplemented with 10% heat-inactivated horse serum (Sigma, St.
Louis, Mo.), penicillin G (100 U/ml), streptomycin (100 g/ml), and glutamine (2 mM).
Monocytes were adhered to plastic for 4 h at 37ЊC and washed twice with Hanks’ balanced salt solution (GibcoBRL) to remove nonadherent cells.
The medium containing the nonadherent cells was clariﬁed by centrifugation at 3,500 ϫ g and then placed back on the monocytes.
Fetal equine kidney (FEK) cells were grown at 37ЊC in minimum essential medium (GibcoBRL) supplemented with 10% heat-inactivated fetal bovine serum (Atlanta Biologicals, Atlanta, Ga.), nonessential amino acids (GibcoBRL), and penicillin, streptomycin, and glutamine as described above.
The infectious, nonpathogenic molecular clone pEIAV19-2 (31; GenBank accession number U01866) was used as a backbone for the generation of chimeric proviral clones by restriction fragment exchanges.
Since the predominant LTR sequence of EIAVPV was identical to the EIAV65 LTR (31), the latter was used for LTR restriction fragment exchanges.
The 3Ј end of the gp45 gene and the entire 3Ј LTR of EIAV19-2 (contained on a BstXI-EcoRI restriction fragment) was replaced with the corresponding fragment of pEIAV65 to generate the proviral clone pEIAVPV100.
A second chimeric proviral clone that contained both the predominant gp90 and LTR sequences of EIAVPV was created.
A 736-bp HindIII-HindIII restriction fragment from clone EIAVPV-A2, which was identical to the EIAVPV consensus sequence and encoded all but the ﬁrst 24 amino acids of the region that we sequenced, was used to replace the corresponding region of pEIAVPV100, such that the proviral clone pEIAVPV101 was created.
Viruses and infections.
Virus stocks were prepared by harvesting the medium from calcium phosphate-transfected FEK cells (39).
The stocks were assayed by using a micro-reverse transcriptase assay (18) and titered for infectivity with an infectious center assay based on an enzyme-linked immunosorbent assay (11).
Virus was absorbed to equine monocytes for 4 h at 37ЊC, after which the virus inoculum was washed off and fresh medium was added to the cells.
Medium was replaced daily, and the clariﬁed supernatants were assayed for reverse transcriptase activity as described previously (18).
Seronegative Shetland ponies were inoculated intravenously with 107.0 infectious center doses of EIAVPV100 or EIAVPV101.
Rectal temperatures were monitored daily, and serum and plasma samples were taken periodically.
Derivation of EIAVPV and plasma samples.
Serial back-passage of the celladapted strain of Wyoming in Shetland ponies was described previously (27).
Plasma from a second back-passage pony (pony 82) was passaged a third time in pony F135 (37).
To generate a neutralization escape mutant, virus isolated from pony F135 at 16 days postinfection (during the ﬁrst disease episode) was serially passaged 13 times in FEK cells in the presence of autologous neutralizing serum isolated 203 days postinfection (37).
The resulting neutralization-resistant virus was biologically cloned twice in FEK cells by endpoint dilution and ampliﬁed one time in FEK cells to generate the virus stock termed EIAVPV.
The titer of EIAVPV is 106.5 horse infectious doses.
Two seronegative Shetland ponies were intravenously infected with 300 horse infectious doses of EIAVPV.
Plasma taken from each pony during the ﬁrst disease episode (day 19 for pony 06 and day 33 for pony 11) was used in this study.
The titer of pony 06 plasma taken on day 19 is 105.0 50% tissue culture infectious doses.
The titer of pony 11 plasma taken on day 33 is unavailable.
However, a quantitative reverse transcription (RT)-PCR assay (18) showed that the concentration of viral RNA in the plasma sample was approximately 3 ϫ 107.0 molecules per ml of plasma.
RNA puriﬁcation and RT-PCR.
Viral RNA was puriﬁed and RT-PCR was performed as described previously (18).
Brieﬂy, concentrated virus, pelleted from plasma or tissue culture supernatant at 120,000 ϫ g, was extracted with the Trizol reagent (GibcoBRL) to purify the viral RNA.
Reverse transcription of 2 l of viral RNA (equivalent to 200 l of EIAVPV stock or pony plasma) was performed with the Superscript Preampliﬁcation System (GibcoBRL) as speciﬁed by the manufacturer except that ﬁrst-strand synthesis was initiated with an EIAV-speciﬁc primer (0.4 M).
CDNA synthesis for ampliﬁcation of the gp90 variable region or the U3-LTR was performed with the Var2 primer (5Ј GAG CAGTTATATTGGTTAAAGCTTTGG 3Ј) or the LTR6 primer (5Ј AGGC CTTTTCAGCCCAGCAGA 3Ј), respectively.
PCRs were carried out by using a hot-start procedure with a mixture that contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, 0.001% gelatin, 0.05 mM each deoxynucleoside triphosphate, 0.3 M each primer, 1 U of AmpliTaq DNA polymerase (PerkinElmer, Norwalk, Conn.), and 2 l of cDNA in a ﬁnal volume of 25 or 100 l.
Primers used for ampliﬁcation of the variable region of gp90 were Var1 (5Ј GTTCCTTCCCGGGGTGTAGACC 3Ј) and Var2.
Primers used for ampliﬁcation of the U3-LTR were LTR5 (5Ј CCCCTCATAAAAACCCCACA 3Ј) and LTR6.
PCR was carried out under the following cycling conditions: 1 min at 94ЊC, 1 min at 55ЊC, and 1 min at 72ЊC for 35 cycles; 10 min at 72ЊC for one cycle; and hold at 4ЊC.
Cloning of RT-PCR products.
Except as noted below, a single RT reaction provided cDNA for multiple independent PCRs.
PCR products were pooled prior to puriﬁcation (Wizard PCR Preps DNA Puriﬁcation System; Promega, Madison, Wis.) and then T-A cloned into pGEM5Zf(ϩ) (Promega).
The EIAVPV env clones were generated in two separate experiments, each of which consisted of 12 independent PCRs from a single RT reaction.
Similar results were obtained after analysis of 8 or 12 clones from the ﬁrst or second set of reactions, respectively.
Five independent PCRs were used to generate the EIAVPV LTR clones.
The pony 06 env clones were generated from 10 independent PCRs.
The pony 11 env clones, the pony 06 LTR clones, and 12 of the pony 11 LTR clones were derived from a single RT-PCR.
An additional 8 pony 11 LTR clones were generated from 8 independent PCRs (that were not pooled) from a single RT reaction.
Analysis of the latter pony 11 clones conﬁrmed the results obtained with the former group of pony 11 clones.
White colonies were screened for the proper size insert by restriction enzyme analysis without prior screening or following hybridization of the appropriate 32P-labeled probe to a colony blot (39).
Sequencing of RT-PCR clones.
For sequencing, double-stranded plasmid DNA and 1.0 l of the appropriate primer (0.5 pmol/l) were denatured together in a total volume of 10.0 l in the presence of dimethyl sulfoxide (10%) and NaOH (0.1 M) for 10 min at 68ЊC.
Samples were neutralized with 0.4 volume of neutralizing buffer [0.28 M N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid (TES), 0.08 M MgCl2, 0.2 M NaCl (pH 1.1)] and then kept at room temperature for 10 min.
Samples were sequenced by the chain termination method (40, 41) with a Sequenase version 2.0 sequencing kit (United States Biochemical, Cleveland, Ohio) and [␣-thio-35S]dATP (1,000 to 1,500 Ci/mmol; DuPont NEN, Boston, Mass.).
Both strands of each gp90 insert were sequenced with the following primers: M13 (Ϫ40) forward primer (United States Biochemical), M13 (Ϫ24) reverse primer (New England Biolabs, Beverly, Mass.), Var3 (5Ј GTCTTCTTGCACAGATAG 3Ј), Var4 (5Ј GTGTGCCTGTCCTATAAC 3Ј), Var5 (5Ј GGGATACATCCAATCAGGC 3Ј), Var6 (5Ј ATCTTCTAAAAC CCCAAG 3Ј), Var7 (5Ј CACTAACATAACTTCCTGC 3Ј), and Var8 (5Ј TTCAGGTACTAATATAGT 3Ј).
For the LTR clones, the M13 forward and reverse primers were used such that both strands of the insert were sequenced.
All PCR and sequencing primers were designed according to the published sequence (35).
The Genetics Computer Group (Madison, Wis.) package of sequence analysis software was used for sequence analysis (10).
To estimate the Taq DNA polymerase error rate, 10 fg of a previously sequenced gp90 variable region clone was reampliﬁed by PCR, and the products were cloned and sequenced as described above.
Six clones were sequenced, and the error rate was determined to be 0.06% (three substitutions per 4,842 bp sequenced).
Nucleotide sequence accession numbers.
The sequences analyzed were submitted to GenBank and have been assigned accession numbers U35178 through U35224 and U43948 through U43950. Downloaded from http://jvi.asm.org/ on May 15, 2013 by guest RESULTS To examine EIAV genomic quasispecies that are associated with the initiation of infection and disease, we compared viral RNA sequences present in a pathogenic stock of virus termed EIAVPV with the sequences circulating in the plasma during the initial disease episode in two ponies infected with EIAVPV. 3348 LICHTENSTEIN ET AL. J.
VIROL. Downloaded from http://jvi.asm.org/ on May 15, 2013 by guest FIG. 1.
U3-LTR sequences of EIAVPV, pony 06, and pony 11.
Sequences were aligned with the program Pileup from the Genetics Computer Group sequence analysis package (10).
Dashes indicate a nucleotide identical to the EIAVPV consensus sequence (EIAVPV Con).
For each sample, only unique clones are shown, and the number of clones with a given sequence is shown to the right.
The nucleotide numbering refers to the position of each sequence with respect to the EIAVPr proviral sequence. Two portions of the viral genome reported to be variable were sequenced: (i) a 219-bp segment of the U3-LTR (6) and (ii) an 807-bp segment of the env gene encoding the variable region of the SU glycoprotein gp90 (30).
The portion of the LTR that was sequenced contained nearly the entire U3 region, including the hypervariable region (6).
The segment of the env gene encoding gp90 that was sequenced contained the variable region ﬂanked by portions of two conserved regions.
Located within the gp90 variable region are the hypervariable region and the PND, the latter of which is composed of two adjacent neutralizing epitopes (DNT and ENT) that were deﬁned by monoclonal antibodies (see Fig. 3 and references 3 and 14).
The EIAVPV stock was derived as described in Materials and Methods.
Pony 06 and pony 11 were each infected with EIAVPV, and viral RNA was isolated from plasma during the ﬁrst clinical episode in each pony (day 19 for pony 06 and day 33 for pony 11).
We sequenced viral genomic RNA isolated directly from plasma, as opposed to sequencing proviral DNA, since previous results with EIAV (38) and recent results with human immunodeﬁciency virus (HIV) (13, 44) suggest that lentivirus infections are quite dynamic.
Thus, examining the virus present in plasma is the best way to capture the currently replicating viral species.
Genomic viral RNA was subjected to RT-PCR, then the ampliﬁed products were cloned, and multiple independent clones were sequenced.
Sequence analysis of the U3 region of the LTR.
The sequences of the LTR clones from the EIAVPV stock and the pony-derived samples (12 to 20 clones from each sample) were nearly all identical (Fig. 1).
The consensus sequences from all three samples were identical, and clones which varied from the consensus sequence did so by only one or two nucleotides.
The level of variation observed for the LTR sequences (Table 1) was only slightly greater than the calculated Taq DNA polymerase error rate (see Materials and Methods).
Sequence comparison of the previously identiﬁed hypervariable region of the U3-LTR was used to determine the relationship of the EIAVPV, pony 06, and pony 11 LTRs to those of pathogenic and nonpathogenic strains of EIAV (Fig. 2).
The hypervariable region of the EIAVPV LTR and that of the pony-derived LTR had signiﬁcant homology with two virulent strains of EIAV (the consensus Wyoming LTR [Fig. 2] and the Th-1 LTR [data not shown]) and were more divergent from the LTRs of the avirulent isolates EIAV1369 and EIAV19-2.
Inter- TABLE 1.
Mean percentage variation in gp90 and the U3-LTR Mean % variation Sample Nucleotide gp90 LTR, nucleotide Amino acid PV stock Pony 06 Pony 11 0.20 (33/16,140a) 0.40 (36/9,684) 0.30 (29/9,684) 0.48 (26/5,380b) 0.80 (27/3,228) 0.77 (25/3,228) 0.11 (3/2,628a) 0.08 (2/2,628) 0.14 (6/4,380) a Total number of nucleotide substitutions, with respect to each sample’s consensus sequence, divided by the total number of nucleotides sequenced.
B Total number of deduced amino acid substitutions, with respect to each sample’s consensus sequence, divided by the total number of amino acids. VOL. 70, 1996 EIAV GENOMIC QUASISPECIES 3349 FIG. 2.comparison of the U3-LTR hypervariable region of various virus isolates.
A portion of the EIAVPV LTR consensus sequence was aligned with the corresponding region of other known LTR sequences, using the program Pileup (10).
The pony-derived consensus sequences are not shown since they were identical to that of EIAVPV.
Potential cis-acting sequences (PEA-2, CAAT, ets, AP-1, and TATA) are underlined.
A lowercase letter indicates a nucleotide position for which at least one of the sequences differs from the others.
Gaps (indicated by dots) were introduced to align the sequences. Downloaded from http://jvi.asm.org/ on May 15, 2013 by guest estingly, the EIAVPV and pony-derived LTRs lacked a second CAAT box motif and a third core ets binding motif that are present in the Wyoming and Th-1 LTRs (6, 21, 34).
It has been suggested that two CAAT boxes and core ets binding motifs may be important in viral pathogenesis and replication in macrophages, respectively (6, 21).
Sequence analysis of the SU protein gp90.
Limited sequence variation was observed within the portion of the env gene that was ampliﬁed from the EIAVPV stock (20 clones) and from the plasma virus samples isolated from pony 06 (12 clones) and pony 11 (12 clones) (Fig. 3).
Sequence variation among the latter two samples was slightly higher than among the EIAVPV clones, indicating that the EIAVPV stock underwent variation in vivo (Table 1).
The level of variation observed for all three samples was signiﬁcantly higher (between three- and sevenfold) than the calculated Taq error rate (see Materials and Methods), indicating that the majority of nucleotide changes were not the result of the Taq DNA polymerase.
Among all three samples, the large majority of sequence changes were nucleotide substitutions.
Of the 44 env clones examined, only three contained a 1-bp deletion resulting in a defective env gene (one clone from EIAVPV and two clones from pony 11).
In addition, one clone from pony 06 contained an in-frame 3-bp deletion.
There was no evidence for G-to-A hypermutation (data not shown) as described previously for EIAV and other lentiviruses (5, 9, 15, 34, 42, 43).
Nonsynonymous changes within gp90 were disproportionately located in the PND and/or the hypervariable region (HVR).
Although the PND comprised 7% of the total length of the region sequenced, this segment accounted for 38% (10 of 26), 22% (6 of 27), and 48% (12 of 25) of the nonsynonymous changes that were found in EIAVPV, pony 06, and pony 11, respectively (Fig. 3).
In addition, 41% (11 of 27) of the nonsynonymous changes found in pony 06 were located in the HVR, a segment that constituted only 13% of the total length of the region sequenced (Fig. 3).
These data suggested that the pattern of mutations was not random.
The position and type of Taq-induced errors (one nonsynonymous change in each of the two conserved regions and one synonymous change in the PND) in the control experiment (see Materials and Methods) make it unlikely that the observed pattern of mutations in our samples was due to the presence of a hot spot for Taq-induced misincorporations.
The rate of ﬁxation of mutations was determined from the formula R ϭ D/2T, where R is the number of nucleotide substitutions per site per year, D is the mean pairwise nucleotide distance between EIAVPV and the pony-derived samples (data not shown), and T is the length of time after infection that the samples were isolated (pony 06 ϭ 0.05 year and pony 11 ϭ 0.09 year).
The mutation rates were calculated to be 5.6 ϫ 10Ϫ2 and 3.2 ϫ 10Ϫ2 mutations per site per year for pony 06 and pony 11, respectively.
The EIAVPV and pony 06 consensus sequences were identical to each other, while that of pony 11 differed from them by only two amino acids, both of which were located within the PND (Fig. 4).comparison of all three consensus sequences with other known gp90 amino acid sequences revealed significant differences, particularly within the PND and the HVR (Fig. 4).
All sequences examined were only distantly related to the Wyoming sequence, diverging signiﬁcantly in the PND and HVR.
Our gp90 sequences showed greater homology to the prototype (EIAVPr) strain of virus, from which EIAVPV was originally derived (see Materials and Methods), although sequence divergence was still signiﬁcant in the PND and HVR.
The EIAVPV and pony-derived consensus sequences differed from the EIAVPr and EIAV19-2 sequences by seven or eight amino acids (approximately 3%).
In addition, all 44 env clones contained two silent nucleotide changes with respect to the EIAVPr and EIAV19-2 sequences (data not shown).
Five of the amino acid changes, four within the HVR and one within the PND, resulted in additional potential N-glycosylation sites within our consensus sequences compared with the EIAVPr and EIAV19-2 sequences (Fig. 4; note that 6 of the 44 clones contained an additional mutation which resulted in loss of the additional potential N-glycosylation site within the PND [see Fig. 3]).
These results are not surprising since differences in the tryptic glycopeptide maps among various EIAV isolates have been noted previously (33, 38), and sequence analysis of EIAV and other lentiviruses has yielded similar changes in potential N-glycosylation sites (4, 5, 22, 28, 30).
The addition of a potential N-glycosylation site within the PND of EIAVPV could account, in part, for its altered neutralization properties as described previously (37).
Infection with chimeric viruses.
Chimeric viruses were constructed to examine the pathogenic potential of the predominant LTR and gp90 variable region sequences present in the EIAVPV stock.
The predominant sequences present in the pathogenic EIAVPV stock might be associated with the generation of disease, since these sequences were very similar to those present during the ﬁrst disease episode in two EIAVPVinfected ponies and differed signiﬁcantly from the corresponding sequences of the avirulent strains EIAVPr and EIAV19-2.
Two chimeric proviral clones were constructed such that the only known difference between the parental and recombinant viruses was the U3-LTR (clone pEIAVPV100) or this region plus the gp90 variable region (clone pEIAVPV101).
The aviru- Downloaded from http://jvi.asm.org/ on May 15, 2013 by guest 3350 LICHTENSTEIN ET AL. J.
VIROL. VOL. 70, 1996 EIAV GENOMIC QUASISPECIES 3351 Downloaded from http://jvi.asm.org/ on May 15, 2013 by guest FIG. 3.
Deduced amino acid sequences of the EIAV gp90 variable and ﬂanking regions of EIAVPV, pony 06, and pony 11.
Sequences were aligned as described in the legend to Fig. 1.
The single-letter amino acid code shows the consensus sequence above each group of clones.
Dashes indicate an amino acid identical to the EIAVPV consensus sequence (EIAVPV Con).
An asterisk indicates a silent nucleotide substitution.
An X indicates the position of a single nucleotide deletion that shifts the sequence out of the proper reading frame.
A gap (indicated by a dot) was introduced into clone P06-18 to align the sequences.
The variable and hypervariable regions are indicated by a single and double underlines, respectively.
The boxed area shows the PND, which contains two adjacent neutralizing epitopes, ENT and DNT.
The numbering refers to the position of each sequence with respect to the ﬁrst amino acid of gp90 (35).
Only unique clones are shown, and the number of clones with a given amino acid sequence is shown to the right.
The silent mutations shown at positions 150 and 317 of clone EIAVPV-I4 were present only in one of the two clones with that amino acid sequence.
The silent mutation shown at position 342 of clone EIAVPV-A2 was present only in one of the four clones with that amino acid sequence.
Two nucleotide substitutions were responsible for the arginine change at position 200 of clone P11-9. lent, infectious molecular clone EIAV19-2 was used as the backbone for restriction fragment exchanges (see Fig. 2 and 4 for the EIAV19-2 LTR and gp90 sequences, respectively).
Sequences that were identical to the consensus EIAVPV LTR and gp90 variable region sequences were used to generate the chimeric viruses (see Fig. 1 and 3 for the EIAVPV LTR and gp90 consensus sequences, respectively).
Virus stocks were generated by transfection of FEK cells with the infectious molecular proviral clones.
Cycle sequencing of viral RNA conﬁrmed the presence of the expected LTR and gp90 variable region sequences in the recombinant viruses (data not shown).comparison of the kinetics of virus replication in equine monocyte cultures showed that the parental and recombinant viruses replicated to the same level and with the same kinetics (Fig. 5).
Shetland ponies (one for each chimera) were infected with EIAVPV100 or EIAVPV101.
Rectal temperature and platelet counts were monitored as indicators of disease.
Although both ponies seroconverted by 21 days postinfection, which indicated that they were infected, no clinical symptoms have been noted up to 120 days postinfection.
These data suggest that the determinants of pathogenesis may not be associated with the predominant variable region of the env gene and the LTR of EIAVPV. DISCUSSION Analysis of EIAV sequence diversity present in a pathogenic stock of virus (EIAVPV) and during the initial disease episode in two ponies infected with EIAVPV was carried out to investigate the correlation between disease and the presence of particular genomic sequences.
Sequence diversity in the env gene of the EIAVPV stock was low and was only slightly higher in the ponies infected with EIAVPV.
While the former result is likely due to the fact that EIAVPV is a biological clone (37), the latter ﬁnding was somewhat surprising since previous studies have demonstrated rapid variations in the antigenic (14, 33, 38) and pathogenic (27) properties of EIAV during persistent infections in Shetland ponies.
In fact, the predominant virus isolates from successive disease episodes separated by as few as 14 days have been shown to contain different gp90 and gp45 peptide maps (27, 33).
The level of EIAV env variation that we determined was similar to that seen early after experimental infection of monkeys with molecular clones of simian immunodeﬁciency virus (2, 28).
Our estimate of the rate of mutation (3.1 ϫ 10Ϫ2 to 5.4 ϫ 10Ϫ2 mutations per site per year) was in accordance with the rates previously estimated for EIAV, HIV, and simian immunodeﬁciency virus (5, 12, 15, 30).
Thus, 3352 LICHTENSTEIN ET AL. J.
VIROL. Downloaded from http://jvi.asm.org/ on May 15, 2013 by guest FIG. 4.
Sequence alignment of the EIAV gp90 variable and ﬂanking regions of various isolates.
The consensus sequences (Con) determined in Fig. 3 were aligned with other known gp90 sequences, using the program Pileup (10).
See the legend to Fig. 3 for an explanation of the sequence numbering, symbols, and features.
Gaps (indicated by dots) were introduced to align the sequences.
Arrows indicate the addition of potential N-glycosylation sites to the EIAVPV and the pony-derived sequences relative to the EIAVPr and EIAV19-2 sequences.
The following sequences were obtained from GenBank and have the accession numbers shown in parentheses: EIAVPr (M16575), CL22 (M87581), EIAV19-2 (U01866), and Wyoming (this is a consensus sequence derived from the sequences assigned accession numbers M87576, M87577, M87579, M87580, M87582, M87583, and M87586 to M87589). the low level of sequence variation detected in samples from the ﬁrst disease episode may simply reﬂect the rate of mutation and the fact that we analyzed samples taken shortly after infection (19 or 33 days postinfection).
In contrast to our results, Alexandersen and Carpenter (1) showed that signiﬁcant sequence diversity in portions of gp90 was present during the ﬁrst disease episode in a sample isolated from a horse infected with a ﬁeld isolate of EIAV.
These workers sequenced proviral DNA isolated from macrophages after one passage of the virus in culture, as opposed to examining viral RNA isolated directly from plasma.
Passage of the virus in culture may have altered the population of quasispecies as was reported previously for HIV (23).
In addition, the uncharacterized, uncloned ﬁeld isolate probably contained a more complex quasispecies than the biologically cloned EIAVPV stock, which could account for the discrepancy in the levels of variation reported by the two studies.
The observed sequence changes in the replicating EIAV quasispecies isolated from infected ponies were not randomly distributed; they were localized to either the PND (pony 11) or the PND and HVR (pony 06) (Fig. 3).
This localized sequence variation was detected concomitantly with (pony 06) or shortly after (pony 11) seroconversion (data not shown), which sug- FIG. 5.
Infection of equine monocytes with parental and chimeric EIAV.
Equine monocytes were mock infected (}) or infected with parental (EIAV19-2) (s), EIAVPV100 (å), or EIAVPV101 (x) at a multiplicity of 0.01 infectious center dose per cell.
After 4 h at 37ЊC, the virus inoculum was removed and replaced with fresh medium.
The medium was replaced daily.
The harvested medium was clariﬁed at 800 ϫ g, and the supernatants were assayed for reverse transcriptase activity. VOL. 70, 1996 EIAV GENOMIC QUASISPECIES 3353 gested that selection by the humoral immune response could account for the genetic changes.
It is important to note, however, that neutralizing antibodies were not detected at the time the plasma samples were taken.
Although anti-EIAV T-cell responses have not been studied in detail, these responses may also contribute to genetic variation if the situation for EIAV is analogous to that of HIV, in which case T-cell responses are seen prior to seroconversion in HIV-infected individuals (8, 17).
Rwambo et al. (37) reported changes in the antigenic and biochemical properties of EIAVPV gp90, which supported the proposal that antigenic variation is responsible for escape from neutralization.
The consensus EIAVPV gp90 sequence differed substantially from the EIAVPr sequence, particularly within the PND, where there were three amino acid changes and the addition of a potential N-glycosylation site (Fig. 4).
This may account for the fact that two neutralizing monoclonal antibodies directed against the EIAVPr PND did not react with EIAVPV gp90 (37).
In addition, the EIAVPV consensus sequence differed from the EIAVPr sequence at four amino acid positions within the hypervariable region (three of which introduced potential N-glycosylation sites), suggesting that changes within this region may also be important for escape from immune surveillance.
Together, these data strongly suggest that escape from neutralization was correlated, in part, to sequence changes within the PND and possibly the hypervariable region as well.
All three samples had identical consensus LTR sequences (Fig. 2), and the variation within samples was very low (Table 1).
This ﬁnding is in contrast to those of previous reports which showed that this region of the LTR is hypervariable (6, 30, 31).
Since the EIAVPV stock was biologically cloned twice, the lack of diversity for this sample was not surprising.
In regard to the uniformity exhibited among the pony-derived LTRs, it is important to note that much less variability is seen when ﬁeld isolates (or pony isolates) are compared with each other than when they are compared with tissue culture-adapted strains (6, 31).
Our sequencing results support the idea elaborated previously (21) that selection for replication in different host cells (ie, macrophages in vivo versus tissue culture cells in vitro) is responsible for the hypervariability seen among the LTRs of different virus stocks.
The EIAVPV and pony-derived LTR sequences were most similar to the Wyoming LTR sequence (Fig. 4) (34).
However, analysis of potential transcription factor binding motifs showed that the latter contained a second CAAT box motif and a third potential core ets motif that were absent in the former.
These two motifs have been suggested to be important in virus pathogenesis and replication.
Two CAAT boxes are present in the LTRs of several viruses associated with disease (eg, Wyoming, Th-1, and several virus isolates recovered during disease episodes [6, 30, 34]), whereas a single CAAT box is usually present in avirulent viruses (EIAVPr, MA-1, and EIAV19-2 [6, 31, 35]).
Our LTR sequencing results showed that viruses containing a single CAAT box predominated during a disease episode and that a pathogenic stock of virus need not contain an LTR with two CAAT boxes.
The macrophage- and Blymphocyte-restricted ets transcription factor family member PU.1 binds the EIAV LTR and is important for high-level Tat-transactivated transcription in macrophages in vitro (7, 2).
Since an LTR with only two ets core motifs was present in all of the samples that were sequenced, including those isolated from two independent disease episodes, the data suggest that three ets core motifs were not required for virus-induced disease.
Payne et al. (30) also reported an LTR sequence containing one CAAT box and two core ets motifs that was iso- lated from a pony during a disease episode.
However, this sequence may have been altered during biological cloning of the virus in FEK cells.
The possibility that our samples contained a low abundance (Յ1 to 8%) of sequences with two CAAT boxes and/or three core ets motifs cannot be ruled out.
Our results showed that the LTR and env sequences of EIAVPV were similar to those found in the plasma of two different ponies during a disease episode and were signiﬁcantly different from the corresponding sequences of nonpathogenic strains (Fig. 2 and 4).
Thus, we reasoned that the predominant EIAVPV LTR and env sequences might be associated with disease.
To test this hypothesis, we constructed chimeric viruses that differed from the parental nonpathogenic molecular clone EIAV19-2 only in the U3-LTR or this region in combination with the gp90 variable region.
Both chimeric viruses replicated in equine monocytes to the same extent as the parental virus.
However, by 120 days postinfection, neither chimera had induced disease in Shetland ponies.
While a limited number of ponies were used in this study, the amount of recombinant chimeric virus used for these infections (107.0 infectious center doses) was at least 104.0-fold greater than the standard EIAVPV inoculum that produces disease in 100% of inoculated ponies.
These data suggested that the viral determinants of pathogenesis may lie outside the regions used to make the chimeric viruses.
Alternatively, development of disease may involve a mixture of viral quasispecies rather than a particular viral strain, perhaps in agreement with the lack of selection of particular EIAVPV quasispecies during the initial stages of infection and disease as observed here.
Studies using chimeras between nonpathogenic and pathogenic molecular clones of simian immunodeﬁciency virus indicate that although the env gene is an important determinant of pathogenesis, it is not the only determinant (20, 26).
These results indicate that other portions of the EIAV genome should be assessed for their roles in pathogenesis. ACKNOWLEDGMENTS We thank John Mellors and Sharon Harrold for helpful discussions and Phalguni Gupta for critically reading the manuscript.
This work was supported by National Institutes of Health grant 5R01CA49296 and by funds from the Lucille Markey Charitable Trust and the Kentucky Agricultural Experimental Station. REFERENCES 1.
Alexandersen, S., and S.
Characterization of variable regions in the envelope and S3 open reading frame of equine infectious anemia virus.
Virol. 65:4255–4262. 2.
Baier, M., M.
Cichutek, and R.
Development in vivo of genetic variability of simian immunodeﬁciency virus.
USA 88:8126–8130. 3.
Issel, and R.
Detailed mapping of the antigenicity of the surface unit glycoprotein of equine infectious anemia virus by using synthetic peptide strategies.
Virol. 66:732–742. 4.
Schipper, and M.
Insertion of N-linked glycosylation sites in the variable regions of the human immunodeﬁciency virus type 1 surface glycoprotein through AAT triplet reiteration.
Virol. 68:7566–7569. 5.
P., and R.
Selection of genetic variants of simian immunodeﬁciency virus in persistently infected rhesus monkeys.
Virol. 65:1843–1854. 6.
Carpenter, S., S.
Perryman, and B.
Identiﬁcation of a hypervariable region in the long terminal repeat of equine infectious anemia virus.
Virol. 65:1605–1610. 7.
Carvalho, M., and D.
The PU.1/Spi-1 proto-oncogene is a transcriptional regulator of a lentivirus promoter.
Virol. 67:3885–3890. 8.
Couillin, I., P.
Guillet, and S.
Impaired cytotoxic T lymphocyte recognition due to genetic variations in the main immunogenic region of the HIV-1 nef protein.
Med. 180:1129–1134. 9.
Gao, F., L.
Shaw, and B.
Read more about Pony 1 : Plasma taken from each pony during the ﬁrst disease episode….: