Persistence of attenuated HIV-1 rev alleles in an epidemiologically linked cohort of long-term survivors infected with nef-deleted virus

Background The Sydney blood bank cohort (SBBC) of long-term survivors consists of multiple individuals infected with nef-deleted, attenuated strains of human immunodeficiency virus type 1 (HIV-1). Although the cohort members have experienced differing clinical courses and now comprise slow progressors (SP) as well as long-term nonprogressors (LTNP), longitudinal analysis of nef/long-terminal repeat (LTR) sequences demonstrated convergent nef/LTR sequence evolution in SBBC SP and LTNP. Thus, the in vivo pathogenicity of attenuated HIV-1 strains harboured by SBBC members is dictated by factors other than nef/LTR. Therefore, to determine whether defects in other viral genes contribute to attenuation of these HIV-1 strains, we characterized dominant HIV-1 rev alleles that persisted in 4 SBBC subjects; C18, C64, C98 and D36. Results The ability of Rev derived from D36 and C64 to bind the Rev responsive element (RRE) in RNA binding assays was reduced by approximately 90% compared to Rev derived from HIV-1NL4-3, C18 or C98. D36 Rev also had a 50–60% reduction in ability to express Rev-dependent reporter constructs in mammalian cells. In contrast, C64 Rev had only marginally decreased Rev function despite attenuated RRE binding. In D36 and C64, attenuated RRE binding was associated with rare amino acid changes at 3 highly conserved residues; Gln to Pro at position 74 immediately N-terminal to the Rev activation domain, and Val to Leu and Ser to Pro at positions 104 and 106 at the Rev C-terminus, respectively. In D36, reduced Rev function was mapped to an unusual 13 amino acid extension at the Rev C-terminus. Conclusion These findings provide new genetic and mechanistic insights important for Rev function, and suggest that Rev function, not Rev/RRE binding may be rate limiting for HIV-1 replication. In addition, attenuated rev alleles may contribute to viral attenuation and long-term survival of HIV-1 infection in a subset of SBBC members.


Background
The Sydney blood bank cohort (SBBC) of long-term survivors (LTS) consists of multiple individuals who became infected with attenuated strains of human immunodeficiency type 1 (HIV-1) via contaminated blood products from a common blood donor between 1981 and 1984 [1][2][3]. Long-term prospective studies showed convergent evolution of nef/long-terminal repeat (LTR) sequences in virus harbored by SBBC members, characterized by progressive sequence deletions toward a minimal nef/LTR structure retaining only sequence elements required for viral replication [4]. Thus, gross deletions in the nef/LTR region of the HIV-1 genome contribute to viral attenuation and slow progression of HIV-1 infection in SBBC members. Despite convergent nef/LTR sequence evolution, after 22 to 26 years of infection SBBC members comprise antiretroviral therapy (ART)-naïve long-term nonprogressors (LTNP) as well as slow progressors (SP) who eventually commenced ART, suggesting that other viral and/or host factors may contribute to the in vivo pathogenicity (or lack thereof) of SBBC HIV-1 strains [3,4].
HIV-1 Rev is a 116 amino acid (aa), ~18 kD regulatory protein whose primary function is to mediate the nucleocytoplasmic transport, and therefore expression, of unspliced and singly spliced HIV-1 mRNA transcripts encoding viral structural proteins, via binding to the Rev response element (RRE) which is a complex RNA stemloop structure present in these transcripts [reviewed in [[18-21]]. Therefore, Rev activity is essential for HIV-1 replication. Extensive mutational analysis of Rev has identified 2 distinct functional domains [reviewed in [21]]. These include an arginine-rich N-terminal region at aa positions 34 to 50 which contains the nuclear localization signal (NLS) and the RNA-binding domain (RBD) that mediates direct binding of Rev to the RRE, and a highly conserved leucine-rich C-terminal activation domain at aa positions 75 to 83 which contains the nuclear export signal (NES). The N-terminal NLS/RBD is flanked on both sides by less well defined sequences that are required for multimerization [22][23][24][25].
A previous study of rev alleles isolated from a subject with long-term nonprogressive HIV-1 infection showed a persistent Leu to Ile change at position 78 in the activation domain which attenuated Rev function and HIV-1 replication capacity [10], providing the first evidence that defective rev alleles may contribute to long-term survival of HIV-1 infection in some patients. A subsequent study of naturally occurring rev alleles with rare sequence variations in the activation domain showed variable reductions in Rev activity [26], although it was unclear from this study whether the reductions in Rev activity observed would be sufficient to attenuate HIV-1 replication capacity. In the present study, we undertook a genetic and functional analysis of HIV-1 rev alleles isolated from 4 SBBC subjects to determine whether defects in viral genes other than nef/LTR contribute to attenuation of HIV-1 strains harbored by SBBC members.

Subjects
The clinical history of the study subjects, results of laboratory studies and antiretroviral therapies have been described in detail previously [3,4,27]. The results of laboratory studies relevant for the longitudinal samples used in this study are summarized in Table 1. Briefly, D36 acquired HIV-1 sexually in December 1980. C18, C64 and C98 acquired HIV-1 by receiving blood products donated by D36 in August 1983, April 1983 and February 1982, respectively. After 19 years of asymptomatic infection without ART, D36 was placed on highly active ART (HAART) in January 1999 after evidence of HIV-1 progression. C98 was also placed on HAART in November 1999 after 18 years of HIV-1 infection, and died of causes unrelated to HIV-1 in March 2001. C64 has been infected for 24 years without ART, and has stable CD4 T-cells and below detectable viral load. C18 died of causes unrelated to HIV-1 in November 1995, but prior to death was asymptomatic with stable CD4 T-cell count for 12 years without ART. Thus, D36 and C98 are SP, and C18 and C64 are LTNP [3,4]. CCR5Δ32 genotyping by PCR showed that all subjects carried CCR5 (wt/wt) alleles ([28], and J. S. Sullivan, personal communication). CCR2-64I genotyping by PCR-RFLP showed that C64 and C98 carried the CCR2-64I (wt/wt) genotype [28]. The CCR2-64I genotype of C18 and D36 has not been determined.

Persistence of unique rev alleles in SBBC members
Peripheral blood mononuclear cells (PBMC) isolated from blood samples longitudinally collected on 4 occasions between 1995 and 2001 were available from D36, C64 and C98 for this study (Table 1). Only one blood sample collected in 1993 was available from C18. Blood was taken from subjects in accordance with guidelines endorsed by the Australian Red Cross Blood Service human ethics committee. Multiple, independent fulllength Rev clones containing the first and second Rev coding exons were generated from genomic DNA of each PBMC sample and sequenced. Phylogenetic analysis showed that all Rev sequences were clade B (data not shown). The dominant Rev aa sequence from each PBMC sample, which represents the consensus sequence from 10 independent clones, is shown in Additional file 1. In each subject where longitudinal PBMC samples were available (D36, C64 and C98), the persistence of a dominant rev allele was evident over a 4-to 6-year period. Figure 1 shows an aa sequence alignment of these dominant and persistent rev alleles as well as the dominant rev allele in the single C18 PBMC sample. Single aa changes at positions 74, 104, 106, 108 and 112 in sequence encoding Rev exon 2 segregated the dominant C18 and C98 Revs from the dominant C64 and D36 Revs. However, each dominant Rev sequence contained unique, distinguishing aa changes. In addition, C18, C64 and C98 Revs had a 3 aa extension at the Rev C-terminus, and D36 Revs had a 13 aa extension at this position. Similar C-terminal extensions were not identified in 164 Rev sequences available in the Los Alamos data base and other published studies [10]. Thus, the dominant and persistent Revs harbored by these SBBC members are unique. The following studies functionally characterized Rev proteins derived from these dominant and persistent SBBC rev alleles.

Rev proteins derived from subjects C64 and D36 have attenuated RRE binding capacity
The ability of His-tagged Rev proteins derived from the dominant and persistent SBBC rev alleles to bind the RRE was quantified by electrophoretic mobility shift assays with [ 32 P]-labelled RNA transcripts bearing the RRE ( Fig.  2A). His-tagged Rev and Matrix proteins derived from HIV-1 NL4-3 were used as positive and negative controls, respectively. Compared to His-tagged Rev from HIV-1 NL4-

Rev amino acid sequences associated with attenuated RRE binding
Attenuated RRE binding was not due to mutations in the N-terminal RBD, since the amino acid sequences across this region were conserved among all SBBC rev alleles and were identical to HIV-1 NL4-3 (Fig. 1). This was somewhat surprising, since previous studies showed that the RBD of Rev was the principal determinant of RRE binding [23-25,29-32]. The C-terminal 3 aa extensions present in C18, C64 and C98 Revs ( Fig. 1) had no effect on RRE binding, since RRE binding by C18 and C98 Revs was similar to HIV-1 NL4-3 . Three amino acid changes that were conserved among D36 and C64 rev alleles and that were not present in C18 and C98 rev alleles were identified outside the RBD; Gln to Pro at position 74 immediately N-terminal to the Rev activation domain, and Val to Leu and Ser to Pro at positions 104 and 106 at the Rev C-terminus, respectively (Fig. 1). Amino acid changes also occurred at positions 108 and 112 which segregated C64 and D36 Revs from C18 and C98 Revs, but database analysis showed that amino acid variation is frequent at these positions (data not shown  Table 2). The frequency of any 2 of these residues being present was 0.006. None of the available sequences had all 3 amino acid changes. Thus, the amino acid changes occurring in D36 and C64 Revs are unique. However, the presence of one or more of these amino acid changes was not able to discriminate between subjects with progressive or non-progressive HIV-1 infection (

Rev derived from D36, but not C64, has impaired function
To determine whether SBBC Revs have impaired function, D36, C64, C18 and C98 Revs were subcloned into the pcDNA3.1 expression vector. Western blot analysis of Rev protein expression using sheep polyclonal anti-Rev antiserum showed equivalent levels of Rev in lysates of transfected CEM cells (Fig. 3A). Rev function in mammalian cells was investigated using the Rev-dependent reporter plasmid pDM128 [31], which expresses the chloramphenicol acetyltransferase (CAT) gene in the presence of Rev, as described previously [36] (Fig. 3B). In this assay, the Rev expression plasmids were first titrated to determine an amount to use that was within the linear response range of the assay (data not shown). Levels of CAT activity were compared to those present in lysates of cells co-transfected with pDM128 and HIV-1 NL4-3 Rev. Cells cotransfected with pDM128 and empty pcDNA3.1 vector or pcDNA3.1 expressing HIV-1 NL4-3 Matrix protein were included as negative controls. Levels of CAT activity in lysates of cells transfected with C18 or C98 Revs were not significantly different to those in lysates of cells transfected with HIV-1 NL4-3 Rev. In contrast, levels of CAT activity in lysates of cells transfected with C64 or D36 Revs were reduced by approximately 20% and 50%, respectively (P < 0.01). Similar results were obtained using 293 cells (data not shown). In addition, similar results were obtained using a Rev-dependent HIV-1 env reporter system as a measure of Rev function, as described previously [37] (data not shown). These data suggest efficient Rev function by C18 and C98 Revs, a modest reduction in the activity of C64 Rev, but significant impairment in the activity of D36 Rev.

Amino acid sequences associated with impaired D36 Rev function
Efficient C18 and C98 Rev function is consistent with results of the Rev/RRE binding studies that showed efficient RRE binding by these Revs (Fig. 2). However, the modest or significant impairment in C64 or D36 Rev activity, respectively, is discrepant with results of the Rev/ RRE binding studies that showed equivalent reductions in RRE binding by these Rev variants (Fig. 2). Therefore, in C64 Rev, the reduced levels of RRE binding appear to be sufficient for the majority of Rev function to be retained.
Additional sequence changes that differentiate C64 and D36 Revs are likely to impair D36 Rev function. Longitudinal sequence analysis showed that the presence of an unusual 13 aa extension at the D36 Rev C-terminus was the only genetic alteration that consistently differentiated D36 Rev from C64 Rev (Fig. 1), [see also Additional file 1]. The otherwise isogenicity of the D36 and C64 Revs used in the functional studies identifies the C-terminal 13 aa extension as the primary determinant underlying impaired D36 Rev function.
It is presently unclear how this sequence alteration may affect D36 Rev function, but the additional 13 aa at the Rev C-terminus may affect Rev structure. One hypothesis is that such structural changes may interfere with the recruitment of cellular proteins to the NES such as eIF-5A [38], nucleoporins including Rip/Rab [39-43] and CRM1/ exportin 1 [44-46], which could potentially affect nuclear export. The presence of Pro at position 74 immediately Nterminal to the Rev NES may induce further structural changes contributing to this interference, which might also account for the modest reduction in C64 Rev activity. Further studies are required to fully elucidate the importance of amino acid alterations that impair D36 Rev function.

Rev cloning and sequencing
Full-length HIV-1 Rev clones containing the first and second Rev coding exons were generated from genomic DNA of patient PBMC samples by PCR using Expand high fidelity DNA polymerase (Roche Diagnostics, Basel, Switzerland) as follows; The first Rev coding exon was amplified using primers 5RevE2 (5'-GGGTGTCGACATAGCA-GAATAG-3'; corresponding to nt positions 5781 to 5802 of HIV-1 NL4-3 ) and 3RevE2 (5'-CTGCTTTGATAGA-GAAGCTTG-3'; corresponding to nt positions 6024 to 6044 of HIV-1 NL4-3 ) that spans a SalI restriction site 5' to the Rev start codon. The second Rev coding exon was Analysis of Rev protein expression and function in mamma-lian cells Results are representative of three independent experiments. *P < 0.01, Student's t test.