Potent and Stable Attenuation of Live-HIV-1 by Gain of a Proteolysis-resistant Inhibitor of NF-κB (IκB-αS32/36A) and the Implications for Vaccine Development*

Live-attenuated human immunodeficiency viruses (HIVs) are candidates for Acquired Immunodeficiency Syndrome (AIDS) vaccine. Based on the simian immunodeficiency virus (SIV) model for AIDS, loss-of-function (e.g. deletion of accessory genes such as nef) has been forwarded as a primary approach for creating enfeebled, but replication-competent, HIV-1/SIV. Regrettably, recent evidence suggests that loss-of-function alone is not always sufficient to prevent the emergence of virulent mutants. New strategies that attenuate via mechanisms distinct from loss-of-function are needed for enhancing the safety phenotype of viral genome. Here, we propose gain-of-function to be used simultaneously with loss-of-function as a novel approach for attenuating HIV-1. We have constructed an HIV-1 genome carrying the cDNA of a proteolysis-resistant nuclear factor-κB inhibitor (IκB-αS32/36A) in the nef region. HIV-1 expressing IκB-αS32/36A down-regulates viral expression and is highly attenuated in both Jurkat and peripheral blood mononuclear cells. We provide formal proof that the phenotypic and attenuating characteristics of IκB-αS32/36A permit its stable maintenance in a live, replicating HIV-1 despite 180 days of forced ex vivopassaging in tissue culture. As compared with other open-reading frames embedded into HIV/SIV genome, this degree of stability is unprecedented. Thus, IκB-αS32/36A offers proof-of-principle that artifactually gained functions, when used to attenuate the replication of live HIV-1, can be stable. These findings illustrate gain-of-function as a feasible strategy for developing safer live-attenuated HIVs to be tested as candidates for AIDS vaccine.

Live-attenuated human immunodeficiency virus, type-1 (HIV-1) 1 viruses are vaccine candidates for AIDS based on the assumption that poorly replicating viruses are reduced in cytopathicity while eliciting an efficient immune response in vivo. In fact, in the SIV model of AIDS, viruses deleted of accessory genes, such as nef, are attenuated and confer immune protection in adults (1)(2)(3)(4)(5)(6). Recent results have raised concerns about the safety of such live-attenuated viruses because of the emergence of virulent mutants (7)(8)(9)(10)(11). Hence, it is reasonable that new mechanistically distinct strategies should be considered in combination with loss-of-function to develop safe live-attenuated vaccine candidates. Stable attenuation of lentivirus is difficult. Reasons for this include the high mutability of retroviruses as a consequence of poor fidelity in reverse transcription and high rates of viral replication. Virus replication is one of several parameters used to measure HIV disease progression. To this end, endowing the viral genome with an inhibitory function that represses replication is, in principle, a valid strategy for attenuation.
AIDS pathogenesis involves complex host-pathogen interactions (12)(13)(14)(15)(16)(17). HIV-1 entry occurs through binding of gp120 envelope glycoprotein to CD4 and chemokine co-receptors (18,19). Later, after reverse transcription and integration, regulated HIV-1 gene expression is dependent on cellular transcription factors and on the viral Tat protein acting through several cis regulatory sequences in the HIV-1 LTR (20,21). Prominent among these sequences are two NF-B sites located upstream of the Sp1-TATAA motif (22) and a third NF-B site co-incident with the TAR sequence (23). The prevalence of NF-B sites in all HIV isolates (24) suggests that NF-B function is fundamentally important for virus replication and represents an attractive target to attenuate HIVs. NF-B is regulated by IB inhibitors (25). In response to activating stimuli, IB proteins become phosphorylated, ubiquinated, and degraded by proteasomes. This releases active NF-B complexes leading to transcriptional activation of responsive genes. An IB-␣ mutant protein, IB-␣S32/36A (defective for serine 32-and serine 36phosphorylation), was previously shown to resist proteolysis (26,27). IB-␣S32/36A was found to be a particularly potent inhibitor of NF-B-dependent gene transcription (26,27). Based on the role of NF-B in virus transcription, we reasoned that expression of IB-␣S32/36A could be one way to attenuate HIV-1.
Here, we have investigated the feasibility of a stable bimodal attenuation of HIV-1 through both gain-and loss-of-function.
We show that a proteolysis-resistant dominant negative IB-␣ molecule (IB-␣S32/36A) can be inserted into nef of HIV-1, is maintained in the viral genome despite prolonged passaging in tissue culture, and contributes to a strong attenuation of HIV-1. These results indicate that gain-of-function should be considered together with loss-of-function in strategies for constructing safe live-attenuated HIV-1 vaccine candidates. In addition, they demonstrate that NF-B inhibition is a major target for HIV-1 attenuation.

EXPERIMENTAL PROCEDURES
Construction of IB-␣ Recombinant HIV-1 Viruses-pNL⌬nef was generated by replacing the first 117 nucleotides of nef sequence of pNL4 -3 (28) with a linker specifying for the unique restriction sites XhoI, XbaI, and NotI. This resulted in the frameshift of the remaining nef sequence. cDNAs for IB-␣ and IB-␣S32/36A were amplified by PCR using a 5Ј-primer flanked by SalI site and a 3Ј-primer flanked by the FLAG sequence followed by stop codon and XbaI site. The 5Ј-primer was 5Ј-ACGCGTCGACATGTTCCAGGCGGCCGAGCGC-3Ј; the 3Јprimer was 5Ј-GCTCTAGATCACTTGTCGTCATCGTCTTTGTAGTCT-AACGTCAGACGCTGGCCTCCAAA-3Ј. The PCR products for IB-␣ and IB-␣S32/36A were digested with SalI and XbaI and ligated into pNL⌬nef to generate pNLIB-wt and pNLIB-M, respectively. IB-␣S32/36A was also cloned in antisense orientation using a 5Ј-primer flanked by XbaI site and a 3Ј-primer flanked by SalI site to generate pNLIB-as. Molecular clones were confirmed by DNA restriction analysis and DNA sequencing.
Transfections and Viral Stocks-293T cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% v/v heat-inactivated fetal bovine serum and 3 mM glutamine. Viral stocks were produced by transfecting 293T cells (5 ϫ 10 6 ) with viral plasmids (10 g) using calcium phosphate. Forty hours later, the cell culture supernatant was passed through a 0.45-m filter and measured for RT activity.
Immunoblot Analysis-293T cells were transfected with viral plasmids (10 g) and were lysed in RIPA buffer 24 h later. Proteins (10 g) were separated by electrophoresis in 10% SDS-polyacrylamide gel and transferred to Immobilon-P (Millipore). Filters were blotted with AIDS patient serum or with anti-FLAG monoclonal antibody (Eastman Kodak, Rochester, NY) and anti-IB-␣ antiserum recognizing the amino acids 51-64 of IB-␣ (gift from N. Rice, National Cancer Institute-Frederick Cancer Research and Development Center, MD) using Western-Light Chemiluminescent Detection System (Tropix, Bedford, MA).
RNase Protection Assay-293T cells were transfected with viral plasmids (10 g) using calcium phosphate, and total RNA was isolated by acid guanidinium thiocyanate/phenol/chloroform extraction 3 days after transfection. RNA was treated with RQ1 DNase from Promega (2 units/50 g of RNA) at 37°C for 15 min, extracted by phenol:chloroform (25:24) and precipitated in ethanol. RNase protection assay was performed using RPA II kit (Ambion, Austin, TX). Briefly, aliquots of 10 g of RNA from different samples were hybridized with 10 5 cpm of HIV-1 and actin RNA probes for 15 h at 44°C. Yeast RNA was used as control. Samples were treated with RNase A (2.5 units/ml) and RNase T1 (100 units/ml) for 30 min at 37°C, precipitated in ethanol, resuspended in gel loading buffer (95% formamide, 0.025% xylene cyanol, 0.025% bromphenol blue, 0.5 mM EDTA, 0.025% SDS), and separated by electrophoresis on 8 M urea, 8% acrylamide gel in Tris borate-EDTA. HIV-1 antisense probe to the pNL4 -3 sequence from ϩ350 to ϩ535 nt corresponding to NF-B, Sp1, TATAA, and TAR sequences of HIV-1 LTR was transcribed from the T7 promoter in the pGEM LTR vector. Protected HIV-1-specific bands of 80 and 185 bp correspond to the 5Ј-LTR (ϩ455 to ϩ535 nt of pNL4 -3 sequence) and the 3Ј-LTR (ϩ9425 to ϩ9610 nt of pNL4 -3 sequence), respectively. Actin antisense probe was transcribed using T7 RNA-polymerase from vector pTRI-␤-actin-125-human (Ambion). Protected actin transcript corresponds to a band of 127 bp. T7 RNA polymerase-derived transcripts were produced by using the Riboprobe in vitro Transcription System (Promega). Transfection efficiencies were normalized to CAT activity measured from co-transfections with pBLCAT2 (2 g), which contains the cat gene fused to the herpes simplex virus tk minimal promoter, followed by CAT assay (23).
Viral Growth-Jurkat cells were cultured in RPMI supplemented with 10% v/v heat-inactivated fetal bovine serum and 3 mM glutamine. PBMCs were stimulated with PHA (0.5 g/ml) or mAb OKT3 (1 g/ml) in RPMI 1640 with 10% v/v heat-inactivated fetal bovine serum for 3 days and then washed and cultured in RPMI 1640 with 10% fetal bovine serum with interleukin-2 (20 units/ml). Cells were infected with equivalent doses of viral stocks normalized by RT activity. Cell supernatants were collected every 3 days for RT assay; equal volumes of fresh medium were replaced into the cultures at the same time.
RT-PCR Analysis-Viral RNA was purified from cell culture supernatants using QIAampViral RNA kit (Qiagen). RNA extracts (1 g) were reverse transcribed, and PCR was amplified using Titan One Tube RT-PCR System (Roche Molecular Biochemicals). The IB-␣S32/36A sequence was amplified with primers flanking the cDNA insert from ϩ8732 to ϩ8749 nt (5Ј-primer; env region) and from ϩ9017 to ϩ9034 nt FIG. 1. Recombinant HIV-1 viruses that express IB-␣. Panel A, structure of HIV-1 genomes that express IB-␣ wild-type (pNLIB-wt), IB-␣S32/36A in sense (pNLIB-M), and antisense (pNLIB-as) orientations, respectively. Panel B, immunoblot analysis of total extracts (10 g each) from 293T cells 24 h after transfection with the indicated HIV-1 plasmids (10 g) using a hyperimmune AIDS patient serum. Panel C, immunoblot analysis of total extracts (10 g) from 293T cells 24 h after transfection with the indicated HIV-1 plasmids (10 g) using anti-FLAG monoclonal antibody followed by an anti-IB-␣ antiserum. This mixture identifies both IB-␣-FLAG expressed from the viral genomes and endogenous IB-␣.
(3Ј-primer; nef region) of pNL4 -3. As control, the integrase region was amplified with 5Ј-primer from ϩ4339 to ϩ4359 nt and 3Ј-primer from ϩ4894 to ϩ4914 nt of pNL4 -3. Amplification was carried out according to the following protocol: incubation at 50°C for 30 min, preheating at 95°C for 2 min followed by 30 cycles of 95°C for 45 s, 58°C for 45 s, and 72°C for 3 min.
Flow Cytometry Analysis-Cells were stained with phycoerythrinconjugated anti-p24 mouse monoclonal antibody (Ortho Diagnostic, Raritan, NJ) and Annexin-V-Fluos (Roche Molecular Biochemicals). Flow cytometry was performed with a Coulter Epics XL Cytometer. Ten thousands cells were analyzed under each condition.

RESULTS
HIV-1 Expressing IB-␣S32/36A Down-regulates Viral Expression-To determine whether the attenuation of HIV-1 could be obtained via cis-expression of a proteolysis-resistant inhibitor of NF-B, FLAG-tagged IB-␣S32/36A cDNA was positioned in sense and antisense orientations into nef of HIV-1 pNL4 -3 generating pNLIB-M and pNLIB-as genomes, respectively (Fig. 1A). As a control, an otherwise isogenic molecular clone of pNL4 -3 expressing the proteolysis-sensitive wildtype IB-␣ (pNLIB-wt) cDNA was also constructed (Fig. 1A). These insertions also resulted in the deletion of 39 amino acids from the N terminus of Nef and a translational frameshift for the remaining Nef-encoding codons. Hence, the chimeric genomes gained an IB-␣S32/36A function coupled with a simultaneous loss of Nef function. From these molecular genomes, different virus stocks were generated through independent transfections of 293T cells. Expression in the transfected cells of HIV-1 proteins and the heterologous IB-cDNAs was verified by immunoblotting (Fig. 1, B and C). To assess the functional impact of IB-␣S32/36A in the context of HIV-1 gene expression, viral transcription in 293T cells transfected individually with pNL4 -3, pNL⌬nef, pNLIB-M, pNLIB-as, or pNLIB-wt was analyzed by RNase protection assays. Viral RNAs were efficiently produced from all genomes except for the IB-␣S32/ 36A-expressing virus, pNLIB-M ( Fig. 2A). This finding points to a strong inhibition of LTR-directed transcription by IB-␣S32/36A. A normalized simultaneous co-transfection into cells of pNLIB-M genome with pNL4 -3 genome fused to the luciferase reporter gene still resulted in markedly reduced viral transcription (Fig. 2B). The inhibitory mechanism is likely to be a dominant trans-repressive effect from the IB-␣S32/36A protein rather than a cis-destabilizing phenomenon arising as a consequence of an insertion into nef. This is further supported by the fact that the insertion of IB-␣ cDNA either in antisense orientation, or as wild-type sequence does not inhibit viral expression ( Fig. 2A, lanes 7 and 8). Repressed expression would have been expected if there were a cis insertional effect.
Attenuation of HIV-1 Expressing IB-␣S32/36A-Next, the replication properties of the recombinant HIV-1 genomes were measured. We monitored for viral growths in established (Jurkat) and primary (PBMCs) T-cells. Based on normalized amounts of input viruses, NLIB-M was found to be highly attenuated in Jurkat cells when compared with the NL⌬nef and NLIB-as (Fig. 3A). A reduced replication capacity was also observed for NLIB-M in human PBMCs stimulated either with anti-CD3 mAb OKT3 (Fig. 3B) or PHA (Fig. 3C). In parallel, the chimeric HIV-1 which expresses the proteolysis-sensitive wild-type IB-␣ produced a replication profile that was only slightly attenuated when compared with control NLIB-as virus (Fig. 4, A-C). Taken together, these results indicate a critical contribution of proteolysis-resistant mutant IB-␣ toward attenuated viral growth.
Attenuation of NLIB-M Is Not Because of Apoptosis-Previously, IB-␣S32/36A was shown to enhance the apoptosis induced by tumor necrosis factor-␣ (30 -32). Because tumor necrosis factor-␣ is produced during HIV-1 infection (33,34), we asked whether a pro-apoptotic effect of IB-␣S32/36A might account, in part, for attenuation of viral growth. To this end, Jurkat cells infected separately with NL4 -3, NLIB-M, or NLIB-as viruses were analyzed by flow cytometry for the expression of p24 and for binding to annexin V, a marker of an early event in apoptosis (35). Because of the different viral growth kinetics, to account fully for equivalent levels of virus production, cell samples were analyzed at variant times postinfection. As shown in Table I, the percentage of apoptotic cells among p24-positive population was 17 and 13% for NL4 -3 and NLIB-as, respectively, at day 10 post-infection, as compared with 19% for NLIB-M at day 18 post-infection. Thus, the expression of IB␣S32/36A did not affect apoptosis significantly because the percentage of apoptotic cells among those infected with NL4 -3, NLIB-as, or NLIB-M was similar.
NLIB-M Is Stably Attenuated through Cell Passaging-Retroviruses rapidly delete heterologous genes which do not confer benefits for replication (36 -38). One indication of such a loss in gained function would be a change in the kinetics of replication in later (compared with earlier) passages of viruses. In examining NLIB-M virus after prolonged culturing in cells, we noted that its attenuated phenotype bred true despite extended re-passaging (Fig. 5, A-C). No changes in replication phenotype were observed with up to 180 days (9 passages) of continuously forced propagation in ex vivo tissue cultures. This stability of IB-␣S32/36A contrasts sharply with documented findings in other systems where rapid deletion of gained func-

TABLE I Flow cytometric analysis of Jurkat cells infected with NL4 -3, NLIB-as, or NLIB-M
The cells were analysed by flow cytometry for the expression of p24, as marker of viral infection, and for binding to annexin V, as marker of apoptosis. Ten thousand cells were analysed under each condition. Values are expressed as percentages. tion commenced as early as 48 h after viral passaging (36). To verify biochemically that this phenotypic stability was from the maintenance of IB-␣S32/36A in nef, RT-PCR analysis of virion RNAs was performed using samples isolated after various times of culturing in Jurkat cells (Fig. 5D). Viral RNAs from first passage (20 days in culture) as compared with sixth (120 days in culture) and ninth (180 days in culture) passage were indistinguishable, providing no evidence for deletion of IB-␣S32/36A (Fig. 5D, compare lanes 5-7). Thus, by contrast with other cDNAs inserted into nef (36 -38) in the setting of replicating HIV-1/SIV, IB-␣S32/36A is unprecedently stable. DISCUSSION Various approaches have been proposed for generating liveattenuated HIV-1 (39). Based on findings from the SIV model of AIDS, loss-of-function (e.g. deletion of accessory genes such as nef, vpr, and U3) has been forwarded as a primary approach for creating enfeebled, but replication-competent, HIV/SIV (1-6). Regrettably, recent evidence indicates that loss-of-function alone is not always sufficient (and is perhaps largely insufficient) to prevent virulence (7)(8)(9)(10)(11). Hence, one is faced with either abandoning live-attenuated virus as a potential route for HIV-1 vaccine or devising another approach which is mechanistically distinct from and could be used additively with lossof-function in attenuating the AIDS virus. Attenuation of HIV-1 by gain-of-function was first broached four years ago (40). However, an issue that has impeded the useful application of gain-of-function is uncertainty over the stability of functions added to a replicating HIV-1 genome. To date, all exogenous open-reading frames, such as thymidine kinase, ␥-interferon, interleukin-2 (36 -38), when artifactually embedded into the HIV-1/SIV genome have been unstable to prolonged ex vivo passaging, suggesting a rule of rapid deletion of gained genes upon virus replication. However, because the sampling size for gained functions has been small, it remains unclear whether this is indeed a universal rule. Our findings from IB-␣S32/36A formally challenge the rigidity of this rule.
An ability by viruses in general and by HIV-1 specifically to maintain stably gained functions should not be inherently surprising. For example, piracy of large numbers of cellular genes by herpesviruses has been well documented (see Ref. 41 and references cited therein). In simple animal retroviruses, only Gag, Pol, and Env open-reading frames are required for virus propagation. Hence, the evolution of tat, rev, nef, vpu, vif, and vpr in the existing HIV-1 genome represents likely illustrations of how this virus has anciently through natural means acquired and continues to maintain stably added reading frames. Conceivably, IB-␣S32/36A has certain cellular characteristics that benefit the virus-cell interaction during HIV-1 replication. What might be the propitious characteristics specified by IB-␣S32/36A await further investigation. It is possible that the inhibition of virus transcription by IB-␣S32/36A allows the cell to survive HIV-1 infection, while cell killing occurs in the case of recombinant viruses that carry exogenous genes which do not confer virus attenuation. A second possible explanation for the insert stability in NLIB-M is that IB-␣S32/36A could provide an "anti-mutator" effect to the chimeric HIV-1 genome. This "anti-mutator" effect would confer a relative resistance to the deletion of the gained function. In this regard, the expression of a dominant negative IB-␣ has been shown to correct aberrant DNA synthesis which causes genetic instability in ataxia telangectasia cells (42).
In conclusion, we demonstrate here the feasibility of a gainof-function strategy for stably attenuating live HIV-1. Specifically, we show that the addition of IB-␣S32/36A to an HIV-1 genome conferred a highly attenuated replication phenotype that is genetically stable for a minimum of 180 days of growth in tissue culture. Attenuated replication in ex vivo tissue culture is by no means a perfect predictor of in vivo viral robustness; however, the unprecedented observation that emerges from IB-␣S32/36A is its durability in settings where other artifactually gained open-reading frames (36 -38) have failed. An important extension to the experiments here would be to explore the in vivo stability of IB-␣S32/36A in chimeric SIVs propagated in macaques. However, the fact that transcription from the SIV LTR is not ruled by NF-B in the same manner as that from the HIV-1 LTR (43-45) undermines the feasibility of such a heterologous in vivo test for IB-␣S32/36A. In this regard, further studies on SIV LTR-driven transcription are warranted to identify analogous cellular inhibitors of SIV replication. That not withstanding, the findings from IB-␣S32/ 36A represent a first-step proof that gain-of-function could be a stably useful approach for attenuation. This first step should spur further stepwise exploration for other useful attenuating markers that might be more amenable to incremental testing in nonhuman primate models. If so, a combination approach that incorporates gain-of-function with loss-of-function could possibly contribute toward the development of a live-attenuated HIV-1 vaccine.