High Attenuation and Immunogenicity of a Simian Immunodeficiency Virus Expressing a Proteolysis-resistant Inhibitor of NF-κB*

NF-κB/IκB proteins play a major role in the transcriptional regulation of human immunodeficiency virus, type-1 (HIV-1). In the case of simian immunodeficiency virus (SIV) the cellular factors required for the viral transcriptional activation and replication in vivo remain undefined. Here, we demonstrate that the p50/p65 NF-κB transcription factors enhanced the Tat-mediated transcriptional activation of SIVmac239. In addition, IκB-αS32/36A, a proteolysis-resistant inhibitor of NF-κB, strongly inhibited the Tat-mediated transactivation of SIVmac239. Based on this evidence, we have generated a self-regulatory virus by endowing the genome of SIV-mac239 with IκB-αS32/36A; the resulting virus, SIVIκB-αS32/36A, was nef-deleted and expressed the NF-κB inhibitor. We show that SIVIκB-αS32/36A was highly and stably attenuated both in cell cultures and in vivo in rhesus macaque as compared with a nef-deleted control virus. Moreover, the high attenuation was associated with a robust immune response as measured by SIV-specific antibody production, tetramer, and intracellular IFN-γ staining of SIV gag-specific T cells. These results underscore the crucial role of NF-κB/IκB proteins in the regulation of SIV replication both in cell cultures and in monkeys. Thus, inhibitors of NF-κB could efficiently counteract the SIV/HIV replication in vivo and may assist in developing novel approaches for AIDS vaccine and therapy.

Live attenuated simian immunodeficiency virus (SIV) 1 strains deleted of the nef region have provided a sterilizing immunity against homologous and heterologous virus challenge in rhesus macaque (1)(2)(3)(4)(5). Unfortunately, the continuous and error-prone replication of the nef-deleted virus gave rise to pathogenic escape mutants (6 -9). To improve safety, multiply deleted SIV strains lacking the accessory genes nef, vpr, vpx, vif (10), the regulatory rev and tat genes (11,12), or the leader RNA sequences (13) were generated with higher levels of attenuation. However, viral attenuation in vivo correlated inversely with the degree of the SIV-specific antibody response and protection from the subsequent viral challenge (14). Attenuation was also achieved by deleting N-linked glycosylation sites in gp120 envelope protein (15,16), suggesting a more efficient eradication of deglycosylated viruses by antibody-and T cell-mediated immune response. Nevertheless, reversion of glycosylation site mutants was observed (15), indicating that viral replication still occurred despite the efficient immune control of deglycosylated SIV. Here, we describe a gain-offunction strategy to generate a highly attenuated SIV strain endowed with a transcriptional repressor for studies in primate models. SIV and HIV-1 share a similar genome organization and modality of infection. In particular, the LTR of both retroviruses contain a TATA box, Sp1, and NF-B cis sequences (17) and transcribe for a 5Ј-untranslated region that generates the RNA stem loop binding to Tat, the viral elongation factor (18,19). NF-B plays a major role in the transcriptional activation and replication of HIV-1 (20). In fact, NF-B proteins activate the expression of HIV-1 by binding to NF-B sites of LTR (21,22) and synergize with Tat-mediated transactivation (23). Moreover, the deletion of the NF-B sites in the LTR (24) as well as the expression of NF-B inhibitors (25,26) strongly impair HIV-1 replication. In the case of SIVmac239, the NF-B-Sp1 core enhancer in the LTR was described to be either dispensable (27) or required (28,29) for efficient viral transcription and growth. In this study, we report that IB-␣S32/ 36A, a proteolysis-resistant inhibitor of NF-B (30), potently represses the expression and replication of SIVmac239 and can be inserted into the viral genome in order to generate an in vivo highly attenuated and immunogenic virus.

EXPERIMENTAL PROCEDURES
Plasmids and Gene Expression Assays-To generate pSIVLTRluc, the sequence of SIVmac239 (GenBank TM accession number M33262) from nucleotides 257-950 was amplified by PCR from the plasmid p239SpE3Ј (NIH AIDS Research & Reference Reagent Program) with the forward primer 5Ј-GCTCTAGATGGAAGGGATTTATTACAG-3Ј and the reverse primer 5Ј-CCGCTCGAGTACTTCTAAAATGGCAGCT-* This work was supported by grants from Istituto Superiore di Sanità-National Research Program on AIDS, and Ministero dell'Istruzione, dell'Università e della Ricerca. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Viral Stocks and Cell Culture Infections-Viral stocks were produced by transfecting 293 T cells with viral plasmids and evaluated by RT assay and p27 gag antigen using the SIV core antigen kit (Coulter Corp., Hialeah, FL) as previously reported (26). For cell culture infections, CEMx174 (10 5 cells) and rhesus macaque PBMC (2 ϫ 10 5 cells) stimulated with concanavalin A were infected with 10 5 and 2 ϫ 10 5 cpm RT activity of viral stocks, respectively. Viral growth was monitored by RT activity, as previously described (26). For in vitro passages, CEMx174 (10 5 cells) were infected with 10 5 cpm RT activity. At the RT peak of each infection, virus was collected and used for infections in the next passage. For analysis of nef region, viruses were collected at the RT peak of each passage and the viral RNA was analyzed by RT-PCR using the primers SIV15, 5Ј-TCTGCGACAGAGACT-3Ј (9377-9391 nucleotides of SIVmac239) and SIV16, 5Ј-TCCTGCCAATCTGGT-3Ј (9791-9805 nucleotides of SIVmac239). Titers of viral stocks for monkey infection were measured as TCID 50 by using CEMx174 cells.
Animal Infection Studies-Mamu-A*01-positive juvenile rhesus macaques were maintained in accordance with the American Association for Accreditation of Laboratory Animal Care Standards and housed in a biosafety level 2 facility. Viral infection was performed by intravenous inoculation of 100 TCID 50 of viral stocks. The infected monkeys were monitored for plasma viral load by detecting SIV RNA using a real-time reverse transcription-PCR with a threshold sensitivity of 200 viral RNA copies Eq/ml (31). The gag region was amplified by using the following primers and probe: SIV-F, 5Ј-AGTATGGGCAGCAAATGAAT-3Ј; SIV-R, 5Ј-TTCTCTTCTGCGTGAATGC-3Ј; SIV-P, 6FAM-AGATTTGGATTAG-CAGAAAGCCTGTTGGA-TAMRA. For cocultures, PBMC and lymph node cells (10 6 up to 10 7 ) from infected animals were stimulated with 10 g/ml PHA (Sigma-Aldrich) for 2 days and cultured with PHA-activated PBMC from uninfected monkeys or with CEMx174 cells. Cultures were monitored by RT activity and p27 production over 5 weeks of coculture.
To analyze the nef region, viral RNA was reverse-transcribed and amplified using Superscript one-step RT-PCR for long template (Invitrogen Life Technologies) with the primers SIV15 and SIV16 above described. The PCR product (1 l) was re-amplified with the primers SIV 17, 5Ј-GCGTGGGGAGACTTAT-3Ј (9401-9416 nucleotides of SIV-mac239) and SIV 18,5Ј-CTTGCACTGTAATAAATCCC-3Ј (9723-9742 nucleotides of SIVmac239). The PCR product was analyzed by liquid hybridization with a ␥-32 P-labeled IB-␣ probe followed by electrophoresis on 10% acrylamide gel and autoradiography. The IB-␣ probe was: 5Ј-TCCTGACCTGGTGTCACTC-3Ј. For gag analysis, viral RNA was amplified by RT-PCR with primers SIV-F and SIV-R above described, and the PCR products were detected by liquid hybridization with ␥-32 Plabeled SIV-P probe.
Attenuation of SIVIB-␣ S32/36A in Cell Cultures-Next, we generated a SIVmac239 expressing IB-␣S32/36A, SIVIB-␣S32/36A, by inserting the IB-␣S32/36A cDNA fused to the FLAG epitope into the nef region of SIVmac239 (Fig. 2a). As a control, we generated a SIVmac239 carrying the IB-␣S32/36A insert in antisense orientation (Fig. 2a). In both recombinant viruses, we deleted the nef coding region from nucleotides 9501 to 9681 of SIVmac239 leaving in place the sequence that overlaps with the env gene and codes for the first 56 amino acids of Nef (Fig. 2a). We detected the expression of viral proteins and IB-␣S32/36A by Western blotting of cell extracts upon transfection of 293 T cells with the viral plasmids (Fig. 2, b and c). As expected, IB-␣S32/36A was produced exclusively by SIVIB-␣S32/36A (Fig. 2c).
Next, we analyzed the stability of the IB-␣ insert in plasma virions and lymph node cells. At week 3 of primary infection, the IB-␣S32/36A insert was completely (WBA, 374) or partially (825, 599) lost when placed in antisense orientation (Fig.  5, lanes 1-4), while it was maintained in the viral genome when placed in sense orientation (529, 904, 540, 893) (Fig. 5,  lanes 12-15). At week 18 post-infection, virions produced in the antisense-infected animals had completely lost the IB-␣S32/ 36A antisense insert with the exception of 599 (Fig. 5, lanes  6 -9). At the same time, animals 904, 540, and 893 infected with the sense construct tested negative in this assay (Fig. 5, lanes TABLE I Virus production in PBMC and lymph nodes of infected monkeys PBMC or lymph node cells (10 6 up to 10 7 ) obtained from infected animals were cultured (1:1 ratio) with PHA-activated PBMC from uninfected monkeys for five weeks. Culture supernatants were analyzed for viral production by measuring RT activity and p27 production.  4. Attenuation of SIVIB-␣S32/36A in rhesus macaques. Animals were intravenously injected with 100 TCID 50 of SIVIB-antisense or SIVIB-␣S32/36A. The plasma viremia (left panels) and CD4 ϩ T cell counts (right panels) were measured in rhesus macaques infected with SIVIB-antisense or SIVIB-␣S32/36A as detailed under "Experimental Procedures." 18 -20) consistently with the lack of detectable plasma viremia shown in Fig. 4; virions produced in 529 were depleted of the IB-␣ insert (Fig. 5, lane 17). These findings indicate that viral attenuation correlated with the maintenance of IB-␣S32/36A expressed in sense orientation; in fact, the lack of the sense IB-␣S32/36A insert was associated with viremia in 529, while the maintenance of the antisense IB-␣S32/36A insert did not provide attenuation in 599. At 1 year post-infection, the genomic DNA from two inguinal lymph nodes of each monkey was analyzed for virus integration. In the animals infected with SIVIB-antisense, the nef region was largely deleted (Fig.  6, upper panel, lanes 1-4). In animals infected with SIVIB-␣S32/36A, the nef region was amplified as a band of the expected size; smaller bands were also detected in 529, 904, and 540 (Fig. 6, upper panel, lanes 5-8). By liquid hybridization, the IB-␣S32/36A insert was present in the nef-amplified region of animals infected with SIVIB-␣S32/36A (Fig. 6, lower  panel, lanes 5-8), while it was absent in animals infected with the control antisense virus with the exception of 599 (Fig. 6,  lower panel, lanes 1-4). These findings are consistent with the higher genomic stability of SIVIB-␣S32/36A as compared with the SIVIB-antisense control virus.
Analysis of Immune Response-Next, we analyzed the immune response elicited in infected monkeys. SIV-specific antibodies were detected at week 2 post-infection in all the animals and increased to higher levels throughout 2 years of postinfection observation (Fig. 7a). Antibody production did not correlate with the ability to control the viremia since the group of animals infected with antisense construct showed similar or even higher titers of SIV-specific antibodies as compared with the group of animals infected with SIVIB-␣S32/36A (Fig. 7a).
In further studies, we measured the number of p11C Gagspecific CD8 ϩ T lymphocytes. Over 2 years of observation, the percentages of p11C Gag-specific CD8 ϩ T lymphocytes ranged between 0.1 and 2.15 in the group of animals infected with SIV-IB␣S32/36A, and between 0.1 and 0.6 in the group of animals infected with antisense construct (Fig. 7b) with no significant difference between the two groups (p ϭ 0.14 by the Wilcoxon ranksum test). In parallel, we analyzed the frequency of CD8 ϩ T cells producing intracellular IFN-␥ in response to the p11C Gag peptide. At 2 months post-infection, a substantial population of IFN-␥ ϩ -CD8 ϩ T cells was detected in three animals (904, 540, and 893) of the group infected with the sense construct, and in one animal (WBA) of the group infected with the antisense control construct (Fig. 7c). In the following months, the population of IFN-␥ ϩ -CD8 ϩ T cells appeared at variable levels in all animals of both groups. At 24 months post-infection, all animals maintained a substantial number of IFN-␥ ϩ -CD8 ϩ T cells with the exception of animal WBA (Fig.  7c). No significant difference was observed in the number of IFN-␥ ϩ -CD8 ϩ T cells between the two groups of animals (p ϭ 0.89 by the Wilcoxon ranksum test). Consistent with previous reports in SIV-infected monkeys (34) and HIV-1-infected subjects (35), no statistically significant correlation was observed between the ability to control viremia and the levels of p11C Gag-specific T lymphocytes (p ϭ 0.11 by the Spearman rank test) or IFN-␥ ϩ -CD8 ϩ T cells (p ϭ 0.14 by the Spearman rank test).
These results show that SIV-specific antibody production and virus-specific T cell activity were induced in both groups of animals at similar levels, thus indicating that the ability to control viremia in SIVIB-␣S32/36A-infected animals was due to the poor growth ability of the virus. Of interest, a longlasting immune response was maintained in SIVIB-␣S32/ 36A-infected monkeys even in the absence of detectable viremia over 2 years post-infection. DISCUSSION SIV strains deleted of the nef gene are attenuated in vivo (6) and have been proposed as live attenuated vaccine viruses (1)(2)(3)(4)(5); however, they can reconstitute the Nef function as a consequence of rearrangements of the nef region (7-9). The genomic instability of nef-deleted live attenuated viruses has precluded their use as vaccine candidates for AIDS. To overcome these hurdles, we have developed a novel strategy of attenuation by coupling the loss of Nef with the gain of a repressor of viral expression. First, we have shown that p50/ p65 NF-B complex increased the Tat-mediated transactivation of the SIV LTR (Fig. 1b), while IB-␣ S32/36A, a proteolysis-resistant inhibitor of NF-B, potently inhibited the expression and production of SIV (Fig. 1, a and c). IB-␣ was previously shown to inhibit the NF-B-mediated expression and replication of HIV-1 (25,26,30) and to interfere with the Rev function (36). Thus, as in the case of HIV-1 (21-26, 30, 36), NF-B/IB proteins may play a major role in the regulation of SIV transcription and replication. The role of NF-B in the SIV regulation has been controversial due to studies with LTRdeleted SIV strains. In fact, the NF-B-Sp1 enhancer in SIV-mac239 LTR was shown to be required (28) and sufficient (29) for viral growth, or dispensable (27). In contrast to previous studies, we have analyzed the effect of NF-B/IB proteins on SIVmac239 expression without modifying the complex organi-zation of LTR; indeed, deletions of the LTR may create alternative regulatory elements that could be relevant in the context of a mutated virus. The redundancy of NF-B and Sp1 sites in the HIV/SIV LTR could ensure the activation of viral expression in different cell types. In this regard, NF-B and Sp1 can individually contribute to the HIV-1 promoter activity (37). Moreover, SIV strains that lack the NF-B enhancer (27,38) or the Sp1 enhancer (27) are still able to replicate efficiently. Altogether, these findings indicate that HIV/SIV may redundantly use either NF-B or Sp1 for transcription.
To evaluate the inhibition of viral growth by IB-␣, we endowed the SIVmac239 genome with IB-␣S32/36A in sense or antisense orientation. Only the sense orientation allowed the expression of IB-␣S32/36A. We observed that SIV expressing IB-␣S32/36A, SIVIB-␣S32/36A, was potently attenuated in cell cultures and in monkeys. Differently from the control antisense virus, SIV expressing IB-␣S32/36A was unable to grow in PBMC (Fig. 3a), showed 10-fold lower levels of viremia in the course of acute primary infection, and was undetectable during the following 2 years of post-acute infection (Fig. 4). Moreover, SIVIB-␣S32/36A-infected monkeys showed conserved levels of peripheral CD4 ϩ T cells with absence of clinical abnormalities over the 2 years of post-infection observation (Fig. 4). Differently, two of the four animals infected with the control SIVIBantisense (WBA and 374) experienced decreasing levels of CD4 ϩ T-cells (Fig. 4). In particular, in animal WBA the decay in CD4 ϩ T cells indicated that the SIVIB-antisense replication was persistently ongoing despite the lack of detectable plasma viremia from week 19 post-infection. Accordingly, cells from inguinal lymph nodes produced virus only in the case of SIVIB-antisense-infected animals including the monkey WBA with undetectable viremia in late infection (Table I). Thus, in the same genetic background, SIVIB-␣S32/36A showed a higher level of attenuation as compared with SIVIB-antisense. Since the two viruses differ for the ability to express IB-␣S32/36A, the stronger attenuation of SIV IB-␣S32/36A was essentially due to the IB-␣ repressor inhibiting the SIV expression.
The attenuation of SIVIB-␣S32/36A correlated with the maintenance of the IB-␣S32/36A insert as shown by serial passages in cell cultures (Fig. 3c). This pattern was also observed in monkeys, where SIVIB-␣S32/36A kept the IB-␣S32/36A insert as observed in plasma virions produced in the course of primary infection (Fig. 5) and in lymph node cells at 1 year post-infection (Fig. 6). At the same time of observation, the control SIVIB-antisense rapidly lost the IB insert (Figs. 3c, 5, and 6). These results underscore the higher genomic stability of SIVIB-␣S32/36A as compared with SIVIB-antisense and suggest that IB-␣S32/36A exerted anti-mutator activity by reducing the replicative rate of the virus.
In multiply-deleted SIV strains the degree of attenuation correlated inversely with the level of SIV-specific antibodies (14). The lack of crucial viral epitopes in multiply-deleted viruses may have enabled the induction of a broadly protective immune response (14). Differently from previous approaches, the strategy of attenuation by gain-of-repressor decreased the replication rate of SIV while preserving all of the viral proteins except Nef. Indeed, SIVIB-␣S32/36A-infected animals elicited a robust antibody and T cell-mediated immune response that was comparable to the control SIVIB-antisense (Fig. 7a). The strong immunogenicity of SIVIB-␣S32/36A represents a novel attribute of a highly attenuated virus. In fact, a highly attenuated SIVmac239 lacking nef, vpr, vpx, and upstream sequences in U3 showed a pattern of viral attenuation comparable to SIVIB-␣S32/36A (10) but was unable to raise an efficient SIV-specific antibody response (14). The effective im-FIG. 6. Analysis of virus integration in lymph node cells. The nef region was amplified from genomic DNA of lymph node cells with the primers SIV22 and SIV25 followed by nested PCR with primers SIV17 and SIV18. The viral plasmids pSIVIB-␣S32/36A and pSIVIBantisense were amplified as controls. The nef PCR product is 1147 bp in SIVIB-␣S32/36A and SIVIB-antisense (upper panel). The nef product was analyzed by liquid hybridization with the IB-␣ probe (lower panel). mune response coupled with the intrinsically slow replication of the SIVIB-␣S32/36A may have contributed to the eradication of possible escape mutants lacking the insert, as documented in monkey 529. In fact, monkey 529 that showed an occasional viremia at weeks 17 and 18 as a consequence of a deletion of the IB insert, was aviremic during the following 2 years of post-infection observation with PBMC, and lymph node cultures tested negative for viral production (Figs. 4 and 5 and Table I).
In summary, we have demonstrated that IB-␣S32/36A is a potent repressor of SIVmac239 replication. Consistent with these results, we have previously reported that an HIV-1 strain expressing IB-␣S32/36A showed a strong attenuation in primary cultures of human PBMC (26). Thus, the in vivo studies reported here may have a direct relevance for HIV-1 infection and may assist in developing new strategies of HIV-1 attenuation in vivo. Indeed, our studies underscore the in vivo relevance of NF-B/IB proteins in the regulation of SIV/HIV replication and indicate that inhibition of NF-B signaling interferes with viral progression in vivo.