NF-IL6 (C/EBPβ) induces HIV-1 replication by inhibiting cytidine deaminase APOBEC3G

Abstract

T cell activation is crucial for the productive HIV-1 infection of primary T cells; however, little is known about the host molecules involved in this process. We show that the host transcription factor NF-IL6 (also called C/EBPβ) renders primary CD4⁺ T cells highly permissive for HIV-1 replication. NF-IL6 facilitates reverse transcription of the virus by binding to and inhibiting the antiviral cytidine deaminase APOBEC3G. A mutation in NF-IL6 at Ser-288 weakened its binding to APOBEC3G and strongly inhibited HIV-1 replication. NF-IL6 also induced the replication of a Vif-deficient strain of HIV-1 in nonpermissive HUT78 cells. These data indicate that NF-IL6 is a natural inhibitor of APOBEC3G that facilitates HIV-1 replication. Host factors, such as NF-IL6, that are involved in early HIV-1 replication are potential targets for anti-HIV-1 therapy. Our findings shed light on the activation of HIV-1 replication by T cell host molecules and reveal a unique regulation of DNA deamination by APOBEC3G and NF-IL6.

A productive HIV-1 infection spontaneously occurs in many CD4⁺ T cell lines but is never observed in primary quiescent CD4⁺ T cells. In primary T cells, a productive HIV-1 infection occurs only after T cell activation. These observations suggest that, in T cell lines, host molecules important for HIV-1 replication are constitutively active but these molecules are available only after activation in primary CD4⁺ T cells (1–3). Dissecting the nature of such differences represents an opportunity to understand the biology of HIV-1 infection and may reveal potential therapeutic targets.

APOBEC3G (initially called CEM15) was identified as a host restriction factor for HIV-1 replication by a subtractive hybridization screen using cell lines that differ in their abilities to support the replication of an HIV-1 strain deficient in the virion infectivity factor (Vif) (4). APOBEC3G belongs to the APOBEC family of cytidine deaminases that edit RNA and mutate DNA. The APOBEC family consists of APOBEC1, APOBEC2, APOBEC3A-G, and activation-induced deaminase (AID) (5, 6). APOBEC3 family members inhibit infection by various viruses, including retroviruses (7) and also DNA viruses such as hepatitis B virus (HBV) (8) and adeno-associated virus (9). In the case of HIV-1, APOBEC3G catalyzes cytidine deamination in the newly synthesized, minus-strand viral DNA and induces guanosine (G)-to-adenosine (A) hypermutation in viral plus strand during reverse transcription (7, 10–12). These mutations may result in incomplete reverse transcription or lead to the production of nonfunctional viral proteins. The HIV-1 accessory protein Vif binds to APOBEC3G and blocks its antiviral function. Vif binding induces the degradation of APOBEC3G by a proteasome-mediated pathway and inhibits virion encapsidation of APOBEC3G (13–15). In addition to its activity against HIV-1, APOBEC3G inhibits infection by human T cell leukemia virus type-1 (HTLV-1) (16), murine leukemia virus (MLV) (10), HBV, and primate foamy virus (PFV) (17). These viruses do not have a Vif analog, and how these viruses overcome the antivirus function of APOBEC3G is still unclear.

As reported here, we have used a cDNA complementation screen to demonstrate that the host transcription factor NF-IL6 (also known as C/EBPβ) associates with the host antiviral cytidine deaminase APOBEC3G. This association prevents G-to-A hypermutations in HIV-1 DNA normally caused by APOBEC3G, allowing HIV-1 reverse transcription and subsequent viral replication to proceed. Our findings indicate that NF-IL6 is an inhibitor of APOBEC3G and, therefore, plays a role in regulation of DNA deamination and maintenance of genomic stability.

Results

Screening of Host Permissive Factors for HIV-1 Replication.

We used a cDNA complementation screen with a retrovirus cDNA library derived from the Jurkat CD4⁺ T cell line to identify host molecules permissive for HIV-1 replication in primary CD4⁺ T cells. We infected ≈5 × 10⁷ primary CD4⁺ T cells with a retroviral Jurkat T cell cDNA library at ≈30% infection rate. Fourteen days later, these cells were challenged with HXB-CD5, an engineered HIV-1 construct, in which mouse CD5 is fused to the nef reading frame. Seven days after HIV-1 challenge, we collected CD5-positive cells by using anti-CD5-conjugated magnetic beads and prepared total DNA from these cells. Using primers specific to constant regions flanking the cDNA insert, we amplified this DNA by using PCR, subcloned the PCR fragment into the pBMN retrovirus vector, and prepared plasmid DNA. We prepared 50 independent retrovirus DNAs and transfected these DNAs into Phoenix-Ampho to prepare recombinant retroviral supernatants for retesting. These retroviral supernatants were transduced into primary CD4⁺ T cells as in the original screen. In this retesting, we obtained 16 clones that induced HIV-1 replication in primary CD4⁺ T cells; 7 carried unique cDNA inserts. Sequence analysis indicated that it was NF-IL6 that rendered primary CD4⁺ T cells permissive for HIV-1 replication. NF-IL6 was originally identified as a transactivator of the IL-6 gene and exhibits homology to the CCAAT/enhancer binding protein (C/EBP), which belongs to the basic leucine zipper family of transcription factors (18). NF-IL6 is involved in regulation of expression of various acute-phase proteins, cytokines, and viruses (19) and has been called C/EBPβ, AGP/EBP, LAP, IL-6DBP, CRP2, and NF-M (19–21).

NF-IL6 Induces HIV-1 Replication in Primary T cells.

To characterize the function of NF-IL6 in HIV-1 replication, we measured HIV-1 replication levels in primary CD4⁺ T cells that ectopically express NF-IL6. We prepared NF-IL6-expressing CD4⁺ T cells and control CD4⁺ T cells by using a retroviral gene delivery system (pBMN-NF-IL6-IRES-Lyt2α′ or pBMN-control-IRES-Lyt2α′). Lyt2α′-expressing primary CD4⁺ T cells were selected after transduction by flow cytometry to result in cell populations that were >98% pure. Twelve days after transduction, both NF-IL6-expressing CD4⁺ T cells and control CD4⁺ T cells were challenged with the HIV-1 T-tropic strain NL4-3 and viral replication was measured by p24-based ELISA during the subsequent 18 days. As previously reported, HIV-1 replication was undetectable in control CD4⁺ T cells (22). HIV-1 replication was dramatically induced in CD4⁺ T cells expressing NF-IL6 (Fig. 1A).

Fig. 1.

NF-IL6 induces HIV-1 replication in primary T cells. (A) Primary CD4⁺ T cells transduced with either an NF-IL6-expressing (NF-IL6) or a control retrovirus (control) were challenged with HIV-1 (NL4-3; 400 TCID₅₀/10⁵ cells). P24^gag levels in culture supernatants were assayed from five wells on various days after infection, as indicated. P24^gag levels were normalized for cell numbers measured by using the XTT assay. Data represent the average (± SE) per 10⁶ cells. Similar results were observed in three independent experiments. (B) PCR amplification using the indicated primer pairs was performed by using cellular DNA from the cells in A. β-actin was used to normalize DNA levels. The asterisks indicate the position of PCR bands of interest. N.D., not detected.

To define the NF-IL6-dependent step necessary to establish a productive HIV-1 infection, we performed DNA-dependent PCR with primer sets able to distinguish between salient stages of reverse transcription. The primer pair R/U5 was designed to amplify the earliest reverse-transcription product (minus-strand strong stop DNA); primer pairs U3/5NC and R/5NC detect only full-length, double-stranded viral DNA (22, 23). Total cellular DNA was prepared 7 days after HIV-1 challenge from control CD4⁺ T cells and from CD4⁺ T cells expressing NF-IL6. Although we observed a specific band for the earliest reverse-transcription product in both the control CD4⁺ T cells and in NF-IL6-expressing CD4⁺ T cells, the U3/5NC and R/5NC primer sets generated specific bands only in the NF-IL6-expressing CD4⁺ T cells (Fig. 1B). This result indicates that NF-IL6 is critical for the production of full-length, double-stranded viral DNA in primary CD4⁺ T cells. Although several laboratories have reported that NF-IL6 can activate HIV-1 gene transcription (24–26), this is an indication that it is also involved in HIV-1 reverse transcription. Thus NF-IL6 regulates HIV-1 replication during two discrete points in the viral life cycle: during reverse transcription upon infection and during gene transcription after integration. It is possible that NF-IL6 expression triggers a degree of T cell activation sufficient to facilitate HIV-1 replication.

NF-IL6 Binds to APOBEC3G and Facilitates HIV-1 Replication.

It is likely that the cytidine deaminase activity of APOBEC3G inhibits HIV-1 replication by introducing G-to-A mutations during reverse transcription (7, 10–12). These mutations might interfere with the completion of reverse transcription or may lead to the production of nonfunctional viral proteins that block replication. Because our results indicate that NF-IL6 facilitates the production of the full-length reverse-transcription product and induces HIV-1 replication in primary CD4⁺ T cells, we hypothesized that NF-IL6 associates with and inactivates APOBEC3G. We cotransfected 293T cells with GST-NF-IL6 and APOBEC3G-HA and immunoprecipitated the cell lysates with anti-HA antibody. Western blotting of the lysates with anti-GST antibody indicated that NF-IL6 specifically associated with APOBEC3G (Fig. 2A). We confirmed that APOBEC3G does not bind to GST by using the GST-p65 fusion protein as a negative control. Because phosphorylation of NF-IL6 is important for its activity, we repeated the immunoprecipitations with APOBEC3G-HA and either NF-IL6(S220A, S221A) or NF-IL6(S288A), proteins with mutations at NF-IL6 phosphorylation sites. NF-IL6(S220A, S221A) contains alanines in place of serines in a serine-rich domain (20). In NF-IL6(S288A), a serine located in a basic domain is replaced with alanine; this mutation disrupts the protein's ability to exit the nucleus (27). We found that although APOBEC3G and NF-IL6(S220A, S221A) were specifically associated, the association between APOBEC3G and NF-IL6(S288A) was very weak (Fig. 2A). These data indicate that NF-IL6 physically associates with APOBEC3G and that the Ser-288 residue in NF-IL6 is crucial for this association.

Fig. 2.

Ser-288 of NF-IL6 is essential for the association between NF-IL6 and APOBEC3G and for optimal HIV-1 replication enhancement. (A) 293T cells were cotransfected with the indicated combinations of expression vectors: APOBEC3G-HA, GST-p65 (control), GST-NF-IL6, GST-NF-IL6(S220A, S221A), or GST-NF-IL6(S288A). Cell lysates were immunoprecipitated with anti-HA mAb and immunoblotted with anti-GST mAb. Purified GST (p65) and GST NF-IL6 are shown as controls and marked with asterisks. (B) Primary CD4⁺ T cells expressing NF-IL6 or its mutants, NF-IL6(S220A, S221A) and NF-IL6(S288A), were challenged with HIV-1 (NL4-3; 600 TCID₅₀/10⁵ cells). P24^gag levels in culture supernatants were assayed from five wells on the indicated days after infection. P24^gag levels were normalized for cell numbers measured by using the XTT assay. Data are presented as the average (±SE) per 10⁶ cells. Similar results were observed in three independent experiments.

To determine whether the association we observed between NF-IL6 and APOBEC3G affects HIV-1 replication, we used a retroviral gene delivery system [pBMN-NF-IL6-IRES-Lyt2α′, pBMN-NF-IL6(S220A, S221A)-IRES-Lyt2α′, and pBMN-NF-IL6(S288A)-IRES-Lyt2α′] to stably overexpress NF-IL6 and its mutants in primary CD4⁺ T cells. Lyt2α′-positive cells were selected by flow cytometry to create cell populations that were >98% pure. The HIV-1 T-tropic strain NL4-3 was then used to challenge the transduced and sorted cells, and HIV-1 replication levels were determined as described above. We found that HIV-1 replication levels were similar in the NF-IL6-expressing and the NF-IL6(S220A, S221A)-expressing primary CD4⁺ T cells during the 18 days after challenge. In contrast, HIV-1 replication levels were substantially lower in CD4⁺ T cells overexpressing NF-IL6(S288A), which did not bind tightly to APOBEC3G (Fig. 2B). Our results indicate that NF-IL6 facilitates HIV-1 replication by binding to APOBEC3G and interfering with its natural antiviral function. Moreover, Ser-288 in NF-IL6 plays an important role in this association and, by extension, in the activation of HIV-1 replication. NF-IL-6 does not inhibit APOBEC3G expression, as shown by a comparison of the levels of APOBEC3G expression in primary CD4⁺ T cells and in primary CD4⁺ T cells engineered to ectopically express NF-IL6 (data not shown).

NF-IL6 Is a Natural Inhibitor of APOBEC3G.

The natural antiviral defense mechanism of APOBEC3G is neutralized upon binding by the HIV-1 accessory protein Vif; the interaction with Vif directs APOBEC3G degradation and elimination via a proteasome-dependent pathway (13, 14). Vif-deficient HIV-1(Δvif HIV-1) cannot replicate in nonpermissive cells such as primary T cells, macrophages, and certain CD4⁺ T cell lines that express APOBEC3G (23, 28, 29).

We then sought to determine whether host factors like NF-IL6 can substitute for Vif and induce the replication of Δvif HIV-1 in nonpermissive cells. We therefore transduced nonpermissive HUT78 cells with NF-IL6 or mutants NF-IL6(S220A, S221A) or NF-IL6(S288A). Nontransduced HUT78 cells (HUT78) and HUT78 cells transduced with the control retrovirus (control HUT78) were used as controls. These cells were then challenged with Δvif HIV-1, and HIV-1 replication levels were determined by p24 ELISA after 8 days. Δvif HIV-1 replication levels were ∼1000-fold higher in HUT78 cells overexpressing NF-IL6 and NF-IL6(S220A, S221A) than in nontransduced HUT78 cells or HUT78 cells transduced with the control retrovirus. Furthermore, Δvif HIV-1 replication levels were ∼100-fold higher than controls, even in HUT78 cells overexpressing NF-IL6(S288A), the mutant that binds only weakly to APOBEC3G (Fig. 3A). These data indicate that NF-IL6 dramatically enhances HIV-1 replication in nonpermissive HUT78 cells in the absence of Vif.

Fig. 3.

NF-IL6 inhibits APOBEC3G activity and induces HIV-1 replication. (A) Nontransduced HUT78 cells; HUT78 transduced with a control retrovirus; and HUT78 cells expressing NF-IL6, NF-IL6(S220A, S221A), or NF-IL6(S288A) were challenged with Δvif HIV-1 (NL 4–3; 600 TCID₅₀/5 × 10⁴ cells). P24^gag levels in culture supernatants were assayed from five wells 8 days after infection. P24^gag levels were normalized for cell numbers by using the XTT assay. Data are presented as the average (±SE) per 10⁶ cells. Similar results were observed in three independent experiments. (B) Cellular DNA was prepared from the cells in A and amplified with the U3 and 5NC primer pair, and the sequences of the R, U5, and PBS regions (450–670) were analyzed. The NL4-3 sequence is shown at the top. Four sequences are shown from each sample and only the mutated nucleotides are shown. Dots indicate sequence identity.

It was reported that expression of APOBEC3G is correlated with an increase in the prevalence of G-to-A hypermutations in newly synthesized Δvif HIV-1 cDNA (15). We therefore examined whether NF-IL6 interferes with the cytidine deaminase activity of APOBEC3G by examining the extent of G-to-A hypermutation in various HUT78 cells infected with Δvif HIV-1 (Fig. 3B). We prepared cellular DNA from HUT78 and control-transduced HUT78 cells and from HUT78 cells overexpressing either NF-IL6 or its mutants 8 days after challenge with Δvif HIV-1. HIV-1 cDNA was amplified by PCR using the U3 and 5NC primer pair. We detected an increased prevalence of G-to-A mutations in the R and U5 regions, including the primer binding site (PBS), in both HUT78 and control HUT78 cells infected with Δvif HIV-1. Five of six mutation hotspots contained the sequence TGG, which is known to be targeted by APOBEC3G. In contrast, we detected almost no G-to-A mutations in Δvif HIV-1-infected HUT78 cells overexpressing NF-IL6 or its mutants (Fig. 3B). Our results indicate that NF-IL6 blocks cytidine deaminase activity of APOBEC3G by binding to APOBEC3G and facilitates complete reverse transcription of HIV-1.

Direct Effects of NF-IL6 on Single-Cell Infectivity by Δvif HIV-1.

To examine whether the induction of replication of Δvif HIV-1 is a direct or indirect effect of ectopically expressed NF-IL6 in HUT78 cells, we analyzed the expression of NF-IL6 (via the Lyt2α′ surrogate marker) and Δvif HIV-1-GFP [via the expression of green fluorescent protein (GFP)] by dual-color, multiparameter FACS. As expected, there were no Δvif HIV-1-GFP-infected (GFP-positive) cells in HUT78 cells or in control HUT78 cells not expressing NF-IL6. In contrast, Δvif HIV-1-GFP-infected (GFP-positive) cells were observed in cultures of HUT78 cells that overexpressed NF-IL6, NF-IL6(S220A, S221A), or NF-IL6(S288A) (Fig. 4). These findings confirm that HIV-1 infection correlates strongly with the expression of NF-IL6 at the single-cell level.

Fig. 4.

Single-cell analysis of Δvif HIV-1 replication in nonpermissive cells expressing NF-IL6. Flow cytometry analysis of cells ectopically expressing NF-IL6 genes (as indicated by Lty2α′ expression) and cells infected by Δvif HIV-1-GFP (as indicated by GFP expression). (A) nontransduced HUT78 cells. (B) HUT78 cells transduced with a control retrovirus. (C) HUT78 cells expressing NF-IL6. (D) NF-IL6(S220A, S221A)-expressing HUT78 cells. (E) NF-IL6(S288A)-expressing HUT78 cells were challenged with Δvif HIV-1-GFP (NL4-3; 10⁵ TCID₅₀/10⁶ cells). GFP expression was measured 5 days after infection and plotted against Lyt2α′ expression. Numbers in quadrants indicate cell percentages.

Discussion

The cytidine deaminase function of APOBEC3G reportedly results in its broad antiviral activity against HTLV-1, MLV, HBV, PFV, and others (7, 8, 10, 16, 17). To establish a productive infection, APOBEC3G must be inhibited. HIV-1 Vif binds to APOBEC3G and leads to its proteosome-dependent degradation (13, 14). This degradation eliminates APOBEC3G from the cytoplasm, facilitating accurate HIV-1 reverse transcription and enhancing HIV-1 replication. Vif is produced after the HIV-1 provirus is established in the host cell, during the translation of viral protein. Newly infected cells have negligible amounts of Vif before integration (30). Therefore, early steps of HIV replication and viruses that do not express Vif likely require host factors to overcome inhibition by APOBEC3G.

We show here that NF-IL6 inhibits the antiviral function of APOBEC3G. APOBEC3G is expressed in primary T cells and macrophages (4). Productive HIV-1 infection of primary T cells occurs only in activated T cells; HIV-1 reverse transcription terminates prematurely after synthesis of the strong stop minus-strand DNA in quiescent CD4⁺ T cell (2, 22). In contrast, macrophages host a productive HIV-1 infection regardless of their activation state (31). This variation in the relative ability of HIV-1 to infect different types of nondividing cells might be explained by the mechanism of inhibition of APOBEC3G by NF-IL6. It is likely that NF-IL6, which is constitutively expressed in macrophages (32), inhibits APOBEC3G function in these cells but not in quiescent primary T cells.

We demonstrate that phosphorylation of S288 on NF-IL6 is crucial for binding to APOBEC3G. The ectopic expression of NF-IL6(S288A), which cannot be phosphorylated Ser-288 of NF-IL6, decreased HIV-1 replication by >95% in comparison with wild-type NF-IL6 expression (Fig. 2 A and B). The mutation at position 288 in NF-IL6 disrupted transport of the protein to the nucleus (27). The mutation also inhibited its binding to DNA, and thus gene transcription usually activated by NF-IL6 did not occur (21). In comparison with wild-type NF-IL6, NF-IL6(S288A) did not significantly enhance HIV-1 reverse transcription. The mutant was unable to bind tightly to APOBEC3G. Our results indicate that the phosphorylation of Ser-288 in NF-IL6 is required for binding with APOBEC3G. This mutant only weakly induced HIV-1 replication; in the presence of the mutant NF-IL6, HIV-1 replication was inhibited at two different steps, both before and after integration: during reverse transcription and transcription. Our data suggest that NF-IL6 should be a potent anti-HIV-1 drug target. A further analysis of the interactions between APOBEC3G and either NF-IL6 or its mutants will provide important information, which may lead to the design and development of new anti-HIV-1 drugs that are based on the association of these host proteins used by HIV for its replication.

It was reported that NFAT is involved in HIV-1 activation at pre- and postintegrative steps of the HIV-1 life cycle, as is NF-IL6 (3, 22, 33). We verified that NFAT can bind to APOBEC3G, but we did not observe a specific association between them (S.M.K., unpublished data). Our result suggests that both NF-IL6 and NFAT are involved in the regulation of HIV-1 reverse transcription; however, the mechanisms by which these two host factors activate viral reverse transcription differ. Interestingly, phosphorylation is important for NF-IL6 activation (20, 27), but NFAT is activated by dephosphorylation (34). These phosphorylation and dephosphorylation events occur during T cell activation.

Single-cell analysis of Δvif HIV-1 replication indicated that NF-IL6-induced replication is a direct effect of NF-IL6 expression in nonpermissive cells (Fig. 4). For example, in HUT78 NF-IL6 cells (Fig. 4C), the NF-IL6-expressing cell population is 38.56% (38.3 + 0.26) of the population and the NF-IL6-unexpressing cell population is 61.43% (61.4 + 0.029). Thus, there are 1.59-fold (61.43/38.56) more cells that do not express NF-IL6 than cells that do. However, NF-IL6-expressing cells showed an approximately ninefold (0.26/0.029) higher infection rate with Δ-vif HIV-1-GFP than NF-IL6-unexpressing cells. These data clearly show that NF-IL6 is a host factor that allows Δvif HIV-1 replication in nonpermissive cells.

APOBEC family members function as cytidine deaminases that affect a wide range of physiological functions by editing both RNA and DNA. For example, one member of the APOBEC family, AID, acts on Ig genes to bring about the gene diversification, somatic hypermutation, and class-switch recombination steps that are important in adaptive immunity (35, 36). Although the mutations in RNA and DNA caused by cytidine deaminases are an important way to maximize the diversity of information available from a limited number of genes, overexpression of AID or APOBEC3G can cause malignancy (37, 38).

The mutation of deoxycytidine to deoxyuridine by cytidine deaminases is repaired by the base excision repair or the mismatch repair system (39, 40). These comprehensive and nonspecific DNA repair systems are found in the cells of almost every living organism, both prokaryotic and eukaryotic. It has been reported that hypermutation is caused by AID and further mutations caused during DNA repair are involved in the generation of antibody diversity (39). Interestingly, somatic hypermutation happens at a rate 10⁶-times higher than the rate of mutation in housekeeping genes even though DNA repair systems are active in these cells (40). How these radically different mutation rates are maintained is still unknown. The specific regulation mechanisms for the activity of cytidine deaminases are also not understood, despite the importance of these enzymes for genome stability. We have observed that NF-IL6 binds to AID (S.M.K., unpublished data). If NF-IL6 inhibits the cytidine deaminase activity of AID and other members of the APOBEC family, as it does that of APOBEC3G, NF-IL6 may be involved in the regulation of mutation frequency.

NF-IL6 may have evolved to negatively regulate APOBEC3G, to limit DNA mutations, and to ensure the genomic stability necessary for proper physiological function. However, this protective mechanism appears to backfire during HIV-1 infection. In this situation, inhibition of APOBEC3G by NF-IL6 actually facilitates viral reverse transcription and the ensuing replication of HIV-1. It appears that HIV-1 exploits the interaction between APOBEC3G and NF-IL6, which is critical for genomic stability, for viral activation.

Materials and Methods

Plasmids and Viruses.

The coding sequences of NF-IL6, NF-IL6(S220A, S221A), and NF-IL6(S288A) were inserted into the retroviral expression vector pBMN-IRES-Lyt2α′ by using BamHI and SalI restriction sites. pBMN-control-IRES-Lyt2α′ was used as control (22). The human APOBEC3G-HA vector was a kind gift from N.R. Landau (New York University, New York) (15). HXB-CD5 was derived from HXB2 by fusion of the mouse CD5 to the nef reading frame. Δvif HIV-1 (NL4-3) was created by the deletion of amino acids 83–170 in Vif by PCR. The deletion fragment was amplified with the following primers: 5′-ATCGCCATAGAATGGCCTAGTGTTAGGAAACTGACA-3′ and 5′-TTCTGAAATGGATAAACAGCAGTT-3′. This fragment of vif was inserted into the PflMI (nucleotide 5297) and EcoRI (nucleotide 5743) sites of pNL4-3. Δvif HIV-1-GFP was created by the insertion of GFP into the BlpI (nucleotide 8853) and XhoI (nucleotide 8887) sites of the pNL4-3 containing the vif deletion. Δvif HIV-1 and Δvif HIV-1-GFP transfected into 293T cells and levels were quantified by p24 ELISA using SupT1 cells.

Cell Culture and Primary CD4⁺ T Cell Isolation.

Human peripheral blood mononuclear cells were isolated by Ficoll gradient centrifugation. Depletions of B cells (Dynabeads CD19, DYNAL), monocytes (Dynabeads CD14, DYNAL), and CD8+ T cells (Dynabeads CD8, DYNAL) were accomplished with biomagnetic beads as described by the manufacturer. Cells were cultured in RPMI 1640 containing 10% FCS with 10 U/ml human recombinant IL-2 (Roche). HUT78 cells were cultured in RPMI 1640 containing 10% FCS.

Screening.

A cDNA library (5 × 10⁶ primary transformants) was prepared from Jurkat T cells with the pBMN vector (41). Production of and infection by retroviruses was performed as described (22). The retrovirus-transduced primary CD4⁺ T cells were cultured in RPMI 1640 containing 10% FCS with 10 U/ml human recombinant IL-2 (Roche). After 14 days, cells were challenged with HXB-CD5 (1000 TCID₅₀/10⁵ cells) in 0.5 ml of culture medium at 37°C for 16 h. Seven days after HIV-1 challenge, CD5-positive cells were collected by using CD5 Microbeads (Miltenyi Biotec) and total cellular DNA was prepared by using the QIAamp Blood Kit (Qiagen). PCR was performed by using primers designed to hybridize to constant regions flanking the cDNA insert (5′-ACGTGAAGGCTGCCGA-3′ and 5′-TAGCTTGCCAAACCTACAGGT-3′). The PCR fragment was subcloned into the pBMN-IRES-Lyt2α′ retrovirus vector, plasmid DNA was transfected into Phoenix-Ampho, and the recombinant retroviral supernatant was purified for retesting in primary CD4⁺ T cells.

Preparation of Cells Stably Expressing NF-IL6 and Its Mutants.

Two days before retrovirus infection, CD4⁺ T cells were stimulated with 2 μg/ml phytohaemagglutinin (PHA). Recombinant retroviruses were prepared as previously described (3, 22). These viruses were transduced 2–3 days after PHA stimulation by spin infection with 5 μg/ml polybrene. Lyt2α′ positive cells were isolated by flow cytometry 7 days after infection. Cells were challenged with HIV-1 14 days after PHA stimulation. During this time, primary CD4⁺ T cells grew slowly and HIV-1 could not replicate in these cells without stimulation or ectopically expressed NF-IL6. We also prepared HUT78 cells expressing NF-IL6 and its mutants as described previously (3).

HIV-1 Infection and Detection.

Cells were infected with HIV-1 by incubating cells with NL4-3 (400 TCID₅₀/10⁵ cells or 600 TCID₅₀/10⁵ cells) in 0.5 ml of culture medium at 37°C for 4 h. After HIV-1 challenge, cells were washed with culture medium and plated in five wells in a 48-well plate (1 × 10⁵ cells per well). Virus replication was measured every 3 days after HIV-1 challenge by p24 ELISA according to the manufacturer's protocol (ZeptoMetrix Corporation). P24^gag levels were normalized for the cell number measured by using the XTT assay (3). HUT78 cells expressing the gene indicated were infected with Δvif HIV-1 (600 TCID₅₀/5 × 10⁴ cells) in 0.5 ml of culture medium at 37°C for overnight. P24^gag levels in culture supernatants from five wells were measured 8 days after HIV-1 challenge. For HIV-1 infection and single-cell FACS analysis, unsorted cells expressing the genes indicated were infected overnight with Δvif HIV-1-GFP (10⁵ TCID₅₀/10⁶ cells) in 2 ml of culture medium at 37°C.

Viral DNA PCR.

Total cellular DNA from the cells indicated was prepared by using QIAamp Blood Kit (Qiagen) 7 days after HIV-1 challenge. DNA was amplified by 35 cycles of denaturation (94°C, 30 s), annealing (55°C, 30 s), and extension (72°C, 60 s) under standard conditions. Primer pairs for amplification were previously described (22, 23). PCR products were analyzed by electrophoresis in a 1.5% agarose gel with a 1-kb DNA ladder (Invitrogen) as a marker.

Immunoprecipitation and Immunoblotting.

293T cells were seeded at 1.5 × 10⁶ cells per 60-mm dish. Twenty-four hours later, cells were transfected with 3 μg of various plasmids by using the calcium phosphate coprecipitation technique. Two days later, the cells were lysed for 20 min in ice in buffer containing 10 mM CHAPS, 50 mM NaCl, 20 mM Tris (pH 7.5) and cleared by centrifugation. The lysates were immunoprecipitated with anti-HA mAb (clone HA-7, Sigma) and immunoblotted with anti-GST mAb (G1160, Sigma) and anti-mouse-HRP conjugate (A9044, Sigma). Blots were visualized with ECL Plus Western Blotting Detection System (GE Healthcare). GST fusion proteins were purified according to the manufacturer's protocol (Amersham Pharmacia).

HIV-1 DNA Sequencing.

A 220-bp fragment (nucleotides 450–670 of pNL4-3, M19921) was amplified with Pfu Ultra DNA polymerase (Stratagene) by using the R and 5NC primer pair (23). The amplicon was cloned into pCR-Blunt II-TOPO (Invitrogen) and sequenced by using the flanking M13 reverse and forward primers.