|Home | About | Journals | Submit | Contact Us | Français|
The human and simian immunodeficiency viruses (HIV and SIV) primarily infect lymphocytes, which must be activated for efficient viral replication. We show that the cytoplasmic domain of the transmembrane glycoprotein gp41 (gp41CD) of both HIV-1 and SIV induces activation of NF-κB, a cellular factor important for proviral genome transcription and lymphocyte activation. This NF-κB activating property localized to a region 12–25 (SIV) or 59–70 (HIV-1) residues from the gp41 membrane-spanning domain. An siRNA-based screen of 42 key NF-κB regulators revealed that gp41CD-mediated activation occurs through the canonical NF-κB pathway via TGF-β-activated kinase 1 (TAK1). TAK1 activity was required for gp41CD-mediated NF-κB activation, and HIV-1-derived gp41CD physically interacted with TAK1 through the same region required for NF-κB activation. Importantly, an NF-κB activation-deficient HIV-1 mutant exhibited increased dependence on cellular activation for replication. These findings demonstrate an evolutionarily conserved role for gp41CD in activating NF-κB to promote infection.
Members of the genus Lentivirus, family Retroviridae, have transmembrane envelope glycoproteins (TM) with unusually long cytoplasmic domains. The cytoplasmic domain of lentiviral TM is typically 150–200 amino acids in length, compared to only 30–40 amino acids found in the members of other Retroviridae genera (Evans et al., 2002). For the human and simian immunodeficiency viruses, most of these cytoplasmic domain sequences are dispensable for viral replication in certain cell types (Akari et al., 2000; Hirsch et al., 1989; Murakami and Freed, 2000b; Wilk et al., 1992; Yuste et al., 2004). Sequences in the cytoplasmic domain (CD) of the TM protein of HIV and SIV, commonly designated as gp41, have been shown to regulate the rate of endocytosis of the viral envelope complex from the surface of the cell (Berlioz-Torrent et al., 1999; LaBranche et al., 1995; Rowell et al., 1995) and to associate with the matrix protein in the process of virion assembly (Freed and Martin, 1996; Gonzalez et al., 1996; Lopez-Verges et al., 2006; Manrique et al., 2008; Murakami and Freed, 2000a). Additionally, several cellular interaction partners of gp41CD have been reported, including calmodulin (Srinivas et al., 1993), α-catenin (Kim et al., 1999), the guanine nucleotide exchange factor p115-RhoGEF (Zhang et al., 1999), the prenylated Rab acceptor PRA1 (Evans et al., 2002), the transcription factor luman (Blot et al., 2006), and, most recently, prohibitin 1 and 2 (Emerson et al., 2010). The physiological relevance of some of these interactions remains poorly understood; other functional contributions of the long gp41CD have not been described.
Nuclear factor-kappaB (NF-κB) consists of a family of transcription factors which act as central regulators in various cellular processes, including innate and adaptive immunity, cell survival, proliferation and differentiation (Oeckinghaus and Ghosh, 2009). The family includes five members, namely RelA (also referred to as p65), c-Rel, RelB, p105/p50 (NF-κB1) and p100/p52 (NF-κB2). These proteins form homo- and heterodimers, each with a specific set of target genes. NF-κB has been shown to play an important role in the replication of HIV and SIV. HIV-1 contains two NF-κB-binding elements in the transcriptional control element of its long terminal repeat region (LTR) (Nabel and Baltimore, 1987); the LTR of SIVmac features one such element (Regier and Desrosiers, 1990). While these NF-κB-binding sites are not absolutely required for viral replication due to considerable redundancy of cis-acting elements in this region (Ilyinskii and Desrosiers, 1996; Leonard et al., 1989; Ross et al., 1991), it has been demonstrated that NF-κB can drastically enhance LTR-dependent transcription (Nabel and Baltimore, 1987; Osborn et al., 1989; Perkins et al., 1993), and that stimulation of infected cells with NF-κB-activating cytokines can strongly increase replication of HIV-1 (Alcami et al., 1995; Folks et al., 1989; Griffin et al., 1989).
In this study, we demonstrate that gp41CD has the previously unrecognized capacity to induce NF-κB activation. This activation is mediated through the canonical pathway of NF-κB, with the TGF-β-activated kinase 1 (TAK1) being a critical component of this signal-transduction event. In fact, we show here that gp41CD interacts physically with TAK1, and this interaction requires the same sequences that are required for NF-κB activation. To establish the physiological relevance of these observations, we also show that a mutant of HIV-1 deficient in gp41CD-mediated NF-κB activation exhibits markedly impaired replication kinetics under conditions of suboptimal cellular stimulation and that its replication can be rescued by external stimulation of the host cell.
To isolate gp41CD from the remainder of the envelope protein (Env), expression cassettes were constructed in which full-length or truncated derivatives of HIV-1 or SIVmac239 gp41CD were fused to the external and transmembrane domain of human CD8α. When co-transfected with an NF-κB-, AP-1-, or NF-AT-luciferase reporter plasmid in HEK293T cells, the CD8 fusion constructs allowed us to conveniently and specifically measure activation of these transcription factors by gp41CD. Use of truncated derivatives of gp41CD (CD8-gp41Δ) minimized cytotoxic effects and thereby facilitated unambiguous interpretation of transcription factor reporter measurements.
The gp41CDs of three different HIV-1 strains truncated at CD amino acid 71 consistently induced NF-κB activity 35- to 80-fold depending upon the strain (Figure 1A). One of the HIV-1 gp41CDs that was used was derived from the laboratory strain NL4-3 (Adachi et al., 1986), while the other two were derived from clade B primary isolates 92HT593.1 and 92HT596.4 (Gao et al., 1996). gp41CD of SIVmac239 (Regier and Desrosiers, 1990) truncated at CD amino acid 26 (Env residue 737) consistently induced NF-κB activity approximately 20-fold above background levels (Figure 1B). Neither AP-1-nor NF-AT-dependent transcription was induced by these constructs (Supplemental Figure S1A–F). As expected, CD8-gp41 constructs with the full-length gp41CDs also were able to induce NF-κB, as was an expression-optimized construct of full-length SIVmac239 Env (Supplemental Figure S1G). The lower levels of NF-κB activation observed with the constructs featuring the complete gp41CD compared to the truncated versions are likely due at least in part to the cytopathic effects associated with expression of the full cytoplasmic domain (Chernomordik et al., 1994; Ishikawa et al., 1998; Miller et al., 1993; Srinivas et al., 1992).
To examine whether gp41CD is also capable of inducing NF-κB activation in CD4+ T cells, we transfected Jurkat JPM50.6 cells with CD8-gp41Δ constructs. This cell line is a CARD11-deficient Jurkat clone that is stably transfected with an NF-κB-dependent GFP reporter plasmid (Wang et al., 2002). Cells transfected with CD8-gp41Δ derived from HIV-1 NL4-3 (Supplemental Figure S1H, left panel) or from SIVmac239 (Supplemental Figure S1H, right panel) consistently expressed higher levels of GFP than control-transfected cells, as measured by flow cytometry. Similarly, JPM50.6 cells transfected with an expression cassette encoding full-length Env of HIV-1 NL4-3 exhibited higher levels of GFP fluorescence than cells transfected with Env truncated immediately C-terminal to the membrane-spanning domain, at residue G709 (Supplemental Figure S1I).
Progressive C-terminal deletions of 10–12 amino acids were introduced into CD8-gp41Δ constructs in order to define sequences required for the observed activation of NF-κB. Truncation of an additional 12 amino acids from the three already-truncated HIV-1 gp41CDs shown in Figure 1A led to a dramatic loss of the NF-κB-inducing activity (Figure 1C). These 12 amino acids are CLFSYHRLRDLL in all three HIV-1 strains (Figure 1E). Similarly, truncation of SIVmac239 gp41CD sequences by an additional 13 amino acids also led to the loss of the induced NF-κB activity (Figure 1D). The sequences in SIVmac239 important for the NF-κB induction by these criteria are VFSSPPSYFQQTH (Figure 1E).
Having identified these regions as required for NF-κB activation, we endeavored to identify specific amino acids of importance for this signaling activity. We therefore mutated several amino acids alone or in combination based on their biochemical properties and their potential for being part of a signaling motif (Figure 2). Tyrosine 766 in the NL4-3 sequence (the Y of YHRL) was particularly important for NF-κB activation (Figure 2A, B). The tyrosine at this position is extremely well conserved among group M, group N, and group O strains of HIV-1 and SIVcpz; it is present in 174 of 176 sequences examined (Kuiken et al., 2009). The critical amino acids tyrosine 766 and leucine 769 are contained within a sequence, YHRL, that conforms to the consensus motif for clathrin-mediated endocytosis, YXXΦ (where X describes any and Φ describes a hydrophobic amino acid) (Marks et al., 1997). However, it has been shown that this particular HIV-1 gp41CD YHRL sequence does not significantly mediate Env endocytosis (Boge et al., 1998; Rowell et al., 1995). Moreover, disruption of the established dominant endocytosis motif of gp41CD, 712YSPL715, did not affect gp41CD-mediated NF-κB activation (Figure 2A, mutant Y712A). Therefore, the sequence requirements for gp41CD-induced activation of NF-κB appear to be unrelated to clathrin-mediated endocytosis.
The motif YXXΦ also describes a common consensus sequence for SH2 binding (Machida and Mayer, 2005). SH2-binding requires the tyrosine of the consensus sequence to be phosphorylated, and mutating this tyrosine to a phenylalanine typically abrogates SH2 binding and signaling. However, mutating tyrosine 766 to a phenylalanine did not decrease NF-κB activation; in fact, it increased activation slightly (Figure 2A). These results indicate that conventional SH2 binding is not involved in the observed phenomenon. Conversely, introducing a negative charge by mutating tyrosine 766 to a glutamate or aspartate strongly reduced gp41CD-induced NF-κB activation, as did a mutation to serine (Figure 2B).
Similarly, mutating the central tyrosine of the sequence critical for NF-κB activation by SIVmac239 gp41CD, tyrosine 731, to the negatively charged glutamate reduced the observed NF-κB activation markedly (Figure 2C). Tyrosine 731 is also well-conserved across SIVmac, HIV-2, SIVmne, SIVsmm, and SIVstm, being present in 65 out of 68 sequences examined (Kuiken et al., 2009). As we saw with gp41CD of HIV-1, disruption of the membrane-proximal endocytosis motif 721YRPV724 of SIV by mutating tyrosine 721 to alanine (LaBranche et al., 1995) did not affect gp41CD-induced NF-κB activation (data not shown).
We next used a pre-validated siRNA collection to characterize the specific NF-κB pathway utilized by gp41CD. Expression levels of 42 key regulators of NF-κB were individually knocked down in HEK293T cells, after which NF-κB induction by CD8-gp41Δ was quantified in luciferase reporter assays (Figure 3 and Supplemental Figure S2). The knock-downs affected NF-κB activation by gp41CD of HIV-1 NL4-3 (Figure 3A) and SIVmac239 (Figure 3B) in a very similar pattern, indicating conservation of pathway utilization despite a lack of significant sequence similarities between the NF-κB-activating regions of gp41CD of HIV-1 and SIV (Figure 1E). Most notably, inhibition of expression of the proteins TGF-β-activated kinase 1 (TAK1), NF-κB essential modulator (NEMO or IKKγ, encoded by gene IKBKG), and of the NF-κB subunit RelA markedly reduced NF-κB activation. The specificity and efficiency of the knock-down of these three genes was verified by Western blot (Figure 3C). TAK1 is a serine/threonine kinase which phosphorylates and thereby activates the inhibitor of κB kinase (IKK) complex (Wang et al., 2001), while NEMO forms the regulatory subunit of the IKK complex (Yamaoka et al., 1998). Both TAK1 and NEMO are indispensable components of canonical NF-κB activation, but neither is involved in the non-canonical pathway. This implies that gp41CD activates NF-κB via the canonical pathway.
As the siRNA screen suggested RelA to be the dominant NF-κB subunit activated by gp41CD, we tested whether expression of CD8-gp41Δ leads to an increase in nuclear RelA capable of binding to the NF-κB consensus nucleotide sequence. Consistent with the siRNA results, an oligonucleotide pull-down demonstrated that substantially more RelA in the nuclear fraction of cells expressing CD8-gp41Δ binds to the NF-κB-binding sequence than in the nuclear fraction of cells expressing CD8 without a cytoplasmic domain (CD8-stop) or of untransfected cells (Figure 4). Importantly, the NF-κB-binding sequence of the biotinylated probe used in these experiments corresponds to the sequence of the second NF-κB-binding site in the HIV-1 LTR (Nabel and Baltimore, 1987). No RelA could be detected when a tenfold excess of unlabelled probe was added to the biotinylated NF-κB-binding probe, or when the biotinylated NF-κB-binding probe was mutated to include a single-nucleotide substitution which disrupts RelA binding.
To further confirm the central role of TAK1 suggested by the siRNA screens, the effect of the dominant-negative TAK1 mutant K63W (Yamaguchi et al., 1995) on gp41CD-mediated NF-κB activation was assessed. Indeed, overexpression of TAK1 K63W reduced NF-κB activation by ca. 80% (Figure 5A). Similarly, the chemical TAK1 inhibitor (5Z)-7-Oxozeaenol (Ninomiya-Tsuji et al., 2003) was able to markedly diminish the NF-κB activation induced by expression of CD8-gp41Δ (Figure 5B). Together, these results unambiguously corroborate the importance of TAK1 for gp41CD-mediated NF-κB activation.
Next, we sought to identify cellular partners of gp41CD responsible for the initiation of the signal transduction cascade. FLAG-tagged versions of candidate proteins were co-expressed in HEK293T cells with gp41CD derived from NL4-3 and truncated at Env residue 774, fused to mammalian glutathione-S-transferase (GST-gp41Δ L774*). No co-precipitation of the proteins retinoic acid-inducible gene-I (RIG-I), casein kinase 1α1 (CSNK1A1), tumor necrosis factor receptor-associated factor 6 (TRAF6), or tumor necrosis factor receptor type 1-associated DEATH domain (TRADD) with the gp41CD sequences was observed. However, TAK1 specifically and efficiently co-precipitated with the HIV-1 gp41CD sequences (Figure 6A). Further truncation of GST-gp41Δ by 12 amino acids to Env residue 762 resulted in complete loss of association with TAK1 (Figure 6B), precisely mirroring the loss of NF-κB activation by CD8-gp41Δ constructs with the very same truncation (Figure 1C). Specific co-precipitation of TAK1 was also observed with GST-gp41 constructs with the full-length gp41CD of NL4-3 (Figure 6C). TAK1 did not detectably co-precipitate with a GST-gp41Δ construct based on SIVmac239 (data not shown), consistent with the low similarity between the NF-κB-activating gp41CD sequences of HIV-1 and SIV (Figure 1E).
We further studied this interaction of HIV-1 gp41CD with TAK1 by examining their subcellular localization in transiently transfected HeLa cells by confocal microscopy (Supplemental Figure S3). Consistent with the specific co-precipitation of TAK1 by gp41CD, CD8-gp41Δ L774* (HIV-1 NL4-3) co-localized with TAK1, whereas this co-localization was strongly reduced in cells transfected with the non-signaling construct CD8-gp41Δ C762* (Supplemental Figure S3A). Furthermore, full-length Env of HIV-1 NL4-3 also co-localized with TAK1 (Supplemental Figure S3B).
An IL-2-dependent, CD4+ human T-cell line, WE17/10 (Willard-Gallo et al., 1990), was used to evaluate the physiological relevance of the NF-κB-activating function of gp41CD by measuring the ability of virus to replicate under conditions of suboptimal cellular stimulation. The parental HIV-1 NL4-3 virus was compared to the Env mutant Y766S. This mutation significantly impairs the ability of gp41CD sequences to activate NF-κB (Figure 2B), but it does not affect the amino-acid sequence encoded by the overlapping reading frame of the second exon of rev. The Env Y766S mutation did not significantly influence the ability of virus to infect LTR_SEAP-CEMx174 cells (Means et al., 1997) or the TZM-bl reporter cell line (Derdeyn et al., 2000) (Figure 7A, B). Furthermore, wild-type and mutant virus replicated at similar levels in fully activated, human peripheral blood mononuclear cells (PBMC) isolated from two separate individuals (Figure 7C, D). However, when cultured with a low concentration of IL-2, the replication of the Env Y766S mutant was severely impaired in WE17/10 cells, while wild-type virus replicated efficiently (Figure 7E). Importantly, this impairment of the Env Y766S mutant could be largely overcome by stimulating the cells with phytohemagglutinin (PHA) (Figure 7F), a plant lectin capable of activating lymphocytes. This result implies an important role for gp41CD-mediated NF-κB activation in overcoming the restrictions posed to viral replication by a suboptimally activated cell.
It has long been established that HIV-1 replicates optimally in T cells that are fully activated (McDougal et al., 1985; Stevenson et al., 1990; Zagury et al., 1986). The data presented here establish a role for gp41CD of both HIV-1 and SIVmac in inducing NF-κB activation, a hallmark of lymphocyte activation. For HIV-1, this induction is mediated by an interaction with TAK1, and the NF-κB-activating capacity is evidently important for the virus’ ability to replicate in cells under conditions of limited cellular activation.
TAK1 is a highly conserved key regulator of NF-κB activation. Several branches of the canonical NF-κB pathway converge on TAK1, including signal transduction cascades induced by tumor necrosis factor α (TNFα), interleukin-1β (IL-1β), lipopolysaccharide, viral RNA, cellular DNA damage, as well as complexes of antigen and major histocompatibility complex. According to the current model of canonical NF-κB activation, NF-κB dimers are retained in the cytoplasm by a family of inhibitor of κB (IκB) proteins during the inactive state. An external stimulus can induce a signal transduction cascade culminating in the recruitment of TAK1 and NEMO, the regulatory subunit of the IKK complex, into close proximity of each other through a scaffold of K63-linked polyubiquitin chains. This allows TAK1 to phosphorylate IKKβ and thereby activate the IKK complex, which then phosphorylates IκB proteins, leading to proteasomal degradation of IκB and nuclear translocation of NF-κB to activate transcription of its target genes (Adhikari et al., 2007; Liu and Chen, 2011; Wertz and Dixit, 2010). gp41CD of HIV-1 might intersect this pathway by acting as a molecular bridge between TAK1 and the IKK complex; alternatively, gp41CD might nucleate a larger signaling complex including an E3 ubiquitin ligase capable of providing K63-linked polyubiquitination. The observation that neither TRAF6 nor TRADD, both NF-κB pathway components situated “upstream” of TAK1 in the IL-1β- and TNFα-induced signaling cascades respectively, detectably co-precipitated with gp41CD lends credence to the notion that HIV-1 gp41CD circumvents the initial steps of canonical NF-κB activation and directly intersects this pathway at the level of TAK1.
Interestingly, recent reports indicate that at least two other viruses may also have evolved means of co-opting TAK1 for the purpose of NF-κB activation (Bottero et al., 2011; Wu and Sun, 2007). Although the exact regions of TAK1 interaction for these two instances have not been mapped, it is noteworthy that a remarkable sequence similarity exists between the region of HIV-1 gp41CD reported here to be crucial for NF-κB activation (762CLFSYHRLRDLL773), and a region of HHV-8 protein vGPCR, also reported to activate NF-κB via an interaction with TAK1 (264CFPYHVLNLL273) (Bottero et al., 2011). In addition, two groups have recently shown that the cellular restriction factor TRIM5, which efficiently restricts retroviral replication in a capsid-dependent manner (Stremlau et al., 2004), also interacts functionally with TAK1 and induces NF-κB activation (Pertel et al., 2011; Tareen and Emerman, 2011). Further studies will be needed to determine if there is a functional connection between the observation of TAK1 as a target of TRIM5 activity and the interaction between gp41CD and TAK1 to induce NF-κB activation which we report here. For example, gp41CD could conceivably subvert TRIM5 function by recruiting and sequestering TAK1 away from TRIM5.
While NF-κB activation by gp41CD is evolutionarily conserved between HIV-1 and SIVmac, NF-κB activation by SIVmac gp41CD is apparently not mediated by binding to TAK1, and the sequences used by SIVmac gp41CD are quite different from those used by HIV-1 gp41CD. Nonetheless, the siRNA screen reported here indicates an important role for TAK1 in SIVmac gp41CD-induced NF-κB activation. It therefore seems likely that gp41CD of SIVmac manipulates the NF-κB pathway at the level of a regulatory enzyme located “upstream” of TAK1. The exact mechanisms by which SIVmac gp41CD induces NF-κB will require further investigation.
An incoming virus brings only about 7–16 envelope trimers to a cell following fusion of the virion with the plasma cell membrane (Chertova et al., 2002; Yuste et al., 2004; Zhu et al., 2003). While this may be sufficient, it seems like a small number to induce significant levels of NF-κB activation. It is also possible that low levels of Env expression inside a resting or minimally activated T cell may induce a small degree of activation, thus augmenting the levels of viral gene expression, resulting in a positive feedback loop of steadily increasing cellular activation and viral gene expression. One of the functions of the virally encoded early gene product Nef is to increase the state of cellular activation (Alexander et al., 1997; Fenard et al., 2005; Simmons et al., 2001). Sequential expression of Nef and Env, a late gene product, may contribute to such a feedback loop of increasing NF-κB activation. It is also possible that Env may serve to complement or modify the lymphocyte activation induced by Nef at a later point in the virus’ life cycle, possibly by activating a different combination of NF-κB homo- or heterodimers and thereby a different subset of target genes. This notion is consistent with the observation that Nef induces a transcriptional profile in T cells which is remarkably similar to the pattern induced by a single activating stimulus (Simmons et al., 2001), whereas T cells typically require two independent stimuli for full activation (Janeway and Bottomly, 1994; Linsley and Ledbetter, 1993). Such a feedback mechanism might contribute significantly to the residual replication of HIV-1 observed in unstimulated lymphocytes (McDougal et al., 1985), and play an important role for viral replication in vivo, where the virus is likely to continuously encounter non- or suboptimally activated lymphocytes.
HEK293T/17 cells (ATCC, Manassas, VA), TZM-bl cells (Derdeyn et al., 2000) (NIH AIDS Research and Reference Reagent Program [NIH ARRRP]), and LTR_SEAP-CEMx174 cells (Means et al., 1997) were maintained using standard tissue-culture techniques and conditions. WE17/10 cells (Willard-Gallo et al., 1990) (NIH ARRRP) were maintained in RPMI 1640 medium supplemented with 20% Fetal Bovine Serum (FBS) (Invitrogen, Carlsbad, CA) and 50 U/ml interleukin-2 (IL-2) (Lahm and Stein, 1985) (NIH ARRRP). Note that 50 U/ml IL-2 is only half the concentration recommended for this cell line. This low concentration was chosen to maintain a state of suboptimal cellular activation, unless stimulated by the addition of PHA.
CD8-gp41 constructs were created in vector pcDNA3.1(+) (Invitrogen) by fusing the extracellular and transmembrane domain of human CD8α to the cytoplasmic-domain sequences of gp41 of SIVmac239 (Regier and Desrosiers, 1990), HIV-1 NL4-3 (Adachi et al., 1986), HIV-1 92HT593.1 or HIV-1 92HT596.4 (Gao et al., 1996) (all from NIH ARRRP). Truncations and point mutations of CD8-gp41 constructs were introduced by site-directed mutagenesis, using the QuikChange II kit (Agilent Technologies, Santa Clara, CA). An HA tag flanked by glycine linkers was inserted immediately C-terminally to the signal peptide of CD8α in CD8-g41Δ constructs. GST-gp41 constructs were created analogously, by cloning the respective gp41CD sequences into vector pEBG, which was a kind gift from Dr. Michaela Gack (Harvard Medical School). The cDNA sequences of CSNK1A1, TRAF6, TRADD, and TAK1 were obtained from Open Biosystems (Thermo Fisher Scientific, Huntsville, AL), cloned into pcDNA3.1(+) and furnished with an N-terminal FLAG tag. The pFLAG-TAK1 K63W mutant was created by site-directed mutagenesis. Sequence fidelity of every plasmid was verified by DNA sequencing (Retrogen, San Diego, CA). Expression plasmid pEF-IRES-RIG-I-FLAG was also kindly provided to us by Dr. Michaela Gack (Gack et al., 2007). The following plasmids were generously provided to us by Dr. Jae Jung (University of Southern California): pEF-IRES-STP-A, the NF-κB-luciferase, AP-1-luciferase, and NF-AT-luciferase reporter plasmids, as well as the β-galactosidase, Fyn, Lck, and Src expression plasmids (Garcia et al., 2007). The RNA-optimized SIVmac239 envelope expression vector pEEO239 was a kind gift from Dr. George Pavlakis (NCI Frederick) (Rosati et al., 2005). The F725* truncation was introduced into pEEO239 by site-directed mutagenesis.
For luciferase reporter assays, HEK293T cells were transfected with the indicated plasmids using FuGENE 6 (Roche, Indianapolis, IN). Thirty-seven h after transfection, cells were harvested and lysed, and activity of luciferase (Luciferase Assay System, Promega, Madison, WI) and β-galactosidase (Galacto-Light Plus Assay System, Applied Biosystems, Carlsbad, CA) were quantified. Each luciferase reporter experiment was performed with triplicate samples. Additionally, results given are the average of at least three independent experiments.
siRNA screens were performed using SureSilencing siRNA Arrays for Human NFkB Signaling (SABiosciences, Frederick, MD). HEK293T cells were transfected with the respective pre-coated siRNA pairs (“reverse transfection”) using SureFECT transfection reagent (SABiosciences). After 24 h, cells were co-transfected with CD8-gp41Δ, luciferase reporter plasmid, and β-galactosidase using FuGENE 6. Thirty-seven h after the second transfection, luciferase and β-galactosidase activity were quantified. Results are given as the percentage of normalized luciferase activity relative to transfection with nonsense siRNA. Values are the average of at least three independent experiments. For NEMO knock-downs, cells were transfected with IKBKG-specific Silencer siRNA 139260 (Applied Biosystems) using Dharmafect I (Thermo Fisher Scientific) and then treated analogously.
Oligonucleotide pull-down experiments were performed essentially as described previously (de Jong et al., 2010). In brief, HEK293T cells were transfected with CD8-gp41Δ L774* (HIV-1 NL4-3) or CD8-stop, using FuGENE 6. After ca. 48 h, cells were harvested and the nuclear fraction isolated with the BioVision Nuclear/Cytosol Fractionation Kit (BioVision, Mountain View, CA), according to the manufacturer’s instructions. Nuclear extracts were normalized for protein concentration and incubated on ice with 50 pmol double-stranded, 5’-biotinylated oligonucleotide probe (Sigma-Aldrich, St. Louis, MO) in Band Shift Buffer (0.5% NP-40 [Thermo Scientific, Rockford, IL]; 50 mM KCL; 50 mM Tris, pH 7.5) supplemented with 50 µg/ml poly(dI:dC) (Sigma-Aldrich) and 1 mM DTT (BioVision). In specificity controls, 50 pmol 5’-biotinylated oligonucleotide probe was supplemented with 500 pmol unbiotinylated oligonucleotide probe of the same sequence. Specificity of binding was further verified in a separate set of controls by using 50 pmol of 5’-biotinylated oligonucleotide probe with a single-nucleotide substitution which abrogates NF-κB binding. After incubation with nuclear extract, biotinylated probes were precipitated with streptavidin-coated sepharose beads (GE Healthcare, Piscataway, NJ). Relative amount of RelA protein bound to oligonucleotide probes and precipitated with them was assessed by Western blot.
For precipitation experiments, HEK293T cells were transfected with the indicated constructs, using FuGENE 6. Thirty-seven h later, cells were harvested and lysed with NP-40 lysis buffer (1% NP-40, 50 mM HEPES [Invitrogen], 150 mM NaCl) supplemented with Complete protease inhibitor cocktail (Roche). Cellular debris was pelleted by centrifugation for 25 min. at 4 °C and 1.6·104 rcf. An aliquot of supernatant was stored as whole-cell lysate sample.
For glutathione precipitation, the remaining cell lysate was incubated with Glutathione Sepharose 4B beads (GE Healthcare) for 4–6 h at 4 °C. Beads were then washed 5 times with 1 ml NP-40 lysis buffer (supplemented with Complete protease inhibitor cocktail). Finally, bound protein was eluted by resuspending the beads in Laemmli Buffer (Sigma-Aldrich) and used for Western blotting.
The following primary antibodies were used in this study: mouse anti-GST, mouse anti-FLAG, rabbit anti-FLAG (all Sigma-Aldrich), mouse anti-HA, mouse anti-NEMO, rabbit anti-β-Tubulin, rabbit anti-TAK1, rabbit anti-RelA (all Cell Signaling Technology, Danvers, MA), mouse anti-β-Actin (Abcam, Cambridge, MA), mouse anti-SP1 (Santa Cruz Biotechnology, Santa Cruz, CA). As secondary antibodies, goat anti-mouse IgG-HRP and goat anti-rabbit IgG-HRP (both Santa Cruz Biotechnology) were used. Western blots were performed using standard techniques.
The Env Y766S mutation (nucleotide A8517C) was introduced into the proviral genome of HIV-1 NL4-3 (NIH ARRRP, Genbank ID AF324493.2) (Adachi et al., 1986) by site-directed mutagenesis and verified by DNA sequencing. Virus stocks were generated as previously described (Bixby et al., 2010). Concentration of virus in the supernatant was determined by a p24 antigen-capture assay (Advanced BioScience Laboratories, Kensington, MD).
Viral infectivity was determined by infection of the reporter cell lines LTR_SEAP-CEMx174 (Means et al., 1997) and TZM-bl (Derdeyn et al., 2000), essentially as described previously (Bixby et al., 2010). Cells were infected with virus at series of two-fold dilutions. On day 3 after infection, activity of SEAP in the supernatant (Phospha-Light kit, Applied Biosystems) or luciferase (Luciferase Assay System, Promega) was quantified.
Human PBMC of healthy volunteers were isolated as described previously (Bixby et al., 2010) and infected with 10 ng p24 equivalent of either HIV-1 NL4-3 or HIV-1 NL4-3 Env Y766S. Supernatant concentrations of p24 were measured by p24 antigen-capture assay every 3–4 days.
WE17/10 cells were resuspended in fresh medium on the day of infection, either with or without 2 µg/ml PHA. Both sets of cells were then infected with 10 ng p24 equivalent of either HIV-1 NL4-3 or HIV-1 NL4-3 Env Y766S. Supernatant concentrations of p24 were measured by p24 antigen-capture assay every 3–4 days.
This work was supported by grant 5RO1AI025328 from the National Institutes of Health (NIH) to RCD, by the graduate college program GRK1071 of the Deutsche Forschungsgemeinschaft, and by a scholar stipend from GlaxoSmithKline to TSP. This work was also supported by base grant RR00168 from the NIH to the New England Primate Research Center. We are grateful to Dr. Jae Jung (University of Southern California), Dr. Michaela Gack (Harvard Medical School), and Dr. George Pavlakis (NCI Frederick) for kindly sharing their expression plasmids, and to Dr. Xin Lin (University of Texas) for use of his JPM50.6 cell line. We would also like to thank Jacqueline Bixby and Patrick Neilan for technical assistance, and Dr. Michael Farzan, Dr. Ellen Cahir-McFarland and Dr. Michaela Gack (all Harvard Medical School) for helpful advice and critical discussions. Lastly, we are also grateful to Karen Dalecki Boisvert and Carolyn Sweeney (both Harvard Medical School) for help with confocal microscopy.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
For more details, see Extended Experimental Procedures in the Supplemental Information.