HIV-1 Myristoylated Nef Treatment of Murine Microglial Cells Activates Inducible Nitric Oxide Synthase, NO2 Production and Neurotoxic Activity

Background The potential role of the human immunodeficiency virus-1 (HIV-1) accessory protein Nef in the pathogenesis of neuroAIDS is still poorly understood. Nef is a molecular adapter that influences several cellular signal transduction events and membrane trafficking. In human macrophages, Nef expression induces the production of extracellular factors (e.g. pro-inflammatory chemokines and cytokines) and the recruitment of T cells, thus favoring their infection and its own transfer to uninfected cells via exosomes, cellular protrusions or cell-to-cell contacts. Murine cells are normally not permissive for HIV-1 but, in transgenic mice, Nef is a major disease determinant. Both in human and murine macrophages, myristoylated Nef (myr+Nef) treatment has been shown to activate NF-κB, MAP kinases and interferon responsive factor 3 (IRF-3), thereby inducing tyrosine phosphorylation of signal transducers and activator of transcription (STAT)-1, STAT-2 and STAT-3 through the production of proinflammatory factors. Methodology/Principal Findings We report that treatment of BV-2 murine microglial cells with myr+Nef leads to STAT-1, -2 and -3 tyrosine phosphorylation and upregulates the expression of inducible nitric oxide synthase (iNOS) with production of nitric oxide. We provide evidence that extracellular Nef regulates iNOS expression through NF-κB activation and, at least in part, interferon-β (IFNβ) release that acts in concert with Nef. All of these effects require both myristoylation and a highly conserved acidic cluster in the viral protein. Finally, we report that Nef induces the release of neurotoxic factors in the supernatants of microglial cells. Conclusions These results suggest a potential role of extracellular Nef in promoting neuronal injury in the murine model. They also indicate a possible interplay between Nef and host factors in the pathogenesis of neuroAIDS through the production of reactive nitrogen species in microglial cells.

Introduction specific target sequences located in the inos promoter region, thus inducing the synthesis of the enzyme [13][14][15]. In contrast to neuronal and endothelial NOS (nNOS and eNOS, respectively), iNOS is constitutively active once synthesized and acts in a Ca2 + -independent manner [16,17].
HIV-1 Nef is a small (MW 27-34 kDa) myristoylated, cytoplasmic, multifunctional virulence factor acting as an adaptor molecule inside the cell. It is partially associated with the cell membrane and plays multiple roles during HIV-1 replication [18][19][20]. Nef-defective viruses lead to an attenuated clinical phenotype with reduced viral load in mouse models, monkeys, and humans [21][22][23][24][25]. More recently, it has been shown that this viral protein can be transferred to uninfected cells via cellular nanotubes, cell-to-cell contacts and release of exosomes. These findings lead to the idea that Nef is able to regulate both the endocytotic and exocytotic cell pathways thereby inducing specific effects also in non-infected cells [26].
In human monocyte-derived macrophages (MDMs), both Nef expression within the cell and cell treatment with the recombinant protein induce a pro-inflammatory response characterized by synthesis and release of specific cytokines and chemokines [27][28][29][30][31][32]. Nef-induced pro-inflammatory state in macrophages is largely due to NF-κB activation [28,[32][33][34]. In addition, we reported that Nef treatment of MDMs activates IRF-3, the main transcriptional regulator leading to the synthesis of IFNβ [32] and, ultimately, to the induction of IRF-1. Based on these two premises, we hypothesized that Nef promotes synthesis and activation of iNOS in microglial cells as a result of its pro-inflammatory properties. Consequently, iNOS-derived nitrogen reactive species might play a role in neuronal loss in a Nef-dependent manner. Due to the lack of an available in vitro system based on human-derived microglial cells, we resorted to a well characterized murine microglial cell line (i.e. BV-2, see [35]) and tested the above hypotheses by treating cells with myristoylated (myr + ) Nef and by analyzing iNOS and RNS induction, as well as the ability of RNS to induce neuronal death. Murine macrophages respond to Nef treatment as human MDMs by activating IKKα and IKKβ, JNK, and p38 MAP kinases [36]. Activation of the NF-κB pathway is mandatory for the tyrosine phosphorylation of signal transducer and activator of transcription (STAT)-1, STAT-2, and STAT-3, which is induced within 2 h in an autocrine and paracrine manner. These data confirmed that murine and human macrophages respond similarly to myr + Nef treatment [36].
Although murine cells are not permissive for HIV-1 infection, several investigators have engineered transgenic (Tg) mice to model HIV-1-induced diseases and overcome such restriction [37][38][39]. The generation of Tg mice expressing selected HIV-1 genes revealed that Nef represents a major disease determinant and that the murine system is a suitable model to investigate the mechanisms of Nef activity [37]. Nevertheless, in these studies the functions affected by Nef in monocytes/macrophages have not been extensively investigated. To the best of our knowledge, only two studies dealt with Nef expression in the brain of Tg mice: in the first one Nef was expressed in oligodendrocytes [40], whereas in the second Nef was expressed in macrophages/microglial cells but pro-inflammatory responses were not investigated [41].
In the present study, we describe that Nef treatment of murine microglial cells induces iNOS expression that requires both NF-κB activation and ex novo synthesis of IRF-1, an event dependent on IFNβ release. We also show that, similarly to other proinflammatory stimuli, such as LPS, extracellular Nef cooperates with IFNβ to induce iNOS. The myristoylation site and the acidic cluster of the viral protein are required for these effects. Finally, one or more factor(s), released in the supernatants of Nef-treated BV-2 microglial cells, induce neuronal death in a Nω-Nitro-L-arginine methyl ester (L-NAME) sensitive way.

Extracellular Nef induces STATs phosphorylation, Iκ-B degradation and IRF-1 expression in BV-2 microglial cells
Two main transcription factors are responsible for iNOS/NOS2 induction in murine as well as human phagocytic cells, i.e. IRF-1 and NF-κB. Studies of the iNOS murine promoter revealed the presence of two NF-κB and one IRF-1 binding sites [13][14][15]. We previously reported that Nef activates the NF-κB signaling pathway, and synthesis of IRF-1 both in primary human MDMs and murine macrophages [28-30, 32, 36]. Therefore, we assessed whether this was also the case in microglial cells. First, we confirmed that myr + Nef SF2 treatment induced tyrosine phosphorylation of STAT-1, -2 and -3 in BV-2 cells in a time-and dose-dependent manner. In particular, tyrosine phosphorylation signal increased starting from 2 h of cell treatment with 100 ng/ml of the viral protein and was clearly detectable after 5 h of treatment with 50 ng/ml myr + Nef SF2 (Fig 1A).
Subsequently, cells were challenged for different times (from 15 to 120 min) with myr + Nef SF2 and total cellular lysates were analyzed for the expression of I-κB-α, the alpha isoform of the inhibitor of κB. As shown in Fig 1B, myr + Nef SF2 treatment induced I-κB-α degradation that became clearly evident after 60 to 75 minutes. At later time points, protein levels started to recover, likely due to the neo-synthesis of I-κB which is itself encoded by a NF-κB regulated gene [42,43]. After 4/6 h of myr + Nef SF2 treatment the expression of IRF-1, encoded by a STAT-1 responsive gene, was clearly increased (Fig 1C). These results indicate that in murine microglial cells Nef positively affects the two transcriptional factors involved in iNOS regulation.

Nef treatment of murine microglial cells induces iNOS expression and NO 2 production
Based on the above results, we asked whether extracellular Nef was able to regulate iNOS expression. First, BV-2 cells were exposed to myr + Nef SF2 and iNOS mRNA levels were evaluated by Real Time RT-PCR from 1 to 24 h post-treatment. As shown in Fig 2A, iNOS mRNA expression reached a peak at 6 h post-treatment, declined thereafter and increased again at 16 and 24 h, similarly to the kinetics of iNOS mRNA induction in cytokines-stimulated DLD-1 cells [44]. Induction of iNOS mRNA expression was also confirmed in primary murine microglial cells (Fig 2B). To test whether iNOS expression and NO 2 production were both affected, BV-2 cells were treated for 24 h with 100 to 500 ng/ml myr + Nef SF2 . iNOS protein levels were measured in total cellular extract, whereas quantification of NO 2 released into the supernatants was performed using the Griess reagent. As depicted in Fig 2C and 2D, iNOS expression and NO 2 release were induced in a dose-dependent manner by myr + Nef treatment of microglial cells. The effect required the integrity of the protein, as heat-denaturated Nef barely induced iNOS synthesis and NO 2 release. As a positive control, we evaluated NO 2 production induced by LPS which is not sensitive to heat inactivation. To verify iNOS induction in a human context, we polarized monocytes into M1 inflammatory macrophages by cultivating them for 6 days in the presence of GM-CSF ( Fig 2E). M1 macrophages were then treated with myr + Nef in the presence of IFNγ for 24, 48 or 72 h. We were unable to detect nitrite accumulation in the supernatants of Nef-treated M1 macrophages, due to the well known low NO 2 -/ NO 3 ratio production in human cells [45], nevertheless we found that iNOS was slightly induced in cells treated for 48 h with myr + Nef plus IFNγ compared to IFNγ treatment alone ( Fig  2F and S1 File). Stimulation with LPS plus IFNγ, used as a positive control, induced iNOS at 24 and 48 h declining thereafter at 72 h ( Fig 2F).

iNOS upregulation induced by extracellular Nef requires both NF-κB and IFNβ
As previously reported, iNOS expression is regulated by both NF-κB and IRF-1. To test whether the NF-κB pathway was involved in Nef-dependent iNOS induction, BV-2 cells were treated with recombinant myr + Nef SF2 in the presence of BMS-345541, a highly specific IKKα/IKKβ inhibitor. Pre-treatment with BMS-345541 at 5 μM concentration greatly reduced Nef-dependent induction of iNOS mRNA levels ( Fig 3A) and completely inhibited iNOS expression ( Fig  3B). However, due to BMS-345541 cytotoxicity upon prolonged cell treatment, NO 2 production was evaluated in supernatant of BV-2 cells treated with Nef for 24 h in the presence of lower doses of the inhibitor (i.e. 1 and 2.5 μM). Nonetheless, these doses were still able to reduce NO 2 production in a dose-dependent manner ( Fig 3C). In human MDMs we observed that Nef treatment induces both the synthesis and the release of IFNβ followed by IRF-1 production [30,32]. We also reported that Nef treatment induced IRF-1 in RAW264.7, a murine macrophage cell line [36] and in BV-2 microglial cells ( Fig 1C). Based on these findings, we hypothesized that IFNβ blockade could affect iNOS upregulation. BV-2 cells were incubated with myr + Nef SF2 in presence of anti-IFNβ neutralizing antibodies (NAb). As shown in Fig 4, anti-IFNβ NAb started to inhibit the induction of iNOS mRNA at 8 h post-treatment ( Fig 4A). This is in agreement with the kinetics of IRF-1 protein induction, that follows Nef-induced degradation of Iκ-B (see Fig 1B). Consequently, iNOS expression was also affected by anti-IFNβ NAb starting from 6 h ( Fig 4B, compare signals at 6 h in anti-IFNβ treated and untreated cells). NO 2 detected in the culture supernatants appeared also reduced after 24 h of treatment ( Fig 4C). Taken together, these results demonstrated that Nef-induced iNOS upregulation is primarily dependent on NF-κB signalling and, to a lesser extent, on IFNβ production.

Extracellular Nef cooperates with IFNβ to induce iNOS expression
It is well recognized that stimuli that induce IRF-1 expression do not induce per se neither iNOS expression nor NO 2 production, rather, they "prime" the cells to respond to NF-κB-activating stimuli, improving their effect on iNOS regulation. This is, paradigmatically, the case of LPS and IFNγ combined treatment [46][47][48]. Therefore, we sought to test whether IFNβ has a priming effect in promoting Nef-induced iNOS expression and function. The results shown in Fig 5 demonstrate that combined treatment induced iNOS mRNA expression (Fig 5A), iNOS protein levels ( Fig 5B) and NO 2 production ( Fig 5C) to a greater extent compared to what is observed in cells exposed to myr + Nef alone.

Nef myristoylation and conserved acidic cluster are essential to induce iNOS
We previously demonstrated that Nef-mediated interference with cell signalling in human and murine macrophages treated with the viral protein required the integrity of both the N-terminal myristoylation site and the acidic cluster (AC) which consists of four glutamates [28,32,36]. To determine whether these motifs are also required for iNOS induction, BV-2 cells were treated with wild type myr + Nef SF2 , a mutant in the myristoylation site (i.e. G2A), two different preparations of a myristoylation positive, AC-mutated viral protein (4EA' and 4EA"), and a Nef-deletion mutant lacking the N-terminal anchor domain (ΔN-term). The results shown in Fig 6 indicate that all the mutants tested failed to induce iNOS mRNA expression (Fig 6A), iNOS protein (Fig 6B), and NO 2 production alone or in IFNβ combined treatments ( Fig 6C).
Conversely, other Nef mutants we tested retained their pro-inflammatory capabilities and . Total cellular RNAs were isolated and real time RT-PCR analysis was performed as reported in the materials and methods section. Results were expressed using the 2 -ΔΔCT method using basal mRNA level in untreated cells (Ctr, open circles) at T = 0 as a calibrator and GAPDH level as an internal loading control. (C, D) BV-2 cells were treated for 24 h with the indicated amounts of myr + Nef SF2 , heat inactivated myr + Nef SF2 (Inactiv. recNef), LPS or heat treated LPS. Total cell lysates were analyzed by Western Blot for iNOS (C, upper panel) levels and, as internal loading control, β-Tubulin expression (C, lower panel). (D) NO 2 content in the supernatants was quantified using the Griess colorimetric assay as reported in the materials and methods section. Dark gray bars: native myr + Nef or LPS. Light gray bars: heat pre-treated (Inactiv.) myr + Nef or heat pre-treated LPS. Results from one of five independent experiments are shown. (E) Cell phenotyping by flow cytometry of human monocytes and M1 macrophages obtained as described in materials and methods. According to [77], human monocytes were CD14 + /CD163 + /CD86 + whereas M1 macrophages were CD14 -/CD163 -/CD86 bright . (F) M1 human macrophages were left untreated or treated for 24, 48 and 72 h with IFNγ, 100 ng/ml wild type myr + Nef SF2 plus IFNγ, heat pre-treated myr + Nef SF2 and IFNγ or 100 ng/ml LPS plus IFNγ. Total cellular extracts were analyzed by Western Blot to evaluate iNOS expression using specific antibodies as described in materials and methods. β-tubulin expression as internal loading control. Blots are representative of two independent experiments. doi:10.1371/journal.pone.0130189.g002 therefore their ability to induce iNOS expression and RNS production alone or in combined treatment with IFNβ. We tested Nef mutants affecting functional motifs involved in the interaction and down-regulation of CD4 (CAWL), the interaction with SH3-containing proteins (PxxPxP) or with proteins involved in endocytic pathways such as the V1H subunit of the vacuolar membrane ATPase or the adaptor protein complexes APs (DD and LL), as well as a loop mutant lacking the C-terminal flexible loop (data not shown).These results assign the myristoylation site and the AC as essential structural motifs for Nef-mediated iNOS regulation.

Extracellular Nef induced the release of iNOS-dependent neurotoxic factors in murine microglial cells
It is thought that both viral products and cellular derived factors are mainly responsible for the neuronal loss observed in HAD [49]. To address whether Nef-dependent iNOS induction could have neurotoxic effects, the supernatant of BV-2 cells exposed to myr + Nef SF2 was collected and used to verify induction of cell death in NB41A3 murine neuroblastoma cell line. Supernatant collected from microglial cells treated for 48 h with myr + Nef, or with LPS used as positive control, induced cell death in NB41A3 cells (Fig 7C). This effect was completely abolished by the pre-treatment of BV-2 microglial cells with the NOS inhibitor L-NAME (Fig 7C), that prevented the Nef-dependent NO 2 production without affecting iNOS protein levels ( Fig   7A and 7B). Treatment of NB41A3 cells for 24 h with myr + Nef, as well as LPS or L-NAME did not induce any cytotoxic effects (Fig 7D). Again, supernatants collected from BV-2 cells treated with G2A or 4EA mutants did not cause cell death in neuroblastoma cells (Fig 7E), demonstrating that the myr + Nef neurotoxic effects also require these domains. All together, these results show that Nef-stimulated microglial cells release NOS-dependent neurotoxic factor(s), whereas Nef itself is not neurotoxic. myr + Nef SF2 with or without IFNβ (50 IU/ml). Cells were also incubated with heat-inactivated recNef SF2 (inactiv. recNef, 500 ng/ml) or, as control, with LPS or pre-heated LPS (500 ng/ml each). Total cellular lysates were analyzed by Western Blot for iNOS expression. β-Tubulin expression was used as an internal loading control. (C) Cells were treated like in (B) and NO 2 content in supernatants was measured using the Griess colorimetric assay. Empty diamond: IFNβ; filled circles: myr + Nef SF2 ; gray empty circles: heat-inactivated myr + Nef SF2 ; filled squares: recNef SF2 plus IFNβ; gray empty squares: heatinactivated myr + Nef SF2 plus IFNβ.

Discussion
Compelling evidence indicates that the neurocognitive disorders associated with the development and/or progression of AIDS are mainly due to the presence of infected macrophages/ microglial cells characterized by a pro-inflammatory M1 profile, in the central nervous system (CNS). Indeed, neuroAIDS severity is directly correlated with the levels of pro-inflammatory cytokines and chemokines, whereas no direct correlation with the viral load has been demonstrated [50,51]. Based on the theory of the "Trojan Horse" [52], HIV-1 is carried into the CNS by infected macrophages, the only immune cells able to cross the Blood Brain Barrier; the infection is then spread to other susceptible CNS-resident target cells, either competent (i.e. microglial cells) or defective (i.e. astrocytes) for HIV-1 replication [53]. The demonstrated presence of macrophage-tropic strains into the CNS further supports this hypothesis [54,55]. Even if to the best of our knowledge, there is no evidence of Nef secretion from HIV-1 infected microglial cells and/or astrocytes, it has been reported that HIV-1 infected cells release Nef into the extracellular microenviromment. Nef was detected in the medium from the MOLT-4 lymphocyte cell line transfected with a Nef-producing baculovirus vector and it was quantified in 32 patient serum samples [56]. This evidence suggests that infected cells would release Nef through a non classical secretory pathway or after lysis. Then bystander cells like uninfected microglial cells in the case of CNS, might internalize Nef via endocytosis, pinocytosis or other unknow mechanisms, as we and others already reported for MDMs, dendritic and B cells [29,[57][58][59][60][61]. Once internalized, Nef could provoke the production of ROS and RNS as an integral part of a pro-inflammatory response [62,63].
Our previous studies demonstrated that Nef treatment of human MDMs induces the proinflammatory response mainly through the activation of the NF-κB and IRF-3 transcription factors that lead to synthesis and release of IL-1β, IL-6, TNFα, CCL3/macrophage inflammatory protein 1a (MIP-1α), CCL4/MIP-1β and IFNβ [28,29,32]. Here we describe the ability of Nef to influence also the production of nitrogen reactive species through the induction of type 2 NOS, in an in vitro murine model of microglial cells exposed to the recombinant protein. The results reported in Fig 2 indicate a direct correlation between RNS production and Nef-induced upregulation of iNOS, both at the mRNA and protein level. As already reported for other canonical inducers, such as LPS and IFNγ, also in the case of Nef, iNOS induction is achieved through the activation of NF-κB and is reinforced by IRF-1 upregulation (Figs 3 and 4, respectively). Even if we cannot formally exclude the hypothesis that a surface receptor for Nef may activate a signalling cascade leading to iNOS induction, the observation that iNOS mRNA upregulation was inhibited by both cytochalasin D and 5-(N,N-dimethyl)-amyloride, two compounds which inhibit membrane ruffling and macropinocytosis [64,65] argue against this hypothesis (data not shown).
In line with previous results, we show that iNOS induction and RNS production require the N-terminal myristoylation of Nef that influences Nef localization to the cellular membrane as well as the integrity of its conserved acidic cluster (Fig 6), which was shown to be pivotal for pro-inflammatory macrophage activation through the interaction with TNF receptor adapter factor 2 [28]. Interestingly, protein alignment studies on neurotropic strain-derived Nef from post-mortem biopsies revealed that some conserved amino acidic substitutions characterized these viruses compared to patient-matched blood-derived samples [66]. Two of them are 71R/ K/T residue, in which the brain-derived viruses had arginine and lysine, whereas peripheral blood-derived HIVs had a mixture of threonine, lysine and arginine and 83G/A residue, in which brain-derived viruses displayed a glycine whereas blood-derived ones contained an alanine [66]. These residues are located immediately downstream of the Nef acidic cluster and they likely play a role in the stability and/or surface accessibility of this part of the protein that might also mediate the "pro-inflammatory potential" of the acidic cluster.
Studies performed to identify the phenotype of infected macrophages in the CNS revealed that both CD14 + /CD45 + perivascular macrophages and multinucleated giant cells represent the major source of infected cells in macaques [67], whereas CD14 -/CD45parenchymal microglial cells do not express any viral factor. From this point of view, the capacity of Nef to be transferred from infected cells to non-infected cells by exosomes and nanotubes or possibly myr + Nef SF2 (200 ng/ml) in the presence (light gray bars) or absence (dark gray bars) of L-NAME at 10 μM concentration. Supernatants were collected and NO 2 content was measured using the Griess colorimetric assay. (B) iNOS expression was evaluated on total cell lysates by Western Blot using specific anti-iNOS antibodies (upper panel) and anti-β-tubulin antibodies as an internal loading control (lower panel). (C) NB41A3 cells were incubated for 24 h with supernatants from (A). (D) NB41A3 cells were treated for 24 h with myr + Nef SF2 (200 ng/ml) or with LPS (100 ng/ml). Black line: No L-Name treatment; Gray line: L-Name (10 μM). (E) NB41A3 cells were incubated for 24h with conditioned supernatants derived from untreated BV-2 cells or treated for 48 h with 200 ng/ml of wild type (WT), G2A, 4EA', 4EA" and ΔN-terminal deleted myr + Nef SF2 proteins. In (C) to (E), cell viability was measured by flow cytometry using Propidium Iodide (PI) dye exclusion and data were expressed as percentage of PI positive cells. One out of three independent experiments is shown.
doi:10.1371/journal.pone.0130189.g007 via endocytosis as free protein (reviewed in [26]) should be emphasized. If this route of transmission is active also in the CNS, Nef-dependent iNOS induction and production of RNS might be achieved in non-infected microglial cells as well. As a consequence and regardless of the nature of infected cells in the CNS, a deregulated proinflammatory response could be induced, thereby promoting and/or accelerating the clinical onset of neurological disease [68]. Indeed, the production of matrix metalloproteinases MMP-2, -7, and -9, as well as IL-1, IL-6, TNFα, nitric oxide (NO), L-cysteine, and Ntox further disseminate immune activation in noninfected phagocytic cells.
Our results suggest a role of Nef in this macrophage-and/or microglial-based proinflammatory environment, possibly contributing to the induction of neuronal death and neuronal damage observed in the brain of AIDS patients [69]. Results obtained using the NB41A3 neuroblastoma cell line (Fig 7) demonstrated that the secreted factors induced by Nef in BV-2 cells, rather than Nef itself, are responsible for neuronal death assessed by PI uptake. These results are in conflict with reports indicating a direct neurotoxic activity of Nef due to its similarity to scorpion peptides [70]. L-NAME-dependent inhibition of neuronal death suggests that this phenomenon is dependent on NOS activity and NO production. Interestingly, NO 2 -, provided to NB41A3 cells as NaNO 2 , was not toxic per se and supernatants collected from microglia treated with myr + Nef for only 24 h were unable to induce NB41A3 cell death (data not shown). The explanation of this observation is unknown at present, as the NO 2 content in supernatants collected from BV-2 cells treated for 24 and 48 h with Nef did not differ significantly (data not shown). It is conceivable that longer incubation times are needed to allow the release of other products that in concert with NO 2 induce cell death in NB41A3 cells. In this regard, it is interesting to note that both human and murine Nef-expressing astrocytes release CXCL10/IP-10, a chemokine shown to induce neuronal death, and that its neurotoxicity is more pronounced at 48 h than at 24 h after infection [71]. Further studies will be needed to evaluate iNOS induction and production of neurotoxic factor in a bona-fide human microglial cell system upon recombinant Nef treatment and/or Nef endogenous expression.

Materials and Methods
Cells cultures, recombinant Nef preparations and reagents BV-2 cells, derived from primary murine microglial cells immortalized by transduction with vraf and v-myc expressing J2 retrovirus, were characterized before [72]. BV-2 cells were cultured in Dulbecco's modified essential medium (DMEM, Lonza, Milan, Italy) supplemented with 10% heat inactivated Fetal Calf Serum (iFCS, Lonza). NB41A3 neuroblastoma cell line [73,74] was a kind gift from Ada Maria Tata, Department of Cellular and Developmental Biology, Sapienza University of Rome, Italy and was cultured in DMEM plus 10% iFCS. Primary mouse microglia were purified as previously described [75]. In brief, mixed glial cultures were established from the forebrain of 1-day old newborn CD1-Swiss mice. After 10 days in vitro microglial cells were detached from the astroglial monolayer by gentle manual shaking of the culture flasks; the supernatants were collected and centrifuged, and the cells were then reseeded on plastic surfaces, at the density of 10 5 cells/cm 2 . After 1 h, the medium was replaced to remove non-adherent cells and microglial cells were allowed to grow for additional 24 h before starting experiments. The experimental procedures related to the use of newborn CD1 Swiss mice for establishment of primary microglial cultures have been reviewed and approved by the Italian Ministry of Health and the "Istituto Superiore di Sanitá" according to the italian law (Article 7, Legislative Decree 116/92) and Council Directive 86/609/EC. All surgery was performed under sodium pentobarbital anesthesia, and all efforts were made to minimize suffering.
Monocytes were purified as in [28]. No ethical approval from our and University "La Sapienza" ethics committees nor formal or verbal informed consent from blood donors were necessary to use buffy coats as sources of primary monocytes. Blood samples are collected for routine medical purpose and not specifically for this study. None of the authors collected the blood samples nor had any direct contact with the donors, or had access to any identifying information. M1 polarization was performed as in [76] by culturing monocytes for 6 days in RPMI, 10% FCSi supplemented with GM-CSF (Peprotech, Rocky Hill, NJ) at 50 ng/ml. According to [77], M1 phenotype was checked by flow cytometry using the following fluorchome-conjugated antibodies: CD14-FITC (clone UCHM1, Becton Dickinson, Research Triangle Park, NC), CD86-APC (clone IT2.2, BioLegend, San Diego, CA) and CD163-PE (clone GHI/ 61, BioLegend). At least ten thousand events were recorded using a FACs ARIA II sorter (Becton Dickinson) and the obtained data were analyzed using Flowing software (v.2.5.1, University of Turku, Finland).
Wild-type (WT) myristoylated Nef SF2 , the mutant in the acidic cluster (E 66 EEE 69 !AAAA), the myristoylation deficient mutant (G2A) and the mutant lacking the first 44 amminoacids (ΔN-Term), were co-expressed with a N-myristoyl-transferase expression vector in E.coli, induced in presence of myristic acid and purified as C-terminal hexahistidine-tagged fusion proteins as previously described [78]. WT as well as mutants preparations of the viral protein were checked by SDS-PAGE and Coomassie staining after purification and titered by extinction coefficient measurement. Nef preparations were analyzed for the presence of endotoxin using the chromogenic Limulus amebocyte lysate endpoint assay QCL-1000 and, if required, purified using the EndoTrap endotoxin removal system (both from Lonza). Escherichia coli 0111:B4 lipopolysaccharide (LPS), used as positive control, and L-NAME were purchased from Sigma-Aldrich (Milan, Italy).
The highly specific IKKα/β inhibitor BMS-345541 [79,80] was a kind gift from Dr. James R. Burke, Department of Immunology, Inflammation and Pulmonary Drug Discovery, Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, NJ.
Both murine IFNβ and anti-IFNβ NAb (neutralizing titer 4x10 6 UI/ml) were a kind gift from Dr. Paola Borghi, Department of Cell Biology and Neurosciences, Istituto Superiore di Sanità, Rome, Italy.

NO 2 production and quantification
NO 2 production was measured using the Griess colorimetric assay on supernatant of BV-2 microglial cells. Cells were treated in DMEM without red phenol plus 10% iFCS. Supernatants were collected, clarified by centrifugation (3,500 g, 5', 4°C) and stored at -80°C. Seventy microlitres of supernatants were incubated in duplicate for 5' at RT with 10 μl of 10 mM sulfanilamide, 10 μl of 10 mM HCl and 10 μl of 10 mM NEDA (N-1-napthylethylenediamine dihydrochloride, all from Sigma Aldrich). Samples absorbance at 550 nm was evaluated with a ELISA reader (EL800 Bio-Tek instruments, Inc., VT). Data were expressed as the percentage of maximum NO 2 release obtained by treating cells with LPS 100 ng/ml.

Neuronal cell death assay
Supernatant collected from BV-2 cells treated with wt recNef (200 ng/ml) or with G2A and 4EA Nef mutants for 48 h in presence or not of L-NAME (10 μM) were centrifugated at 3,500 g, 5', RT, filtered using a 0.22 μm filter and added to 50% confluent NB41A3 cells. Twenty four hours later cells were collected and analyzed evaluating by cytofluorimetry the percentage of propidium iodide (PI) positive cells. Ten thousand events were recorded using a DAKO Galaxy flow cytometer (Dako, Glostrup, Denmark). Data analysis was performed using Flowing software (v2.5.1, Turku Centre for Biotechnology, Finland).
Supporting Information S1 File. Western Blot densitometric analysis of MDMs treated with Nef for 48 h. Panel of Fig 1F corresponding to human MDMs treated for 48 h with IFNγ, 100 ng/ml wild type myr + Nef SF2 plus IFNγ, heat pre-treated myr + Nef SF2 and IFNγ or 100 ng/ml LPS plus IFNγ was analyzed by densitometry using ImageJ software (v.1.48). Data were normalized using β-Tubulin expression and expressed as fold of induction using the value of untreated MDMs as reference. (PDF)