The Tat Protein of HIV-1 Induces Tumor Necrosis Factor-α Production

Human immunodeficiency virus (HIV) infection may cause a dementing illness. HIV-mediated dementia is clinically and pathologically correlated with the infiltration of activated macrophages and elevated levels of tumor necrosis factor (TNF)-α, both of which occur in an environment of small numbers of infected cells. We examined the possibility that HIV protein Tat, which is released extracellularly from infected cells, may induce the production of TNF-α. Tat induced TNF-α mRNA and protein production dose-dependently, primarily in macrophages but also in astrocytic cells. The TNF-α induction was NF-κB-dependent and could be eliminated by inhibiting protein kinase A or protein tyrosine kinase activity. In addition, Tat-induced TNF-α release was also linked to phospholipase C activation. However, Tat effects were independent of protein kinase C. These observations suggest that Tat may provide an important link between HIV and macrophage/glial cell activation and suggest new therapeutic approaches for HIV dementia.

Studies of neurological complications of HIV 1 infection such as dementia, myelopathy, and neuropathy suggest an important pathogenic role for macrophage infiltration and glial cell activation. In fact, the infection of the brain is limited compared with the severity of clinical presentation, while macrophage infiltration and activation most closely corelate with neuronal cell loss and the severity of dementia (1,2). Soluble products released from these cells have been implicated in producing neurotoxicity (3). One such substance is a cytokine called TNF-␣ (4 -7). TNF-␣ mRNA and protein levels are increased in brains of individuals with HIV infection, and the severity of dementia is closely correlated with TNF-␣ levels (8,9). Although under certain conditions, TNF-␣ may have neuroprotective properties (10,11), this molecule also has multiple damaging effects on cells in the central nervous system. For example, TNF-␣ may cause apoptosis in human neuronal cultures (12) and damage myelin (13). TNF-␣ mediated signaling events such as ceramide formation, tyrosine kinase activation, NF-B activation, calcium mobilization and release, reactive oxygen species formation (14) have been associated with neurotoxicity. Additionally, excess levels of TNF-␣ inhibit glutamate uptake by astrocytes (15). TNF-␣ may cause immune dysregulation by inducing interleukin-6, interleukin-8, and colony-stimulating factors (16) and by inducing the expression of adhesion molecules on endothelial cells (17). TNF-␣ expression, however, is not up-regulated in HIV-infected macrophages unless they are co-cultured with astrocytes, upon which high levels of TNF-␣ are released, suggesting that an intermediary factor is required for TNF-␣ production (4).
The Tat protein of HIV-1 is a potent transactivator of viral replication as well as a transactivator of some host genomes. Tat is actively released from HIV-infected cells (18,19) and may act on uninfected brain-derived cells to cause a variety of effects, including activation of NF-B (20,21) and neurotoxicity (22)(23)(24)(25). Tat may act directly on neurons to cause excitotoxicity (24,26) and cell death by apoptosis (27). Additionally, Tat may act on glial cells to cause the release of neurotoxic substances (28). This study examines this later possibility by determining the ability of Tat to induce the production of TNF-␣ in various cell types and determine the mechanisms involved during TNF-␣ induction by Tat. Previous studies have shown that TNF-␣ production is regulated by NF-B induction (29) and that Tat can induce NF-B activation in glial cells (20). Furthermore, intracerebral injection of a peptide derived from Tat into rat brain caused increased cytokine production, including TNF-␣ (30). Thus, we performed experiments to determine the extent to which Tat induces TNF-␣ production and identify mechanisms involved in TNF-␣ overexpression.

EXPERIMENTAL PROCEDURES
Cell Culture and Primary Cell Purifications-The human astrocytoma cell line U373 and monocytic cell line THP-1 were obtained from American Type Culture Collection (Rockville, MD). Human peripheral blood mononuclear cells (PBMC) were obtained from healthy volunteers, and whole PBMC and macrophage cultures were prepared as described previously (28). Human fetal astrocyte cultures were prepared from human fetal brain specimens of 12-17 weeks gestation (31). U373 cells and astrocyte cultures were maintained in minimal essential medium with heat-inactivated 10% (v/v) FBS. The PBMC and macrophages cells were cultured in complete RPMI 1640 medium with 15% FBS. THP-1 cells were cultured in RPMI medium with 10% FBS and 5.5 M ␤-mercaptoethanol. All cell types were supplemented with 100 g of streptomycin/ml and 0.25 g of amphotericin/ml. The U373 cells and astrocytes were cultured to approximately 95% confluence in 24-well plates (Ϸ1 ϫ 10 6 cells/ml). PBMC, macrophage, and THP-1 cells were cultured at a density of 2.5 ϫ 10 6 , 2.0 ϫ 10 5 , and 1.5 ϫ 10 6 cells/ml, respectively.
HIV-1 Tat Treatment-Highly purified (Ͼ95%) recombinant Tat protein was prepared from the tat gene encoding the first 72 amino acids as outlined previously (21), and the functional properties of this protein were confirmed using a transactivation assay in HL3T1 cells containing the HIV-1 long terminal repeat, chloramphenicol acetyltransferase con-struct (21). To establish dose profiles of each cell type, Tat was used at 0, 10, 100, and 1000 nM concentrations for 4 h or 1 h for U373 cells and astrocytes. For time course experiments in PBMC, macrophages, and THP-1 cells, Tat was used at 100 nM concentration, and for U373 cells and astrocytes, Tat was used at 1000 nM concentration for 0, 1, 2, 4, 6, 12, 24, and 48 h. Cells were stimulated with LPS from Escherichia coli type 055:B5 (Sigma) 1.0 g/ml as positive controls. Negative controls included mock (PBS)-treated cells and cells that were treated with either solutions from which Tat had been immunoadsorbed or Tat had been digested with trypsin followed by inactivation with a trypsin inhibitor as described previously (24,25).
RNA Extraction, RT-PCR, and Southern Blot Analysis-Total cellular RNA was prepared as described previously (32) from which first strand cDNA was synthesized as per the manufacturer's protocol (Pharmacia Biotech Inc.). PCR was conducted using published primers for TNF-␣ and ␤-actin as described previously (33). ␤-Actin primers served as internal controls in each reaction. PCR products were resolved in a 1.5% agarose gel and transferred to a nylon membrane and probed with [ 32 P]ATP end-labeled oligonucleotide probes to TNF-␣ or ␤-actin as described (34). TNF-␣ and ␤-actin oligonucleotide probes were designed based on the products amplified when using the above primers (TNF-␣, 5Ј-CAAGCTGAGGGGCAGCTCCAGTGG-3Ј; ␤-actin, 5Ј-GAGACCTTC-AACACCCCAGCCATGT-3Ј).
Quantitative Immunoassay for TNF-␣-TNF-␣ in tissue culture supernatants was determined by a sandwich ELISA (35). Briefly, microtiter plates were coated with 2 g/ml purified monoclonal antiserum to human TNF-␣ (PharMingen). A biotin-labeled mouse anti-human TNF-␣ antiserum (0.5 g/ml) directed against a different epitope (PharMingen) identified TNF-␣ in the supernatants. The assay detected TNF-␣ at concentrations as low as 5 pg/ml. Serial doubling dilutions of human recombinant TNF-␣ (1250 to 4.5 pg/ml) were used to generate a standard curve. All samples were analyzed in triplicates and expressed as mean Ϯ S.D.
Intracellular Signaling Pathway Blockers-THP-1 cells cultured in 24-well plates were pretreated for 2 h at 37°C with either 100 M TLCK; (Sigma), 20 M H7 (Calbiochem), 10 M H89 (Calbiochem), 2 M HA (Sigma), or 10 M U73122 (Calbiochem), following which they were incubated with 100 nM Tat. Separate samples were analyzed for TNF-␣ mRNA and protein. Each experiment was conducted in triplicate and repeated at least three times. Cell viability was monitored by trypan blue exclusion.

RESULTS
To determine if Tat could stimulate TNF-␣ production, we conducted initial experiments with macrophages, since previous studies have shown that macrophages are the major source of TNF-␣ and observed that Tat could induce TNF-␣ in the culture supernatants. TNF-␣ induction was dose-and time-dependent (Fig. 1, A and B), with significant increases seen at concentrations of 10 nM Tat (7.03 Ϯ 1.84 ng/ml versus 0.46 Ϯ 0.25/ng/ml; p Ͻ 0.05). TNF-␣ release occurred as early as 1 h after stimulation and peaked by 4 h (Fig. 1A).
In brain, TNF-␣ is primarily produced by tissue macrophages similar to that in blood (1,8). To determine if Tat could release TNF-␣ in astrocytes, we used highly purified human fetal astrocyte cultures. Although Tat could induce TNF-␣ in these cells, the dose and time profile was different than that in macrophages (Fig. 1, C and D). Significant (p Ͻ 0.05) increases were observed at only dosages of 1 M Tat, and the peak levels occurred within 1 h after which there was a progressive decline in TNF-␣ production (Fig. 1, C and D).
Although we used highly purified cultures of macrophages and astrocytes, we could not exclude the possibility of small amounts of other contaminating cell types in these cultures. Hence we also used a human monocytoid (THP-1) and astrocytic (U373) cell line to determine TNF-␣ induction in response to Tat. Both cell lines responded to Tat. The effect of Tat (1 M) on U373 cells (0.80 Ϯ 0.07 ng of TNF-␣/ml/10 6 cells) was comparable with that of fetal astrocytes (1.20 Ϯ 0.11 ng of TNF-␣/ ml/10 6 cells), while THP-1 cells responded to Tat 1 ⁄10 the dosage needed for the astrocyte cell lines (Fig. 1, C and E). Primary macrophages produced 6-fold more TNF-␣ (15.00 Ϯ 1.05 ng/ml/ 10 6 cells) than the THP-1 cells (2.03 Ϯ 0.23 ng/ml/10 6 cells) with 100 nM Tat (Fig. 1, B and F). The time course of TNF-␣ induction in both cell lines was similar (Fig. 1, A and E). With LPS (positive control), TNF-␣ was induced in the macrophages (18.20 Ϯ 0.89 ng/ml/10 6 cells; Fig. 2) and THP-1 cells (9.27 Ϯ 0.17 ng/ml/10 6 cells). In astrocytes and U373 cells, although there was a small induction of mRNA for TNF-␣ with LPS ( Fig.  3), levels of TNF-␣ in culture supernatants was below the level of detection for the ELISA. Tat was thus a more potent inducer of TNF-␣ protein production as compared with LPS.
All experiments were conducted with highly purified recombinant Tat protein. Yet we considered the possibility that TNF-␣ induction could be due to a contaminant in the Tat preparation. We prepared a trypsin digest of Tat and incubated it with macrophages, since they were most sensitive to induction of TNF-␣ by Tat. No response was noted in these cells (Fig.  2). Cellular response to Tat was also eliminated by immunoabsorption of Tat with Tat antisera, but not with preimmune rabbit sera, confirming the specificity of Tat action (Fig. 2). Similar results were also seen with THP-1 cells (data not shown).
To determine if the induction of TNF-␣ by Tat occurred at the level of transcription or translation, we measured TNF-␣ mRNA from THP-1 cells using semiquantitative RT-PCR. Tat induced TNF-␣ mRNA expression in a dose-and time-dependent manner similar to TNF-␣ protein detection (Fig. 3, A and  B). 10 nM Tat produced increases in TNF-␣ mRNA as early as 1 h, peaking by 4 h and returning to base line by 12 h (Fig. 3A). A small induction in of TNF-␣ mRNA was also seen in the U373 cells in response to 100 nM Tat (Fig. 3C).
Previous studies had shown that Tat induces NF-B activa- tion (20) and that NF-B regulates TNF-␣ production (29). Hence, we conducted experiments to determine if TLCK, a specific blocker of NF-B activation, could affect Tat-induced TNF-␣ expression. Pretreatment of THP-1 cells with TLCK caused complete block of TNF-␣ mRNA (Fig. 4A) and protein production (Fig. 4C). Since IB␣ is phosphorylated by protein kinases, we used specific blockers for protein kinase C (H7), protein kinase A (H89), and for tyrosine kinase (HA) to identify signaling cascades that are Tat-mediated and activate NF-B. Both H89 and HA were able to block Tat-induced TNF-␣ mRNA and protein production while H7 had no effect (Fig. 4, B and C).
To determine the role of phospholipase C in this cascade, we pretreated THP-1 cells with U73122, which also blocked the effect of Tat on TNF-␣ (Fig. 4, B and C). Treatment of the cells with Tat or any of the above pharmacological agents did not effect cell viability (data not shown). DISCUSSION We have demonstrated that the Tat protein of HIV-1 induces TNF-␣ expression in macrophages and astrocytes. Since TNF-␣ up-regulates HIV expression in infected macrophages (36,37) and glial cells (38), and there is an increased expression of TNF-␣ receptors on macrophages and microglial cells in the brains of patients with HIV infection (39), perhaps a positive feedback loop exists whereby only small numbers of infected cells may be necessary to produce substantial amounts of Tat and TNF-␣. Furthermore, since both Tat and TNF-␣ are implicated in neurotoxicity, the neurotoxic effects of Tat would be amplified by its ability to induce TNF-␣.
TNF-␣ mRNA levels are elevated in the brains of patients with HIV infection and dementia (6, 8, 40). Viral transcripts are also increased in patients with HIV dementia (40). These findings support our observations that the expression of viral transcripts, such as Tat, and cytokines, such as TNF-␣, are tightly linked and interdependent.
Tat-induced TNF-␣ production was noted in cells of monocytic lineage and in astrocytes. The macrophages, however, responded to much lower dosages of Tat and produced larger amounts of TNF-␣. Previous studies using different stimuli have also shown that TNF-␣ is predominantly produced by cells of monocytic lineage (14,41). Furthermore, brain tissue from adults with AIDS shows that TNF-␣ immunoreactivity is primarily present in microglial cells and occasionally in astrocytes (1,9). However, since astrocytes are the most numerous cell type in the brain and outnumber microglia by several logfold Tat-induced TNF-␣ production by astrocytes may be biologically significant. In fact, astrocytes can induce HIV replication in macrophages, which is dependent upon the production of TNF-␣ by astrocytes (42). This may particularly be important for the pathogenesis of HIV dementia in children, since in vitro studies have shown that virtually all human fetal astrocytes are capable of producing TNF-␣ as opposed to adult astrocytes where only a minority of cells could be induced to express TNF-␣ in response to agents such LPS and interferon ␥ (41). Since HIV dementia occurs much more frequently in children than adults, differential TNF-␣ induction by fetal and adult astrocytes may, in part, account for these differences.
Consistent with previous studies demonstrating that TNF-␣ production is regulated by NF-B (29), we observed that Tatinduced TNF-␣ production was similarly regulated. However, in contrast to previous observations where macrophage TNF-␣ production may be be regulated by protein kinase C (43), we found that inhibition of protein kinase C had no effect on Tat-induced TNF-␣ mRNA or protein production. Instead, Tatinduced TNF-␣ production was protein kinase A-and protein tyrosine kinase-dependent. Thus, the regulation of TNF-␣ gene induction by protein kinases is stimulation-specific. It is likely that these protein kinases are directly or indirectly linked to phospholipase C, since inhibition of phospholipase C also blocked Tat-induced TNF-␣ production. Although the cascade of events leading to TNF-␣ production by Tat needs to be further examined, it has been suggested that Tat-mediated inhibition of manganese-dependent superoxide dismutase may also play a role (44).
Tat is formed from two exons. The first exon contributes to the first 72 amino acids and the second to another 14 -22 amino acids. Here we show that Tat derived from the first exon (Tat1-72) is sufficient to induce TNF-␣ induction. We have previously shown that Tat1-72 is poorly internalized (21) and that a number of physiological effects of Tat1-72, such as neuronal excitation (25), increases in intracellular calcium, and NF-B induction (19), are mediated by interactions of Tat with the cell membrane. It is thus likely that Tat-induced TNF-␣ production is also a membrane-triggered event; however, further studies are necessary to determine the mechanism of TNF-␣ induction by Tat.
In conclusion, the HIV-1 protein, Tat, is an important stimulus for TNF-␣ production by macrophages and glial cells and that even transient expression of Tat may have detrimental effects on the brain of HIV-infected patients. Thus, therapeutic approaches using combined antiretroviral and anti-TNF-␣ therapy should be considered for the treatment of HIV dementia.