Anti-Human Immunodeficiency Virus Activity of Tau Interferon in Human Macrophages: Involvement of Cellular Factors and β-Chemokines

Abstract

Tau interferon (IFN-τ) is a noncytotoxic type I IFN responsible for maternal recognition of the fetus in ruminants. IFN-τ inhibits human immunodeficiency virus (HIV) replication more strongly than human IFN-α, particularly in human monocyte-derived macrophages. In this study performed in human macrophages, IFN-τ efficiently inhibited the early steps of the biological cycle of HIV, decreasing intracellular HIV RNA and inhibiting the initiation of the reverse transcription of viral RNA into proviral DNA. Two mechanisms induced by IFN-τ treatment in macrophages may account for this inhibition: (i) the synthesis of the cellular antiviral factors such as 2′,5′-oligoadenylate synthetase/RNase L and MxA protein and (ii) an increased production of MIP-1α, MIP-1β, and RANTES, which are natural ligands of CCR5, the principal coreceptor of HIV on macrophages. Our results suggest that IFN-τ induces the same antiviral pathways in macrophages as other type I IFNs but without associated toxicity.

The interferon (IFN) system is the first line of defense against viral infection in mammals and is induced by viral infection in most types of cell, particularly monocytes and macrophages (for a review, see reference 57). By activation of the cellular Jak-STAT signaling pathway, type I IFNs induce the synthesis of three major antiviral factors (for a review, see reference 49). The first is 2′,5′-oligoadenylate synthetase (2′,5′-OAS), which is activated by the TAR sequence on human immunodeficiency virus (HIV) RNA (28) and activates the latent RNase L. The 2′,5′-OAS/RNase L pathway was found to be involved in inhibiting HIV replication by IFNs, by strategies involving 2′,5′-oligoadenylates (2-5A) (39, 47, 54) or the RNase L (29). The second factor is RNA-dependent serine-threonine kinase R (PKR), which is known to impair HIV replication in chronically infected cells (1, 35), and the third is MxA protein of the GTPase superfamily (20). No study has yet provided insights into the effects of MxA protein on HIV replication, but it may act at different levels of the virus replication cycle, as for other viruses (19, 24).

IFN-α is active against HIV infection when administered with other antiretroviral drugs to HIV-infected patients in the primary (14) and chronic (34) stages of infection, or in monotherapy to treatment-naive patients (21). In contrast to currently available anti-HIV drugs, type I IFNs act on several steps of the biological cycle of HIV, such as viral entry (61), proviral DNA synthesis (51) or protein synthesis, maturation, and budding (9, 40). Their antiviral mechanism and the lack of cross-resistance with existing antiretroviral drugs, render these compounds potentially valuable for HIV therapy. Nevertheless, IFN toxicity limits IFN-α treatment because of side effects, which may lead to low levels of compliance and treatment failure.

The trophoblast IFN-τ is found only in ruminant ungulate species and forms part of the hormonal environment required for embryonic development (26, 31). This antiluteolytic IFN structurally related to IFN-α and IFN-ω is secreted in large quantities by extraembryonic trophectoderm cells during the peri-implantation period (30, 45). In addition, IFN-τ displays antiviral and antiproliferative properties, like other IFNs (41, 42), but is less toxic in vitro and in vivo than IFN-α and IFN-β (11, 56, 58). It could therefore be considered a possible alternative to IFN-α. Interestingly, IFN-τ inhibits HIV and feline immunodeficiency virus replication (11, 43) without cytotoxicity, and its anti-HIV effects in human macrophages are stronger than those of IFN-α2a (11). Macrophages constitute a reservoir of HIV provirus in infected individuals. Preventing their infection and the production of viral particles in these cells may make it possible to decrease efficiently the spread of HIV.

In this study, we investigated the molecular mechanisms involved in the anti-HIV activity of IFN-τ in human monocyte-derived macrophages.

Human monocytes were isolated from healthy seronegative donors by countercurrent elutriation (16). Macrophages were obtained by 7-day differentiation and cultured as previously described (44). Cells were infected with the reference macrophagetropic HIV-1/Ba-L strain (17), and HIV replication was assessed in cell culture supernatants by quantifying reverse transcriptase (RT) activity (RetroSys kit; Innovagen, Lund, Sweden). Ovine recombinant IFN-τ was produced in yeast (11). Human IFN-α2b (Schering-Plough, Kenilworth, N.J.) was used as a control. Both IFNs had a specific antiviral activity of 10⁸ IU/mg against vesicular stomatitis virus in MDBK cells (25). Data were analyzed with Student's unpaired t test (Statview F 4.5, SAS Institute Inc., Cary, N.C.). Differences were considered significant if P was ≤0.01 or ≤0.05.

In in vitro-infected macrophages treated with 100 IU of IFN-τ per ml throughout culture, HIV replication was almost completely inhibited (Fig. 1A). If IFN-τ treatment was maintained for only 24 h after infection, then HIV-1 replication was decreased by 85% (Fig. 1A). If IFN-τ was administered as late as 15 days after infection, it did not decrease the rate of HIV replication (Fig. 1B). Thus, IFN-τ seemed to affect preferably early events in the biological cycle of HIV. This hypothesis was confirmed by a dose-dependent decrease in the number of HIV DNA copies in cells treated with IFN-τ (Fig. 1C). These results are consistent with the study performed by Meylan et al. (33) in HIV-1/Ba-L-infected macrophages treated with IFN-α or IFN-β before infection. Those authors demonstrated a decrease in the number of copies of the viral env gene and concluded that IFN exerted its major effects between absorption and reverse transcription in macrophages. They also found that IFNs inhibited the synthesis of HIV transcripts (i.e., late steps in the HIV cycle), but this inhibition was correlated with the effect on viral DNA and reflected the inhibition of early steps. Shirazi and Pitha found that in the CD4 T lymphocyte line CEM-174, adding IFN-α more than 22 h after infection had no further effect on virus production (50). However, on the promonocytic U1 cell line chronically infected with HIV, IFN-τ (data not shown) and IFN-α (40) inhibited the de novo synthesis of viral particles, suggesting that the nature of the inhibition depends on the sensitivity of the cell to the antiviral effects of IFNs.

FIG. 1.

Antiviral effects of IFN-τ against HIV-1/Ba-L. (A) Effects of short-term and long-term treatments with IFN-τ (100 IU/ml). Treatment was maintained throughout culture (long term) or stopped at the time of infection (short term). Cells were infected on day 0 with HIV-1/Ba-L (MOI, 0.01). Results are expressed as means ± standard deviations for three macrophage cultures. (B) Effects of IFN-τ on chronically infected macrophages. Cells were infected for 15 days before treatment with IFN-τ (100 IU/ml) or indinavir (IDV) (10 nM). RT activity was measured every 3 or 4 days, and a cumulative total for the 25 days of culture was calculated. Results are expressed as percentages of the cumulative RT activity of the positive control (means ± standard deviations). (C) Effects of IFN-τ on HIV proviral DNA synthesis. Macrophages that had been pretreated for 24 h were infected with HIV-1/Ba-L (MOI, 0.1) for 24 h. The gag (SK38-SK39) and β-globin genes were amplified by PCR.

We took interest in the reverse transcription of viral RNA to generate proviral DNA using HIV-specific primers designed to enable us to follow the progression of reverse transcription: LTR/RU5, LTR/U3-R, LTR-gag (46), and gag (SK38-SK39). In untreated macrophages, the RU5 sequence was detected as early as 1 h after infection, whereas the U3R and gag DNA sequences remained undetectable for 6 h, and the long terminal repeat (LTR)-gag sequence and integrated forms remained undetectable for 24 h (data not shown). We analyzed the various forms of HIV DNA present 24 h after infection, as previously described (46). In the untreated control, fewer copies of late than of early and intermediate forms were detected (RU5, 242 ± 48 copies per 100 cells; U3R, 235 ± 9; gag, 95 ± 1; LTR/gag, 26 ± 6) (Fig. 2A, top). The inhibitory effects of zidovudine (AZT) increased with the progression of reverse transcription (from 48% ± 7% inhibition for RU5 to 88% ± 1% for U3R, 99% ± 0.1% for gag, and 100% for LTR-gag; P versus untreated control, 0.09 [not significant {NS}], 0.0001, 0.003, and 0.004, respectively) (Fig. 2A, bottom), demonstrating the inhibition of DNA proviral strand elongation consistent with the mode of action of AZT. Significant inhibition was observed with IFN-τ and IFN-α as early as the initiation of reverse transcription: RU5 synthesis was inhibited by 65% ± 10% (P = 0.015) by IFN-τ and by 94% ± 1% (P = 0.003) by IFN-α. However, neither IFN inhibited elongation of the proviral DNA strand further, as the percent inhibition for the sequences synthesized late in the reverse transcription step was similar to that for sequences synthesized early. IFN-α seemed to be a stronger inhibitor, although only a few significant differences were found between IFN-τ and IFN-α (P = 0.02 for RU5, P = 0.03 for U3R, P is NS for gag, and P = 0.05 for LTR-gag). Interestingly, a study performed by Baca-Regen et al. on human monocytes (3) showed marked inhibition only of the intermediate and late reverse transcription intermediates of HIV_ADA and HIV_DJC strains, whereas no inhibition of RU5 synthesis was observed even at high IFN-α doses (5,000 IU/ml). In contrast, Shirazi and Pitha (51) suggested that the initiation of reverse transcription was inhibited by IFN-α in CEM-174. The molecular nature of the inhibition seems to depend on the cell type.

FIG. 2.

Reverse transcription (A) and integration (B) of HIV proviral DNA. Macrophages that had been pretreated for 24 h (IFN-τ or IFN-α [100 IU/ml] or AZT [10 μM]) were infected with HIV-1/Ba-L (MOI, 1) for 24 h. (A) Reverse transcription of HIV viral RNA to generate proviral DNA. Two dilutions of each lysate were subjected to PCR amplification, using HIV-1-specific primers. (Top) Results are expressed as follows: 100 × [(number of proviral DNA copies)/(number of β-globin gene copies/2)]. (Bottom) Results are expressed as percent inhibition with respect to the untreated control (means ± standard deviations for three amplified lysates). This experiment was repeated twice, with blood from two other donors, and similar results were obtained in each case. (B) Integrated forms of HIV-1 DNA. Untr., untreated and infected macrophages; carry over, performed on initial untreated template DNA. We diluted 50, 10, and 2 ng of each DNA sample by mixing with 1 μg of uninfected human DNA and subjected this mixture to PCR amplification (first reaction) (top). Amplified samples were diluted (1:100) and subjected to amplification. (Bottom) Signals were quantified, and results are expressed as the number of copies of integrated HIV-1 proviral DNA per 100 cells. This experiment was performed with samples from three blood donors.

As IFN-τ did not inhibit the elongation of the proviral DNA strand, we next determined whether the nuclear translocation or the integration of HIV proviral DNA was affected, by quantifying integrated proviral DNA. As HIV proviral DNA integrates into the host cell genome in a random manner, we first amplified sequences between the ubiquitous human Alu repeated sequences and the viral LTR, as previously described (4). Alu sequences are disseminated throughout the human genome, and distances between Alu sequences and integrated LTR sequences are variable, as shown by the smears in Fig. 2B (top; amplification 1, Alu-LTR). A nested PCR, performed with the products of amplification 1 (Fig. 2B; amplification 2), was specific to the HIV LTR, making it possible to amplify only integrated HIV proviral DNA forms (4). For the carryover control, the amplification 1 was performed without Taq polymerase. The absence of signal in the carryover lines for amplification 2 (Fig. 2B, top) confirms that the secondary amplification did not use initial unintegrated viral DNA as a template. At a concentration of 100 IU/ml, IFN-τ decreased the number of copies of integrated proviral DNA by 89% ± 1% (Fig. 2B, bottom), whereas only 69% ± 9% inhibition was obtained for LTR-gag forms (Fig. 2A). IFN-τ seemed to induce an additional mechanism to prevent either the nuclear import of HIV proviral DNA or the integration of proviral DNA into the host cell genome. IFN-τ and IFN-α gave similar levels of inhibition (89% ± 1% versus 86% ± 1%; NS) (Fig. 2B, bottom).

IFN-τ did not prevent the reverse transcription step. However, the inhibition observed as early as the initiation of reverse transcription (sequence RU5) may be partially due to inhibition of an earlier step in the biological cycle of HIV. We therefore investigated the intracellular accumulation of viral RNA a short time after infection.

Macrophages were treated with IFN-τ for 24 h before infection, except for the HSA-CD4 control, for which treatment and infection were simultaneous. One hour after infection, cells were treated with trypsin to eliminate extracellular viruses. Total RNA was extracted and subjected to DNase. We amplified the cDNA generated from viral RNA by PCR, using the LTR U3/R primers described above (Fig. 3, top). No signal was detected following amplification reactions in which the cDNA of macrophages infected with inactivated HIV-1/Ba-L (1 h at 60°C) was used as a template. Both IFNs significantly decreased the early accumulation of intracellular viral RNA: 73% ± 14% inhibition with IFN-τ and 66% ± 16% inhibition with IFN-α (P = 0.007 and 0.015, respectively) (Fig. 3A, bottom). Vieillard et al. (61) showed that 77% of the RT activity in human IFN-β-transformed T-cell lines remained extracellular 2 h after infection, demonstrating a block of viral entry. Thus, whatever the cell type, IFNs induce a molecular mechanism (i) preventing the entry of HIV into the host cell, (ii) leading to the degradation of viral genomic RNA, and/or (iii) inhibiting the initiation of reverse transcription.

FIG. 3.

Intracellular genomic HIV RNA 1 h after infection. Macrophages that had been pretreated for 24 h (100 IU/ml IFN-τ or IFN-α) were infected with HIV-1/Ba-L (MOI, 1) for 1 h. Three dilutions of each cDNA sample were amplified by PCR using LTR U3/R primers. The positive control was untreated and infected macrophages; the negative control was untreated and uninfected macrophages. “In vs” refers to untreated macrophages infected with inactivated virus. Results are expressed as the ratio of number of copies of viral cDNA to number of copies of GAPDH cDNA (means ± standard deviations). This experiment was performed with samples from three donors. Differences were considered significant if P was ≤0.01 (**) or ≤0.05 (*).

To investigate a possible mechanism induced by IFN-τ to block viral entry, we evaluated the expression of CD4 and CCR5, which are the receptor and major coreceptor, respectively, of HIV-1/Ba-L in macrophages (2, 13).

Macrophages treated with IFN-τ at doses of 1 to 1,000 IU/ml were incubated for 30 min at 4°C with phycoerythrin-conjugated monoclonal antibodies against CD4 (Becton Dickinson Biosciences, Mountain View, Calif.) and phycoerythrin-conjugated monoclonal antibodies against CCR5 (BD Biosciences) or their isotype-matched controls. Fluorescence was analyzed by flow cytometry with an LSR apparatus (BD Biosciences). The expression of both receptors was not affected by IFN-τ at any of the doses used (data shown for 100 IU/ml in Fig. 4A).

FIG. 4.

Effects of IFN on the expression of CD4 and CCR5 at the macrophage membrane (A) and on the synthesis of natural ligands of CCR5 by macrophages (B and C). (A) Results are expressed as the means for seven culture wells of macrophages and were generated from blood samples taken from two donors. (B) MIP-1α, MIP-1β, and RANTES were quantified by enzyme-linked immunosorbent assay in the supernatants of cells treated for 24 h with IFN-τ or IFN-α (100 IU/ml). (C) MIP-1α and RANTES production in the 6 h following a second treatment with IFN (100 IU/ml) and/or infection with HIV-1/Ba-L (MOI, 0.1). Similar results were obtained for two other blood donors. Differences were considered significant if P was ≤0.01 (**) or ≤0.05 (*).

Several studies have demonstrated the antiviral effects of β-chemokines, which inhibit HIV entry into macrophages by occupying the coreceptor CCR5 (7, 8, 60). We therefore quantified the synthesis of MIP-1α, MIP-1β, and RANTES in cell culture supernatants of macrophages treated with 100 IU of IFN-τ or IFN-α per ml for 24 h using Quantikine immunoassay kits (R&D Systems, Abingdon, United Kingdom). The results were as follows: 276 ± 51 pg/ml for MIP-1α (P = 0.007 versus control [93 ± 2 pg/ml]), 327 ± 42 pg/ml for MIP-1β (P = 0.006 versus control [133 ± 29 pg/ml]), and 15.4 ± 2.5 pg/ml for RANTES (P = 0.017 versus control [8.2 ± 0.3 pg/ml]) (Fig. 4B). IFN-α induced MIP-1α production significantly more strongly than did IFN-τ (482 ± 31 versus 276 ± 51 pg/ml; P = 0.008). At the end of this first incubation for 24 h, cells were re-treated and infected or not infected with HIV-1/Ba-L (multiplicity of infection [MOI], 0.1) to evaluate the effect of infection on the production of β-chemokines. Regardless of infection status, only RANTES levels remained increased after 6 h of treatment for both IFNs (8.3 ± 0.5 pg/ml [IFN-τ; P = 0.02] and 12.6 ± 2.6 pg/ml [IFN-α; P = 0.0329] versus 6.4 ± 0.4 pg/ml for controls) (Fig. 4C). In contrast, no significant increase in levels of MIP-1α (Fig. 4C) or MIP-1β (data not shown) was detected following IFN-τ treatment. IFN-α treatment led to a decrease in the production of MIP-1α (17 ± 3 pg/ml [IFN-α] versus 57 ± 3 pg/ml [control]; P < 0.0001). Early HIV infection did not affect the production of RANTES (Fig. 4C) in untreated or IFN-treated macrophages. MIP-1α levels in infected macrophages were not affected by IFN-τ treatment, whereas they decreased in response to IFN-α treatment, as observed for uninfected cells (values for infected cells: 38 ± 1 pg/ml [IFN-α] versus 66 ± 1 pg/ml [control]; P < 0.0001).

IFNs are known to inhibit T-tropic HIV infection of lymphocytes by downmodulating CXCR4 expression on the lymphocyte surface (52). However, the differentiation of monocytes into macrophages in vitro is associated with the increase in expression of CCR5 parallel to the decrease in CXCR4 (15, 59), and CXCR4 is undetectable on our mature macrophages. Moreover, HIV-1/Ba-L is an R5-tropic strain and uses CCR5 to enter target cells (5). Cremer et al. (10) demonstrated an increase in RANTES production, correlated with CCR5 downregulation, in IFN-β-transduced-macrophages derived from CD34⁺ cells from umbilical cord blood. Other studies showed that RANTES strongly downregulates CCR5 on monocytes and lymphocytes (27) but not on macrophages differentiated under experimental conditions similar to our model (without added growth factors), even at higher doses (62). The inhibitory effect of RANTES on HIV infection of differentiated macrophages could also occur mainly by preventing virus interaction with CCR5 (53, 55). However, based on the results of previous studies, 100 ng/ml seems to be the minimum concentration of exogenous recombinant β-chemokines required for HIV replication on entry into macrophages to be strongly limited (7, 23). In our experimental system, a treatment with 100 ng/ml recombinant MIP-1α, MIP-1β, and RANTES efficiently inhibited HIV replication (data not shown). In parallel, HIV replication was not increased when the natural endogenous productions of β-chemokines were neutralized in cell culture supernatants (data not shown). Thus, the concentrations of β-chemokines released by macrophages here were unlikely to have strong antiviral activity.

Other mechanisms could account for the inhibition of HIV replication, such as the induction of the cellular antiviral factors. IFN-τ induces 2′,5′-OAS (22) and MxA protein (38) in the physiological environment of pregnancy. In human macrophages, these antiviral factors mRNA were produced in small amounts in untreated cells (Fig. 5A). After treatment for 24 h with IFN-τ, we detected high levels of mRNA for 2′,5′-OAS (the 40-kDa [P < 0.0001] and 69-kDa forms), MxA protein (P < 0.001), and PKR (P < 0.001) (Fig. 5B). No significant difference in induction was observed between IFN-τ and IFN-α (Fig. 5B). The induction of these antiviral proteins was confirmed by Western blotting using specific antibodies against PKR (clone K-17; Santa Cruz Biotechnology, Santa Cruz, Calif.) and MxA protein (generously provided by M. A. Horisberger, Novartis, Basel, Switzerland) (Fig. 5C).

FIG. 5.

Induction by IFN of the cellular antiviral factors MxA protein, 2′,5′-OAS, and PKR, showing mRNA (A and B) and protein (C) levels. Macrophages were treated for 24 h with 100 IU of IFN per ml for total RNA extraction or for 48 h for Western blotting. (A) Two dilutions of each cDNA sample were specifically amplified in triplicate by PCR for each of the three cellular antiviral factors. (B) mRNA levels were normalized on the basis of GAPDH mRNA levels. Results are representative of those obtained with samples from three blood donors.

Thus, in our macrophages treated for 24 h with IFN-τ, the levels of 2′,5′-OAS were increased at the time of infection. In many viral infections, the synthesis of 2-5A by 2′,5′-OAS starts rapidly after infection thanks to intracellular viral double-stranded RNA. Nanomolar concentrations of 2-5A were detectable in HeLa cells infected with reovirus as soon as 2 h after infection (36). The levels of 2-5A increased up to 6 h postinfection but declined afterwards. In parallel, a nuclease activity could be detected in these cells, which could be accounted for by RNase L (36). In the case of HIV infection, the viral TAR RNA sequence with a stem-bulge-loop structure produced a six-fold stimulation of 2′,5′-OAS activity in an in vitro 2-h-incubation assay (28). First, 2-5A is known to inhibit the initiation of the reverse transcription by preventing RT-primer complex formation (54) and may take part in the inhibition of the synthesis of RU5 sequence observed in our study. Second, 2-5A binds to and activates the latent endonuclease RNase L known to inhibit HIV replication (29). This endoribonuclease cleaves RNA at the 3′ side of UpNp sequences, which are very frequent in the HIV U3R DNA sequence. One hour may be sufficient to have an endonuclease activity leading to a beginning of degradation of HIV RNA. Thus, the activation of the 2-5A pathway, which interferes with the early steps in the HIV cycle, is consistent with the effects observed in this study at the molecular level in the biological cycle of HIV.

However, the inhibitory effect of the RNase L on HIV infection seems transient. 2-5A levels and RNase L activity were shown to increase during first days of infection and to decrease afterward (32, 48) to return under its basal level as early as 3 days postinfection (32). This may be correlated with the induction of the RNase L inhibitor (32). An increase in levels of RNase L inhibitor mRNA was observed from day 7 postinfection in our macrophages (data not shown), consistent with the absence of inhibition of HIV replication by IFN-τ when treatment was administered 15 days after infection. PKR is known to impair HIV replication in chronically infected cells (1, 35). However, HIV has also developed various mechanisms to circumvent this problem, including, in particular, a Tat/PKR interaction (6). As observed for La Crosse virus, MxA protein may interact with viral or cellular protein from reverse transcription complexes (24), preventing the normal initiation of this step, or it may block nuclear transport, as seen for hepatitis B virus (19).

The lack of major differences in the antiviral mechanisms induced by these two types of IFNs could be surprising, as IFNτ and IFN-α recognize the type I receptor differently (58). However, despite having a lower binding affinity for the receptor, IFN-τ was found to have specific antiviral properties similar to those of IFN-α in MDBK cells (58). Low-affinity interaction with the receptor is sufficient for antiviral responses, whereas a higher-affinity interaction was required for growth inhibition (18). Thus, the difference in receptor binding between IFN-α and IFN-τ probably results in the activation of different signaling pathways, with the exception of Jak/Stat activation, which is responsible for the antiviral effects, as demonstrated for IFN-α and IFN-β (12).

In conclusion, the antiviral effects of ovine recombinant IFN-τ cross the species barrier. Although this IFN is not involved in antiviral defenses in vivo, it induces the same antiviral pathways as human IFNs in human cells, except that the antiviral pathway triggered is not associated with the toxic response observed for human IFNs. A phase I study performed to determine the safety and biological effects of IFN-τ showed no clinically significant toxicity (37), and this compound therefore has considerable potential for therapeutic application.