Human Macrophages Support Persistent Transcription From Unintegrated HIV-1 DNA

Abstract

Retroviruses require integration of their RNA genomes for both stability and productive viral replication. In HIV infection of non-dividing, resting CD4 T cells, where integration is greatly impeded, the reverse transcribed HIV DNA has limited biological activity and a short half-life. In metabolically active and proliferating T cells unintegrated DNA rapidly diminishes with cell division. HIV also infects the non-dividing, but metabolically active macrophage population. In an in vitro examination of HIV infection of macrophages, we find that unintegrated viral DNA not only has an unusual stability, but also maintains biological activity. The unintegrated linear DNA, 1-LTR, and 2-LTR circles are stable for at least 30 days. Additionally there is persistent viral gene transcription, which is selective and skewed towards viral early genes such as nef and tat with highly diminished rev and vif. One viral early gene product Nef was measurably synthesized. We also find that independent of integration, the HIV infection process in macrophages leads to generation of numerous chemokines.

Introduction

As with all retroviruses, the HIV RNA genome needs to be reverse transcribed into a double-stranded DNA molecule that is subsequently integrated into the host chromatin. This proviral DNA serves as a template for viral transcription and progeny production. Integration is believed to be an obligatory step for viral replication (Donehower and Varmus, 1984; Engelman et al., 1995; Schwartzberg, Colicelli, and Goff, 1984; Wiskerchen and Muesing, 1995), and yet the natural process of HIV infection generates excessive amounts of unintegrated viral DNA, which accumulates in the lymphatic tissue, the brain, as well as the peripheral T cells (Kim et al., 1989; Muesing et al., 1985; Pang et al., 1990; Pauza, Galindo, and Richman, 1990; Robinson and Zinkus, 1990; Shaw et al., 1984; Stevenson et al., 1990b; Teo et al., 1997). The biological implication for this excessive presence of unintegrated viral DNA is still not fully understood. There have been numerous experimental inquiries into possible significance of unintegrated DNA. Early studies largely focused on using integration defective viruses which carry low level biological activity (Engelman et al., 1995; Stevenson et al., 1990a; Stevenson et al., 1990b; Wiskerchen and Muesing, 1995). It has been demonstrated that a subset of integrase mutants, with mutations in the catalytic residues, are capable of mediating transactivation of an indictor gene linked to the viral LTR promoter, suggesting the synthesis of Tat (Engelman et al., 1995; Wiskerchen and Muesing, 1995). Subsequently, transcription from unintegrated viral DNA was shown to also occur without altered integrase activity (Brussel and Sonigo, 2004; Wu and Marsh, 2001; Wu and Marsh, 2003). This preintegration transcription was suggested to be a normal early process in HIV-1 infection, which is characterized by the transcription of all viral genes. However, only early proteins such as Nef were detected (Wu and Marsh, 2003). The restriction to early protein syntheses does not appear to be absolute since expression of Rev in trans can rescue the translation of unspliced transcripts (Wu and Marsh, 2003). Another HIV-1 accessory protein, Vpr, has also been suggested to be involved in preintegration transcription. Deletion of Vpr decreases transcription from unintegrated HIV DNA 10- to 20-fold (Poon and Chen, 2003).

Most of these studies have been limited to T cells or proliferating cell lines, where unintegrated HIV DNA is short-lived (Wu and Marsh, 2003; Zhou et al., 2005), and pre-integration transcription was largely viewed as a transient process. Natural HIV targets also include CD4 macrophages and brain microglia. These non-cycling cells are derived from bone marrow and perform similar functions (Hickey, Vass, and Lassmann, 1992; Kennedy and Abkowitz, 1997; Lawson, Perry, and Gordon, 1992; Santambrogio et al., 2001). In the brain, HIV infection is largely limited to perivascular macrophages and parenchymal microglia (Koenig et al., 1986; Lackner et al., 1991; Shaw et al., 1985). HIV infection of macrophages is commonly viewed as non-cytopathic, in contrast to the high turnover of T cells following HIV infection (Ho et al., 1995). However, infected macrophages serve as an important viral reservoir (Crowe, Zhu, and Muller, 2003) and initiate aberrant immunological responses that may directly contribute to pathogenesis. Secreted products from infected macrophages, such as chemokines, may recruit not only the desirable antiviral cytotoxic T lymphocytes but also susceptible uninfected CD4 T cells (Agostini et al., 2000; Fantuzzi et al., 2001; Foley et al., 2005; Schmidtmayerova et al., 1996; Swingler et al., 1999; Tedla et al., 1996). Furthermore, secreted inflammatory factors, and possibly viral products, are believed to be responsible for neuronal death and the impairment of cognitive function in HIV-infected individuals (Ciborowski and Gendelman, 2006; Kaul and Lipton, 2006).

Macrophages also harbor large quantities of unintegrated viral DNA. For example, levels of unintegrated viral DNA were found to be more than 10 times higher than integrated DNA in the brain (Pang et al., 1990). In T cells, unintegrated HIV-1 DNA has been shown to modulate resting T cell activity (Wu and Marsh, 2001), and to down-regulate CD4 receptors (Gillim-Ross, Cara, and Klotman, 2005b). Compared to T cells, much less is understood concerning the stability and transcriptional activity of unintegrated viral DNA in macrophages. In this report we demonstrate that macrophages have the capacity to support sustained viral transcription from unintegrated DNA. We also demonstrate that similar to proviral DNA, the unintegrated DNA has a capacity to induce chemokines such as CXCL9 and CXCL10. These data suggest that macrophages can mediate long-term, low level viral and virus-mediated cellular activities from unintegrated DNA, which may aid viral dissemination and virus-mediated pathogenesis.

Results

Persistence of unintegrated viral DNA in macrophages

To investigate possible accumulation and transcription of unintegrated HIV DNA in macrophages, we set up an in vitro macrophage culture system and examined viral DNA synthesis following infection. Macrophages were derived from peripheral monocytes by culturing in macrophage colony stimulating factor (M-CSF) (10 ng/ml) for two weeks. Following differentiation, cells were infected with HIV-1_AD8 (Wt) or its integration negative mutant, HIV-1_AD8/D116N (D116N) (Englund et al., 1995), using equal p24 viral input. HIV-1_AD8/D116N was derived from HIV-1_AD8 by introducing a single point mutation, Asp-to-Asn substitution, into the integrase catalytic domain, which affects the invariant D-116 residue within the integrase D(35)E motif (Engelman et al., 1995). This single point mutation has been shown to completely abolish viral DNA integration without affecting other known functions such as reverse transcription and nuclear targeting (Engelman et al., 1995; Wiskerchen and Muesing, 1995). We followed viral replication for approximately 30 days. No viral replication was observed in D116N-infected cells, whereas robust viral replication was seen in the wild type-infected cells, with a peak of virus release at around day 15 (data not shown). Our results confirm a previous demonstration of an essential role for integration in HIV-1 infection of macrophages (Englund et al., 1995), although a conflicting report exists (Cara et al., 1995).

Although unable to integrate, viral DNA synthesized from the T tropic D116N was found to be present transiently in T cells (Engelman et al., 1995; Wiskerchen and Muesing, 1995; Wu and Marsh, 2001; Wu and Marsh, 2003). However, in macrophages, we found that the unintegrated DNA from HIV-1_AD8/D116N persisted for at least 30 days until the termination of the cell culture (Fig. 1A). Alu-PCR amplification resulted in no detection of integration in D116N infection at any time point, whereas in the wild type-infected cells, integrated viral DNA was readily detectable at 48 hour and increased with time (Fig. 1B). This is consistent with previous in vitro assays demonstrating that the D116N mutation completely abolishes all enzymatic activities of the integrase, such as 3′ processing, strand transfer and disintegration (Engelman and Craigie, 1992). Two of the unintegrated viral DNA forms, both the 1-LTR and 2-LTR circles, were found to persist with viral infection (Fig. 1C). The D116N-infected cells had higher levels of 2-LTR circles, similar to previous findings of higher levels of 2-LTR circles in D116N-infected T cells (Engelman et al., 1995; Wiskerchen and Muesing, 1995; Wu and Marsh, 2003). However, the difference was not as significant as in infected T cells, particularly at later stage of viral infection (after 20 days). The non-integrated viral DNA detected in D116N-infected macrophages is mostly newly synthesized since the addition of the reverse transcriptase inhibitor, AZT, diminished the viral DNA synthesis (Fig. 1E).

Fig. 1.

Persistence of unintegrated HIV-1 DNA in macrophages. Total cellular DNA harvested at indicated time points post infection with Wt or D116N (equal p24 level) was PCR amplified with primers specific for HIV-1, as described in Materials and Methods. For all PCR amplification, series dilutions were made to ensure amplification is within the liner range (data not shown). All these PCR conditions did not amplify products from uninfected cells (first lanes on the left, Cell). Total cellular DNA was amplified with primers for viral late reverse transcription products (A). The DNA was also amplified for detection of integration by Alu-PCR (B), and unintegrated circular viral DNA forms, both 1-LTR-circle and 2-LTR-circle as indicated (C). For relative comparison, cellular β-actin pseudo-gene was amplified to ensure equal amount of cellular DNA used for amplification (D). To demonstrate that the unintegrated viral DNA was newly synthesized, cells were treated with AZT (50 μM) for 12 hours, and then infected with D116N in the continuous presence of AZT for 5 days. Total cellular DNA was purified and amplified by real-time PCR for viral DNA (E). AZT-untreated, D116N-infected cells were used for comparison.

Among the non-integrated, circular viral DNA forms, the 1-LTR circles appeared to be far more abundant than 2-LTR circles (Fig. 1C). However, it was likely that when 1-LTR and 2-LTR circles were amplified together, 1-LTR circles were preferentially amplified because of their smaller size. We further quantified individual forms of unintegrated DNA by separate methods. Full-length viral DNA and the 2-LTR circle were measured by real-time PCR, whereas the 1-LTR circles can not be directly measured due to the lack of specific primers to distinguish it from the 2-LTR circle. Thus, 1-LTR circle was measured by a competitive PCR approach we developed (Fig. 2). As shown in Table 1, in the wild type infected cells, the 1-LTR and 2-LTR circles were roughly at the same level (5.5-14.3% for 1-LTR circles; 11.8% for 2-LTR circles) and constituted about 20% of the total viral DNA, whereas in D116N infected cells, the 1-LTR- and 2-LTR circles were about 10% and 37.4%, respectively. We assume that the rest of the viral DNA was the linear DNA plus small amounts of irregular DNA forms normally associated with retrovirus infection (Farnet and Haseltine, 1991; Lee and Coffin, 1990; Shoemaker et al., 1980; Shoemaker et al., 1981).

Fig. 2.

Quantification of 1-LTR circles by competitive PCR. (A) Construction of the 1-LTR circle competitor. Cloning of the competitor was described in Materials and Methods. The competitor carrying a new PacI site was used to distinguish it from 1-LTR circles. Total cellular DNA from D116N (B) or Wt (C) infected cells was mixed with serially diluted, known copies of the competitor DNA. The copy numbers of the competitor DNA were determined through three independent measurements by real-time PCR, using the same DNA standard for measuring linear DNA and 2-LTR-circle shown in Table 1. Following PCR amplification with LTR-nef2 and LTR-gag primers, products were digested with PacI. DNA amplified from 1-LTR circles showed a size about 1 kb (a), whereas DNA amplified from the competitor was digested into two small fragments about 0.6 and 0.4 kb respectively (b, c). When the molar ratio between “a” and “b” is one, 1-LTR circle and the competitor have an equal copy number.

Table 1.

Quantification of HIV DNA in macrophages infected for 30 days

	Total DNAa	1-LTR Circleb	2-LTR Circlec	1-LTR Circle (%)d	2-LTR Circle (%)e
Wt	5557.3 ± 1.2	310.0 - 792.8	656.0 ± 1.4	5.5 - 14.3 %	11.8%
D116N	2478.4 ± 1.1	196.8 - 310.0	926.4 ± 1.2	7.9 - 12.5 %	37.4%

^aLate products of HIV reverse transcription measured by real-time PCR (copies in 20 ng total cellular DNA, three independent measurements).

^b1-LTR circles measured by competitive PCR as described in Fig. 3 (copies in 20 ng total cellular DNA). Shown are estimated ranges of 1-LTR-cricles.

^c2-LTR-circles measured by real-time PCR (copies in 20 ng total cellular DNA, three independent measurements).

^dPercentage of 1-LTR Circle in total HIV DNA.

^ePercentage of 2-LTR Circle in total HIV DNA.

Persistence of transcriptional activity from unintegrated viral DNA in macrophages

To determine possible transcriptional activity of unintegrated viral DNA in macrophages, total cellular mRNA was amplified by RT-PCR as previously described (Wu and Marsh, 2001; Wu and Marsh, 2003) (Fig. 3A and 3B). Consistent with the persistence of unintegrated DNA, we observed the persistence of viral transcription in D116N-infected macrophages; the unintegrated viral DNA remained transcriptionally active for a minimum of 30 days (Fig. 3C). Multiple viral transcripts, including all classes of viral splicing isoforms, were detected. In comparison with the wild type, the pattern of viral transcription from D116N was different and selective. The unintegrated DNA predominately transcribed nef, tat, env and gag-pol with diminished rev and vif transcripts. In addition, the ratio between two of the tat isoforms, tat1 and tat2, was reversed, with equal or slightly higher tat 2 amplified, whereas in cells infected with the wild type virus, after three days tat 1 was strongly amplified (compare Fig. 3C lanes 5-8 with lanes 13-16).

Fig. 3.

Persistence of transcription from unintegrated HIV DNA in infected macrophages. (A) Schematic representation of RT-PCR amplification of viral transcripts and primers used for RT-PCR. (B) Total cellular RNA from infected cells (12 hours post infection) was amplified with RT-PCR as described in Materials and Methods. Shown are products from RT-PCR using primers specific for gag-pol, vif, env, tat, rev and nef. None of these primer pairs amplified human sequences from uninfected cells (Cell). To confirm that the amplification of the gag-pol transcript did not result from DNA contamination, RNA was also directly amplified in the absence of reverse transcription (- RT). (C) Time course of viral transcription in D116N and Wt infection. Total cellular RNA from infected cells was harvested at different time, and then amplified as in (B). To ensure that the comparisons were quantitative, samples were normalized by the content of the co-amplified cellular β-actin transcript using QuantumRNA β-actin Internal Standards with a ratio of 1/9 for actin primer/competitor. To confirm that the transcripts detected in D116N infection was derived from newly synthesized viral DNA, cells were treated with AZT (50 μM) for 12 hours, and then infected with D116N. Total cellular DNA and RNA were purified at day 1 post infection and amplified for viral DNA by real-time PCR (D116N, 3721.7±409.7 copies; +AZT, 78.8±4.6 copies) (D) or the nef transcript by RT-PCR (E). To control for DNA carryover from virion particles, D116N virions were also treated with Benzonase (100 U/ml) for 30 min at 37°C or not treated. DNA from 300 ng (p24) of Benzonase-treated or untreated virions were purified and subjected to real-time PCR measurement of viral DNA (D116N, 2.7 × 10⁹ copies; + Benzonase, 1.1 × 10⁶ copies) (F). Benzonase-treated or untreated virions were also used for infection of macrophages. RNA from infected cells (day 1 post infection) was extracted and subjected to RT-PCR amplification for the nef transcript (G).

At early time post infection (4 to 12 hours), transcription from the wild type virus shared a similar pattern as that from D116N (compare Fig. 3C lanes 1-2 with lanes 9-10). Presumably this early transcription in the wild type infection may also be derived from unintegrated DNA. Similar pre-integration transcription has been reported in HIV infection of T cells (Wu and Marsh, 2001; Wu and Marsh, 2003). Indeed, viral integration was undetectable before 12 hours when these transcripts were present (Fig. 2B). After 12 hours, viral transcription increased significantly in the wild type infection. Presumably, this enhanced transcriptional activity in the wild type infection was initiated from provirus following integration.

The generation of viral transcripts from D116N infection is the result of de novo viral DNA synthesis. Treatment of macrophages with AZT during infection drastically inhibited viral DNA synthesis (Fig. 3D), as well as the synthesis of the nef transcript (Fig. 3E). Secondly, although infected macrophages contain some plasmid DNA carried over by virion particles, these contaminating DNA molecules do not appear to play a significant role in mediating viral transcription. Treatment of virion particles with Benzonase prior to infection reduced carryover DNA by three log[10] (Fig. 3F), but no significant reduction in viral transcription was observed (Fig. 3G). This is in great contrast to the inhibitory effect of AZT (Fig. 3E), demonstrating that active transcription in D116N infection depends on newly synthesized viral DNA.

Viral protein syntheses from unintegrated DNA

We also examined the capacity of unintegrated viral DNA to direct viral protein syntheses in macrophages. Infected cells were harvested for western blot, which was probed with human anti-HIV-1 antisera as described (Wu and Marsh, 2003). Viral p24 was detected early (4 hours post infection) both in D116N and wild type HIV-infected cells (Fig. 4A). This protein diminished with the D116N infection, suggesting that it was residual from input virion particles in the D116N infection. The antisera also detected a band with a similar size as the viral p24 precursor, p55gag. The same band was detected from uninfected cells (Fig. 4A, lane 2) and its level did not increase over the course of D116N infection (Fig. 4A lanes 3 to 8), suggesting it is a cellular protein. Thus, we conclude that there was no measurable viral structural protein synthesis in cells infected with D116N. In contrast, in cells infected with the wild type virus, p55gag, along with other viral structural proteins, accumulated with time (Fig. 4A, lanes 9 to 14). The blot was reprobed with a sheep Nef antiserum (Pandori et al., 1996). We detected a protein similar to the size of Nef in wild type virus-infected cells, and this Nef band peaked at day 2 in D116N infection (Fig. 4B). Because the Nef antiserum also recognized a weak band similar to the size of Nef in uninfected macrophages, we performed additional experiments to determine whether the Nef band detected in D116N infection was indeed Nef. We first deleted the nef gene from D116N (Fig. 4G and 4H), and then used the D116NΔnef to infect macrophages. Cell lysates were harvested at day 2 and 3 post infection and analyzed by western blotting with a different Nef antiserum from rabbit (Shugars et al., 1993). Consistently, we observed a similar Nef band that was much stronger than the background band in uninfected cells (Fig. 4I). Indeed, deletion of Nef reduced this band to the uninfected, background level (Fig. 4I). These data confirm that the protein detected by these two Nef antisera in D116N infection was Nef. Nevertheless, the amount of Nef in D116N-infected cells was much lower than that in the wild type-infected cells (Fig. 4B). Quantification of the Nef protein generated by D116N revealed that it was approximately 6 to 10% of that generated by the wild type at day 2 (Fig. 4E and 4F). The other two early viral products, Tat and Rev, are poorly detectable by western, and remain undetected here, but given the previous demonstration of functional expression of Tat by D116N HIV (Wiskerchen and Muesing, 1995), as well as the inability to detect Tat and Rev even in wild type HIV infection by western, we can not exclude expression of these early genes with the D116N virus. The restricted syntheses of only viral early proteins by D116N were similar to what we have observed in D116N infection of T cells (Wu and Marsh, 2003). Previously we have shown that the restriction on structural protein syntheses was due to a lack of Rev function (Wu and Marsh, 2003). Rev is essential for the transition from early to late protein syntheses (Pollard and Malim, 1998). The Rev transcript was at a low level over the course of D116N infection of macrophages. Presumably, this minimal amount of Rev was not adequate for its function to turn on viral structural protein syntheses in the absence of integration.

Fig. 4.

Viral protein synthesis from unintegrated DNA in infected macrophages. Lysates from infected cells were resolved and blotted with human anti-HIV antisera (A), or a sheep Nef antiserum (B), or a monoclonal antibody against human actin as a loading control (D). To serve as a Nef control, 0.5 ng of purified HIV-1 Nef protein (lane 1) and uninfected macrophages (lane 2) were applied to the gel. Protein bands were detected by incubation with corresponding secondary antibodies conjugated to peroxidase and visualized by chemiluminescence. (C) is a shorter exposure of (B). Quantification of the Nef protein in (B) was performed as previous described (Liu et al., 2001). Integration of chemiluminscence signal yielded the relative intensities of the Nef protein (arbitrary units): lane 1 (purified Nef), 2537 units; lane 4 (D116N, day 2), 113 units; lane 5 (D116N, day 3), 120 units; lane 10 (Wt, day 2), 1835 unites; lane 11 (Wt, day 3), signal saturated. The Nef protein standard and a repeat of the Nef western blot from a separate infection with either Wt or D116N (Dn) are shown in (E) and (F). A Nef-expression Jurkat cell line was used as a control. To confirm the synthesis of the Nef protein from D116N, a D116N nef mutant was constructed by inserting a Nef-stop-linker sequence (G). The nef deletion was confirmed by sequencing analysis and western blotting of infected Rev-CEM indicator cells (H). (H) Rev-CEM was infected with an equal p24 level (3 μg per million cells) of D116N or D116NΔnef, and cell lysates were prepared at 48 hours for western blotting for Nef, HIV-1 p24, or β-actin for loading control. Lane 1 is 0.5 ng of purified Nef protein. (I) Macrophages were also infected with an equal p24 level of D116N or D116NΔnef. Cell lysates were prepared at days 2 and 3 post-infection. Blots were probed, stripped, and reprobed similarly as in (H) with antibodies specific to Nef, p24, or β-actin.

Chemokine induction by unintegrated viral DNA in macrophages

In a prior study with quiescent T cells, we have demonstrated that transcription from unintegrated HIV DNA can alter the outcome of T cell activation and HIV replication (Wu and Marsh, 2001). The biological response of macrophages to HIV infection includes the generation of various chemokines and inflammatory cytokines (Canque et al., 1996; Clouse et al., 1991; Fantuzzi et al., 2001; Gruber et al., 1995; Mengozzi et al., 1999; Schmidtmayerova et al., 1996). To evaluate possible contribution of unintegrated HIV DNA to the biological response in macrophages, we assayed for various chemokine inductions following HIV infection. Initially, we performed assays on the supernatant of infected cells with a cytometric bead array for chemokines and inflammatory cytokines. This assay included the chemokines CXCL8 (IL-8), CCL5 (RANTES), CXCL9 (MIG), CCL2 (MCP-1), and CXCL10 (IP-10), and the inflammatory cytokines TNF, IL-10, IL-1β, IL-6, IL-8 and IL-12p70. We infected macrophages with equivalent levels of HIV-1_AD8 and the integrase mutant HIV-1_AD8/D116N, followed by chemokine and cytokine measurement. Many of the previous studies cited above examined macrophage infection for periods in excess of 1-2 weeks; however, we noted that with these long periods of experimental incubation there was greater variability in cell viability and large-cell formation, along with varied donor-dependent chemokine expression (data not shown). When we looked at earlier time points (2-4 days), we found reproducible (from donor to donor) chemokine responses. CCL2 levels were elevated in all macrophage preparations independent of HIV exposure (data not shown), and this may be related to M-CSF, which was used here to differentiate the monocytes to macrophages, and has been shown to also induce CCL2 (Shyy et al., 1993). Furthermore, CCL5 was induced in HIV-infected macrophages in only one of four donors (data not shown). With the exception of CXCL8 (IL-8, see below), none of the cytokines from the cytometric bead arrays were notably induced by day 4 (data not shown). However, the chemokines CXCL10 and CXCL9, as well as cytokine/chemokine CXCL8, were found to be induced in macrophages from all donors within 4 days, following infection with wild type HIV or the integrase mutant HIV (see Table 2). From p24 staining of wild type HIV-infected cells, we would estimate that the infection levels approximated ten percent of cells (10.2% ± 2.5, mean ± SE, n = 5). Additionally, the inclusion of AZT, which inhibits de novo viral DNA synthesis, reduced the induction of CXCL9 and CXCL10 at day four to levels comparable to uninfected populations (Table 2). Relative to wild type virus, the addition of AZT also diminished CXCL8, but for this chemokine, the induced levels in the presence of AZT approached the levels that were achieved by the addition of D116N.

Table 2.

Induction of chemokine secretion by HIV-1

Chemokine	Wild type HIV	Integrase-negative HIVa	Not Infected	Wild type HIV plus AZT
CXCL8	1275 ± 561 pg/ml (n = 5)	717 ± 334 pg/ml (n = 3)	248 ± 57 pg/ml (n = 5)	773 ± 333 pg/ml (n = 6)
CXCL9	297 ± 84 pg/ml (n = 5)	60 ± 18 pg/ml (n = 3)	21 ± 10 pg/ml (n = 5)	11 ± 4 pg/ml (n = 6)
CXCL10	2902 ± 699 pg/ml (n = 6)	1172 ± 892 pg/ml (n = 3)	323 ± 69 pg/ml (n = 6)	135 ± 55 pg/ml (n = 6)

Monocyte-derived macrophages from multiple donors (n = 3 - 6) were infected with either wild type HIV-1_AD8, or integrase mutant HIV-1_AD8/D116N or not infected, with or without AZT. After four days incubation, cell supernatants from infected and non-infected populations were assayed by cytometric bead array. The level of chemokine is listed as the mean value in pg/ml ± SE. In assays that individual donor samples were not measured in replicate, the variation (standard error) represents differences among donors. The number of donors examined is defined by n.

^aPercentage of chemokine production in cells infected with D116N versus Wt (after subtracting uninfected cell background): CXCL8, 46%; CXCL9, 14%; CXCL10, 33%. -

We reproduced the same experiment in monocyte-derived macrophages from three additional donors, but this time performed three independent infections of cells from each donor and measured CXCL10 by ELISA (Fig. 5). This procedure permits multiple measurements of a chemokine induction from the same donor, something we did not perform on the costly bead array. Induction of CXCL10 secretion occurred with wild type virus, and this induction was again prevented by the inhibition of reverse transcriptase. The addition of similar levels of D116N also resulted in the induction of CXCL10, an effect also diminished by AZT. The CXCL10 secretion seen with the integrase mutant was reduced, relative to wild type virus, and this reduction is presumably due to the loss of secondary infection by the mutant virus, and limited viral protein synthesis observed (Fig. 5). The ability of AZT to diminish HIV induction of the chemokines suggests that the effect is mediated by post-reverse transcription activity, but does not require completion of integration. Our result is consistent with the findings of Foley et al. (Foley et al., 2005), who noted that HIV infection of macrophages rapidly induced CXCL10 transcription. The authors reported that this response of macrophages to HIV appeared to be independent of the viral envelope engagement, but did require a post-reverse transcription process (Foley et al., 2005).

Fig. 5.

Wild type and integrase-negative HIV induction of CXCL10 in macrophages. Monocyte-derived macrophages from healthy donors were infected with an equal level of either wild type or the integrase mutant, D116N, and cultured for 4 days either in the presence or absence of AZT. Supernatants from cell cultures were collected at 4 days post infection and assayed by ELISA for CXCL10 secretion. Each donor cell population underwent three independent infections (n = 3) and the data represent mean ± standard error.

Discussion

In this article, we demonstrate that unintegrated HIV-1 DNA can persist and remain active for transcription in infected macrophages. Multiple unintegrated viral DNA forms, the linear and the 1-LTR- and 2-LTR circles, were found to persist for at least 30 days. Transcription in the absence of integration is selective and skewed towards certain viral early genes such as nef and tat, with highly diminished rev and vif. While no structural proteins were measurably synthesized, one viral early protein, Nef, was detectable. These results were reproduced in macrophages cultured from four different donors. The persistence of unintegrated viral DNA in macrophages was in contrast to its transient presence in proliferating T cells (Wu and Marsh, 2003), in which pre-integration transcription peaks at about 12 hours and diminishes within 3 days, and the unintegrated viral DNA such as 2-LTR circles diminishes as a result of T cell division (Butler, Johnson, and Bushman, 2002; Pierson et al., 2002).

The nature of the transcribing, unintegrated HIV DNA has not yet been determined. Previous studies have suggested that the 2-LTR circles could be the template (Engelman et al., 1995; Wiskerchen and Muesing, 1995). This was largely based on the fact that higher than wild type levels of 2-LTR circles were found in T cells infected with integrase mutants (Engelman et al., 1995; Wiskerchen and Muesing, 1995). We also found higher levels of 2-LTR circles in D116N-infected macrophages (Fig. 2C). Although the 2-LTR circle could be a template, we cannot exclude the possibility that other DNA forms such as the 1-LTR circle or the linear viral DNA may function as a template. Both DNA forms, along with the 2-LTR circle, persisted with viral transcription in D116N-infected macrophages. Aside from these unintegrated viral DNA, virion particles prepared by transfection carry some plasmid DNA on the surface that could be introduced into the cell. The plasmid DNA can be removed by treatment with Benzonase to a level reportedly undetectable by PCR (Sastry et al., 2004) or three log[10] lower as demonstrated in this report (Fig. 3F). We observed no decrease in viral transcriptional activity following Benzonase treatment of the D116N virions (Fig. 3G), excluding any role of contaminating plasmid DNA in mediating transcription. It is likely that even the carryover plasmid DNA can enter cells through non-specific attachment to virion particles, these molecules may not enter the nucleus for transcription.

The production of chemokines and inflammatory cytokines by macrophages and microglia, following HIV infection, has been attributed to both enhanced infection of T cells (Canque et al., 1996; Swingler et al., 1999; Tedla et al., 1996) as well as HIV-mediated brain diseases, such as the AIDS dementia complex (Hickey and Kimura, 1988; Koenig et al., 1986; Lassmann, Rinner, and Hickey, 1994; Lassmann et al., 1993). Human monocytes/macrophages exposed to gp120 or inactivated HIV have been shown to secret elevated levels of TNF, IL-1, IL-6, IL-10, MCP-1, MIP-1, or Rantes (Borghi et al., 1995; Clouse et al., 1991; Fantuzzi et al., 2001; Lee et al., 2005). Infectious virus have been reported to be necessary for induction of MIP-1, MCP-1, and M-CSF, since AZT or heat-inactivation of the HIV resulted in diminished production of these cytokines/chemokines (Canque et al., 1996; Kutza et al., 2000; Mengozzi et al., 1999; Schmidtmayerova et al., 1996). The induction of CXCL8 (IL-8) is not as well characterized, but in one study it was found that CXCL8 was generated from many p24-negative cells in an HIV-infected population (Glienke et al., 1994), implying that the inducing factor could be a viral (or cellular) product secreted into the media. This is consistent with our finding that the addition of non-replicating D116N HIV or wild type HIV plus AZT (also non-replicating) induced similar levels of CXCL8, with both populations generating lower CXCL8 levels than a macrophage population supporting spreading HIV (Table 2).

The inductions of CXCL9 and CXCL10 by HIV are likely to be different. Work by Foley et al has demonstrated that the induction of these chemokines by HIV was eliminated by AZT, and could be mediated by VSV-pseudotyped HIV (Foley et al., 2005). These findings suggest that post reverse transcription processes are responsible, and Foley et al have excluded an interferon-based induction. Our findings concerning CXCL10 are in agreement with these findings, and in addition it is noted here that integration is not essential for induction of this chemokine. A mechanism for HIV induction of CXCL10 is under investigation.

The induction of CXCL10 by HIV results in a process by which macrophage activity could affect the in vivo spread of R5 HIV. It was reported that R5 HIV-infected macrophages could recruit resting CD4 T cells, through secretion of soluble ICAM and CD23, and then enhance T cell infection with X4 HIV (Swingler et al., 2003; Swingler et al., 1999). CXCL10 largely recruits active CCR5⁺ CD4 T cells (Foley et al., 2005), which could be directly infected by the R5 HIV-1 released by infected macrophages. Active T cells are highly susceptible to the R5 virus (Brenchley et al., 2004; Mattapallil et al., 2005).

A definitive role for unintegrated DNA in the pathogenesis of HIV infection is unresolved. In other retroviruses, unintegrated DNA has been implicated in viral pathogenesis. Keshet and Temin were the first to suggest a correlation between cell killing and accumulation of unintegrated DNA in spleen necrosis virus infection (Keshet and Temin, 1979). A similar association has been seen in avian leukosis virus-induced osteoporosis, feline leukemia virus-induced feline AIDS, and equine infectious anemia virus infection of horses (Mullins, Chen, and Hoover, 1986; Rice et al., 1989; Weller, Joy, and Temin, 1980). In HIV infection of T cells, accumulation of unintegrated viral DNA was reported to correlate with the extent of syncytia formation (Pauza, Galindo, and Richman, 1990), but not the occurrence of single-cell killing (Bergeron and Sodroski, 1992). However, it is not clear whether the mere presence of unintegrated DNA may trigger certain cellular processes, or products from the DNA may aid viral pathogenesis to some extent.

In addition to induction of inflammatory proteins, pathogenic effects from viral early products such as Nef exist. Nef-transduced macrophages, when grafted into the rat hippocampus, induced monocyte/macrophage recruitment, expression of TNF-α, and astrogliosis (Mordelet et al., 2004). Nef was also found to regulate the release of superoxide anion from Nef-transduced human macrophages (Olivetta et al., 2005; van Marle et al., 2004). Even very low levels of Nef (barely detectable) expressed in oligodendrocytes were sufficient to mediate vacuolar myelopathy and elicit disease (Radja et al., 2003). The production of proinflammatory cytokines has been associated with multiple viral factors such as gp120, Nef and Tat (Koka et al., 1995; Merrill et al., 1992). We would assume that stimulation of proinflammatory cytokines by D116N could be a combined effect of multiple viral factors. Nevertheless, our finding reinforced the notion that non-replicative HIV with limited activity is still capable of inducing neuronal damage (Pang et al., 1990).

Unintegrated HIV DNA lacks a functional origin of replication normally present in DNA viruses such as SV40. Introduction of such an origin leads to DNA amplification and productive viral replication (Lu et al., 2004). This observation highlights the full transcriptional potential of unintegrated DNA, and is consistent with a notion that unintegrated circular DNA is highly stable. These DNA forms decrease in concentration only as a function of dilution resulting from cell division (Butler, Johnson, and Bushman, 2002; Pierson et al., 2002). Thus, it is not surprising that in non-dividing cells, when cell cycle is arrested, unintegrated viral DNA can persist (Gillim-Ross, Cara, and Klotman, 2005a; Saenz et al., 2004). The ability of unintegrated viral DNA to persist and to transcribe has stimulated a recent interest in using unintegrating lentiviral vector to transduce non-dividing cells, such as ocular or neuronal cells, for gene therapy (Philippe et al., 2006; Saenz et al., 2004; Yanez-Munoz et al., 2006). These vectors have demonstrated a surprising efficiency in mediating high level, stable expression from internally promoted transgenes in non-dividing cells (Philippe et al., 2006; Saenz et al., 2004; Yanez-Munoz et al., 2006). Currently we are also examining an unintegrating Rev-dependent lentiviral vector (Wu, Beddall, and Marsh, 2007b) for targeting HIV-1-infected macrophages (unpublished data). Such vectors would be advantageous for minimizing integration-mediated mutagenesis. Additionally, the capacity of unintegrated HIV DNA to persist and express viral genes in macrophages, a professional antigen-presenting cell, could be exploited in the development unintegrating lentiviral vectors for the delivery of vaccine or therapeutic genes (Lu et al., 2003; Smith, 2002).

Materials and Methods

Viruses and cells

Viral stocks of the HIV-1_AD8 and the integrase mutant HIV-1_AD8/D116N (kindly provided by Dr. Malcolm A. Martin) (Englund et al., 1995) were prepared by transfection of HEK293 T cells as previously described (Wu and Marsh, 2001). To remove plasmid DNA contamination, viral stock was treated with Benzonase (Novagen, Madison, WI) (50 to 100 U/ml) at 37 °C for 15 to 30 min. To measure the extend of plasmid DNA degradation, DNA from treated viruses was purified, and then subjected to real-time PCR amplification (see below) for measuring full-length viral DNA. Viral titer was measured by TCID₅₀ assay with a cloned CEM indicator line with R5 susceptibility (Rev-CEM) (Wu, Beddall, and Marsh, 2007a; Wu, Beddall, and Marsh, 2007b). Macrophages were differentiated from human monocytes from the peripheral blood of HIV-1 negative donors. All protocols involving human subjects were reviewed and approved by the George Mason University IRB. Briefly, two million peripheral blood mononuclear cells were plated into each well of six plates in serum free RPMI medium for one hour. Adherent cells were cultured in RPMI plus 10% heat inactivated fetal bovine serum (FBS) with 10 ng/ml macrophage colony stimulating factor (M-CSF) (R&D System, Minneapolis, MN) for two weeks with medium change for every two days. Macrophages were infected with wild type virus at a multiplicity of infection (m. o. i) 0.2. The integrase mutant was used for infection with an equivalent level of p24 as the wild type. Infection was carried out at 37 °C for 4 hours. Infected cells were washed three times and continuously cultured in RPMI plus 10% FBS without M-CSF. Fresh medium was added every two days. Viral replication was monitored by harvesting supernatant and measuring the content of viral p24 as recommended by the manufacture (Beckman Coulter, Miami, FL).

Generation of the D116N Nef deletion mutatnt, HIV-1_{AD8/D116N/Δnef}

Nef deletion was first introduced into HIV-1(KFS) (Kindly provided by Dr. Eric Freed) by inserting a Nef-STOP-Linker (TGATTGATGACGGCCGTAGTAGTGA) into the unique BlpI site. The Nef deletion mutant of HIV-1_AD8/D116N was then generated by replacing the XhoI-BamHI fragment of HIV-1_AD8/D116N with the XhoI-BamHI fragment of HIV-1(KFS)Δnef. The Nef deletion was confirmed by DNA sequencing.

DNA and RNA detection

Total cellular DNA and RNA were purified using SV total RNA isolation kit as recommended by the manufacture (Promega, Madision, WI). Purified RNA was further treated with DNase I (DNA-free kit, Ambion Inc, Austin, TX) as recommended by the manufacture. Quantitative real-time PCR analyses of viral DNA and 2-LTR circles were carried out by using ABI Prism 7700 sequence detection system as described previously (Wu and Marsh, 2003), using the forward primer, 5′LTR-U5: 5′-AGATCCCTCAGACCCTTTTAGTCA-3′; the reverse primer, 3′ gag: 5′-TTCGCTTTCAAGTCCCTGTTC-3′; the probe, FAM-U5/gag: 5′-(FAM)-TGTGGAAAATCTCTAGCAGTGGCGCC-(TAMRA)-3′. DNA standard used for both late DNA and 2-LTR circle quantification was constructed by using a plasmid containing a complete 2 LTR region (pLTR-2C, cloned by amplification of infected cells with 5′-TGGGTTTTCCAGTCACACCTCAG-3′ and 5′-GATTAACTGCGAATCGTTCTAGC-3′). Measurement was run in triplicate ranging from 1 to 10⁶ copies of pLTR-2C mixed with DNA from uninfected cells. For the detection of viral late reverse transcription products by PCR, forward primer: 5′ GGTTAGACCAGATCTGAGCCTG 3′ and reverse primer: 5′ TTAATACCGACGCTCTCGCACC 3′ were used. PCR was carried out in 1 × Ambion PCR buffer, 125 μM dNTP, 50 pmol each primer, 1U SuperTaq Plus (Ambion Inc. Austin, TX) with 30 cycles at 94 °C for 20 seconds, 68 °C for 40 seconds. For detection of circular viral DNA forms, both the 1-LTR and 2-LTR circle, primer LTR-nef2 (5′ TGGGTTTTCCAGTCACACCTCAG 3′) and LTR-gag (5′ GATTAACTGCGAATCGTTCTAGC 3′) were used as previously described (Wu and Marsh, 2003). Viral DNA integration was measured by Alu-PCR as previously described (Wu and Marsh, 2001). Briefly, nuclear DNA was purified and subjected to amplification by Alu-LTR PCR, using the Alu primer (5′ TCCCAGCTACTGGGGAGGCTGAGG3′) and HIV-1 LTR primer L1 (5′ AGGCAAGCTTTATTGAGGCTTAAGC3′). Following Alu-LTR PCR, a second round of PCR was carried out using an aliquot equivalent to 1/500 of the PCR products, using LTR specific primer pairs, L2 (5′CTGTGGATCTACCACACACAAGGCTAC3′) and L3 (5′GCTGCTTATATGTAGCATCTGAGGGC3′). Construction of the 1-LTR circle competitor was achieved by PCR amplification and cloning of DNA from infected cells with primer LTR-nef2 and LTR-gag. The resulting plasmid was digested with AfIII, then treated with Klenow fragment of DNA polymerase I and self-ligated to generate a new PacI site which was used to distinguish the competitor from the 1-LTR circle.

Quantitative RT-PCR analysis of viral transcripts was carried out as previously described (Wu and Marsh, 2001; Wu and Marsh, 2003). Briefly, reverse transcription was accomplished with random decamers as the first-strand primers. Following cDNA synthesis, PCR was carried out using primer F2 (5′TAATCGGCCGAACAGGGACTTGAAAGCGAAAG3′) and B4 (5′CCATCGATTGCGTCCCAGAAGTTCCACAATCC3′) to amplify viral transcripts, specifically, doubly spliced transcripts such as nef, tat and rev. For relative quantification of RT-PCR, cellular β-actin transcripts were co-amplified using QuantumRNA β-actin Internal Standards (Ambion Inc. Austin, TX), with a ratio of 1/9 for actin primer/competitor. To analyze singly spliced viral transcripts such as the env transcript, primer F2 and B3 (5′CCCATCTCCACAAGTGCTGATACTTC3′) were used, whereas for the vif transcript, F2 and B2 (5′CTAGGTCAGGGTCTACTTGTGTGC3′) were used. For analysis of full-length viral transcripts, mRNA was treated with DNase I (DNA-free kit, Ambion, Inc. Austin, TX), then subjected to RT-PCR amplification using primer pair F2 and B1 (5′ TCTGAAGGGATGGTTGTAGCTGTCC3′).

Immunodetection of viral proteins

Viral proteins were detected by resolving proteins on 4-20% SDS-polyacrylamide gel and electroblotting onto 0.2 μm nitrocellulose membrane. A 1:1000 dilution of a human anti-HIV antiserum (obtained through the NIH AIDS Research and Reference Program) or a monoclonal antibody against gag (p24) (Wehrly and Chesebro, 1997) (obtained through the NIH AIDS Research and Reference Program) was incubated with the membrane, followed by a secondary goat antihuman antiserum (1:2000 dilution) or goat antimouse antiserum conjugated with peroxidase. The Nef protein was detected by using 1:1000 dilution of a sheep anti-Nef antiserum as previously described (Wu and Marsh, 2001), as well as a rabbit Nef antiserum (Shugars et al., 1993) (1:4000 dilution) provided by the NIH AIDS Research & Reference Reagent Program.

Detection of the immunoreactive product by chemilumilence was described previously (Wu and Marsh, 2001). The light signal was captured on a cooled CCD camera using chemiluminescent SuperSignal West Dura substrate (Pierce, Rockford, IL). Staining of p24 to identify HIV-infected macrophages was accomplished as previously described (Wu, Beddall, and Marsh, 2007b).

Chemokine assays

Macrophages were cultured in six well plates as described above and infected with equivalent p24 levels of either HIV-1_AD8 (Wt) or the integrase mutant, HIV-1_AD8/D116N (D116N) in the presence or absence of 50 μM 3′azido-3′deoxythimidiine (AZT) (Calbiochem, San Diego, CA). Following four days of culture, supernatants were assayed by Cytometric Bead Array (BD Biosciences, San Jose, CA) or by ELISA (BioSource, Camarillo, CA) following the manufacturers’ recommendations. We made use of the CBA inflammatory cytokine and CBA chemokine kits (BD Biosciences, San Jose, CA) to survey chemokines CXCL8 (IL-8), CCL5 (RANTES), CXCL9 (MIG), CCL2 (MCP-1), and CXCL10 (IP-10), and cytokines TNF, IL-1β, IL-12p70, and IL-6. In addition, we measured chemokine induction in triplicate by ELISA from multiple infections of cells from additional donors.