|Home | About | Journals | Submit | Contact Us | Français|
Institute for Virology, University Hospital Ulm, Germany.
This work was supported by a Wellcome Trust fellowship (WT077161) to MN.
Macrophages contribute to HIV-1 infection at many levels. They provide permissive cells at the site of inoculation, augment virus transfer to T cells, generate long-lived viral reservoirs and cause bystander cell apoptosis. A body of evidence suggests that the role of macrophages in cellular host defence is also compromised by HIV-1 infection. In this respect, macrophages are potent cells of the innate immune system that initiate and regulate wide-ranging immunological responses. This study focuses on the effect of HIV-1 infection on innate immune responses by macrophages at the level of signal transduction, whole genome transcriptional profiling and cytokine secretion. We show that in an ex vivo model, M-CSF differentiated monocyte-derived macrophages (MDM) uniformly infected with replicating CCR5 tropic HIV-1, without cytopathic effect, exhibit selective attenuation of the NF-κB activation pathway in response to TLR4 and TLR2 stimulation. However, functional annotation clustering analysis of genome-wide transcriptional responses to lipopolysaccharide stimulation suggests substantial preservation of gene expression changes at the systems level, with modest attenuation of a subset of up-regulated LPS responsive genes, and no effect on a selection of inflammatory cytokine responses at the protein level. These results extend existing reports of inhibitory interactions between HIV-1 accessory proteins and NF-κB signalling pathways, and whole genome expression profiling provides comprehensive assessment of the consequent effects on immune response gene expression. Unexpectedly, our data suggest innate immune responses are broadly preserved with limited exceptions, and pave the way for further study of the complex relationship between HIV-1 and immunological pathways within macrophages.
Macrophages are permissive to CCR5-tropic HIV-1 and in most models can host active viral replication without significant cytopathic effect1,2. This interaction may contribute to HIV-related disease during HIV-1 transmission when R5-tropic virus predominates3, by transfer of virus to T-cells4, as a long-lived viral reservoir5 and by provoking bystander cell apoptosis that may be involved in T-cell or neuronal death6,7. In addition the effect of HIV-1 infection on macrophage function is of significant interest. HIV-1 mediated inhibition of complement and Fc-receptor phagocytosis8,9 and attenuated intracellular killing of Leishmania10 suggest that the role of macrophages as host defence effector cells may be compromised. Importantly, macrophages also function as sentinel cells of innate immunity involved in recognition of microbial pathogens that leads to both activation and regulation of host immune responses to wide ranging microbial pathogens11. Innate immune stimulation of monocyte derived macrophages (MDM) and related monocyte derived dendritic cells from HIV-infected subjects have frequently been reported to show attenuated or altered host cell responses12, but these observations are unlikely to represent direct effects since the extent of HIV-1 infection of monocytes in vivo, is estimated to be very low13,14. The same may be true of reports of deficient innate immune responses in alveolar macrophages from HIV-1 infected subjects15-19. Nonetheless two additional lines of evidence support the hypothesis that HIV-1 inhibits innate immune responses in macrophages. Firstly, the latently infected myeloid leukaemia cell line, U1, in which HIV-1 replication can be activated by PMA-induced differentiation into a macrophage-like adherent phenotype, shows reduced inflammatory cytokine production compared with the parental non-infected U937 cell line20,21. This effect has been attributed to upregulation of mitogen activated protein kinase phosphatase (MKP)-1 by the HIV-1 accessory protein nef, and consequent inhibition of the innate immune signalling cascade through extracellular regulated kinase (ERK)1/222. Secondly, in a variety of models, other interactions between HIV accessory proteins and components of the innate immune signalling pathway have also been reported to inhibit the function of the nuclear factor(NF)-κB family of transcription factors. The HIV accessory protein vpu can interact with β-transducin repeat-containing protein (βTrCP), involved in E3 ubiquitin ligase complex-mediated degradation of intracellular proteins23,24. This interaction competitively inhibits TrCP-dependent degradation of the inhibitor of κB (IκB)D, and in Drosophila, transgenic expression of vpu inhibits Toll-mediated degradation of the IκB homologue, cactus, and consequent activation of the NF-κB homologues, dorsal and dif25. In addition, another HIV-1 accessory protein, vpr interacts with the glucocorticoid receptor and generates a novel biological effect, inhibiting nuclear translocation of the NF-κB coactivating factor poly(ADP-ribose) polymerase (PARP)-126. The NF-κB family of heterodimeric transcription factors are a major target for innate immune receptor mediated signalling in general27,28. The most abundant component of this family is the p65 (RelA)/ p50 heterodimer in which p50 principally functions as a regulatory element interacting with IκBα. This complex shuttles between nucleus and cytoplasm but in unstimulated cells shows relative cytoplasmic sequestration. Activation of the classical NF-κB pathway induces phosphorylation and degradation of IκBα and nuclear translocation of NF-κB, where RelA exerts potent transcription factor activity29. Inhibition of this pathway by microbial pathogens is increasingly evident30 and currently a key focus of research on host-pathogen interactions that may provide opportunities for novel therapeutic interventions in the future.
In this study, in view of the limited evidence showing HIV-1 induced inhibition of innate immune cellular activation and in particular, possible inhibition of the NF-κB activation pathway that plays a central role in mediating innate immune cellular responses, we used ex vivo replication competent HIV-1 infection of human MDM in order to investigate the effect of HIV-1 on activation of selected innate immune signalling pathways and downstream immune responses. Importantly, innate immune signalling in macrophages induces complex and wide-ranging transcriptional responses31, that include expression of cytokines, inducible intracellular enzymes, cell surface molecules, plasma proteins, cytoskeletal components and factors that regulate cell cycle or apoptosis. Despite previous mechanistic reports of HIV-1 mediated inhibition of innate immune signalling, the effect of HIV-1 infection in macrophages on the broad repertoire of innate immune response elements has not previously been assessed. Therefore, in addition to testing the hypothesis that HIV-1 inhibits innate immune signalling in a more physiological macrophage model, we have extended the assessment of effects on downstream immune response genes using whole genome transcriptional profiling.
Human blood samples were obtained from healthy volunteers for isolation of peripheral blood mononuclear cells (PBMC) and production of MDM cultures. The study was approved by the joint University College London/University College London Hospitals National Health Service Trust Human Research Ethics Committee and written informed consent was obtained from all participants. PBMC were prepared by density-gradient centrifugation of heparinised blood with Lymphoprep™ (Axis-Shield) according to the manufacturer's instructions and MDM were prepared as previously described32. PBMC were seeded (2 ×106/cm2) for adhesion onto tissue culture plastic (Nunc). After one hour (h) at 37°C non-adherent cells (lymphocytes) were removed and adherent monocytes were incubated in RPMI 1640 (GIBCO Invitrogen) with 10% autologous heat-inactivated human serum (HS) supplemented with 20 ng/mL macrophage colony stimulating factor (M-CSF) (R&D systems) for three days. The media was then refreshed (without additional M-CSF), removing any remaining non-adherent cells. Typically, this protocol yields 105 MDM/cm2. After 6 days culture 10% autologous HS was replaced with 5% normal (N)HS (Sigma-Aldrich).
The CCR5-tropic HIV-1 strain, Ba-L was propagated in peripheral blood lymphocytes (PBL). Non-adherent PBLs from MDM preparations were cultured for 3 days in RPMI 1640 with 20% FCS and 0.5 μg/mL phytohaemagglutinin (PHA) (Sigma) to generate activated T cells. These cells were then inoculated with HIV-1 Ba-L, using a multiplicity of infection (MOI) of 1, and subsequently cultured in RPMI 1640 with 20% FCS and 20 U/mL interleukin (IL)-2 (Peprotech). At 3-4 day intervals, the cell culture supernatants were collected and additional PHA-stimulated PBMC were added to maintain the cell density at 1 ×106/mL. Cell culture supernatants containing PBMC-derived HIV-1 were filtered through 0.45μ filters (Millipore) and used to inoculate 6-day old MDM cultures overnight (MOI 1), refreshing the media on the following day. Culture supernatants from infected MDM, containing MDM-derived HIV-1 Ba-L, were collected at weekly intervals, centrifuged at 400g for 5 minutes (min) and filtered (0.45μ Millipore filter) to remove cellular debris. The CCR5/CXCR4 dual-tropic HIV-1 strain, 89.6, and the CXCR4-tropic HIV-1 strain, NL4-3 were derived from infectious clones by transient transfection of HEK293t producer cell cultures using Fugene® 6 transfection reagent (Roche) according to manufacturer's instructions, and collecting culture supernatants 72 hours later. All virus suspensions were ultracentrifuged through a 20% sucrose buffer and resuspended in RPMI 1640 with 5% NHS, for subsequent infection of MDM. All virus preparations were titrated on the NP2 astrocytoma cell line stably transfected with CD4 and CCR5 or CXCR4 as previously described33. Briefly, adherent NP2/CD4/CCR5 and NP2/CD4/CXCR4 cells, cultured in Dulbecco's Modified Eagle Medium (GIBCO Invitrogen) with 5% FCS, 1 μg/mL puromycin (Sigma-Aldrich) and 100 μg/mL G418 (Sigma-Aldrich) were inoculated with serial log-fold dilutions of viral stocks. The cells were incubated with the virus for 2 hours at 37°C before removing the inoculum and replacing the media. Infection was detected by p24 immunostaining 72 hours later (described below). MDM-derived HIV-1 Ba-L, and HEK293t-derived HIV-1 NL4-3 and 89.6 strains were used for overnight inoculation (MOI 3-5) of 6-day old MDM cultures in the experimental model presented here.
Cell cultures were fixed with ice cold methanol:acetone (1:1) for 5 minutes, then incubated with mouse anti HIV-1 gag (p24) monoclonal antibodies (E365/366, National Institute for Biological Standards and Control, UK) followed by goat anti-mouse immunoglobulin (Ig) antibody conjugated to β-galactosidase (gal) (Southern Biotechnology Associates) for 1 hour each at room temperature. The β-gal substrate solution (0.5 mg/mL 5-bromo-4-chloro-3-indolyl-β-galactopyranoside in PBS containing 3 mmol/l potassium ferricyanide, 3 mmol/l potassium ferrocyanide and 1 mmol/L magnesium chloride) was then added and incubated overnight at 37°C to develop a blue stain. Positively staining cells were counted microscopically to provide a virus titre (focus forming units/mL) or to calculate the number and proportion of infected cells.
MDM culture lysates collected in RLT buffer (Qiagen) were used to purify total RNA with RNAeasy spin columns (Qiagen) according to manufacturer's instructions. 1μg of total RNA was used to generate cDNA with the First Strand cDNA synthesis kit (New England Biolabs) using oligo dT primers. The product was then heat-inactivated (95°C, 5 min) and subjected to PCR (95°C denaturing temperature, 55°C annealing temperature and extension temperature 72°C for 30 cycles) using primers (Table 1) for a conserved unspliced HIV-1 gag product (SK38 and SK39)34 and spliced HIV-1 tat, rev, nef product (413MOD and P659)35. Samples from HIV-1 Ba-L infected MDM were also subjected to sequencing reactions (Lark Technologies) for the HIV accessory genes nef, vpu and vpr using primers (Table 1) based on the published HIV-1 Ba-L sequence (accession no: AB253432). All three full-length accessory genes were sequenced successfully and had 95-100% amino acid homology to the expected sequences (data not shown).
Relative viability of HIV infected and uninfected MDM cultures was compared by using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) assay. Cell culture supernatants were removed and replaced with 1 mg/mL MTT (Sigma-Aldrich) in serum-free media (1mL/106 cells) and incubated for 3 hours at 37°C. This was then replaced with dimethyl sulfoxide (Sigma-Aldrich) to permeabilise cells in a plate shaker for 5 minutes (room temperature). Relative formazan concentration was quantified by spectrophotometry (OD540nm) within 0.1-1.8 linear range.
Ultrapure lipopolysaccharide from E. coli O111:B4 (Invivogen) and Pam3CSK4 (Axis-Shield) were used as minimal innate immune stimuli of TLR4 and TLR2 respectively. Both stimuli were prepared in RPMI with 5% NHS. MDM cultures were stimulated for 0-60 min for detection of innate immune cellular activation and 3-24 hours for detection of transcriptional responses and cytokine production. In selected experiments MDM were stimulated with 10 ng/ml recombinant human Interferon (IFN)γ (Peprotech) 24 hours before or after infection with HIV-1.
Cell lysates from MDM cultures were collected directly into SDS sample buffer (62.5 mM Tris HCl, pH 6.8, 10 % Glycerol, 2 % SDS, 0.01 % bromphenol blue and 5 % 2-mercaptoethanol), sonicated and heated (100°C for 5 min) before polyacrylamide gel electrophoresis (4-12% gradient gels) and transfer on to Amersham Hi-bond™ membranes (GE healthcare). Membranes were blocked for 1 h in 5% milk powder in Tris-buffered saline (TBS) with 0.05% Tween-20 (Sigma-Aldrich) and then immunoblotted sequentially with primary antibody overnight (4°C), biotin-conjugated secondary antibody for 2 h (room temperature) and horseradish peroxidase-conjugated streptavidin for 1 h (room temperature), all prepared in TBS/Tween with 1 % milk powder. Membranes were washed with TBS/Tween after each step. Immunostains were developed with Amersham ECL™ reagent (GE healthcare) and visualized on Amersham Hyperfim™ ECL (GE healthcare) according to manufacturer's instructions. Mouse anti HIV-1 gag (p24) antibody (E365/ 366, NIBSC, UK), rabbit anti IκB-α (Cell Signalling Technology), rabbit anti phosphorylated Erk1/2 (clone 197G2, Cell Signalling Technology), rabbit anti phosphorylated p38 MAP kinase (Cell Signalling Technology) and rabbit anti actin (Sigma-Aldrich) were used as 1° antibodies. Biotin conjugated sheep anti mouse IgG and sheep anti rabbit IgG (Dako) were used as secondary antibodies. HRP conjugated streptavidin was obtained from R&D systems.
Immunofluorescence staining of NF-κB RelA (p65) was performed as described previously36. HIV-1 infected and control MDM cultured on glass cover slips were subjected to immunofluorescence staining with rabbit polyclonal anti NF-κB RelA (C-20) (Santa Cruz Biotechnology) with Alexa-Fluor (AF)655-conjugated goat anti rabbit IgG (Invitrogen), mouse anti HIV-1 Gag (p24) antibody (E365/366, NIBSC, UK) with Alexa-Fluor (AF)488-conjugated goat ant mouse IgG (Invitrogen), and nuclear counter staining with 4,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich). Sub-saturating fluorescence images were captured on a Leica SP2 confocal microscope with a pin hole of 1 Airy (114.5 μm), scan speed of 400 Hz and 4 frame averaging. Sequential image acquisition was used to give separate image files for DAPI (excitation 405 nm, emission 400-450 nm), AF488 (excitation 488 nm, emission 510-530 nm) and AF555 (excitation 543 nm, emission 560–580 nm). Image analysis was performed with Metamorph v7.17 (Molecular Devices) to quantify nuclear: cytoplasmic ratios NF-κB RelA staining and proportion of cells demonstrating positive co-localization of DAPI/ RelA (AF655) staining (correlation coefficient >0.5) as markers of NF-κB nuclear translocation37.
Total RNA was collected from MDM cultures as described above. RNA integrity and concentrations were analysed electrophoretically (Agilent RNA 6000 Nano assay/Agilent 2100 bioanalyzer)38 and used to generate firstly amplified cDNA and subsequently Cy5-labelled cRNA using the Agilent Low RNA Input Linear Amplification Kit. This was mixed in equal measure with Cy3-labelled reference cRNA derived similarly from a universal human reference RNA mix (Stratagene), and then used to hybridise Agilent 4x44K whole human genome cDNA microarrays according to manufacturer's instructions (www.agilent.com). Array images were acquired with Aglient's dual-laser microarray scanner G2565BA (5μ resolution) and signal data were collected with dedicated Agilent Feature Extraction software (v9.5.1). Log2 transformed Red (Cy5) and green (Cy3) data from each channel were first normalised separately, to give equal mean and standard deviation. Preliminary analysis of these normalised data showed that there were no significant differences when comparing any gene on the reference channel between all samples. No further normalisation was therefore carried out on the data. Gene expression values were compared by significance analysis of microarray (SAM) using MultiExperiment Viewer v4.039. Significant differences in gene expression across three separate experiments (with different donor cells) were identified using a 1% false detection rate (FDR). SAM tests were performed both with and without pairing of individual experimental/donor samples. Gene lists of interest were annotated using the functional annotation clustering module of the on-line bioinformatics database, DAVID (http://david.abcc.ncifcrf.gov)40,41. In this analysis gene lists were restricted to those with refseq accession numbers for which contemporary functional annotation is available. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MEXP-1904.
The concentration of cytokines in MDM culture supernatants was quantified using a cytometric bead array inflammation kit for tumour necrosis factor (TNF)α, IL-1, IL-6, IL-8, IL-10 and IL-12p70 (Becton Dickinson) using the FACSArray Bioanalyzer System (Becton Dickinson), according to manufacturer's instructions.
The first objective of this study was to establish a sustainable model of productive HIV-1 infection in macrophages. Day 6 MDM cultures, inoculated overnight with the CCR5 tropic strain Ba-L (MOI 3-5), and subsequently stained for intracellular HIV-1 gag (p24) 7 days later, typically showed 100% p24-positive cells (Figure 1A). All subsequent experiments were performed at this time point. On occasion, there was incomplete (30-70%) p24-positive staining of cell cultures. Therefore we performed this assessment of successful HIV-1 infection of MDM in all our experiments and excluded heterogeneously infected (as judged by p24-positivity) cell cultures from this study. Productive HIV-1 infection in this model was demonstrated further by reverse transcriptase (RT)-PCR detection of spliced HIV-1 transcript (Figure 1B) in total RNA samples and by detection of precursor (Pr)55gag protein (Figure 1C) using immunoblotting of protein lysates from HIV-1 Ba-L inoculated MDM cultures 7 days after infection. In this model, MDM were not able to support productive HIV-1 infection with the prototypic CXCR4-tropic strain, NL4-3. HIV-1 spliced transcript and Pr55gag protein were not detected, and p24 staining was positive in <1% of cells in NL4-3 inoculated cultures. There was no appreciable cytopathic effect of HIV-1 infection in MDM, demonstrated by comparable mitochondrial function in uninfected cells and HIV-1 infected MDM 7 days after infection (Figure 1D), an observation which is in keeping with the majority of previous studies42.
The range of innate immune cellular activation pathways employ a complex network of receptors, adaptor molecules and kinases which appear to converge onto selected intracellular signalling events, detectable by Western blotting. Therefore we used this strategy to examine degradation of IκBα, and phosphorylation of p38, Erk1/2 and JNK in a time-course study of MDM stimulated with LPS (Figure 2A). In uninfected MDM the IκBα signal is diminished at 20 and 60 minutes, and regenerated by 120 minutes. Rapid p38 and increased Erk1/2 phosphorylation is also evident, but we did not detect JNK phosphorylation (not shown). This pattern of signalling events was replicated exactly in control MDM cultures inoculated with HIV-1 NL4-3, a strain that is unable to establish host cell infection. In HIV-1 Ba-L infected MDM, p38 and Erk1/2 phosphorylation was comparable to controls, but degradation of IκBα was attenuated, albeit with the same time course profile. These results support the hypothesis that HIV-1 infection of MDM may have a selective inhibitory effect on innate immune signalling pathways down-stream of the LPS receptor complex and MyD88 adaptor protein. IκBα binds to the NF-κB p50-RelA heterodimer and inhibits the transcription factor activity of RelA, in large part by promoting its cytoplasmic sequestration. In the classical NF-κB activation pathway, IκBα is phosphorylated by the IκB kinase (IKKα) and is degraded by ubiquitination, allowing for nuclear translocation of the NF-κB complex. Therefore we performed quantitative confocal immunofluorescence assays of NF-κB RelA nuclear translocation, to assess the functional consequence of attenuated IκBα degradation in HIV-1 infected MDM.
RelA exhibits mostly cytoplasmic staining in control and HIV-1 Ba-L infected MDM before stimulation (Figure 2B). In response to increasing concentrations of LPS, greater nuclear staining is clearly evident. Quantitative image analysis, by measurement of the ratio of nuclear: cytoplasmic RelA staining and the proportion of cells showing nuclear RelA staining, was then used to compare the dose-response to LPS in HIV-1 infected and uninfected MDM cultures (Figure 3A-D). We found that nuclear translocation of RelA was attenuated significantly in HIV-1 infected MDM across the LPS dose range (1-100 ng/ml). This effect was evident in 9/10 separate experiments with cells from different donors (Figure 3E). However, testing multiple donors in this way provided a powerful illustration of the natural variance of this response and the effect of HIV-1 infection. The proportion of cells showing RelA nuclear translocation in control (uninfected) MDM cultures stimulated with 10 ng/ml LPS ranged 17-97%, and reduced by a mean value of 28.9% (95% confidence interval of 9.8-46.8% inhibition) in HIV-1 infected MDM. The specificity of our observations was also tested in MDM infected with the dual tropic HIV-1 strain 89.6 and a synthetic TLR2 stimulus (Pam3CSK4) and showed similarly attenuated NF-κB RelA nuclear translocation (Figure 3F), suggesting that HIV-1 dependent attenuation of NF-κB activation in response to innate immune stimuli may show broad strain and stimulus specificity.
Innate immune responses by macrophages are known to be augmented by IFNγ from T cells in the classical TH1 type paradigm for adaptive immune responses. Therefore, we also investigated the effect of IFNγ priming on activation of the NF-κB pathway in HIV-1 Ba-L infected MDM. We found that RelA nuclear translocation in HIV-1 infected MDM was enhanced by pre-stimulating MDM (24 hours earlier) with 10 ng/ml IFNγ and equivalent to control (HIV-1 uninfected) MDM (Figure 4A-B). In view of this finding we considered the possibility that some variability in HIV-1 mediated inhibition of NF-κB activation may be due to the presence of IFNγ, potentially as a result of minor T cell contamination of MDM cultures. We therefore analysed all cell culture supernatants from the experiments performed in this study for IFNγ but found none detectable by ELISA at a sensitivity of 10 pg/mL (data not shown). We also considered the mechanism by which IFNγ attenuated NF-κB activation in response to LPS is corrected in HIV-1 infected cells. This may be due to priming of the innate immune activation pathway directly or by an effect of IFNγ on HIV-1. Comparison of LPS induced nuclear translocation of NF-κB RelA in uninfected MDM with and without IFNγ pre-stimulation showed no significant differences (Figure 4C-D), suggesting that there is no direct effect independent of HIV-1. However, in keeping with previous reports43, we found that addition of IFNγ to HIV-1 Ba-L infected MDM cultures potently inhibited HIV-1 replication (Figure 4E-F), supporting the hypothesis that IFNγ priming of NF-κB activation in this model is mediated through an inhibitory effect on HIV-1. Further detailed study of this mechanism is not addressed here.
Innate immune stimulation of MDM is known to invoke wide-ranging transcriptional responses44, mediated by NF-κB and other transcription factors. In view of the apparent attenuation of NF-κB activation in HIV-1 infected MDM, we tested the hypothesis that resultant transcriptional responses would also by affected. Therefore genome-wide transcriptional profiling was performed in unstimulated cells, and after 3 h and 24 h stimulation with LPS (10 ng/ml), to make a comprehensive assessment of the effects of HIV-1 on this response. Significant gene expression changes were identified across 3 separate (different donor) experiments by SAM (1% FDR) comparison of transcriptional profiles from unstimulated MDM with those from 3 h and 24 h stimulated cells, for HIV-1 infected and uninfected cultures separately. Gene expression changes were clearly greater at 3 h (approximately 400 genes up-regulated and 100 genes down-regulated) when compared with 24 h (approximately 40 genes up-regulated and 15 genes down-regulated). The overall profile of significant changes in gene expression levels, as assessed by frequency and magnitude compared to unstimulated cells, was comparable in HIV-1 infected and uninfected MDM (Figure 5A-B, and supplementary figure S1). The transcriptional responses to LPS were then interrogated qualitatively by aligning all significantly affected genes in an expression matrix showing mean fold change in 3 or 24 hour LPS stimulated cells compared to corresponding HIV-1 infected or uninfected unstimulated MDM (Figure 5C). Functional annotation of these genes was performed by identifying statistically over-represented gene ontology clusters using the on-line DAVID bioinformatics database 40,41 and indicating the alignment of individual genes within the expression matrix. This analysis shows LPS induced upregulation of a wide repertoire of genes that cluster together within functionally related immune response ontology groups, which demonstrate highly significant enrichment in comparison to the whole human genome (Figure 5C, supplementary figure S2 and supplementary table 1). By contrast, LPS-induced down-regulated genes cluster within ontological groups that are not directly related to immune responses, and not significantly enriched. Genome-wide transcriptional responses of HIV-1 infected and uninfected MDM after LPS stimulation for 3h or 24 h appear to be similar in this analysis also (Figure (Figure5C5C and 6A-B).
In order to assess the effect of HIV-1 infection on LPS responses directly, the expression levels of the cumulative set of 581 LPS responsive (up and down-regulated) genes were compared in HIV-1 infected and uninfected MDM at 3 h and 24 h after LPS stimulation (Figure 7). Statistically significant differences in gene expression were identified in 85 of the LPS responsive genes by paired testing of HIV-1 infected and uninfected MDM from individual donors. The major effect was attenuation of a proportion of up-regulated LPS responsive genes in HIV-1 Ba-L infected MDM compared with controls (Figure 7A-B) and the extent of attenuation varied between 20 and 70% of control values, corresponding to 1.2-3 fold reduction in expression levels. Alignment of these genes with the gene ontology groups identified by the functional annotation clustering analysis of LPS responsive genes showed that LPS responses attenuated by HIV-1 included a range of pro-inflammatory cytokines and immune-related genes (Fig 7A-B). These differences were not evident when comparing HIV-1 infected and uninfected MDM as two groups without pairing individual donor samples, suggesting that differences in gene expression attributable to HIV-1 infection were smaller than the variability between different donors.
The release of inflammatory cytokines represents one of the major functional outcomes of macrophage innate immune responses. It was therefore of significant interest that expression levels of IL-1, IL-6, IL-8 and IL-12 were included amongst the LPS transcriptional responses attenuated by HIV-1 infection. In order to establish whether these transcriptional response differences were translated to significant differences in protein levels, we quantified inflammatory cytokine concentrations in culture supernatants collected from the same transcriptional profiling experiments in unstimulated, 3 and 24 h LPS stimulated MDM with and without 7 days HIV-1 infection (Figure 8). In addition to the cytokines highlighted by the transcriptional profiling differences, TNFα and IL-10 levels were measured, as well established components of MDM cytokine responses to LPS that were not found to have attenuated gene expression levels in HIV-1 infected MDM. Significant concentrations of IL-8, but none of the other cytokines tested, were detectable in unstimulated MDM culture supernatants and were unchanged by HIV-1 infection. Concentrations of TNFα, IL-6, IL8 and IL-10 clearly increased following LPS stimulation, but were equivalent in HIV-1 infected and control samples. Similarly, no significant difference was evident in levels of IL-1β, which increased just above the detection threshold in response to LPS. IL-12p70 was not detected in any samples, an observation in keeping with previous analyses of this MDM model45,46. We undertook additional experiments to measure the same cytokines in culture supernatants after 6 h stimulation with LPS. These experiments also showed no significant difference attributable to HIV-1 infection (data not shown).
In the present study we aimed to address the hypothesis that HIV-1 infection of macrophages inhibits innate immune cellular activation and may therefore have significant effects on down-stream immunological and host-defence responses, and hence contribute to immunodeficiency.
An ex vivo model was established using replication competent HIV-1 for uniform infection of MDM cultures. This model avoids some of the limitations of previous reports using recombinant HIV-1 proteins, expression vector systems with individual HIV-1 components, myeloid leukaemia cell lines used as models for macrophages and primary cells from HIV-1 infected patients which generally, are infected with HIV-1 at low frequency47. We principally used the long established laboratory CCR5 tropic HIV-1 Ba-L strain and found that we were able to generate reproducible productive infection without cytopathic effect. This strain was favoured over primary CCR5-tropic HIV-1 strains because although such primary strains are readily accessible, they are not easily propagated to adequate high titre virus required for the experimental scale used here. HIV-1 Ba-L is not available as a molecular clone, but is propagated in PBMC and therefore prone to genetic drift. Like some others laboratories48, we alternately passage Ba-L in MDM in order to retain its macrophage tropism and exert some selection pressure against significant genetic mutations or deletions. Selected accessory protein genes (nef, vpu, vpr), that have been previously reported to interact with cellular innate immune signalling pathways, were sequenced from infected MDM samples to ensure that they were expressed and did not show marked differences with published Ba-L sequences. It is also important to note that in ex vivo models, MDM display heterogenous biological properties as a result of alternative differentiation protocols49. We used M-CSF differentiated MDM, that are known to support HIV-1 replication better than GM-CSF differentiated MDM50-52, but typically show less inflammatory capacity53. This feature is illustrated by the reported deficiency of IL12 cytokine secretion by LPS stimulated M-CSF differentiated MDM that is also evident in our results54.
Innate immune responses to LPS were studied as this is the best characterised and most potent minimal stimulus for MDM, which is known to signal mainly through TLR4. The effect of HIV-1 infection on TLR signalling pathways, genome wide transcriptional responses and inflammatory cytokine secretion was investigated. Our results show that established HIV-1 infection in MDM has a selective inhibitory effect on the classical NF-κB activation pathway, reflected in diminished IκBα degradation and NF-κB RelA nuclear translocation. This effect is also evident in MDM infected with an alternative HIV-1 strain 89.6 and alternative innate immune stimulation with Pam3CSK4, a TLR2 ligand. By contrast the MAP kinase pathway involving phosphorylation of p38 is not affected. These findings are consistent with existing data in other model systems that show inhibitory interactions between the HIV-1 accessory proteins55. However, inhibition of NF-κB activation is not complete. The inhibitory effect exhibits considerable experimental/donor variability and is rescued by prior exposure to IFNγ, possibly reflecting the strong inhibition of viral replication induced by this cytokine. The attenuation of innate immunity may therefore be aggravated by the widespread deficiency of TH1 responses that is the hall mark of progressive HIV-1 immunodeficiency. Importantly we did not detect any endogenously produced IFNγ that may have been released by contaminating T cell populations and confounded our results.
The inhibition of the NF-κB signal was associated with attenuation of the transcriptional response to LPS. However, this decrease was restricted, both in terms of the number of LPS-responsive genes affected (85 of 581 gene expression changes), and in the magnitude of the effect. Indeed, this effect was smaller than the inter-donor variability observed, since the differences could only be detected using paired sample analysis (with/without HIV-1 infection in cell cultures from same experiment/donor). The differences in transcription observed for three key inflammatory cytokines, IL1β, IL6 and IL8 could not be detected at protein level, suggesting that either the magnitude of transcriptional differences detected are not biologically significant or the existence of compensatory post-translational regulatory mechanisms, which are acting to preserve the core innate immune response. Furthermore, genome-wide bioinformatic analysis shows that the functional clusters of genes in the innate immune transcriptional response to LPS are substantially the same in HIV-1 infected and uninfected MDM, suggesting that from a systems perspective LPS responses are preserved in our model of HIV-1 infected MDM. The classical NF-κB activation pathway is thought to be a key signalling pathway for innate immune activation of macrophages and can regulate inducible expression of wide ranging genes involved in immune responses and cell cycle regulation, including apoptosis56. The effects of HIV-1 infection on LPS induced gene expression in this model were therefore unexpectedly modest. Our observations suggest that sufficient redundancy may exist in innate immune signalling to compensate for relative inhibition of the NF-κB pathway in macrophages. This hypothesis is supported by the finding that other signal transduction pathways, such as the MAP kinase pathway involving p38 phosphorylation57, that contribute to innate immune responses are not affected by HIV-1 infection in our model.
Our results contrast with reports of deficient TNFα responses to LPS in the myeloid leukaemia HIV-1 infected U1 cell line compared with the uninfected parental U937 cell line58,59. This difference is most likely to reflect differences in the biology of leukemic cell lines and MDM60, or heterogeneity between U937 sub-clones. Indeed, the variability observed in our extended experiments using cells from 10 different donors highlight the limitations of cell lines to model host-pathogen interactions in a physiologically relevant fashion. The results presented here are not incompatible with many reports that show impaired innate immune responses in ex vivo MDM or alveolar macrophages from HIV-infected patients. These cells are likely to harbour only low frequency HIV-1 infection61-64. Therefore their phenotype is unlikely to be due directly to HIV-1 infection but result from the in vivo effects of HIV-1 on other components of the immune system. For example, the effect of IFNγ in our experiments, shows clearly the close interrelationship between adaptive and innate immunity. T-cell depletion in progressive HIV-1 infection is therefore very likely to compromise innate immune activation of macrophages.
In conclusion, our study shows preservation of innate immune transcriptional responses by HIV-1 infected macrophages in genome-wide functional clustering analysis, but also supports the hypothesis that HIV-1 infection of macrophages impairs both the major pathway of innate immune cellular activation and a subset of subsequent transcriptional responses, and leaves open the extent to which the ability of HIV-1 to manipulate macrophage immune responses may contribute to immunodeficiency and viral persistence. Further studies are in progress to investigate the ability of the HIV-1 infected macrophages to respond effectively to the wider range of minimal innate immune stimuli as well as bacterial, fungal and viral co-pathogens.
Publisher's Disclaimer: This is an author-produced version of a manuscript accepted for publication in The Journal of Immunology (The JI). The American Association of Immunologists, Inc. (AAI), publisher of The JI, holds the copyright to this manuscript. This manuscript has not yet been copyedited or subjected to editorial proofreading by The JI; hence it may differ from the final version published in The JI (online and in print). AAI (The JI) is not liable for errors or omissions in this author-produced version of the manuscript or in any version derived from it by the United States National Institutes of Health or any other third party. The final, citable version of record can be found at www.jimmunol.org.