Macrophages and HIV-1 – an unhealthy constellation

Abstract

Lentiviruses have a long-documented association with macrophages. Abundant evidence exists for in vitro and, in a tissue-specific manner, in vivo infection of macrophages by the primate lentiviruses HIV-1 and SIV. However, macrophage contribution to aspects of HIV-1 and SIV pathogenesis, and their role in viral persistence in individuals on suppressive antiretroviral therapy remains unclear. Here we discuss recent evidence implicating macrophages in HIV-1-mediated disease, and highlight directions for further investigation.

Macrophages – new insight into origins

Macrophages are myeloid lineage cells of the innate immune system that are required for multiple functions, from tissue homeostasis and repair to sensing and eliminating microbial pathogens and tumor cells (Okabe and Medzhitov, 2015). Macrophages can be found in virtually every tissue in the body, and have been shown to originate from two distinct sources: self-renewing tissue-resident macrophages derived from primitive embryonic precursors, and infiltrating monocyte-derived macrophages (MDM) (Haldar and Murphy, 2014). This knowledge overturns the previous paradigm that tissue macrophages derive substantially from infiltrating monocytes. This also influences how we view macrophage function in relation to location and origin, in both normal physiology and disease such as HIV-1 infection. Macrophage phenotype and function differs between and within tissues, varying from brain microglia to to lung alveolar macrophages. Since most in vitro models of HIV-1 macrophage infection are based on blood-derived MDM, it is timely to re-evaluate the significance of this evidence in relation to exciting in vivo, ex vivo and in vitro experimental findings of the last few years.

HIV-1 infection of macrophages

Macrophage infection by HIV-1 was first described in the 1980s. Ex vivo analyses of post-mortem AIDS patient samples revealed infected macrophages in multiple tissues, particularly frequently observed in brain, lung and secondary lymphoid tissues. More recent studies have revealed infection of tissue macrophages at all stages of disease, which persists under combination antiretroviral therapy (cART) (Cory et al., 2013). These ex vivo data are consistent with in vitro models of MDM infection. Macrophages express the entry receptors required for HIV-1 infection - CD4, and the viral entry chemokine coreceptors CCR5 and CXCR4. Despite CXCR4 expression however, most CXCR4-using (X4) viruses are unable to establish a productive infection in macrophages. CCR5 expression is therefore important, but is not the only criterion for macrophage tropism. Increased affinity of the viral envelope glycoprotein Env for CD4 and more efficient exposure of the CCR5-binding bridging sheet upon CD4-gp120 engagement correlate with enhanced macrophage tropism (Arrildt et al., 2015; Mefford et al., 2015; Musich et al., 2015; Salimi et al., 2013).

Viruses that initiate inter-host transmission, so-called transmitted/founder (T/F) viruses, are very weakly macrophage tropic (Ochsenbauer et al., 2012). This finding is consistent with ex vivo data that CD4⁺ T cells and not macrophages are the primary viral targets at very early time points after transmission (King et al., 2013; Li et al., 2009). Moreover, these results are broadly consistent with in vitro models that show inefficient infection of MDM compared to T cells resulting largely but not entirely from low-level expression of CD4 and CCR5. Subsequent to receptor-mediated fusion, there are post-entry obstacles that must be overcome before infection can be established. Despite this, macrophages are permissive for infection even in the non-proliferating state (Weinberg et al., 1991). This ability to establish infection in non-dividing cells (Bukrinsky et al., 1992; Lewis et al., 1992) distinguishes primate lentiviruses from other retroviral genera, and HIV-1-based lentiviral vectors are employed specifically to transduce non-dividing cellular targets. That primate lentiviruses have maintained the ability to infect non-dividing cells suggests a central role for macrophage-lineage cells in virus biology.

Macrophage tissue heterogeneity is also manifest by different levels of permissiveness to HIV/SIV. Thus gut-associated lymphoid tissue (GALT) macrophages are relatively refractory to infection (Shen et al., 2011), although this is modulated by precise location since rectal macrophages appear more susceptible (King et al., 2013; McElrath et al., 2013), whereas lung macrophages are relatively permissive (Jambo et al., 2014). The variety of macrophage tissue phenotypes may in part be recapitulated in vitro by exposure to different polarizing cytokines. Unpolarized MDM are relatively susceptible to HIV-1 infection, whereas M1- or M2-polarized macrophages, exposed to pro- or anti-inflammatory cytokines respectively, are relatively more refractory (Cassol et al., 2010). These models may globally relate to in situ tissue macrophage phenotypes at steady state and under inflammatory conditions, although the in vivo situation will probably be more complex.

New models for macrophage infection

Macrophages maintain tissue homeostasis by recognizing and disposing of apoptotic cells in a non-proinflammatory manner, a process termed efferocytosis (Martin et al., 2014). This essential function is carried out during normal ontogeny and development, and its interruption leads to accumulation of dead cells driving inflammation. Macrophages also provide a critical front line of defense against intracellular pathogens including viruses, bacteria and eukaryotic parasites by eliminating pathogen-infected cells. ‘Eat me’ signals for macrophage recognition and capture of infected cells include apoptotic and pyroptotic cell death signals (Martin et al., 2014). In most cases, macrophage encounter with a pathogen-infected cell leads to its engulfment and destruction. However in some cases intracellular pathogens escape from the target cell to infect the macrophage. This is exemplified by the intracellular bacterium Listeria monocytogenes that induces phosphatidyl serine exposure on infected cells, providing an ‘eat-me’ signal to macrophages via the aopoptotic cell recognition machinery leading to infected cell engulfment and macrophage infection (Czuczman et al., 2014). It was recently shown that MDM selectively capture and engulf HIV-1-infected T cells, leading to macrophage infection Figure 1 (Baxter et al., 2014). The signals for infected T cell capture remain to be defined, but cell death appears to be an important component of the recognition process. This mode of viral spread can be very efficient, since MDM were rapidly and robustly infected after transient exposure to infected T cells. Moreover, the weak tropism of T/F viruses for macrophages was overcome leading to robust infection of MDM via this route.

Figure 1. In vitro models of macrophage exposure to HIV-1-infected T cells.

CD4⁺ T cells infected by cell-free HIV-1 (A) or by cell-cell spread from macrophages that are infected or have captured HIV-1 (B) may be engulfed by uninfected macrophages (C). (D) Shows an Imagestream® image of a macrophage that has engulfed multiple Gag⁺ Caspase-3⁺ T cells. This may result in macrophage infection as evidenced by extensive macrophage cytoplasmic Gag label after coculture of infected T cells with macrophages (E), or false positive infection (F) in which the infection does not pass from the T cell to the macrophage, in this case because the macrophage is treated with AZT. Imagestream® images (D, E, F) were obtained by A. Baxter.

Phagocytosis of SIV-infected T cells was reported in the SIV macaque model, demonstrated by the presence of rearranged T cell receptor sequence signatures associated with myeloid cells extracted from macaque immune tissues (Calantone et al., 2014). Two interesting possibilities arise from these observations: i) that HIV-1 may spread efficiently from infected CD4⁺ T cells to macrophages via phagocytosis at sites of high CD4⁺ T cell infection and death; ii) that phagocytes containing HIV-1 nucleic acids and/or proteins may not necessarily be infected, but may simply have engulfed infected cells or their debris (Figure 1). This model might also explain the detection of T cell receptor sequences associated with apparently infected blood and GALT myeloid cells sorted from HIV-1-infected patients on cART (Josefsson et al., 2013).

Macrophage infection and onward viral spread

The efficiency of HIV-1 infection is strongly influenced by the mode of viral spread. Direct transfer of HIV-1 between contacting cells is substantially more effective than the cell-free route. This appears to be true not only for spread of HIV-1 from infected T cells to MDM (above), but also from infected macrophages to other cells. As noted by several groups, in vitro infection of MDM leads to accumulation of infectious particles in a surface-connected vesicular compartment termed the virus-containing compartment (VCC) (Deneka et al., 2007; Jouve et al., 2007; Welsch et al., 2007). Infectious virus may be stored within the VCC for extended periods (Sharova et al., 2005) and then transferred rapidly to contacting CD4⁺ T cells (Giese and Marsh, 2014; Gousset et al., 2008; Groot et al., 2008). Such high multiplicity CD4⁺ T cell infection reduced the efficacy of some antiretroviral drugs and neutralizing antibodies (Duncan et al., 2013; Duncan et al., 2014). Synaptic HIV-1 transfer between T cells extracted from lymphoid tissue has recently been shown to drive abortive infection and pyroptotic cell death rather that productive infection (Galloway et al., 2015). Whether this is also the case for MDM-mediated viral spread to T cells remains to be elucidated. Direct spread between MDM via actin containing nanotubes has also been observed. These projections are formed in a Nef-dependent manner and facilitate viral spread (Eugenin et al., 2009).

A related but distinct phenomenon is transfer of HIV-1 infection in trans – a situation in which virus may be captured by a myeloid-lineage (macrophage or dendritic) cell without necessarily resulting in infection, and subsequently transferred to a target CD4⁺ T cell. This mode of spread has been observed for HIV-1 using in vitro models (Cameron et al., 1992), ex vivo explants (Ballweber et al., 2011) and very recently, in vivo using intravital microscopy (Sewald et al., 2015). Sewald and colleagues demonstrated in both unaltered and human immune system (HIS)-reconstituted mice that sinus-lining macrophages capture HIV-1 virions via viral envelope-associated gangliosides interacting with the lectin CD169. This interaction promoted in vivo infection in HIS mice, which was reduced in the presence of blocking CD169 antibody (Sewald et al., 2015). CD169 capture of HIV-1 promotes macrophage infection (Zou et al., 2011). These results imply that HIV-1 infection may be similarly propagated in vivo by macrophage presentation in trans, or potentially in cis, to CD4⁺ T cells.

Macrophages as viral tissue reservoirs

Macrophages are infected in vivo in a tissue-specific manner. The brain has been recognized as a target organ for macrophage infection since the 1980s. The principal target cells in the brain are microglia and perivascular macrophages, which are infected rapidly after transmission, although the source of infecting virus remains unclear (Rappaport and Volsky, 2015). The difficulty in obtaining tissue from the human central nervous system (CNS) in acute infection has meant that most work has been done on late stage and postmortem tissue, or in the SIV macaque model. Nevertheless, it seems likely that once infected, the CNS is likely to remain so for the lifetime of the host, implying that this is an important viral reservoir (Fois and Brew, 2015). Consistent with this, recent studies have demonstrated that genetically distinct HIV-1 variants exist in the CNS and plasma, and viral RNA can be detected in the CSF of cART-treated patients when undetectable in the plasma (Dahl et al., 2014).

Aside from the CNS, most of the limited information on infection status of human tissue macrophages has been derived from analysis of relatively few anatomic sites; ones that are accessible with minimally invasive procedures. Lung alveolar macrophages, isolated from bronchoalveolar lavage (BAL), are targeted by HIV-1 with implications both for local tissue damage and viral reservoir formation (Costiniuk and Jenabian, 2014). In the SIV macaque model, alveolar macrophages were SIV RNA⁺ within 10 days post-infection onwards, demonstrating rapid seeding of lung tissue, and were the major cell type infected (Li et al., 2015). In this study, macrophage numbers remained relatively constant whereas CD4⁺ T cell numbers declined. The Kuroda lab observed that both interstitial and alveolar macrophages were SIV infected in macaques, however whereas infected alveolar macrophages were long-lived, their interstitial counterparts died rapidly (Cai et al., 2015). Similar to these findings in the SIV/macaque model, a sizeable percentage of alveolar macrophages were found to harbor HIV-1 in untreated viremic patients (Jambo et al., 2014). Fluorescence in situ hybridization revealed macrophages transcribing HIV-1 preferentially in small alveolar macrophages. Analysis of HIV-1-infected cART-treated patients revealed that whilst most of these patients had negative plasma viral loads, 16/23 had detectable proviral DNA in alveolar macrophages, of which 8 had detectable viral RNA (Cribbs et al., 2015). Compartmentalization of HIV-1 infection within the lung of treated patients suggests that this tissue may well comprise an important persistent viral reservoir.

Gut-associated lymphoid tissue (GALT) has long been associated with HIV-1 infection, and is rich in macrophages. As with other tissues, the primary GALT target cells are CD4+ T cells, which are rapidly depleted during HIV-1 or SIV infection. Ex vivo analyses of human GALT macrophages revealed that they were relatively refractory to HIV-1 infection compared to MDM, and this phenotype was found to correspond to exposure in situ to TGF-β which downregulated CD4 and CCR5 expression (Shen et al., 2011). Nevertheless, tissues obtained from infected human duodenum confirmed macrophage infection, even in the context of highly suppressive cART (Zalar et al., 2010). Interestingly, macrophages proximal to the rectum showed increased expression of CCR5 and greater HIV-1 susceptibility compared to colon-resident cells (King et al., 2013; McElrath et al., 2013), reinforcing the tissue-specific nature of the macrophage phenotype.

In a study to examine the relative contribution of central memory and effector memory T-cells to viral persistence in cART-treated aviremic individuals, the amount of total DNA in lymphoid biopsies was found to exceed that measured in CD4⁺ T-cells, suggesting the presence of a non-T-cell reservoir in those tissues (Yukl et al., 2013). Further analysis demonstrated the presence of proviral DNA in macrophages purified from rectal and ileal tissue (Yukl et al., 2014). In another study (Josefsson et al., 2013), quantification of the cellular reservoir in ART-treated patients revealed 0.008–0.03% of myeloid cells isolated from GALT were HIV-1 DNA⁺. However, a complication of this finding as described above was that T cell receptor sequences were associated with the myeloid cells, potentially reflecting uptake of infected T cells by myeloid cells as previously suggested (Baxter et al., 2014; Calantone et al., 2014).

Urethral macrophages have also been reported to maintain infection in the face of therapy. SIV RNA was detected in urethral macrophages from macaques, and the extent of infection was minimally impacted by initiation of cART treatment (Matusali et al., 2015). Consistent with this, urethral macrophages in penile tissue explants from individuals undergoing gender reassignment could be infected by HIV-1 (Ganor et al., 2013).

Contribution of macrophages to HIV-1 pathogenesis

Aside from their role in HIV-1 spread and persistence, infection of macrophages may directly promote disease, the principal mechanism of which appears to be via activation of inflammatory processes. HIV-1 infection of the central nervous system is associated with inflammation leading to pathology ranging from encephalitis to mild but progressive cognitive dysfunction, associated with neuronal death. cART has generally reduced the burden of CNS disease, but low-level residual inflammation resulting from persistent infection of resident macrophages may continue to cause progressive disease.

HIV-1 infection is associated with respiratory dysfunction and increased susceptibility to lower respiratory tract infections. Of relevance, infected alveolar macrophages have impaired phagocytic activity both in cART-treated (Cribbs et al., 2015) and untreated (Jambo et al., 2014) patients, implicating reduced macrophage function in HIV-1-associated lung disorders. Analysis of macrophage infection in the SIV macaque model has revealed that interstitial macrophages, not recovered by BAL and therefore previously unstudied in this context, are targeted by SIV and die, causing inflammatory damage to the lung tissue (Cai et al., 2015).

The GALT is a major target during acute HIV-1 infection, and damage to the immunological and physical barriers results in subsequent microbial translocation driving chronic immune activation. At steady state, GALT macrophages are subject to, and actively maintain, an anti-inflammatory homeostasis by constitutively secreting regulatory cytokines such as IL-10 and TGF-β. Whether this tissue conditioning can be modified in the context of HIV-1 infection remains to be seen. However it would be surprising if exposure to a highly pro-inflammatory local environment consisting of HIV-1-infected CD4⁺ T cells undergoing apoptosis, secondary necrosis and potentially pyroptosis (Galloway et al., 2015), would not modify macrophage polarity towards a more inflammatory M1 phenotype, as observed in untreated AIDS patients (Cassol et al., 2015). Moreover, since inflammation recruits monocytes to the site of tissue damage and polarizes the incoming cells according to the local environment, this would tend to amplify ongoing inflammatory processes in the GALT.

Inflammation associated with chronic HIV-1 infection, treated or not, is a risk factor for cardiovascular disease (CVD), and macrophages are likely to play a central role in the pathogenesis. Studies in humans have revealed correlations between duration of HIV-1 infection, markers of inflammation, the presence of inflammatory macrophages in the vasculature, and CVD (Crowe et al., 2010). The SIV macaque model has allowed more direct insight into the functional of macrophages in CVD: Walker and colleagues used anti-α4 integrin monoclonal antibody (natalizumab) to inhibit monocyte/macrophage traffic, and observed decreased cardiac fibrosis (Walker et al., 2015). These data not only provide a functional correlate but also suggest the possibility of pharmacologically reducing HIV-1-mediated CVD.

A recent finding of relevance to both reservoir formation and pathogenesis of osteolytic bone disease, observed in both ART naïve and treated patients, is the infection of osteoclasts. These are myeloid-lineage cells that can be derived in vitro from blood monocytes, express CD4 and CCR5, and are readily infected by HIV-1 (Gohda et al., 2015). Infected osteoclasts upregulated markers of activation and increased bone resorption activity, suggesting a direct role in bone disease.

Animal models for macrophage infection

The anatomical location and density of the tissues in which macrophages reside makes their isolation and analysis challenging. The SIV macaque model is ideally placed to overcome this challenge, as animals can be infected under precisely defined conditions via relevant routes, and tissues can be sampled at will. In many ways the model reflects human infection and pathogenesis in that SIV uses the same receptors, has the same tissue tropism and causes the same, if somewhat accelerated, disease. However one important caveat of this system relates to the expression of the Vpx accessory protein that is lacking in HIV-1. SAMHD1 is a deoxynucleoside triphosphate triphosphohydrolase that reduces cellular dNTPs to a level that is unfavorable for viral reverse transcription. Vpx is a viral accessory protein encoded by HIV-2 and most SIV strains but not by HIV-1 that promotes proteasomal degradation of SAMHD1 thereby elevating dNTP to levels favorable for reverse transcription (Hrecka et al., 2011; Laguette et al., 2011). The absence of vpx in HIV-1 and the apparent lack of an analogous SAMHD1-counteracting activity in HIV-1 suggest that SAMHD1-expressing cells are likely to be similarly refractory to HIV-1 as vpx-deleted SIV variants. However, this has not been substantiated in parallel comparisons of HIV-1 and SIV for macrophage infectivity. Furthermore, it has been suggested that HIV-1-SAMHD1 co-evolution has led to a reverse transcriptase that can efficiently polymerize in the low dNTP environment created by SAMHD1 (Lenzi et al., 2015). One would predict that Vpx expression dictates the cellular tropism of SIV in the host, but studies on this are conflicting. Using in situ approaches to visualize infected cells, Vpx was found to be essential for SIV infection of macrophages, particularly in GALT, but not for CD4⁺ T-cell decline and development of AIDS (Westmoreland et al., 2014). However when the presence of viral DNA was used to detect infected cells, Calantone and colleagues failed to find a significant difference in mucosal tissue myeloid cell-associated viral DNA in Δvpx SIV compared to wild-type virus-infected animals (Calantone et al., 2014). It is possible that CD4⁺ T-cell phagocytosis by macrophages masked differences in the relative frequency of wild-type and vpx-deleted SIV-infection of macrophages. Further comparison of the SIV macaque model with human samples is required to fully define the role of Vpx in macrophage infection in the host.

An alternative to the macaque model is the humanized immune system (HIS) mouse. First generation HIS mouse models reconstituted the lymphoid compartment with some fidelity, but showed only modest myeloid compartment reconstitution largely because of lack of cross-reactivity between murine myeloid growth factors and human myeloid precursors. More recent attempts have knocked in human innate immune cell growth factors allowing reconstitution of elements of the myeloid tissue compartment (Rongvaux et al., 2014). Future work will determine how suitable these models are for HIV-1 research into tissue macrophage infection and pathogenesis.

Conclusions and perspectives

Macrophages are clearly a highly heterogeneous population of cells with phenotypic characteristics defined by their tissue environment. This phenotypic variability is played out in a diverse range of tissue-dependent susceptibilities to HIV-1 infection, and a variety of ensuing responses that will influence viral persistence, spread and pathogenic outcomes. There is convincing recent evidence that HIV-1 and SIV infection of macrophages establishes a reservoir in brain and lung tissue that persists in the face of cART, bringing further complexity to the concept of viral eradication leading to cure.

A clear lesson from in vitro studies on macrophages is that although informative, they fail to adequately address the missing element of tissue environment, and more reliance on animal models will be required to interrogate this. Nevertheless in vitro model systems have utility in driving in vivo experiments, one example of which is the finding that in vitro, macrophages selectively engulf HIV-1 infected T cells. This finding has implications both for macrophage infection at tissue sites of CD4⁺ T cell infection and for potential false-positive signals relating to phagocytosed infected T cells and debris. Given the potential confounding interpretations surrounding capture and uptake of infected CD4⁺ T-cells by macrophages, approaches to reveal the presence of macrophage reservoirs in individuals on suppressive cART clearly need to extend beyond demonstrating the presence of viral DNA in purified macrophage samples. Might the presence of viral DNA in macrophages be solely a consequence of the ingestion of infected CD4⁺ T-cells as suggested by (Calantone et al., 2014)? Neither in vivo infection of macaques with highly pathogenic SHIV (Igarashi et al., 2001) or SIV (Micci et al., 2014) in which viraemia is maintained by tissue macrophage infection in the absence of CD4⁺ T cells, nor in vitro macrophage infection (Baxter et al., 2014) are consistent with this proposal. These studies combined with numerous others that identify infected cells by in situ and ex vivo approaches, indicate that macrophages support bona fide infection of macrophages in vivo. As the HIV/AIDS research field investigates strategies to eliminate reservoirs that persist in the face of ART, the nature of the cellular reservoirs that sustain viral persistence becomes a central question. Most of the attention in this regard has focused on CD4⁺ T-cell reservoirs in which the virus can reside in a latent state that will be difficult to eliminate (Siliciano and Siliciano, 2015). There is scant information on whether macrophages can similarly maintain HIV-1 in a transcriptionally latent form. Most of the earlier studies employed monocytic cell lines whose physiologic relevance to tissue macrophages is unclear. As such, new studies are needed to determine whether macrophages can sustain HIV-1 latency and whether these cells pose a major obstacle in the pursuit of a cure for HIV-1 infection.

Powerful technologies can now be applied to the in vivo and ex vivo analysis of HIV-1 and SIV-infected macrophages including RNA- and DNAscope® (eg. DOI: 10.1126/science.aac8908), single cell laser capture and single cell sequencing and transcriptomics. These approaches will provide important new information on the role of macrophages as persistent viral reservoirs, the influence of infection on macrophage phenotype and on the tissue in which they are resident. In situ approaches such as RNA- and DNAscope® have the potential to reveal the relative contributions of CD4⁺ T-cells and macrophages to viral persistence under suppressive cART. Use of Scope techology to compare the levels of RNA versus DNA positive cells in individuals on suppressive ART could also be very informative in ascertaining the presence of latently-infected macrophages. Intravital imaging, that has been used extensively to visualize immune system cellular interactions, can now be applied to pathogens such as HIV-1 in the context of HIS mice or macaques, and will reveal details of macrophage infection and its outcome in living tissue.

Figure 2. The VCC in an in vitro infected macrophage.

A monocyte-derived macrophage infected in vitro and labeled for CD4 (green) shows individual virions labeled for Gag (red) within the virus-containing compartment at low magnification by confocal microscopy (box, bar = 2.5 μm) and high magnification by stimulated emission-depletion (STED) super-resolution fluorescence microscopy (main figure, bar = 400nm). Sample prepared by T. Do and images acquired by J. Chojnaki.