Astrocytes as an HIV Reservoir: Mechanism of HIV Infection

Abstract

If we have any hope of achieving a cure for HIV infection, close attention to the cell types capable of getting infected with HIV is necessary. Of these cell types, astrocytes are the most ideal cell type for the formation of such a reservoir. These are long-lived cells with a very low turnover rate and are found in the brain and the gastrointestinal tract. Although astrocytes are evidently resistant to infection of cell-free HIV in vitro, these cells are efficiently infected via cell-to-cell contact by which immature HIV virions bud off lymphocytes and have the ability to directly bind to CXCR4, triggering the process of fusion in the absence of CD4. In this review, we closely examine the evidence for HIV infection of astrocytes in the brain and the mechanisms for viral entry and regulation in this cell type, and discuss an approach for controlling this viral reservoir.

INTRODUCTION

The current focus of HIV research has shifted towards the possible eradication of HIV reservoirs, prevention of their formation by early treatment or keeping them in a dormant state once they are formed. While much attention has been devoted to resting CD4 T cells as a reservoir, the role of other cell types as a reservoir has not been well studied. There is abundant evidence that HIV can enter the brain early in the course of infection and establish a reservoir in resident microglia and astrocytes. Unless we develop an understanding of the mechanism of HIV entry and replication in these cell types, we will never be able to develop therapies for controlling the reservoirs. The current review is focused on HIV infection of brain astrocytes; however, it is important to note that these cells can also be found in the gastrointestinal tract [1–3] which may be another site of HIV persistence.

HIV EARLY AND PERSISTENT INFECTION OF THE BRAIN

Early infection of the central nervous system (CNS) was reported in patients with primary HIV infection in the 1980s (4–6) and HIV was isolated from the cerebrospinal fluid (CSF) of these patients [4, 5]. This was further confirmed in two patients with iatrogenic HIV inoculation, in the brains of whom the viruses were recovered 15 days post-infection [7, 8]. A recent study showed that HIV RNA was detected in the CSF from 15 of 18 subjects as early as 8 days after estimated HIV transmission, with an average detection period of 15 days after exposure [9]. In the macaque model of simian immunodeficiency virus (SIV) infection, the virus could be isolated from CSF within one week and the proviral DNA/viral RNA was detected in brain tissues within 2 weeks post-intravenous inoculation [10–13]. Similarly, early infection of the CNS is also observed in the cats with feline immunodeficiency virus (FIV) [14]. Viral replication may be low or suppressed in the brain during the asymptomatic period after primary infection [4, 11]; however, the viral DNA remains stable from acute through asymptomatic infection [11, 15, 16]. Also, low levels of CSF HIV-1 RNA can even be detected in 17% (12 of 70) of the CSF samples taken after up to 10 years of suppressive antiretroviral therapy [17]. These findings strongly indicate that the brain is an important reservoir for HIV where the virus establishes residence soon after infection.

ASTROCYTE AS ONE OF HIV TARGET CELLS IN THE BRAIN

Major target cells for HIV infection in the brain are perivascular macrophages, microglia and astrocytes, and both microglia [18–20] and astrocytes [21] have an extremely low turnover rate. Cellular immune response to HIV infection in the CNS is inefficient and many antiretroviral drugs cannot efficiently penetrate through the blood-brain barrier [22]. No significant cytopathic effect (CPE) has ever been observed in HIV-infected astrocytes [23, 24] or in microglia at the asymptomatic stage of HIV infection. Both of these cells maintain persistent HIV infection or latency in the CNS. Therefore, as a critical sanctuary or reservoir of HIV, the CNS becomes an insurmountable obstacle for an HIV cure.

Astrocytes are the most abundant cell type in the brain that outnumber neurons by 5–10 fold [25, 26]. Even if HIV only infected a very small percentage of astrocytes, it would still significantly contribute to establishing a reservoir in the brain. HIV infection in astrocytes has been a controversial topic for many years. In part this may be attributed to the difficulty in detection of HIV infection in vivo due to low level of viral replication in these cells and the inefficient infection in astrocytic cell lines in vitro, properties of which are quite different from primary astrocytes. This review mainly focuses on the mechanisms by which HIV infects primary astrocytes.

IN VIVO EVIDENCE OF HIV INFECTION IN ASTROCYTES

Although numerous studies show that HIV infection in the brain mainly occurs in macrophages and microglia [27, 28], in vivo evidence indicates that astrocytes can also be infected by HIV [29]. A late 1980s study detected HIV-1 structural protein p25 in astrocytes from patients with AIDS dementia complex by immunohistochemistry [30]. However, since the authors used only morphological features to identify cell type instead of specific markers, it was not conclusively proven that the stained cells were indeed astrocytes. Stronger evidence was provided in subsequent investigations by applying immunohistochemical staining to detect both glial fibrillary acidic protein (GFAP) and HIV early/structural proteins or applying in situ hybridization (ISH)/PCR-ISH to detect HIV mRNA/proviral DNA as well [31–36]. These studies convincingly confirmed that HIV does infect astrocytes in vivo. Further evidence was added by detecting HIV DNA in astrocytes in brain tissues using immunohistochemistry, laser capture microdissection (LCM) and PCR techniques [16, 37, 38]. Churchill and colleagues showed that the frequency of HIV-infected astrocytes correlated with the severity of neuropathological changes and proximity to brain blood vessels and perivascular macrophages [38]. These findings are somewhat controversial since the possibility of contamination during nested/triple PCR or from surrounding cellular material cannot be excluded [39]. Nevertheless, detection of HIV DNA in astrocytes across all brain regions examined, even those distant from other HIV-susceptible cell types, strongly suggests that these cells can be infected under the right circumstances. In addition, the infection of astrocytes is also observed in SIV macaque models [40, 41]. SIV proviral DNA or early proteins Nef and Rev can be detected in the ex vivo cultures of astrocytes isolated from the macaques infected with SIV for 1 – 3 years [42]. Although it has been difficult to accurately estimate the HIV infection rate of astrocytes in vivo [29, 38], this ex vivo study detected SIV Nef and Rev in 0.05% to 0.1% of GFAP-positive cells [42]. In conclusion, HIV infection of astrocytes is strongly supported by cumulative in vivo and ex vivo evidence. The frequency of infection appears to be quite low overall but may be enhanced in certain conditions, as has been observed in post-mortem tissues from encephalitis and dementia patients.

CAPACITY OF ASTROCYTES TO SUPPORT PRODUCTIVE HIV INFECTION

Although there is little doubt that astrocytes are infected by HIV based on the in vivo evidence, the debate continues whether HIV generates productive or abortive infection after its entry into astrocytes [43]. Some studies show that HIV/SIV structural proteins are detected in the infected astrocytes [30, 32, 41]; however, most studies, both in vivo and in vitro, demonstrate that only viral early gene products (e.g., Nef) can be reliably detected [31, 36, 40] and productive infection is only transiently observed after the infection [44–46]. This indicates that astrocytes might have some intrinsic restriction mechanisms to limit HIV production. However, no obstacle is observed and persistent infection can be established very well when primary astrocytes are infected with VSV-G pseudotyped HIV [47, 48] or cell lines of astrocytic origin are transfected with HIV proviral plasmids [23, 49]. Productive infection with CPE can be extensively generated in the astrocytic cell lines that express CD4 and co-receptors CXCR4/CCR5 [50–52]. Our data also confirmed that persistent infection and stable viral production could be easily established in primary astrocytes when the cells were infected with VSV-G pseudotyped HIV or transfected with HIV proviral DNA, and similar results were further generated from primary astrocytes that were pre-transfected with CD4 plasmid [T4-pMV7 [53]] and then infected with cell-free HIV. Recent studies [48, 54] as well as our data showed that productive, persistent infection can be established in primary astrocytes when HIV particles or viral RNA are able to escape from degradation in endosomes/endolysosomes and get released into the cytosol in presence of lysosomotropic agents (e.g., chloroquine). Thus, it can be concluded that astrocytes do not have significant barriers to the intracellular process of the HIV life cycle.

RESTRICTION AND LATENCY OF HIV INFECTION IN ASTROCYTES

Several groups have identified post-entry blocks to productive HIV replication in astrocytes, focusing primarily on expression of cellular co-factors for viral proteins and host restriction factors. These include low expression of Sam68, a Src-associated protein that is required for Rev function [55, 56]; high expression of the Rev-interacting domain (Risp) protein family, which inhibit Rev function [57]; low levels of TRBP, an inhibitor of double-stranded RNA-dependent kinase PKR [58–61], as well as the Rev-interacting RNA helicases DDX1/DDX3 [62–64]; and abundant expression of TCF-4, the downstream effector of the canonical Wnt/β-catenin pathway [65–68]. However, it is worth noting that these studies were primarily conducted in in vitro culture systems with astrocytic cell lines or fetal cells and may not accurately represent characteristics of astrocytes in vivo. Indeed, few studies have shown that Rev function is unimpaired in human fetal astrocyte cultures [54, 69] and that Rev-dependent transcripts from transiently transfected plasmids are efficiently transported into the cytoplasm of astrocytes [70, 71]. Several groups reported persistent, productive HIV infection in astrocytes when entry restrictions were bypassed by transfection [54, 72] or pseudotyping of HIV [47, 54, 73]; these findings are supported by our own data. Therefore, while the majority of in vitro and ex vivo studies support restricted HIV infection in astrocytes, it is evident that these cells are capable of supporting productive replication. There is also significant evidence that viral replication can be significantly enhanced in nonproductively infected astrocytes under appropriate circumstances such as exposure to pro-inflammatory cytokines (TNFα, IFNγ, IL-1β) [29, 46, 66, 74–76] or lysosomotropic agents [48, 77]. Thus, HIV infection in astrocytes may be restricted by a variety of intracellular mechanisms when the infection is very low; however, the restrictions can be completely overcome when barrier of HIV entry is bypassed.

There is a convincing body of literature that indicates astrocytes support latent HIV infection that can be reactivated from dormancy to produce infectious virions [38, 74, 78–81]. In fact, latently infected astrocytes can transmit HIV to activated lymphocytes in co-culture [24, 78, 81]. Using a variety of latency-reversing agents (LRAs), Narasipura and colleagues recently reported that the HIV LTR in chronically infected progenitor-derived astrocytes is silenced by epigenetic changes, particularly histone modifications performed by class I histone deacetylases (HDACs) and the histone trimethyltransferase (HMT) SU(VAR)3–9 which trimethylates histone 3 at lysine 9 (H3K9) [81]. Interestingly, other LRAs such as 5-aza-2-deoxycytidine (5-aza-CdR), BIX-01294 (a specific inhibitor of G9a, an H3K9 dimethyltransferase) and UNC0628 (an inhibitor of both G9a and GLP, an H3K9 monomethyltransferase) had no effect on HIV expression in astrocytes. In contrast, 5-aza-CdR and BIX-01294 have been shown to reactivate latent HIV replication in T-cell (ACH-2, CEM) and monocytic (OM10.1) cell lines [82]. Additionally, BIX-01294 reactivated HIV in resting T cells isolated from HIV patients on HAART [83]. These findings are particularly interesting because they suggest the epigenetic modifications that regulate HIV latency are not uniform across cell types, which may be relevant to either future eradication strategies or enforced viral quiescence in the CNS. However, Chauhan and colleagues showed that latency-reversing agents actually had no effect on HIV reactivation in astrocytes persistently and productively infected with VSV-G pseudotyped HIV [84]. We also found no increase of HIV production by HDAC inhibitors in long-term, productively HIV-infected astrocytes, in which the infection was established with treatment of chloroquine (unpublished data). Therefore, HIV latency may exist in some circumstances in which the infection is extremely low.

RECEPTORS MEDIATING HIV INFECTION IN ASTROCYTES

Unlike microglia that express CD4 and chemokine coreceptors CCR5 and CCR3 for HIV infection [85], no detectable level of CD4 has been demonstrated in astrocytes [86, 87] and anti-CD4 antibodies or soluble CD4 (sCD4) cannot block HIV infection [46, 74, 88, 89]. While CXCR4 is well expressed on astrocytes [90–92] and can be upregulated by pro-inflammatory cytokines [93], CCR5 is only present under some circumstances. Therefore, other receptors or non-receptor mechanisms have been implicated in HIV entry into astrocytes [46]. These include a CC chemokine receptor, D6, which can function as a co-receptor for various primary dual-tropic strains [94]. An early study showed that antibodies against galactosyl ceramide (galactocerebroside, or GalC) could inhibit HIV-1 entry into a glioma cell line, U373-MG, suggesting that GalC might serve as a receptor for HIV infection in astrocytes [95]; however, this finding has not been confirmed so far. Two independent studies demonstrated that membrane proteins 260 kD or 65 kD in size on the surface of astrocytes were responsible for specific binding to HIV-1 gp120 [86, 96], but these proteins have not been further characterized. Later, another study identified human mannose receptor (hMR), a 165 kD protein, as a mediator of HIV entry into astrocytic cell line where HIV-1 viral particles bound to hMR via the abundant and highly mannosylated sugar moieties of glycoprotein gp120 [97]. Another study indicated that DC-SIGN as well as CCR5 mediates HIV endocytosis in astrocytes [98], however the mRNA of both receptors was barely detectable in primary fetal astrocytes in our study. Altogether, there is still no consensus on the primary receptors responsible for HIV infection in astrocytes.

CD4 mRNA can be detected in astrocytes though CD4 protein is undetectable on the membrane [99]. A residual level of CD4 may be still involved in the process of HIV entry into astrocytes in some circumstances because anti-CD4 antibody can partially inhibit the chloroquine-mediated infection of astrocytes (data not shown; further discussed below). This is also supported by the finding that pretreatment of HIV virions with sCD4 results in a decrease of HIV infection in astrocytes [96].

ENDOCYTOSIS OF HIV IN ASTROCYTES AND THE BARRIER TO VIRAL INFECTION

Viral entry mechanisms can be generally classified as endocytic or initiated by membrane fusion. Receptor-mediated endocytosis and viral escape from the endosome-lysosome pathway serve as a common mechanism for a variety of viruses to enter into specific target cells [100–102]. In the acidic environment of endosomes/lysosomes, low pH triggers a conformational change in the viral envelope that induces fusion of viral and endosomal/lysosomal membranes, resulting in release of viral DNA/RNA into the cytoplasm. Vesicular stomatitis virus (VSV) is one example of a virus that utilizes this mechanism to infect target cells [103]. In the circumstance of HIV infection, membranous fusion classically occurs as a result of receptor engagement, independent of pH [104, 105]. Particularly, the direct fusion of viral envelope with cell membrane is commonly seen in CD4 T lymphocytes with high expression of CD4 and co-receptors [106]. However, there is significant evidence that membrane fusion is not the only avenue of entry for HIV; endocytosis has also been described as a pathway of HIV infection [107, 108], especially in non-lymphocytic cells [109–111], and HIV endosomal fusion can be seen to follow after the endocytosis [111]. This indicates that the low pH may still facilitate HIV endosomal fusion in the cells that express low levels of CD4 or co-receptors [112] though HIV gp120 is not as sensitive to low pH as other viral envelopes such as VSV-G.

The exact mechanism of HIV infection in astrocytes remains elusive. Although endocytosis has been proposed as a pathway of HIV entry into astrocytes [48, 54, 89, 96, 97], the infection is extremely inefficient. This indicates that the majority of internalized viruses are trapped in endosomes/lysosomes and finally degraded there. In general, ligands or viruses internalized via receptor-mediated endocytosis are trafficked to early sorting endosomes, and then rapidly to late endosomes and endolysosomes for degradation [100, 113–115]. Therefore, very few HIV particles are able to escape from degradation in lysosomes and finally gain entry into the cytosol. We observed the full process of HIV endocytosis in astrocytes via electron microscopy and found that infection of astrocytes was significantly enhanced while the cells were simultaneously treated with chloroquine, a lysosomotropic agent that elevates the pH in the lumen of endosomes/lysosomes and therefore impairs the integrity of the endosomal/lysosomal membrane. Although a previous study [99] as well as our data (not yet published) demonstrated that astrocytes express CD4 mRNA and residual levels of its protein, the low pH is still not able to efficiently trigger endosomal fusion while CD4 is only expressed at a minimal level. Thus, the barrier of HIV infection in astrocytes is at entry and the release of viral RNA into the cytosol is blocked in endosomes/endolysosomes where the majority of viral particles are finally degraded.

EFFICIENT HIV INFECTION IN ASTROCYTES CAN OCCUR VIA CELL-TO-CELL CONTACT

Previous studies have shown that astrocytes can be infected with HIV by co-cultivation with HIV-infected lymphocytes [24, 116]. While primary astrocytes cultured with cell-free virus are largely resistant to infection [44–46], infectivity is significantly increased in co-cultures of astrocytes with HIV-infected lymphocytes [117, 118]. This is consistent with studies that show HIV infection by cell-to-cell transmission between lymphocytes is 3 – 4 orders of magnitude more efficient than using cell-free virus [119, 120]. However, the mechanism of cell-to-cell HIV transmission does not appear to be shared across all cell types. Although there are studies showing that cells expressing HIV co-receptors can promote the infection of CD4-negative cells by transcomplementation of receptor deficiency [121, 122], we recently showed that, in the transmission from lymphocytes to astrocytes, HIV may only infect astrocytes before the process of its maturation is completed [117]. Only immature virions enter astrocytes via this route, while entry of mature particles is blocked. This phenomenon can be specifically attributed to a variety of contacts and/or virological synapses formed between astrocytes and lymphocytes [117, 123]. This unique mechanism is CD4-independent but requires CXCR4 and the model of infection has been proposed (Fig. 1) [117]. Classical receptor-mediated fusion occurs via a stepwise process in which binding of gp120 on a mature virion to CD4 on a target cell induces a conformational change in gp120, exposing the CXCR4-interacting domain. The new model proposes that in immature/budding virions the gp120 molecule remains in an “open” conformation that allows interaction with CXCR4, resulting in occurrence of the fusion step in absence of CD4. Thus, cell-to-cell contact might be a possible mode by which HIV can infect astrocytes in vivo.

Fig. (1).

(A) Serial, discontinuous sections of EM were aligned. Multiple tight contacts (block arrow) are observed between HIV-infected JKT cell and astrocyte. The viruses from the infected cell (thin arrow) are seen budding into intercellular spaces or directly onto the membrane of astrocyte. (B) Diagram shows that HIV particles are budding into virological synapses between the areas of tight contacts or directly onto the membrane of astrocyte. The immature, budding viral particles make contact with an astrocyte. (C) A hypothesis is proposed based on the EM observations that CXCR4-binding sites on the surface of the envelope of budding or immature HIV is in an “open” state which allows the virus to directly bind to CXCR4 on the membrane of astrocytes, and hidden following a conformational change that is triggered during HIV maturation. However, the pre-bound virus would trigger the fusion process of HIV envelope with the astrocyte membrane while the maturation is completed, leading to HIV transmission from lymphocytes to astrocytes. This process cannot occur with cell-free mature HIV in astrocytes that lack CD4 expression since the CXCR4-binding sites are hidden in the envelope of mature HIV particles. T – T lymphocyte, A – astrocyte.

Based on these findings, it seems that only X4 or dualtropic R5X4 viruses are able to infect astrocytes via cell-to-cell contact because CCR5 is not detectable in cultured astrocytes [117]. One argument against this hypothesis is that most HIV strains isolated from the brain are R5-tropic [124]; however, some studies have reported that X4-tropic or R5X4 dual-tropic HIV-1 viruses are present in the brain or CSF of patients with HIV-associated dementia [124–127]. Additionally, there is evidence that suggests some R5X4 viruses preferentially use CXCR4 for entry [126, 128–130]. Combined antiretroviral therapy induces a switch in HIV tropism from R5 to X4 usage [131]. This switch may appear late in the CNS compartment compared to the periphery. Also, cell-cell fusion assays show that many bioinformatic prediction programs underestimate CXCR4 usage by R5X4 viruses in the brain [132]. Furthermore, studies show that lymphocytes actually migrate into the brain and the frequency of migration is significantly higher in asymptomatic carriers compared to patients with early or late stage AIDS [133, 134]. The migration of lymphocytes into the brain is also observed in SIV/FIV-infected animal models [135, 136]. HIV-infected and/or immune activated macrophages produce IL-1β that further induces secretion of SDF-1 from astrocytes [137]. Thus, SDF-1 can trigger HIV-infected lymphocytes to migrate through the blood-brain barrier (BBB) and contact with astrocytes leading to HIV infection of astrocytes. This is also supported by the finding that HIV infection of astrocytes occurs predominantly in perivascular regions [38].

PROSPECT

Although studies have shown that infection of astrocytes with cell-free HIV is inefficient due to poor viral entry, cell-to-cell contact can facilitate HIV transmission from lymphocytes to astrocytes leading to productive infection. Further, in patients with HIV-associated neurological disorders the inflammatory environment associated with activation of different factors (e.g., cytokines, chemokines) [93, 138–140] may upregulate expression of HIV receptor CD4 and coreceptors in astrocytes that make the cells permissive to HIV infection. Thus, we propose that HIV infection of astrocytes in the brain occurs by two major mechanisms as depicted in the diagram (Fig. 2). Therefore, in addition to infection by immature HIV virions, cell-free mature HIV may also be able to infect the astrocytes that have been induced to upregulate CD4/coreceptors in vivo.

Fig. (2).

Two major mechanisms that might be involved in HIV infection of astrocytes in vivo are shown. One occurs via the cell-to-cell contact with HIV-infected lymphocytes, by which immature HIV particles are transmitted to astrocytes. The second may require inflammatory factors (e.g., cytokines, chemokines, etc.) produced from HIV-infected perivascular macrophages and/or microglia that up-regulate expression of HIV receptor and co-receptors in astrocytes making the cell (highlighted) permissive to HIV infection. Mø - macrophage

Astrocytes, by the very nature of being long-lived cells that do not undergo any CPE while persistently infected, are an important reservoir for HIV infection in the CNS. The ability of these cells to respond to cytokines with a burst of HIV replication suggests that, unless this reservoir is prevented from being formed, it will be able to reseed the periphery even if a cure is achieved in lymphocytes. Viral entry into astrocytes can be achieved by immature virons where it is capable of engaging co-receptor CXCR4 in the absence of CD4. Thus, blockers of this co-receptor have the potential for preventing the establishment of the astrocyte reservoir. This may be important for the brain as well as the gut since astrocytes are found in both organs.

Biography

Guan-Han Li