Innate immunity in the pathogenesis of polytropic retrovirus infection in the central nervous system

Abstract

Neuroinflammation, including astrogliosis, microgliosis, and the production of proinflammatory cytokines and chemokines is a common response in the central nervous system (CNS) to virus infection, including retrovirus infection. However, the contribution of this innate immune response in disease pathogenesis remains unresolved. Analysis of the neuroinflammatory response to polytropic retrovirus infection in the mouse has provided insight into the potential contribution of the innate immune response to retrovirus-induced neurologic disease. In this model, retroviral pathogenesis correlates with the induction of neuroinflammatory responses including the activation of astrocytes and microglia, as well as the production of proinflammatory cytokines and chemokines. Studies of the neuroviru-lent determinants of the polytropic envelope protein as well as studies with knockout mice suggest that retroviral pathogenesis in the brain is multifaceted and that cytokine and chemokine production may be only one mechanism of disease pathogenesis. Analysis of the activation of the innate immune response to retrovirus infection in the CNS indicates that toll-like receptor 7 (TLR7) is a contributing factor to retrovirus-induced neuroinflammation, but that other factors can compensate for the lack of TLR7 in inducing both neuroinflammation and neurologic disease.

Introduction

The activation of astrocytes and microglia (termed gliosis), along with increased production of proinflammatory cytokines and chemokines is a common finding among virus infections in the central nervous system (CNS), including HIV infection in adult and pediatric patients [1–5]. Similar neuroinflammatory responses are observed in animal models of retrovirus infection [6–13]. This innate immune response is also observed in other neurologic diseases of viral, bacterial, or unknown etiologies. The innate immune response may play an important role in limiting virus replication in the CNS. However, the initiation of the innate immune response may be a two-edge sword, as proinflammatory cytokines and chemokines may impact the function of intrinsic brain cells including neurons. This review discusses the potential influence of the innate immune response to retroviral-induced neuropathogenesis, focusing on studies of the polytropic retrovirus mouse model of neuroinflammation and neuropathogenesis.

Cells of the innate immune response in the central nervous system

The central nervous system (CNS) has limited interactions with the peripheral immune system due to the lack of lymphatics and the presence of tight junctions at the blood–brain and blood–cerebrospinal fluid (CSF) barriers that limit the influx of cells and protein to the CNS in both adults and neonates [14–16]. As such, immune cells, including cells involved in the innate immune response, have limited access to the CNS. In addition, the brain does not have a resident population of dendritic cells to trigger innate immune responses. Thus, the phagocytic cells intrinsic to the brain, including brain macrophages, parenchymal microglia and astrocytes, play important roles in the innate immune response to pathogen infection in the CNS.

Brain macrophages and parenchymal microglia cells are considered to be the primary resident immune cells in the CNS and are bone-marrow derived [17]. Monocytes that differentiate into parenchymal microglia cells generally enter the brain during embryonic and post-natal development [17, 18]. Following entry, microglial cells undergo morphological changes and develop into resting ramified microglia with long processes. When microglia becomes activated, they retract their processes and take an ameboid or macrophage-like morphology. Perivascular macrophages can be distinguished from resting microglia as they are elongated, located in close proximity to capillary endothelial cells, express slightly different cell surface markers including higher levels of CD45 and have a higher turnover rate than microglia [19, 20]. Both perivascular macrophages and brain microglia produce proinflammatory cytokines and chemokines in response to infections in the CNS (reviewed in [21, 22]).

Astrocytes, once considered to serve only as scaffolding for neurons, have multiple roles in the CNS. They regulate transendothelial cell migration across the blood–brain barrier (BBB) [23] and have an accessory role to neurons, modulating glutamate levels in the extracellular space and preventing glutamate-induced neurotoxicity [24–26]. Upon activation, astrocytes increase production of glial fibrillary acidic protein (GFAP), and undergo process extension and interdigitation. Activated astrocytes are a common finding in a number of neurologic disorders (reviewed in [27]). The process by which these cells become activated during infections and the downstream effects of astrocyte activation remain important questions in understanding pathogenesis in the CNS.

Use of the polytropic retrovirus model to study innate immunity in retroviral pathogenesis

The mouse model of polytropic retrovirus infection provides a valuable animal model to study the contribution of the innate immune response to viral neuropathogenesis because there is limited lymphocytic infiltration into the brain in this model [9, 28, 29]. This allows for the analysis of how the innate immune response affects intrinsic brain cells and the induction of neurologic disease. Polytropic murine retroviruses are members of murine leukemia viruses (MuLVs), a subfamily of the gamma retroviruses. The MuLV family can be further divided depending on the type of cellular receptor utilized by the virus, which also affects the host range of the virus. For example, ecotropic retroviruses, which primarily infect mouse cells, utilize the cationic amino acid transporter (solute carrier family 7 member 1, SLC7A1) protein as a receptor. Amphotropic retroviruses, which infect a broad range of species, utilize the phosphate transporter solute carrier family 20 member 2 (SLC20A2) protein as a receptor. Polytropic as well as xenotropic retroviruses infect cells through the xenotropic/polytropic receptor 1 (XPR1), a member of the G-protein coupled receptor protein signaling pathway family [30, 31]. Polymorphisms in XPR1 between mammalian species mediate the host range of xenotropic and polytropic viruses [32, 33].

Differences between the subfamilies of MuLVs are also observed with virus infection and pathogenesis of the CNS. Ecotropic retroviruses can induce either intracerebral hemorrhages or widespread spongiform degeneration depending on determinants in the ecotropic envelope protein [34–38]. In contrast, the brain pathology following infection with neurovirulent polytropic retroviruses is more limited and primarily consists of reactive astrogliosis, white matter microglial infection with microgliosis and microglial nodules [28, 29]. There is only minimal vacuolation and only occasional neuronal death. In this regard, polytropic retrovirus infection of the CNS has some similarities with HIV-encephalopathy as HIV infection of the CNS also induces reactive astrogliosis, microglia nodules without an extensive infiltration of inflammatory cells from outside the CNS [1–5].

Polytropic retrovirus infection of the CNS and clinical disease

Infection of newborn Inbred Rocky Mountain White (IRW) or 129S6 mice by intraperitoneal (ip) injection of polytropic retroviruses leads to virus infection in hematopoietic cells of bone marrow and spleen and then progression via blood to the brain. In the brain, the polytropic viruses productively infect endothelia, microglia, and macrophages with only rare infection of oligodendrocytes [29]. Flow cytometry analysis of microglia/macrophages from infected mice revealed that approximately 20–40% of CD45^high, F4/80+ macrophages are infected with poly-tropic retroviruses, while only 1–5% of CD45^low, F4/80+ microglia are positive for virus. Thus, a larger portion of the macrophage population is infected with virus compared to microglia. Virus-infected cells are normally found in the white matter tracts of the cerebellum, as well as the thalamus, hippocampus, and corpus callosum [28, 29]. Polytropic retroviruses do not productively infect neurons or astrocytes indicating that effects on these cell types are indirect.

The most well-studied neurovirulent polytropic retrovirus is a molecular clone, Fr98. Infection with Fr98 induces neurologic disease in 100% of IRW or 129S6 mice within 2–4 weeks post-infection [28, 29]. Mice have initial clinical signs of tremors and hind limb weakness, followed by ataxia, seizures, and ultimately death [28, 29]. As all animals develop neurologic disease, infecting knockout mice with Fr98 provides a useful method to determine if individual inflammatory genes affect the disease process.

The neuroinflammatory response to Fr98 infection of the CNS

The lack of a clear mechanism of pathogenesis by histologic examination of Fr98-infected mice suggests that changes at the molecular level may play an important role in the disease process. Analysis of mRNA expression of genes involved in innate immune responses indicated a strong upregulation of proinflammatory cytokine and chemokine gene expression in Fr98-infected mice with clinical signs of disease [9]. The increase in mRNA expression correlated with increased protein expression of these cytokines and chemokines in Fr98-infected mice compared to mock-infected controls [39, 40].

Analysis of gene expression over the time course of infection often provides clues as to whether the produced protein may contribute to pathogenesis or be a response to the disease process. Fr98 viral RNA can be detected in the CNS at 3 dpi and increases logarithmically over the course of infection [41]. Astrocyte activation, as detected by increased expression of Gfap mRNA, is first detected at 10–11 dpi and then increases over the course of infection [41]. Increased mRNA expression of Tnfs1a (Tnf, Tnfα), Ccl2, Ccl3(MIP-1α), Ccl4(MIP-1β), Ccl5(RANTES), Cxcl1(MIP-2) and Cxcl10(IP-10) is also observed at 10–11 dpi, peaks at 12–13 dpi, and remains elevated throughout the course of infection [9]. The activation of astrocytes and production of these proinflammatory cytokines and chemokines prior to the onset of clinical disease suggest that these responses may contribute to the disease process. In contrast, increased expression of Tnfsf1b (Tnfβ, Lymphotoxin A) and Il1α mRNA is only consistently found in Fr98-infected mice with clinical disease, suggesting that these cytokines may be a response to disease rather than a cause of pathogenesis [9]. Interestingly, mRNA expression of two additional chemokines, Ccl7(MCP-3), and Ccl12(MCP-5), are upregulated 3–4 days prior to clinical disease development, but return to basal levels at the time of clinical disease [42].

In situ hybridization analysis showed that the majority of chemokines are produced by uninfected cells. Ccl3, Ccl4, and Ccl5 mRNA are expressed primarily by uninfected microglia or macrophages, while Ccl2 and Cxcl10 mRNA are expressed by astrocytes, which are also uninfected [39, 42] (unpublished observations). Infiltrating Ccl12-positive cells are found in brain tissue of pre-clinical mice, but are difficult to detect in mice with clinical disease [42]. The varied response of cytokines and chemokines by multiple cell types including uninfected cells indicate the complexity of the innate immune response to retrovirus infection and suggest interactions between different cell types. The expression of several proinflammatory factors as well as the activation of astrocytes prior to the development of clinical disease suggests that these innate immune responses may be contributing factors to disease development.

Influence of the envelope gene on neurovirulence and innate immune responses

The envelope gene of the polytropic retroviruses is responsible for mediating neurovirulence. The Fr98 virus was generated by cloning the envelope region of neurovirulent biologic virus isolate, FMCF98, into the backbone of a rapidly replicating, but avirulent ecotropic retrovirus, FB29. The subsequent virus, Fr98, has the same polytropic host range as the biological isolate, but induces disease at a faster rate [28] (Table 1). A similar procedure using the envelope from a non-neurovirulent polytopic biological isolate, FMCF54, resulted in an avirulent polytropic retrovirus, Fr54, that differs from Fr98 only in the envelope gene and the 3′ end of the polymerase gene [28] (Table 1). Subsequent mapping studies identified the neurovirulent determinants of Fr98 to be encoded in the env gene and not in the 3′ end of the pol gene [43]. The neurovirulent Fr98 virus and the non-neurovirulent Fr54 virus infect the same cell types in the same region of the brain, although Fr98 viral protein is detected at 2- to 3-fold higher levels than Fr54 in the brain following ip inoculation [28].

Table 1.

Comparison of polytropic retroviruses

^aRange of disease onset in IRW and 129S6 mice following ip innoculation of 10⁴ focus forming units of virus. In 129S6 mice, only 60–80% of BE infected mice and 50–70% of EC infected mice develop clinical signs of ataxia and seizures

^bInnate response as defined by the upregulation of proinflammatory cytokines and chemokines. Similar cytokines and chemokines are upregulated by Fr98, EC and BE infected mice.+ is an indicator of the level of cytokine/chemokine expression in the brain relative to mock-infected controls

^cGene name reflects gene knock out mouse. A straight line indicates no effect on disease kinetics in knock out mice compared to wildtype controls. A single arrow indicates a significant decrease on the kinetics and/or incidence of disease in knock out mice compared to wildtype mice. Three arrow indicates minimal disease in knockout animals. n.d. indicates not determined

The difference between Fr98 and Fr54 in inducing neurologic disease is also observed in the induction of neuroinflammatory responses. Fr98 induces a strong innate immune response, while Fr54 induces only minimal upregulation of proinflammatory cytokines and chemokines [9, 44] (Table 1). This difference in cytokine production is at least partially due to virus levels in the CNS. Increased expression of Fr54 in the CNS, through the use of infected neural stem cells, results in increased production of proinflammatory cytokines and chemokines as well as the development of neurologic disease [45]. This suggests that the influence of the envelope on pathogenesis may be through increasing virus spread in the CNS and/or inducing neuroinflammatory responses.

Generation of chimeric viruses that encode complementing regions of the Fr98 and Fr54 envelope sequence identified two separate neurovirulent determinants, both encoded in the envelope region [46]. One neurovirulent determinant is encoded within the N terminal region of the envelope gene, between a BbsI and an EcoRI restriction site (Table 1). Insertion of the BbsI to EcoRI sequence from Fr98 virus into the genome of the Fr54 resulted in the recombinant virus BE, which induces neurologic disease with similar clinical signs to Fr98, but with slightly slower kinetics (3–5 weeks post-infection) [43, 44]. Seven envelope residues encoded by the BE region influence neurovirulence (residues 84, 113, 126, 127, 134, 136, and 141) [43]. When projected on a three dimensional model of the receptor binding domain (RBD) of the closely related Friend ecotropic MuLV envelope protein, these residues are clustered in two spatially separated groups [43]. One group is in variable region B of the receptor binding site, and the other group (residues 84 and 113) is located on the opposite side of the RBD. Importantly, residues from both sites are required for BE-mediated pathogenesis.

Another neurovirulent determinant of the Fr98 envelope maps within the EcoRI to ClaI region of the Fr98 envelope [46]. The recombinant EC virus encoding the EcoRI to ClaI region also induces disease similar to Fr98, but with much slower kinetics (4–10 weeks post-infection) and with a lower virus burden in the CNS [47]. Two amino acids, at positions 165 and 168 of the envelope protein from the EC region, mediate pathogenesis [47]. Surprisingly, these two residues are on the same surface/pocket of the envelope protein as residues 84 and 113 from BE [43], suggesting that this pocket of the envelope protein may regulate an important aspect of virus spread, such as envelope trimerization or interaction of the envelope protein with another cellular protein. The neurovirulent residues in BE and EC do not appear to enhance virus entry, as all viruses have similar efficiencies of virus entry in primary cortical cultures in vitro [43].

BE and EC viruses differ substantially in virus replication and neuroinflammatory responses in the brain. BE-virus levels in the brain at the time of clinical disease are comparable to Fr98 levels, whereas EC virus levels at the time of clinical disease are much lower and more comparable to Fr54 levels (Table 1) [44, 47]. The difference between BE and EC in virus levels correlates with differences in the induction of innate immune responses. BE infection of the CNS induces a much more pronounced proinflammatory cytokine and chemokine response compared to EC infection [9, 40, 44]. The differences in disease kinetics and inflammatory responses suggest that BE and EC may mediate neuropathogenesis through two disparate mechanisms.

Influence of individual cytokines and chemokines on polytropic retroviral pathogenesis

The strong correlation between neurologic disease and the intensity of the innate immune response suggests that several proinflammatory cytokines and/or chemokines may contribute to the disease process. To examine the role individual genes played in the disease process, we utilized knockout mice. We examined mice deficient in TNF, a key player in a number of neurologic disorders. In addition, we examined three different chemokine receptor deficient mice, CCR1, CCR2, and CCR5.

CCR1 and CCR5 bind multiple beta-chemokines (CCL3, CCL4, and/or CCL5) upregulated by Fr98 infection. However, mice deficient in either of these receptors develop disease with the same kinetics as wildtype mice [39, 42]. The absence of these receptors did not influence virus replication in the CNS, nor the induction of chemokine expression. This suggests that despite the early upregulation of CCL3, CCL4, and CCL5, these chemokines may not play an important role in Fr98-induced pathogenesis.

Mice deficient in CCR2, the primary receptor for CCL2, have reduced kinetics and/or frequency of Fr98-induced disease compared to wildtype controls [42]. This suppression of disease is observed in CCR2-deficient 129S6 mice infected by ip inoculation as well as in CCR2-deficient Balb/c mice inoculated intracerebroventricularly (icv) with neural stem cells infected with Fr98 (unpublished observations) [42]. Despite the influence of CCR2 on disease induction, CCR2 does not influence virus replication in the CNS, nor astrocyte activation [42]. It is possible that the influence of CCR2 on disease induction is through a direct effect on neurons, since neurons express CCR2 and can respond to CCL2 stimulation [48–50]. However, it is also possible that the influence of CCR2 on neurologic disease is indirect and mediated through activation of other intrinsic brain cells or the recruitment of macrophages to the CNS.

TNF deficiency delays the onset of Fr98-induced neurologic disease, but does not prevent disease induction indicating that TNF also plays a role in disease pathogenesis [44] (Table 1). Analysis of the two chimeric viruses, EC and BE, demonstrate that EC pathogenesis is dependent on TNF, while BE-mediated pathogenesis is not [44]. This demonstrates that the impact of an individual host protein on disease pathogenesis can vary between highly related viruses. A possible explanation for the different influences of TNF on disease pathogenesis may reside with the kinetics of virus infection. Both BE and Fr98 are found at higher levels in the brain than EC and induce stronger proinflammatory cytokine and chemokine responses. Thus other factors, perhaps, other cytokines, may compensate for the lack of TNF for BE or Fr98-induced disease in TNF knockout mice. In contrast, EC-induced disease is a slower process, with a reduced neuroinflammatory response compared to the other viruses. In EC infection, TNF may be a key mediator in the progression of the inflammatory response and the development of neurologic disease. This is supported by the lack of upregulation of the microglia/macrophage marker, F4/80, in TNF deficient mice infected with EC [44].

BE pathogenesis was not altered in any of the knockout mice studied [44] (unpublished observations) (Table 1). This was surprising since BE-virus infection induces a pronounced upregulation of proinflammatory cytokines and chemokines. The lack of an effect of cytokine/chemokine receptor knockouts on BE-mediated pathogenesis may indicate that BE pathogenesis is mediated by another mechanism that does not involve the innate immune response. A non-immune mechanism of BE pathogenesis would correlate with the hypothesis of two separate mechanisms for EC- and BE-mediated pathogenesis, which are additive in Fr98-mediated pathogenesis.

Regulation of the innate immune response to retrovirus infection

One of the important considerations in understanding the impact of the innate immune response in retroviral pathogenesis is determining the events and molecules that are necessary to induce the innate immune response. Innate immune responses are initiated by the recognition of pathogen-associated molecular patterns (PAMPs), repeated structural motifs that are unique to microorganisms. PAMPs are recognized by pattern recognition receptors (PRR), include transmembrane-bound toll-like receptors (TLRs) as well as cytoplasmic PRRs [51, 52]. The life cycle of retroviral infection produces large amounts of HIV ssRNA, but also generates some HIV-DNA. Although both of these products could be stimulatory; HIV ssRNA, but not HIV-DNA, is responsible for HIV-triggering of IFNα responses by plasmacytoid dendritic cells [53], suggesting a role for TLR7 or TLR8 in initiating anti-retroviral innate immune responses. Analysis of Fr98-induced interferon β (Ifnb1) mRNA responses in primary cortical cells from TLR7 wildtype and deficient mice also indicate that retroviral-induced IFNβ responses, in vitro, are mediated through TLR7 [41].

TLR7 and retroviral pathogenesis in the brain

Toll-like receptor 7 is expressed by brain capillary endothelial cells as well as ependymal cells in the neonatal brain [41]. During Fr98 infection, TLR7 is not upregulated on virus-infected microglia and/or macrophages to detectable levels [41]. However, TLR7 is detected on a population of non-virus infected cells in Fr98 infected mice that are not present in mock-infected controls. These cells may possibly be activated astrocytes, non-infected macrophages or microglia or recruited peripheral cells. The presence of these cells in the CNS correlates with increased expression of Tlr7 mRNA in Fr98-infected mice compared to mock controls [41]. Interestingly, the influence of TLR7 on Fr98-induced neuroinflammation is temporal, with TLR7 playing an important role in the neuroinflammatory response in mice at the pre-clinical stage of infection. Many, but not all, chemokines are reduced and cellular responses limited in non-symptomatic TLR7 deficient mice infected with Fr98 [41]. However, TLR7 deficiency does not significantly suppress neuroinflammatory responses in mice with clinical disease nor does it alter the kinetics of disease development [41]. Thus, TLR7 appears to have only a minor role in the response to Fr98-infection in the CNS.

There are several potential mechanisms that could explain the dichotomy of the influence of TLR7 on the early and late stages of the neuroinflammatory response. It is possible that the cytokines and chemokines not dependent on TLR7 are sufficient to initiate the neuroinflammatory response, induce the production of the other cytokines as well as astrocyte activation. It is also possible that other PRRs may be able to compensate for the lack of TLR7 during retrovirus infection in the CNS. TLR9 polymorphisms were shown to substantially influence rapid progression of AIDS [54], suggesting that TLR9 may be a contributing factor to retroviral pathogenesis. TLR4, which was shown to recognize retrovirus glycoprotein [55], may also play a contributing role to retroviral pathogenesis in the CNS.

Another possible PRR that could play a role in retroviral neuropathogenesis is TLR8. Agonists that stimulate both human TLR7 and TLR8, did not induce neuroinflammatory responses in TLR7 deficient mice [56]. This correlates with other studies, where the response to viral infection or TLR7/8 agonist were suppressed in TLR7 deficient mice [57, 58]. These studies suggest a lack of function of TLR8 in the mouse and suggest that TLR8 does not contribute to retrovirus-induced neuroinflammatory response. However, as TLR8 is functional in humans, it is highly possible that TLR8 functions as a mediator of innate immunity in HIV pathogenesis in the CNS.

Influence of TLR7 on astrocyte activation

Astrocyte activation, as measured by GFAP expression and/or CCL2 expression is a common finding following retrovirus infection in the brain [13, 42, 45, 59]. The influence of TLR7 on astrocyte activation during the early stages of Fr98 infection was surprising since astrocytes are not productively infected with Fr98 [29]. TLR7-dependent astrocyte activation could be mediated by cytokines and chemokines that are produced following TLR7 stimulation in macrophages or microglia. Alternatively, TLR7-dependent activation of astrocytes may be direct, following astrocytic endocytosis of virus particles or viral RNA. Intracerebral inoculation of TLR7 agonists induced Gfap mRNA expression within 12 h of inoculation indicating that stimulation through TLR7 can induce astrocyte activation [56]. TLR7 agonist administration also induced strong CCL2 production by astrocytes [56]. Thus, it is possible that viral ssRNA, either released from infected cells or from endocytosed virus are responsible, in part, for astrocyte activation during retrovirus infection.

Future directions

The above studies indicate that the neuroinflammatory response to retrovirus infection is complex, both in the initiation of the neuroinflammatory response as well as the contribution of this response to the disease process. Future studies utilizing mice deficient in signal transduction proteins of the PRR pathways will help determine the mechanisms by which retroviruses induce neuroinflammation in the brain and determine which responses are necessary for disease induction.

The different effects of TNF deficiency on BE and EC-mediated neurologic disease demonstrates that multiple mechanisms can contribute to the disease process. As the innate immune response appears to play a more important role in EC-mediated pathogenesis, this model will provide the ability to analyze the pathways by which proinflammatory cytokines and chemokines contribute to retroviral neuropathogenesis. In the instance of BE-mediated neurologic disease, it will be important to determine other mechanisms by which the virus infection induces disease and whether the neuroinflammatory response has any influence on this mechanism. Understanding the mechanisms behind the induction of the neuroinflammatory response as well as the impact of the neuroinflammatory response on disease pathogenesis will provide new insights for targeting pathways to inhibit retroviral pathogenesis as well as other neurologic diseases.