NeuroAIDS: The Blood-Brain Barrier in HIV-associated Dementia

Monocytes have been shown to infiltrate brain tissue during various neurological disorders including HIV-associated dementia. The presence of an excess of activated macrophages in brain tissue is accompanied by a loss in neuronal function and viability. Therapeutic options against such neurological disorders could be aimed at the prevention of monocyte infiltration across the blood-brain barrier. Therefore, a better understanding of these processes is needed. Recent insights in cellular processes between monocytes/macrophages and brain microvascular endothelial cells in the neuropathogenesis of HIV-1 infection demonstrate that monocytes roll along endothelial cells via the inducible endothelial adhesion molecule E-selectin. Binding of these cells is mainly mediated via the endothelial adhesion molecule named vascular cell adhesion molecule-1. The transmigration through the blood-brain barrier is facilitated by both endothelial and monocyte/macrophage-derived nitric oxide, as well as by the increased production of gelatinase B activity by HIV-infected monocytes/macrophages. Chemokines produced within the brain regulate the traffic of infiltrating monocytes through the brain parenchyma. In addition, endothelial cells also produce monocyte-attracting chemokines during their interactions with HIV-infected monocytes/macrophages, thus promoting additional influx of mononuclear phagocytes into the brain. Furthermore, excessive infiltration of monocytes is accompanied by endothelial damage resulting in the loss of tight junctions. In summary, brain microvascular endothelial cells might contribute to the neuropathogenesis of HIV-1 infection.

Introduction

Abstract

Introduction

HIV-associated dementia (HAD) is the most severe neurological manifestation of HIV infection of the brain. In the era before treatment with anti-HIV compounds 30% of the adults with AIDS and 50% of the pediatric AIDS cases were affected by HIV-induced neurological complications in the western world. Interestingly, in the adult brain HIV only productively replicates in macrophages/microglia and not in neurons, astrocytes, oligodendrocytes or brain microvascular endothelial cells (BMEC) (1)(2)(3). Neuropathological features of HIV-1 encephalitis include productive HIV-1 infection of brain cells of monocyte/macrophage lineage, multinucleated giant cell formation, monocyte infiltration of the central nervous system (CNS), astrogliosis, and myelin pallor. Some neurons undergo dendritic pruning, simplification of synaptic contacts, and frank cell loss (4). These neuronal changes are clinically manifested by cognitive and motor dysfunctions that affect many individuals with AIDS. The finding that neurological symptoms are a consequence of dendritic injury and synaptic loss (5), and occur in absence of neuronal cell loss (6), suggest that treatment may retard the onset of cognitive and motor defects.

HIV- l is a haematogenously spread virus that most likely gains entry into the brain inside blood-derived macrophages (7). Indeed, HIV-1 is selectively localized within perivascular and infiltrated parenchymal blood-derived macrophages and microglia (1)(2)(3). Axonal spread of HIV as a mechanism to enter the CNS is ruled out since direct neuronal infection has not been demonstrated in pathologic specimens of AIDS patients with HIV-1 encephalitis. Theoretically, cell-free HIV might infect BMEC and subsequent release of virus at the abluminal membrane of the blood-brain barrier (BBB) would result in HIV entry into the brain parenchyma. Although direct in vitro infection of brain endothelium by lymphotropic HIV-l has been described, the relevance of that finding remains questionable (8). First, macrophagetropic HIV-1 is the predominant viral phenotype isolated during the acute seroconversion reaction, thus excluding a role for lymphotropic HIV-1 in CNS infiltration (9). Second, brain-derived viral strains are macrophagetropic rather than lymphotropic (10)(11). Third, there is no in vivo evidence for HIV infection of brain endothelial cells (1)(2)(3). In addition, several other investigators could not find in vitro evidence for permissiveness of these cells to either lymphotropic or macrophagetropic HIV-1 (12)(13)(14). These findings and the ample evidence that only blood-derived macrophages and microglia support productive viral replication within the CNS suggest that HIV-1 gains entry into the brain within macrophages or as as cell-free virions (15). Although the early transmigration of HIV-infected monocytes/macrophages (M/M) into the CNS might result in meningitis, the occurrence of severe neurological symptoms including dementia take place at a relatively late stage of HIV infection. In these stages the immune system is deteriorated and activated. This has consequences for the mechanisms involved in the brain infiltration by HIV-infected M/M and the subsequent development of HAD as outlined hereafter.

Transmigration of HIV-infected M/M through the blood-brain barrier

Abstract

Introduction

Transmigration
through BBB

During the later symptomatic stages of HIV-1 infection elevated levels of pro-inflammatory cytokines such as tumor necrosis factor-a (TNF-a) and interleukin-1b (IL-1b) have been detected in serum of AIDS patients and in macrophages isolated from HIV-infected individuals (16)(17). Both cytokines are mediators of meningitis and regulate BBB permeability in a reversible manner. Indeed, intracisternal injection of TNF-a and IL-1b into rats increased BBB permeability for systemically administrated radioactive tracers (18). In addition, these cytokines induce the expression of endothelial adhesion molecules that facilitate rolling and binding of leukocytes. Indeed, activated HIV-infected M/M induce the highest levels of E-selectin and vascular cell adhesion molecule-1 (VCAM-1) on brain microvascular endothelial cells (14) as compared to non-activated HIV-infected M/M and activated uninfected M/M (Figure 1). Since E-selectin mediates rolling and VCAM-1 mediates binding of leukocytes to endothelium, HIV-infected cells have an advantage in these processes when compared to normal M/M. The in vitro studies are supported by the observation that endothelial expression of E-selectin and to a lesser degree VCAM-1 was the highest in brain tissue of demented AIDS patients when compared to non-demented AIDS patients (14). In addition, monocyte binding to encephalitic brain could be prevented by blocking monoclonal antibodies directed against both adhesion molecules (14). Since recombinant tat induces E-selectin expression on other endothelial cells, the induction of this adhesion molecule is most likely mediated by the HIV-1 tat molecule that is secreted by HIV-infected M/M (19). Furthermore, HIV-1 infection of M/M induces the production of low levels of nitric oxide (NO), a potent vasoactive molecule (20)(21). Since NO can slow down the speed of blood cells passing through the blood vessel, the production of NO might increase the change that mononuclear cells roll on endothelium via E-selectin (Figure 1).

(same in color)

Figure 1 (Enlarge): Mechanisms for HIV-1 entry into brain during neurologic disease.
(1) During the later symptomatic stages of HIV-1 infection elevated levels of pro-inflammatory cytokines such as TNF-a and IL-1b are secreted by perivascular HIV-1-infected M/M. TNF-a, IL-1b and HIV-1 tat induce the expression of E-selectin on BMEC. The change that these cells indeed roll on brain endothelium via E-selectin might increase due to the secretion of NO, a potent vasodilator, by HIV-infected macrophages. E-selectin then mediates rolling of HIV-infected M/M on the vessel wall.
(2) TNF-a and IL-1b also induce the expression of VCAM-1 on BMEC. The induction of this adhesion molecule allows binding of HIV-infected M/M to brain endothelium.
(3) The interactions between HIV-infected M/M and BMEC result in the production of the M/M-derived chemokines MIP-1a, MIP-1b and the endothelial-derived chemokine MCP-1 (22). These chemokines will attract more mononuclear phagocytic cells into the CNS.
(4) The overexpression of the pro-inflammatory mediators TNF-a, and IL-1b by HIV-infected M/M will result in an increase in blood-brain barrier (BBB) permeability. The overexpression of gelatinase B by HIV-infected M/M will result in degradation of abluminal extracellular matrix proteins. All these factors will lead to the transmigration of HIV-infected M/M across the BBB.
(5) Intercellular interactions between astrocytes and infiltrated HIV-infected M/M within the brain parenchyma will result in the induction of several chemokines, TGF-b, MIP-1a and MCP-1. These chemokines will attract more HIV-infected and uninfected M/M into the brain parenchyma resulting in an expansion of the viral load within the CNS.

However, besides these subtle changes between M/M and endothelial cells there is also evidence of endothelial damage that might underlie the excessive monocyte infiltration into the brain and even contribute to the severe stages of HAD. For instance, in vitro HIV-1 infection of M/M resulted in superoxide anion production (21), and endothelial NO synthase was induced in cocultures of M/M and endothelial cells (unpublished data). Since superoxide anion and NO form the highly toxic molecule peroxynitrite, immunohistochemical staining for nitrotyrosine, the footprint of peroxynitrite, was performed and showed extensive immunoreactivity in perivascular areas of brain tissue of demented AIDS patients (21). Furthermore, the architecture of many bloodvessels was damaged as shown by immunohistochemical staining for zonula occludens-1, a tight junction protein (23). Thus, during the intimate contact with HIV-infected M/M endothelial cells may contribute to their own damage by the production of NO. These findings might explain the increased permeability of artificial BBB for radioactive albumin when put in contact with HIV-infected M/M (24).

After their migration through the endothelial cell layer, the monocytes will encounter the basement membrane that surrounds the abluminal side of the BBB. HIV-infected M/M produce higher levels of gelatinase B as compared to uninfected M/M. These cells therefore have an advantage in the process of diapedesis since gelatinase B degrades the extracellular matrix proteins that underlie the endothelial cell monolayer and in such a way increases the permeability of the BBB (25). Interestingly, incubation of HIV-infected M/M with tissue inhibitor of metalloproteinase-1 (TIMP-1) and TIMP-2 partially inhibited the increased permeability of endothelial cell monolayers to ¹²⁵I albumin. These proteinase inhibitors antagonize the effects of matrix metalloproteinases, such as the HIV-induced gelatinase B activity (25). Importantly, matrix metalloproteinase-9 activity in the cerebrospinal fluid of HIV-infected patients was significantly higher in patients with neurological deficits than in patients without neurological deficits (26). This finding suggest that, in addition to the effects of HIV-infected M/M on endothelial cells, interactions between HIV-infected M/M and the extracellular matrix also affect BBB permeability (Figure 1). Indeed, cocultivation of HIV-infected M/M with BMEC resulted in an increased permeability of the endothelium, measured as ¹²⁵I albumin passage through endothelial cell monolayers in transwell systems (24). In addition, immune-activated M/M placed on BMEC cultured on the upper chamber of a transwell migrated in larger numbers through the artificial BBB than non-activated cells (27).

Migration of HIV-infected M/M through the brain parenchyma

Abstract

Introduction

Transmigration through
brain parenchym

The distribution of infected macrophages in the brain demonstrates viral predisposition for cerebral white matter, deep gray matter (basal ganglia and thalamus), and the brain-stem. Although the exact mechanisms that underlie this distribution remain unknown, temporal expression of adhesion molecules and chemokines in the brain has been suggested to guide specific trafficking of HIV-infected M/M to these brain areas (28). In any event, chemoattractants have been detected in brain tissue of AIDS patients with HIV- l encephalitis, and might therefore play a role both in infiltration and subsequent migration of HIV-infected M/M into the brain (Figure 1). For instance, transforming growth factor-b (TGF-b), a cytokine that induces the migration of monocytes at femtomolar concentrations, was clearly identified within AIDS brain tissues in association with macrophages and astrocytes, but not in control brain tissue (29). More recently, elevated expression of the b chemokines macrophage inflammatory protein- (MIP-1a) and MIP-1b was detected in brain tissue of demented AIDS patients as compared to non-demented AIDS patients (30)(31). Cells expressing these chemokines were identified morphologically as brain macrophages and astrocytes. Interestingly, MIP-1b levels were much more prevalent than the MIP-la levels (30)(31). Since the monocyte chemotactic activity of MIP-la is much more potent than that of MIP-1b, the elevated levels of MIP-1a in HIV-infected brain tissue might result in further monocyte infiltration of brain tissue and subsequent spread of viral CNS infection. In addition, monocyte chemoattractant protein-1 (MCP-1), another chemokine with relatively selective chemoattractant properties for monocytes, was found to be elevated in astrocytes in brain tissue of demented AIDS patients (31)(32). Furthermore, elevated levels of MCP-1 in CSF of demented individuals have also been described (32). Indeed, using a coculture of human endothelial cells and astrocytes to model the BBB, it was recently demonstrated that tat-induced, astrocyte-derived MCP-1 directs the transmigration of monocytes (33). And, finally, microglial expression of RANTES (Regulated upon Activation, Normal T-cell Expressed and Secreted) was found to be elevated in brain tissue of demented AIDS patients as compared to non-demented AIDS patients (31).

Interestingly, endothelial cells themselves might also play a role in the recruitment of monocytes into the CNS by the production of chemokines. For instance, these cells elicit the production of several chemokines by uninfected as well as HIV-infected M/M (22). The levels produced by the HIV-infected cocultures are significantly higher than the levels produced by the uninfected cocultures. Importantly, endothelial expression of MCP-1 was readily detected in brain tissue of AIDS patients with HAD when compared to tissue of non-demented AIDS patients (22).

Conclusions

Abstract

Introduction

Conclusions

HIV-1 enters the brain in a haematogenously fashion early during HIV-1 infection. Experimental evidence has been provided by several scientific investigators that it gains entry via the Trojan horse model by modulating both monocyte/macrophage and endothelial functions. HIV takes advantage of cellular machineries involved in rolling and binding processes between both cells and also increases the ability for monocytes to transmigrate into the CNS. Apparently the brain is able to limit extensive HIV-1 replication within the CNS or is able to suppress inflammatory processes. However, when the immune system is deteriorated, neurological complications start to occur and will eventually lead to dementia. During these events circulating monocytes and possibly also macrophages resident within the brain are immune-activated leading to enhanced production of molecules that increase BBB permeability, endothelial adhesion molecules expression, and chemokine expression. Eventually, when there is an excess of infiltrating monocytes into the brain the BBB is damaged with concomitant loss of tight junctions. At these stages, the BMEC might, in addition to the HIV-infected monocyte/macrophages, also function as instigators of the BBB loss and also contribute to the neurological complications.

Acknowledgments
I would like to thank Jeena Middel for her indispensable help in realizing figure 1. Hans S.L.M. Nottet is a fellow of the Royal Netherlands Academy of Sciences and Arts. Candida da Fonseca Pereira is a Marie Curie fellow.

References

Abstract

Introduction