In this study qPCR was utilized to identify HIV-1 DNA within 66 brain tissues from 13 autopsies of HIV-infected individuals with and without clinical and histological indications of HAD. These results indicated that most tissues from patients without HAD contained little amplifiable HIV-1 DNA, whereas two patients with both clinical and histological HAD contained amplifiable HIV-1 DNA in all brain tissues analyzed. Three other patients also contained positive qPCR results in brain tissues, each with a different pathology: CVD, MAC co-infection, and non-Hodgkins AIDS-related lymphoma.
Histological staining of frontal lobe tissues for HIV-1 showed varied degrees of replicating and resident macrophages among patients with differing pathologies. In patient AZ, the CD68+ staining was negative; these results suggest that brain viruses found in patient AZ’s frontal lobe had either not been present long enough to activate brain macrophages or that those viruses that had entered the frontal lobe had migrated from an atherosclerotic plaque, a rich pathogenic macrophage process. Either interpretation is consistent with late viral entry into brain tissues due to atherosclerosis. The next three cases (BW, DY, GA) were interesting in that they all showed a moderate degree of HIV-1 infected macrophages and varying degrees of replicating virus. It is also noteworthy that patient DY staining results indicated the existence of a population of activated macrophages; however, neuropathology reports describe the brain as grossly normal and unremarkable. As in patient AZ, this may indicate that the infected cell population within the frontal lobe was relatively new. The fact that subjects BW, DY and GA all show less positive CD68 and P24 results than patient CX may be due to the effect of HAART. Certainly it is well reported that manifestations of HIV-associated CNS involvement have generally become less severe and more manageable in patients on HAART.
A variety of sequence analysis highlighted differences in viral evolution between HAD and non-HAD patients. Recombination rates were more prevalent in both HAD patients, indicating that there were more viruses replicating within super-infected macrophages in brain tissues. This observation was previously noted in gp120 DNA sequences from damaged tissues (Lamers et al, 2009
). The positive correlation seen in the HAD patients between protein length and the number of recombinants may indicate less selective constraints among proteins, random generation of recombinant events across the HIV-1 genome, or that most sequences were derived from viruses that were sequestered within a macrophage reservoir for a much longer period of time. The fact that a correlation between domain length and number of recombinants was not present in the three non-HAD individuals may indicate selective constraints on particular proteins evolving under particular conditions, such as populations of viruses that had recently been exposed to a more complex immune environment. It should be noted that the small number of variables in this regression analysis (three data points) makes it impossible to say with certainty that either conclusion is correct, still, if it is true that the viruses from these patients entered the brain at significantly different times, the results provide another piece of evidence that indicates different evolutionary genetic patterns in lymphoid rather than non-lymphoid tissues. A full genome study would help confirm these results. Patients BW and DY showed a statistical increase in diversity in gp120 over gp41, while both HAD patients demonstrated a significant increase in diversity of gp41 over gp120 sequences. This difference may be due to different selective pressures on the surface glycoprotein (gp120) and the transmembrane/fusion glycoprotein (gp41) during different disease pathologies.
Phylogenetic trees for both HAD subjects contained multiple highly supported sub-clades containing sequences from brain consistent with a longer period of infection within brain tissues during HAD development. This finding is also supported by the larger median root height of the posterior distribution of sequences in CX and GA. On the other hand, the primary brain clades for patients DY and AZ contained no supported sub-clades. The phylogeny for subject BW differed from the other four in that brain and non-brain sequences each branched in highly supported monophyletic clades. However, patient BW was similar to the HAD patients in that the tree showed well-supported sub-clades within brain sequences. Various mechanisms can contribute to the degree of compartmentalization of HIV-1 populations seen in a phylogenetic analysis, such as the high rate of mutation of HIV-1 in vivo, differences in selective pressures imposed by the immune system, differences in local concentrations of antiviral drugs, cellular tropism, drug resistance, and level of pathogenesis (Zarate et al, 2007
). The lack of compartmentalization in subject AZ and DY is not surprising if the viruses that populated the brain were due to arterial leakage near death. The compartmentalization of subject BW brain viruses may reflect the nature of lyphomagenesis in patients with HIV-1 infection; BW’s brain may have been populated near death with an aggressive strain of HIV-1 closely associated with metastatic tissues. It is also possible that some of the sequences represented archival proviruses from latently infected cells. Yet, the substantial genetic heterogeneity observed, coupled with p24 positive staining discussed in indicate that at least some viruses are replicating within productively infected brain cells. Overall, discerning which sequences reflect latent or productive infection does not impact the principal finding that different viral evolutionary patterns can be observed in patients with different pathology.
Williams and Hickey (Williams and Hickey, 2002
) proposed a popular model to describe the development of HAD: during the early phase of the disease, the immune system still controls the infection in brain macrophages; later the immune system becomes unable to prevent the emergence of productively infected cells as they continue to amass and damage surrounding tissues. The continuous accumulation in the brain of activated macrophages initiates a self-amplifying cycle that eventually leads to the development of HAD. The model explains latency, extensive brain damage observed at autopsy and the separate evolution of brain from body viral isolates (Salemi et al, 2005
; Williams and Hickey, 2002
). Later, Fischer-Smith et al. (Fischer-Smith et al, 2008
; Fischer-Smith and Rappaport, 2005
) described another model for HIV-1 infection of the brain; they proposed that brain tissues may also become infected during late-stage disease via the expansion of more virulent viruses, uncontrolled viral replication and alterations in the myeloid differentiation pathway that results in an activated monocyte subset capable of CNS tissue invasion. In vitro
experiments have implicated direct infection of brain astrocytes by HIV-1 as contributing to HAD development (Zheng et al, 1999
). Changes in cell environment, like the elevation in the level of cytokines such as TNF-α and IL-1β, might also activate virus production within CNS tissues (Gorry et al, 2005
; Kramer-Hammerle et al, 2005
HIV-associated dementia generally develops in late-stage HIV-1 disease (CD4 cells/μL<50). The onset is usually subtle, developing over weeks or months. At autopsy, the brains from patients with clinically identified HAD commonly show marked astrocytosis, microgliosis and a numerical increase of resident macrophages actively replicating virus. The macrophage infiltrate is most prominent in the white matter of the frontal lobe, although usually no area of the central nervous system is spared (Williams and Hickey, 2002
). The general theory for the development of HAD is that infected macrophages carry HIV-1 into brain tissues where the infection evolves primarily in microglia and macrophages (McGrath, 1997
; Williams et al, 2001
). Macrophages bind viruses that are specific for the CCR5 and CCR3 chemokine receptors (Alkhatib et al, 1996
; Deng et al, 1996
). Once infected, macrophages can produce a variety of proinflammatory cellular neurotoxins, including tumor necrosis factor-alpha, cytokines, interleukins, chemokines, nitric oxide, and excitatory amino acids (Dou et al, 2004
). Some studies have suggested that the damage caused by macrophages within tissues may be more correlated to their release of inflammatory mediators in the CNS than to the actual viral load in the brain (Hult et al, 2008
; Williams et al, 2008
Two of the HAD patients studied (GA and CX), one on HAART and the other not on therapy, presented with a slower-progressing clinical dementia and showed classic histological signs of HAD at the time of death. Phylogenetic analysis, recombination, distance and charge analysis in these patients all support the theory of Williams and Hickey as well as earlier observations from our group (Salemi et al, 2005
), wherein the brain is infected early, sequences in the brain evolve separate from the body, and a self-amplifying macrophage infection ensues.
Another patient, AZ, did not develop HAD; however, the patient did have a severe case of CVD. CVD is the most common non-AIDS-defining illness leading to death during HAART. It is uncertain if this is a result of long-term survival during HAART therapy or an indirect toxic effect of the therapy (Friis-Moller et al, 2003
; Holmberg et al, 2002
; Sudano et al, 2006
). What is clear is that HIV-infected patients in the HAART era are at a significantly increased risk for myocardial infarction (Friis-Moller et al, 2003
), endothelial dysfunction (Stein et al, 2001
), osteonecrosis (Monier et al, 2000
), coronary artery calcification (Meng et al, 2002
), and atherosclerosis (Hsue et al, 2004
). During the evolution of atherosclerosis, fatty lipids are deposited within arteries. Macrophages called in to consume the lipids are transformed into activated fat containing vacuoles called foam cells. The sites of damaged endothelium attract more macrophages to the site of lipid deposition and eventually form an early atherosclerotic site, also called a fatty streak. Fatty streaks cause vascular breakdown, bleeding and a self-propagating inflammatory response to injury (Linton and Fazio, 2003
). In certain ways, this process is similar to what has been described during the perivascular macrophage inflammatory response associated with HAD (Salemi et al, 2005
; Williams et al, 2001
; Williams and Hickey, 2002
); however, the CVD process differs in that the perivascular HIV-1 infected macrophages in CVD are associated with arteries, rather than central nervous system veins involved in HAD. Patient AZ exhibited little diversity in sequence populations between brain and body isolates and little brain tissue damage other than widespread atherosclerosis. Phylogenetic, distance, and recombination analysis confirm a late-stage brain infection likely due to arterial leakage from atherosclerosis.
A fourth patient, DY, suffered for several years with reoccurring MAC infection, one of the most common bacterial opportunists. Dual infection in HIV-1 patients with other opportunistic infections increases viral replication in both macrophages and CD4+ T-cells (Wahl et al, 2000
; Wahl et al, 1998
; Wahl et al, 1999
), especially in late-stage disease when the CD4+ T-cells are depleted and HIV-1 continues to replicate within macrophage populations. Wahl et al. (Wahl et al, 1999
) found that MAC up-regulates TNF-α expression, which subsequently enhances HIV-1 replication within macrophages. Other studies have also observed this finding (Denis and Ghadirian, 1994
; Havlir et al, 2001
; MacArthur et al, 2000
). Additionally, patient DY likely harbored dual-tropic viruses. Infection of an individual with dual-tropic HIV-1 has been linked to drug resistance, rapid progression to AIDS and hastened death. Although brain autopsy specimens were normal, this patient developed a severely aggressive dementia in the days prior to death. This was the only patient observed with no apparent histology abnormalities in which amplifiable HIV-1 DNA was present. Phylogenetic, recombination, distance analysis and primary sequence analysis all suggest the rapid expansion of an aggressive form of virus that evolved near the time of death. One plausible explanation for this patient’s pathology could be the Fisher-Smith Hypothesis; however, another interesting hypothesis was presented in 1999 concerning the infection of astrocytes with HIV-1 (Zheng et al, 1999
). This study showed through in vitro
experiments that brain astrocytes are capable of binding viruses via the CXCR4 co-receptor. Once infected, astrocytes cause the release of a variety of pro-inflammatory cytokines that could cause neuronal dysfunction (Zheng et al, 1999
). Astrocyte infection is associated with moderate to severe dementia (Ranki et al, 1995
). By combining genetic analysis, pathology, histology and theories concerning HIV-1 infection of the CNS, an interesting hypothesis for the rapid development of HAD minus brain tissue damage in this patient can be constructed, 1) co-infection with MAC super-activated macrophages that were producing dual-tropic viruses, 2) dual tropic viruses could enter the brain via macrophages, 3) because of the ability to utilize both the CCR5 and CXCR4 co-receptors, these viruses could bind both macrophages and astrocytes directly, causing a rapid and amplified production of cytotoxic and neurotoxic proteins from several cell-types and 4) this amplified cascade of events would disrupt neuronal signaling faster and could cause observed dementia before an accumulation of macrophages within brain tissues.
The fifth patient in the study, patient BW, had developed NH-ARL in the weeks prior to death that had metastasized into the CNS where it formed a neoplasm in the meninges. Since the development of HAART, the occurrence of NH-ARL has dropped approximately 50%, but it still occurs at a rate higher than in the normal population. NH-ARL in HIV-1 infected patients is diffusely aggressive and almost always fatal. In addition to common symptoms of lymphoma, HIV-1 infected patients with CNS lymphoma may present with seizures, headache, altered mental status, or other focal neurological deficits. Although symptoms of dementia may be present, the subsequent histological examination will be different than seen in patients with clinical HAD, showing less overall macrophage infiltrate into deep brain tissues. Patient BW was interesting in comparison to patients DY and AZ in that the individual’s brain viruses were highly compartmentalized when compared body sequences, which could indicate the spread of a potentially later-stage, lymphoma-specific virus that travelled within metastatic cells to the brain meninges where it spread to other brain tissues. A previous study showed that macrophages containing HIV-1 taken from a human NH-ARL tumor induced a tumor in a SCID model that containing related HIV, thus showing that HIV-1 infected macrophages were implicated in tumorogenesis in ARL (Zenger et al, 2002
). As ARL is an explosive disease process, this could also explain BW’s high rate of viral production as indicated by p24 staining.
Theories on how and when HIV-1 enters CNS tissues have been presented for many years. In the pre-HAART era, HIV-1 infection progressed faster and it is likely that fewer mechanisms for viral CNS entry existed. Additionally, studies on the prevalence of HAD within different subtypes of HIV-1 in similar populations points to the fact that a genetic determinant for CNS invasion of HIV-1 exists (Sacktor et al., 2009). In developed countries where HAART therapy is common, greatly reduced cases of HAD have been noted; however, the evolution of different mechanisms for the invasion of HIV-1 into brain tissues may be occurring and may further complicate the debate over how and when HIV-1 evolves within the CNS. Although the number of case studies examined here are small, the number of sequences examined are large and the results from the analysis presented support several mechanisms of HIV-1 viral entry into CNS tissues, including some that are associated with non-AIDS defining illnesses. The study highlights the fact that HIV-1 infection in the brain is likely specific to the host and to the host’s developing pathologies and physical reaction to HAART therapy.