| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH COMMUNICATION |
a Laboratory of Neurotoxicology, National Institute of Mental Health, Bethesda, Maryland 20892, USA
b New England Regional Primate Research Center, Harvard Medical School, Southborough, Massachusetts 01772, USA
c Neuropathology Division, Presbyterian-University Hospital, Pittsburgh Pennsylvania 15213, USA
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: macrophages • astrocytes • indoleamine-2,3-dioxygenase • kynurenine-3-hydroxylase • kynureninase • kynurenine aminotransferase • L-tryptophan • L-kynurenine • kynurenic acid • retrovirus • blood–brain barrier • cytokines • nonhuman primates
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Motor abnormalities, cognitive deficits, and dementia (encephalopathy) are frequent complications of infection with the human immunodeficiency virus (HIV) and can occur independent of opportunistic CNS infections (reviewed in ref 1). The most frequent neuropathologic substrate associated with neurologic symptoms is HIV encephalitis, an inflammatory condition characterized by the presence of HIV-infected macrophages, astrogliosis, white matter pallor, and neuronal injury (loss of neurons and synapses). Neurologic deficits can also occur in early stage patients, although their relationship to a neuropathologic substrate remains unclear. The mechanisms responsible for neurologic dysfunction and neurodegeneration are unknown, but because HIV is localized predominantly in macrophage/microglia and macrophage-tropic isolates are associated with neurologic disease to a greater extent than T cell tropic isolates (2), the production of toxins by macrophage/microglia has been hypothesized. Potential host-coded neurotoxins include the NMDA receptor agonist quinolinic acid (QUIN), cytokines, nitric oxide, platelet activating factor, autoantibodies, and prostaglandin metabolites (3–8); potential virus-coded neurotoxins include gp120, fragments of gp120, gp41, components of the structural proteins env and gag, and the regulatory proteins tat, nef, and rev (9–11). Additional unidentified toxic mechanisms have also been implicated (12–15).
QUIN is an excitotoxic metabolite of the tryptophan–kynurenine pathway (
Fig. 1;
ref 16). Sustained increases in the concentrations of QUIN occur in cerebral spinal fluid (CSF) and blood of HIV-infected patients and macaques infected with the simian immunodeficiency virus (SIV), and begin soon after primary infection. Elevated CSF QUIN is associated with motor deficits and virus recovery from the CNS in the early asymptomatic stages of disease, and correlate with quantitative measures of neuropsychologic deficits, striatal and limbic atrophy, and markers of intrathecal immune activation (CSF ß2-microglobulin and neopterin concentrations) in late stage patients (3, 17–25). Because CNS QUIN levels are not elevated in noninflammatory disorders such as Huntington's disease (18, 26), Alzheimer's disease (18, 27), and complex partial seizures (28), the elevations of QUIN in inflammatory diseases cannot be attributed to motor, cognitive, or convulsant symptoms or to neurodegeneration and astrogliosis per se (18, 19). QUIN accumulations are hypothesized to reflect combinations of increased QUIN production within brain parenchyma as well as entry of QUIN into the CNS from the blood (3). Macrophages and microglia activated by proinflammatory stimuli such as cytokines, HIV itself, or opportunistic conditions were postulated to be important cellular sources of QUIN (29–35).
|
Recently, Giulian and co-workers (13) reported that brain parenchymal QUIN concentrations are not elevated in HIV-infected patients and that the elevations in CSF QUIN levels are attributable to QUIN production by the meninges or diffusion of QUIN into the brain from the blood. They also reported that an HIV-infected macrophage cell line (THP-1) did not produce QUIN.
The present study has determined whether and to what degree QUIN concentrations are increased in the brain parenchyma of AIDS patients. We also evaluated whether the increases in CSF QUIN levels are explained by the production of QUIN by brain tissue, by the meninges, or by entry of QUIN into the brain from the blood. HIV-infected patients usually die with cytomegalovirus infection of the brain, which confounds the interpretation of results. We therefore used the retrovirus-infected macaque model of HIV-associated anatomical neurologic disease (without opportunistic CNS infections) to evaluate the enzymatic mechanisms involved in accelerating QUIN production within the CNS and systemic tissues. Activities of the kynurenine pathway enzymes (
Fig. 1) indoleamine-2,3-dioxygenase (IDO), kynureninase, kynurenine-3-hydroxylase, and 3-hydroxyanthranilate-3,4-dioxygenase were quantified in the cerebral cortex of macaques with and without local SIV encephalitis. The activities of kynurenine pathway enzymes in the lung were also determined. Hepatic tryptophan-2,3-dioxygenase activity, an important influence on systemic kynurenine pathway metabolites (36), was measured. The relationships between CNS and systemic responses were then assessed in the context of CNS encephalitis. The activity of kynurenine aminotransferase was quantified to investigate mechanisms that may be involved in the elevated CNS levels of kynurenic acid, an L-kynurenine metabolite and antagonist of excitatory amino acid receptors (16) that is elevated in retrovirus infection (17, 19) and may also contribute to neuronal dysfunction by interfering with excitatory amino acid neurotransmission. Finally, we directly measured the amounts of QUIN within the brain derived from blood vs. QUIN synthesized locally by brain tissue (37) in a noninfected control macaque as compared to an SIV-infected macaque with histologically verified encephalitis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Animals and viral infections
Rhesus macaques (Macaca mulatta) were housed in accordance with standards of the American Association for Accreditation of Laboratory Animal Care. Studies were approved by the Animal Care and Use Committees of the California Regional Primate Center (Davis, Calif.) or the New England Regional Primate Center (Southborough, Mass.).
Study 1
To obtain a broad spectrum of neurologic disease severity, samples from macaques infected with different retroviruses and serotypes were studied.
Macaques were divided into the following groups. Controls (n=27): healthy uninfected macaques (n=6); macaques with either chronic colitis or failure to thrive not associated with retroviral infection (n=15); antibody positive/virus negative for type-D retrovirus, (recovered, n=6). Simian retrovirus type D (SRV-D) (n=10): infected with type D retrovirus (D/1/California: simian AIDS, n=8; simian AIDS-related complex, n=2). SIV (n=23): infected with SIV: SIVsm (n=7), SIVmac-251, (n=11), SIVmac-239 (n=3), or SIVmac-1A11 (n=2).
Type D retroviruses are neuroinvasive oncornavirinae that induce an AIDS-like syndrome. SIV are lentiviruses with extensive sequence homology and with genomic organization, morphology, and biologic properties similar to that of HIV. The following SIV serotypes were studied: uncloned biological isolates of SIV (SIVsm and SIVmac-251); pathogenic (SIVmac-239) or nonpathogenic (SIVmac1A11) molecular clones. Macaques were infected via intravenous, intramuscular, or vaginal routes. Macaques infected with SIVsm, SIVmac-251, or SIVmac-239 were moribund from simian AIDS at the time of sample collection but were without opportunistic CNS infections. Macaques infected with SIVmac-1A11 did not develop AIDS, but several had colitis unrelated to lentiviral infection (colitis is common in captive macaques). Samples of cerebral cortex, lung, liver, CSF, and blood were collected at the time of euthanasia within 10 min of death and stored at -70°C until enzyme analysis. Additional samples of different brain regions were fixed in 10% buffered formalin for histopathologic examination. The time between death, sample freezing, and storage prior to the biochemical analyses were comparable among the four groups of macaques.
Study 2
Macaques were inoculated intravenously with 1000 animal infectious doses of SIVmac-251 (n=6) or tissue culture media (n=6 control). Samples (50–300 mg) of lung, liver, spleen, axillary lymph node, duodenum, plasma, and CSF were collected 4 wk later. None showed clinical signs or histopathological evidence of encephalitis, but were persistently viremic.
Study 3
The rate of QUIN excretion in the urine was determined in uninfected macaques (n=6) and macaques infected with SIVmac-239 (n=6). Urine was collected for 24 h.
Study 4
The fraction of QUIN derived from blood vs. de novo synthesis within the brain, CSF, and systemic tissues was determined in one control and one SIV-infected macaque with histologically verified encephalitis. Four osmotic pumps (Model 2ML1; ALZA Corporation, Palo Alto, Calif.) were implanted subcutaneously 4 days prior to death and infused with [13C7]QUIN (72 mM solution at 10 µl/h). The fraction of QUIN derived from de novo synthesis within the tissue is given by the formula (37):
 |
Biochemical analyses
The QUIN assay used was an electron capture negative chemical ionization-gas chromatography method that used either [18O4]QUIN or [2H3]QUIN as internal standard (38, 39). Quantification of enzymes and metabolites within any given set of tissues or experiments was run within the same assay with appropriate standard curves by methods referenced in ref 40. Urea and creatinine were measured by routine assays.
Histopathology, immunohistochemistry, and in situ hybridization: definition of encephalitis
Encephalitis is a generalized term that covers a broad spectrum of inflammatory responses including macrophage infiltration, microglial activation, and astrocytosis. Generally, patients with encephalitis show neurologic deficits, although not every patient with neurologic deficits has encephalitis. HIV encephalitis is distinct from encephalitits due to opportunistic infections, and is defined histopathologically as an inflammatory condition with HIV in the brain (measured by gp41 antigen) without evidence of opportunistic CNS infections. Both, one, or neither form of encephalitis may be present postmortem. Cytomegalovirus infection of the brain is common in AIDS patients, and is a confounding variable in the attribution and interpretation of CNS responses specifically to HIV encephalitis. SIV encephalitis differs from HIV encephalitis in that its definition includes the presence of multinucleated giant cells, which are much less common in HIV-infected brain. Cytomegalovirus infection is also much less common in the CNS of SIV-infected macaques with AIDS. Therefore, studies of SIV-infected macaques can allow a clearer discrimination of CNS responses to retrovirus infection and SIV encephalitis vs. encephalitis due to opportunistic conditions.
Human tissue
Tissue blocks from the frontal cortex gray and white matter and the basal ganglia were collected (41). To assess histopathology, routine hematoxylin- and eosin-stained sections from all blocks were examined. A monoclonal antibody obtained from Genetic Systems (Seattle, Wash.) that specifically recognizes the transmembrane portion of the HIV envelope protein gp41 was used to immunocytochemically identify HIV. Sections were deparaffinized in Histo-Clear (National Diagnostics, Manville, N.J.), treated with 1% H2O2 in methanol (to block endogenous peroxidase), rehydrated through graded alcohols, and rinsed in phosphate-buffered saline (PBS); tissues were incubated with primary antibody for 2 h at 37°C or overnight at 4°C, followed by PBS rinses and another incubation with biotinylated secondary antiserum (goat anti-mouse from Tago, Burlingame, Calif.) for 30 min at room temperature. After rinsing and a 30 min incubation with avidin-biotin-peroxidase complex (ABC kit, Vector Laboratories, Burlingame, Calif.) at room temperature, the enzyme was developed with 3-amino-9-ethylcarbazole and the sections were counterstained with Meyer's hematoxylin. Tissue sections were studied for the presence of HIVgp41; levels of HIV antigen expression were assessed separately for each region and scored on a scale from 0 to 2; 0 = no cells stained for gp41, 1 = less than three cells stained for gp41 in five 20x fields, and 2 = more than two cells stained for gp41 in five 20x fields. All assessments were done blindly and independently by two of the authors (C.A.W. and C.L.A.).
Macaque tissue
Macaque necropsy specimens were immediately frozen in liquid nitrogen-cooled freon and stored at -70°C. An adjacent block of tissue was fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned at 6 µm, and stained with hematoxylin and eosin or used for in situ hybridization to localize SIV nucleic acid. Macrophage infiltration and major histocompatability class II expression in the CNS were determined in cryostat sections (6 µm) by monoclonal antibodies specific for CD68 (EBM-11, Dako Corporation, Carpinteria, Calif.) and rhesus major histocompatability class II, respectively. Controls were normal uninfected animals, omission of the primary antibody, and substitution of an isotype-matched monoclonal antibody of irrelevant specificity. The SIV nucleic acid probe was a combination of two plasmids that provides essentially the entire SIVmac genome. The probes were labeled with digoxigenin-dUTP by random priming. Probe size and labeling were determined by electrophoresis, blotting, and immunostaining of the DNA with anti-digoxigenin antibodies. Sections were immunostained by an unlabeled polyclonal sheep antidigoxigenin antibody. As a negative control, sections were hybridized with the plasmid PUC19 labeled with digoxigenin at the same time as the probes. In addition, as a negative control for immunostaining, representative sections were processed identically using equivalent concentrations of sheep gamma globulin in place of the primary antibody. All sections were lightly counterstained with Mayer's hematoxylin.
Statistical analyses
Results were analyzed by one- and two-way analyses of variance with either Dunnett's t test or the Kruskall-Wallis test as the post hoc analysis, paired and unpaired t tests, or the chi-square test. Linear regression analysis was performed by the method of least squares. Statistical analyses of QUIN, L-kynurenine, kynurenic acid, and IDO data were done after log10 transformation. Values presented are mean ± 1 SEM. A P value of <0.01 was considered statistically significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Fig. 2,
bars).
|
In HIV-infected patients, average CNS QUIN levels were markedly elevated in basal ganglia (291-fold), cerebral cortex white matter (338-fold), and cerebral cortex gray matter (324-fold) compared to controls ( Fig. 2; note: the lines connect individual patients' values). Patients had higher QUIN levels in basal ganglia compared to corresponding values in both cortical white matter (80% of patients by 43±10%, P<0.01 by chi-square) and cortical gray matter (67% of patients by 37±14%, P<0.01). QUIN levels in basal ganglia were correlated with those in cortical white matter (r=0.90; P<0.01) and cortical gray matter (r=0.94; P<0.01); QUIN levels in cerebral cortex white matter also correlated with those in cortical gray matter (r=0.89; P<0.01). There were significant correlations between local ß2-microglobulin concentrations and QUIN levels in basal ganglia (r=0.37, P<0.02), cortical white matter (r=0.33; P<0.05), and cortical gray matter (r=0.31; P<0.05). There were no correlations between local QUIN levels and gp41 scores within any brain region.
CSF QUIN concentrations were also elevated in HIV-infected patients (171-fold). Measures of CSF QUIN were an index of parenchymal QUIN concentrations, as demonstrated by the close correlations between CSF and QUIN levels in basal ganglia (r=0.54; P<0.01), cortical white matter (r=0.52; P<0.001), and cortical gray matter (r=0.53; P<0.001), although parenchymal concentrations were underestimated by 7.8 ± 1.1-fold from the CSF values.
In the 16 deceased AIDS patients where brain, CSF, and serum QUIN levels were quantified, serum QUIN concentrations averaged 37.3 ± eightfold higher than controls. The ratio of QUIN concentrations in serum:basal ganglia, cortical gray matter, cortical white matter, and CSF averaged 1.92 ± 0.66, 1.86 ± 0.55, 1.48 ± 0.49, and 0.70 ± 0.34, respectively. There were no significant correlations between serum QUIN concentrations and QUIN levels in basal ganglia, cortical gray matter, cortical white matter, or CSF.
Kynurenine pathway metabolism in retrovirus-infected macaques
Histology
There were no inflammatory lesions in the brain of uninfected macaques or type D antibody-positive/virus-negative macaques or a significant difference in any biochemical measure between these two groups (control). Inflammatory lesions that defined SIV encephalitis were found in either cerebral cortex or subcortical areas of 9 of 21 macaques (5 of 7 infected with SIVsm, 3 of 11 infected with SIVmac-251, and 1 of 3 infected with SIVmac-239) but in neither of the two macaques infected with SIVmac-1A11. The inflammatory brain lesions were characterized by multifocal, perivascular aggregates of macrophages and multinucleated giant cells, and were visible by hematoxylin and eosin. By immunohistochemistry for the macrophage-specific marker CD68, these histologically discrete lesions were found to be part of larger, more diffuse infiltrates of CD68-positive macrophages or activated microglia ( Fig. 3B).
Furthermore, major histocompatability class II expression was associated with these infiltrates ( Fig. 3C), indicating CNS immune activation. Because immunohistochemistry for CD68 revealed much more extensive evidence of macrophage infiltration or microglia activation than was visible by routine histology, this analysis was also performed on macaques without encephalitis to determine whether there were subtle infiltrates that were not detected by routine histology. These animals did have increased numbers of CD68-positive macrophages or activated microglia throughout the white matter ( Fig. 3.A). Similar, although less extensive, perivascular infiltrates of CD68-positive cells were seen in animals dying from type D retrovirus infection ( Fig. 3D). Thus, retroviral infection of the CNS in macaques can be associated with increased numbers of CD68-positive cells that are not detected by routine histology. These CD68-positive cells are either recruited peripheral blood monocyte/macrophages or activated microglia and are associated with expression of major histocompatability class I antigen.
|
Classification of macaques for data analysis (Study 1)
Analysis of group plasma and CSF metabolite concentrations and of systemic enzyme responses were done comparing control, SRV-D (with no encephalitis), SIV without encephalitis, and SIV with encephalitis (in either cerebral cortex or subcortical areas). Because of the hypothesized importance of macrophage infiltrates in the kynurenine pathway responses, cerebral cortex QUIN concentrations and enzyme activities were then analyzed after further subclassification: SIV encephalitis cortex: inflammatory lesions in cerebral cortex immediately adjacent to sample collected for biochemical analyses (n=4); SIV encephalitis subcortex: inflammatory brain lesions located in subcortical areas (n=5).
CNS responses (Study 1)
CSF Elevations in CSF L-kynurenine, QUIN, and kynurenic acid concentrations ( Fig. 4)
and the increased ratio of QUIN:kynurenic acid were highest in the macaques with SIV encephalitis. Although there were significant correlations between QUIN levels in the CSF and plasma in both uninfected controls (r=0.81, P<0.01) and SRV-D-infected macaques (r=0.91, P<0.01), there were no correlations between CSF and plasma QUIN levels in SIV-infected macaques due to the presence of encephalitis. Also, three of the SIV-infected macaques with encephalitis had higher QUIN levels in both CSF and brain tissue than in the blood.
|
Cerebral cortex QUIN concentrations in brain parenchyma exceeded CSF in all macaques studied. Cerebral cortex QUIN and enzyme activity measurements were analyzed after further classification into macaques with inflammatory lesions in the cerebral cortex vs. subcortical areas. Cerebral cortex QUIN concentrations ( Fig. 5)
and the activities of IDO ( Fig. 6A)
and kynurenine-3-hydroxylase ( Fig. 6B) in cerebral cortex were increased in all infected macaques. These cortical increases were significantly larger in macaques with cortical inflammatory lesions (designated by a dagger). The activity of cerebral cortex kynureninase ( Fig. 6C) was significantly increased only in macaques with cortical inflammatory lesions. There were no significant differences in cerebral cortex 3-hydroxyanthranilate-3,4-dioxygenase ( Fig. 6D) or cortical kynurenine aminotransferase activities (control, 59±6 nmol·g-1·h-1; no CNS lesions, 59±6 nmol·g-1h-1; subcortical encephalitis, 59±7 nmol·g-1h-1; cortical encephalitis, 52±4 nmol· g-1·h-1).
|
|
In the data set taken collectively, cerebral cortex QUIN levels correlated with local activities of cerebral cortex IDO (r=0.85, P<0.01), kynurenine-3-hydroxylase (r=0.60, P<0.01), and kynureninease (0.65, P<0.01), but not 3-hydroxyanthranilate-3,4-dioxygenase (r=0.29, NS). CSF L-kynurenine levels correlated with CSF QUIN (r=0.89, P<0.01) and kynurenic acid (r=0.86, P<0.01) levels but not with kynurenine aminotransferase activity.
Systemic responses (Studies 1–3)
Plasma QUIN and L-kynurenine levels and the activities of IDO, kynurenine-3-hydroxylase, kynureninease, and 3-hydroxyanthranilate-3,4-dioxygenase in the lung ( Table 1)
were significantly increased in the three groups of retrovirus-infected macaques. In striking contrast to CNS changes, there was no significant difference in the magnitude of these responses when comparing macaques with encephalitis vs. those without. Liver tryptophan-2,3-dioxygenase activity did not differ between uninfected controls and retrovirus-infected macaques.
|
In the separate study of systemic tissue QUIN levels (Study 2; Fig. 7),
SIV-infected macaques had no demonstrable CNS lesions according to routine histologic examination. QUIN levels were elevated in SIV-infected macaques in plasma, duodenum, liver, lung, lymph node, and spleen compared to controls. Note that systemic tissue QUIN concentrations exceed blood levels. Four of the six SIV-infected macaques had inflammatory lung lesions that consisted of large numbers of activated macrophages and multinucleated giant cells. QUIN levels within the lesion area of the lung were substantially higher than in areas from the same macaque that contained no such lesions.
|
Elevated serum QUIN levels could not be attributed to impaired renal excretion rates. SIV-infected macaques had significantly higher excretion rates of QUIN in the urine than uninfected controls and significantly higher serum QUIN levels (Study 3; Table 2).
Clearance rates of both QUIN and creatinine, however, were not significantly different between controls and SIV-infected macaques.
|
Fractions of QUIN in brain, CSF, and systemic tissue derived from tissue vs. blood (Study 4)
Serum QUIN and [13C7]QUIN concentrations remained stable on the second, third, and fourth day after implantation of the osmotic pumps, demonstrating that steady-state infusion rates and equilibrium had been achieved as predicted (37). In the control macaque, the percentage of QUIN derived from tissue was 
68% in all CNS regions (median: 85%; Fig. 8)
and was 99% in spleen (16107 pmol/g), 98% in lung (14410 pmol/g), 91% in liver (716 pmol/g), 83% in duodenum (778 pmol/g), 84% in skeletal muscle (358 pmol/g), 65% in heart (458 pmol/g), and 50% in kidney (1838 pmol/g). Serum QUIN concentrations were 1531 nM and CSF QUIN levels were 44 nM. In the SIV-infected macaque, brain QUIN levels were up to 100-fold higher than in the control macaque ( Fig. 8). At least 92% of QUIN in brain was derived from tissue (median: 98%) and was 97% in spleen (116390 pmol/g), 98% in lung (73232 pmol/g), 94% in liver (25894 pmol/g), 37% in duodenum (5592 pmol/g), and 57% in kidney (37479 pmol/g). Serum QUIN concentrations were 4851 nM and CSF QUIN levels were 5582 nM. Average brain tissue QUIN values were highest in areas of encephalitis: severe, 13502±2240 pmol/g (n=7); moderate, 12137±3147 pmol/g (n=6); mild, 10775 ± 3070 (n=3); minimal, 7405 pmol/g (n=1); none, 2477 ± 652 pmol/g (n=2).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Fig. 2) and, notably, achieved concentrations that exceeded levels reported to be neurotoxic in vitro (20–100 nM) by up to three orders of magnitude (12, 13, 42). Blood QUIN levels were also increased, but there was no correlation between blood and brain tissue QUIN levels. Because of the confounding variables of cytomegalovirus infection of the CNS and severe systemic diseases in the HIV-infected patients, we used retrovirus-infected macaques to specifically investigate the relationships between kynurenine pathway responses to both encephalitis and systemic responses. The analyses showed that the largest increases in macaque brain tissue ( Fig. 5and Fig. 8) and CSF QUIN levels ( Fig. 4), as well as enzyme activities ( Fig. 6), occurred in conjunction with local inflammation and encephalitis ( Fig. 3). There was, however, no difference in the magnitude of systemic kynurenine pathway responses between macaques with SIV encephalitis vs. those without ( Table 1). The increases in blood levels are attributable to QUIN synthesis in lung and systemic tissues ( Fig. 4) and not to impairments in renal QUIN excretion ( Table 2). Direct measures of the amount of QUIN derived locally vs. originating from blood showed that the overwhelming source of QUIN in SIV encephalitis was the brain tissue itself ( Fig. 8).
In contrast to CNS kynurenine pathway responses, the small increases in plasma QUIN and L-kynurenine concentrations with increased enzyme activities in lung occurred in all three classifications of retrovirus-infected macaques to the same degree ( Table 1). Therefore, the enhanced responses in areas of encephalitits cannot be attributed to differences in systemic responses per se. The elevations in QUIN and L-kynurenine levels in plasma are explained by the increases in kynurenine pathway enzyme activities. The lung may be a particularly important source of blood QUIN as it receives the cardiac output, has a high anabolic capacity for QUIN compared to other systemic organs (43), and shows large elevations in QUIN levels in the areas of inflammation ( Fig. 5). Additional systemic tissues likely contribute to increased blood QUIN levels because the elevated tissue concentrations in SIV-infected macaques exceeded blood levels in all systemic tissues examined ( Fig. 7) and all systemic organs examined derived QUIN by de novo synthesis ([13C7]QUIN infusion results). Although hepatic tryptophan-2,3-dioxygenase is a regulatory kynurenine pathway enzyme (36), no changes in activity were found ( Table 1), an observation consistent with mouse and gerbil data, which show that tryptophan-2,3-dioxygenase is not induced by immune stimuli (43, 44). The small increase in hepatic QUIN levels ( Fig. 7) is explained by induction of hepatic IDO or entry of QUIN precursors such as L-kynurenine from the blood ( Table 1). Because the biological half-life of QUIN in blood and CSF, and the half time for QUIN in blood to distribute and reach equilibrium in brain and systemic tissues, is less than 3 h (17, 37, 45), QUIN must be continuously synthesized and released to persistently elevate brain and blood levels and increase urinary QUIN excretion ( Table 2).
The similarities in systemic responses between macaques with or without SIV encephalitis ( Table 1) contrasts with the enhanced accumulations of QUIN, L-kynurenine, and kynurenic acid in the CSF of macaques with SIV encephalitis ( Fig. 4), as well as the exacerbated elevations in parenchymal QUIN concentrations ( Fig. 5) and higher activities of IDO, kynurenine-3-hydroxylase, and kynureninase specifically in areas of cortical encephalitis ( Fig. 6). An analogous disassociation between CNS and systemic metabolism has also been reported in macaque, gerbil, guinea pig, and mouse models of CNS-localized injury and inflammation independent of HIV infection (34, 37, 40, 46–49). There are, however, conditions of systemic immune activation where brain QUIN increases correlate with blood QUIN levels and where direct entry of QUIN into the brain from the blood is a quantitatively important source of QUIN in brain (37, 43, 49–52). Systemic immune activation is, of course, present in retrovirus infection and accounts for the elevations in blood QUIN levels in terminal AIDS patients ( Fig. 2; ref 3) and infected macaques ( Table 1). We have also reported that CSF and serum QUIN concentrations are correlated in later stage HIV-infected patients (3), with a modest correlation between blood–brain barrier permeability (CSF:plasma albumin ratio) and CSF QUIN levels in one subgroup of HIV-infected patients (demented, later stage patients). Therefore, some QUIN within the brain parenchyma ( Fig. 2and Fig. 6) may have originated from blood, particularly in areas of increased blood–brain barrier permeability, although it is not possible to quantify the contribution of blood to the elevations in brain and CSF QUIN in HIV-infected patients as we have done in the SIV-infected macaque. In the present study, however, there was no correlation between blood QUIN levels and CNS QUIN levels in terminal AIDS patients or in macaques with encephalitis as would be expected if blood was the main determinant of CNS QUIN levels. These observations indicate that factors other than blood QUIN levels are principal determinants of the kynurenine pathway responses in areas of encephalitis. Direct measures of tissue vs. blood contributions to QUIN levels in the SIV-infected macaque with encephalitis showed that almost all of the elevated QUIN in the brain parenchyma originated by local do novo synthesis and not from blood ( Fig. 8).
In the human postmortem data ( Fig. 2), significant correlations were observed between local QUIN concentrations and ß2-microglobulin levels, a marker of immune activation. These results are consistent with our findings of highly significant correlations between CSF neopterin and ß2-microglobulin levels with CSF QUIN, L-kynurenine, and kynurenic acid concentrations, but reciprocal correlations to CSF L-tryptophan levels (19). Furthermore, previous studies have demonstrated in models of CNS-localized inflammation that brain tissue from areas of brain inflammation has increased activities of IDO, kynurenine-3-hydroxylase, and kynureninase, similar to those observed in areas of SIV encephalitits, as well as increased ability to directly convert L-tryptophan to QUIN (40, 46, 48). These observations highlight the important role of local inflammation in producing exacerbated elevations in parenchymal QUIN levels in areas of encephalitis. Thus, the contribution of blood QUIN to brain QUIN likely varies according to the degree of local inflammation, the degree of blood elevations, and changes in blood–brain-barrier permeability (53). However, in the case of a neuron sensitive to the neurotoxic effects of QUIN, the source of QUIN is irrelevant provided QUIN has access to the mechanisms that mediate toxicity (namely, NMDA receptors accessed via the extracellular space). We have established that increases in local brain QUIN biosynthesis due to inflammation result in large elevations of QUIN in the extracellular fluid (49). The observation that CSF QUIN levels are elevated in retrovirus infection ( Fig. 2and Fig. 4) is consistent with elevations in QUIN in the extracellular fluid compartment.
The significant correlations between cortical QUIN levels and local IDO activity support the hypothesis that induction of IDO is an important initiating event in increasing the production of kynurenine pathway metabolites (40), a response facilitated by small increases in local kynureninase and kynurenine-3-hydroxylase activities ( Fig. 6). The induction of IDO and production of L-kynurenine in brain has been replicated in vitro in human microglia, macrophages, and astrocytes by stimulating such cells with endotoxin or physiologic concentrations of interferon 
and tumor necrosis factor 
(35, 46, 54–56). Thus, many cell types are sources of L-kynurenine and kynurenic acid within the CNS. Entry of L-kynurenine into the brain as a result of elevated blood levels with conversion to kynurenic acid (57) may also occur. The enhanced accumulations of QUIN in areas of encephalitis most likely occurred in activated microglia as well as macrophage infiltrates into the brain parenchyma, as both cell types have a high capacity to synthesize QUIN (29, 31, 35, 40, 54, 58). Activated astrocytes synthesize relatively small amounts of QUIN (31, 40) because of their low activities of kynurenine-3-hydroxylase, kynureninase, and 3-hydroxyanthranilate-3,4-dioxygenase compared to monocyte/macrophages (32). QUIN production by activated microglia (35) can also account for the smaller elevations of QUIN in macaques without encephalitis, as immunocytochemical studies showed microglial activation in these macaques ( Fig. 3D).
Several stimuli are likely responsible for the enhanced kynurenine pathway metabolism, including retrovirus infection, opportunistic infections, and cytokines. In cultures of human macrophages, infection with HIV enhances the production of [13C6]QUIN from [13C6]L-tryptophan, with the largest QUIN accumulations occurring in cultures infected with either macrophage-tropic HIV strains or wild-type strains isolated from patients with HIV-associated neurologic disease (30). Additional studies have also reported QUIN production by macrophages in response to HIV infection (29, 59). Increased QUIN immunoreactivity has also been reported in peripheral blood macrophages from SIV-infected macaques (60). The concentrations of QUIN achieved in the media of HIV-1-infected macrophages (30) were comparable to parenchymal values ( Fig. 1) and ranged from 1,000 to 10,000 nM, amounts well in excess of in vitro neurotoxic concentrations (13). Indeed, 6-chloro-tryptophan, an inhibitor of IDO activity and QUIN production (40, 54), reduces HIV-induced macrophage QUIN production and neurotoxicity in vitro (61). Lymphotropic strains of HIV, including the IIIBx strain used by Giulian and co-workers (13) to infect THP-1 cells, did not induce the production of QUIN (30). It has been demonstrated, however, THP-1 cells are capable of producing [13C6]QUIN from [13C6]L-tryptophan in response to interferon stimulation (40, 54).
On the other hand, whereas brain QUIN concentrations in HIV-1-infected patients correlated with local ß2-microglobulin levels, there was no correlation with gp41 scores. These observations indicate that factors other than HIV infection mediate the increase in local kynurenine pathway metabolism, including non-HIV encephalitis (59). We have also reported a broad spectrum of conditions independent of retrovirus infection with inflammation localized within the brain (physical trauma and injury, microbial infections, and autoimmune conditions) where 1) local brain QUIN levels are selectively elevated in proportion to the severity of injury and inflammation; 2) blood QUIN levels are unchanged; and 3) brain tissue QUIN levels exceed blood concentrations (18, 34, 40, 44, 46–49, 51, 55, 62–66). Therefore, HIV infection, either latent or productive, is clearly not a requirement for the QUIN accumulations in HIV-infected patients. Several studies have demonstrated that bacterial endotoxins and cytokines, particularly interferon , ß, and and tumor necrosis factor in combination, are potent stimuli of QUIN production by primary peripheral blood macrophages, brain microglia, and THP-1 cells (29, 31, 54, 58, 59). Therefore, cytokines produced in response to retrovirus infection, opportunistic infections, or injury may be important links leading to increased kynurenine pathway metabolism.
The high CSF QUIN concentrations previously reported in patients with meningitis (3, 18), coupled with the claims by Giulian and co-workers (13) that QUIN levels are not elevated in the brain tissue of HIV-infected patients, raise the possibility that CSF QUIN originates from the meninges and does not enter the brain parenchyma. In that case, CSF QUIN measures would not be an index of parenchymal values, and QUIN could not induce neuronal dysfunction and neurodegeneration. Clearly, however, brain parenchyma QUIN levels are elevated in HIV-infected patients ( Fig. 2) as compared to victims of suicide and patients with dementia, motor disturbances, and neurologic disease caused by Alzheimer's disease, Huntington's disease, and complex partial seizures (26–28). Despite the fact that 38 of the 42 HIV-infected patients studied had mild to severe meningitis, parenchymal QUIN concentrations exceeded CSF values by almost an order of magnitude. These results directly refute the notion that parenchymal levels are not elevated in HIV-infected patients and do not support the hypothesis that the meninges are the source of CSF QUIN (13). This does not mean that the meninges cannot be a source of QUIN in the CSF. As reported previously, HIV-infected patients with aseptic meningitis do have large elevations in CSF QUIN levels (3), but not necessarily neurologic deficits or elevated parenchymal QUIN levels (67). Also, certain bacteria are capable of synthesizing QUIN. Therefore, there may be conditions of elevated QUIN concentrations due to meningeal or bacterial production with minimal or no elevations of parenchymal levels.
In conjunction with increased CSF L-kynurenine levels was a proportional increase in CSF kynurenic acid concentrations ( Fig. 4), but no significant increase in cerebral cortex kynurenine aminotransferase activity. Studies in normal rats have shown that conversion of L-kynurenine to kynurenic acid can be substrate driven within the CNS (57). Therefore, the significant correlation between CSF L-kynurenine and kynurenic acid level suggests that elevations in CSF kynurenic acid are dependent on changes in IDO activity and the availability of L-kynurenine. It is also possible that L-kynurenine may have entered the CNS from blood. In a poliovirus model of CNS inflammation in nonhuman primates, we found highly significant correlations between brain IDO activity, CSF L-kynurenine concentrations, and CSF kynurenic acid levels, without changes in kynurenine aminotransferase activity (40). Accumulations of kynurenic acid may attenuate the excitotoxic effects of enhanced NMDA activity, but may also contribute to neurologic dysfunction by interfering with excitatory amino acid neurotransmission. Such a response may be a caveat in chronic therapeutic strategies designed to reduce QUIN synthesis and enhance kynurenic acid formation.
Compared to previous values in HIV-infected adults (3), the elevated CSF QUIN concentrations spanned the range typical of AIDS ( Fig. 1), with the majority of levels comparable to patients with severe dementia or opportunistic CNS conditions. The high correlation between CSF and parenchymal levels also demonstrates that CSF is an index of parenchymal values. Giulian and co-workers (13) also assessed CSF QUIN levels in a subgroup of HIV-infected patients (unspecified disease stage, type, and severity of neurologic deficits, antiretroviral, and illicit drug histories) and compared their elevated levels to `control' patients with multiple sclerosis, an inflammatory neurologic disease with CSF values that averaged 74 nM, a value already over threefold higher than neurologically normal control (22 nM; see refs 18, 26–28, 68, 69)). Giulian and co-workers (13) did not study more severely affected individuals or include any quantitative measures of neurologic deficits, and so could not examine the relationship between CSF QUIN levels and severity of neurologic symptoms. We have reported that, in the early stages of SIV and HIV infection (Walter Reed stages 1 through 4), elevated CSF QUIN levels are associated with subtle motor slowing (21–23), an observation in accordance with the involvement of the basal ganglia in HIV-associated neurologic disease and the sensitivity of the striatum to the neurotoxic effects of QUIN (16). Recently, we have also found significant correlations between CSF QUIN concentrations and the degree of brain atrophy in the striatum (caudate nucleus, putamen, and globus pallidus) and limbic system (mesial temporal lobe uncus, amygdala, hippocampus, and parahippocampal gyrus) as assessed by in vivo brain imaging in HIV-infected adult patients levels (L. Ryan, R. Ellis, M. P. Heyes, and T. Jernigan, unpublished observations). In this latter cohort of patients, there were also significant correlations between measures of learning and CSF QUIN levels (L. Ryan, R. Ellis, M. P. Heyes, and T. Jernigan, unpublished observations). In later stage patients, where CSF QUIN levels are increased still further, the scope and severity of neurologic deficits increase and, again, CSF QUIN levels are quantitatively correlated to severity of neurologic deficits (3). Similarly, CSF QUIN levels in HIV-infected children correlate with quantitative measures of delays in neurologic development (20), an observation we have since confirmed in an independent cohort of patients (M. P. Heyes and P. Brouwers, unpublished observations). Thus, QUIN remains the only neuropathogenic mechanism implicated in HIV-associated neurologic disease unequivocally identified within the brain whose CNS concentrations correlate quantitatively with the severity of functional neurologic deficits, as assessed in independent cohorts of patients and evaluated at different stages of disease.
The notion that HIV-associated neurologic disease results from the effects of virus- or host-coded toxins produced within the brain by HIV-infected macrophages and microglia has received increased attention, and both NMDA receptor-dependent (10, 12, 70) and -independent mechanisms have been implicated (4–8, 11, 14, 15). Certain putative toxins remain to be identified or shown to be present in physiologically meaningful levels in actual patients. The present results support the following model to explain the progressive increases in kynurenine pathway metabolites within the CNS during retrovirus infection. The small (two- to fourfold) elevations in CSF QUIN in the early stages of infection may reflect increased synthesis within the brain itself as well as entry of QUIN into the brain from the blood. Microglia and/or macrophage infiltrates, activated either by direct retrovirus infection or in response to cytokines released in response to HIV, may be important sites of QUIN and kynurenic acid production. In later stages of infection, when encephalitis is more evident, local activated microglia and macrophage infiltrates contribute to the substantially larger accumulations of QUIN (10- to 1000-fold). These accumulations cannot be attributed simply to diffusion of QUIN into the brain across a breached blood–brain barrier or to production of QUIN by the meninges, but result from local de novo synthesis in response to HIV or opportunistic infections. The concentrations of QUIN achieved in brain tissue exceeded levels required for the neurotoxicity of susceptible neurons. QUIN and kynurenic acid may also produce noncytolytic and potentially reversible neurologic dysfunction by interfering with excitatory amino acid neurotransmission.
The therapeutic implications of our collective studies are that strategies to attenuate the accumulations of QUIN, targeted at predominantly central sources, may reduce the degree of NMDA receptor activation and symptom progression in HIV-infected patients. Because intracerebral immune activation and accumulations of QUIN are a hallmark of many neurologic diseases, such as physical trauma, stroke, CNS infection, and autoimmune conditions, QUIN may have a central neuropathogenic role in a broad spectrum of inflammatory neurologic conditions (18, 71). QUIN may act as an `amplification' mediator that exaggerates the impact of relatively small amounts of virus, injury, and other immune stimuli to produce neurologic dysfunction. Protean symptoms are accounted for by QUIN being present in variable amounts in brain regions that mediate variable neurologic functions and contain different amounts and sensitivities of NMDA receptors. In accordance with this hypothesis are our observations that attenuation of local QUIN accumulations after spinal cord contusion injury by treatment with 4-chloro-3-hydroxyanthranilate also significantly reduce the severity of delayed neurologic deficits that develop in association with inflammatory processes (47, 72).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Abbreviations: CNS, central nervous system; CSF, cerebrospinal fluid; HIV, human immunodeficiency virus; IDO, indoleamine-2,3-dioxygenase; n, number of patients or samples studied; QUIN, quinolinic acid; SIV, simian immunodeficiency virus; SRV-D, simian retrovirus type-D; PBS, phosphate-buffered saline.
Received for publication November 14, 1997. Accepted for publication .
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, and nitric oxide in glal cultures ferived from fetal, neonatal, and adult human brain. J. Exp. Med. 182, 941–952
P < 0.01 compared to SIV-infected macaques with no CNS lesions.
