Persistent hijacking of brain proteasomes in HIV-associated dementia

Abstract

Immunoproteasome induction sustains class 1 antigen presentation and immunological vigilance against HIV-1 in the brain. Investigation of HIV-1-associated alterations in brain protein turnover by the ubiquitin-proteasome system was performed by (1) determining proteasome subunit changes associated with persistent brain inflammation due to HIV-1; (2) determining whether these changes are related to HIV-1 neurocognitive disturbances, encephalitis, and viral loads; and (3) localizing proteasome subunits in brain cells and synapses. On the basis of neurocognitive performance, virological, and immunological measurements obtained within 6 months before death, 153 autopsy cases were selected. Semiquantitative immunoblot analysis performed in the dorsolateral prefrontal cortex revealed up to threefold induction of immunoproteasome subunits LMP7 and PA28alpha in HIV-1-infected subjects and was strongly related to diagnoses of neuropsychological impairment and HIV encephalitis. Low performance on neurocognitive tests specific for dorsolateral prefrontal cortex functioning domains was selectively correlated with immunoproteasome induction. Immunohistochemistry and laser confocal microscopy were then used to localize immunoproteasome subunits to glial and neuronal elements including perikarya, dystrophic axons, and synapses. In addition, HIV loads in brain tissue, cerebrospinal fluid, and blood plasma were robustly correlated to immunoproteasome levels. This persistent "hijacking" of the proteasome by HIV-1-mediated inflammatory response and immunoproteasome induction in the brain is hypothesized to impede turnover of folded proteins in brain cells. This would disrupt neuronal and synaptic protein dynamics, contributing to HIV-1 neurocognitive disturbances.

Figure 1

Western blots illustrate altered proteasome subunit concentrations in DLPFC and WM from eight HIV⁺ subjects with HIVE and/or neurocognitive impairment compared with eight HIV⁻ subjects. A: Constitutive 20S proteasome subunits X(β5), Y(β1), and Z(β2) were not changed in HIV⁺ subjects in DLPFC or WM. B: Three inducible 20S immunoproteasome β subunits were increased in DLPFC and WM of HIV⁺ subjects. C: The majority of constitutively expressed 19S proteasome regulatory complex subunits remained unchanged, except for Rpn2, which showed a decrease in the DLPFC of HIV⁺ subjects. D: The immunoproteasome 11S regulatory subunit PA28α was increased in DLPFC and WM of HIV⁺ subjects. E: 20S α subunits were not changed.

Figure 2

Immunoproteasome subunit proteins in 88 HIV⁺ and 65 HIV⁻ subjects were quantified by using densitometry of calibrated Western blots with loading controls. Averaged LMP7 was increased 99% in DLPFC (A) and 184% in WM (B). PA28α was increased by 204% in DLPFC (C) and 233% in WM (D). Statistics: Student’s t test; *P < 10⁻⁷; **P < 10⁻¹⁰. OD = optical density.

Figure 3

Significantly increased immunoproteasome subunit LMP7 and PA28α concentration in HIV⁺ subjects with and without HIVE. A: DLPFC LMP7 levels in HIV⁺ subjects with and without HIVE were increased 173% and 81%, respectively, compared with HIV⁻ subjects. B: WM LMP7 in HIVE was increased 414% and 114% compared with HIV⁻ subjects and HIV⁺ subjects without HIVE, respectively. LMP7 levels of HIV⁺ subjects without HIVE were 140% greater than HIV⁻ subjects. C: DLPFC PA28α levels of HIV⁺ subjects with and without HIVE were 349% and 166% higher than HIV⁻ subjects, respectively. D: In WM, PA28α was increased 443% and 185% in HIV⁺ subjects with and without HIVE, respectively, compared with HIV⁻ subjects. Statistics: one-way analysis of variance with Tukey-Kramer Multiple Comparisons Test; *P < 0.05; ***P < 0.001. OD = optical density.

Figure 4

Increased immunoproteasome subunit LMP7 and PA28α concentration significantly related to NPI. Subunit level increases in HIV⁺ subjects with NPI-O were less pronounced. No substantial change was observed with HIV⁺ subjects without NPI. A: DLPFC LMP7 levels in HIV⁺ NPI subjects were increased 133% and 110% compared with HIV⁻ subjects and HIV⁺ subjects without NPI, respectively. B: WM LMP7 increased 243% in those with NPI and 154% in those with NPI-O compared with HIV⁻ subjects. C: DLPFC PA28α in HIV⁺ NPI subjects was increased 259% compared with HIV⁻ subjects and 216% compared with HIV⁺ subjects without NPI. HIV⁺ NPI-O subjects had a 126% increase compared with HIV⁻ subjects. D: WM PA28α was increased 85% in HIV⁺ subjects without NPI, 305% in HIV⁺ NPI subjects, and 159% in HIV⁺ NPI-O subjects compared with HIV⁻ subjects. Statistics: one-way analysis of variance with Tukey-Kramer Multiple Comparisons Test; *P < 0.05; **P < 0.01; ***P < 0.001. OD = optical density.

Figure 5

Immunoproteasome induction in DLPFC, but not WM, was correlated with worse performance in the abstract and executive functioning and speed of information processing neurocognitive domains. Decreased performance on the WCST-64 reflects abstract and executive function, and it was correlated with PA28α (A–C) and LMP7 (not illustrated). Scores for (A) categories completed, (B) total errors made, and (C) perseverative responses are shown. PA28α, but not LMP7, also was correlated with decreased performance on the Wechsler Adult Intelligence Scale III (WAIS-III) scores for (D) Digit-Symbol and (E) Symbol Search subtests, which reflects a decrease in the speed of information processing. Other neurocognitive domains that were tested were not correlated significantly. Statistics: Pearson’s correlation.

Figure 6

Immunoproteasome induction in DLPFC (A–D) and WM (E–H) was correlated significantly with virological and immunological status of HIV-infected subjects. HIV-1 RNA concentrations in brain (A and E), CSF (B and F), and blood plasma (C and G) were positively correlated with PA28α levels. CD4⁺ lymphocyte counts in blood plasma were negatively correlated with PA28α levels in DLPFC (D), but not WM (H). Equivalent results for LMP7 were obtained (not illustrated). Statistics: Pearson’s correlation.

Figure 7

Immunoperoxidase histochemistry illustrates immunoproteasome subunit staining for PA28α in glial and neuronal elements that are pathological in HIV encephalitis. Microglia, macrophages, and oligodendrocyte nuclei are stained in a microglial nodule (A). Neuronal cell bodies in neocortical neurons were stained in focal pathology (B) and were evident in laminae III and IV generally, though admittedly lighter than focal pathology (D). White matter axons often contained immunoproteasome subunits (C). Low immunoreactivity was observed in the (E) cortex and (F) white matter of HIV⁻ subjects.

Figure 8

Immunoproteasome subunits expressed in neurons in HIV encephalitis. Dual indirect immunofluorescence staining for LMP2 (A) or PA28α (B) shows colocalization with the neuronal markers NeuN and neurofilament in single optical sections from confocal microscopy. LMP2 is present in granular deposits in the parikaryon (A). PA28α is more prominent in the nucleus (B). A pathologically swollen axon in white matter contains PA28α (arrows in B), as do several other neurofilament-containing processes. Immunofluorescence stainings of HIV⁻ sections (insets) reveal an absence of LMP2 and PA28α. Scale bar = 10 μm.

Figure 9

Immunofluorescence and laser confocal microscopy reveal that immunoproteasome subunit LMP2 is localized to some neocortical synapses in HIV encephalitis. A: LMP2 was colocalized within punctate deposits of synaptophysin, which is an established cell marker of presynaptic boutons (arrows). About 12% of labeled synapses in the figure contain immunoproteasome antigenicity. Complete overlapping of these two antigens was evident in some synapses and suggests that immunoproteasomes often are present in presynaptic boutons (B). Some synapses had incomplete overlapping (C and D), which suggests that immunoproteasomes may be present in adjacent structures, including the postsynaptic density or in synaptic astrocyte foot processes. Scale bar = 10 μm.

Figure 10

Persistent “hijacking” of brain proteasomes in HIV-1-infected people may lead to neuronal dysfunction. Normally, proteasome complexes rapidly turn over ubiquitinylated proteins. Persistent infection with HIV-1 produces inflammatory cytokines such as IFN-γ that induce a temporary shift to immunoprotesome synthesis. In turn, the protein substrate repertoire is temporarily shifted toward the processing of unfolded peptides for class I antigen presentation. “Borrowing” the proteasome apparatus for heightened antigen presentation persists until the pathogen is eradicated, after which normal brain protein turnover should resume. Since HIV-1 infection is not eradicated in the brain, there is a persistent “hijacking” of the proteasome. Normal turnover of folded proteins is disrupted chronically. This leads eventually to the accumulation of pathologically misfolded proteins, neuronal dysfunction, and dementia.