Direct interaction of HIV gp120 with neuronal CXCR4 and CCR5 receptors induces cofilin-actin rod pathology via a cellular prion protein- and NOX-dependent mechanism

Abstract

Nearly 50% of individuals with long-term HIV infection are affected by the onset of progressive HIV-associated neurocognitive disorders (HAND). HIV infiltrates the central nervous system (CNS) early during primary infection where it establishes persistent infection in microglia (resident macrophages) and astrocytes that in turn release inflammatory cytokines, small neurotoxic mediators, and viral proteins. While the molecular mechanisms underlying pathology in HAND remain poorly understood, synaptodendritic damage has emerged as a hallmark of HIV infection of the CNS. Here, we report that the HIV viral envelope glycoprotein gp120 induces the formation of aberrant, rod-shaped cofilin-actin inclusions (rods) in cultured mouse hippocampal neurons via a signaling pathway common to other neurodegenerative stimuli including oligomeric, soluble amyloid-β and proinflammatory cytokines. Previous studies showed that synaptic function is impaired preferentially in the distal proximity of rods within dendrites. Our studies demonstrate gp120 binding to either chemokine co-receptor CCR5 or CXCR4 is capable of inducing rod formation, and signaling through this pathway requires active NADPH oxidase presumably through the formation of superoxide (O2-) and the expression of cellular prion protein (PrPC). These findings link gp120-mediated oxidative stress to the generation of rods, which may underlie early synaptic dysfunction observed in HAND.

Fig 1. HIV gp120 induces the formation of aberrant, rod-shaped cofilin-actin inclusions (rods).

After 18 h of gp120 exposure, dissociated mouse hippocampal neurons cultures were immunostained for cofilin, actin, and/or the growth cone antigen 2G13. (A) Dual tropic gp120MN (II), R5-tropic gp120 CM (III), or X4-tropic gp120IIIB (IV) at 250 pM caused a robust induction of rod-shaped cofilin-actin inclusions (arrow heads) in neurites neurons compared to untreated control (I). Cofilin immunoreactivity in growth cones is indicated by asterisk. (B) Rod-shaped inclusions were immunostained for both cofilin and actin (arrowheads). (C) Notably, the growth cone antigen 2G13 reveals growth cone-associated cofilin indicated by arrows in expanded boxed areas 1 and 2, structures clearly distinct from rods (arrowheads), which are in regions not immunoreactive to the growth cone antigen.

Fig 2. Different tropic gp120 strains induce dose- and time-dependent rod formation in hippocampal neurons.

(A) Increasing concentrations of dual tropic gp120_MN, R5-tropic gp120_BaL, or X4-tropic gp120_IIIB revealed a dose-dependent formation of rods in processes of hippocampal neurons quantified either as the percentage of neurons forming rods (dual-tropic) or as rod index (* p<0.01, *** p = 0.0002, **** p<0.0001). (B) Time-course of rod formation measured as a percent of neurons with rods upon exposure to 500 pM dual-tropic gp120_MN. Rod induction was significantly above control from as early as 6 h and remained sustained for the duration of the experiment (*p<0.01). (C) A wash-out of gp120_MN after 20 h for a 4 h time period significantly reduced rod formation (**p<0.01 compared to no wash-out for 24 h). It is noteworthy that no additive effect was measured upon incubation of hippocampal neurons with a combination of 500 pM gp120_MN and 50 ng/ml TNFα or 500 pM gp120_MN and 1 nM Aβ_d/t. Note, all manipulations showed significant rod formation compared to control (*p<0.01).

Fig 3. Microglia cells are virtually absent in dissociated cultures of hippocampal neurons.

Microglia and astrocytes in dissociated cultures of mouse hippocampal neurons were characterized by immunoreactivity against Iba-1 or GFAP, respectively. (A) Immunostaining of adult mouse brain slices with Iba-1 antibody revealed the presence of microglia indicating that our fixation, methanol-permeabilization and immunostaining protocol worked well for the Iba-1 antibody. Immunoreactivity is shown in the cytoplasm of a microglial cell co-stained with DAPI (nucleus). A ramified extension of the microglial cell is visible (arrowhead in overlaid image, 100x confocal microscopy). Identical staining was obtained in slices following the more extensive but unnecessary citrate buffer antigen retrieval [35]. (B) Dissociated cultures of mouse hippocampal neurons (DIV 7) were fixed, methanol permeabilized and immunostained (identical conditions as in panel A). Confocal images were acquired (20x) to reveal nuclei (DAPI), astrocytes (GFAP), microglia (Iba-1), and cofilin and an overlay image generated. No Iba-1 staining was observed indicating cultures were devoid of microglia, a finding confirmed by scanning with 60x and 100x objectives as well. GFAP-positive cells (astrocytes) comprised about 40% of total DAPI nuclei within this particular preparation.

Fig 4. CXCR4 receptors predominantly mediate gp120-dependent rod formation.

(A) Mouse hippocampal neurons (C57BI/6) at DIV 7 express both CXCR4 and CCR5 receptors on the surface of neuronal soma and processes as revealed by immunostaining with receptor-specific antibodies in the absence of a permeabilization step. (B) Neuronal cultures were incubated with the CCR5-specific inhibitor maraviroc (100 nM) or the CXCR4-specific inhibitor AMD3100 (50 nM) for 1h prior to and during exposure with 250 pM of either dual- or monotropic gp120. The presence of AMD3100 blocked rod induction by X4-tropic gp120_IIIB and dual tropic gp120_MN but not by R5-tropic gp120_CM. In contrast, the CCR5 antagonist maraviroc reduced rod formation in neurons exposed to R5-tropic gp120_CM but was ineffective in blocking rod formation in response to X4-tropic gp120_IIIB or dual tropic gp120_MN (*p<0.01 unless indicated otherwise). Hippocampal neurons obtained from PrP^C-null mouse line expressed both CXCR4 and CCR5 chemokine receptors indistinguishable from neurons derived from wild type mice (S3 Fig).

Fig 5. Gp120-mediated rod induction requires the expression of the cellular prion protein PrP^C.

(A) Dissociated hippocampal neurons from PrP^C-null mice (6 DIV) were expose to dual-tropic gp120_MN, R5-tropic gp120_BaL, or X4-tropic gp120_IIIB for 16 h (250 pM each) and rod formation was quantified as rod index. The lack of PrP^C abolishes rod induction to levels indistinguishable from spontaneous rod formation. (B) PrP^C-null neurons were infected with adenovirus (50 moi) for expressing EGFP-PrP^C or EGFP (control) for 60 h prior to gp120 exposure for an additional 16 h (250 pM, see strains above) followed by rod quantification (rod index) in EGPF-expressing neurons. Restoring PrP^C expression resulted in a robust gp120-mediated rod formation above control levels for all tropic forms tested (**p = 0.0053, ***p = 0.0003, ****p<0.0001).

Fig 6. Gp120-induced rod formation requires NADPH oxidase (NOX).

NOX activity was inhibited using pharmacological, molecular biological and genetic approaches in dissociated cultures of mouse hippocampal neurons. (A) The NOX inhibitor TG6-277 blocked dual-tropic gp120MN-induced rod formation (16 h exposure, 250 pM) compared to an absence of the pharmacological inhibitor (*p<0.05). TG6-277 alone did not alter basal levels of spontaneous rod formation. (B) Adenoviral-mediated expression of a dominant-negative mutant of the small NOX membrane subunit p22PHOX (DNp22^PHOX) completely abolished rod formation upon exposure to dual-tropic gp120_MN (250 pM) using MOIs (MOI: multiplicity of infection) of 30 to 100. Note, expression of the reporter gene LacZ (control) did not induce rods nor interfere with gp120_MN-mediated rod formation (*p<0.05). (C) Hippocampal neurons lacking the cytosolic subunit p47^PHOX crucial for NOX activity (p47^PHOX-null mouse line) were exposed to dual-tropic gp120_MN, R5-tropic gp120_BaL, or X4-tropic gp120_IIIB (250 pM each, 16 h) and rod induction quantified (rod index). Neither gp120 tropic strain induced rod formation above control levels in p47^PHOX-null neurons.

Fig 7. Inhibition of CCR5 and CXCR4 signaling abolishes Aβ_d/t-mediated rod formation.

Dissociated rat hippocampal neurons (DIV 6) were exposed to 1 nM Aβ_d/t for 16 h followed by immunostaining against cofilin and rod quantification. As expected, the presence of 1 nM Aβ_d/t induced a 4-fold increase in rod-containing neurons compared to control (Con), which was abolished by the presence of either the CXCR4 or CCR5 inhibitors, AMD3100 (60 nM) or maravirac (60 nM), respectively (*p<0.05).

Fig 8. Signaling components in gp120-mediated cofilin-actin rod formation and some putative downstream effectors.

PrP^C is associated with lipid-rafts, plasma membrane domains enriched in cholesterol, gangliosides, sphingolipids and phosphatidyl inositol phosphates (PIP2) all of which properly organize chemokine receptors (CXCR4, CCR5), large and small membrane subunits of NADPH oxidase (NOX and p22^PHOX, respectively), and other raft proteins such as caveolin (Cav), neutral sphingomyelinase (nSMase), and fyn, a src-family kinase. Interaction of gp120 with chemokine receptors and potentially PrP^C stimulates NOX activity to generate ROS through one or more signaling pathways likely involving heterotrimeric G-proteins, Rho family GTPases, caveolin, and fyn. The small GTPase Rac1 is essential for NOX activation whereas Cdc42 is implicated in cofilin activation [66]. The oxidative environment favors activation of the cofilin phosphatase slingshot-1L [67] and dephosphorylation of inactive, phospho-cofilin (pCof) to the active form (Cof). Alterations in phosphoinositides can release PIP2-sequestered phospho and dephospho-cofilin [68]. The increasing pool of active cofilin can locally saturate F-actin, sever filaments, and form cofilin-saturated F-actin fragments now susceptible to ROS-induced formation of intermolecular disulfide-linked cofilin dimers ultimately forming thick bundles of cofilin-actin filaments or rods [20]. Note, lipid rafts (grey lipid bilayer) are thicker in diameter compared to the more fluid plasma membrane (orange bilayer). Dashed lines indicate multi-step pathways whereas solid lines refer to single-step pathways.