HIV-1 counteracts an innate restriction by amyloid precursor protein resulting in neurodegeneration

HIV-1 counteracts an innate restriction by amyloid precursor protein resulting in neurodegeneration ARTICLE Corrected: Author correction DOI: 10.1038/s41467-017-01795-8 OPEN HIV-1 counteracts an innate restriction by amyloid precursor protein resulting in neurodegeneration 1 1 1 2 3 1 Qingqing Chai , Vladimir Jovasevic , Viacheslav Malikov , Yosef Sabo , Scott Morham , Derek Walsh & 1,2 Mojgan H. Naghavi While beta-amyloid (Aβ), a classic hallmark of Alzheimer’s disease (AD) and dementia, has long been known to be elevated in the human immunodeficiency virus type 1 (HIV-1)-infected brain, why and how Aβ is produced, along with its contribution to HIV-associated neuro- cognitive disorder (HAND) remains ill-defined. Here, we reveal that the membrane- associated amyloid precursor protein (APP) is highly expressed in macrophages and microglia, and acts as an innate restriction against HIV-1. APP binds the HIV-1 Gag poly- protein, retains it in lipid rafts and blocks HIV-1 virion production and spread. To escape this restriction, Gag promotes secretase-dependent cleavage of APP, resulting in the over- production of toxic Aβ isoforms. This Gag-mediated Aβ production results in increased degeneration of primary cortical neurons, and can be prevented by γ-secretase inhibitor treatment. Interfering with HIV-1’s evasion of APP-mediated restriction also suppresses HIV-1 spread, offering a potential strategy to both treat infection and prevent HAND. 1 2 Department of Microbiology-Immunology, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USA. Department of Biochemistry and Molecular Biophysics, Howard Hughes Medical Institute, Columbia University, New York, NY 10032, USA. MesaGen, LLC, South Salt Lake City, UT 84115, USA. Correspondence and requests for materials should be addressed to M.H.N. (email: Mojgan.naghavi@northwestern.edu) NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 1 | | | 1234567890 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 n addition to causing acquired immunodeficiency syndrome The gradual accumulation of amyloid plaques is associated (AIDS), HIV-1 crosses the blood–brain barrier (BBB) and with neurodegenerative conditions such as AD in uninfected Ienters the CNS in around 80% of infected individuals leading individuals . Antibodies that target Aβ aggregates have to disorders ranging from mild cognitive impairment to severe strengthened support for amyloid as a causative factor and 1,2 9 HIV-associated dementia (HAD) . While widespread use of therapeutic target in AD . Neurotoxic Aβ is generated by combination antiretroviral therapy (cART) has increased the life sequential site-specific proteolytic cleavage of the ubiquitously span of people living with HIV-1/AIDS, an estimated 50% of HIV expressed type I trans-membrane protein, APP. APP processing is patients on cART exhibit milder forms of HAND . The persis- mediated by four types of secretases (α, β, γ and η) via three tence of HAND is thought to involve poor antiretroviral drug alternative pathways (amyloidogenic, non-amyloidogenic, and η- 8,10 penetration and incomplete viral suppression in the CNS, as well secretase) (Fig. 1a) . Most APP processing is mediated by α- as possible toxic effects of therapy itself . Although HIV-1 does secretase, primarily at the plasma membrane, resulting in release not infect neurons, inside the CNS, it establishes infection in of a large N-terminal soluble fragment (sAPPα) into the extra- perivascular macrophages, microglia, and possibly astrocytes . cellular space and a short C-terminal fragment (α-CTF) into the These infected cells secrete a mix of host and viral proteins that cytoplasm. This process is referred to as the non-amyloidogenic contribute to inflammation and the complex events leading to pathway. Less frequently, in the amyloidogenic pathway, pro- 6,7 HIV-1-induced neuronal damage . However, one poorly cessing of APP by β-secretase generates a soluble ectodomain understood, yet potentially significant host contributor is Aβ. (sAPPβ) and a C-terminal fragment (β-CTF). CTFs can be further a b Antibodies: sAPPα sAPPβ sAPPη sAPPη sAPPα sAPPβ Aβ- APP - 130 Amyloid APP η-CTF Aη-β β-CTF - 55 Aη-α Pr55 Gag Aβ β β p3 α-CTF α α Lumen - 130 APP - 55 Pr55 Gag Cytosol AICD AICD Non-amyloidogenic Amyloidogenic η-secretase pathway pathway pathway Nucleus APP Gag Merge 0.6 *** 0.5 0.4 0.3 0.2 0.1 0.0 - 130 Anti-APP Anti-GAPDH - 34 Fig. 1 APP is highly expressed in macrophages and microglia and binds HIV-1 Gag. a APP processing through amyloidogenic, non-amyloidogenic and η- secretase pathways involves α-, β-, γ- and η-secretases. The Aβ peptide resulting in toxic amyloid oligomers and plaques is generated by sequential cleavages by β- and γ-secretases via the amyloidogenic pathway (central). b Human APP (APP-Flag) binds HIV-1 Gag (Gag-HA) in anti-APP co-IP from transfected 293T cells. c Endogenous APP and Gag colocalize in CHME3 cells infected with HIV-1 carrying vesicular stomatitis virus G (VSV-G) envelope glycoprotein at 16 h post infection (h.p.i). Nuclei were stained with Hoechst (blue). All images were obtained using a 100× oil objective of a spinning disk confocal microscope. Scale bar = 10 μm. d Quantification of APP and Gag as determined by Pearson’s coefficient in at least 10 random fields of view from samples as in c, shown as Mean ± SEM. The Pearson’s correlation coefficients were r = 0.512 ± 0.02. e Endogenous APP levels in glioblastoma (U87), normal human dermal fibroblasts (NHDF), microglia (CHME3), 293T and differentiated THP-1 cells. Molecular weight markers (in kDa) are shown to the right of WBs 2 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | Colocalized Non-colocalized U87 NHDF CHME3 293T THP-1 HIV-1 infected CHME3 Pearson’s coefficient APP-Flag Gag-HA APP-Flag+ Gag-HA IP: Anti-APP Input NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE processed by γ-secretase to create either a non-toxic peptide p3 Here we reveal that APP is highly expressed in macrophages from α-CTF, or Aβ monomers of various lengths from β-CTF, and microglia, natural target cells for HIV-1 infection in the which can self-associate to form toxic Aβ oligomers. γ-secretase brain, and acts as an innate restriction factor that sequesters the cleavage of α-or β-CTFs in the plasma membrane also releases HIV-1 Gag polyprotein in lipid rafts to block virus production fragments of varying sizes from the cytosolic APP intracellular and spread. To evade this restriction, HIV-1 Gag subverts host domain (AICD) into the cytoplasm. Amyloidogenic Aβ peptides secretases to cleave APP and clear membrane-associated CTFs, range from 30 to 42 amino acids (aa) in length, with two main but in doing so also results in increased Aβ production that toxic Aβ species, Aβ40 and Aβ42. Although Aβ40 accounts for causes the degeneration of cultured primary cortical neurons. Our 90% of all Aβ produced, the smaller Aβ42 fraction is more prone findings explain how and why infection leads to Aβ production to aggregation. While Aβ increases in the brain during normal and its contribution to neuronal damage, revealing an antiviral aging, Aβ accumulation is accelerated by HIV-1 infection and activity of APP that could potentially be exploited to treat both correlates with viral loads and the onset of HAND .Aβ also acts HIV-1 infection and neurodegeneration. as a biomarker for HAND, while drugs that inhibit Aβ produc- 11–14 tion may have therapeutic potential . Notably, distinct dif- Results ferences in Aβ deposition patterns between AD and HAND have APP is elevated in macrophages and microglia and binds HIV- been observed, suggesting that HIV-1 specifically alters Aβ 1 Gag. In screens for HIV-1 Gag-interacting host factors, we metabolism and this likely contributes to unique features of HAD identified APP , which was verified by transfecting 293T cells 7,15 and HAND . Indeed, several studies suggest soluble amyloid with plasmids encoding flag-tagged human APP or HA-tagged oligomers represent the primary pathological structure by per- HIV-1 Gag polyprotein alone or in combination. Co- meabilizing cellular membranes, leading to neuronal loss immunoprecipitation (co-IP) assays revealed that HIV-1 Gag observed in AD , and intraneuronal amyloid accumulation is a was only present in immune complexes when APP was present 17,18 predominant feature in HIV-infected brains . Despite this, (Fig. 1b), confirming their interaction. Immunofluorescence (IF) fundamental questions remain about how and why HIV-1 causes analysis of transfected cells also showed that APP co-localized Aβ production, and whether this directly contributes to neuronal with HIV-1 Gag in 293T cells (Supplementary Fig. 1a, b) and damage during infection. CHME3, a human microglia cell line and natural target cell type a c Input Bound APP binding domain Pr55 Gag -55 MA CA NC P6 Gag-HA HA Gag-N-Myr-HA HA Gag-MA-20-HA HA . . . -26 Gag-MA-40-HA p24 CA HA . . . . . . Gag-MA-60-HA HA p17 MA -17 . . . . . . . . . Gag-MA-80-HA HA Pr55 Gag -55 . . . . . . . . . . . . Exogenous APP in 293T Input Bound Anti-HA (Pr55 Gag) -55 -26 p24 CA Anti-APP -130 -17 p17 MA GFP-APP ++ + + + + +++ + + + Gag-HA + – –– –– + ––––– Anti-APP -130 Gag-N-Myr-HA – + –– – – – + –– – – Anti-GFP Gag-MA-20-HA – – + – –– – – + – –– (GFP) -26 Gag-MA-40-HA – – – + – – – – – + – – PARP-1 –95 Gag-MA-60-HA – –– –– + – –– – + – GFP +– –– + –– Gag-MA-80-HA – – – – – + – – – – – + GFP-APP – + ++ – + + Endogenous APP in CHME3 Gag-HA ++ – – + – – Input IP: Anti-APP MA-HA – – + – – + – -55 Anti-HA (Pr55 Gag) CA-HA – –– + – – + -130 Anti-APP Endogenous APP in HIV-1 infected CHME3 4x4 Gag-HA + – –– –– + – –– –– Gag-N-Myr-HA – + –– – – – + –– – – * -55 Input Anti-Pr55 Gag Gag-MA-20-HA – – + – –– – – + – –– -55 Gag-MA-40-HA – – – + – – – – – + – – IP: Anti-APP -130 Gag-MA-60-HA – –– – + – – –– – + – Anti-APP Gag-MA-80-HA – – – – – + – – – – – + Mock HIV-1 Fig. 2 APP binds the MA region of HIV-1 Gag. a GFP-tagged human APP (GFP-APP), but not GFP control, binds HIV-1 Gag (Gag-HA) or Matrix (MA- HA), but not Capsid (CA-HA) in GBP-binding assays. b Endogenous APP interacts with Pr55 Gag in WT HIV-1-infected CHME3 4 × 4 cells in anti-APP co- IP. *indicates unspecific bands detected in cell lysates. c Schematic of the HIV-1 Gag polyprotein used in binding assays, including the c-terminal HA tag. X indicates a point mutation in the N-terminal myristoylation site (Gag-N-Myr-HA). Sequential 20 aa deletions are indicated. d Gag mutants lacking aa 72-111 of MA (Gag-MA-40-HA, Gag-MA-60-HA, or Gag-MA-80-HA) fail to bind GFP-APP in co-transfected 293T cells in GBP-binding assay. e Gag mutants lacking aa 72-111 of MA (Gag-MA-40-HA, Gag-MA-60-HA, or Gag-MA-80-HA) fail to bind endogenous APP in CHME3 cells in anti-APP co-IP. Molecular weight markers (in kDa) are shown to the right of WBs NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 3 | | | Anti-Pr55 Gag Anti-HA ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 for HIV-1 infection in the brain (Supplementary Fig. 1c). Vali- (U87) and microglia (CHME3) also express high levels of APP dating co-transfection approaches, endogenous APP also co- compared with primary normal human dermal fibroblasts localized with Gag in HIV-1-infected CHME3 cells (Fig. 1c, d). (NHDFs) or 293T cells (Fig. 1e). Beyond brain-resident microglia, While APP and Gag were often expressed at high levels and co- human monocyte-derived macrophage cell lines (THP-1) also localized broadly throughout the cell, identifying cells that expressed high levels of APP, which exhibited altered mobility in expressed lower levels of both proteins more clearly illustrated SDS-PAGE, suggestive of an alternative isoform or post- their co-localization at distinct cellular regions, discussed below translational modification (Fig. 1e). This suggested that high (Fig. 1c and Supplementary Fig. 1a). APP is ubiquitously levels of APP expression in macrophages and microglia, natural expressed in many cell types but highly expressed in neurons . target cell types for HIV-1 infection, and its interaction with Gag We found that other human brain cell lines such as glioblastoma could be of particular biological significance during HIV-1 ac 293T (μg DNA) CHME3 4x4 THP-1 pCDNA3.1 1.8 1.4 1.0 0.7 0.4 0 1.0 0.7 0.4 0 130 - -130 APP pGAPDH-HA 0 0 00 0 0 0.4 0.7 1.0 1.4 55 - Pr55 Gag -55 APP-Flag 0 0 0.4 0.7 1.0 1.4 00 0 0 pNL4-3 -26 0 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 26 - eIF4E 130- -34 APP-Flag GAPDH-HA -26 26 - p24 CA Pr55 Gag -55 35,000 ** p24 CA 30,000 -130 25,000 ** 4000 PARP-1 -55 20,000 * β-tubulin 15,000 -26 eIF4E 10,000 p24 CA -26 siRNA: d5 p17 MA -17 Days post- infection (d.p.i): d3 d5 1 2 3 4 5 67 8 9 10 b e p24 ELISA 1,000,000 CHME3 4x4 THP-1 600 60 *** *** ** d.p.i: d3 d.p.i: d5 d.p.i: d5 *** *** 500 500 100,000 400 400 *** 300 300 200 200 10,000 100 100 0 0 APP-Flag siRNA: pGAPDH-HA 0 0.4 0.7 1.1 1.4 APP-Flag or pGAPDH (μg) f HeLa-TZM-bl CHME3 4x4 THP-1 8000 30,000 10,000 *** *** d.p.i: d3 d.p.i: d5 ** d.p.i: d5 25,000 6000 8000 *** ** 20,000 *** 4000 15,000 10,000 0 0 0 siRNA: Fig. 3 APP inhibits production of HIV-1 particles. a Increasing expression of human APP (APP-Flag) reduces the levels of MA/CA-containing HIV-1 particles in culture supernatants compared with human GAPDH (GAPDH-HA) controls. b Infectious virion yields from APP or GAPDH expressing 293T cells measured using TZM-bl indicator cells. c RNAi-mediated depletion of APP relieves an endogenous block to production of extracellular virus particles in CHME3 4 × 4 cells and differentiated THP-1 cells infected with pNL4-3-derived HIV-1 at 3 and/or 5 days post-infection (d.p.i: d3 and d5). d Quantification of p24 CA intensity in supernatants from c. e Measurements of p24 CA levels in supernatants from c by ELISA. f Infectious virus yield in samples from c was measured using TZM-bl indicator cells. The data in b, d, e, f represent average of 3 replicates, and are represented as mean +/− SEM (one-way ANOVA; *P < 0.05, **P < 0.01, ***P < 0.001). Molecular weight markers (in kDa) are shown either to the right or to the left of WBs 4 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | Infectious virion yield (Log RLU) Virions Cell lysate Infectious virion yield p24 (pg/ml) p24 CA intensity Cell lysate Virions Control Control Control APP-V1 APP-V1 APP-V1 APP-V2 APP-V2 APP-V2 Control APP-V1 APP-V2 Control Control p24 CA intensity APP-V1 APP-V1 APP-V2 APP-V2 Control APP-V1 APP-V2 Control Control Virions Cell lysate APP-V1 APP-V1 APP-V2 APP-V2 NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE Total lysate or sup Fractionated lysate GAPDH + WT Gag APP + WT Gag Anti-APP Anti-Pr55 Gag 55 Anti-GAPDH Anti-GAPDH Pr55 Gag Anti-Flotilin-1 55 GAPDH Anti-PARP-1 95 Anti-Caveolin-1 p24 CA Anti-Nicastrin Anti-PEN2 12 3 4 5 12 3 4 5 67 8 910 67 8 910 WT Gag Top Bottom Top Bottom GAPDH + Gag-MA-60-HA (Δ APP binding site) APP + Gag-MA-60-HA (Δ APP binding site) Anti-APP Gag-MA-60-HA Anti-HA 34 GAPDH Anti-Flotilin-1 34 Anti-GAPDH Anti-elF4E Anti-PEN2 p24 CA 123 4 5 678 910 12 3 4 5 67 8 910 Top Bottom Top Bottom Gag-MA-60-HA CHME3 transfected with control siRNA CHME3 transfected with APP-V2 siRNA Anti-APP Anti-Pr55 Gag 55 Anti-Pr55 Gag Anti-Caveolin-1 Anti-eIF4E Anti-Nicastrin Anti-PARP-1 Anti-PEN2 10 12 3 4 5 67 8 910 12 3 4 5 67 8 910 Top Bottom Top Bottom 293T co-transfected with APP + pcDNA3.1 293T co-transfected with APP + WT Gag Anti-APP Anti-Pr55 Gag 55 f 293T (μg DNA) Anti-Flotilin-1 0.3 0.2 0.1 0 0.3 0.2 0.1 0 130 pcDNA3.1 Anti-Nicastrin 0000 0.1 0.2 0.3 0.4 pGAPDH-HA Gag-HA 0.1 0.2 0.3 0.4 00 0 0 Anti-PEN2 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 APP-Flag 123 4 5 6789 10 123 4 56789 10 Anti-Pr55 Gag Anti-HA Top Bottom Top Bottom CHME3 4x4 infected with mock CHME3 4x4 infected with HIV-1 6E10 130 Anti-APP 130 LN27 55 Anti-Pr55 Gag Anti-eIF4E Anti-Nicastrin Anti-PEN2 123 4 56789 10 123 4 56789 10 Top Bottom Top Bottom Fig. 4 APP causes Gag retention in membrane domains. a, b Analysis of cell lysates or supernatant VLPs (left panels) or fractionated lysates from flotation assays (right panels). a Left: In cells expressing WT Gag, APP expression reduces VLP levels in cell supernatants. Right: HIV-1 Gag is present in both membrane-free (8–10) and membrane-bound fractions containing lipid raft and γ-secretase components (2–4) in control samples, but is solely membrane- bound in APP-expressing 293T cells. b Left: In cells expressing mutant Gag that does not bind APP, APP expression does not affect VLP levels in cell supernatants. Right: HIV-1 Gag lacking the APP-binding site is present in membrane-free fractions in both control and APP-expressing 293T cells. c Left: WB verification of APP depletion in cell lysates. Right: RNAi-mediated depletion of APP increases the release of HIV-1 Gag into membrane-free fractions in CHME3 cells transfected with HIV-1 Gag expression plasmid. d, e HIV-1 Gag expression does not release APP from membrane-bound fractions containing lipid raft and γ-secretase components in transfected 293T cells (d) or in CHME3 4 × 4 cells infected with WT HIV-1 (e). f Increasing Gag-HA expression results in decreased APP levels in transfected 293T cells compared with control GAPDH-HA expression. Molecular weight markers (in kDa) are shown either to the right or to the left of WBs NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 5 | | | GAPDH APP GAPDH APP Control APP-V2 Anti-HA VLPs Cell lysate VLPs Cell lysate Anti-HA Anti-APP ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 infection in the brain. Moreover, the levels of APP expression in (Supplementary Fig. 2). Notably, further increasing APP expres- APP-transfected 293T cells resembled those found naturally in sion to very high levels also affected intracellular Pr55 Gag macrophages and microglia (Figs. 1b, e and 2), demonstrating abundance (Supplementary Fig. 2). These findings suggested that that transfected 293T cells offered a tractable system to under- APP primarily affected HIV-1 or VLP production and/or release, stand APP function at physiologically relevant levels. but at very high levels APP exerted secondary effects on intra- To independently verify this interaction and identify the APP- cellular Gag accumulation. Validating findings in transfected interacting domain in Gag, 293T cells were transfected with 293T cells, RNAi-mediated depletion of APP in CHME3 4 × 4 plasmids encoding GFP-tagged APP together with HA-tagged cells resulted in an increase in extracellular mature virus particles Pr55 Gag polyprotein, p24 capsid (CA) or p17 matrix (MA). in supernatants from cells infected with WT (pNL4-3-derived) GFP-APP was recovered from soluble cell extracts on GFP- HIV-1, as detected by either p24 CA WB or ELISA analysis binding protein (GBP)-conjugated sepharose . APP was found (Fig. 3c–e, and Supplementary Fig. 3a, b). Although RNAi- to specifically interact with Gag or the MA portion of Gag, but mediated depletion was less efficient, reducing APP expression in not CA (Fig. 2a). To confirm these results using endogenous APP THP-1 differentiated to macrophages also significantly increased in the context of infection in natural target cell types, CHME3 the levels of mature virions in culture supernatants (Fig. 3c–e). 4 × 4 cells, which express higher levels of CD4 and CXCR4 for Applying supernatants to TZM-bl indicator cells confirmed that more efficient infection with WT HIV-1 envelope , were infected this corresponded to an increase in production of extracellular with HIV-1 followed by anti-APP co-IP. In line with findings in infectious virus particles from APP-depleted microglia or mac- co-transfected 293T cells, endogenous APP also co- rophages infected with WT (pNL4-3-derived) HIV-1 (Fig. 3f). immunoprecipitated with Pr55 Gag in HIV-1-infected cells siRNA-mediated silencing of APP did not affect cell viability as (Fig. 2b and below). To further define the region of MA involved, detected by the lack of PARP-1 cleavage in these cells (Supple- Gag expression plasmids with serial truncations or mutations in mentary Fig. 3c). As such, APP overexpression in 293T cells or MA (Fig. 2c) were tested for binding to either exogenous GFP- naturally high levels of APP in natural target cells such as APP in transfected 293T cells or endogenous APP in CHME3, microglia or monocyte-derived macrophages suppressed the using two independent approaches. In 293T cells co-transfected production of extracellular HIV-1 particles. with different mutants of Gag-HA together with GFP-APP, GBP- binding assays revealed that mutations in the N-terminal myristoylation site (Gag-N-Myr-HA) or deletion of the last 20 APP retains Gag in lipid rafts. We next determined whether aa of the C-terminus of MA had no effect on APP binding APP influenced Gag localization using membrane flotation assays (Fig. 2d). However, larger deletions of 40, 60, or 80 aa in the MA where-in cell lysates are fractionated to separate membrane- C-terminus impaired Gag binding to APP. Validating these bound and membrane-free fractions . In control cells co- findings using endogenous APP, CHME3 cells were transfected transfected with GAPDH, Gag was present in both membrane- with the same HA-tagged forms of wild-type (WT) or mutated bound (2–4) and membrane-free (8–10) fractions (Fig. 4a and Gag followed by anti-APP co-IP. Similar to observations in co- Supplementary Fig. 4a), which is in line with previous reports . transfected 293T cells, APP again failed to interact with Gag-MA- By contrast, all detectable Gag was found exclusively in 40-HA, Gag-MA-60-HA or Gag-MA-80-HA mutants, but membrane-bound fractions in 293T cells overexpressing APP, efficiently bound myristoylation or Gag-MA-20-HA mutants and again decreased the levels of p24 CA in culture supernatants (Fig. 2e). This demonstrated that residues 72-111 of MA were (Fig. 4a and Supplementary Fig. 4a). In line with its general localization to diverse cellular compartments, GAPDH in either required for Gag binding to either transfected, tagged forms of APP in 293T cells, or endogenous APP in microglia. GAPDH- or APP-overexpressing cells was broadly distributed across both membrane and free cytosolic fractions. HIV-1 Gag/ MA has been reported to associate with membrane lipid rafts 28,29 APP inhibits HIV-1 virion production and spread. APP is during virion assembly and release , and membrane- membrane-associated , while the MA domain mediates plasma- associated fractions containing Gag were found to contain lipid membrane association of the Gag precursor protein during raft markers, Flotillin-1 and Caveolin-1 (Fig. 4a). This was 23–25 assembly and budding of new HIV-1 particles . To test observed in both APP- and GAPDH-expressing cells, demon- whether APP could affect virus production, 293T cells were strating that APP overexpression did not disrupt lipid raft com- transfected with an infectious cDNA clone of HIV-1 (pNL4-3) position, but retained Gag at these sites. Notably, lipid rafts are 30–33 together with increasing amounts of either APP or GAPDH also enriched in β- and γ-secretases that process APP , and control plasmids. Western blot (WB) analysis confirmed membrane-associated fractions containing Gag were also speci- increasing expression of APP or GAPDH in each case, while the fically enriched for the γ-secretase components, Nicastrin and levels of housekeeping proteins (β-tubulin and eIF4E) or PARP-1, Presenilin Enhancer 2 (PEN2) (Fig. 4a). To determine if binding an apoptosis indicator, confirmed no adverse effects on cell via- to APP mediated Gag sequestration at these sites and production bility (Fig. 3a). While intracellular Pr55 Gag or p24 CA levels of extracellular virus particles, we compared the effects of over- were moderately elevated with increasing APP expression expressing APP versus GAPDH on the distribution of the APP- (Fig. 3a), supernatant levels of p24 CA and p17 MA components binding mutant, Gag-MA-60-HA. In contrast to WT Gag, the of mature particles revealed a highly potent, dose-dependent Gag-MA-60-HA mutant exhibited a similar distribution to both reduction in APP-expressing cells (Fig. 3a). Applying these membrane-bound and membrane-free fractions in both GAPDH supernatants to TZM-bl indicator cells confirmed that APP control and APP-overexpressing cells (Fig. 4b and Supplementary expression blocked production of extracellular infectious virus Fig. 4b). Moreover, while APP expression resulted in a potent particles (Fig. 3b). Whether this reduction in extracellular virus in decrease in VLPs in supernatants from cells expressing WT Gag APP-expressing cells reflects decreased maturation, assembly or (Fig. 4a, left panels), VLP production and release from cells budding remains to be determined. To confirm that these effects expressing Gag-MA-60-HA was unaffected by APP (Fig. 4b, left on infectious HIV-1 replication reflected effects of APP on the panels). These findings demonstrated that APP binding did Gag polyprotein, 293T cells were transfected with HIV-1 Gag, indeed influence Gag membrane localization and the production which produces and releases virus-like particles (VLPs). Expres- of extracellular virions. In line with findings in 293T over- sion of APP potently blocked production of extracellular VLPs expression systems, the converse approach of RNAi-mediated 6 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | d.p.i: d3 NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE a b c d e APP-FL 130 130 Diffrentiated Diffrentiated THP-1 (HIV-1-VSV-G) Anti-APP lighter CHME3 4x4 (WT HIV-1) THP-1 (WT-HIV-1) APP-FL 55 5 130 Anti-Aβ 42 Pr55 Gag APP-FL Anti- Aβ 42 Anti-Aβ 40 α-CTF 34 1.4 GAPDH 1.8 ** 1.2 β-CTF 1.5 1.0 0.8 α-CTF Anti-GAPDH 1.2 0.6 0.9 55 0.4 Pr55 Gag 95 β-CTF 0.6 Anti-PARP-1 10 0.2 α-CTF 26 0.3 0.0 Anti-eIF4E Pr55 Gag 0.0 Mock HIV-1 Anti-Aβ 42 d.p.i: d5 p24 CA 5 GAPDH Anti-Aβ 40 95 Anti-PARP-1 d.p.i: d3 d5 N.I. Anti-eIF4E γ-secretase d.p.t: d3 d5 inhibitor Anti- 5 - Aβ 42 f gi CHME3 4x4 (WT HIV-1) HeLa-TZM-bl CHME3 4x4 (WT HIV-1) DMSO γ-secretase Nicastrin 130 APP-FL inhibitor APP-FL α-CTF 10 CHME3 4x4 (WT HIV-1) Pr55 Gag APP eIF4E 0 Pr55 Gag β-CTF α-CTF p24 CA GAPDH 55 34 d.p.i: d5 siRNA: 4000 Pr55 Gag p24 CA 26 PARP-1 2000 p24 ELISA p24 CA *** *** *** 500 *** 200 400 *** 1500 *** d.p.i: d7 *** *** *** siRNA: 400 ** 150 300 100 200 DMSO γ-secretase 50 100 inhibitor 0 0 N.I. N.I. N.I. γ-secretase γ-secretase γ-secretase γ-secretase inhibitor inhibitor inhibitor inhibitor d.p.i: d3 d5 d7 Fig. 5 HIV-1 Gag promotes APP processing into Aβ isoforms. a Decreased APP levels in Gag-expressing 293T cells correlates with increased secretion of Aβ40 and Aβ42 compared to GAPDH (Ctrl) at days 3 (d3) and 5 (d5) post-transfection (d.p.t). b Gag-induced processing of APP and Aβ42 accumulation in transfected 293T cells is blocked by γ-secretase inhibitor treatment. Full-length APP: APP-FL, C-terminal fragments: α-CTF and β-CTF. c, d Aβ40 and/or Aβ42 secretion from CHME3 4 × 4 cells c or differentiated THP-1 cells d is enhanced upon infection with WT HIV-1 (pNL4-3 derived), detected by either WB analysis (top panels) or Congo Red staining of amyloid (lower graphs). e Infection of differentiated THP-1 cells with HIV-1 pseudotyped with VSV-G envelope to increase infection efficiency results in production of p24 CA and a reduction in cellular APP levels, and both process are inhibited by γ- secretase inhibitor. f Depletion of Nicastrin suppresses APP processing, as seen by the accumulation of APP and CTFs, and reduces the levels of WT HIV-1 particles in supernatants of infected CHME3 4 × 4 cultures. g, h γ-secretase inhibitor treatment blocks APP processing, as seen by the accumulation of α- CTF and β-CTF, and reduces the levels of WT HIV-1 particles in supernatants of infected CHME3 4 × 4 cultures at d3-d7 d.p.i., detected either by WB using anti-p24 CA antibody (g) or ELISA (h). i As in g, h, except that infectious virus yield was measured using TZM-bl indicator cells. j siRNA-mediated depletion of APP in CHME3 4 × 4 renders infection with WT HIV-1 insensitive to γ-secretase inhibitors as determined by measurements of p24 CA levels in culture supernatants by either WB analysis or ELISA. The data in c, d (lower panels) as well as h–j (lower panel) represent average of 3 replicates, and are represented as mean +/− SEM (one-way ANOVA; *P < 0.05, **P < 0.01, ***P < 0.001). Molecular weight markers (in kDa) are shown either to the right or to the left of WBs depletion of APP in CHME3 cells followed by transfection with with WT HIV-1 (Fig. 4e and Supplementary Fig. 4e). Although HIV-1 Gag increased the proportion of Gag present in APP distribution was unaffected by Gag expression, the levels of membrane-free fractions (Fig. 4c and Supplementary Fig. 4c). APP expression were notably lower in fractions from Gag- This demonstrated that the integral membrane protein, APP expressing cells compared with controls (Fig. 4d). Indeed, APP binds and retains Gag in membrane regions rich in lipid raft overexpression was less efficient in the presence of Gag in earlier markers, which would explain the reduction in extracellular experiments (Supplementary Fig. 2), suggesting an antagonistic infectious virions and VLPs in cells expressing high levels of APP relationship between Gag and APP. To test this further, we (Figs. 3, 4a, b, and Supplementary Fig. 2, left panels). transfected 293T cells with a constant amount of APP together We next addressed the reciprocal question of whether Gag with increasing amounts of either Gag-HA or GAPDH-HA influenced APP membrane localization by co-transfecting cells control. WB analysis of cell lysates revealed that increasing levels with APP and either HIV-1 Gag or empty vector control. In of Gag resulted in a dose-dependent reduction in APP expression control samples, APP localized to the same membrane fractions in cells (Fig. 4f), possibly by enhancing APP’s turnover in lipid as Gag, which again contained lipid raft and γ-secretase rafts to circumvent APP-mediated restriction. components (Fig. 4d and Supplementary Fig. 4d). Validating findings in transfected 293T cells, endogenous APP also localized to membrane-associated fractions containing Gag as well as lipid HIV-1 Gag promotes processing of APP into neurotoxic Aβ raft and γ-secretase components in CHME3 4 × 4 cells infected isoforms. Given that HIV-1 Gag decreases APP levels both inside NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 7 | | | Ctrl Gag Ctrl Gag Control Ctrl Nicastrin A Gag Nicastrin B Ctrl Gag DMSO 1 μM 2.5 μM DMSO 1 μM 2.5 μM Anti-CTF (Y188) Anti-HA Virions Cell lysate Supernatant Cell lysate Cell lysate Sup Cell lysate p24 (pg/ml) Virions Anti-HA Anti-CTF (Y188) Congo DMSO 490 650 Red A –A 1 μM 2.5 μM Mock HIV-1 DMSO Mock 1 μM HIV-1 2.5 μM Congo DMSO 490 650 Red A –A 1 μM 2.5 μM Infectious Infectious Infectious virion yield (RLU) virion yield (RLU) virion yield (RLU) Cell lysate Virions p24 (pg/ml) Control Control Darker APP-V1 exposure APP-V2 Virious Cell lysate ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 Supernatants Gag transfected 293T HIV-1 infected CHME3 4x4 from: a d fj HIV-1 infected PBMCs 130 130 Immunodepletion 120 120 ## 130 110 110 Beads-bound APP 100 100 * 55 Pr55 Gag 90 90 5 Supernatants 80 80 GAPDH 34 70 70 Anti-Aβ Anti-HA 60 60 eIF4E (6E10) 50 50 30 p24 CA ** 20 1.0 10 60,000 0 0.8 50,000 0.6 40,000 b e ** 0.4 30,000 **** 0.2 160 20,000 160 ‡‡ 0.0 10,000 140 ## ## 0 siRNA: *** ** 0 20 60 2.50 Ctrl Gag Ctrl Gag 40 2.25 *** Mock HIV-1 ** DMSO γ-secretase 2.00 inhibitor 1.75 1.50 Anti-Aβ Anti-HA (6E10) 1.25 1.2 1.00 0.75 1 0.35 0.50 0.3 0.25 0.8 0.00 0.25 0.2 0.6 l 0.18 *** 0.15 Y = –0.2465X + 0.4304 0.16 0.1 R = 0.18557 0.4 0.14 y = –0.0012x + 0.9375 Ctrl + DMSO N = 20 0.05 0.12 R = 0.37785 Gag + DMSO P < 0.00001 0.2 N = 44 Ctrl + inhibitor 0.10 0 0.2 0.4 0.6 0.8 1 1.2 P < 0.00001 Gag + inhibitor 0.08 490 650 Congo Red A –A 0.06 0 50 100 150 200 250 300 (% of control) Aβ42 (pg/ml) 0.04 0.02 0.00 siRNA: Fig. 6 HIV-1 Gag-induced Aβ42 production causes neurodegeneration. a Cortical neurons treated with supernatants from 293T cells co-transfected with # ## APP-Flag along with either pGAPDH-HA (Ctrl) or HIV-1 Gag (Gag) reveals neurotoxicity caused by Gag expression. *P < 0.05 vs. Ctrl; P < 0.05, P < ## 0.01 vs. gag (one-way ANOVA; n = 11; F = 5.665; P < 0.05). b ELISA analysis of Aβ42 levels in denatured supernatants from a. **P < 0.01 vs. Ctrl, P < 40,3 ‡‡ 0.01 vs. gag, P < 0.01 vs. Ctrl + inhibitor (one-way ANOVA; n = 11; F = 53.340; P < 0.001). Bars represent SEM. c Regression plot showing the 40,3 correlation between cell viability and Aβ42 levels from a-b (F = 25.508; P < 0.00001; R = 0.378). d Cortical neurons treated with supernatants from 42,1 mock or WT HIV-1-infected CHME3 4 × 4 cells reveals neurotoxicity caused by HIV-1. *P < 0.05 vs. mock; (n = 11; t = 2.49; P < 0.05). e Congo Red staining of amyloid levels in supernatants from d. ****P < 0.0001 vs. mock (n = 11; t = -9.37; P < 0.0001). Bars represent SEM. f Bead-bound and supernatant levels of Aβ in 6E10 or anti-HA antibody-treated samples from WT HIV-1-infected CHME3 4 × 4. g Congo Red staining of amyloid levels in supernatants from f.(n = 10, **P < 0.01). h Cortical neurons treated with supernatants from f.(n = 10, ***P < 0.001). i Regression plot showing the correlation between cell viability and Aβ levels from samples in f–h (F = 3.874; P < 0.00001; R = 0.186). j–l APP depletion in PBMCs infected with WT 17,1 HIV-1 causes an increase in p24 CA j, a decrease in Aβ levels determined by Congo Red staining (k) and a corresponding increase in viability of cortical neurons treated with supernatants from j, l. The data in j (lower panel), k, l represent mean +/− SEM. k Congo Red staining of amyloid levels in supernatants from j.**P < 0.01 vs. Ctrl, ***P < 0.001 vs. Ctrl, (one-way ANOVA; n = 10, P < 0.001). l Cortical neurons treated with clarified supernatants from PBMCs treated as in j, k reveals APP depletion reduces neurotoxicity caused by infection and its inverse correlation with amyloid levels. ***P < 0.001 vs. Ctrl (one-way ANOVA; n = 10, P < 0.001). Molecular weight markers (in kDa) are shown to the right of WBs the cell and in γ-secretase enriched lipid rafts, we tested whether controls (Fig. 5a). This decrease was accompanied by a corre- Gag stimulated APP processing into Aβ isoforms. Co-transfection sponding increase in the levels of secreted Aβ42, and to a lesser of 293T cells with vectors expressing APP along with either HIV- extent Aβ40 in the supernatants of Gag-expressing cells (Fig. 5a). 1 Gag or GAPDH again revealed a large decrease in the intra- This suggested that HIV-1 Gag enhanced APP processing, most cellular levels of APP in Gag-expressing cells compared to likely through secretases. In agreement with this, the decrease in 8 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | Control APP-V1 APP-V2 Control APP-V1 APP-V2 Aβ42 (pg/ml) Viability (% of control) Congo Red 490 650 A –A (% of control) Viability (% of control) Viability Congo Red 490 650 (% of Aβ depletion) A –A (% of 6E10) Congo Red P24 CA intensity 450 490 650 Viability (A ) A –A (% of control) Virions Cell lysate NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE APP levels induced by Gag expression, and accompanying neurons, those taken from Gag-expressing cells resulted in a increase in Aβ42 secretion, could be blocked by treating trans- statistically significant decrease in cell viability compared to fected cells with γ-secretase inhibitor (Fig. 5b). In line with control samples (Fig. 6a). ELISA analysis of supernatants showed blocking γ-secretase activity, inhibitor-treated cells were also that this effect paralleled increased Aβ42 levels in Gag-expressing found to contain higher levels of α- and β-CTF processing cell supernatants (Fig. 6b), similar to Aβ42 WB analysis (Fig. 5b). intermediates (Fig. 5b). Levels of Gag as well as host eIF4E or Notably, ELISA was performed using antibody suitable for PARP-1 were unaffected, demonstrating that inhibitor treatment detection of monomeric Aβ42 and failed to detect Aβ42 in did not block APP processing by indirectly affecting Gag supernatants until samples were extensively denatured, demon- expression or cell viability, establishing that Gag enhanced γ- strating that the Aβ42 being produced and detected, represents secretase-dependent processing of APP. In line with observations soluble amyloid oligomers. In control samples treated with γ- in transfected 293T cells, infection of CHME3 4 × 4 cells or dif- secretase inhibitor, reducing Aβ42 levels to below normal levels ferentiated THP-1 cells with WT HIV-1 resulted in elevated Aβ40 observed in DMSO-treated cells had no significant impact on and Aβ42 in culture supernatants, as detected by either WB neuronal viability. By contrast, treatment of Gag-expressing cells analysis or staining with the amyloid-binding detection reagent, with γ-secretase inhibitor blocked Gag-induced increases in Aβ42 Congo Red (Fig. 5c, d, respectively). production and prevented Gag-induced neurotoxicity (Fig. 6a, b). Exploring this further, and validating the biological relevance of Indeed, linear regression analysis showed significant correlation secretase-mediated APP processing in Gag-transfected 293T cells, between the neurotoxic effects of clarified supernatants and the infection of differentiated THP-1 cells with HIV-1 pseudotyped with levels of Aβ42 present in supernatants under each condition VSV-G envelope, to attain high multiplicity of infection (m.o.i), also (Fig. 6c). To validate this Gag-mediated neurodegeneration in the resulted in decreased abundance of APP and CTFs that could be context of WT infection, the effects of clarified supernatants from blocked by inhibiting γ-secretase (Fig. 5e). In addition, blocking this mock- or WT HIV-1-infected CHME3 4 × 4 cells were also decrease in APP using the γ-secretase inhibitor resulted in a assessed. Similar to Gag-transfected 293T cells, supernatants from corresponding decrease in mature virions in culture supernatants, as HIV-1-infected cells contained statistically significant increases in determined by WB analysis of p24 CA (Fig. 5e). Independently Aβ42 levels and resulted in a corresponding increase in validating observations using the γ-secretase inhibitor in THP-1 neurodegeneration compared to mock-infected control samples cells, RNAi-mediated depletion of the γ-secretase subunit, Nicastrin (Fig. 6d, e). To confirm that Aβ did indeed cause neurotoxic in CHME3 4 × 4 cells infected with WT HIV-1 resulted in increased effects of HIV-1-infected culture supernatants, Aβ was immuno- APPand CTFlevels, andacorrespondingdecreaseinp24 CA in depleted using the antibody, 6E10 prior to clarification of culture supernatants (Fig. 5f). This suggested that protecting APP supernatants from WT HIV-1-infected CHME3 4 × 4 followed and CTFs from processing suppressed HIV-1 infection. In line with by assessment of effects on neuronal viability. Successful Aβ this, a β-secretase inhibitor also increased APP and CTF expression depletion from clarified supernatants was confirmed by WB and suppressed p24 CA levels in supernatants of CHME3 4 × 4 analysis, which showed a large increase in bead-bound Aβ over infected with WT HIV-1 (Supplementary Fig. 5a, b). Testing background levels in control antibody-treated samples, and a potential effects on HIV-1 spread in natural target cells, treatment of corresponding reduction in supernatant levels of Aβ (Fig. 6f). WB CHME3 4 × 4 with γ-secretase inhibitor to prevent APP processing analysis of immune-depleted samples was further confirmed by suppressed the production of extracellular virus particles, as Congo Red staining of Aβ (Fig. 6g). When clarified, immuno- determined by p24 CA levels detected by either WB analysis or depleted supernatants from HIV-1-infected cells were applied to ELISA (Fig. 5g, h, respectively), as well as infectious HIV-1 virions cortical neurons, Aβ depletion resulted in a significant increase in (Fig. 5i) between 3 and 7 d.p.i. It must be noted that HIV-1-induced neuronal viability compared with controls (Fig. 6h, i). changes in APP are less obvious in low m.o.i. spreading assays due Finally, although challenging to work with we addressed the to a mix of uninfected and infected cells, but efficacy of the inhibitor question of whether APP was functionally important in primary can be seen in the accumulation of α-and β-CTFs, as well as the human natural target cells. We were unable to test whether resulting decreases in p24 CA and infectious virus in culture primary human microglia express APP due to the enormous supernatants (Fig. 5g). Demonstrating that the antiviral activity of ethical and technical challenges involved in obtaining, let alone the γ-secretase inhibitor was mediated by APP, depletion of APP in working with such cells. It is also worth noting that the few CHME3 4 × 4 rescued HIV-1 spread and levels of p24 CA in culture studies that do use primary human brain cells acquire material supernatants in γ-secretase inhibitor-treated cultures, as determined during biopsy or autopsy, meaning cells are isolated under by either WB analysis or ELISA (Fig. 5j). This demonstrated that pathological or trauma/hypoxic conditions that would likely alter APP mediated the effects of secretase inhibitors on infection, and microglia biology in unpredictable ways. However, primary that APP’s antiviral activity could be harnessed using secretase rodent microglia do express APP , and we were able to confirm inhibitors to suppress HIV-1 replication. In addition, these findings APP expression and functionality in primary human peripheral demonstrated that HIV-1 Gag was sufficient to stimulate γ- blood mononuclear cells (PBMCs), the blood-stream counter- secretase-mediated APP processing, as a means to escape this parts of microglia that notably also enter the brain during HIV-1 35,36 restriction, offering a mechanistic explanation for elevated Aβ levels infection . Validating our overall findings, RNAi-mediated released from HIV-infected cells. depletion of APP in PBMCs infected with WT HIV-1 resulted in Although Aβ is elevated in HIV-1-infected patients, whether a statistically significant increase in the production of extra- Aβ production by infected cells in the brain contributes to cellular HIV-1 particles (Fig. 6j). Moreover, APP depletion also neurodegeneration remains unclear . To determine whether Gag- resulted in a decrease in Aβ production in infected PBMCs as induced Aβ production caused neurodegeneration, we developed determined by Congo Red staining (Fig. 6k), and resulted in a an assay using primary mouse cortical neurons. 293T cells were corresponding increase in neuronal viability compared to control transfected with APP together with either HIV-1 Gag or GAPDH siRNA-treated samples (Fig. 6l). in the presence of DMSO solvent control or γ-secretase inhibitor. Supernatants were then collected and clarified through 10 and 3 kDa cutoff filters to remove many other proteins that might also Discussion contribute to neuronal toxicity and confound data interpretation. Although APP has been implicated in various neuronal and When clarified supernatants were added to cultured cortical synaptic processes, its primary function remains unknown . APP NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 9 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 dimerization domains have been shown to interact with various affect Aβ clearance and uptake . While this may contribute to Aβ proteins, including binding to and activation of death receptor 6, overproduction, Tat is secretory and toxic, and proposed effects 5,7 resulting in axonal pruning and inhibition of synapse of Tat on Aβ remain unclear . It is also unclear why HIV-1 37–39 formation . Here, we identify a function for APP as an innate would evolve such a function and how this might benefit viral antiviral defense factor in macrophages and microglia that fitness, or whether it is simply a detrimental side effect. Here, we restricts HIV-1 release. By also identifying a viral evasion reveal that APP has a biological function to restrict HIV-1 release mechanism that leads to the production of neurotoxic Aβ42, our from brain-resident microglia and macrophages that carry HIV-1 findings address fundamental questions about how and why Aβ across the BBB, and Gag-mediated evasion of this restriction levels are elevated in the brains of HIV-1-infected patients, and results in Aβ production. Interestingly, primary rat microglia whether this contributes to neuronal damage. express APP but limit its processing . Gag-induced processing of Mammals have evolved a number of strategies to protect APP in HIV-1-infected microglia may contribute to the under- against retroviral infection, including the expression of antiviral lying differences in Aβ deposition patterns that have been proteins or restriction factors . These include Tetherin, which reported between AD and HAND, which have led to suggestions 41,42 prevents budding of new viral particles . Our findings reveal that HIV-1 alters Aβ metabolism in a manner that contributes 7,15 that APP also targets a late stage of the viral lifecycle required for to unique features of HAD and HAND . While this remains the production of extracellular virions, either through effects on to be explored further, our findings reveal the biological reason assembly, maturation, or release of new virus particles. APP acts for why HIV-1 causes Aβ overproduction, and provides direct by trapping the Gag polyprotein that is processed at the plasma evidence that elevated Aβ caused by Gag expression or HIV-1 23,25 membrane during viral maturation and release within lipid infection contributes to neurodegeneration. Although these rafts. Indeed, these membrane domains have been previously effects might appear small in the conventional sense of cyto- suggested to function in Gag maturation and virion toxicity, neurodegeneration caused by Aβ is a slow, gradual 29,43,44 budding , and have independently been shown to serve as process and the extent of effects from Gag-induced Aβ are in 45,46 51 regions of APP sorting and processing . APP is trafficked line with studies of amyloid in cultured neurons . Moreover, through ESCRT pathways via TSG101, directing APP to lyso- combined with inflammatory responses to infection and cyto- 47–49 somes and for processing into Aβ isoforms for secretion . toxicity of proteins like Tat, Gag-mediated APP processing Intriguingly, TSG101 is also a regulator of HIV-1 budding .As and production of Aβ would be an important contributing factor such, HIV-1 Gag and APP appear to have evolved to utilize the to the overall process of HIV-1-induced neurodegeneration. same subcellular compartments during their maturation. This is Intriguingly, a γ-secretase inhibitor not only prevented the clearly detrimental to HIV-1 production and release, and to evade increase in Aβ42 production induced by Gag and protected APP-mediated restriction, HIV-1 enhances APP processing. An against Gag-induced neurotoxicity, it also protected APP and early report suggested that HIV-1 protease activity leads to CTFs from degradation and suppressed HIV-1 replication in cleavage of APP , but this was not subsequently followed up. We microglia. There are a number of secretase inhibitors being find that APP processing induced by Gag does not require HIV-1 developed as potential therapeutics in AD, and our data suggest protease activity, which is encoded within the pol gene of the Gag/ that these may have the potential to serve dual purposes of both Pol fusion protein, as expression of Gag alone was sufficient to suppressing HIV-1 replication and preventing neuronal damage reduce APP expression (Figs. 4f and 5a, b). Instead, Gag induces by interfering with HIV-1’s attempts to evade this restriction APP processing through host secretases. While the molecular imposed by APP in brain-resident target cells. basis by which this is achieved remains to be determined, Gag might directly interact with secretases or act as a chaperone to Methods enhance APP processing. Alternatively, Gag might induce post- Cells. 293T, U87, NHDF, CHME3, and CHME3 4 × 4 cells were described pre- 21,52 viously . Peripheral blood mononuclear cells (PBMCs) were isolated from a translational modifications of APP. Indeed, around 10% of APP LifeSource Buffy Coat blood using Ficoll-Plaque (GE), and monocytes were isolated undergoes palmitoylation, which has been suggested to enhance from PBMCs using CD14 Microbeads (Miltenyi Biotec). Ethical approval for the amyloidogenic processing by targeting APP to membrane lipid study was obtained from the Institutional Review Board of Northwestern Uni- 28,46 rafts, promoting its β-secretase-mediated cleavage . While versity and all donors provided their written, informed consent. THP-1 cells were kindly provided by Thomas Hope. Primary mouse cortical neurons were purchased these aspects of the underlying mechanism remain to be deter- from Gibco (Cat # A15586). HeLa TZM-bl cells expressing CD4 and CCR5 as well mined, secretases clearly play a central role in this evasion as a LacZ and a luciferase reporter gene under control of the HIV-1 LTR (AIDS mechanism as inhibiting either β-or γ-secretase prevented fur- 26 Reagent Repository number 8129) were maintained in DMEM containing 10% ther processing of CTFs into Aβ isoforms. These protected CTFs, fetal bovine serum (FBS) and 1% Pen/Strep. and full-length APP, contain trans-membrane and cytosolic regions that most likely interfere with aspects of Gag processing Viruses and drugs. WT HIV-1 was generated by transfection of 293T cells with or assembly in lipid rafts. infectious clone pNL4-3 (AIDS Reagent Repository number 114). To generate HIV-1 carrying VSV-G envelope glycoprotein, pNL4-3.Luc.R−.E− plasmid (AIDS While APP likely plays a role in limiting virus spread by Reagent Repository number 3418) was transfected into 293T cells together with a macrophages in the blood, this process and the viral evasion VSV-G-expressing construct (pVSV-G) as described . γ-secretase inhibitor L- strategy is likely to be of particular importance in the brain. 685,458 (Cat # 2627) was purchased from Tocris. β-secretase (BACE1) inhibitor Although HIV-1 does not infect neurons, which lack appropriate Verubecestat (MK-8931, Cat # S8173) was purchased from Selleckchem. CHME3 4 × 4 or 293T cells were treated with DMSO or γ-secretase inhibitor or BACE1 receptors for virus entry, expression of APP in brain-resident inhibitor reconstituted in DMSO at 1 μM and/or 2.5 μM 4 h or 6 h post trans- microglia that become infected and in macrophages that carry fection or infection, respectively, and maintained throughout the entire experiment. 35,36 HIV-1 across the BBB would serve as a critical restriction to protect against HIV-1 spread in the brain. Thus, exploitation of Generation of expression constructs and viral vectors. For generation of host secretases not only provides a mechanistic basis for viral N-terminally Flag-tagged APP (pCAGOSF-APP ), APP (NM_201413; 770 770 evasion of this restriction but also explains increased Aβ pro- OriGene) was amplified from human cDNA using the sense 5′-CCCGGGAT GCTGCCCGGTTTGGCACTGC-3′ and the antisense 5′-GTCGACCTAGTT duction in infected cells. Although several HIV-1 proteins CTGCATCTGCTCAAAG-3′ primers. The restriction enzyme sites are shown in broadly cause inflammation and cytotoxicity, the onset of HAND bold. The PCR product was digested using SmaI and SalI and ligated into the has been shown to correlate with HIV-1-induced accumulation of pCAGOSF plasmid (DNASU), which was digested with SmaI and XhoI and host Aβ, which is also associated with AD. In attempts to explain contained an N-terminus One STrEP FLAG tag. For generation of the N-terminally this, studies have suggested that HIV-1 Tat may bind APP or GFP-tagged APP (N’-GFP-APP ), the APP cDNA from above was amplified 770 770 10 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE using the sense 5′-GCTAGCATGCTGCCCGGTTTGGCACTGCTCCTGCTG using 15 μl Turbofect (Thermo scientific). Total DNA amount in each group was GCC-3′ and the antisense 5′-GTCGACGTTCTGCATCTGCTCAAAGAACT kept constant with addition of empty vector pcDNA3.1 . Soluble cell extracts were TGTAGG-3′ primers. The restriction enzyme sites are shown in bold. The PCR prepared 48 h post-transfection as described and precleared with protein G- product was digested and cloned into the pEGFP-N1 expression vector (Clontech) sepharose. A concentration of 27 μl of the input samples was taken and the using NheI and SalI restriction enzyme sites. For generation of C-terminally remainder of the cell extract was incubated with 2 μl of the mouse anti-APP hemagglutinin (HA)-tagged GAPDH expression construct (pGAPDH-HA), antibody (Invitrogen, 130200) and protein G-sepharose for 1 h at 4 °C. Immune human GAPDH was amplified using cDNA from primary fibrobalsts and following complexes were then washed and boiled with Laemmli buffer and subjected to WB primers; forward primer, hGAPDH-S, 5′-GCAACTGCGGCCGCCATGGGGA analysis. For co-IP of endogenous APP with Gag, 1 × 10 CHME3 4 × 4 cells were AGGTGAAGGTCGGA-3′ and reverse primer, hGAPDH-A 5′-GCTTGAGGATC infected with WT HIV-1 or mock virus 48 h before cell lysates were subjected to CTTAAGCGTAATCTGGAACATCGTATGGGTACTCCTTGGAGGCCATG co-IP as described above. For GBP-binding assay, 293T cells (3 × 10 ) were TG-3′. The restriction enzyme sites are shown in bold, and the HA peptide transfected with 2 μg of a GFP-expressing control vector (GFP) or N′-GFP-APP sequence is underlined. The PCR product was then cloned into the expression (GFP-APP) along with 2 μg of HA-tagged forms of HIV-1 Gag (Gag-HA, Gag-N- vector pcDNA3.1 (Invitrogen). The inserts of all the expression constructs were Myr-HA or Gag-MA: -20-HA, -40-HA, -60-HA, and -80-HA) using 15 μl Tur- confirmed by sequencing. Expression constructs encoding C-terminally HA-tagged boFect (Thermo scientific). Two days post-transfection, cells were lysed with 1 ml Rev-independent HIV-1 Gag (Gag-HA), Matrix (MA-HA), Capsid (CA-HA) and cold NP-40 lysis buffer, subjected to GBP-binding assay followed by WB analysis as Gag-HA containing a single point mutation (Gag-MA-N-Myr) MA or serial described in ref. . G1A truncations of 20 aa in the C-terminus of MA (Gag-MA: -20-HA, -40-HA, -60-HA, and -80-HA) were described previously . Membrane flotation assay. A total of 1.5 × 10 293T cells were transfected with 3 μg of APP-Flag or pGAPDH-HA together with 1 μg Gag-HA or Gag-MA-60-HA Virion yield assays. A total of 1 × 10 293T cells were seeded in 60 mm dishes and followed by replacement of the media to fresh media 24 h post-transfection. The − − co-transfected with 0.4 μg of pNL4-3 or 0.7 μg of pNL4.3-luc.R E (Gag-Pol) along following day, cells were either lysed and subjected to WB analysis or collected and with increasing amounts of pCAGOSF-APP (APP-Flag) or either of the two subjected to membrane flotation assay as follows. The cells were washed with cold controls, pGAPDH-HA or empty vector pCAGOSF using Turbofect (Thermo 1×PBS and resuspended in cold 1×TE buffer with a Complete mini protease scientific). Total amount of DNA in each group was kept constant with addition of inhibitor tablet (Thermo Fisher). The cells were disrupted by Dounce homno- empty vector pcDNA3.1 or pCAGOSF. Following transfection, media was genizer with 10 pulses for 2 s/pulse. The cell lysates were then centrifuged at 500 × g replaced with fresh DMEM and supernatants were collected and filtered through a for 5 min at 4 °C to pellet the nuclei and unlysed cells. The resulting postnuclear 0.45 μm filter (Millipore) at 48 h post-transfection. The cells were then lysed and supernatant (PNS) was loaded under a discontinuous 10% to 75% sucrose gradient subjected to WB analysis for detection of Pr55 Gag, APP-Flag, GAPDH-HA and made in TNE buffer, and centrifuged to equilibrium in Beckman (Beckman the housekeeping proteins (PARP-1, β-tubulin and eIF4E) using antibodies specific Coulter, Optima XE-90) SW41Ti rotor overnight at 100,000 × g at 4 °C as described to each protein or to their tag as described below. Infectious virus yields were previously . After ultracentrifugation, 10 fractions were collected from the top to 26 4 determined by inoculating HeLa TZM-bl indicator cells seeded at 1.5 × 10 cells bottom of the density gradient. Equal volumes of samples from each fraction were in 96-well plates with 100 μl of serially diluted supernatants followed by mea- loaded onto an SDS-PAGE gel and subjected to WB analysis. For infected cell surements of beta-galactosidase activity 48 h post-infection using GalactoStar membrane flotation assays, 1 × 10 CHME3 4 × 4 cells were infected with WT reagent per manufacturer’s instructions (Life Tech). Physical particle yields were HIV-1 or mock virus 48 h before cell lysates were subjected to membrane flotation determined by WB analysis of virion containing supernatants either directly assay as described above. (supernatant from pNL4-3 transfected cells), or after pelleting the virion (super- − − natant from pNL4.3-luc.R E transfected cells) by passing the supernatants RNA interference. For transient knockdown, cells were transfected with siRNA through a 25% sucrose cushion by centrifugation at 100,000 g for 2 h at 4 °C, using anti HIV-1-Pr55/p24/p17 described below. For measurements of replication duplexes from Ambion using oligofectamine RNAiMAX (Invitrogen) as descri- bed . Briefly, cells were transfected with the control non-targeting siRNA (NC, competent HIV-1 virion yields, CHME3 4 × 4 cells were mock infected or infected with pNL4-3-derived HIV-1. Supernatants were collected and cells were lysed at ID# AM4635) or the APP-specific select siRNA duplexes (APP-V1 and APP-V2, days 3 and 5 post-infection followed by WB analysis as described above. ID# S1500 and S1501, respectively) each at 10 pmol. For western blotting and supernatant analysis, 5 × 10 CHME3 4 × 4 cells per well of 12-well plates were transfected with siRNAs. Twenty-four hours post-transfection, cells were trypsi- Western blotting. For WB analysis, cells were lysed in laemmli buffer and resolved nized and seeded at 2 × 10 cells per well on fresh 12-well plates. The following day on 10% SDS-PAGE gels. The levels of Aβ isoforms in supernatants were detected cells were infected with WT HIV-1, followed by collection of supernatants and cell by separating lysates on 4–12% Bis-tris Nu-PAGE gels (Invitrogen, NP0323BOX), lysis for measurements of infectious virions and intracellular protein levels, proteins were transferred to PVDF transfer membrane (Immobilon, IPVH00010) respectively, as described. For membrane flotation assays, two wells of a 6-well and blocked with 3% non-fat milk before incubating with primary antibodies. The plate, each containing 4 × 10 CHME3 cells, were transfected with siRNAs. uncropped western blots of all figures are shown in Supplementary Figs 6–13. Twenty-four hours post-transfection, both wells of CHME3 cells were trypsinized Antibodies used for WB were Flag (F7425) and HA (H3663) from Sigma; APP and pooled into a 10 cm dish. The following day, the cells were transfected with 1 (6E10, 803001) from Biolegend; HIV-1-Pr55/p24/p17 (ab63917) (labeled as anti- μg of Gag-HA or the empty vector pcDNA3.1 using 15 μl TurboFect (Thermo Pr55 Gag in the Fig.s), HIV-1-p24 (ab9071) (for detection of Pr55 Gag and/or p24 scientific). The cells were then collected and subjected to membrane flotation assay CA) and anti-APP antibody (Y188) (ab32136) from Abcam; eIF4E (610269) from as described above. For PBMCs, 1.5 × 10 PBMCs on a 60 mm dish were trans- BD Biosciences; GAPDH (sc-25778) from Santa Cruz; Aβ40 (PA3-16760) from fected with siRNA duplexes using oligofectamine RNAiMAX (Invitrogen) before Thermo Fisher; Aβ42 (700254) from Life Technologies; Caveolin-1 (D46G3), infection and analysis as described in figure legends and below. flotillin-1 (D2V27J), GFP (2555), PARP-1 (9542), Nicastrin (D38F9, 5665) and PEN2 (D6G8, 8598) from Cell Signaling. All primary antibodies were used at 1:1000 dilution and detected using the appropriate HRP-conjugated secondary Neurotoxicity assay. A total of 1 × 10 293T cells on a 60 mm dish were co- antibodies. transfected with 0.4 μg of APP-Flag and 1.6 μg of either the control pGAPDH-HA or Gag-HA using 8 μl TurboFect (Thermo scientific). Total DNA amount in each group was kept constant with addition of empty vector pcDNA3.1 . Alternatively, IF and congo red staining. For IF analysis, 4 × 10 293T or CHME3 cells grown on 5× 10 CHME3 4 × 4 cells on a 60 mm dish were infected with pNL4-3-derived glass coverslips in 6-well plate were co-transfected with equal amounts of HIV-1. Four hours post-transfection, or 1 day post-infection, the media were − − pCAGOSF-APP (APP-Flag) and Gag-HA or pNL4.3-luc.R E . Forty-eight replaced with fresh media without antibiotics containing either DMSO or 2.5 μMof hours post-transfection, cells were fixed with 3.7% paraformaldehyde, then blocked γ-secretase inhibitor L-685,458 (Tocris, Cat # 2627) reconstituted in DMSO. The and permeabilized with PBS supplemented with 0.1% Triton X-100 as described . following day, the media were replaced with neurobasal media (Gibco, 21103-049) Samples were incubated with anti-HIV-1-Pr55/p24/p17 (Abcam, ab63917) at 1:200 supplemented with B27 (Gibco, 17504-044) and Glutamax (Gibco, 35050-061) and anti-APP (LN27, Invitrogen, 130200) at 1:150 overnight at 4 °C. The next day, (2 ml/dish) containing either DMSO or 2.5 μMof γ-secretase inhibitor. 2 days samples were washed and incubated with the appropriate Alexa Fluor-conjugated post-transfection, supernatants were collected, filtered through 0.45 μm filters and secondary antibodies for 1 h at room temperature. Nuclei were stained with 1:3000 10 kDa Centrifugal Filter Unit (Abcam, ab93349), concentrated with Ultra Cen- Hoechst 33342. Images were acquired using a motorized spinning-disc confocal trifugal Filter concentrators (Millipore, UFC900308) prior to passing through a microscope (Leica DMI 6000B) with Yokogawa CSU-X1 A1 confocal head. For 3 kDa filter. A concentration of 2 μl of the concentrated supernatants from Congo Red staining, 10 μl supernatant was mixed with 200 μl Congo Red solution 293T cells were boiled in 2% SDS buffer for 10 min and subjected to ELISA analysis (Sigma) and incubated for 10 min at room temperature. Amyloid stained with for detection of Aß42 (Invitrogen, KHB3441) according to the manufacturer’s Congo Red was pelleted for 5 min at 18,407 g, the supernatant was discarded, the protocol. A total of 10 μl concentrated supernatants from CHME3 4 × 4 cells were pellet was then dissolved in 100 μl DMSO and the OD value of the solution was used for Congo Red staining as described above. For neurotoxicity assays, 0.5 × 10 determined at 490/650 nm. primary mouse cortical neurons (Life Technoogies)/well were plated in a poly-D- Lysine-coated 46-well plate and maintained in 500 μl of neurobasal medium sup- Co-IP and GBP-binding assay. For co-IP, CHME3 cells (3 × 10 ) were transfected plemented with 2% B-27 and 0.5 mM glutamax (Life Technologies). One half of the with 2 μg of APP-Flag or HA-tagged forms of HIV-1 Gag alone, or in combination medium was replaced with fresh medium every 3 days. On day 10, a concentration NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 11 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 of 200 μl of the concentrated in vitro supernatants from transfected 293T cells or 10. Willem, M. et al. eta-Secretase processing of APP inhibits neuronal activity in infected CHME3 4 × 4 cells described above were added onto the neurons while the hippocampus. Nature 526, 443–447 (2015). still in their own medium. It must be noted that, although supernatants were 11. Rempel, H. C. & Pulliam, L. HIV-1 Tat inhibits neprilysin and elevates amyloid initially concentrated as part of the clarification process, this served two additional beta. AIDS 19, 127–135 (2005). purposes. First, it allowed culture medium from transfected or infected cultures to 12. Andras, I. E. et al. Simvastatin protects against amyloid beta and HIV-1 Tat- be added to neuronal cultures without removing neuronal culture medium and induced promoter activities of inflammatory genes in brain endothelial cells. disturbing neuronal cultures. The final concentration of the added supernatants Mol. Pharmacol. 73, 1424–1433 (2008). after re-dilution upon addition to neuronal cultures was 3×. The second purpose of 13. Andras, I. E. et al. HIV-1-induced amyloid beta accumulation in brain this concentration was simply to collect all Aβ produced by cultured cells as the endothelial cells is attenuated by simvastatin. Mol. Cell. Neurosci. 43, 232–243 total amount produced is diluted into large volumes of culture medium, while (2010). in vivo neurons are directly exposed to factors secreted by neighboring microglia/ 14. Kurapati, K. R., Atluri, V. S., Samikkannu, T. & Nair, M. P. Ashwagandha macrophages. 3 days later the viability of the neurons was determined using Vita- (Withania somnifera) reverses beta-amyloid1-42 induced toxicity in human Orange Cell Viability Reagent (Biotool.com) following the suppliers protocol: to neuronal cells: implications in HIV-associated neurocognitive disorders each well 50 μl of the Vita-orange reagent was added and cells were incubated until (HAND). PLoS ONE 8, e77624 (2013). the orange color was developed. Wells without neurons, containing only the 15. Andras, I. E. & Toborek, M. Amyloid beta accumulation in HIV-1-infected medium, were used as a baseline. The results were recorded using a microplate brain: the role of the blood brain barrier. IUBMB. Life 65,43–49 (2013). reader, measuring the absorbance at 450 nm. Cell viability was calculated relative to 16. Braidy, N. et al. Recent rodent models for Alzheimer’s disease: clinical control group, treated with the supernatants from cells transfected with pGAPDH- implications and basic research. J Neural Transm 119, 173–195 (2012). HA or infected with mock virus, using the following formula: (AS-A0)/(AC-A0). 17. Achim, C. L. et al. Increased accumulation of intraneuronal amyloid beta in AS – sample absorbance; A0 – baseline absorbance, AC – control absorbance. For HIV-infected patients. J. Neuroimmune. Pharmacol. 4, 190–199 (2009). assays using PBMCs, 1.5 × 10 PBMCs on a 60mm dish were transfected with 18. Ances, B. M. et al. 11C-PiB imaging of human immunodeficiency virus- siRNA duplexes as described above. The following day, cells were infected with WT associated neurocognitive disorder. Arch. Neurol. 69,72–77 (2012). HIV-1 before supernatants were processed and used to perform neurotoxicity 19. Muller, U. C. & Zheng, H. Physiological functions of APP family proteins. Cold assays as described above. Spring Harb Perspect. Med. 2, a006288 (2012). 20. Malikov, V. et al. HIV-1 capsids bind and exploit the kinesin-1 adaptor FEZ1 Immuno-depletion of Aß. For immuno-depletion of Aβ, supernatants were col- for inward movement to the nucleus. Nat. Commun. 6, 6660 (2015). lected from HIV-1-infected CHME3 4 × 4 cells 48 h post infection, filtered through 21. Donahue, D. A. et al. SUN2 overexpression deforms nuclear shape and inhibits 0.45 μm filters and incubated at 4 °C for 2 h with 2 μl of anti-6E10 antibody bound HIV. J. Virol. 90, 4199–4214 (2016). to 15 μl of protein G-sepharose. The same amount of control mouse anti-HA 22. Sabo, Y. et al. HIV-1 induces the formation of stable microtubules to enhance antibody was used as a mock depletion. After incubation, beads were removed by early infection. Cell. Host. Microbe. 14, 535–546 (2013). centrifugation at 21,130 × g for 30 s. Beads and 20 µl of supernatant were boiled in 23. Sundquist, W. I. & Krausslich, H. G. HIV-1 assembly, budding, and laemmli buffer and subjected to WB analysis. maturation. Cold Spring Harb Perspect. Med. 2, a006924 (2012). 24. Dick, R. A. & Vogt, V. M. Membrane interaction of retroviral Gag proteins. Statistical analyses. Statistical analyses for two groups were performed in Front. Microbiol. 5, 187 (2014). Microsoft Excel 2016 using two-tailed Student’s t test. Statistical analyses for three 25. Freed, E. O. HIV-1 assembly, release and maturation. Nat. Rev. Microbiol. 13, or more groups were performed using Astatsa online statistical calculator. One-way 484–496 (2015). analysis of variance (ANOVA) was followed by Tukey’s test for post-hoc com- 26. Platt, E. J., Wehrly, K., Kuhmann, S. E., Chesebro, B. & Kabat, D. Effects of parisons of three or more experimental groups (only when ANOVA was sig- CCR5 and CD4 cell surface concentrations on infections by macrophagetropic nificant). Homogeneity of variance was confirmed with Levene’s test for equality of isolates of human immunodeficiency virus type 1. J. Virol. 72, 2855–2864 variances. To quantify co-localization of APP and Gag, the Pearson’s correlation (1998). coefficient (R ) was used. The samples were analyzed by using Fiji-ImageJ-NIH. R r r 27. Ono, A. & Freed, E. O. Binding of human immunodeficiency virus type 1 Gag ranges between −1 (perfect negative correlation) to +1 (perfect positive correlation) to membrane: role of the matrix amino terminus. J. Virol. 73, 4136–4144 with 0 meaning no correlation. All the data are expressed as mean ± SEM. The P (1999). value for all cases was set to < 0.05 for significant differences. *P < 0.05, **P < 0.01, 28. Simons, K. & Toomre, D. Lipid rafts and signal transduction. Nat. Rev. Mol. ***P < 0.001, ****P < 0.0001. Cell. Biol. 1,31–39 (2000). 29. Ono, A. & Freed, E. O. Plasma membrane rafts play a critical role in HIV-1 Data availability. The authors declare that the data supporting the findings of this assembly and release. Proc. Natl Acad. Sci. USA 98, 13925–13930 (2001). study are available within the article and its Supplementary Information files, or are 30. Lee, S. J. et al. A detergent-insoluble membrane compartment contains A beta available from the authors upon request. in vivo. Nat. Med. 4, 730–734 (1998). 31. Riddell, D. R., Christie, G., Hussain, I. & Dingwall, C. Compartmentalization of beta-secretase (Asp2) into low-buoyant density, noncaveolar lipid rafts. Curr. Received: 25 April 2017 Accepted: 16 October 2017 Biol. 11, 1288–1293 (2001). 32. Wahrle, S. et al. Cholesterol-dependent gamma-secretase activity in buoyant cholesterol-rich membrane microdomains. Neurobiol. Dis. 9,11–23 (2002). 33. Simons, K. & Vaz, W. L. Model systems, lipid rafts, and cell membranes. Annu. Rev. Biophys. Biomol. Struct. 33, 269–295 (2004). 34. Haass, C., Hung, A. Y. & Selkoe, D. J. Processing of beta-amyloid precursor References protein in microglia and astrocytes favors an internal localization over 1. Hellmuth, J., Milanini, B. & Valcour, V. Interactions between ageing and constitutive secretion. J. Neurosci. 11, 3783–3793 (1991). NeuroAIDS. Curr. Opin. HIV AIDS 9, 527–532 (2014). 35. Ivey, N. S., MacLean, A. G. & Lackner, A. A. Acquired immunodeficiency 2. Nightingale, S. et al. Controversies in HIV-associated neurocognitive disorders. syndrome and the blood-brain barrier. J. Neurovirol. 15, 111–122 (2009). Lancet. Neurol. 13, 1139–1151 (2014). 36. Rahimy, E. et al. Blood-brain barrier disruption is initiated during primary HIV 3. Rappaport, J. & Volsky, D. J. Role of the macrophage in HIV-associated infection and not rapidly altered by antiretroviral therapy. J. Infect. Dis. 215, neurocognitive disorders and other comorbidities in patients on effective 1132–1140 (2017). antiretroviral treatment. J. Neurovirol. 21, 235–241 (2015). 37. Nikolaev, A., McLaughlin, T., O’Leary, D. D. & Tessier-Lavigne, M. APP binds 4. Clifford, D. B. & Ances, B. M. HIV-associated neurocognitive disorder. Lancet DR6 to trigger axon pruning and neuron death via distinct caspases. Nature Infect. Dis. 13, 976–986 (2013). 457, 981–989 (2009). 5. Gonzalez-Scarano, F. & Martin-Garcia, J. The neuropathogenesis of AIDS. Nat. 38. Olsen, O. et al. Genetic analysis reveals that amyloid precursor protein and Rev. Immunol. 5,69–81 (2005). death receptor 6 function in the same pathway to control axonal pruning 6. Xu, Y. et al. HIV-1-mediated apoptosis of neuronal cells: Proximal molecular independent of beta-secretase. J. Neurosci. 34, 6438–6447 (2014). mechanisms of HIV-1-induced encephalopathy. Proc. Natl Acad. Sci. USA 101, 39. Xu, K., Olsen, O., Tzvetkova-Robev, D., Tessier-Lavigne, M. & Nikolov, D. B. 7070–7075 (2004). The crystal structure of DR6 in complex with the amyloid precursor protein 7. Ortega, M. & Ances, B. M. Role of HIV in amyloid metabolism. J. provides insight into death receptor activation. Genes Dev. 29, 785–790 (2015). Neuroimmune. Pharmacol. 9, 483–491 (2014). 40. Simon, V., Bloch, N. & Landau, N. R. Intrinsic host restrictions to HIV-1 and 8. van der Kant, R. & Goldstein, L. S. Cellular functions of the amyloid precursor mechanisms of viral escape. Nat. Immunol. 16, 546–553 (2015). protein from development to dementia. Dev. Cell. 32, 502–515 (2015). 41. Neil, S. J., Zang, T. & Bieniasz, P. D. Tetherin inhibits retrovirus release and is 9. Sevigny, J. et al. The antibody aducanumab reduces Abeta plaques in antagonized by HIV-1 Vpu. Nature 451, 425–430 (2008). Alzheimer’s disease. Nature 537,50–56 (2016). 12 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE P01GM105536 to D.W. and grant R01GM101975, P01GM105536 and R01NS099064 to 42. Van Damme, N. et al. The interferon-induced protein BST-2 restricts HIV-1 M.H.N. and with help from the Third Coast Center for AIDS Research (cFAR), an NIH release and is downregulated from the cell surface by the viral Vpu protein. Cell funded center (P30AI117943). Host Microbe. 3, 245–252 (2008). 43. Brugger, B. et al. The HIV lipidome: a raft with an unusual composition. Proc. Natl Acad. Sci. USA 103, 2641–2646 (2006). Author contributions 44. Rousso, I., Mixon, M. B., Chen, B. K. & Kim, P. S. Palmitoylation of the HIV-1 Q.C., V.J., V.M., Y.S. and S.M., performed the research; Q.C., V.J., V.M., Y.S., S.M., envelope glycoprotein is critical for viral infectivity. Proc. Natl Acad. Sci. USA D.W. and M.H.N. designed the research; Q.C., V.J., D.W. and M.H.N. analyzed the data; 97, 13523–13525 (2000). M.H.N. wrote the manuscript, and D.W. edited the manuscript. 45. Haass, C., Kaether, C., Thinakaran, G. & Sisodia, S. Trafficking and proteolytic processing of APP. Cold Spring Harb Perspect. Med. 2, a006270 (2012). Additional information 46. Bhattacharyya, R., Barren, C. & Kovacs, D. M. Palmitoylation of amyloid precursor protein regulates amyloidogenic processing in lipid rafts. J. Neurosci. Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- 33, 11169–11183 (2013). 017-01795-8. 47. Choy, R. W., Cheng, Z. & Schekman, R. Amyloid precursor protein (APP) traffics from the cell surface via endosomes for amyloid beta (Abeta) Competing interests: The authors declare no competing financial interests. production in the trans-Golgi network. Proc. Natl Acad. Sci. USA 109, Reprints and permission information is available online at http://npg.nature.com/ E2077–E2082 (2012). reprintsandpermissions/ 48. Morel, E. et al. Phosphatidylinositol-3-phosphate regulates sorting and processing of amyloid precursor protein through the endosomal system. Nat. Commun. 4, 2250 (2013). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 49. Edgar, J. R., Willen, K., Gouras, G. K. & Futter, C. E. ESCRTs regulate amyloid precursor protein sorting in multivesicular bodies and intracellular amyloid- beta accumulation. J. Cell. Sci. 128, 2520–2528 (2015). 50. Tomasselli, A. G. et al. Actin, troponin C, Alzheimer amyloid precursor protein Open Access This article is licensed under a Creative Commons and pro-interleukin 1 beta as substrates of the protease from human immunodeficiency virus. J. Biol. Chem. 266, 14548–14553 (1991). Attribution 4.0 International License, which permits use, sharing, 51. Wang, Y., Liu, L., Hu, W. & Li, G. Mechanism of soluble beta-amyloid 25–35 adaptation, distribution and reproduction in any medium or format, as long as you give neurotoxicity in primary cultured rat cortical neurons. Neurosci. Lett. 618, appropriate credit to the original author(s) and the source, provide a link to the Creative 72–76 (2016). Commons license, and indicate if changes were made. The images or other third party 52. Haedicke, J., Brown, C. & Naghavi, M. H. The brain-specific factor FEZ1 is a material in this article are included in the article’s Creative Commons license, unless determinant of neuronal susceptibility to HIV-1 infection. Proc. Natl Acad. Sci. indicated otherwise in a credit line to the material. If material is not included in the USA 106, 14040–14045 (2009). article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Acknowledgements licenses/by/4.0/. We thank Olivier Schwartz, Thomas Hope and Robert Vassar for reagents and advice and Katherine Sadleir and Kayla Schipper for technical help. The following reagent was obtained through the NIH AIDS Research and Reference Reagent Program, Division of © The Author(s) 2017 − − AIDS, NIAID, NIH: pNL4-3 from Dr. Malcom Martin, pNL4-3.Luc.R .E from Dr. Nathaniel Landau, and TZM-bl from Dr. John C. Kappes, Dr. Xiaoyun Wu and Tran- zyme Inc. This study was supported by National Institute of Health (NIH) grant NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 13 | | | http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Communications Springer Journals

HIV-1 counteracts an innate restriction by amyloid precursor protein resulting in neurodegeneration

Free
13 pages

Loading next page...
 
/lp/springer_journal/hiv-1-counteracts-an-innate-restriction-by-amyloid-precursor-protein-00OUs0XtHu
Publisher
Nature Publishing Group UK
Copyright
Copyright © 2017 by The Author(s)
Subject
Science, Humanities and Social Sciences, multidisciplinary; Science, Humanities and Social Sciences, multidisciplinary; Science, multidisciplinary
eISSN
2041-1723
D.O.I.
10.1038/s41467-017-01795-8
Publisher site
See Article on Publisher Site

Abstract

ARTICLE Corrected: Author correction DOI: 10.1038/s41467-017-01795-8 OPEN HIV-1 counteracts an innate restriction by amyloid precursor protein resulting in neurodegeneration 1 1 1 2 3 1 Qingqing Chai , Vladimir Jovasevic , Viacheslav Malikov , Yosef Sabo , Scott Morham , Derek Walsh & 1,2 Mojgan H. Naghavi While beta-amyloid (Aβ), a classic hallmark of Alzheimer’s disease (AD) and dementia, has long been known to be elevated in the human immunodeficiency virus type 1 (HIV-1)-infected brain, why and how Aβ is produced, along with its contribution to HIV-associated neuro- cognitive disorder (HAND) remains ill-defined. Here, we reveal that the membrane- associated amyloid precursor protein (APP) is highly expressed in macrophages and microglia, and acts as an innate restriction against HIV-1. APP binds the HIV-1 Gag poly- protein, retains it in lipid rafts and blocks HIV-1 virion production and spread. To escape this restriction, Gag promotes secretase-dependent cleavage of APP, resulting in the over- production of toxic Aβ isoforms. This Gag-mediated Aβ production results in increased degeneration of primary cortical neurons, and can be prevented by γ-secretase inhibitor treatment. Interfering with HIV-1’s evasion of APP-mediated restriction also suppresses HIV-1 spread, offering a potential strategy to both treat infection and prevent HAND. 1 2 Department of Microbiology-Immunology, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USA. Department of Biochemistry and Molecular Biophysics, Howard Hughes Medical Institute, Columbia University, New York, NY 10032, USA. MesaGen, LLC, South Salt Lake City, UT 84115, USA. Correspondence and requests for materials should be addressed to M.H.N. (email: Mojgan.naghavi@northwestern.edu) NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 1 | | | 1234567890 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 n addition to causing acquired immunodeficiency syndrome The gradual accumulation of amyloid plaques is associated (AIDS), HIV-1 crosses the blood–brain barrier (BBB) and with neurodegenerative conditions such as AD in uninfected Ienters the CNS in around 80% of infected individuals leading individuals . Antibodies that target Aβ aggregates have to disorders ranging from mild cognitive impairment to severe strengthened support for amyloid as a causative factor and 1,2 9 HIV-associated dementia (HAD) . While widespread use of therapeutic target in AD . Neurotoxic Aβ is generated by combination antiretroviral therapy (cART) has increased the life sequential site-specific proteolytic cleavage of the ubiquitously span of people living with HIV-1/AIDS, an estimated 50% of HIV expressed type I trans-membrane protein, APP. APP processing is patients on cART exhibit milder forms of HAND . The persis- mediated by four types of secretases (α, β, γ and η) via three tence of HAND is thought to involve poor antiretroviral drug alternative pathways (amyloidogenic, non-amyloidogenic, and η- 8,10 penetration and incomplete viral suppression in the CNS, as well secretase) (Fig. 1a) . Most APP processing is mediated by α- as possible toxic effects of therapy itself . Although HIV-1 does secretase, primarily at the plasma membrane, resulting in release not infect neurons, inside the CNS, it establishes infection in of a large N-terminal soluble fragment (sAPPα) into the extra- perivascular macrophages, microglia, and possibly astrocytes . cellular space and a short C-terminal fragment (α-CTF) into the These infected cells secrete a mix of host and viral proteins that cytoplasm. This process is referred to as the non-amyloidogenic contribute to inflammation and the complex events leading to pathway. Less frequently, in the amyloidogenic pathway, pro- 6,7 HIV-1-induced neuronal damage . However, one poorly cessing of APP by β-secretase generates a soluble ectodomain understood, yet potentially significant host contributor is Aβ. (sAPPβ) and a C-terminal fragment (β-CTF). CTFs can be further a b Antibodies: sAPPα sAPPβ sAPPη sAPPη sAPPα sAPPβ Aβ- APP - 130 Amyloid APP η-CTF Aη-β β-CTF - 55 Aη-α Pr55 Gag Aβ β β p3 α-CTF α α Lumen - 130 APP - 55 Pr55 Gag Cytosol AICD AICD Non-amyloidogenic Amyloidogenic η-secretase pathway pathway pathway Nucleus APP Gag Merge 0.6 *** 0.5 0.4 0.3 0.2 0.1 0.0 - 130 Anti-APP Anti-GAPDH - 34 Fig. 1 APP is highly expressed in macrophages and microglia and binds HIV-1 Gag. a APP processing through amyloidogenic, non-amyloidogenic and η- secretase pathways involves α-, β-, γ- and η-secretases. The Aβ peptide resulting in toxic amyloid oligomers and plaques is generated by sequential cleavages by β- and γ-secretases via the amyloidogenic pathway (central). b Human APP (APP-Flag) binds HIV-1 Gag (Gag-HA) in anti-APP co-IP from transfected 293T cells. c Endogenous APP and Gag colocalize in CHME3 cells infected with HIV-1 carrying vesicular stomatitis virus G (VSV-G) envelope glycoprotein at 16 h post infection (h.p.i). Nuclei were stained with Hoechst (blue). All images were obtained using a 100× oil objective of a spinning disk confocal microscope. Scale bar = 10 μm. d Quantification of APP and Gag as determined by Pearson’s coefficient in at least 10 random fields of view from samples as in c, shown as Mean ± SEM. The Pearson’s correlation coefficients were r = 0.512 ± 0.02. e Endogenous APP levels in glioblastoma (U87), normal human dermal fibroblasts (NHDF), microglia (CHME3), 293T and differentiated THP-1 cells. Molecular weight markers (in kDa) are shown to the right of WBs 2 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | Colocalized Non-colocalized U87 NHDF CHME3 293T THP-1 HIV-1 infected CHME3 Pearson’s coefficient APP-Flag Gag-HA APP-Flag+ Gag-HA IP: Anti-APP Input NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE processed by γ-secretase to create either a non-toxic peptide p3 Here we reveal that APP is highly expressed in macrophages from α-CTF, or Aβ monomers of various lengths from β-CTF, and microglia, natural target cells for HIV-1 infection in the which can self-associate to form toxic Aβ oligomers. γ-secretase brain, and acts as an innate restriction factor that sequesters the cleavage of α-or β-CTFs in the plasma membrane also releases HIV-1 Gag polyprotein in lipid rafts to block virus production fragments of varying sizes from the cytosolic APP intracellular and spread. To evade this restriction, HIV-1 Gag subverts host domain (AICD) into the cytoplasm. Amyloidogenic Aβ peptides secretases to cleave APP and clear membrane-associated CTFs, range from 30 to 42 amino acids (aa) in length, with two main but in doing so also results in increased Aβ production that toxic Aβ species, Aβ40 and Aβ42. Although Aβ40 accounts for causes the degeneration of cultured primary cortical neurons. Our 90% of all Aβ produced, the smaller Aβ42 fraction is more prone findings explain how and why infection leads to Aβ production to aggregation. While Aβ increases in the brain during normal and its contribution to neuronal damage, revealing an antiviral aging, Aβ accumulation is accelerated by HIV-1 infection and activity of APP that could potentially be exploited to treat both correlates with viral loads and the onset of HAND .Aβ also acts HIV-1 infection and neurodegeneration. as a biomarker for HAND, while drugs that inhibit Aβ produc- 11–14 tion may have therapeutic potential . Notably, distinct dif- Results ferences in Aβ deposition patterns between AD and HAND have APP is elevated in macrophages and microglia and binds HIV- been observed, suggesting that HIV-1 specifically alters Aβ 1 Gag. In screens for HIV-1 Gag-interacting host factors, we metabolism and this likely contributes to unique features of HAD identified APP , which was verified by transfecting 293T cells 7,15 and HAND . Indeed, several studies suggest soluble amyloid with plasmids encoding flag-tagged human APP or HA-tagged oligomers represent the primary pathological structure by per- HIV-1 Gag polyprotein alone or in combination. Co- meabilizing cellular membranes, leading to neuronal loss immunoprecipitation (co-IP) assays revealed that HIV-1 Gag observed in AD , and intraneuronal amyloid accumulation is a was only present in immune complexes when APP was present 17,18 predominant feature in HIV-infected brains . Despite this, (Fig. 1b), confirming their interaction. Immunofluorescence (IF) fundamental questions remain about how and why HIV-1 causes analysis of transfected cells also showed that APP co-localized Aβ production, and whether this directly contributes to neuronal with HIV-1 Gag in 293T cells (Supplementary Fig. 1a, b) and damage during infection. CHME3, a human microglia cell line and natural target cell type a c Input Bound APP binding domain Pr55 Gag -55 MA CA NC P6 Gag-HA HA Gag-N-Myr-HA HA Gag-MA-20-HA HA . . . -26 Gag-MA-40-HA p24 CA HA . . . . . . Gag-MA-60-HA HA p17 MA -17 . . . . . . . . . Gag-MA-80-HA HA Pr55 Gag -55 . . . . . . . . . . . . Exogenous APP in 293T Input Bound Anti-HA (Pr55 Gag) -55 -26 p24 CA Anti-APP -130 -17 p17 MA GFP-APP ++ + + + + +++ + + + Gag-HA + – –– –– + ––––– Anti-APP -130 Gag-N-Myr-HA – + –– – – – + –– – – Anti-GFP Gag-MA-20-HA – – + – –– – – + – –– (GFP) -26 Gag-MA-40-HA – – – + – – – – – + – – PARP-1 –95 Gag-MA-60-HA – –– –– + – –– – + – GFP +– –– + –– Gag-MA-80-HA – – – – – + – – – – – + GFP-APP – + ++ – + + Endogenous APP in CHME3 Gag-HA ++ – – + – – Input IP: Anti-APP MA-HA – – + – – + – -55 Anti-HA (Pr55 Gag) CA-HA – –– + – – + -130 Anti-APP Endogenous APP in HIV-1 infected CHME3 4x4 Gag-HA + – –– –– + – –– –– Gag-N-Myr-HA – + –– – – – + –– – – * -55 Input Anti-Pr55 Gag Gag-MA-20-HA – – + – –– – – + – –– -55 Gag-MA-40-HA – – – + – – – – – + – – IP: Anti-APP -130 Gag-MA-60-HA – –– – + – – –– – + – Anti-APP Gag-MA-80-HA – – – – – + – – – – – + Mock HIV-1 Fig. 2 APP binds the MA region of HIV-1 Gag. a GFP-tagged human APP (GFP-APP), but not GFP control, binds HIV-1 Gag (Gag-HA) or Matrix (MA- HA), but not Capsid (CA-HA) in GBP-binding assays. b Endogenous APP interacts with Pr55 Gag in WT HIV-1-infected CHME3 4 × 4 cells in anti-APP co- IP. *indicates unspecific bands detected in cell lysates. c Schematic of the HIV-1 Gag polyprotein used in binding assays, including the c-terminal HA tag. X indicates a point mutation in the N-terminal myristoylation site (Gag-N-Myr-HA). Sequential 20 aa deletions are indicated. d Gag mutants lacking aa 72-111 of MA (Gag-MA-40-HA, Gag-MA-60-HA, or Gag-MA-80-HA) fail to bind GFP-APP in co-transfected 293T cells in GBP-binding assay. e Gag mutants lacking aa 72-111 of MA (Gag-MA-40-HA, Gag-MA-60-HA, or Gag-MA-80-HA) fail to bind endogenous APP in CHME3 cells in anti-APP co-IP. Molecular weight markers (in kDa) are shown to the right of WBs NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 3 | | | Anti-Pr55 Gag Anti-HA ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 for HIV-1 infection in the brain (Supplementary Fig. 1c). Vali- (U87) and microglia (CHME3) also express high levels of APP dating co-transfection approaches, endogenous APP also co- compared with primary normal human dermal fibroblasts localized with Gag in HIV-1-infected CHME3 cells (Fig. 1c, d). (NHDFs) or 293T cells (Fig. 1e). Beyond brain-resident microglia, While APP and Gag were often expressed at high levels and co- human monocyte-derived macrophage cell lines (THP-1) also localized broadly throughout the cell, identifying cells that expressed high levels of APP, which exhibited altered mobility in expressed lower levels of both proteins more clearly illustrated SDS-PAGE, suggestive of an alternative isoform or post- their co-localization at distinct cellular regions, discussed below translational modification (Fig. 1e). This suggested that high (Fig. 1c and Supplementary Fig. 1a). APP is ubiquitously levels of APP expression in macrophages and microglia, natural expressed in many cell types but highly expressed in neurons . target cell types for HIV-1 infection, and its interaction with Gag We found that other human brain cell lines such as glioblastoma could be of particular biological significance during HIV-1 ac 293T (μg DNA) CHME3 4x4 THP-1 pCDNA3.1 1.8 1.4 1.0 0.7 0.4 0 1.0 0.7 0.4 0 130 - -130 APP pGAPDH-HA 0 0 00 0 0 0.4 0.7 1.0 1.4 55 - Pr55 Gag -55 APP-Flag 0 0 0.4 0.7 1.0 1.4 00 0 0 pNL4-3 -26 0 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 26 - eIF4E 130- -34 APP-Flag GAPDH-HA -26 26 - p24 CA Pr55 Gag -55 35,000 ** p24 CA 30,000 -130 25,000 ** 4000 PARP-1 -55 20,000 * β-tubulin 15,000 -26 eIF4E 10,000 p24 CA -26 siRNA: d5 p17 MA -17 Days post- infection (d.p.i): d3 d5 1 2 3 4 5 67 8 9 10 b e p24 ELISA 1,000,000 CHME3 4x4 THP-1 600 60 *** *** ** d.p.i: d3 d.p.i: d5 d.p.i: d5 *** *** 500 500 100,000 400 400 *** 300 300 200 200 10,000 100 100 0 0 APP-Flag siRNA: pGAPDH-HA 0 0.4 0.7 1.1 1.4 APP-Flag or pGAPDH (μg) f HeLa-TZM-bl CHME3 4x4 THP-1 8000 30,000 10,000 *** *** d.p.i: d3 d.p.i: d5 ** d.p.i: d5 25,000 6000 8000 *** ** 20,000 *** 4000 15,000 10,000 0 0 0 siRNA: Fig. 3 APP inhibits production of HIV-1 particles. a Increasing expression of human APP (APP-Flag) reduces the levels of MA/CA-containing HIV-1 particles in culture supernatants compared with human GAPDH (GAPDH-HA) controls. b Infectious virion yields from APP or GAPDH expressing 293T cells measured using TZM-bl indicator cells. c RNAi-mediated depletion of APP relieves an endogenous block to production of extracellular virus particles in CHME3 4 × 4 cells and differentiated THP-1 cells infected with pNL4-3-derived HIV-1 at 3 and/or 5 days post-infection (d.p.i: d3 and d5). d Quantification of p24 CA intensity in supernatants from c. e Measurements of p24 CA levels in supernatants from c by ELISA. f Infectious virus yield in samples from c was measured using TZM-bl indicator cells. The data in b, d, e, f represent average of 3 replicates, and are represented as mean +/− SEM (one-way ANOVA; *P < 0.05, **P < 0.01, ***P < 0.001). Molecular weight markers (in kDa) are shown either to the right or to the left of WBs 4 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | Infectious virion yield (Log RLU) Virions Cell lysate Infectious virion yield p24 (pg/ml) p24 CA intensity Cell lysate Virions Control Control Control APP-V1 APP-V1 APP-V1 APP-V2 APP-V2 APP-V2 Control APP-V1 APP-V2 Control Control p24 CA intensity APP-V1 APP-V1 APP-V2 APP-V2 Control APP-V1 APP-V2 Control Control Virions Cell lysate APP-V1 APP-V1 APP-V2 APP-V2 NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE Total lysate or sup Fractionated lysate GAPDH + WT Gag APP + WT Gag Anti-APP Anti-Pr55 Gag 55 Anti-GAPDH Anti-GAPDH Pr55 Gag Anti-Flotilin-1 55 GAPDH Anti-PARP-1 95 Anti-Caveolin-1 p24 CA Anti-Nicastrin Anti-PEN2 12 3 4 5 12 3 4 5 67 8 910 67 8 910 WT Gag Top Bottom Top Bottom GAPDH + Gag-MA-60-HA (Δ APP binding site) APP + Gag-MA-60-HA (Δ APP binding site) Anti-APP Gag-MA-60-HA Anti-HA 34 GAPDH Anti-Flotilin-1 34 Anti-GAPDH Anti-elF4E Anti-PEN2 p24 CA 123 4 5 678 910 12 3 4 5 67 8 910 Top Bottom Top Bottom Gag-MA-60-HA CHME3 transfected with control siRNA CHME3 transfected with APP-V2 siRNA Anti-APP Anti-Pr55 Gag 55 Anti-Pr55 Gag Anti-Caveolin-1 Anti-eIF4E Anti-Nicastrin Anti-PARP-1 Anti-PEN2 10 12 3 4 5 67 8 910 12 3 4 5 67 8 910 Top Bottom Top Bottom 293T co-transfected with APP + pcDNA3.1 293T co-transfected with APP + WT Gag Anti-APP Anti-Pr55 Gag 55 f 293T (μg DNA) Anti-Flotilin-1 0.3 0.2 0.1 0 0.3 0.2 0.1 0 130 pcDNA3.1 Anti-Nicastrin 0000 0.1 0.2 0.3 0.4 pGAPDH-HA Gag-HA 0.1 0.2 0.3 0.4 00 0 0 Anti-PEN2 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 APP-Flag 123 4 5 6789 10 123 4 56789 10 Anti-Pr55 Gag Anti-HA Top Bottom Top Bottom CHME3 4x4 infected with mock CHME3 4x4 infected with HIV-1 6E10 130 Anti-APP 130 LN27 55 Anti-Pr55 Gag Anti-eIF4E Anti-Nicastrin Anti-PEN2 123 4 56789 10 123 4 56789 10 Top Bottom Top Bottom Fig. 4 APP causes Gag retention in membrane domains. a, b Analysis of cell lysates or supernatant VLPs (left panels) or fractionated lysates from flotation assays (right panels). a Left: In cells expressing WT Gag, APP expression reduces VLP levels in cell supernatants. Right: HIV-1 Gag is present in both membrane-free (8–10) and membrane-bound fractions containing lipid raft and γ-secretase components (2–4) in control samples, but is solely membrane- bound in APP-expressing 293T cells. b Left: In cells expressing mutant Gag that does not bind APP, APP expression does not affect VLP levels in cell supernatants. Right: HIV-1 Gag lacking the APP-binding site is present in membrane-free fractions in both control and APP-expressing 293T cells. c Left: WB verification of APP depletion in cell lysates. Right: RNAi-mediated depletion of APP increases the release of HIV-1 Gag into membrane-free fractions in CHME3 cells transfected with HIV-1 Gag expression plasmid. d, e HIV-1 Gag expression does not release APP from membrane-bound fractions containing lipid raft and γ-secretase components in transfected 293T cells (d) or in CHME3 4 × 4 cells infected with WT HIV-1 (e). f Increasing Gag-HA expression results in decreased APP levels in transfected 293T cells compared with control GAPDH-HA expression. Molecular weight markers (in kDa) are shown either to the right or to the left of WBs NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 5 | | | GAPDH APP GAPDH APP Control APP-V2 Anti-HA VLPs Cell lysate VLPs Cell lysate Anti-HA Anti-APP ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 infection in the brain. Moreover, the levels of APP expression in (Supplementary Fig. 2). Notably, further increasing APP expres- APP-transfected 293T cells resembled those found naturally in sion to very high levels also affected intracellular Pr55 Gag macrophages and microglia (Figs. 1b, e and 2), demonstrating abundance (Supplementary Fig. 2). These findings suggested that that transfected 293T cells offered a tractable system to under- APP primarily affected HIV-1 or VLP production and/or release, stand APP function at physiologically relevant levels. but at very high levels APP exerted secondary effects on intra- To independently verify this interaction and identify the APP- cellular Gag accumulation. Validating findings in transfected interacting domain in Gag, 293T cells were transfected with 293T cells, RNAi-mediated depletion of APP in CHME3 4 × 4 plasmids encoding GFP-tagged APP together with HA-tagged cells resulted in an increase in extracellular mature virus particles Pr55 Gag polyprotein, p24 capsid (CA) or p17 matrix (MA). in supernatants from cells infected with WT (pNL4-3-derived) GFP-APP was recovered from soluble cell extracts on GFP- HIV-1, as detected by either p24 CA WB or ELISA analysis binding protein (GBP)-conjugated sepharose . APP was found (Fig. 3c–e, and Supplementary Fig. 3a, b). Although RNAi- to specifically interact with Gag or the MA portion of Gag, but mediated depletion was less efficient, reducing APP expression in not CA (Fig. 2a). To confirm these results using endogenous APP THP-1 differentiated to macrophages also significantly increased in the context of infection in natural target cell types, CHME3 the levels of mature virions in culture supernatants (Fig. 3c–e). 4 × 4 cells, which express higher levels of CD4 and CXCR4 for Applying supernatants to TZM-bl indicator cells confirmed that more efficient infection with WT HIV-1 envelope , were infected this corresponded to an increase in production of extracellular with HIV-1 followed by anti-APP co-IP. In line with findings in infectious virus particles from APP-depleted microglia or mac- co-transfected 293T cells, endogenous APP also co- rophages infected with WT (pNL4-3-derived) HIV-1 (Fig. 3f). immunoprecipitated with Pr55 Gag in HIV-1-infected cells siRNA-mediated silencing of APP did not affect cell viability as (Fig. 2b and below). To further define the region of MA involved, detected by the lack of PARP-1 cleavage in these cells (Supple- Gag expression plasmids with serial truncations or mutations in mentary Fig. 3c). As such, APP overexpression in 293T cells or MA (Fig. 2c) were tested for binding to either exogenous GFP- naturally high levels of APP in natural target cells such as APP in transfected 293T cells or endogenous APP in CHME3, microglia or monocyte-derived macrophages suppressed the using two independent approaches. In 293T cells co-transfected production of extracellular HIV-1 particles. with different mutants of Gag-HA together with GFP-APP, GBP- binding assays revealed that mutations in the N-terminal myristoylation site (Gag-N-Myr-HA) or deletion of the last 20 APP retains Gag in lipid rafts. We next determined whether aa of the C-terminus of MA had no effect on APP binding APP influenced Gag localization using membrane flotation assays (Fig. 2d). However, larger deletions of 40, 60, or 80 aa in the MA where-in cell lysates are fractionated to separate membrane- C-terminus impaired Gag binding to APP. Validating these bound and membrane-free fractions . In control cells co- findings using endogenous APP, CHME3 cells were transfected transfected with GAPDH, Gag was present in both membrane- with the same HA-tagged forms of wild-type (WT) or mutated bound (2–4) and membrane-free (8–10) fractions (Fig. 4a and Gag followed by anti-APP co-IP. Similar to observations in co- Supplementary Fig. 4a), which is in line with previous reports . transfected 293T cells, APP again failed to interact with Gag-MA- By contrast, all detectable Gag was found exclusively in 40-HA, Gag-MA-60-HA or Gag-MA-80-HA mutants, but membrane-bound fractions in 293T cells overexpressing APP, efficiently bound myristoylation or Gag-MA-20-HA mutants and again decreased the levels of p24 CA in culture supernatants (Fig. 2e). This demonstrated that residues 72-111 of MA were (Fig. 4a and Supplementary Fig. 4a). In line with its general localization to diverse cellular compartments, GAPDH in either required for Gag binding to either transfected, tagged forms of APP in 293T cells, or endogenous APP in microglia. GAPDH- or APP-overexpressing cells was broadly distributed across both membrane and free cytosolic fractions. HIV-1 Gag/ MA has been reported to associate with membrane lipid rafts 28,29 APP inhibits HIV-1 virion production and spread. APP is during virion assembly and release , and membrane- membrane-associated , while the MA domain mediates plasma- associated fractions containing Gag were found to contain lipid membrane association of the Gag precursor protein during raft markers, Flotillin-1 and Caveolin-1 (Fig. 4a). This was 23–25 assembly and budding of new HIV-1 particles . To test observed in both APP- and GAPDH-expressing cells, demon- whether APP could affect virus production, 293T cells were strating that APP overexpression did not disrupt lipid raft com- transfected with an infectious cDNA clone of HIV-1 (pNL4-3) position, but retained Gag at these sites. Notably, lipid rafts are 30–33 together with increasing amounts of either APP or GAPDH also enriched in β- and γ-secretases that process APP , and control plasmids. Western blot (WB) analysis confirmed membrane-associated fractions containing Gag were also speci- increasing expression of APP or GAPDH in each case, while the fically enriched for the γ-secretase components, Nicastrin and levels of housekeeping proteins (β-tubulin and eIF4E) or PARP-1, Presenilin Enhancer 2 (PEN2) (Fig. 4a). To determine if binding an apoptosis indicator, confirmed no adverse effects on cell via- to APP mediated Gag sequestration at these sites and production bility (Fig. 3a). While intracellular Pr55 Gag or p24 CA levels of extracellular virus particles, we compared the effects of over- were moderately elevated with increasing APP expression expressing APP versus GAPDH on the distribution of the APP- (Fig. 3a), supernatant levels of p24 CA and p17 MA components binding mutant, Gag-MA-60-HA. In contrast to WT Gag, the of mature particles revealed a highly potent, dose-dependent Gag-MA-60-HA mutant exhibited a similar distribution to both reduction in APP-expressing cells (Fig. 3a). Applying these membrane-bound and membrane-free fractions in both GAPDH supernatants to TZM-bl indicator cells confirmed that APP control and APP-overexpressing cells (Fig. 4b and Supplementary expression blocked production of extracellular infectious virus Fig. 4b). Moreover, while APP expression resulted in a potent particles (Fig. 3b). Whether this reduction in extracellular virus in decrease in VLPs in supernatants from cells expressing WT Gag APP-expressing cells reflects decreased maturation, assembly or (Fig. 4a, left panels), VLP production and release from cells budding remains to be determined. To confirm that these effects expressing Gag-MA-60-HA was unaffected by APP (Fig. 4b, left on infectious HIV-1 replication reflected effects of APP on the panels). These findings demonstrated that APP binding did Gag polyprotein, 293T cells were transfected with HIV-1 Gag, indeed influence Gag membrane localization and the production which produces and releases virus-like particles (VLPs). Expres- of extracellular virions. In line with findings in 293T over- sion of APP potently blocked production of extracellular VLPs expression systems, the converse approach of RNAi-mediated 6 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | d.p.i: d3 NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE a b c d e APP-FL 130 130 Diffrentiated Diffrentiated THP-1 (HIV-1-VSV-G) Anti-APP lighter CHME3 4x4 (WT HIV-1) THP-1 (WT-HIV-1) APP-FL 55 5 130 Anti-Aβ 42 Pr55 Gag APP-FL Anti- Aβ 42 Anti-Aβ 40 α-CTF 34 1.4 GAPDH 1.8 ** 1.2 β-CTF 1.5 1.0 0.8 α-CTF Anti-GAPDH 1.2 0.6 0.9 55 0.4 Pr55 Gag 95 β-CTF 0.6 Anti-PARP-1 10 0.2 α-CTF 26 0.3 0.0 Anti-eIF4E Pr55 Gag 0.0 Mock HIV-1 Anti-Aβ 42 d.p.i: d5 p24 CA 5 GAPDH Anti-Aβ 40 95 Anti-PARP-1 d.p.i: d3 d5 N.I. Anti-eIF4E γ-secretase d.p.t: d3 d5 inhibitor Anti- 5 - Aβ 42 f gi CHME3 4x4 (WT HIV-1) HeLa-TZM-bl CHME3 4x4 (WT HIV-1) DMSO γ-secretase Nicastrin 130 APP-FL inhibitor APP-FL α-CTF 10 CHME3 4x4 (WT HIV-1) Pr55 Gag APP eIF4E 0 Pr55 Gag β-CTF α-CTF p24 CA GAPDH 55 34 d.p.i: d5 siRNA: 4000 Pr55 Gag p24 CA 26 PARP-1 2000 p24 ELISA p24 CA *** *** *** 500 *** 200 400 *** 1500 *** d.p.i: d7 *** *** *** siRNA: 400 ** 150 300 100 200 DMSO γ-secretase 50 100 inhibitor 0 0 N.I. N.I. N.I. γ-secretase γ-secretase γ-secretase γ-secretase inhibitor inhibitor inhibitor inhibitor d.p.i: d3 d5 d7 Fig. 5 HIV-1 Gag promotes APP processing into Aβ isoforms. a Decreased APP levels in Gag-expressing 293T cells correlates with increased secretion of Aβ40 and Aβ42 compared to GAPDH (Ctrl) at days 3 (d3) and 5 (d5) post-transfection (d.p.t). b Gag-induced processing of APP and Aβ42 accumulation in transfected 293T cells is blocked by γ-secretase inhibitor treatment. Full-length APP: APP-FL, C-terminal fragments: α-CTF and β-CTF. c, d Aβ40 and/or Aβ42 secretion from CHME3 4 × 4 cells c or differentiated THP-1 cells d is enhanced upon infection with WT HIV-1 (pNL4-3 derived), detected by either WB analysis (top panels) or Congo Red staining of amyloid (lower graphs). e Infection of differentiated THP-1 cells with HIV-1 pseudotyped with VSV-G envelope to increase infection efficiency results in production of p24 CA and a reduction in cellular APP levels, and both process are inhibited by γ- secretase inhibitor. f Depletion of Nicastrin suppresses APP processing, as seen by the accumulation of APP and CTFs, and reduces the levels of WT HIV-1 particles in supernatants of infected CHME3 4 × 4 cultures. g, h γ-secretase inhibitor treatment blocks APP processing, as seen by the accumulation of α- CTF and β-CTF, and reduces the levels of WT HIV-1 particles in supernatants of infected CHME3 4 × 4 cultures at d3-d7 d.p.i., detected either by WB using anti-p24 CA antibody (g) or ELISA (h). i As in g, h, except that infectious virus yield was measured using TZM-bl indicator cells. j siRNA-mediated depletion of APP in CHME3 4 × 4 renders infection with WT HIV-1 insensitive to γ-secretase inhibitors as determined by measurements of p24 CA levels in culture supernatants by either WB analysis or ELISA. The data in c, d (lower panels) as well as h–j (lower panel) represent average of 3 replicates, and are represented as mean +/− SEM (one-way ANOVA; *P < 0.05, **P < 0.01, ***P < 0.001). Molecular weight markers (in kDa) are shown either to the right or to the left of WBs depletion of APP in CHME3 cells followed by transfection with with WT HIV-1 (Fig. 4e and Supplementary Fig. 4e). Although HIV-1 Gag increased the proportion of Gag present in APP distribution was unaffected by Gag expression, the levels of membrane-free fractions (Fig. 4c and Supplementary Fig. 4c). APP expression were notably lower in fractions from Gag- This demonstrated that the integral membrane protein, APP expressing cells compared with controls (Fig. 4d). Indeed, APP binds and retains Gag in membrane regions rich in lipid raft overexpression was less efficient in the presence of Gag in earlier markers, which would explain the reduction in extracellular experiments (Supplementary Fig. 2), suggesting an antagonistic infectious virions and VLPs in cells expressing high levels of APP relationship between Gag and APP. To test this further, we (Figs. 3, 4a, b, and Supplementary Fig. 2, left panels). transfected 293T cells with a constant amount of APP together We next addressed the reciprocal question of whether Gag with increasing amounts of either Gag-HA or GAPDH-HA influenced APP membrane localization by co-transfecting cells control. WB analysis of cell lysates revealed that increasing levels with APP and either HIV-1 Gag or empty vector control. In of Gag resulted in a dose-dependent reduction in APP expression control samples, APP localized to the same membrane fractions in cells (Fig. 4f), possibly by enhancing APP’s turnover in lipid as Gag, which again contained lipid raft and γ-secretase rafts to circumvent APP-mediated restriction. components (Fig. 4d and Supplementary Fig. 4d). Validating findings in transfected 293T cells, endogenous APP also localized to membrane-associated fractions containing Gag as well as lipid HIV-1 Gag promotes processing of APP into neurotoxic Aβ raft and γ-secretase components in CHME3 4 × 4 cells infected isoforms. Given that HIV-1 Gag decreases APP levels both inside NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 7 | | | Ctrl Gag Ctrl Gag Control Ctrl Nicastrin A Gag Nicastrin B Ctrl Gag DMSO 1 μM 2.5 μM DMSO 1 μM 2.5 μM Anti-CTF (Y188) Anti-HA Virions Cell lysate Supernatant Cell lysate Cell lysate Sup Cell lysate p24 (pg/ml) Virions Anti-HA Anti-CTF (Y188) Congo DMSO 490 650 Red A –A 1 μM 2.5 μM Mock HIV-1 DMSO Mock 1 μM HIV-1 2.5 μM Congo DMSO 490 650 Red A –A 1 μM 2.5 μM Infectious Infectious Infectious virion yield (RLU) virion yield (RLU) virion yield (RLU) Cell lysate Virions p24 (pg/ml) Control Control Darker APP-V1 exposure APP-V2 Virious Cell lysate ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 Supernatants Gag transfected 293T HIV-1 infected CHME3 4x4 from: a d fj HIV-1 infected PBMCs 130 130 Immunodepletion 120 120 ## 130 110 110 Beads-bound APP 100 100 * 55 Pr55 Gag 90 90 5 Supernatants 80 80 GAPDH 34 70 70 Anti-Aβ Anti-HA 60 60 eIF4E (6E10) 50 50 30 p24 CA ** 20 1.0 10 60,000 0 0.8 50,000 0.6 40,000 b e ** 0.4 30,000 **** 0.2 160 20,000 160 ‡‡ 0.0 10,000 140 ## ## 0 siRNA: *** ** 0 20 60 2.50 Ctrl Gag Ctrl Gag 40 2.25 *** Mock HIV-1 ** DMSO γ-secretase 2.00 inhibitor 1.75 1.50 Anti-Aβ Anti-HA (6E10) 1.25 1.2 1.00 0.75 1 0.35 0.50 0.3 0.25 0.8 0.00 0.25 0.2 0.6 l 0.18 *** 0.15 Y = –0.2465X + 0.4304 0.16 0.1 R = 0.18557 0.4 0.14 y = –0.0012x + 0.9375 Ctrl + DMSO N = 20 0.05 0.12 R = 0.37785 Gag + DMSO P < 0.00001 0.2 N = 44 Ctrl + inhibitor 0.10 0 0.2 0.4 0.6 0.8 1 1.2 P < 0.00001 Gag + inhibitor 0.08 490 650 Congo Red A –A 0.06 0 50 100 150 200 250 300 (% of control) Aβ42 (pg/ml) 0.04 0.02 0.00 siRNA: Fig. 6 HIV-1 Gag-induced Aβ42 production causes neurodegeneration. a Cortical neurons treated with supernatants from 293T cells co-transfected with # ## APP-Flag along with either pGAPDH-HA (Ctrl) or HIV-1 Gag (Gag) reveals neurotoxicity caused by Gag expression. *P < 0.05 vs. Ctrl; P < 0.05, P < ## 0.01 vs. gag (one-way ANOVA; n = 11; F = 5.665; P < 0.05). b ELISA analysis of Aβ42 levels in denatured supernatants from a. **P < 0.01 vs. Ctrl, P < 40,3 ‡‡ 0.01 vs. gag, P < 0.01 vs. Ctrl + inhibitor (one-way ANOVA; n = 11; F = 53.340; P < 0.001). Bars represent SEM. c Regression plot showing the 40,3 correlation between cell viability and Aβ42 levels from a-b (F = 25.508; P < 0.00001; R = 0.378). d Cortical neurons treated with supernatants from 42,1 mock or WT HIV-1-infected CHME3 4 × 4 cells reveals neurotoxicity caused by HIV-1. *P < 0.05 vs. mock; (n = 11; t = 2.49; P < 0.05). e Congo Red staining of amyloid levels in supernatants from d. ****P < 0.0001 vs. mock (n = 11; t = -9.37; P < 0.0001). Bars represent SEM. f Bead-bound and supernatant levels of Aβ in 6E10 or anti-HA antibody-treated samples from WT HIV-1-infected CHME3 4 × 4. g Congo Red staining of amyloid levels in supernatants from f.(n = 10, **P < 0.01). h Cortical neurons treated with supernatants from f.(n = 10, ***P < 0.001). i Regression plot showing the correlation between cell viability and Aβ levels from samples in f–h (F = 3.874; P < 0.00001; R = 0.186). j–l APP depletion in PBMCs infected with WT 17,1 HIV-1 causes an increase in p24 CA j, a decrease in Aβ levels determined by Congo Red staining (k) and a corresponding increase in viability of cortical neurons treated with supernatants from j, l. The data in j (lower panel), k, l represent mean +/− SEM. k Congo Red staining of amyloid levels in supernatants from j.**P < 0.01 vs. Ctrl, ***P < 0.001 vs. Ctrl, (one-way ANOVA; n = 10, P < 0.001). l Cortical neurons treated with clarified supernatants from PBMCs treated as in j, k reveals APP depletion reduces neurotoxicity caused by infection and its inverse correlation with amyloid levels. ***P < 0.001 vs. Ctrl (one-way ANOVA; n = 10, P < 0.001). Molecular weight markers (in kDa) are shown to the right of WBs the cell and in γ-secretase enriched lipid rafts, we tested whether controls (Fig. 5a). This decrease was accompanied by a corre- Gag stimulated APP processing into Aβ isoforms. Co-transfection sponding increase in the levels of secreted Aβ42, and to a lesser of 293T cells with vectors expressing APP along with either HIV- extent Aβ40 in the supernatants of Gag-expressing cells (Fig. 5a). 1 Gag or GAPDH again revealed a large decrease in the intra- This suggested that HIV-1 Gag enhanced APP processing, most cellular levels of APP in Gag-expressing cells compared to likely through secretases. In agreement with this, the decrease in 8 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | Control APP-V1 APP-V2 Control APP-V1 APP-V2 Aβ42 (pg/ml) Viability (% of control) Congo Red 490 650 A –A (% of control) Viability (% of control) Viability Congo Red 490 650 (% of Aβ depletion) A –A (% of 6E10) Congo Red P24 CA intensity 450 490 650 Viability (A ) A –A (% of control) Virions Cell lysate NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE APP levels induced by Gag expression, and accompanying neurons, those taken from Gag-expressing cells resulted in a increase in Aβ42 secretion, could be blocked by treating trans- statistically significant decrease in cell viability compared to fected cells with γ-secretase inhibitor (Fig. 5b). In line with control samples (Fig. 6a). ELISA analysis of supernatants showed blocking γ-secretase activity, inhibitor-treated cells were also that this effect paralleled increased Aβ42 levels in Gag-expressing found to contain higher levels of α- and β-CTF processing cell supernatants (Fig. 6b), similar to Aβ42 WB analysis (Fig. 5b). intermediates (Fig. 5b). Levels of Gag as well as host eIF4E or Notably, ELISA was performed using antibody suitable for PARP-1 were unaffected, demonstrating that inhibitor treatment detection of monomeric Aβ42 and failed to detect Aβ42 in did not block APP processing by indirectly affecting Gag supernatants until samples were extensively denatured, demon- expression or cell viability, establishing that Gag enhanced γ- strating that the Aβ42 being produced and detected, represents secretase-dependent processing of APP. In line with observations soluble amyloid oligomers. In control samples treated with γ- in transfected 293T cells, infection of CHME3 4 × 4 cells or dif- secretase inhibitor, reducing Aβ42 levels to below normal levels ferentiated THP-1 cells with WT HIV-1 resulted in elevated Aβ40 observed in DMSO-treated cells had no significant impact on and Aβ42 in culture supernatants, as detected by either WB neuronal viability. By contrast, treatment of Gag-expressing cells analysis or staining with the amyloid-binding detection reagent, with γ-secretase inhibitor blocked Gag-induced increases in Aβ42 Congo Red (Fig. 5c, d, respectively). production and prevented Gag-induced neurotoxicity (Fig. 6a, b). Exploring this further, and validating the biological relevance of Indeed, linear regression analysis showed significant correlation secretase-mediated APP processing in Gag-transfected 293T cells, between the neurotoxic effects of clarified supernatants and the infection of differentiated THP-1 cells with HIV-1 pseudotyped with levels of Aβ42 present in supernatants under each condition VSV-G envelope, to attain high multiplicity of infection (m.o.i), also (Fig. 6c). To validate this Gag-mediated neurodegeneration in the resulted in decreased abundance of APP and CTFs that could be context of WT infection, the effects of clarified supernatants from blocked by inhibiting γ-secretase (Fig. 5e). In addition, blocking this mock- or WT HIV-1-infected CHME3 4 × 4 cells were also decrease in APP using the γ-secretase inhibitor resulted in a assessed. Similar to Gag-transfected 293T cells, supernatants from corresponding decrease in mature virions in culture supernatants, as HIV-1-infected cells contained statistically significant increases in determined by WB analysis of p24 CA (Fig. 5e). Independently Aβ42 levels and resulted in a corresponding increase in validating observations using the γ-secretase inhibitor in THP-1 neurodegeneration compared to mock-infected control samples cells, RNAi-mediated depletion of the γ-secretase subunit, Nicastrin (Fig. 6d, e). To confirm that Aβ did indeed cause neurotoxic in CHME3 4 × 4 cells infected with WT HIV-1 resulted in increased effects of HIV-1-infected culture supernatants, Aβ was immuno- APPand CTFlevels, andacorrespondingdecreaseinp24 CA in depleted using the antibody, 6E10 prior to clarification of culture supernatants (Fig. 5f). This suggested that protecting APP supernatants from WT HIV-1-infected CHME3 4 × 4 followed and CTFs from processing suppressed HIV-1 infection. In line with by assessment of effects on neuronal viability. Successful Aβ this, a β-secretase inhibitor also increased APP and CTF expression depletion from clarified supernatants was confirmed by WB and suppressed p24 CA levels in supernatants of CHME3 4 × 4 analysis, which showed a large increase in bead-bound Aβ over infected with WT HIV-1 (Supplementary Fig. 5a, b). Testing background levels in control antibody-treated samples, and a potential effects on HIV-1 spread in natural target cells, treatment of corresponding reduction in supernatant levels of Aβ (Fig. 6f). WB CHME3 4 × 4 with γ-secretase inhibitor to prevent APP processing analysis of immune-depleted samples was further confirmed by suppressed the production of extracellular virus particles, as Congo Red staining of Aβ (Fig. 6g). When clarified, immuno- determined by p24 CA levels detected by either WB analysis or depleted supernatants from HIV-1-infected cells were applied to ELISA (Fig. 5g, h, respectively), as well as infectious HIV-1 virions cortical neurons, Aβ depletion resulted in a significant increase in (Fig. 5i) between 3 and 7 d.p.i. It must be noted that HIV-1-induced neuronal viability compared with controls (Fig. 6h, i). changes in APP are less obvious in low m.o.i. spreading assays due Finally, although challenging to work with we addressed the to a mix of uninfected and infected cells, but efficacy of the inhibitor question of whether APP was functionally important in primary can be seen in the accumulation of α-and β-CTFs, as well as the human natural target cells. We were unable to test whether resulting decreases in p24 CA and infectious virus in culture primary human microglia express APP due to the enormous supernatants (Fig. 5g). Demonstrating that the antiviral activity of ethical and technical challenges involved in obtaining, let alone the γ-secretase inhibitor was mediated by APP, depletion of APP in working with such cells. It is also worth noting that the few CHME3 4 × 4 rescued HIV-1 spread and levels of p24 CA in culture studies that do use primary human brain cells acquire material supernatants in γ-secretase inhibitor-treated cultures, as determined during biopsy or autopsy, meaning cells are isolated under by either WB analysis or ELISA (Fig. 5j). This demonstrated that pathological or trauma/hypoxic conditions that would likely alter APP mediated the effects of secretase inhibitors on infection, and microglia biology in unpredictable ways. However, primary that APP’s antiviral activity could be harnessed using secretase rodent microglia do express APP , and we were able to confirm inhibitors to suppress HIV-1 replication. In addition, these findings APP expression and functionality in primary human peripheral demonstrated that HIV-1 Gag was sufficient to stimulate γ- blood mononuclear cells (PBMCs), the blood-stream counter- secretase-mediated APP processing, as a means to escape this parts of microglia that notably also enter the brain during HIV-1 35,36 restriction, offering a mechanistic explanation for elevated Aβ levels infection . Validating our overall findings, RNAi-mediated released from HIV-infected cells. depletion of APP in PBMCs infected with WT HIV-1 resulted in Although Aβ is elevated in HIV-1-infected patients, whether a statistically significant increase in the production of extra- Aβ production by infected cells in the brain contributes to cellular HIV-1 particles (Fig. 6j). Moreover, APP depletion also neurodegeneration remains unclear . To determine whether Gag- resulted in a decrease in Aβ production in infected PBMCs as induced Aβ production caused neurodegeneration, we developed determined by Congo Red staining (Fig. 6k), and resulted in a an assay using primary mouse cortical neurons. 293T cells were corresponding increase in neuronal viability compared to control transfected with APP together with either HIV-1 Gag or GAPDH siRNA-treated samples (Fig. 6l). in the presence of DMSO solvent control or γ-secretase inhibitor. Supernatants were then collected and clarified through 10 and 3 kDa cutoff filters to remove many other proteins that might also Discussion contribute to neuronal toxicity and confound data interpretation. Although APP has been implicated in various neuronal and When clarified supernatants were added to cultured cortical synaptic processes, its primary function remains unknown . APP NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 9 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 dimerization domains have been shown to interact with various affect Aβ clearance and uptake . While this may contribute to Aβ proteins, including binding to and activation of death receptor 6, overproduction, Tat is secretory and toxic, and proposed effects 5,7 resulting in axonal pruning and inhibition of synapse of Tat on Aβ remain unclear . It is also unclear why HIV-1 37–39 formation . Here, we identify a function for APP as an innate would evolve such a function and how this might benefit viral antiviral defense factor in macrophages and microglia that fitness, or whether it is simply a detrimental side effect. Here, we restricts HIV-1 release. By also identifying a viral evasion reveal that APP has a biological function to restrict HIV-1 release mechanism that leads to the production of neurotoxic Aβ42, our from brain-resident microglia and macrophages that carry HIV-1 findings address fundamental questions about how and why Aβ across the BBB, and Gag-mediated evasion of this restriction levels are elevated in the brains of HIV-1-infected patients, and results in Aβ production. Interestingly, primary rat microglia whether this contributes to neuronal damage. express APP but limit its processing . Gag-induced processing of Mammals have evolved a number of strategies to protect APP in HIV-1-infected microglia may contribute to the under- against retroviral infection, including the expression of antiviral lying differences in Aβ deposition patterns that have been proteins or restriction factors . These include Tetherin, which reported between AD and HAND, which have led to suggestions 41,42 prevents budding of new viral particles . Our findings reveal that HIV-1 alters Aβ metabolism in a manner that contributes 7,15 that APP also targets a late stage of the viral lifecycle required for to unique features of HAD and HAND . While this remains the production of extracellular virions, either through effects on to be explored further, our findings reveal the biological reason assembly, maturation, or release of new virus particles. APP acts for why HIV-1 causes Aβ overproduction, and provides direct by trapping the Gag polyprotein that is processed at the plasma evidence that elevated Aβ caused by Gag expression or HIV-1 23,25 membrane during viral maturation and release within lipid infection contributes to neurodegeneration. Although these rafts. Indeed, these membrane domains have been previously effects might appear small in the conventional sense of cyto- suggested to function in Gag maturation and virion toxicity, neurodegeneration caused by Aβ is a slow, gradual 29,43,44 budding , and have independently been shown to serve as process and the extent of effects from Gag-induced Aβ are in 45,46 51 regions of APP sorting and processing . APP is trafficked line with studies of amyloid in cultured neurons . Moreover, through ESCRT pathways via TSG101, directing APP to lyso- combined with inflammatory responses to infection and cyto- 47–49 somes and for processing into Aβ isoforms for secretion . toxicity of proteins like Tat, Gag-mediated APP processing Intriguingly, TSG101 is also a regulator of HIV-1 budding .As and production of Aβ would be an important contributing factor such, HIV-1 Gag and APP appear to have evolved to utilize the to the overall process of HIV-1-induced neurodegeneration. same subcellular compartments during their maturation. This is Intriguingly, a γ-secretase inhibitor not only prevented the clearly detrimental to HIV-1 production and release, and to evade increase in Aβ42 production induced by Gag and protected APP-mediated restriction, HIV-1 enhances APP processing. An against Gag-induced neurotoxicity, it also protected APP and early report suggested that HIV-1 protease activity leads to CTFs from degradation and suppressed HIV-1 replication in cleavage of APP , but this was not subsequently followed up. We microglia. There are a number of secretase inhibitors being find that APP processing induced by Gag does not require HIV-1 developed as potential therapeutics in AD, and our data suggest protease activity, which is encoded within the pol gene of the Gag/ that these may have the potential to serve dual purposes of both Pol fusion protein, as expression of Gag alone was sufficient to suppressing HIV-1 replication and preventing neuronal damage reduce APP expression (Figs. 4f and 5a, b). Instead, Gag induces by interfering with HIV-1’s attempts to evade this restriction APP processing through host secretases. While the molecular imposed by APP in brain-resident target cells. basis by which this is achieved remains to be determined, Gag might directly interact with secretases or act as a chaperone to Methods enhance APP processing. Alternatively, Gag might induce post- Cells. 293T, U87, NHDF, CHME3, and CHME3 4 × 4 cells were described pre- 21,52 viously . Peripheral blood mononuclear cells (PBMCs) were isolated from a translational modifications of APP. Indeed, around 10% of APP LifeSource Buffy Coat blood using Ficoll-Plaque (GE), and monocytes were isolated undergoes palmitoylation, which has been suggested to enhance from PBMCs using CD14 Microbeads (Miltenyi Biotec). Ethical approval for the amyloidogenic processing by targeting APP to membrane lipid study was obtained from the Institutional Review Board of Northwestern Uni- 28,46 rafts, promoting its β-secretase-mediated cleavage . While versity and all donors provided their written, informed consent. THP-1 cells were kindly provided by Thomas Hope. Primary mouse cortical neurons were purchased these aspects of the underlying mechanism remain to be deter- from Gibco (Cat # A15586). HeLa TZM-bl cells expressing CD4 and CCR5 as well mined, secretases clearly play a central role in this evasion as a LacZ and a luciferase reporter gene under control of the HIV-1 LTR (AIDS mechanism as inhibiting either β-or γ-secretase prevented fur- 26 Reagent Repository number 8129) were maintained in DMEM containing 10% ther processing of CTFs into Aβ isoforms. These protected CTFs, fetal bovine serum (FBS) and 1% Pen/Strep. and full-length APP, contain trans-membrane and cytosolic regions that most likely interfere with aspects of Gag processing Viruses and drugs. WT HIV-1 was generated by transfection of 293T cells with or assembly in lipid rafts. infectious clone pNL4-3 (AIDS Reagent Repository number 114). To generate HIV-1 carrying VSV-G envelope glycoprotein, pNL4-3.Luc.R−.E− plasmid (AIDS While APP likely plays a role in limiting virus spread by Reagent Repository number 3418) was transfected into 293T cells together with a macrophages in the blood, this process and the viral evasion VSV-G-expressing construct (pVSV-G) as described . γ-secretase inhibitor L- strategy is likely to be of particular importance in the brain. 685,458 (Cat # 2627) was purchased from Tocris. β-secretase (BACE1) inhibitor Although HIV-1 does not infect neurons, which lack appropriate Verubecestat (MK-8931, Cat # S8173) was purchased from Selleckchem. CHME3 4 × 4 or 293T cells were treated with DMSO or γ-secretase inhibitor or BACE1 receptors for virus entry, expression of APP in brain-resident inhibitor reconstituted in DMSO at 1 μM and/or 2.5 μM 4 h or 6 h post trans- microglia that become infected and in macrophages that carry fection or infection, respectively, and maintained throughout the entire experiment. 35,36 HIV-1 across the BBB would serve as a critical restriction to protect against HIV-1 spread in the brain. Thus, exploitation of Generation of expression constructs and viral vectors. For generation of host secretases not only provides a mechanistic basis for viral N-terminally Flag-tagged APP (pCAGOSF-APP ), APP (NM_201413; 770 770 evasion of this restriction but also explains increased Aβ pro- OriGene) was amplified from human cDNA using the sense 5′-CCCGGGAT GCTGCCCGGTTTGGCACTGC-3′ and the antisense 5′-GTCGACCTAGTT duction in infected cells. Although several HIV-1 proteins CTGCATCTGCTCAAAG-3′ primers. The restriction enzyme sites are shown in broadly cause inflammation and cytotoxicity, the onset of HAND bold. The PCR product was digested using SmaI and SalI and ligated into the has been shown to correlate with HIV-1-induced accumulation of pCAGOSF plasmid (DNASU), which was digested with SmaI and XhoI and host Aβ, which is also associated with AD. In attempts to explain contained an N-terminus One STrEP FLAG tag. For generation of the N-terminally this, studies have suggested that HIV-1 Tat may bind APP or GFP-tagged APP (N’-GFP-APP ), the APP cDNA from above was amplified 770 770 10 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE using the sense 5′-GCTAGCATGCTGCCCGGTTTGGCACTGCTCCTGCTG using 15 μl Turbofect (Thermo scientific). Total DNA amount in each group was GCC-3′ and the antisense 5′-GTCGACGTTCTGCATCTGCTCAAAGAACT kept constant with addition of empty vector pcDNA3.1 . Soluble cell extracts were TGTAGG-3′ primers. The restriction enzyme sites are shown in bold. The PCR prepared 48 h post-transfection as described and precleared with protein G- product was digested and cloned into the pEGFP-N1 expression vector (Clontech) sepharose. A concentration of 27 μl of the input samples was taken and the using NheI and SalI restriction enzyme sites. For generation of C-terminally remainder of the cell extract was incubated with 2 μl of the mouse anti-APP hemagglutinin (HA)-tagged GAPDH expression construct (pGAPDH-HA), antibody (Invitrogen, 130200) and protein G-sepharose for 1 h at 4 °C. Immune human GAPDH was amplified using cDNA from primary fibrobalsts and following complexes were then washed and boiled with Laemmli buffer and subjected to WB primers; forward primer, hGAPDH-S, 5′-GCAACTGCGGCCGCCATGGGGA analysis. For co-IP of endogenous APP with Gag, 1 × 10 CHME3 4 × 4 cells were AGGTGAAGGTCGGA-3′ and reverse primer, hGAPDH-A 5′-GCTTGAGGATC infected with WT HIV-1 or mock virus 48 h before cell lysates were subjected to CTTAAGCGTAATCTGGAACATCGTATGGGTACTCCTTGGAGGCCATG co-IP as described above. For GBP-binding assay, 293T cells (3 × 10 ) were TG-3′. The restriction enzyme sites are shown in bold, and the HA peptide transfected with 2 μg of a GFP-expressing control vector (GFP) or N′-GFP-APP sequence is underlined. The PCR product was then cloned into the expression (GFP-APP) along with 2 μg of HA-tagged forms of HIV-1 Gag (Gag-HA, Gag-N- vector pcDNA3.1 (Invitrogen). The inserts of all the expression constructs were Myr-HA or Gag-MA: -20-HA, -40-HA, -60-HA, and -80-HA) using 15 μl Tur- confirmed by sequencing. Expression constructs encoding C-terminally HA-tagged boFect (Thermo scientific). Two days post-transfection, cells were lysed with 1 ml Rev-independent HIV-1 Gag (Gag-HA), Matrix (MA-HA), Capsid (CA-HA) and cold NP-40 lysis buffer, subjected to GBP-binding assay followed by WB analysis as Gag-HA containing a single point mutation (Gag-MA-N-Myr) MA or serial described in ref. . G1A truncations of 20 aa in the C-terminus of MA (Gag-MA: -20-HA, -40-HA, -60-HA, and -80-HA) were described previously . Membrane flotation assay. A total of 1.5 × 10 293T cells were transfected with 3 μg of APP-Flag or pGAPDH-HA together with 1 μg Gag-HA or Gag-MA-60-HA Virion yield assays. A total of 1 × 10 293T cells were seeded in 60 mm dishes and followed by replacement of the media to fresh media 24 h post-transfection. The − − co-transfected with 0.4 μg of pNL4-3 or 0.7 μg of pNL4.3-luc.R E (Gag-Pol) along following day, cells were either lysed and subjected to WB analysis or collected and with increasing amounts of pCAGOSF-APP (APP-Flag) or either of the two subjected to membrane flotation assay as follows. The cells were washed with cold controls, pGAPDH-HA or empty vector pCAGOSF using Turbofect (Thermo 1×PBS and resuspended in cold 1×TE buffer with a Complete mini protease scientific). Total amount of DNA in each group was kept constant with addition of inhibitor tablet (Thermo Fisher). The cells were disrupted by Dounce homno- empty vector pcDNA3.1 or pCAGOSF. Following transfection, media was genizer with 10 pulses for 2 s/pulse. The cell lysates were then centrifuged at 500 × g replaced with fresh DMEM and supernatants were collected and filtered through a for 5 min at 4 °C to pellet the nuclei and unlysed cells. The resulting postnuclear 0.45 μm filter (Millipore) at 48 h post-transfection. The cells were then lysed and supernatant (PNS) was loaded under a discontinuous 10% to 75% sucrose gradient subjected to WB analysis for detection of Pr55 Gag, APP-Flag, GAPDH-HA and made in TNE buffer, and centrifuged to equilibrium in Beckman (Beckman the housekeeping proteins (PARP-1, β-tubulin and eIF4E) using antibodies specific Coulter, Optima XE-90) SW41Ti rotor overnight at 100,000 × g at 4 °C as described to each protein or to their tag as described below. Infectious virus yields were previously . After ultracentrifugation, 10 fractions were collected from the top to 26 4 determined by inoculating HeLa TZM-bl indicator cells seeded at 1.5 × 10 cells bottom of the density gradient. Equal volumes of samples from each fraction were in 96-well plates with 100 μl of serially diluted supernatants followed by mea- loaded onto an SDS-PAGE gel and subjected to WB analysis. For infected cell surements of beta-galactosidase activity 48 h post-infection using GalactoStar membrane flotation assays, 1 × 10 CHME3 4 × 4 cells were infected with WT reagent per manufacturer’s instructions (Life Tech). Physical particle yields were HIV-1 or mock virus 48 h before cell lysates were subjected to membrane flotation determined by WB analysis of virion containing supernatants either directly assay as described above. (supernatant from pNL4-3 transfected cells), or after pelleting the virion (super- − − natant from pNL4.3-luc.R E transfected cells) by passing the supernatants RNA interference. For transient knockdown, cells were transfected with siRNA through a 25% sucrose cushion by centrifugation at 100,000 g for 2 h at 4 °C, using anti HIV-1-Pr55/p24/p17 described below. For measurements of replication duplexes from Ambion using oligofectamine RNAiMAX (Invitrogen) as descri- bed . Briefly, cells were transfected with the control non-targeting siRNA (NC, competent HIV-1 virion yields, CHME3 4 × 4 cells were mock infected or infected with pNL4-3-derived HIV-1. Supernatants were collected and cells were lysed at ID# AM4635) or the APP-specific select siRNA duplexes (APP-V1 and APP-V2, days 3 and 5 post-infection followed by WB analysis as described above. ID# S1500 and S1501, respectively) each at 10 pmol. For western blotting and supernatant analysis, 5 × 10 CHME3 4 × 4 cells per well of 12-well plates were transfected with siRNAs. Twenty-four hours post-transfection, cells were trypsi- Western blotting. For WB analysis, cells were lysed in laemmli buffer and resolved nized and seeded at 2 × 10 cells per well on fresh 12-well plates. The following day on 10% SDS-PAGE gels. The levels of Aβ isoforms in supernatants were detected cells were infected with WT HIV-1, followed by collection of supernatants and cell by separating lysates on 4–12% Bis-tris Nu-PAGE gels (Invitrogen, NP0323BOX), lysis for measurements of infectious virions and intracellular protein levels, proteins were transferred to PVDF transfer membrane (Immobilon, IPVH00010) respectively, as described. For membrane flotation assays, two wells of a 6-well and blocked with 3% non-fat milk before incubating with primary antibodies. The plate, each containing 4 × 10 CHME3 cells, were transfected with siRNAs. uncropped western blots of all figures are shown in Supplementary Figs 6–13. Twenty-four hours post-transfection, both wells of CHME3 cells were trypsinized Antibodies used for WB were Flag (F7425) and HA (H3663) from Sigma; APP and pooled into a 10 cm dish. The following day, the cells were transfected with 1 (6E10, 803001) from Biolegend; HIV-1-Pr55/p24/p17 (ab63917) (labeled as anti- μg of Gag-HA or the empty vector pcDNA3.1 using 15 μl TurboFect (Thermo Pr55 Gag in the Fig.s), HIV-1-p24 (ab9071) (for detection of Pr55 Gag and/or p24 scientific). The cells were then collected and subjected to membrane flotation assay CA) and anti-APP antibody (Y188) (ab32136) from Abcam; eIF4E (610269) from as described above. For PBMCs, 1.5 × 10 PBMCs on a 60 mm dish were trans- BD Biosciences; GAPDH (sc-25778) from Santa Cruz; Aβ40 (PA3-16760) from fected with siRNA duplexes using oligofectamine RNAiMAX (Invitrogen) before Thermo Fisher; Aβ42 (700254) from Life Technologies; Caveolin-1 (D46G3), infection and analysis as described in figure legends and below. flotillin-1 (D2V27J), GFP (2555), PARP-1 (9542), Nicastrin (D38F9, 5665) and PEN2 (D6G8, 8598) from Cell Signaling. All primary antibodies were used at 1:1000 dilution and detected using the appropriate HRP-conjugated secondary Neurotoxicity assay. A total of 1 × 10 293T cells on a 60 mm dish were co- antibodies. transfected with 0.4 μg of APP-Flag and 1.6 μg of either the control pGAPDH-HA or Gag-HA using 8 μl TurboFect (Thermo scientific). Total DNA amount in each group was kept constant with addition of empty vector pcDNA3.1 . Alternatively, IF and congo red staining. For IF analysis, 4 × 10 293T or CHME3 cells grown on 5× 10 CHME3 4 × 4 cells on a 60 mm dish were infected with pNL4-3-derived glass coverslips in 6-well plate were co-transfected with equal amounts of HIV-1. Four hours post-transfection, or 1 day post-infection, the media were − − pCAGOSF-APP (APP-Flag) and Gag-HA or pNL4.3-luc.R E . Forty-eight replaced with fresh media without antibiotics containing either DMSO or 2.5 μMof hours post-transfection, cells were fixed with 3.7% paraformaldehyde, then blocked γ-secretase inhibitor L-685,458 (Tocris, Cat # 2627) reconstituted in DMSO. The and permeabilized with PBS supplemented with 0.1% Triton X-100 as described . following day, the media were replaced with neurobasal media (Gibco, 21103-049) Samples were incubated with anti-HIV-1-Pr55/p24/p17 (Abcam, ab63917) at 1:200 supplemented with B27 (Gibco, 17504-044) and Glutamax (Gibco, 35050-061) and anti-APP (LN27, Invitrogen, 130200) at 1:150 overnight at 4 °C. The next day, (2 ml/dish) containing either DMSO or 2.5 μMof γ-secretase inhibitor. 2 days samples were washed and incubated with the appropriate Alexa Fluor-conjugated post-transfection, supernatants were collected, filtered through 0.45 μm filters and secondary antibodies for 1 h at room temperature. Nuclei were stained with 1:3000 10 kDa Centrifugal Filter Unit (Abcam, ab93349), concentrated with Ultra Cen- Hoechst 33342. Images were acquired using a motorized spinning-disc confocal trifugal Filter concentrators (Millipore, UFC900308) prior to passing through a microscope (Leica DMI 6000B) with Yokogawa CSU-X1 A1 confocal head. For 3 kDa filter. A concentration of 2 μl of the concentrated supernatants from Congo Red staining, 10 μl supernatant was mixed with 200 μl Congo Red solution 293T cells were boiled in 2% SDS buffer for 10 min and subjected to ELISA analysis (Sigma) and incubated for 10 min at room temperature. Amyloid stained with for detection of Aß42 (Invitrogen, KHB3441) according to the manufacturer’s Congo Red was pelleted for 5 min at 18,407 g, the supernatant was discarded, the protocol. A total of 10 μl concentrated supernatants from CHME3 4 × 4 cells were pellet was then dissolved in 100 μl DMSO and the OD value of the solution was used for Congo Red staining as described above. For neurotoxicity assays, 0.5 × 10 determined at 490/650 nm. primary mouse cortical neurons (Life Technoogies)/well were plated in a poly-D- Lysine-coated 46-well plate and maintained in 500 μl of neurobasal medium sup- Co-IP and GBP-binding assay. For co-IP, CHME3 cells (3 × 10 ) were transfected plemented with 2% B-27 and 0.5 mM glutamax (Life Technologies). One half of the with 2 μg of APP-Flag or HA-tagged forms of HIV-1 Gag alone, or in combination medium was replaced with fresh medium every 3 days. On day 10, a concentration NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 11 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 of 200 μl of the concentrated in vitro supernatants from transfected 293T cells or 10. Willem, M. et al. eta-Secretase processing of APP inhibits neuronal activity in infected CHME3 4 × 4 cells described above were added onto the neurons while the hippocampus. Nature 526, 443–447 (2015). still in their own medium. It must be noted that, although supernatants were 11. Rempel, H. C. & Pulliam, L. HIV-1 Tat inhibits neprilysin and elevates amyloid initially concentrated as part of the clarification process, this served two additional beta. AIDS 19, 127–135 (2005). purposes. First, it allowed culture medium from transfected or infected cultures to 12. Andras, I. E. et al. Simvastatin protects against amyloid beta and HIV-1 Tat- be added to neuronal cultures without removing neuronal culture medium and induced promoter activities of inflammatory genes in brain endothelial cells. disturbing neuronal cultures. The final concentration of the added supernatants Mol. Pharmacol. 73, 1424–1433 (2008). after re-dilution upon addition to neuronal cultures was 3×. The second purpose of 13. Andras, I. E. et al. HIV-1-induced amyloid beta accumulation in brain this concentration was simply to collect all Aβ produced by cultured cells as the endothelial cells is attenuated by simvastatin. Mol. Cell. Neurosci. 43, 232–243 total amount produced is diluted into large volumes of culture medium, while (2010). in vivo neurons are directly exposed to factors secreted by neighboring microglia/ 14. Kurapati, K. R., Atluri, V. S., Samikkannu, T. & Nair, M. P. Ashwagandha macrophages. 3 days later the viability of the neurons was determined using Vita- (Withania somnifera) reverses beta-amyloid1-42 induced toxicity in human Orange Cell Viability Reagent (Biotool.com) following the suppliers protocol: to neuronal cells: implications in HIV-associated neurocognitive disorders each well 50 μl of the Vita-orange reagent was added and cells were incubated until (HAND). PLoS ONE 8, e77624 (2013). the orange color was developed. Wells without neurons, containing only the 15. Andras, I. E. & Toborek, M. Amyloid beta accumulation in HIV-1-infected medium, were used as a baseline. The results were recorded using a microplate brain: the role of the blood brain barrier. IUBMB. Life 65,43–49 (2013). reader, measuring the absorbance at 450 nm. Cell viability was calculated relative to 16. Braidy, N. et al. Recent rodent models for Alzheimer’s disease: clinical control group, treated with the supernatants from cells transfected with pGAPDH- implications and basic research. J Neural Transm 119, 173–195 (2012). HA or infected with mock virus, using the following formula: (AS-A0)/(AC-A0). 17. Achim, C. L. et al. Increased accumulation of intraneuronal amyloid beta in AS – sample absorbance; A0 – baseline absorbance, AC – control absorbance. For HIV-infected patients. J. Neuroimmune. Pharmacol. 4, 190–199 (2009). assays using PBMCs, 1.5 × 10 PBMCs on a 60mm dish were transfected with 18. Ances, B. M. et al. 11C-PiB imaging of human immunodeficiency virus- siRNA duplexes as described above. The following day, cells were infected with WT associated neurocognitive disorder. Arch. Neurol. 69,72–77 (2012). HIV-1 before supernatants were processed and used to perform neurotoxicity 19. Muller, U. C. & Zheng, H. Physiological functions of APP family proteins. Cold assays as described above. Spring Harb Perspect. Med. 2, a006288 (2012). 20. Malikov, V. et al. HIV-1 capsids bind and exploit the kinesin-1 adaptor FEZ1 Immuno-depletion of Aß. For immuno-depletion of Aβ, supernatants were col- for inward movement to the nucleus. Nat. Commun. 6, 6660 (2015). lected from HIV-1-infected CHME3 4 × 4 cells 48 h post infection, filtered through 21. Donahue, D. A. et al. SUN2 overexpression deforms nuclear shape and inhibits 0.45 μm filters and incubated at 4 °C for 2 h with 2 μl of anti-6E10 antibody bound HIV. J. Virol. 90, 4199–4214 (2016). to 15 μl of protein G-sepharose. The same amount of control mouse anti-HA 22. Sabo, Y. et al. HIV-1 induces the formation of stable microtubules to enhance antibody was used as a mock depletion. After incubation, beads were removed by early infection. Cell. Host. Microbe. 14, 535–546 (2013). centrifugation at 21,130 × g for 30 s. Beads and 20 µl of supernatant were boiled in 23. Sundquist, W. I. & Krausslich, H. G. HIV-1 assembly, budding, and laemmli buffer and subjected to WB analysis. maturation. Cold Spring Harb Perspect. Med. 2, a006924 (2012). 24. Dick, R. A. & Vogt, V. M. Membrane interaction of retroviral Gag proteins. Statistical analyses. Statistical analyses for two groups were performed in Front. Microbiol. 5, 187 (2014). Microsoft Excel 2016 using two-tailed Student’s t test. Statistical analyses for three 25. Freed, E. O. HIV-1 assembly, release and maturation. Nat. Rev. Microbiol. 13, or more groups were performed using Astatsa online statistical calculator. One-way 484–496 (2015). analysis of variance (ANOVA) was followed by Tukey’s test for post-hoc com- 26. Platt, E. J., Wehrly, K., Kuhmann, S. E., Chesebro, B. & Kabat, D. Effects of parisons of three or more experimental groups (only when ANOVA was sig- CCR5 and CD4 cell surface concentrations on infections by macrophagetropic nificant). Homogeneity of variance was confirmed with Levene’s test for equality of isolates of human immunodeficiency virus type 1. J. Virol. 72, 2855–2864 variances. To quantify co-localization of APP and Gag, the Pearson’s correlation (1998). coefficient (R ) was used. The samples were analyzed by using Fiji-ImageJ-NIH. R r r 27. Ono, A. & Freed, E. O. Binding of human immunodeficiency virus type 1 Gag ranges between −1 (perfect negative correlation) to +1 (perfect positive correlation) to membrane: role of the matrix amino terminus. J. Virol. 73, 4136–4144 with 0 meaning no correlation. All the data are expressed as mean ± SEM. The P (1999). value for all cases was set to < 0.05 for significant differences. *P < 0.05, **P < 0.01, 28. Simons, K. & Toomre, D. Lipid rafts and signal transduction. Nat. Rev. Mol. ***P < 0.001, ****P < 0.0001. Cell. Biol. 1,31–39 (2000). 29. Ono, A. & Freed, E. O. Plasma membrane rafts play a critical role in HIV-1 Data availability. The authors declare that the data supporting the findings of this assembly and release. Proc. Natl Acad. Sci. USA 98, 13925–13930 (2001). study are available within the article and its Supplementary Information files, or are 30. Lee, S. J. et al. A detergent-insoluble membrane compartment contains A beta available from the authors upon request. in vivo. Nat. Med. 4, 730–734 (1998). 31. Riddell, D. R., Christie, G., Hussain, I. & Dingwall, C. Compartmentalization of beta-secretase (Asp2) into low-buoyant density, noncaveolar lipid rafts. Curr. Received: 25 April 2017 Accepted: 16 October 2017 Biol. 11, 1288–1293 (2001). 32. Wahrle, S. et al. Cholesterol-dependent gamma-secretase activity in buoyant cholesterol-rich membrane microdomains. Neurobiol. Dis. 9,11–23 (2002). 33. Simons, K. & Vaz, W. L. Model systems, lipid rafts, and cell membranes. Annu. Rev. Biophys. Biomol. Struct. 33, 269–295 (2004). 34. Haass, C., Hung, A. Y. & Selkoe, D. J. Processing of beta-amyloid precursor References protein in microglia and astrocytes favors an internal localization over 1. Hellmuth, J., Milanini, B. & Valcour, V. Interactions between ageing and constitutive secretion. J. Neurosci. 11, 3783–3793 (1991). NeuroAIDS. Curr. Opin. HIV AIDS 9, 527–532 (2014). 35. Ivey, N. S., MacLean, A. G. & Lackner, A. A. Acquired immunodeficiency 2. Nightingale, S. et al. Controversies in HIV-associated neurocognitive disorders. syndrome and the blood-brain barrier. J. Neurovirol. 15, 111–122 (2009). Lancet. Neurol. 13, 1139–1151 (2014). 36. Rahimy, E. et al. Blood-brain barrier disruption is initiated during primary HIV 3. Rappaport, J. & Volsky, D. J. Role of the macrophage in HIV-associated infection and not rapidly altered by antiretroviral therapy. J. Infect. Dis. 215, neurocognitive disorders and other comorbidities in patients on effective 1132–1140 (2017). antiretroviral treatment. J. Neurovirol. 21, 235–241 (2015). 37. Nikolaev, A., McLaughlin, T., O’Leary, D. D. & Tessier-Lavigne, M. APP binds 4. Clifford, D. B. & Ances, B. M. HIV-associated neurocognitive disorder. Lancet DR6 to trigger axon pruning and neuron death via distinct caspases. Nature Infect. Dis. 13, 976–986 (2013). 457, 981–989 (2009). 5. Gonzalez-Scarano, F. & Martin-Garcia, J. The neuropathogenesis of AIDS. Nat. 38. Olsen, O. et al. Genetic analysis reveals that amyloid precursor protein and Rev. Immunol. 5,69–81 (2005). death receptor 6 function in the same pathway to control axonal pruning 6. Xu, Y. et al. HIV-1-mediated apoptosis of neuronal cells: Proximal molecular independent of beta-secretase. J. Neurosci. 34, 6438–6447 (2014). mechanisms of HIV-1-induced encephalopathy. Proc. Natl Acad. Sci. USA 101, 39. Xu, K., Olsen, O., Tzvetkova-Robev, D., Tessier-Lavigne, M. & Nikolov, D. B. 7070–7075 (2004). The crystal structure of DR6 in complex with the amyloid precursor protein 7. Ortega, M. & Ances, B. M. Role of HIV in amyloid metabolism. J. provides insight into death receptor activation. Genes Dev. 29, 785–790 (2015). Neuroimmune. Pharmacol. 9, 483–491 (2014). 40. Simon, V., Bloch, N. & Landau, N. R. Intrinsic host restrictions to HIV-1 and 8. van der Kant, R. & Goldstein, L. S. Cellular functions of the amyloid precursor mechanisms of viral escape. Nat. Immunol. 16, 546–553 (2015). protein from development to dementia. Dev. Cell. 32, 502–515 (2015). 41. Neil, S. J., Zang, T. & Bieniasz, P. D. Tetherin inhibits retrovirus release and is 9. Sevigny, J. et al. The antibody aducanumab reduces Abeta plaques in antagonized by HIV-1 Vpu. Nature 451, 425–430 (2008). Alzheimer’s disease. Nature 537,50–56 (2016). 12 NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01795-8 ARTICLE P01GM105536 to D.W. and grant R01GM101975, P01GM105536 and R01NS099064 to 42. Van Damme, N. et al. The interferon-induced protein BST-2 restricts HIV-1 M.H.N. and with help from the Third Coast Center for AIDS Research (cFAR), an NIH release and is downregulated from the cell surface by the viral Vpu protein. Cell funded center (P30AI117943). Host Microbe. 3, 245–252 (2008). 43. Brugger, B. et al. The HIV lipidome: a raft with an unusual composition. Proc. Natl Acad. Sci. USA 103, 2641–2646 (2006). Author contributions 44. Rousso, I., Mixon, M. B., Chen, B. K. & Kim, P. S. Palmitoylation of the HIV-1 Q.C., V.J., V.M., Y.S. and S.M., performed the research; Q.C., V.J., V.M., Y.S., S.M., envelope glycoprotein is critical for viral infectivity. Proc. Natl Acad. Sci. USA D.W. and M.H.N. designed the research; Q.C., V.J., D.W. and M.H.N. analyzed the data; 97, 13523–13525 (2000). M.H.N. wrote the manuscript, and D.W. edited the manuscript. 45. Haass, C., Kaether, C., Thinakaran, G. & Sisodia, S. Trafficking and proteolytic processing of APP. Cold Spring Harb Perspect. Med. 2, a006270 (2012). Additional information 46. Bhattacharyya, R., Barren, C. & Kovacs, D. M. Palmitoylation of amyloid precursor protein regulates amyloidogenic processing in lipid rafts. J. Neurosci. Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- 33, 11169–11183 (2013). 017-01795-8. 47. Choy, R. W., Cheng, Z. & Schekman, R. Amyloid precursor protein (APP) traffics from the cell surface via endosomes for amyloid beta (Abeta) Competing interests: The authors declare no competing financial interests. production in the trans-Golgi network. Proc. Natl Acad. Sci. USA 109, Reprints and permission information is available online at http://npg.nature.com/ E2077–E2082 (2012). reprintsandpermissions/ 48. Morel, E. et al. Phosphatidylinositol-3-phosphate regulates sorting and processing of amyloid precursor protein through the endosomal system. Nat. Commun. 4, 2250 (2013). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 49. Edgar, J. R., Willen, K., Gouras, G. K. & Futter, C. E. ESCRTs regulate amyloid precursor protein sorting in multivesicular bodies and intracellular amyloid- beta accumulation. J. Cell. Sci. 128, 2520–2528 (2015). 50. Tomasselli, A. G. et al. Actin, troponin C, Alzheimer amyloid precursor protein Open Access This article is licensed under a Creative Commons and pro-interleukin 1 beta as substrates of the protease from human immunodeficiency virus. J. Biol. Chem. 266, 14548–14553 (1991). Attribution 4.0 International License, which permits use, sharing, 51. Wang, Y., Liu, L., Hu, W. & Li, G. Mechanism of soluble beta-amyloid 25–35 adaptation, distribution and reproduction in any medium or format, as long as you give neurotoxicity in primary cultured rat cortical neurons. Neurosci. Lett. 618, appropriate credit to the original author(s) and the source, provide a link to the Creative 72–76 (2016). Commons license, and indicate if changes were made. The images or other third party 52. Haedicke, J., Brown, C. & Naghavi, M. H. The brain-specific factor FEZ1 is a material in this article are included in the article’s Creative Commons license, unless determinant of neuronal susceptibility to HIV-1 infection. Proc. Natl Acad. Sci. indicated otherwise in a credit line to the material. If material is not included in the USA 106, 14040–14045 (2009). article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Acknowledgements licenses/by/4.0/. We thank Olivier Schwartz, Thomas Hope and Robert Vassar for reagents and advice and Katherine Sadleir and Kayla Schipper for technical help. The following reagent was obtained through the NIH AIDS Research and Reference Reagent Program, Division of © The Author(s) 2017 − − AIDS, NIAID, NIH: pNL4-3 from Dr. Malcom Martin, pNL4-3.Luc.R .E from Dr. Nathaniel Landau, and TZM-bl from Dr. John C. Kappes, Dr. Xiaoyun Wu and Tran- zyme Inc. This study was supported by National Institute of Health (NIH) grant NATURE COMMUNICATIONS 8: 1522 DOI: 10.1038/s41467-017-01795-8 www.nature.com/naturecommunications 13 | | |

Journal

Nature CommunicationsSpringer Journals

Published: Nov 15, 2017

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off