Stimulation of Enveloped Virus Infection by β-Amyloid Fibrils*

Alzheimer's disease is characterized by deposition of β-amyloid peptide (Aβ) into plaques in the brain, leading to neuronal toxicity and dementia. Human immunodeficiency virus type 1 (HIV-1) infection of the central nervous system can also cause a dementia, and amyloid deposition in the central nervous system is significantly higher in HIV-1-infected individuals compared with uninfected controls. Here we report that Aβ fibrils stimulated, by 5–20-fold, infection of target cells expressing CD4 and an appropriate coreceptor by multiple HIV-1 isolates but did not permit infection of cells lacking these receptors. Aβ enhanced infection at the stage of virus attachment or entry into the cell. Aβ fibrils also stimulated infection by amphotrophic Moloney leukemia virus, herpes simplex virus, and viruses pseudotyped with the envelope glycoprotein of vesicular stomatitis virus. Other synthetic fibril-forming peptides similarly enhanced viral infection and may be useful in gene delivery applications utilizing retroviral vectors. These data suggest that Aβ deposition may increase the vulnerability of the central nervous system to enveloped viral infection and that amyloidogenic peptides could be useful in enhancing gene transfer by enveloped viral vectors.

The efficiency of gene delivery using retroviral vectors is often a limiting factor in attempts to express exogenous genes in both cultured mammalian cells and in vivo. Facilitators such as Polybrene (hexadimethrine bromide) (1) and DEAE-dextran (2) have been utilized to increase the efficiency of viral infection. Vectors pseudotyped with the envelope glycoproteins of various viruses are advantageous for targeting exogenous genes to specific cell types that express the cognate receptor molecules. Envelope glycoproteins of vesicular stomatitis virus (VSV) 1 and amphotropic murine leukemia virus (A-MuLV) utilize ubiquitously expressed receptors and are useful for trans-duction of varied cell types (3,4). Envelope glycoproteins that require cell type-specific receptors, such as the gp120-gp41 of human immunodeficiency virus type 1 (HIV-1), can provide a useful tool for targeting exogenous genes to specific cell types. HIV-1 envelope glycoproteins facilitate the fusion of viral and cellular membranes through sequential binding of CD4 and a chemokine receptor, principally CCR5 or CXCR4 (5)(6)(7).
The accumulation of A␤  and A␤ 1-42 proteolytic fragments of Amyloid ␤ precursor protein (APP) is a molecular marker characteristic of Alzheimer's disease (AD) (8 -10).
HIV-1 infection of microglia in the central nervous system leads to HIV-associated dementia (HAD) in ϳ20 -30% of latestage AIDS patients (11), and HIV-1 replication in the brain has been observed to colocalize with sites of APP accumulation (12,13). The occurrence of APP-rich lesions coincides with the presence of HAD (12).
HIV-1 infection of the central nervous system and subsequent infection in the brain occur by mechanisms that remain poorly understood. It is believed that infection of macrophages, microglia, and possibly astrocytes leads to indirect neuronal injury and death, providing the basis for the development of HAD, a syndrome of cognitive and motor dysfunction diagnostically similar to AD-related dementia (11,14,15). A positive relationship between cerebrospinal fluid viral load and the extent to which patients with HAD or minor cognitive/motor disorder experience cognitive dysfunction has been described (16,17). A␤-rich neuritic plaques are also observed to occur with greater prevalence in HIV-1-infected individuals compared with uninfected controls, although an etiological relationship between HAD and plaques has not been established (18). Additionally, HIV-1-infected individuals bearing the ApoE4 allele, a genetic risk factor for AD that correlates with elevated A␤ levels (19), are twice as likely to be demented or have peripheral neuropathy as individuals lacking this allele (20). ApoE4 is an AD susceptibility factor, particularly for individuals harboring herpes simplex virus (HSV) in the brain (21,22). HIV-1 infection of the central nervous system, systemic immune suppression, and increased permeability of the blood brain barrier (11) promote opportunistic HSV (23) and cytomegalovirus infection (24).
Whether the proteolytic fragments of APP, a common molecular marker of dementing disease states including AD, are implicated in the mechanism of HIV-1 brain infection remains unclear. We wished to address whether a relationship exists between the presence of APP proteolytic fragments and HIV-1 infection. Here we report that the amyloidogenic APP fragments A␤ 1-40 and A␤  , as well as other synthetic amyloidogenic peptides, significantly enhanced infection by HIV-1 and viruses with other envelope glycoproteins. The effect was stronger than the enhancement of infection observed using Polybrene. These findings are suggestive of a model that may ex-plain how neuritic damage caused by HIV-1 infection in the brain and subsequent A␤ deposition induced by this damage may facilitate further HIV-1 infection. Additionally, they suggest a use for synthetic amyloidogenic peptides in both laboratory and clinical viral delivery systems.
Recombinant Reporter Viruses-Recombinant HIV-1 reporter viruses were constructed by cotransfection of 293T human embryonal kidney cells with vectors expressing the pCMV⌬P1⌬envpA HIV-1 Gag-Pol packaging construct (29), the envelope glycoproteins of A-MLV, VSV, and HIV-1 isolates (ADA, YU2, JR-FL, and HXBc2), and a reporter gene at a DNA weight ratio of 1:1:3 using Effectene reagents (Qiagen). Cotransfection produced replication-defective (single-round) virions capable of expressing HIV-1 tat and the firefly luciferase gene under control of the HIV-1 long terminal repeat or the green fluorescent protein (GFP) gene under control of the cytomegalovirus immediateearly promoter. Viruses pseudotyped with VSV-G, A-MuLV, and HIV-1 envelope glycoproteins were produced by cotransfecting the pHCMV-G (30), SV-A-MLV-Env (31), or pSVIIIenv (32-35) plasmids, respectively. Production of the VSV-G and A-MuLV recombinant viruses also required cotransfection of pCMV-Rev, a plasmid expressing the HIV-1 Rev protein (36). Thirty h after transfection, the virus-containing cell supernatants were harvested, filtered (0.45 m), and aliquoted, and they were kept frozen until use. The reverse transcriptase activities of all viruses were quantified and normalized by cpm as described previously (37). Replication-deficient A-MuLV (Retropack; Clonetech) and HSV (HD-2) (38) vectors containing ␤-galactosidase reporter genes were produced using NIH-3T3 cells, according to the manufacturer's protocol; infection efficiencies were estimated by reporter gene activity in the target cells. Luciferase and ␤-galactosidase activity was quantitated as described in Promega protocols using an EG&G Berthold Microplate Luminometer LB 96V.
Infection by Single-round Viruses Expressing Luciferase-Target cells for viral entry were seeded in 96-well luminometer-compatible tissue culture plates (Dynex) at a density of 6 ϫ 10 3 cells/well and incubated for 24 h. The medium was removed from the target cells and replaced with fresh complete Dulbecco's modified Eagle's medium containing reverse transcriptase-normalized units of recombinant virus. The amounts of virus varied, depending upon the envelope glycoproteins used for pseudotyping: VSV-G, 1,000 cpm; A-MuLV, 30,000 cpm; and ADA, YU2, JR-FL, 89.6, ADA-⌬ V1/V2, and HXBc2 HIV-1 envelope glycoproteins, 10,000 cpm). Varying amounts of A␤ 1-40 , A␤ 1-42 (1.25-20 M), PPI-2566, or PPI-2480 (1-100 M) were added with the recombinant viruses to a final infection volume of 50 l. The anti-CCR5 antibody 2D7 (39) (BD PharMingen) or TAK-779, a low molecular weight nonpeptide compound that specifically binds CCR5 (40) (Takeda Chemical Industries, Ltd.), was also included in some assays. Target cells were incubated with the infection medium for 48 h. After this incubation, the medium was aspirated from each well, and the cells were lysed by the addition of 30 l of passive lysis buffer (Promega), agitation, and two freeze-thaw cycles. The luciferase activity of each well was measured for 10 s after the addition of 100 l of luciferase buffer (15 mM MgSO 4 , 15 mM KPO 4 , pH 7.8, 1 mM ATP, and 1 mM dithiothreitol) and 50 l of 1 mM D-luciferin potassium salt (BD PharMingen) using an EG&G Berthold Microplate Luminometer LB 96V.
Infection by Single-round Viruses Expressing GFP-SupT1-CCR5 target cells were seeded in 24-well tissue culture plates (Falcon) at a density of 5 ϫ 10 4 cells/well with medium containing reverse transcriptase-normalized units of GFP-expressing recombinant virus (VSV-G, 3,000 cpm; ADA, 150,000 cpm; YU2, 150,000 cpm) and varying amounts of A␤ 1-40 or A␤ 1-42 (62.5 nM to 1 M) in a final volume of 0.4 ml. The infection medium-cell mixture was incubated for 48 h, 1 ml of fresh complete RPMI 1640 was added to each well, and the cells were incubated for an additional 24 h. The cells were then harvested, washed with phosphate-buffered saline, fixed in 10% formalin, and analyzed by fluorescence-activated cell sorting using a Becton Dickinson FACScan with CellQuest software.
Infection by Single-round Viruses Expressing ␤-Galactosidase-Cf2Th cells were infected with an A-MuLV vector expressing ␤-galactosidase without additives, in the presence of 8 g/ml Polybrene or in the presence of 10 M preaggregated A␤ 40 -1 reverse fragment or A␤  . The precipitable fraction of A␤ 1-40 was recovered by pelleting preaggregated A␤ 1-40 at 15,000 ϫ g for 5 min at 4°C, after which the supernatant was removed and retained. The precipitated peptide fibrils were resuspended in phosphate-buffered saline, washed two more times, and resuspended in the starting volume. The ␤-galactosidase expression in the target cells 24 h after infection was estimated using a chemiluminescence assay (Galacto-Star; Tropix, Inc.). Cf2Th cells were also infected with a single-round HSV virus vector (HD-2) (38) containing the ␤-galactosidase reporter gene in the presence of 5 or 10 M preaggregated A␤  . Cells were stained according to the Promega protocol and counted under the microscope 24 h after infection.
Fluorescence-activated Cell-sorting Analysis of Fibril Interactions with Liposomes and Cells-Unilamellar small liposomes (liposomes) similar in size to HIV-1 were prepared from a 2:1 (M/M) mixture of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine and 1-palmitoyl-2oleoyl-sn-glycero-3-phosphoethanolamine supplemented with 1% fluorescent rhodamine-lissamine B-phosphatidylethanolamine (Avanti Polar Lipids), as reported previously (41). Rhodamine-labeled liposomes were incubated with Cf2Th cells in the presence and absence of peptide fibrils in the medium used for the viral entry assay (Dulbecco's modified Eagle's medium ϩ 10% fetal bovine serum) supplemented with 0.02% NaN 3 for 1 h at 37°C. Similarly, fluorescent A␤ 1-40 fibrils (FITC-A␤) were incubated with Cf2Th cells either lacking or expressing CD4 and/or CCR5, in the absence and presence of recombinant HIV-1 gp120 envelope glycoprotein from the JR-FL isolate. After incubation, the cells were washed with phosphate-buffered saline containing 2% bovine serum albumin, and the association of rhodamine-labeled liposomes or FITC-A␤ with cells was analyzed using FACScan, as described above. Cf2Th-CD4/CCR5 cells were also incubated without additives or with unlabeled preaggregated A␤ 1-40 , as described above, and the CD4 and CCR5 cell surface expression was detected using the anti-CD4 antibody RPA-T4-PE (BD PharMingen) and the anti-CCR5 antibody 2D7-PE (BD PharMingen) at a final concentration of 10 nM.  Enhance Signal of Viral Entry-To investigate whether the presence of A␤ affects HIV-1 infection of target cells, recombinant replication-defective HIV-1 vectors expressing firefly luciferase or GFP were used. These singleround viruses were pseudotyped with the envelope glycoproteins of various HIV-1 isolates or with those of VSV or A-MuLV. The receptors for VSV and A-MuLV are ubiquitously expressed. Entry of viruses pseudotyped with the HIV-1 envelope glycoproteins is dependent on the presence of CD4 or a chemokine receptor, CCR5 or CXCR4; the viruses pseudotyped with the VSV or A-MuLV envelope glycoproteins do not require CD4 or chemokine receptor expression on the target cells. Preaggregated A␤ 1-40 and A␤ 1-42 fibrils dramatically increased infection of Cf2Th-CD4/CCR5 cells by HIV-1 pseudotyped with the envelope glycoproteins of three CCR5-using primary HIV-1 isolates (ADA, YU2, and JR-FL) in a dose-dependent manner (Fig. 1). A␤ 1-40 similarly increased infection of GHOST(3)-CD4/ CXCR4 cells by HIV-1 pseudotyped with the envelope glycoproteins of the CXCR4-using isolate, HXBc2 (Fig. 1). Similar results were obtained by infecting a human T-lymphocyte cell line stably expressing CCR5 (SupT1-CCR5) with GFP-expressing viruses pseudotyped with ADA and YU2 envelope glycoproteins (data not shown). A␤ 1-40 was more potent than A␤  and increased the entry of viruses by 2-10 times in a concentration range of 1-5 M and by 5-30 times at a concentration of 20 M. Infection of cells by viruses pseudotyped with the A-MuLV or VSV envelope glycoproteins was also enhanced. The relatively lower enhancement observed with VSV-Gpseudotyped virus may be due to the substantially greater efficiency with which this virus infects cells in the absence of A␤. These data show that A␤ can substantially increase the efficiency of infection of cells by HIV-1 pseudotyped with the envelope glycoproteins of a wide range of HIV-1 isolates, as well as with those of other enveloped viruses.

A␤ 1-40 and A␤
A␤  Enhances an Early Step in Virus Infection-To elucidate whether the enhancement of virus infection by A␤ was mediated by an increased efficiency of early or late events in the virus life cycle, incubation of the recombinant viruses with the Cf2Th-CD4/CCR5 target cells was carried out for only 4 h, followed by washing. The target cells were incubated with 20 M A␤ 1-40 concurrently with virus (ϩ/Ϫ), immediately after virus removal (Ϫ/ϩ), or throughout both time periods (ϩ/ϩ) (Fig. 2). After the wash, the cells were incubated for an additional 48 h, at which time luciferase activity was measured. Enhancement of infection was observed only when A␤ 1-40 was present during the initial 4-h incubation of virus and cells. These data suggest that A␤ exerts its effect at an early stage of viral infection. Because the first 4 h of HIV-1 infection involves virus attachment and entry into the host cell, A␤ likely enhances these processes.

Enhancement of Viral Infection by A␤ Is
Receptor-mediated-A␤ has been shown to exert a destabilizing effect on cellular membranes (42,43). Therefore, A␤ might facilitate fusion of the target cell and viral membrane in a manner that would circumvent the dependence of the virus on its receptors. To investigate this possibility, we examined infection of Cf2Th, Cf2Th-CD4, Cf2Th-CCR5, and Cf2Th-CD4/CCR5 cells by CCR5-dependent HIV-1 isolates. No infection by CCR5-dependent HIV-1 isolates was observed in the presence or absence of A␤ with cells lacking CD4 and/or CCR5 (Fig. 3a), whereas infection by viruses pseudotyped by VSV and A-MuLV envelope glycoproteins, which do not require these cellular receptors, was enhanced by A␤ on all cells examined. A␤-enhanced CCR5dependent viral entry remained sensitive to inhibition by CCR5 ligands, including the 2D7 antibody (39) (Fig. 3b) and the small-molecule antagonist TAK-779 (40) (data not shown). These data demonstrate that HIV-1 infection in the presence of A␤ remains dependent on the expression of CD4 and a chemokine coreceptor.
Other Fibril-forming Peptides Enhance Viral Infection-A␤ aggregates into fibrils (8, 42, 44 -46). We investigated whether other fibril-forming peptides unrelated to A␤ could enhance virus infection. Fig. 4 shows that two such peptides, PPI-2480 (AGAKWSWWELTWVGG) and PPI-2566 (IRQAMCNISRAD-WND), which form fibrils similar to A␤ 1-40 and A␤ 1-42 (Fig. 5,  bϪf), also enhanced the infection efficiency of recombinant HIV-1 pseudotyped with the envelope glycoproteins of the ADA and YU2 HIV-1 isolates by 5-20-fold. The stimulation by these fibrils also required the expression of viral entry coreceptors (data not shown). These compounds enhanced infection of HIV-1 pseudotyped with the VSV-G protein by ϳ2-fold. A number of control peptides of varying sequences and lengths that did not form fibrils had no effect on HIV-1 infection. An example is the peptide PPI-1966 shown in Figs. 4 and 5f. These data demonstrate that the ability of a peptide to enhance viral infection correlates with its propensity to form fibrils in solution. Interestingly, the peptides that most potently enhance infection (A␤ 1-40 and PPI-2480) formed shorter fibrils (Fig. 5, b   FIG. 1. A␤ stimulates infection by recombinant HIV-1 (Fig. 5a). Each of the fibrilforming peptides that enhanced infection also promoted irreversible association of liposomes with cells. The peptides did not cause the formation of syncytia, nor did they promote liposome-to-cell fusion, as judged by the failure of the rhodamine dye in the liposomes to distribute into the cell membrane (data not shown). Consistent with their relative ability to enhance infection, A␤ 1-40 promoted the adherence of liposomes better than A␤  . PPI-1966, which Fig. 5f shows cannot form fibrils, had no effect on the association of liposomes with cells (data not shown). Utilizing fluorescent A␤ 1-40 fibrils (FITC-A␤), we found that FITC-A␤ associated with cell surfaces independent of CD4 or CCR5 expression (data not shown). The presence of recombinant HIV-1 envelope glycoprotein (JR-FL gp120) in the medium did not promote the association of FITC-A␤ with cell membranes (data not shown). Additionally, the presence of A␤ 1-40 did not induce changes in cell surface expression of CD4 or CCR5 (data not shown). These data support a model in which A␤ and other fibril-forming peptides enhance viral infection by mediating a physical association of viral envelopes with the cell lipid bilayer. Figs. 1-3, infection by HIV-1 pseudotyped with the envelope glycoprotein of A-MuLV was enhanced by A␤. Infection by complete A-MuLV was also strikingly enhanced, from 30 -50fold, in the presence of preaggregated Ab 1-40 (Fig. 6a). This effect was 2-3-fold greater than that observed for Polybrene, a cationic polymer commonly used to increase the efficiency of retroviral gene delivery systems (48). In this experiment, Ab 1-40 fibrils were precipitated by multiple centrifugation and washing steps and compared with the supernatant of the first centrifugation. Fig. 6a demonstrates that the precipitable Ab 1-40 fraction, but not any residual soluble peptide, enhanced A-MuLV infection comparably to Ab 1-40 that had not been centrifuged. Conversely, the Ab 40 -1 reverse fragment did not enhance the infection efficiency of A-MuLV. These data underscore the substantial enhancement of retroviral infectivity by Ab 1-40 and demonstrate that the precipitable, and presumably fibril-forming, fraction of A␤ mediates its ability to enhance infection.

A␤ Weakly Stimulates Infection by an Enveloped Virus
Other than a Retrovirus-Because HSV has been suggested to play a role in AD and is a major opportunistic infection observed in late-stage HIV-1 infection, we investigated the ability of A␤ to enhance HSV infection. A dose-dependent enhancement of the infection mediated by an HSV vector was observed (Fig. 6b). However, relative to the enhancement observed with retroviruses, A␤ 1-40 was substantially less efficient in enhancing HSV infection. This less pronounced ability of A␤ to enhance HSV infection may be a consequence of differences in accessibility or composition of the HSV lipid membrane.

DISCUSSION
Here we describe an enhancement of enveloped virus infection by amyloidogenic APP proteolytic fragments A␤ 1-40 and A␤  . The requirement that the A␤ fragments be present during the contact of the virus with the target cell suggests that a very early phase of infection is stimulated by the peptides. Enhancement of infection was observed for viruses containing several different envelope glycoproteins that utilize unrelated receptors, suggesting that enhancement does not require specific protein-protein interactions. Consistent with this, these peptides substantially enhanced the association of liposomes with cells. A common element among the viruses assayed in this study is the presence of a lipid envelope bilayer. It is therefore likely that the mechanism by which these peptides enhance entry includes their propensity to promote an interaction between the viral and cellular lipid membranes. The extent to which the reported membrane-destabilizing properties of A␤ participate in the observed enhancement of viral fusion remains unclear. The requirement for appropriate receptors on the target cell is not bypassed by A␤, suggesting that receptortriggered changes in the envelope glycoproteins are still crucial for achieving the fusion of the viral and target cell membranes.
The entry enhancement observed herein is mediated by the precipitable amyloidogenic fraction of A␤. Interestingly, other synthetic amyloidogenic peptides unrelated to A␤ similarly enhanced viral infection, whereas synthetic nonamyloidogenic peptides had no entry-enhancing effect. These results suggest that fibril formation may be important for the viral enhancement effect. Additional studies will be required to determine the other properties of amyloidogenic peptides that contribute to enhancement of virus infection.
Our observations could have relevance to neuropathogenesis. Neuritic plaques, a primary component of which is A␤, are more detectable in HIV-1-infected individuals than uninfected individuals (18). Additionally, HIV-1-infected individuals are  prone to an HIV-associated dementia that is correlated with high viral loads in the cerebrospinal fluid (16,17). Immune cells, in particular microglia and macrophages, which are important target cells of HIV-1 in the brain, are commonly recruited to neuritic plaques (11). Our observations suggest that regions of high A␤, such as those in the vicinity of plaques, would be a highly favorable environment for virus transmission. It has been observed that sites of HIV-1 replication in the brain colocalize with sites of APP accumulation (12,13), a possible consequence of HIV-1-induced neuronal injury. If high local APP levels also result in the production of A␤, neuronal injury may both recruit immune cells and promote their infection. The observation of more frequent and severe HAD in individuals bearing the ApoE4 allele (20) is also consistent with a role for A␤ in HAD. Taken together with the data herein, these observations suggest that testing the effect of inhibitors of A␤ production in primate models of HAD (49) is warranted.
The infection of viruses pseudotyped with the envelope glycoproteins of VSV, A-MuLV, HIV-1, and HSV was enhanced by A␤. It has been reported that HSV is detectable in a greater percentage of AD patients than in age-matched controls and that the combination of HSV and the ApoE4 allele disposes individuals to AD more than either factor alone (21,22). These observations are consistent with a contribution of HSV to the pathogenesis of AD. Although the enhancement of HSV by A␤ was significantly less pronounced than that observed for the two retroviruses studied, the possibility that an enveloped virus contributes to AD pathology merits further study.
The magnitude of virus entry enhancement by these amyloidogenic peptides raises the possibility that the effect reported herein may be useful in applications such as gene therapy using viral vectors. Particularly in desirable target cells, viral titers and infection rates are frequently limiting. A␤ 1-40 is less neurotoxic than A␤   (20,50), but under the conditions assayed, it is more potent in promoting infection. It is therefore possible that the ability of a peptide to promote infection is independent of its pathogenic properties. Experiments aimed at identifying synthetic peptides that promote infection with the same or greater efficiency as A␤, but with lower cytotoxicity, are under way.