Small Molecule Targets Env for Endoplasmic Reticulum-Associated Protein Degradation and Inhibits Human Immunodeficiency Virus Type 1 Propagation

ABSTRACT Human immunodeficiency virus type 1 (HIV-1) is dependent on its envelope glycoprotein (Env) to bind, fuse, and subsequently infect a cell. We show here that treatment of HIV-1-infected cells with glycyl-prolyl-glycine amide (GPG-NH2), dramatically reduced the infectivity of the released viral particles by decreasing their Env incorporation. The mechanism of GPG-NH2 was uncovered by examining Env expression and maturation in treated cells. GPG-NH2 treatment was found to affect Env by significantly decreasing its steady-state levels, its processing into gp120/gp41, and its mass by inducing glycan removal in a manner dependent on its native signal sequence and the proteasome. Therefore, GPG-NH2 negatively impacts Env maturation, facilitating its targeting for endoplasmic reticulum-associated protein degradation, where Env is deglycosylated en route to its degradation. These findings illustrate that nontoxic drugs such as GPG-NH2, which can selectively target glycoproteins to existing cellular degradation pathways, may be useful for pathogen therapy.

The endoplasmic reticulum (ER) contains a number of molecular chaperones and folding factors that aid in the maturation of proteins that traverse the secretory pathway. This process is strictly monitored by the ER quality control system, which selects properly folded proteins for export to the Golgi (16) and targets misfolded proteins for destruction through the ER-associated protein degradation pathway (ERAD) (4,28). Once an ER protein is selected as a substrate for ERAD, it is translocated from the ER lumen to the cytosol through an ER translocon. This retrotranslocation process is thought to be driven by either the cytosolic AAA-ATPase p97 (39) or the 19S proteasome cap (23). Upon entrance into the cytosol, the ERAD substrate is ubquitinated, and its glycans are removed by an N-glycanase to prepare it for proteasomal degradation (11,28). Viral envelope glycoproteins utilize the host cell secretory pathway for their proper maturation and trafficking to the site of viral assembly. The human immunodeficiency virus type 1 (HIV-1) encodes the envelope glycoprotein (Env), which initiates HIV-1 infections by mediating attachment and fusion of the viral envelope with the host cell membrane (17). Therefore, infectious HIV-1 particle production relies on the ability of Env to pass the rigorous ER quality control system. Env is initially synthesized as a type I membrane precursor glycoprotein termed gp160, which is cotranslationally targeted to the ER by its 30-amino-acid N-terminal signal sequence (24). Within the ER, gp160 receives ϳ30 N-linked glycans and is assisted in its maturation by the chaperones BiP, calnexin, and calreticulin as it undergoes extensive disulfide bond formations (15,21,31). Once gp160 has reached its native state with ten disulfide bonds and its signal sequence has been cleaved posttranslationally (21,25), it assembles into trimers (26) and is exported to the Golgi. Within the Golgi, gp160 is cleaved by cellular endoproteases, yielding the transmembrane protein gp41 and the noncovalently associated surface protein gp120 (27). Thereafter, this complex is transported to the plasma membrane, where it is incorporated into the envelope of assembling HIV-1 particles.
We have previously shown that a tripeptide amide corresponding to a conserved motif of the HIV-1 Env, glycyl-prolylglycine amide (GPG-NH 2 ), suppressed the replication of all 47 HIV-1 laboratory strains and clinical isolates examined with a 50% inhibitory concentration of ϳ10 M, a concentration that is 200-to 2,000-fold less than what affected cell growth or had other toxic effects on peripheral blood mononuclear cells (35). However, this suppression was not, as we had anticipated, due to interactions of the peptide with the early events of the HIV-1 replication cycle, such as attachment or entry (36). In the present study, we demonstrate that GPG-NH 2 reduced Env incorporation into HIV-1 particles during replication by targeting Env toward the ERAD pathway. The ability of GPG-NH 2 to target Env for degradation was dependent on the presence of functional proteasomes and required the fulllength Env signal sequence. These findings illustrate that small molecules may be utilized therapeutically to specifically target unwanted pathogenic proteins for degradation by the existing cellular machinery. viously described (8,13). Peroxidase-conjugated concanavalin A was obtained from Sigma.
Virus expression and precipitation of HIV-1 particles. ACH-2 cells (8 ϫ 10 5 cells/ml) were cultured with or without GPG-NH 2 or GPG-OH for various time prior to the addition of 100 nM 12-phorbol-13-myristate acetate (PMA). Three days later, the cell culture supernatants were collected, cleared by centrifugation at 300 ϫ g for 10 min, passed through 0.45-m-pore-size filters, and the particles were precipitated at 4°C for 48 h in 1:6 (vol/vol) with 40% polyethylene glycol 6000 containing 0.667 M NaCl. The precipitated particles were allowed to sediment at 16,000 ϫ g for 20 min at 4°C. The virus pellets were then dissolved in radioimmunoprecipitation assay (RIPA) buffer containing 50 mM Tris-HCl (pH 7.4), 1% Triton X-100, 1% deoxycholate, 150 mM NaCl, 1 mM EDTA, and 0.1% sodium dodecyl sulfate (SDS) and supplemented with Complete protease inhibitor cocktail (Roche).
Infectivity assay and syncytia formation. For the infectivity assay, 2 ϫ 10 5 TZM-bl cells were infected with virus equivalent to 200 ng of p24 in 0.5 ml of culture medium. After 2 h at 4°C, the infection medium was replaced with virus-free culture medium containing 5 M indinavir. The cells were harvested after 40 h at 37°C and assayed for luciferase activity using the Bright-Glo luciferase assay system (Promega). To test for syncytium formation, ACH-2 cells were cultured in 10 M indinavir with or without 1,000 M GPG-NH 2 . After 24 h, 250,000 SupT1 cells were added to and cocultured with 20,000 ACH-2 cells. Syncytium formation was then monitored by light microscopy 24 and 40 h later.
Transfection, drug treatments, and immunoblotting. HeLa-tat III cells (ϳ5 ϫ 10 5 cells/dish) were treated with the indicated concentrations of GPG-NH 2 prior to cotransfection with the CAT expressing control plasmid and the appropriate construct using FuGENE 6 (Roche). For inhibition of the proteasome (lactacystin [LCT] and epoxomicin), glucosidases (castanospermine [CST]), and mannosidases (deoxymannojirimycin [DMJ] and kifunensine [KIF]), the drugs were added to the cells at 8 to 12 h posttransfection. The isolated virus and cells were lysed in RIPA buffer, standardized to p24 or CAT levels, resolved by SDSpolyacrylamide gel electrophoresis (PAGE), transferred to nitrocellulose membranes, and immunoblotted. The membranes were exposed to film for the appropriate time, and band intensities were quantified by using GeneTools analysis software (SynGene).
UPR, enzyme-linked immunosorbent assay (ELISA), and measurement of RT activity. Induction of unfolded protein response (UPR) after GPG-NH 2 treatment was examined by assaying XBP-1 mRNA splicing as previously described (22). In brief, HeLa-tat III cells were exposed to 5 mM dithiothreitol (DTT) for 3 h or 0 to 1 mM GPG-NH 2 for 48 h. The total cellular RNA was isolated by Env precursor protein (gp160) and the precursor protein p55Gag depicting antibody recognition sites (inverted "Y"), N-linked glycans (trident symbol), and proteolytic cleavage sites (2). (C) HIV-1 particles were produced in the indicated concentrations of GPG-NH 2 . Equivalent p24 amounts of the viral particles were used to infect TZM-bl cells at 4°C for 2 h, the infection medium was removed, and the cells were cultured for 48 h in the presence of 5 M indinavir. The luciferase activity in the lysates was monitored to determine the relative infectivity of the particles. (D) ACH-2 cells were pretreated with 0 to 400 M GPG-NH 2 for 24 h before HIV-1 replication was induced with PMA for 48 h. HIV-1 particle production was determined by measuring the extracellular p24 by ELISA. (E) HIV-1 virus was mixed with 0 to 1,000 M GPG-NH 2 and incubated with TZM-bl cells for 2 h. Upon removal of the infection media the cells were incubated in the presence of 5 M indinavir for 48 h. The cells were subsequently harvested, and the infectivity was determined by measuring the intracellular luciferase activity.
using an RNeasy isolation kit (Qiagen). Standard reverse transcriptase PCR (RT-PCR) was performed by using the oligonucleotide mXBP1 804AS, followed by amplification of spliced or unspliced XBP-1 by PCR with the primers 804AS XBP1 and 383S XBP as previously described (22).
The p24 levels or RT activity in cell culture supernatants were quantified by using p24-ELISA (18) or a Lenti-RT activity assay (Cavidi Tech), respectively. CAT concentrations in cell lysates were quantified by using the CAT ELISA kit (Roche). Subcellular fractionation and alkaline extraction. Cell membrane fractions were prepared from HeLa-tat III cells homogenized with a Dounce homogenizer in 10 mM HEPES (pH 7.4), 1 mM EDTA, and 0.25 M sucrose supplemented with 125 M phenylmethylsulfonyl fluoride and 2.5 g of aprotinin and leupeptin/ml. Nuclei and nondisrupted cells were removed by centrifugation at 1,500 ϫ g for 10 min at 4°C. The supernatants were further centrifuged at 180,000 ϫ g for 1 h at 4°C to separate the membrane vesicles from cytosolic components. The supernatants were precipitated with 15% trichloroacetic acid, and the precipitates rinsed in acetone and dissolved in reducing sample buffer. The membrane pellets were resuspended in 0.5 mM sucrose containing 50 mM TEA (pH 7.5) and 1 mM DTT. For isolation of integral membrane proteins by alkaline extraction, the microsome suspension was first diluted 20 times in 0.1 M NaCO 3 (pH 11.5), incubated for 30 min on ice, and then separated through a 0.5 M sucrose cushion for 1 h at 180,000 ϫ g at 4°C. The pellets were resuspended in reducing sample buffer, while the supernatants were precipitated with trichloroacetic acid as described above.
PNGase F and Endo H digestion. Cell lysates in RIPA buffer were supplemented with 0.5% SDS and 1% ␤-mercaptoethanol and denatured for 10 min at 100°C. The PNGase F reactions were adjusted to 1% NP-40, and the Endo H (endo-␤-N-acetylglucosaminidase H) reactions were adjusted to 50 mM sodium citrate (pH 5.5). Incubation followed at 37°C for 1 h with 16 U per l of lysate of PNGase F or Endo H (New England Biolabs).

RESULTS
GPG-NH 2 decreases HIV-1 particle infectivity. Initially, the effects of GPG-NH 2 (the structure is depicted in Fig. 1A) on infectious HIV-1 particle production was examined. HIV-1 was generated from chronically infected ACH-2 cells in the absence or presence of GPG-NH 2 . TZM-bl cells, which express luciferase from an HIV-1 Tat-dependent promoter, were then infected with equivalent amounts of the produced viral particles as determined by p24 levels. The infectivity of HIV-1 particles produced in 50 M GPG-NH 2 was ϳ50% of those generated in the absence of GPG-NH 2 (Fig. 1C), and it declined in a dose-dependent fashion. However, GPG-NH 2 treatment did not affect HIV-1 particle production, since the extracellular viral capsid protein (p24) levels and the viral RT activity were comparable to the untreated samples ( Fig. 1D and data not shown). To address whether GPG-NH 2 decreased viral infectivity by inhibiting viral binding and entry, TZM-bl cells were infected with untreated virus in the presence of GPG-NH 2 . Even at 1,000 M, GPG-NH 2 showed no effect on luciferase expression, indicating that the drug does not prevent viral binding and entry (Fig. 1E). Together, these findings indicate that HIV-1 particles, produced in the presence of GPG-NH 2 , are less infectious due to inhibition of a step in the replication process that occurs after entry. GPG-NH 2 treatment reduces Env incorporation into HIV-1 particles. The decreased infectivity of HIV-1 particles, produced in the presence of GPG-NH 2 , could be due to a reduced ability of the virus to enter target cells. Since HIV-1 entry into cells is mediated by Env, its content in the particles was examined. ACH-2 cells were treated with 100 M GPG-NH 2 for various times prior to stimulating virus production with PMA. As controls, ACH-2 cells were either untreated or pretreated with the inactive analog GPG-OH (the structure is depicted in Fig. 1A). At 3 days postinduction, the viral particles were collected and examined for the incorporation of p24, unprocessed Env (gp160), and processed Env (gp120/gp41) by immunoblotting ( Fig. 2A). In the presence of GPG-NH 2 the particle incorporation of gp160, gp120, and gp41 decreased by 60 to 80% compared to the controls, whereas the p24 levels remained unaffected ( Fig. 2A, compare lanes 1 and 5 to lanes 2 to 4).
To determine whether GPG-NH 2 reduced the viral incorporation of Env in a cell-type-specific manner, HIV-1 particles produced by the human epithelial derived cell line HeLa-tat III were examined. The cells were untreated or treated with various amounts of GPG-NH 2 20 h prior to transfection with the HIV-1 infectious clone, pNL4-3. The viral particles were collected 3 days posttransfection and standardized to p24 levels by ELISA (data not shown). Similar to ACH-2 cells, GPG-NH 2 treatment reduced the incorporation of gp41 into the particles produced by HeLa-tat III cells (Fig. 2B). This effect appeared to be dose dependent since the gp41/p24 ratio decreased from 0.48 to 0.28 in the presence of 100 and 500 M GPG-NH 2 , respectively (Fig. 2B, compare lanes 1 and 2 to lanes 3 to 6). Decreased levels of processed Env (gp120/gp41) on the viral surface should reduce the capacity of the virus to bind to and fuse with the target cells. To examine this, we developed a cell model that is based on the ability of infected cells, which express gp120/gp41 on their surface, to fuse with uninfected viral target cells. This fusion results in the formation of large multinuclear cells, called syncytia, which can easily be monitored under a light microscope. ACH-2 cells express gp120/ gp41 on their surface, but due to their lack of CD4 expression they cannot form syncytia among themselves. Thus, the ACH-2 cells were used as "virus" since they must be cocultured with CD4-positive target cells to form syncytia. The ACH-2 cells were either untreated or pretreated for 24 h with 1,000 M GPG-NH 2 . In addition, the cell cultures were treated with 10 M indinavir, an HIV-1 protease inhibitor, to prevent production of infectious virus. The ACH-2 cells were subsequently stimulated with PMA, to increase the cell surface levels of Env, 5 h prior to the addition of the CD4 positive cells, SupT1. After 24 and 40 h, a significant number of large syncytia had been formed in the GPG-NH 2 -untreated cocultures, but none were detected in those containing the GPG-NH 2 -pretreated ACH-2 cells (Fig. 2C). Together, these findings show that GPG-NH 2 does not inhibit intact HIV-1 particles from fusing with target cells but rather that HIV-1 particles produced in the presence of GPG-NH 2 have a decreased incorporation of gp120 and gp41. This reduces the ability of the HIV-1 particles to fuse with target cells and thus decreases their infectivity.
GPG-NH 2 decreases the steady-state levels and mass of gp160. To examine the mechanism by which GPG-NH 2 reduced Env incorporation into HIV-1 particles and prevented the ACH-2 cells from fusing with the CD4 positive SupT1 cells, the intracellular expression of Env was studied in a noninfectious system. HeLa-tat III cells were treated with various concentrations of GPG-NH 2 or GPG-OH 20 h prior to the transient expression of gp160 and the internal control protein, CAT. The cells were collected 20 h posttransfection, and the lysates were standardized to the CAT levels, separated by SDS-PAGE, and immunoblotted for Env. Antibodies to the N-terminal region of gp160 and gp120 (see Fig. 1B) revealed a GPG-NH 2 dose-dependent increase in gp160 mobility that coincided with decreases in its steady-state levels (Fig. 3, upper panel, lanes 1 to 6). Since the gp160 mobility change was not discernible from processing to gp120, a C-terminal antibody that recognized gp160 and gp41 was utilized (see Fig. 1B). This antibody revealed a GPG-NH 2 -dependent increase in gp160 mobility and a decrease in its steady-state levels, as well as a reduction in its processing to gp41 (Fig. 3, middle panel, lanes  1 to 6). Given that the smaller gp160 species were observed with antibodies to both termini, at GPG-NH 2 concentrations inhibiting its cleavage to gp41, enhanced processing of Env to gp120 and gp41 was ruled out as the cause of the increased gp160 migration.
Unlike the heavily glycosylated gp160, the steady-state levels of the HIV-1 cytosolic precursor protein p55Gag and its cleavage product p24 were unaffected by GPG-NH 2 treatment (Fig.  3, bottom panel). These results suggested that GPG-NH 2 reduced Env levels in HIV-1 particles by interfering with gp160 biosynthesis rather than by inhibiting its incorporation during the process of viral assembly.
GPG-NH 2 has higher specificity for the HIV-1 glycoprotein than for cellular glycoproteins. Since GPG-NH 2 did not alter the expression or processing of the cytosolic viral precursor protein p55Gag, it was further explored whether GPG-NH 2 had a general or selective effect on proteins within the secretory pathway. First, the effects of GPG-NH 2 were examined on the endogenously expressed glycoprotein LAMP-1 (lysosomeassociated membrane protein 1), which, like gp160, is a heavily glycosylated type 1 membrane protein. A slight reduction in its steady-state levels and increased mobility was observed at 500 M GPG-NH 2 (Fig. 4A, upper panel, lanes 5 and 6). This concentration is ϳ25-fold higher than what is required for similar effects on gp160. Next, the endogenous nonglycosylated type 1 ER membrane protein calnexin was examined and found to be completely unaffected by GPG-NH 2 treatment (Fig. 4A, lower panel). To investigate whether GPG-NH 2 has a general effect on glycoproteins, the glycoprotein expression profile in the HeLa-tat III cells was analyzed with the lectin, concanavalin A. Similar to LAMP-1, the cellular glycoprotein pattern was only slightly altered at the highest GPG-NH 2 concentrations (500 to 1,000 M), where the steady-state levels showed a minor decrease and a high-molecular-mass glycoprotein band of ϳ150 kDa appeared to increase in mobility (Fig. 4B, lanes 5 and 6). These results indicated that GPG-NH 2 has a significantly higher selectivity for gp160 than cellular glycoproteins. During HIV-1 infection the majority of gp160 resides within the ER (21). The high proportion of gp160 within the ER and the observation that GPG-NH 2 reduces its steady-state levels raised the possibility that GPG-NH 2 induced ER stress. The induction of ER stress activates the UPR, which could facilitate enhanced degradation of gp160, resulting in the observed decrease in its steady-state levels. The UPR was followed by monitoring splicing of the XBP-1 mRNA. In contrast to the known UPR inducer, dithiothreitol (DTT), GPG-NH 2 treatment of HeLa-tat III cells did not activate the UPR, even at concentrations up to 1 mM for 48 h (Fig. 4C, compare lane 6 to lanes 1 to 5). Therefore, the ability of GPG-NH 2 to decrease gp160 steady-state levels was not a result of UPR activation.
GPG-NH 2 treatment alters the glycan status of gp160 prior to Golgi transport. Apart from proteolysis, the heterogeneous mass reduction of gp160 after GPG-NH 2 treatment could be caused by N-glycan removal or deficient glycosylation. To examine these possibilities, whole-cell lysates were treated with the N-glycanase PNGase F or Endo H. After deglycosylation with either glycanase, gp160 produced in the absence or presence of GPG-NH 2 coalesced as single bands with identical mobility (Fig. 5A, compare lanes 1 to 3 to lanes 4 to 9). This demonstrated that the reduced mass of gp160 was not caused by proteolysis but was due to variations in its carbohydrate content. Furthermore, the complete deglycosylation of gp160 by Endo H indicated that gp160 had not been exported to the medial Golgi, where complex carbohydrates are added that prevent Endo H deglycosylation.
Similar to gp160, deglycosylation of LAMP-1 with PNGase F resulted in single sharp bands with equal mobilities, regardless of GPG-NH 2 treatment (Fig. 5B). However, LAMP-1 exhibited resistance to Endo H digestion, demonstrating that its transport to the Golgi is not affected by GPG-NH 2 treatment. This indicated that the higher sensitivity of gp160 toward GPG-NH 2 may be due to its longer residence within the ER.
N-linked glycans are trimmed of glucoses and some mannoses before they exit the ER for the Golgi. To determine whether glycan trimming was required for GPG-NH 2 to affect gp160, the glucosidase I and II inhibitor CST or the mannosidase inhibitors DMJ or KIF was added 8 h posttransfection to untreated or GPG-NH 2 -treated cells. Treatment with CST increased the mass of gp160 due to the presence of additional glucoses on each glycan. CST also inhibited gp160 processing to gp41, but it did not prevent the action of GPG-NH 2 since faster-migrating gp160 products were still observed (Fig. 5C,  lanes 1 to 4). Mannosidase inhibition had no influence on the effect of GPG-NH 2 (Fig. 5C, lanes 5 to 10). Taken together, these observations indicated that GPG-NH 2 affected gp160 within the ER, leading to a reduction in the number of Nlinked glycans present on its polypeptide backbone.
Proteasome inhibition prevents GPG-NH 2 -induced deglycosylation of gp160. Glycoproteins targeted for ERAD are deglycosylated in the cytoplasm by N-glycanase prior to proteasomal degradation. To determine whether the decrease in gp160 glycan content and steady-state levels was a result of GPG-NH 2 targeting gp160 to the ERAD pathway, proteasomes were partially inhibited with LCT. LCT was added 10 h posttransfection, and the steady-state levels of gp160 were analyzed by immunoblotting. The partial proteasome inhibition by LCT, in the presence of GPG-NH 2 , stabilized the completely glycosylated gp160 enabling its processing to gp41 and allowed the visualization of smaller gp160 degradation intermediates (Fig. 6A, top panel, compare lane 2 to lanes 4  and 6). Similarly, proteasome inhibition prevented the slight deglycosylation of LAMP-1 in GPG-NH 2 -treated cells (Fig.  6A, bottom panel, compare lane 2 to lanes 4 and 6). Fully glycosylated gp160 could also be rescued by the proteasome inhibitor epoxomicin (Fig. 6B, compare lanes 2 to 4). The rescue of fully glycosylated gp160, by proteasome inhibition, showed that the GPG-NH 2 -induced glycan deficiency was caused by glycan removal and not inefficient addition. Furthermore, proteasomal inhibition increased the steady-state levels of gp160 in the presence of GPG-NH 2 . This shows that GPG-NH 2 targets gp160 toward ERAD, resulting in its proteasomemediated retrotranslocation to the cytoplasm, where it is deglycosylated prior to degradation.
To examine whether the deglycosylated fractions of gp160 that accumulate in the presence of GPG-NH 2 are membrane bound or have been released into the cytosol upon retrotranslocation from the ER, membrane fractions were isolated after cell fractionation. Both gp160 and gp41 derived from cell homogenates of untreated and GPG-NH 2 -treated cells were found entirely in the membrane fractions (Fig. 6C, lanes 3 to  6). Further alkaline extraction of the cellular membranes indicated that the gp160 remained an integral membrane protein after treatment with GPG-NH 2 (Fig. 6C, lanes 7 to 10). As a positive control for integral membrane proteins, gp41 and calnexin were monitored and found only in the membrane pellet after alkaline extraction. Therefore, the gp160 that accumulates after GPG-NH 2 treatment appears to remain associated with the membrane.
GPG-NH 2 requires the native gp160 signal sequence to facilitate its targeting to ERAD. In general, substrates targeted for ERAD are improperly folded. This raises the possibility that GPG-NH 2 acts adversely on a unique step in gp160 maturation. During maturation, gp160 undergoes a rare posttranslational cleavage of its signal sequence, which contributes to its prolonged presence within the ER. The delayed cleavage is partially due to several positively charged amino acids in the n-region of the signal sequence (24,25). To investigate whether the positively charged n-region was important for the action of GPG-NH 2 , a mutant version of gp160 (⌬nSS-gp160) was created lacking the entire n-region of its signal sequence (Fig. 7A). Upon cellular expression ⌬nSS-gp160 had a molecular mass of 160 kDa and was processed into gp41. This indicated that the mutant was still targeted to the ER and transported to the Golgi (Fig. 7B, upper and lower panels, lanes 4  to 7). After a longer exposure, a fraction of ⌬nSS-gp160 appeared to remain in the cytoplasm as a band corresponding to nonglycosylated gp160 (Fig. 7B, lower panel, lanes 4 to 7, ϳ90  kDa). Strikingly, the fully glycosylated ER-targeted ⌬nSS-gp160 was completely unaffected by GPG-NH 2 treatment even at concentrations up to 500 M (Fig. 7B, upper and lower  panels, compare lanes 1 to 3 to lanes 4 to 7). To ensure the lower expression level of ⌬nSS-gp160 did not mask the effect of GPG-NH 2 , the vector expressing wild-type gp160 was titrated during the transfection to reach the expression level of ⌬nSS-gp160. It was found that that the wild-type gp160 was still affected while the ⌬nSS-gp160 remained unaffected even up to 1,000 M GPG-NH 2 (Fig. 7C). These data further indicate that GPG-NH 2 targets gp160 for ERAD by altering a specific step in its maturation process that involves its native signal sequence.

DISCUSSION
HIV-1 readily escapes host immune responses and antiretroviral drugs due to its ability to mutate and adapt to the environment in which it replicates. To counteract this property, the available HIV-1 therapies utilize combinations of drugs targeting the viral RT, protease, and recently also the integrase and viral envelope fusion with the host cell plasma membrane (9,19,33). However, HIV-1 still poses a significant therapeutic problem as it continues to evolve multidrug resistance (20). Thus, development of new antiviral compounds with different modes of action from those currently available drugs is necessary to effectively combat HIV-1 infection.
In the present study, we uncovered a previously unexplored mechanism for suppressing HIV-1 propagation using the drug GPG-NH 2 . The presence of GPG-NH 2 during HIV-1 replication drastically reduced HIV-1 particle infectivity by decreasing the viral incorporation of Env. Our findings are summarized into a model (Fig. 8) in which the viral Env incorporation was reduced due to the specific ability of GPG-NH 2 to direct maturing Env for degradation via the ERAD pathway. Consequently, gp160 is retrotranslocated from the ER lumen into the cytoplasm, where it is deglycosylated prior to its degradation. This correlates with the proteasomal requirement for the observed decrease in gp160 steady-state levels, processing, N-linked glycan status, and subsequently its molecular weight.
The feasibility of developing a drug capable of targeting a specific substrate to the ERAD pathway increases when the substrate has a prolonged ER residency and possesses unique steps in its maturation process that can be disrupted. gp160 undergoes a rare posttranslational signal sequence cleavage in the ER that occurs after substantial maturation and disulfide bond formation (21). The abnormally late removal of the signal sequence is directed in part by several positively charged residues in its n-region that also contribute to the prolonged existence of gp160 within the ER (24,25). Consequently, the native gp160 signal sequence could extend the time period for GPG-NH 2 to act on gp160, leading to its pronounced sensitivity to the drug. To examine the significance of the positively charged region within the signal sequence, for the effect of GPG-NH 2 on gp160, the entire n-region was deleted. The effects of GPG-NH 2 on this construct were entirely abolished, supporting the view that the prolonged ER residency of gp160 contributes to its enhanced GPG-NH 2 sensitivity. Alternatively, GPG-NH 2 may interact directly with the n-region of the signal sequence and adversely affect the maturation of gp160 by disrupting the signal sequence insertion into the ER membrane. However, the removal of the n-region is predicted in silica by SignalP 3.0 (7,29,30) to create a signal anchor sequence. This scenario rather indicates that GPG-NH 2 disrupts gp160 maturation by interacting with the mature n terminus of gp160 after signal sequence cleavage.
The effect of GPG-NH 2 has been shown to be restricted to glycoproteins, since gp160, LAMP-1, and an unspecified glycoprotein (ϳ150 kDa) were affected, while the nonglycosylated ER resident protein calnexin and the viral cytosolic proteins p55Gag and p24 were not. However, a great diversity was observed in the glycoprotein susceptibility to GPG-NH 2 treatment, indicating a certain degree of specificity. The endogenous glycoprotein LAMP-1, as well as the concanavalin A-detected glycoproteins, exhibited a significantly lower sensitivity to GPG-NH 2 than gp160, with only slight effects being observed at 0.5 to 1 mM. These findings imply that GPG-NH 2 acts selectively among glycoproteins, where gp160 is more sensitive to the drug than endogenous proteins. In agreement with this is the finding that GPG-NH 2 treatment did not induce the UPR. Furthermore, the specificity of GPG-NH 2 was readily FIG. 8. Proposed model of how GPG-NH 2 targets gp160 to ERAD. Initially, gp160 is cotranslationally translocated into the ER. (Arrow 1) The presence of GPG-NH 2 or its metabolites prevents the proper folding of gp160, while its signal sequence remains attached. The improper folding of gp160 causes it to be targeted for ERAD. (Arrow 2) gp160 is then retrotranslocated to the cytoplasm in a proteasome-mediated fashion. The polypeptide remains integrated into the ER membrane, likely through its transmembrane region, while (arrow 3) it is deglycosylated by the cytosolic N-glycanase prior to (arrow 4) degradation by the proteasome. apparent by its inability to target gp160 with an N-terminal truncated signal sequence to ERAD.
The use of peptides for therapy raises concerns regarding their short viability due to their high susceptibility to proteases. However, GPG-NH 2 has been found to gain its antiviral activity toward HIV-1 upon being metabolized to ␣-hydroxy glycineamide via the intermediate glycineamide (2,5,6; unpublished data). Furthermore, ␣-hydroxy glycineamide has been found to exhibit the same ability as GPG-NH 2 to reduce Env incorporation into HIV-1 particles, but at a higher efficiency (unpublished data).
We therefore believe this compound offers a potential lead for the development of a new type of drugs for use against HIV-1 and other pathogens. These drugs would be utilized to specifically target the removal of essential pathogenic substrates using the preexisting cellular degradation machinery. The therapeutic benefits of this approach are twofold. First, degradation of an essential pathogenic component would reduce the ability of the pathogen to propagate. Second, enhanced degradation of the foreign protein would lead to increased antigen production for major histocompatibility complex class I presentation, improving the chances of eliciting an immune response to infected cells (38).
Decreasing HIV-1 particle infectivity by targeting Env maturation, using general glucose trimming inhibitors such as CST or deoxynojirimycin, has been tried by others (14). These compounds do not target glycoproteins to the ERAD pathway, nor do they have specificity for viral glycoproteins. Although these inhibitors suppressed HIV-1 replication in vitro by preventing gp160 processing in the Golgi, they did not diminish Env incorporation or enhance its degradation (14). Ultimately, the lack of specificity for these glucosidase inhibitors has prevented their development for therapeutics. We demonstrate here that a drug can selectively induce the degradation of the HIV-1 envelope glycoprotein by targeting it for ERAD, thus limiting its incorporation into viral particles and thereby reducing their infectivity.