The Highly Conserved Glycan at Asparagine 260 of HIV-1 gp120 Is Indispensable for Viral Entry*

Carbohydrate-binding agents bind to the N-glycans of HIV-1 envelope gp120 and prevent viral entry. Carbohydrate-binding agents can select for mutant viruses with deleted envelope glycans. Not all glycosylation motifs are mutated to the same extent. Site-directed mutagenesis revealed that deletions destroying the highly conserved 260NGS262 glycosylation motif resulted in non-infectious virus particles. We observed a significant lower CD4 binding in the case of the N260Q mutant gp120 virus strains, caused by a strikingly lower expression of gp120 and gp41 in the virus particle. In addition, the mutant N260Q HIV-1 envelope expressed in 293T cells was unable to form syncytia in co-cultures with U87.CD4.CXCR4.CCR5 cells, due to the lower expression of envelope protein on the surface of the transfected 293T cells. The detrimental consequence of this N-glycan deletion on virus infectivity could not be compensated for by the creation of novel glycosylation sites near this amino acid, leaving this uncovered envelope epitope susceptible to neutralizing antibody binding. Thus, the Asn-260 glycan in the gp120 envelope of HIV-1 represents a hot spot for targeting suicidal drugs or antibodies in a therapeutic effort to efficiently neutralize a broad array of virus strains.

The envelope of HIV-1, which consists of the surface glycoprotein gp120, non-covalently bound to the transmembrane glycoprotein gp41, is heavily glycosylated. Leonard et al. (2) demonstrated that gp120 of HIV-1 IIIB , expressed in CHO cells, contained 24 N-linked glycans, of which 11 are high mannose or hybrid type, and 13 are complex type. The high number of high mannose type glycans is peculiar, because human cells rarely express proteins carrying this type of glycan (3). Moreover, the high mannose type glycans appear to be clustered on the gp120 envelope, resulting in an unusual density of such glycans in the envelope of HIV-1, discriminating the HIV-1 envelope gp120 from mammalian glycoproteins (4).
We and others previously reported that these carbohydratebinding agents (CBAs) select for a unique drug resistance profile that mainly affects the N-glycosylation pattern of HIV-1 gp120 (9 -12). In order to escape drug pressure, HIV-1 selects for mutant virus strains in whose envelope N-glycosylation sites were deleted by mutating the Asn or the Ser/Thr of the NX(S/T) glycosylation motif. There seems to be a preference to delete high mannose type glycans versus complex type glycans on HIV-1 gp120 (13). Moreover, we observed that among the six complex type N-glycans of the V1/V2-loop of gp120, only three of them have been (rarely) deleted under CBA pressure (14). We could demonstrate that deletion of one or more glycans on the V1/V2-loop of gp120 results in mutant virus strains that showed an increased sensitivity to the inhibitory activity of CBAs, such as HHA and UDA, explaining why they did not appear under escalating CBA pressure. Interestingly, only three N-linked glycans outside the V1/V2 part of HIV-1 gp120 were never found deleted (i.e. Asn-239, Asn-260, and Asn-354) in a wide variety (Ͼ60) of virus strains that had been exposed to a variety of CBAs in cell culture for extended time periods. To address the possibility that elimination of these N-linked glycans might severely compromise HIV-1 replication, which would result in virus strains that are not sufficiently viable and cannot appear upon drug exposure, we constructed mutant HIV-1 NL4.3 strains that lacked each of these glycans on their  Table S1 and Figs. S1 and S2. 1 To whom correspondence should be addressed. Fax: 32-16-337341; E-mail: jan.balzarini@rega.kuleuven.be. 2  envelope. We found that an intact 260 NGS 262 glycosylation motif is indispensable for viral infectivity and that mutations affecting this motif result in a substantially lower expression of gp120 in the viral particle. Deletion of this Asn-260 glycan in HIV gp120 cannot be compensated for by the generation of newly created nearby N-glycans; therefore, the Asn-260 glycan represents a hot spot for a targeted therapeutic intervention by chemotherapeutic means and/or by immunological attack.

EXPERIMENTAL PROCEDURES
Ethics Statement-PBMCs were obtained through, and with the permission of, the Blood Transfusion Center of Leuven. They are derived from anonymous healthy donors, and the material is normally discarded by the Blood Transfusion Center if not needed for the HIV infection experiments.
Test Compounds-The mannose-specific plant lectins from Hippeastrum hybrid (HHA) and from G. nivalis (GNA) and the N-acetylglucosamine-specific plant lectin from U. dioica (UDA) were derived and purified from these plants as described previously (15,16) and kindly provided by Prof. E. J. M. Van Damme and Dr. W. Peumans (Ghent, Belgium). Pradimicin A (PRM-A) was obtained from Prof. T. Oki and Prof. Y. Igarashi (Toyama, Japan).
Viruses-The pNL4.3-⌬env-EGFP construct was used for production of wild-type NL4.3 virus after recombination with env and expresses an enhanced version of green fluorescent protein (EGFP) located between env and nef without affecting the expression of any HIV-1 gene. For this molecular clone, the expression of EGFP in infected cells is a measurement of virus production as described previously (18). The construct pNL4.3-⌬env-EGFP was a kind gift from Dr. M. E. Quiñones-Mateu (Lerner Research Institute, Cleveland, OH).
Construction of Mutant gp120 Virus Strains-The plasmid pBlue-env, which encodes the env gene (18,19), was used to generate gp120 mutant virus strains with a disrupted glycosylation site at amino acid positions Asn-239, Asn-260, and Asn-354, where Asn was replaced by Gln. At amino acid position Ser-262, Ser was replaced by Cys or Ala to delete the 260 NGS 262 glycosylation site motif. At amino acid position Gly-261, Gly was replaced by an Ala, resulting in the glycosylation site motif 260 NAS 262 . These glycan mutations were introduced into the pBlue-env using the QuikChange site-directed mutagenesis kit (Agilent Technologies, Diegem, Belgium). In addition, double mutant gp120 virus strains were constructed that contained V253N to create the glycosylation motif 253 NST 255 ), I270N (to create the glycosylation motif 270 NRS 272 ), Q256N/L258T (to create the glycosylation motif 256 NLT 258 ), and E266N/V268T (to create the glycosylation motif 266 NET 268 ). Each of these new glycosylation site motifs was also combined with a deleted 260 NGS 262 glycosylation site (e.g. 260 QGS 262 ).
Plasmid DNA was purified by the PureLink Quick Plasmid Miniprep Kit (Invitrogen). The presence of glycosylation site mutations was confirmed by sequencing the gp120 gene as described previously (20).
Generation of Mutant Virus by env Chimeric Virus Technology-The generation of mutant virus was performed as described previously (14). Briefly, the PCR fragment was amplified from mutated pBlue-env using Expand High Fidelity Enzyme blend (Roche Applied Science). The PCR products were purified with the QIAgen PCR purification kit (Qiagen, Venlo, The Netherlands). 2 g of PCR product was co-precipitated with 10 g of linearized pNL4.3-⌬env-EGFP and co-transfected into 293T cells using the calcium phosphate method as described (18,19). Positive transfection of 293T cells was detected by fluorescence microscopy. The supernatant, containing the mutant virus, was used to infect U87.CD4.CCR5.CXCR4 cells for the production of stock virus. After 3-5 days, the virus was harvested from the culture supernatant and stored at Ϫ80°C.
Capacity of the Mutant gp120 Virus Strains to Infect Different Susceptible Cell Lines-In order to determine the infection capacity of the mutant gp120 virus strains, equal amounts of virus, corresponding to 5,000 pg of p24, were added to 5 ϫ 10 3 U87.CD4.CCR5.CXCR4 cells or 3 ϫ 10 4 MT4 or C8166 cells in a total volume of 200 l. 3-4 days after infection, the cells were fixed with 3% paraformaldehyde (PFA), and infection was monitored with the FACSCanto II flow cytometer (BD Biosciences). The data were analyzed with FACS Diva Software (BD Biosciences).
In a second experiment, different amounts of virus (5,000, 2,500, 1,250, 625, and 312.5 pg of p24) was added to 30,000 C8166 cells in a total volume of 200 l. 3 days postinfection, the cells were fixed with 3% PFA, and infection was monitored with the FACSCanto II flow cytometer.
PBMCs from healthy donors were stimulated with phytohemagglutinin at 2 g/ml (Sigma) for 3 days at 37°C. The phytohemagglutinin-stimulated blasts were then seeded at 5 ϫ 10 5 cells/well in a 96-well plate. Wild-type or mutant N260Q gp120 virus was added in a 2-fold serial dilution, starting with 750,000 pg of p24 as the highest amount of virus. The cells were incubated for 2 h at 37°C, after which they were thoroughly washed to remove unbound virus. Then the PBMCs were resuspended in medium and incubated for 7 days at 37°C, after which the p24 levels in the supernatant were determined.
Susceptibility of Mutant gp120 HIV-1 Strains to Carbohydrate-binding Agents-The antiviral activity of the CBAs GNA, HHA, UDA, and PRM-A against wild-type and mutant HIV-1 NL4.3_EGFP was measured in C8166 cell culture by determining the EGFP expression of infected cells using the FACSCanto II flow cytometer. 3 days postinfection, the concentration of each compound able to reduce EGFP expression by 50% in the infected cell cultures (i.e. the 50% effective concentration (EC 50 )) was determined.
Binding of Mutant gp120 Virus Strains to Human sCD4-96well Maxi-Sorp plates (Fisher) were coated with 25 ng of human soluble CD4 in 50 mM carbonate buffer, pH 9.6, for 60 min at 37°C. Subsequently, the wells were blocked with phosphatebuffered saline, pH 7.4, with 0.05% Tween 20 (PBST) containing 2% milk powder at 4°C overnight. After washing the plates with PBST, 50,000 pg of p24 of mutant gp120 virus strains, pretreated with 10% Triton X-100 or 500 M aldrithiol-2 (Sigma), was added and incubated at 37°C for 1 h. The wells were thoroughly washed with PBST and incubated with primary sheep anti-gp120 D7324 (Aalto Bio Reagents, Dublin, Ireland) for 1 h at room temperature. After washing away the primary antibody, the wells were incubated with an anti-sheep AP-labeled secondary antibody, which developed color after washing with PBST and the addition of substrate buffer (1 mg/ml p-nitrophenyl phosphate (Sigma) dissolved in 10% diethanolamine, pH 9.8, with 0.5 MgCl 2 ). The absorbance at 405 nm was determined using the Safire 2 microtiter plate reader (Tecan, Männedorf, Switzerland).
Expression of gp120 and gp41 in the Virus Particles and in the Transfected 293T Cells-In order to quantify virion-associated Env protein, supernatant from 293T cells were collected 3 days after virus transfection and concentrated at high speed. The virus pellet was resuspended in lysis buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, and one Complete tablet (Roche Applied Science). Viral protein precipitates containing 50 ng of p24 were separated on a 4 -12% Bis-Tris polyacrylamide gel (Invitrogen) followed by transfer to a PVDF membrane (GE Healthcare). The gp120 protein was detected by Western blot analysis using the polyclonal antibody PA1-7218 (Fisher), gp41 was detected with an anti-gp41 antibody that was obtained through the National Institutes of Health AIDS Research and Reference Reagent Program (21), and the p24 protein was detected with the mouse monoclonal antibody 38/8.7.47 (Abcam, Cambridge, UK).
In order to determine the gp120 expression levels in the transfected 293T cells, 293T cells were seeded and transfected as described above. 3 days post-transfection, the 293T cells were thoroughly washed, twice with PBS containing 1.2 mM CaCl 2 and 1 mM MgCl 2 and once with PBS without CaCl 2 and MgCl 2 . The cell pellet was lysed using radioimmune precipitation assay buffer (Fisher) and sonication. After centrifugation, the supernatant was harvested, p24 levels were determined as described before, and a 2-fold serial dilution of cell extract, starting at 2,000 g of p24, was loaded onto a 4 -12% BisTris polyacrylamide gel as described above. The same antibodies as described above were used for the detection of gp120 and gp41. As a loading control, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was detected in parallel.
Giant Cell Formation between Virus-transfected 293T Cells and Uninfected U87.CD4.CCR5.CXCR4 Cells-293T cells were seeded in glass bottom 35-mm dishes and transfected with DNA encoding wild-type virus, virus containing the N260Q mutation, or virus lacking the envelope protein completely, as described above. 1 day after transfection, the cells were thoroughly washed, and 10 6 U87.CD4.CCR5.CXCR4 cells were added. Less than 24 h later, syncytia formation between the two cell lines was observed in the co-cultures using a fluorescence microscope, and images were taken using a laser-scanning SP5 confocal microscope (Leica Microsystems, Mannheim, Germany). GFP was monitored with the argon laser using the 488-nm line for excitation, and emission was detected between 492 and 558 nm.
Expression of gp120 on the Surface of Transfected 293T Cells as Measured by Anti-gp120 2G12 Antibody Binding-293T cells were seeded in 35-mm glass bottom plates and transfected as described above. 3 days post-transfection, the cells were washed in PBS containing 1.2 mM CaCl 2 , 1 mM MgCl 2 and 2% FCS and incubated for 1 h at room temperature with the 2G12 antibody (Polymun Scientific, Vienna, Austria). After washing away any residual antibody, the cells were incubated with an Alexa Fluor 633-labeled goat anti-human antibody (Invitrogen), and images were taken using a laser-scanning SP5 confocal microscope (Leica). For a quantification of the surface expression of gp120 on transfected 293T cells, the cells were also analyzed using the FACSCanto II flow cytometer.

NGS 262 Result in Non-infectious Virus
Particles-To study the impact of deleting the N-linked glycans at amino acid positions Asn-239, Asn-260, and Asn-354 in the HIV-1 envelope, we constructed virus strains with the asparagine at these positions mutated to a glutamine. Because the pNL4.3⌬Env_EGFP backbone contains the gene that encodes for the green fluorescent protein, successful transfection and/or infection can be easily monitored with a fluorescence microscope (Fig. 1A) and quantified using flow cytometry (Fig. 1B). The supernatant from successfully HIV-1-transfected 293T cells was used to infect several susceptible cell lines, such as the U87.CD4.CXCR4.CCR5 cell line and the T-lymphocytic cell lines C8166 and MT4. As can be seen from Fig. 1, A and B, the virus strain containing the N260Q mutation in gp120 has lost the ability to productively infect the U87.CD4.CXCR4.CCR5 cell line and the two T-lymphocytic cell lines. Identical results were obtained when the asparagine at position Asn-260 was converted into an alanine or a lysine instead of a glutamine (data not shown). To confirm that this lack of infectivity is due to the loss of the N-linked glycan at this position and not to the amino acid substitution in se, we also constructed mutant gp120 virus strains in which the serine of the glycosylation motif 260 NGS 262 was replaced by an alanine or a cysteine or in which the glycine in the middle of the glycosylation motif was replaced by an alanine (Fig. 1B). The virus strains containing mutations in the serine residue, which also result in abrogation of the glycosylation motif, were unable to infect the three susceptible cell lines. In contrast, when the glycine at position 261 was converted into an alanine, we noticed a 2-5-fold lower infectivity, depending on the nature of the infected cell lines. However, this amino acid has no direct influence on the glycosylation motif. This indicates that the presence of the Asn-260 glycan is indispensable for viral infectivity. When C8166 cells were exposed to different amounts of virus, different degrees of infection were obtained with the wild-type and mutant N239Q, N354Q, and G261A gp120 HIV-1 strains, depending on the initial virus input (Fig. 1C). However, neither the mutant N260Q nor the mutant S262A and S262C gp120 HIV-1 strains showed any sign of infection of the C8166 cells.
The infectivity of the wild-type and the mutant N260Q gp120 virus was also examined in a 2-fold dilution series in PBMCs, starting with the virus addition of 750,000 pg of p24. The experiment was performed in duplicate with PBMCs derived from two different healthy donors. In the supernatant of wild-type virus-infected PBMCs, p24 levels ranging from 700 to 6,000 pg/ml could be detected, whereas no p24 was found in the supernatant of the mutant N260Q gp120 virus-infected PBMCs.
Thus, it is clear that the Asn 3 Gln mutation at amino acid position 260 severely reduces the infectivity of the mutant virus in several different cell types, including PBMCs. Although we cannot fully exclude the possibility that changes in the amino acid composition as such in this amino acid region additionally contribute to the lower infection potential of virus strains mutated in the 260NGS262 glycosylation motif, it seems that the N-linked glycan plays a pivotal role in maintaining HIV infectivity.
Virus Strains Containing the N260Q Mutation in gp120 Show a Significantly Lower CD4 Binding Capacity-Because the first step of the entry process is the binding of gp120 to the CD4 receptor, we wanted to investigate whether destroying the glycosylation sequon at positions Asn-239, Asn-260, and Asn-354 would affect the CD4 binding of the mutant virus strains. HIV-1 virions with the N239Q, N260Q, and N354Q mutations in the envelope protein were treated with Triton X-100, which destroys the virus integrity, and with aldrithiol-2, which covalently modifies the zinc fingers in the nucleocapsid protein of HIV-1, thereby inactivating infectivity but conserving conformational and functional integrity of the virus. Fig. 2 shows that the binding of HIV-1 to CD4 under both treatment conditions is not markedly different for the N239Q and N354Q mutant virus strains but also clearly reveals that virus particles containing the N260Q mutant envelope have a significantly lower (p Ͻ 0.05) binding to the recombinant soluble CD4 in comparison with wild-type virus. Although Triton X-100 treated N239Q virus strains also have a lower CD4 binding, this decrease is not statistically significant. If a significantly lower binding of mutant virion to CD4 is observed, as in the case of the N260Q gp120 mutant virus, we would expect that multinuclear syncytia formation between such mutant HIV-1-transfected 293T cells and uninfected U87.CD4.CCR5.CXCR4 cells would not occur. Indeed, in contrast with wild-type HIV-1-transfected 293T cells (Fig. 2B, left), mutant N260Q gp120 HIV-1-transfected 293T cells were unable to induce syncytia when co-cultured with U87.CD4.CCR5.CXCR4 cells (Fig. 2B, middle).
Mutations Affecting the 260 NGS 262 Glycosylation Motif Cause Significantly Lower Env Contents in the Viral Particle-The discovery that the N260Q gp120 mutant virus strain showed a lower CD4 binding was quite unexpected, because this oligosaccharide has not been implicated in CD4 binding (22). Thus, in order to study more in detail how mutations at the asparagine 260 position in gp120 influence HIV-1 infectivity, we determined the expression levels of the envelope proteins gp120 and gp41 in lysed viral particles and came across the unexpected observation that mutations affecting the 260 NGS 262 glycosylation motif, either at the asparagine or at the serine, cause a substantial lower gp120 content in the virus particle, compared with wild-type virus (Fig. 3, A and B). The virtually complete absence of expression of both gp120 and gp41 explains the lack of infectivity of the N260Q mutant virus strain but also the lower CD4 binding observed for these mutant viruses and the inability to induce multinuclear syncytia in co-cultures of mutant N260Q gp120-transfected 293T cells and uninfected U87.CD4CCR5.CXCR4 cells (Fig. 2B).
293T Cells Transfected with pNL4.3-EGFP Containing the N260Q Mutant Envelope Express Lower Levels of gp120 on Their Cellular Surface-The observation that mutations that destroy the glycosylation motif at position Asn-260 in gp120 led to a markedly lower gp120 content in virions made us wonder what might be the underlying cause. For this purpose, 293T cells, transfected with pNL4.3, encoding the wild-type or N260Q envelope, were stained with the antibody 2G12 and an Alexa633-labeled secondary antibody (Fig. 4A). 2G12 is a monoclonal antibody that specifically recognizes well defined N-linked glycans on HIV-1 gp120. 293T cells expressing the wild-type envelope are visible under the confocal microscope as green cells with a red halo, due to 2G12 binding (Fig. 4A, left). However, when the mutant N260Q gp120 virus is transfected in 293T cells, the green cells are not encircled in red (Fig. 4A,  right), indicating that these 293T cells express markedly lower levels of mutant gp120 in their membrane compared with wild type. Although only 0.8% of the fluorescent green cells stained positive for the presence of gp120 on the surface of wild-type transfected 293T cells (Fig. 4B), the virtual absence of gp120 expression on 293T cells transfected by the mutant N260Q gp120 HIV-1 construct is striking (0.1%), confirming the confocal data in Fig. 4A.
The Asn 3 Gln Mutation at Position Asn-260 in gp120 Does Not Cause Premature Degradation of gp120 in the Endoplasmic Reticulum of the Transfected Cells-We finally investigated whether the lower gp120 expression on transfected 293T cells was due to premature degradation of misfolded mutant gp120 in the ER. N-Linked glycans play a pivotal role in the correct folding of glycoproteins in the ER, and proteins that fail to fold correctly are eliminated via ER-associated degradation (23). However, as can be seen from Fig. 4C, the intracellular expression levels of gp120 (left) and gp41 (right) in (mutant) virustransfected 293T cells are comparable in the case of both the wild-type or mutant N260Q envelope. Moreover, both gp160 and gp120 (Fig. 4C, left) or gp160 and gp41 (Fig. 4C, right) can be detected, indicating that the envelope protein was able to migrate from the ER to the Golgi apparatus, where gp160 is cleaved into gp120 and gp41 (24). It leaves open the question why gp120 is hardly expressed on the surface of the cell and suggests that trafficking of the mutant virus envelope gp120/ gp41 from the Golgi to the cell membrane could be affected by the lack of the N-glycan at amino acid position Asn-260.
Creation of Nearby N-Glycans Is Unable to Rescue the Infectivity of the Mutant N260Q Virus Strain-Because mutations affecting the glycosylation motif at position 260 NGS 262 have such a detrimental effect on virus infectivity, we wondered whether compensatory mutations, occurring before or at the same time a N260X mutation is introduced, would be able to restore the infectivity of the virus. For this purpose, a variety of mutant gp120 virus strains were constructed (Fig. 5). We changed the nearby valine at position 253 and the isoleucine at position 270 to an asparagine, which resulted in the creation of two new glycosylation motifs (i.e. 253 NST 255 and 270 NRS 272 , respectively). In addition, introducing the Q256N/L258T and the E266N/V268T mutations in gp120 created two other nearby glycosylation motifs. The generation of these two new glycosylation sites requires two mutations to occur in gp120 and thus is rather less likely to occur in vivo. The infectivity of the mutant virus strains containing the newly created N-glycosylation motifs in gp120 as such or in combination with the N260Q mutation was assessed in U87.CD4.CCR5.CXCR4, MT4, and C8166 cells (Fig. 5). The V253N mutation in gp120 itself had a negative impact on the infection rate of HIV-1, thus being very unlikely to be able to compensate for the N260Q mutation in gp120. Indeed, the V253N/N260Q mutant virus strain was still severely compromised in its infection potential (data not shown). Although changing the isoleucine at position 270 to an asparagine did not turn out to be deleterious to the infectivity of the virus, it was not able to compensate for the loss of infectivity when the N260Q gp120 mutation was combined (Fig. 5). Similar results were obtained for the Q256N/L258T and E266N/V268T gp120 virus mutants containing a newly created N-glycosylation site in proximity of the wild-type Asn-260 or mutant N260Q amino acid. The Q256N/L258T gp120 virus mutant, which was not infectious, was still severely compromised in its infectivity when the N260Q mutation was added (data not shown), whereas the E266N/V268T gp120 virus mutant that was still infectious to the cells lost its infectivity potential in the presence of N260Q. Thus, it seems to be highly unlikely that mutations in proximity of the deleted Asn-260 glycosylation  . The N260Q mutation in HIV-1 gp120 is associated with lower expression of gp120 in the membrane of transfected 293T cells but not with increased intracellular degradation of the mutant envelope. A, expression of gp120 on the surface of 293T cells, transfected with wild-type (left) or N260Q gp120 mutant (right) virus. 3 days post-transfection, the cells were stained with the 2G12 antibody and an Alexa633-labeled secondary antibody. The presence of gp120 can be seen as a red halo surrounding the green transfected cell. B, quantification of the expression of wild-type and mutant N260Q gp120 on the surface of transfected 293T cells. The cells were stained with the same antibodies as in A and analyzed using the FACSCanto II flow cytometer. Data are means Ϯ S.D. of four independent experiments performed in triplicate. C, 293T cells transfected with N260Q gp120 or wild-type virus were harvested 3 days posttransfection, and p24 levels were determined. Equal amounts of p24 were loaded in 2-fold serial dilutions. Western blot analysis using an anti-gp120 antibody (left) or an anti-gp41 antibody (right) revealed that the N260Q mutation has no significant effect on the intracellular expression of gp160 and the cleavage into gp120 and gp41, compared with wild-type virus. The expression level of the cellular GAPDH protein served as a loading control. motif in gp120 could compensate for its detrimental effects on the infectivity potential of the virus.

Mutant gp120 Virus Strains Do Not Show a Higher Susceptibility to the Antiviral Activity of the Carbohydrate-binding
Agents-It was previously observed that the removal of N-linked glycans in the V1/V2-loop of HIV-1 gp120 affords a higher sensitivity to the antiviral properties of CBAs (14). These results could partially explain why the glycans of the V1/V2loop were (almost) never deleted under CBA pressure. Our results provide an explanation why a deletion of the Asn-260 glycan has never been observed. However, the question as to why the Asn-239 and Asn-354 glycans were never found mutated under increasing CBA pressure remains unanswered. To investigate whether a higher susceptibility of such mutant viruses to the CBAs may be the cause, the inhibitory activity of the ␣(1,3)/␣(1,6)-mannose-specific HHA, the ␣(1,3)mannose specific GNA, the GlcNAc-specific UDA, and the ␣(1,2)-specific non-peptidic antibiotic PRM-A was determined for the mutant viruses, derived from the virus-infected U87.CD4.CCR5.CXCR4 supernatants (supplemental Table   S1). The antiviral activity of the CBAs against the mutated virus strains was generally comparable with the antiviral activity against the wild-type virus. Only the lower antiviral activity of HHA against the mutant N239Q gp120 virus strains was significantly (p Ͻ 0.05) but still very modestly different (2.5-fold less active) from that against the wild-type virus. It therefore seems that a potential increased antiviral activity of the CBAs against the mutant virus strains is not the underlying cause of the fact that the Asn-239, Asn-260, and Asn-354 glycans were never found deleted upon increasing CBA pressure in cell culture.

Combinations of the N239Q and N354Q Mutations with Other Glycosylation Site Mutations Can Result in Virus Strains
with Seriously Impaired Infectivity-In order to further study the Asn-239 and Asn-354 glycans, we constructed double mutants by combining the N239Q and N354Q mutations in gp120 with mutations at position Asn-228 and Asn-337. The latter two glycosylation sites were most often found deleted in HIV strains selected under pressure of the CBAs HHA, GNA, UDA, cyanovirin-N, PRM-A, and pradimicin S (supplemental Fig. S1) (9 -11, 25-27), which might indicate that they do not play a vital role in the maintenance of viral infectivity. The rationale for this experiment was to evaluate the possibility that one single glycan deletion may not be essential for compromising virus infectivity, but elimination of the same glycan may become of crucial importance when another glycan is concomitantly eliminated. Indeed, the combination of the N354Q mutation with the N337Q mutation resulted in a complete loss of infectivity in U87.CD4.CCR5.CXCR4, MT4 and C8166 cell cultures (Fig. 6A). Also, the mutant N239Q/N337Q gp120 virus strain had a significantly (p Ͻ 0.05) lower infectivity in MT4 cells compared with wild-type virus. The significantly lower infectivity of the double mutant containing the N354Q/N337Q mutants can be explained by the absence of any gp120 in the virion, as shown by Western blot analysis of virions obtained from transfected 293T cells (Fig. 6, B and C). These findings could, at least in part, explain why the Asn-239 and Asn-354 N-linked glycans are less likely to be deleted under CBA pressure.

DISCUSSION
The selection of HIV-1 strains in cell culture under pressure of CBAs invariably results in the deletion of a variety of different N-linked glycosylation sites in gp120 (9,10,25,26). Because gp120 is one of the most densely glycosylated proteins known so far, containing on average ϳ24 N-linked glycans, one might expect that these glycans are equally important to attain the right conformation and function of the envelope protein. However, we previously reported that the complex-type glycans present in the V1/V2-loop of gp120 are rarely deleted under CBA pressure, probably due to the higher susceptibility to the antiviral activity of CBAs and the compromised replication capacity when the glycosylation motifs of the V1/V2-loop were destroyed (14). Here we report that three other glycans, scattered over the other parts of the HIV-1 gp120 envelope, were also never mutated in more than 60 HIV-1 strains independently selected under increasing pressure of CBAs, such as HHA, GNA, UDA, actinohivin, microvirin, PRM-A, or pradimicin S (9,10,(25)(26)(27)(28)(29). Although our initial research started with the investigation of the effect of mutating the glycans at positions Asn-239, Asn-260, and Asn-354 on viral infectivity and CBA sensitivity, we realized that the Asn-260 glycan of gp120 is of the utmost importance for the preservation of viral infectivity; therefore, we concentrated most of our further work on the study of this important glycan. The N-linked glycan at position 260 is highly conserved among a wide variety of HIV subtypes (10,14), pointing to the pivotal role of this glycan for the structural and/or functional integrity of the virus. It is located in the middle of a peptide stretch (residues 251-265) that is also highly conserved among a wide variety of HIV-1 strains and clades (supplemental Fig. S2).
Several other research groups have already observed the detrimental effects of mutating Asn-260 in the HIV-1 gp120 (30 -32). However, only Willey et al. (30) investigated this interesting phenomenon further in detail. Although they also noticed that replacing the Asn at position 260 by a Gln resulted in noninfectious virus, they came across several other observations that were quite opposite to those reported in our study. For instance, Willey et al. (30) suggested that not the removal of the glycan, but the substitution of the asparagine by a glutamine, caused the lack of infectivity. Our mutagenesis studies in which we replaced the Asn by a Gln, an Ala, or a Lys, together with our data on the mutagenesis of the Gly-261 and Ser-262 residues of the glycosylation motif, probably indicate that the presence of the N-linked glycan is a necessity for a productive HIV-1 infection. It is not clear why the G261A mutation in gp120 also results in a lower infectivity. The glycine at position 261 is highly conserved among a wide variety of HIV-1 strains (supplemental Fig. S2). Perhaps the substitution of the smallest amino acid, glycine, by alanine, containing a bulkier side group, may sterically hinder the optimal amino acid interactions and eventual local conformation of this area of the gp120 molecule.
In our search for an explanation why this Asn-260 glycan is of such importance for viral infectivity, we first looked at the ability of mutant gp120 virus strains to bind recombinant sCD4 and found that the binding was compromised. This observation is in contrast to the results published by Willey et al. (30), who did not observe a lower CD4-binding. A possible reason for these different observations might be the different assays employed to measure CD4 binding (i.e. the use of truncated recombinant monomeric gp120 (30) versus whole virus particles as a source of gp120, which mimics more closely the in vivo situation (our study) (33).
The lower CD4 binding of the N260Q gp120 mutant virus strain provided us with an explanation for the annihilated infectivity and the inability of mutant virus-transfected 293T cells to  (error bars). B, Western blot of virion-associated gp120 expression of double mutant gp120 virions, produced in 293T cells. Viral pellets were obtained as described in the legend to Fig. 3A. The double mutant N354Q/N337Q gp120 virus contained hardly detectable gp120 in its envelope. C, band intensities of the Western blot in B were quantified using the program ImageJ.
induce syncytia when co-cultured with uninfected U87.CD4.CCR5.CXCR4 cells but also raised new issues, because, to our knowledge, the Asn-260 glycan was never reported to be directly involved in CD4 binding (22,34). So how can deletion of a glycan on gp120 that is not involved in CD4binding cause a significant lower CD4-binding? Determining the gp120 levels in the mutant virus solved this issue. Indeed, virions containing the N260Q mutation in their envelope expressed markedly lower levels of gp120 and gp41 in the virus envelope compared with wild-type virions. Thus, a virtual lack of gp120 in the virus envelope obviously prevents efficient CD4 binding.
The lack of significant amounts of gp120 and gp41 in the envelope of mutant N260Q gp120 HIV-1 was also observed for the double N337Q/N354Q gp120 mutant HIV-1 (Fig. 6), pointing to a similar mechanism of lack of virus infectivity with these mutant virus strains. The phenomenon that two single envelope glycan mutations (such as N337Q and N354Q) that do not significantly affect virus infectivity when solely present as a single mutation severely affect virus infectivity when both mutations are concomitantly present in the viral envelope has been previously also observed by Auwerx et al. (14) for glycan mutations in the V1/V2 domain of HIV-1 gp120. In this study, the single N156Q and N186Q envelope mutations virtually did not affect virus infectivity, but when combined, they resulted in a more than 20-fold decreased infectivity of the double mutant virus strain. Thus, certain N-glycans seem to be able to compensate for the absence of other glycan(s) to keep viral infectivity, but when they are absent at the same time, the combination of two well defined glycan deletions may markedly compromise the virus infectivity.
The next step was the investigation of the expression of gp120 and gp41 in the HIV-1-transfected 293T cells. We first examined whether the mutant envelope protein is degraded in the ER due to protein folding problems. Indeed, it has previously been shown that the removal of a single glycan can be sufficient to impair the folding process. For instance, removal of the oligosaccharide at position Asn-81 of influenza hemagglutinin (HA) causes premature degradation of this glycoprotein, due to the inability of the mutant HA to oxidize and form disulfide bridges in the proximity of the glycan (35). Of particular interest is the finding by Pikora et al. (36) that the deletion of a single N-glycan at the glycosylation site 284 NRT 286 of SIV239 gp120 results in a complete loss of infectivity, caused by a lethal defect in Env processing and incorporation. Although HIV-1 and SIV Env share a high degree of sequence similarity (37), the Asn-284-based glycosylation motif in SIV pg120 does not correspond to the Asn-260-based motif in HIV-1 gp120. However, deletion of the oligosaccharide in SIV gp120 that does correspond to the Asn-260 glycan resulted in a markedly decreased infectivity (5% compared with wild type), suggesting that also in SIV, the glycan at this position is vital for viral infectivity. Thus, there are already several examples in the literature of a single glycan deletion causing the premature degradation of an entire glycoprotein. Would this also be the case with the N260Q mutation in HIV-1 gp120? When cell extracts of transfected 293T cells were analyzed, we clearly observed that both gp160 and gp120 were present in 293T cells, transfected with mutant N260Q gp120 virus (Fig. 4C). Because the cleavage of gp160 into gp120 and gp41 was demonstrated to occur late in the Golgi apparatus (24), this implies that the envelope protein had been translocated from the ER to the Golgi apparatus in its intact form and thus was not degraded by ER-associated degradation in the ER. This observation rules out the possibility that protein folding problems were the reason of the lack of gp120 and gp41 in the membrane of virus-transfected 293T cells.
In order to investigate whether the mutant envelope protein is expressed in the ER of the transfected 293T cell but does not efficiently reach the surface of the cell to enable budding virus particles to imbed proper gp120 levels in their envelope, we labeled the envelope proteins on the surface of transfected 293T cells with the glycan-specific monoclonal antibody 2G12. These experiments revealed that 293T cells transfected with the plasmid pNL4.3-EGFP expressing the N260Q mutant envelope expressed lower levels of gp120 in their cell membrane compared with 293T cells transfected with wild-type virus. Although these results might explain our observation that upon co-culturing uninfected U87.CD4.CCR5.CXCR4 cells with mutant N260Q gp120 virus-transfected 293T cells, no giant cell formation is observed, we cannot exclude the possibility that the N260Q mutation in the HIV-1 envelope affects the epitope of the 2G12 antibody. However, several research groups have independently determined the epitope of 2G12, and none have reported the involvement of the Asn-260 glycan (38 -40). Taking all data together, we provided strong evidence that the N-glycan at amino acid position 260 of the HIV-1 gp120 envelope plays an instrumental role in the eventual incorporation of gp120 in the virion membrane rather than preventing correct folding of the gp160 envelope molecule in the endoplasmic reticulum.
The trafficking of Env from the Golgi to the cellular membrane and incorporation into the virion is a process poorly understood. Until now, only gp41 has been reported to interact with several cellular proteins for trafficking to the cellular membrane (for an overview, see Ref. 41). However, glycosylation serves various functions after glycoprotein expression, and the correct trafficking of the glycoprotein to the cell surface is one of these functions. Indeed, in the case of the potassium channels Kv1.1 and Kv1.4, it has been reported that prevention of glycosylation leads to a decreased expression of the channels on the cellular surface (42). In this respect, it would be of interest to investigate a potential altered interaction of wild-type and mutant HIV-1 gp120 with known chaperone molecules of the Golgi apparatus. Further investigations will be needed to reveal how the deletion of the Asn-260 glycan of gp120 affects trafficking and incorporation in the plasma membrane.
Targeting the protective glycan shield of HIV-1 by CBAs has recently generated a lot of attention as a new antiviral therapeutic approach, and several CBAs are being investigated as possible microbicides (43,44). In particular, CBAs with mannose specificity look promising, because the envelope of gp120 has an unusually high density of clustered high mannose type glycans, which is rare, if it exists at all, on human glycoproteins. Our studies now reveal that the high mannose type glycan at amino acid position Asn-260 is a possible target for these CBAs. HIV-1, in an attempt to escape CBA pressure, seems unable to mutate this position, because this would result in a complete loss of viral infectivity. Moreover, the glycan on Asn-260 is located in a peptide stretch of the C2 domain of gp120 that is remarkably conserved among the different HIV-1 strains and clades (supplemental Fig. S2). It would therefore be interesting to develop site-specific antibodies (in particular nanobodies) that specifically target the glycan at position Asn-260 of gp120 or synthetic CBAs that selectively target this glycan position on gp120.
Targeting specific glycans of gp120 is not impossible. The neutralizing antibody 2G12 is able to recognize an epitope constituted by 3-5 N-linked glycans (38,40,45), and very recently, new highly potent broadly neutralizing antibodies with carbohydrate specificity have been described (46). These observations indicate that it should be possible to raise antibodies against the Asn-260-linked glycan and to afford a pronounced neutralizing activity by targeting this newly identified hot spot on HIV-1 gp120. This would represent a novel approach to target the viral gp120 envelope through chemotherapeutic and/or immunological means that are specifically directed against a well defined site of the protective glycan shield of the HIV particle.
In conclusion, we have identified a hot spot N-glycosylation motif on HIV-1 gp120 that is indispensable for an efficient viral entry and infectivity and demonstrated that the lack of infectivity potential of the mutant Asn-260 gp120 virus strain is caused by a severely reduced expression level of the envelope glycoproteins gp120 and gp41 in the viral envelope. The Asn-260 position is located in a highly conserved amino acid stretch of HIV-1 gp120, and its glycan cannot be functionally replaced by the introduction of another N-glycan in the area near this amino acid, further emphasizing the crucial importance of the specific presence of this glycan at amino acid position Asn-260. Selectively targeting the N-glycan at amino acid position 260 of HIV-1 gp120 may be an interesting novel chemo-/immunotherapeutic approach to prevent HIV-1 from binding and entry into its target cells, resulting in suicidal therapy if the virus wants to delete this glycan to escape specific drug pressure.