Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The novel virucidal protein cyanovirin-N (CV-N), an 11-kDa protein originally purified from extracts of the cultured cyanobacterium Nostoc ellipsosporum, irreversibly inactivates diverse strains of human immunodeficiency virus (HIV) (Boyd et al., 1997). CV-N does not share any significant sequence or three-dimensional structural homology with any other known protein (Boyd et al., 1997; Bewley et al., 1998; Yang et al., 1999). CV-N is increasingly of interest for developmental therapeutics and prophylaxis of HIV infection (Boyd et al., 1997; Esser et al., 1999; Dey et al., 2000). Although the exact mechanism by which the protein inactivates HIV has not yet been elucidated, converging evidence indicates that the viral surface envelope glycoprotein gp120 is a molecular target of CV-N (Boyd et al., 1997).

Initial enzyme-linked immunosorbent assay studies from our laboratory indicated that CV-N bound specifically to a soluble form of glycosylated gp120 (sgp120) and the analogous simian immunodeficiency virus proteins sgp130 and sgp140, but not to nonglycosylated recombinant gp120 (Boyd et al., 1997). More recently, isothermal titration calorimetry (ITC) and optical biosensor binding studies showed that CV-N bound recombinant sgp120 with high affinity (K_d = 2-45 nM), and that the CV-N-gp120 binding process was enthalpically driven with large exothermic heats of binding (H = 20.2 kcal/mol) (O'Keefe et al., 2000). The thermodynamics indicates that a large number of bonding interactions (e.g., hydrogen bonds and van der Waals interactions) occur during complex formation, and in the context of the highly glycosylated gp120 such interactions may also be a result of polar/electrostatic contacts between the carbohydrates of gp120 and charged/polar amino acids of CV-N. Recent studies revealed that CV-N preferentially bound to radiolabeled high-mannose oligosaccharides (Bolmstedt et al., 2001). In the present article, we demonstrate that CV-N recognizes oligosaccharides on glycoproteins, and determine the specificity of this recognition. After establishing that CV-N binding by glycoproteins is indeed a carbohydrate-mediated phenomenon, the differential binding by CV-N to glycoproteins, which contain either complex or high-mannose oligosaccharides, is assessed. Following this, the level of CV-N-binding specificity among specific oligosaccharides is tested, to determine whether CV-N has the capacity to discriminate the slight differences in structure among these oligosaccharides. Lastly, the interaction between CV-N and the high-mannose oligosaccharide oligomannose 8 (Man-8) is biophysically characterized, such that a general thermodynamic model can be developed to form the basis for future analysis of the CV-N-sgp120 binding mechanism.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Endoglycosidase H (Endo H) Deglycosylation and Purification of sgp120 and RNase B. HIV-1_IIIB sgp120 (200 µg; ABI, Columbia, MD) or RNase B (1 mg; Sigma, St. Louis, MO) was taken up in 0.4 ml of 0.5 M NaCl, 100 mM sodium acetate buffer, pH = 5.7. To these solutions either 100 mU (gp120) or 20 mU (RNase B) of Endo H (Glyko, Novato, CA) was added in 20 µl and the reaction mixture was incubated overnight at 37°C. Following deglycosylation, the reaction mixtures were centrifuged on a 3-kDa ultrafiltration membrane (Amicon, Beverly, MA) to separate the deglycosylated glycoproteins from released oligosaccharides. The proteins were then fractionated by C18 reverse phase high-performance liquid chromatography and characterized by SDS-polyacrylamide gel electrophoresis, and in the case of RNase B, LC-MS was also used to verify that the deglycosylation was successful.

Isothermal Titration Calorimetry. All experiments were carried out using an Omega titration calorimeter, at 30°C. In a typical experiment 10-µl aliquots of CV-N were injected from a 250-µl syringe into 1.396 ml of macromolecule (₁ acid glycoprotein, soybean agglutinin, RNase B, deglycosylated RNase B, or deglycosylated sgp120), rapidly mixing at 400 rpm with a 5-min equilibration time between injections. A total of 20 injections was made. Heats of reaction were determined by integration of the observed peaks. All isotherms were corrected for the heat of mixing and/or dilution by subtraction of the isotherm obtained following injection of CV-N into buffer. All protein solutions were constituted in 50 mM sodium phosphate, 0.2 M NaCl buffer containing 0.02% NaN₃ and 0.02% Triton X-100, pH 7.5.

Circular Dichroism. CD spectra were obtained with a Jasco J-720 spectropolarimeter scanning from 290 to 190 nm at a rate of 10 nm/min. A 1.0-mm path length optical cell was used. Spectra were obtained at room temperature with a 1.0-nm bandwidth, and a typical averaging time of 0.25 s/step. Sample spectra of 6 µM ₁ acid glycoprotein, 2.5 µM soybean agglutinin, and a range of concentrations of CV-N (2.5-12.5 µM) were individually measured. All spectra were corrected for buffer effects by subtracting the spectra of 25 mM sodium phosphate buffer, pH 7.0, from each sample spectrum. The corrected sample spectra were normalized and the mean residue ellipticity conversion was performed according to the Jasco J-700 software protocol. Mean residue molar concentrations (C_r) of ₁ acid glycoprotein, soybean agglutinin, and CV-N were calculated according to the equation C_r = n × C_p, where n is the number of amino acid residues in a sample protein and C_p is the molar concentration of the protein. Values for n used in the calculations were 183 for ₁ acid glycoprotein, 1080 for soybean agglutinin, (tetramer), and 101 for CV-N.

Affinity Chromatography. Rabbit anti-CV-N polyclonal antibodies (Boyd et al., 1997) were coupled to a Protein A-Sepharose (CL-4B) (Amersham Pharmacia Biotech, Uppsala, Sweden) and loaded into two 3-ml microcolumns according to the manufacturer's instruction. Thereafter, the columns were washed with MAPS II binding buffer (Bio-Rad, Hercules, CA) (30 ml) and 2 ml of 136 µM CV-N (3 mg) was added to just one column; no CV-N was introduced to the mock column. After 2 h, the columns were washed with binding buffer (15 ml). Next, using the stoichiometry value obtained from ITC experiments of CV-N and RNase B, 0.109 µmol of RNase B was added to both columns, after which the columns were washed with binding buffer and the washes collected in 1-ml fractions for MS analysis.

Mass Spectroscopic Analysis. The fractions from both the CV-N and the mock affinity chromatography columns were analyzed by LC-MS using a Hewlett Packard high-performance liquid chromatography/electrospray ionization quadrupole mass spectrometer (model 1100C) running a C18 Zorbax column (2.1 × 110 mm). After mass spectral deconvolution according to manufacturer's protocol, a comparison of the RNase B isoform profiles eluting from the mock or the CV-N column was done. The five RNase B isoforms correspond to RNase B differentially glycosylated with either Man-5, Man-6, Man-7, Man-8, or Man-9. Deconvolution of the total ion current resulted in a relative abundance profile of the RNase B isoforms. An RNase B isoform MS profile of a sample that had not been subjected to affinity chromatography was also measured as a control for the mock column sample. A set of three mass spectroscopic analyses was performed for each of the two separate affinity chromatography experiments. Standard errors were calculated for the data and an unpaired t test was performed to determine whether there was a significant difference in the elution patterns of the mock and CV-N columns.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Binding Studies of CV-N with Glycosylated and Deglycosylated sgp120 and RNase B. The technique of ITC has been previously used in the biophysical characterization of sgp120 binding by CV-N (O'Keefe et al., 2000). The initial experiments had shown that CV-N bound sgp120 with a K_d value of 37 nM and with a stoichiometry of about 5:1. To assess whether the CV-N/sgp120 binding interaction was carbohydrate mediated, calorimetric experiments were similarly used here to study the interaction of CV-N and Endo H-deglycosylated HIV-1 sgp120. Endoglycosidase H is an enzyme that specifically cleaves high-mannose and hybrid-type N-linked oligosaccharides from glycoproteins, leaving intact the complex oligosaccharides. Following treatment with Endo H, SDS-polyacrylamide gel electrophoresis analysis of the sgp120 indicated that the deglycosylated product had a molecular mass of approximately 85 kDa (data not shown). From subsequent ITC analysis, it was apparent from the isotherm of the binding of CV-N and Endo H-deglycosylated sgp120 that, after correction for the heat of dilution of CV-N into buffer, only a basal level of binding energy was detected by the instrument. The nonlinear fitting software did not enable the deconvolution of such a baseline experiment (Fig. 1A). This result showed that sgp120, after Endo H deglycosylation, was no longer recognized by CV-N, indicating that the binding between CV-N and sgp120 is a carbohydrate-mediated process.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   A comparison of the calorimetric titration of CV-N into glycosylated- versus Endo H-deglycosylated sgp120 (A) and RNase B (B). In both panels, CV-N binding of the deglycosylated () glycoproteins result in baseline thermograms in contrast to the tight, sigmoidal curves observed for the CV-N binding of the glycosylated versions () of sgp120 and RNase B.

Binding Studies of CV-N to Glycoproteins with Either All Complex Oligosaccharides or All High-Mannose Oligosaccharides, and Conformational Structure Analysis of the Binding. Previous studies had indicated that complex oligosaccharides on the surface of gp120 might not participate in the CV-N-binding binding process (O'Keefe et al., 2000; Bolmstedt et al., 2001). To further test whether only high-mannose oligosaccharides are recognized by CV-N, ITC binding experiments were designed to look at the interaction of CV-N with glycoproteins that either carry only complex oligosaccharides or only high-mannose oligosaccharides. The protein ₁ acid glycoprotein was chosen for its content of only complex oligosaccharides, and soybean agglutinin was chosen for its homogeneous content of oligomannose 9.

View larger version (14K): [in this window] [in a new window]   Fig. 2.   CD spectra of the CV-N complexes of 1 acid glycoprotein (A) and of soybean agglutinin (B). A, dashed line, theoretical noninteracting spectrum of a mixture of 6.0 µM 1 acid glycoprotein/6.0 µM CV-N; solid line, experimental spectrum of a mixture of 6.0 µM 1 acid glycoprotein/6.0 µM CV-N. B, dashed line, theoretical noninteracting spectrum of a mixture of 2.5 µM soybean agglutinin/2.5 µM CV-N; solid line, experimental spectrum of a mixture of 2.5 µM soybean agglutinin/2.5 µM CV-N. Calculations for theoretical noninteracting spectra are detailed under Materials and Methods.

Specific Binding of CV-N to Individual High-Mannose Oligosaccharides. To determine whether CV-N possessed any differential specificity for binding the different high-mannose oligosaccharides, a novel experimental system was designed using a combination of affinity chromatography and mass spectrometry. Since earlier experiments showed that CV-N binding by RNase B was dependent on the single high-mannose oligosaccharide present on RNase B (Fig. 1B), and since the RNase B (containing Man-5, Man-6, Man-7, Man-8, or Man-9) can be easily characterized by LC-MS, it was possible to dissect CV-N's binding specificity for particular glycoforms of RNase B. The experiments allowed for the characterization of the high-mannose RNase B glycoforms that were specifically retained on a CV-N affinity column. Comparing the RNase B isoform distribution of a sample fraction (Fig. 3B) washing off a CV-N column with that of a fraction from a null column (Fig. 3A), a significant reduction (p < 0.0001) of the Man-8 glycoform and an almost complete reduction (p < 0.0001) of the Man-9 glycoform was observed in the elution profile of the CV-N column.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Mass spectral profile of the various RNase B glycoforms in the wash fractions of the mock affinity (A) and CV-N affinity (B) columns. A, relative abundance distribution of RNase B glycoforms eluting from an affinity column in the absence of CV-N. B, relative abundance distribution of RNase B glycoforms eluting from an affinity column in the presence of CV-N. Results shown are the average of three sets of mass spectroscopic analyses of two separate affinity chromatography experiments.

Characterization of the Binding between CV-N and Man-8. Although affinity chromatography had indicated that Man-8 and Man-9 had approximately equal affinity for CV-N, previous fluorescence polarization experiments (Bolmstedt et al., 2001) had shown that Man-8 was a slightly better inhibitor of the CV-N-sgp120 interaction. Therefore, Man-8 was chosen for further binding experiments. It was apparent from the binding isotherm of Man-8 and CV-N that the binding process was enthalpically driven (H = 28.3 kcal/mol) (Fig. 4; Table 2) with large exothermic heats of binding, similar to those observed in the studies of CV-N binding with sgp120, RNase B, and soybean agglutinin (Table 1). This comparable H value further supports the supposition that the negative enthalpic binding observed with CV-N and glycoproteins is due to interaction of oligosaccharides of the glycoproteins and protein surfaces of CV-N.

View larger version (32K): [in this window] [in a new window]   Fig. 4.   Calorimetric titration of CV-N with oligomannose Man-8. A, instrument feedback (µcal/s) necessary to maintain a constant temperature (30°C) in the calorimeter cell as Man-8 is periodically titrated into a rapidly mixing solution of 46.4 µM CV-N, pH 7.5, at 30°C. These results are then converted (B) to the binding isotherm. Saturation is achieved as the concentration of Man-8 in the calorimeter cell reaches a high molar ratio relative to the amount of CV-N.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The HIV envelope glycoprotein gp120 is heavily glycosylated, with carbohydrates representing approximately half the molecular weight of the glycoprotein (Geyer et al., 1988). The surface carbohydrates of gp120 are known to play a major role in determining the accessibility of the structural binding domains of gp120 to CD4 and other proteins (Biller et al., 1998). It is not entirely surprising therefore that our results with CV-N-gp120 binding also show a carbohydrate dependence. The "canyon theory" of gp120 receptor binding proposes that receptor recognition by gp120 occurs within a molecular canyon, narrowly rimmed with variable amino acids that shield the highly conserved amino acids at the base of the receptor binding pocket (Rossman, 1989). Since the oligosaccharides in the conserved regions are shielded from processing enzymes, they are less processed than the oligosaccharides at the variable regions (Yeh et al., 1993). Thus, the architectures of the variable and conserved domains of gp120 are spatially and chemically different, with highly processed oligosaccharides (complex-type) at the variable regions and less processed oligosaccharides (high-mannose-type) near the conserved regions (Yeh et al., 1993). The canyon theory predicts that proper topological discrimination of these regions is essential for effective neutralization of gp120.

CV-N preferentially recognizes high-mannose, N-linked oligosaccharides on glycoproteins, and binding between CV-N and sgp120 is nil in the absence of these oligosaccharides. Our ITC and CD spectral experiments of CV-N binding to soybean agglutinin and ₁ acid glycoprotein showed that CV-N could bind discriminately to a particular glycoprotein based on that glycoprotein's content of high-mannose oligosaccharides. This is consistent with previous results that showed CV-N binding with equal affinity to both the baculovirus-expressed sgp120 and the H9 cell line-produced sgp120, suggesting that the additional presence of O-linked and complex-type N-linked oligosaccharides on H9 gp120 do not contribute to gp120 binding by CV-N (O'Keefe et al., 2000). The ability of CV-N to distinguish between high-mannose oligosaccharides and complex oligosaccharides may account, at least in part, for the specificity of binding of CV-N to gp120 and the potent neutralization of HIV. In applying the canyon theory to our results, we can speculate that, since CV-N-binding is dependent on the surface presentation of high-mannose oligosaccharides on sgp120, CV-N must bind near or within conserved protein epitopes of gp120. Conceivably, this could account for the broad antiviral activity of CV-N against diverse HIV-1 isolates and related retroviruses (Boyd et al., 1997).

The exact CV-N binding region on gp120 is yet to be determined. However, the accumulated results are all consistent with carbohydrate dependence. Previous studies showed that CV-N did not alter epitope availability in either the V3 loop or the CD4 binding region of gp120 (Boyd et al., 1997; Esser et al., 1999). To date, the binding of only one anti-gp120 monoclonal antibody, namely, that of the glycosylation-dependent antibody 2G12, has been shown to be blocked by CV-N (Esser et al., 1999). The 2G12 antibody is known to bind a neutralization epitope on the heavily glycosylated "silent face" of gp120 (Kwong et al., 1998; Etemad-Moghadam et al., 1999), and recent evidence has indicated that the binding is dependent on the presence of conserved carbohydrate structures on gp120 (Wyatt and Sodroski, 1998). Perhaps the gp120 binding sites of CV-N and 2G12, although not identical (Esser et al., 1999), may overlap in regions containing high-mannose oligosaccharides.

Our present affinity chromatography results showed that the Man-5, Man-6, and Man-7 glycoforms of RNase B did not bind CV-N but that the Man-8 and Man-9 glycoforms were preferentially retained on the CV-N affinity column. Similar evidence from fluorescence polarization experiments showed that only Man-8 and Man-9 selectively blocked the binding interaction between CV-N and sgp120 (Bolmstedt et al., 2001). The collective results suggest that a highly specific oligosaccharide-binding region exists within CV-N that preferentially accommodates only the Man-8 and Man-9 oligosaccharides. The localization of such a binding site on CV-N will be of interest for future structural studies. The present study sets the stage by elucidating the remarkable binding interaction between CV-N and a single high-mannose oligosaccharide.

Our titration calorimetric binding experiments with CV-N and Man-8 revealed that the binding was enthalpically driven and strong (K_d = 0.488 µM). In general, oligosaccharides and protein lectins overcome weak individual interaction free energies of binding (K_d = 1 mM-1 µM) (Mandal et al., 1994) by mediating multivalent binding with one another (Rice and Lee, 1993; Dimick et al., 1999). Such a multivalent binding effect may help to explain the unusually tight binding observed for Man-8 and CV-N. Indeed, examination of the calorimetric cell contents after titration of Man-8 into CV-N revealed a slightly cloudy solution, suggesting that cross-linked precipitates may have formed as a result of multivalent binding. The multivalent effect is further supported by the negative entropy value obtained for the experiment, which indicates that an increase in molecular ordering has occurred. The calculated stoichiometry value of 0.775 (CV-N:Man-8) obtained in our ITC experiments is difficult to interpret. Although the value may suggest an approximate 1:1 (univalent) binding between CV-N and the oligosaccharide, an accurate interpretation of the stoichiometry must await thorough structural examination of CV-N in complex with Man-8.

CD studies of the Man-8-CV-N complex indicated that only a slight conformational change had occurred in CV-N upon binding Man-8 with a calculated approximate gain of 6% -sheet and losses of 4% -helix and 1% random coil structures. Since these estimated structural changes are slight compared with the calculated net gain of 23.3% -sheet and losses of 12.6% -helix and 11% random coil structures in the soybean agglutinin-CV-N complex, we can speculate that CV-N, upon binding the Man-9 oligosaccharide of soybean agglutinin, remains structurally intact while inducing gross conformation changes in the protein backbone of soybean agglutinin. Interestingly, the evidence of a significant decrease in random coil in the structure of the soybean agglutinin-CV-N complex suggests that soybean agglutinin may have gained more protein secondary structure as a consequence of binding CV-N. Similarly, we can compare our present CD analysis of the Man-8-CV-N interaction with that of our previous CD spectral studies of the binding between CV-N and either sgp120 or sgp41 (O'Keefe et al., 2000). Here again, we can speculate that conformational changes occurring in the CV-N complexes of these glycoproteins are largely due to structural rearrangements in sgp120 and sgp41 and not as much in CV-N.

It is possible that, in addition to the interaction of specific oligosaccharides and CV-N, discrete protein-protein interactions may also contribute to the extremely tight binding of CV-N with gp120. Indeed, a comparison of the CV-N-binding abilities of Man-8 and sgp120 (Table 2) indicates that sgp120 mediates a 13-fold greater affinity with CV-N than does Man-8. Protein-protein contacts may play an important ancillary role in the CV-N-sgp120 binding by reinforcing and securing the initial carbohydrate-protein contacts between the high-mannose oligosaccharides of gp120 and CV-N. The possible importance of protein-protein contacts is suggested by the comparatively less negative (more positive) H and S values of the sgp120-CV-N binding. At first glance, the more positive H seems to indicate that sgp120 makes less polar/electrostatic interactions with CV-N than Man-8, which is difficult to conceive since sgp120 has 24 oligosaccharides on its surface and Man-8 has just the one oligosaccharide. However, the calorimeter is a global sensor of heat, and as a result the overall enthalpy measured by the instrument is actually a result of the partial negative enthalpy, contributed by polar/electrostatic interaction, and the partial positive enthalpy, contributed by hydrophobic (protein-protein) contacts. Thus, the overall less negative H value observed for the sgp120/CV-N binding is not likely a result of fewer polar contacts made by these two proteins, but perhaps due to a greater number of hydrophobic contacts made by their protein backbones. A further suggestion that protein-protein contacts may play a significant role in the sgp120-CV-N interaction is found in the less negative overall S value observed for the binding, compared with the S value of the Man-8 experiment (Table 1). The more positive S value may reflect a general increase in entropy of the numerous water molecules in the binding solution, which presumably increases when these molecules are excluded from protein-protein interfaces of two interacting proteins.

In conclusion, we have defined CV-N's preference for binding high-mannose oligosaccharides and have biophysically characterized the binding interaction between Man-8 and CV-N in an attempt to dissect the glycosylation-dependent binding of CV-N and HIV gp120. These results will contribute to and focus future structural elucidation of CV-N binding site(s) on gp120 and oligosaccharide-binding site(s) on CV-N. CV-N binds preferentially to high-mannose oligosaccharides, and as such may only recognize epitopes on gp120, which present these oligosaccharides. Since it is suspected that CV-N blocks binding of the unique HIV-neutralizing monoclonal antibody 2G12 at the immunologically silent face of gp120 (Esser et al., 1999), examination of high-mannose-containing epitopes in this region of gp120 may reveal one or more CV-N binding sites. Future X-ray diffraction or NMR structural studies of CV-N in complex with a high-mannose oligosaccharide may help to uncover the particular architectural basis for the remarkably high-affinity interaction between CV-N and HIV gp120.