Abstract: Immobilised neoglycoconjugates covalently cross-linked into polyacrylamide gel, can be used to detect and characterize carbohydrate-binding proteins. The neoglycoconjugates comprise two active groups, saccharide and allyl, located on a poly(2-hydroxyethylacrylamide) backbone. The allyl group cross-links with the polyacrylamide gel matrix, while the saccharide groups are available for specific protein interactions. This neoglycoconjugate-gel is prepared as a thin layer within the stacking region of a polyacrylamide gel, and electrophoresis is performed according to native, non-denaturing conditions. Carbohydrate-binding proteins, specific for the immobilized neoglycoconjugates, are thus retarded during electrophoresis, while simultaneously permitting the separation of non-binding proteins according to size and charge.

Figure 1. Structure of the neoglycoconjugates used in CA-PAGE. The structure shows the composition of the neoglycoconjugate with a representative saccharide and allyl group linked to the acrylamide polymer backbone (blue coil).

Preparation of Carbohydrate Affinity Gels (CA-PAG)
All electrophoresis experiments were performed using a mini-gel apparatus (SE250, Hoeffer Scientific Instruments, CA). The preparation of single concentration or gradient resolving gels comprising a high pH discontinuous Tris buffer system (pH 9.5) were prepared as described elsewhere [1]. The stacking gel however was made in two polymerisation steps; the affinity gel layer, then the normal stacking gel overlay. A stacking gel solution was prepared comprising 5% acrylamide, 0.03% ammonium persulphate in 0.125M Tris-HCl buffer (pH 6.8), and degassed for 10 minutes. The affinity gel comprised 0.2 ml of this solution mixed with up to 213 mg/ml neoglycoconjugate, together with 0.1% TEMED. The solution was immediately added to the gel cassette to form a 5mm overlay that covers half of the top surface of the resolving gel (Fig. 2). After polymerisation the surface of the gel was rinsed with gel buffer, and the remaining acrylamide solution with 0.1% TEMED but omitting the neoglycoconjugate, was used to form the second layer of stacking gel. Sample wells in the second layer of stacking gel were formed using a teflon comb in the traditional manner.

Source Lectin (common abbreviation) Nominal carbohydrate specificity Molecular weight Isoelectric point Triticum vulgaris Wheat Germ Agglutinin (WGA) b GlcNAc 36-43 kD (18 kD subunit) pH 9.0 Canavalia ensiformis Concanavalin A or Jack bean lectin (ConA) a-Man, GlcNAc 106 kD (26 kD subunit) pH 5.0 Galanthus nivalis Snowdrop lectin (GNA a-Man 52 kD (13 kD subunits) pH 4.6 Glycine max Soybean lectin (SBA) GalNAc 120 kD (30 kD subunit) pH 6.0 Dolichos biflorus Horse gram lectin (DBA) a GalNAc 120 kD (30 kD subunit) pH 5.5 Arachis hypogea Peanut lectin (PNA) b Gal 110 kD (27 kD subunit) not known Ulex europaeus Gorse lectin (UEA-I) a Fuc 63 kD (31 kD subunit not known

Electrophoresis and Western Blotting
A panel of lectins (Oxford Glycosystems, UK) with different carbohydrate binding specificities (see Table) were used as test carbohydrate-binding proteins. Lectins (20 mg) were prepared for electrophoresis in a sample buffer comprising 0.025M Tris-HCl buffer (pH 6.8), 0.05mM CaCl₂, 20% glycerol and trace bromophenol blue. Some lectins are known to require metal ions (usually Mn²⁺ or Ca²⁺) for their interaction with sugars and it may be neccessary under certain conditions to include these in the sample and/or running buffers. Non-reducing native gel electrophoresis was performed at 4^oC at 70 Volts for 5-6 hours. After electrophoresis the gels were either prepared for Western blotting as described below, or the resolving and stacking gels were fixed and stained in coomassie blue. CA-gels and nitrocellulose membranes for Western blotting were pre-soaked in a Tris-glycine blotting buffer adjusted to pH 9.5, containing 0.1M of the competing free saccharide previously used in the CA-gel overlay. Blotting was performed overnight at 4°C at a fixed current of 0.3 amps. Transferred lectins were then visualised on nitrocellulose membranes using 0.1% solution of Amido Black in 45% methanol 10% acetic acid for 10 minutes. The nitrocellulose blots were then destained briefly in 10% acetic acid until the background was clear.

Figure 2. Preparation and orientation of the carbohydrate affinity gel layer in a slab gel electrophoresis format. Carbohydrate-binding proteins (red) are retained by the affinity layer, while non-binding proteins (black) are separated as normal in the resolving gel. Subsequent analyses may include an SDS-PAGE second dimension and/or Western blotting.

Results and Discussion
We have shown that specific allylamine-linked neoglycoconjugates can be used in discontinous polyacrylamide gel electrophoresis for the specific separation of carbohydrate-binding proteins. A major advantage of these polyacrylamide-based neoglycoconjugates is the possibility of creating molecules with predetermined properties, combined with low non-specific protein interaction. No free monomer remains in the gel matrix. These neoglycoconjugates, when co-polymerised with acrylamide, form a unique and stable polyacrylamide affinity matrix (Fig. 1). Moreover, when incorporated as a layer between the resolving and stacking gels (Fig. 2), this affinity gel matrix provides a unique separation phase in polyacrylamide gel electrophoresis. The properties of the polyacrylamide gel, such as the rate of polymerisation, stability and handling, are unaffected by concentrations of neoglycoconjugate sufficient to inhibit the migration of 10 mg of pure lectin (up to 250 mg, Fig. 3). Moreover, the Tris-glycine buffer system used in our approach, permits the separation of proteins within a wide range of isoelectric points (pI), while still allowing the separation and binding of carbohydrate-binding proteins, such as WGA (pI ~9.0).
In model experiments using a panel of lectins as carbohydrate-binding proteins (see Table), the electrophoretic mobility of selected proteins was inhibited by electrophoresis through an affinity gel layer containing immobilised neoglycoconjugates (Fig. 3 and 4). The migration of lectins with no binding specificity for the affinity gel, continued unhindered into the resolving gel. This feature was clearly seen in the electrophoresis of the b-N-acetylglucosamine (b GlcNAc) binding lectin WGA through an affinity gel layer containing the b GlcNAc neoglycoconjugate (Fig. 4A). The remaining lectins (Con A, SBA, PNA, DBA, GNA, and UEA) that are not specific for the b GlcNAc neoglycoconjugate, were unaffected by the affinity gel layer and thus migrate into the native resolving gel. Similarly, only the b-N-acetylgalactosamine (b GalNAc) binding lectin, SBA, was retarded by the b GalNAc neoglycoconjugate, while those lectins not specific for this carbohydrate continued to migrate through the resolving gel (Fig. 4B). Interestingly, this also shows the specificity of interactions detected by this approach, since the aGalNAc binding lectin, DBA, and the b-galactose specific lectin, PNA, were both unaffected by the b GalNAc neoglycoconjugate affinity gel layer.

Figure 3. Native CA-PAGE for SBA with different concentrations of the b GalNAc neoglycoconjugate. The affinity gel layer comprises 0-240 nmol immobilised GalNAc. Note the non-b GalNAc-binding contaminant in the SBA preparation (arrow).

Our approach, in principle similar to affinity chromatography, provides the additional benefit of simultaneous separation of proteins by native gel electrophoresis. Analyses may be performed on crude mixtures of proteins, wherein separation of the carbohydrate-binding protein does not affect the resolution or mobility of other non-binding proteins (Fig. 4C). Furthermore, depending on the concentration of neoglycoconjugate, the specific carbohydrate-binding protein can also be retained as a narrow band in a small volume of acrylamide (~4 mm³). Subsequent analysis of the isolated native proteins is also possible through the electro-transfer of proteins to a membrane support such as nitrocellulose (Western Blotting).
This simple transfer of carbohydrate-binding proteins from the affinity gel matrix permits further analyses such as immunostaining or sequencing.
The CA-PAGE technique, in a single slab gel separates proteins on the basis of functional characteristics in addition to the more traditional size and charge criteria of a native gel electrophoresis fractionation.

Figure 4. Native CA-PAGE of selected lectins with different affinity gels. Note the retarded migration of WGA in the affinity gel layer containing immobilized b GlcNAc (A), and SBA in the affinity gel layer containing immobilized b GalNAc (B). Control lanes are shown with normal stacking gels. Lectins not specific for these saccharides migrate through the affinity gel matrix. Native CA-PAGE of normal serum (S1) with and without SBA (C) shows that the migration of non-binding proteins is unaffected by SBA bound in the affinity matrix. Similarly the binding of the SBA appears to be unaffected in the presence of a complex mixture of non-binding proteins.

However, the technique can also be incorporated into a two dimension analysis (Fig. 2). The region of affinity gel containing the carbohydrate-binding protein(s) can be cut out (6 x 2 mm slice) and placed in a stacking well of a denaturing SDS polyacrylamide gel. The subsequent electrophoresis provides additional information regarding the subunit structure of the carbohydrate-binding protein. One significant advantage is the ability to compare multiple samples in the same second dimension gel.
There are many areas in both clinical and basic research where the CA-PAGE approach could be used. Prion proteins are thought to bind sulphated glycans in brain tissues [2] and these interactions could be facilitated using CA-PAGE. Lectins with identical binding properties, as recently documented in the Galectin family of b-galactoside-binding proteins [3], could be resolved using CA-PAGE as the first dimension. CA-PAGE also provides a rapid approach in screening serum samples, cells or tissues for new carbohydrate-binding proteins that may have prognostic or diagnostic value. The principles of CA-PAGE could also be applied in the study of carbohydrate-carbohydrate interactions.

References

1. Davis, B.J. Disc electrophoresis - II. Method and application to human serum proteins. Ann. NY Acad. Sci., 121:404-427 (1964)

2. Shyng, S.L., Lehman, S., Moulder, K.L., and Harris, D.A. Sulfated glycans stimulate endocytosis of the cellular isoform of the prion protein, PrPc, in cultured cells. J. Biol. Chem., 270: 30221-30229 (1995)

3. Lutomski, D., Caron, M., Cornillot, J-D., Bourin, P., Dupuy, C., Pontet, M., et al. Identification of different galectins by immunoblotting after two-dimensional polyacrylamide gel electrophoresis with immobilized pH gradients. Electrophoresis, 17:600-606 (1996)