Tumor marker-responsive behavior of gels prepared by biomolecular imprinting

Abstract

We report dynamic glycoprotein recognition of gels prepared by biomolecular imprinting using lectin and antibody molecules as ligands for tumor-specific marker glycoproteins. The glycoprotein-imprinted gels prepared with minute amounts of cross-linkers could dynamically recognize tumor-specific marker glycoproteins by lectin and antibody ligands and induce volume changes according to the glycoprotein concentration. The glycoprotein-imprinted gel shrank in response to a target glycoprotein but nonimprinted gel swelled a little. The glycoprotein-responsive shrinking of the imprinted gel was caused by formation of lectin–glycoprotein–antibody complexes that acted as reversible cross-linking points. Glycoprotein-imprinted gels only shrank when both lectin and antibody in the gels simultaneously recognized the saccharide and peptide chains of the target glycoprotein. As shrinking behavior of biomolecularly imprinted gels in response to glycoproteins enables the accurate detection and recognition of tumor-specific marker glycoproteins, they have many potential applications as smart devices in sensing systems and for molecular diagnostics.

Stimuli-responsive gels that undergo volume changes in response to environmental changes such as pH and temperature are promising smart material candidates for drug delivery systems and sensors in biochemical and biomedical fields (1–9). Recently, several researchers have focused on biologically stimuli-responsive gels that can sense specific signal biomolecules, such as glucose (10–18) and protein (19, 20). Signal biomolecule-responsive gels have become increasingly important as smart devices for drug delivery systems and molecular diagnostics. Biomolecule-responsive gels that exhibit swelling/shrinking in response to signal biomolecules like tumor-specific markers have many potential applications as smart devices in sensing systems and molecular diagnostics.

Molecular imprinting is a promising method for creating synthetic hosts (often called “synthetic antibodies”) having molecular recognition sites (21–25). In molecular imprinting, monomers having functional ligand groups for target (template) molecules are copolymerized with a large amount of cross-linkers whereas functional groups interact with template molecules to arrange ligands at suitable positions for recognizing target molecules. In previously reported molecular imprinting studies, monomers of low molecular weight were used as ligands for target molecules, but were not very effective as suitable ligands for macromolecules such as proteins. In addition, most researchers believe that the mobility of polymer chains in molecularly imprinted materials must be depressed by a large amount of cross-linker to create molecular recognition sites. In this article, however, we reveal that proteins can be used as ligands in molecular imprinting and that using minute amounts of cross-linker enables resultant gels to undergo volume changes in response to target biomolecules.

In this study, tumor marker-responsive gels that exhibited volume changes in response to a tumor-specific marker glycoprotein (α-fetoprotein, AFP) were prepared by biomolecular imprinting using protein ligands and minute amounts of cross-linker. We report dynamic glycoprotein recognition of glycoprotein-imprinted gels having lectin and antibody molecules as ligands for tumor-specific marker glycoproteins. This article focuses on tumor marker-responsive behavior of the glycoprotein-imprinted gels prepared with minute amounts of cross-linker.

Results and Discussion

Strategy to Prepare Tumor Marker-Responsive Gels. AFP is a glycoprotein widely used for the serum diagnosis of primary hepatoma (26–28). Serum AFP also exhibits abnormal glycosylation patterns in various diseases states, including primary hepatoma and cirrhosis of the liver, and serves as a diagnostic tumor-specific marker. Our purpose in this study was to prepare AFP-responsive gels that exhibit volume changes in response to a tumor-specific marker glycoprotein, AFP, by biomolecular imprinting using protein ligands and minute amounts of cross-linker. Tumor marker-responsive gels can be strategically prepared by the biomolecular imprinting of lectins and antibodies as ligands for saccharide and peptide chains in target glycoproteins. Association and dissociation behaviors of lectin–AFP–antibody complexes in reversible cross-linking reactions cause AFP-responsive volume changes (Fig. 1). An important point to remember when preparing tumor marker-responsive gels is to use minute amounts of cross-linkers, which enable resultant networks to change from expanded to collapsed states based on reversible changes of cross-linking structures. Our strategy for AFP-responsive behavior of biomolecularly imprinted gels is as follows. As shown in Fig. 1b, AFP-imprinted gels shrink in the presence of free AFP because lectins and antibodies linked within gel networks form lectin–AFP–antibody complexes that play important roles as cross-linking points. Thus, AFP-imprinted gels prepared by biomolecular imprinting can dynamically recognize target glycoproteins by lectin and antibody ligands and induce volume changes according to the glycoprotein concentration.

Fig. 1.

The strategy for the preparation of tumor marker-responsive gels by biomolecular imprinting. (a) Synthesis of the tumor marker-imprinted gels using lectins and antibodies as ligands for template glycoprotein molecules (tumor-specific marker AFP). (b) A schematic representation of glycoprotein-responsive behavior of tumor marker-imprinted and nonimprinted gels.

Bioconjugated gels containing lectins and antibodies to recognize saccharide and peptide chains in AFP were prepared by biomolecular imprinting as follows (Fig. 1a). AFP molecules were used as template glycoproteins to prepare imprinted gels. Lectin (Con A) and polyclonal anti-AFP antibody (anti-AFP) were conjugated with N-succinimidylacrylate (NSA) to introduce polymerizable vinyl groups. After poly(acr ylamide)(PAAm)-grafted lectins were synthesized by the copolymerization of vinyl–lectin with acrylamide (AAm), vinyl–antibodies were copolymerized with AAm and N,N′-methylenebisacrylamide in the presence of template AFP and PAAm-grafted lectins to form lectin–AFP–antibody complexes. AFP-imprinted gels were prepared by removing AFP from resultant networks having lectin–AFP–antibody complexes. To reveal the effect of biomolecular imprinting on the glycoprotein-responsive behavior of resultant gels, we also prepared nonimprinted gels by copolymerizing vinyl–antibodies with AAm and N,N′-methylenebisacrylamide in the presence of PAAm-grafted lectin without AFP.

Tumor Marker-Responsive Behavior. AFP-responsive behaviors of AFP-imprinted, nonimprinted, and PAAm gels were examined after swelling equilibria were attained in 0.2 M phosphate buffer solution (pH 7.4). AFP-imprinted gels began to shrink as soon as they were immersed in phosphate buffer solutions containing AFP. Equilibrium swelling ratios of AFP-imprinted gels gradually decreased with increases in the AFP concentration of buffer solutions (Fig. 2). Swelling behaviors indicated that the AFP-imprinted gels were tumor marker responsive, with glycoprotein-sensing functions. In aqueous AFP solutions, however, PAAm gels did not exhibit volume changes, and the nonimprinted gels experienced slight swelling. It is noteworthy that AFP-imprinted gels shrank and the nonimprinted gels swelled in the presence of AFP. To reveal the mechanism responsible for the AFP-responsive behavior of the gels, cross-linking densities were determined by measuring compressive modulus. Cross-linking densities of the AFP-imprinted gel gradually increased with increasing AFP concentrations in phosphate buffer solutions (Fig. 3). However, cross-linking densities of the nonimprinted and PAAm gels did not change with increasing AFP concentrations. The density differences suggest that the AFP-responsive shrinking of AFP-imprinted gels was caused by increased cross-linking densities, consistent with the mechanism depicted in Fig. 1b. In addition, we investigated adsorption of AFP into AFP-imprinted, nonimprinted, and PAAm gels to reveal the effect of biomolecular imprinting on glycoprotein recognition in gels. The amount of AFP adsorbed into AFP-imprinted gels was a little larger than the amount incorporated into nonimprinted gels (Fig. 5, which is published as supporting information on the PNAS web site). As AFP-imprinted and nonimprinted gels contain both lectins and antibodies with strong affinities for AFP saccharide and peptide chains, AFP was adsorbed into gels. However, a larger amount of AFP adsorbed into AFP-imprinted gels than into nonimprinted gels demonstrates that biomolecular imprinting was effective in constructing glycoprotein recognition sites despite swollen gels having been prepared with minute amounts of cross-linker. Lectins and antibodies in AFP-imprinted gels were distributed at optimal positions for the simultaneous recognition of AFP saccharide and peptide chains because gel networks were formed by biomolecular imprinting using AFP molecules as templates. Therefore, the presence of AFP induced the formation of lectin–AFP–antibody complexes that could play a role at cross-linking points. However, as lectins and antibodies in nonimprinted gels were not distributed at optimal positions for the simultaneous recognition of AFP saccharide and peptide chains, the cross-linking densities of nonimprinted gels did not change because of the absence of lectin–AFP–antibody complexes. Therefore, a slight swelling of nonimprinted gels in the presence of AFP was attributed to the changing osmotic pressure associated with the binding of AFP to lectin or anti-AFP. Consequently, biomolecular imprinting enables gel networks to memorize template biomolecules.

Fig. 2.

AFP concentration effects in phosphate buffer solutions on equilibrium swelling ratios of AFP-imprinted gels (○), nonimprinted gels (•), and PAAm gels (□) in phosphate buffer solutions containing AFP at 25°C. Swelling ratios of the gels were determined by ratios of diameter changes, measured in phosphate buffer solutions with and without free AFP, using an optical microscope.

Fig. 3.

Effect of AFP concentration in phosphate buffer solutions on cross-linking densities of AFP-imprinted gels (○), nonimprinted gels (•), and PAAm gels (□). Cross-linking densities in gels were determined by compressive modulus as follows. Compression modulus tests were performed on gels with a tensile tester. From the compressive modulus of swollen gels, the cross-linking density of the gel can be calculated with Eq. 3.

Glycoprotein Recognition. Because AFP-imprinted gels prepared in this study contained two kinds of ligands (lectins and antibodies) for a glycoprotein AFP, we investigated glycoprotein recognition behaviors by examining swelling ratios of AFP-imprinted and nonimprinted gels in the presence of AFP or ovalbumin. Ovalbumin has a saccharide chain similar to AFP, but has a peptide chain different from AFP. Therefore, ovalbumin is recognized by the lectin ligand (Con A), but not by the antibody ligand (anti-AFP) in AFP-imprinted and nonimprinted gels. Fig. 4 shows swelling ratio changes of AFP-imprinted and nonimprinted gels in phosphate buffer solutions containing AFP and ovalbumin. The presence of AFP and ovalbumin resulted in the slight swelling of nonimprinted gels. On the other hand, AFP-imprinted gels also swelled slightly in the presence of ovalbumin, but immediately shrank in the presence of AFP. Namely, only AFP-imprinted gels reversibly shrank in response to AFP (Fig. 6, which is published as supporting information on the PNAS web site). The aforementioned data demonstrate that AFP-imprinted gels only shrank when both lectins and antibodies in the gels simultaneously recognized the saccharide and peptide chains of the target glycoprotein (Fig. 1b). When lectin–glycoprotein complexes and/or antibody–glycoprotein complexes were formed but the lectin–glycoprotein–antibody complex was not formed, both AFP-imprinted and nonimprinted gels slightly swelled. Therefore, we can conclude that the AFP-imprinted gels showed exact glycoprotein recognition with a double-lock function. In addition, gels swelled when either lectins or antibodies formed complexes with glycoprotein saccharide or peptide chains. Therefore, swelling or shrinking behaviors of AFP-imprinted gels in the presence of glycoproteins enable the accurate detection and recognition of glycoproteins. In molecular diagnostic methods, mutations in AFP saccharide chains are important signals for discriminating hepatic malignancies. An array of AFP-imprinted gels prepared with various lectins and antibodies may permit conventional diagnoses. In addition, our results suggest that gel networks can visualize biomolecular nano-phenomena such as antigen–antibody binding and lectin–saccharide binding. The dynamic recognition of gel networks prepared by biomolecular imprinting enables the precise recognition of target biomolecules and converts biomolecular structural information to gel volumes at a macro level. Therefore, biomolecule-imprinted gels have a high potential in biomedical applications and may be used as sensor devices, because they can recognize tumor-specific markers and induce structural changes; biomolecule-imprinted gels have sensing, processing, and effecting functions for signal biomolecules.

Fig. 4.

Swelling ratio changes of AFP-imprinted gels (a) and nonimprinted gels (b) after the addition of AFP (○) and ovalbumin (•) after swelling had attained equilibrium in phosphate buffer solution at 25°C. The glycoprotein concentration in the phosphate buffer solution was 40 μg/ml. As ovalbumin has a saccharide chain similar to AFP and a peptide chain different from AFP, ovalbumin can be recognized by the lectin (Con A) but cannot be recognized by the antibody (anti-AFP) in the gels.

Materials and Methods

Synthesis of the Tumor Marker-Imprinted Gel. The AFP-imprinted gels were prepared as follows by biomolecular imprinting using lectins and antibodies as ligands (Fig. 1a). First, a type of lectin, Con A, was chemically modified by coupling with NSA in PBS, using a method reported by Shoemaker et al. (29). NSA (0.068 mg) was added to PBS (0.02 M, pH 7.4) containing Con A (20.8 mg) (NSA/Con A molar ratio: 2:1), and the reactions were incubated at 36°C for 1 h to introduce vinyl groups into Con A. The resultant vinyl–Con A products were purified by gel filtration, and PBS containing vinyl–Con A molecules was obtained at a concentration of 6.51 mg/ml. After AAm (20 mg) was added to 15.0 mg of the aqueous vinyl–Con A solution, together with 0.02 ml of 0.1 M aqueous ammonium persulphate and 0.02 ml of 0.8 M aqueous N,N,N′,N′-tetramethylethylenediamine (TEMED) as redox initiators, copolymerizations were performed at 25°C for 6 h to synthesize PAAm-grafted Con A. Furthermore, vinyl–anti-AFP having polymerizable groups was also synthesized by modifying anti-AFP with NSA in the same manner as vinyl–Con A. In biomolecular imprinting, the resultant vinyl–anti-AFP (0.058 mg), template AFP (0.063 mg), AAm (132 mg), and N,N′-methylenebisacrylamide (0.05 wt% relative to AAm) as a cross-linker were dissolved in 880 ml of PBS containing the PAAm-grafted Con A to form lectin–AFP–antibody complexes. As soon as 0.02 ml of 0.1 M aqueous ammonium persulphate and 0.02 ml of 0.8 M aqueous TEMED as redox initiators were added into the mixture and injected into glass capillary tubes with inner diameters of 3 mm, polymerization was performed at 25°C for 6 h. To remove template AFP, the gels obtained by the polymerization were successively immersed in PBS (0.02 M, pH 7.4), phosphoric acid–citric acid buffer solution (0.02 M, pH 4.0), and PBS (0.02 M, pH 7.4) containing 1 mM MnCl₂ and 1 mM CaCl₂. UV spectroscopic analysis revealed that the amount of the glycoprotein removed from the gels was almost the same as that of a template glycoprotein used for their preparation and that no template glycoprotein remained in the AFP-imprinted gels. Furthermore, nonimprinted gels were also prepared as reference materials by the copolymerization of vinyl–anti-AFP, AAm, and N,N′-methylenebisacrylamide in the presence of the PAAm-grafted Con A without template AFP in a similar manner. The bicinchoninic acid protein assay suggested that the lectin and antibody were not removed from the AFP-imprinted and nonimprinted gels during the extraction procedure. As a result, the amount of lectin and antibody immobilized in the AFP-imprinted gels was almost the same as that in the nonimprinted gels. The resulting AFP-imprinted and nonimprinted gels were used in experiments after swelling had attained equilibrium in PBS.

Swelling Measurements. The AFP-imprinted, nonimprinted, and PAAm gels were kept immersed in 0.1 M phosphate buffer solution until equilibrium was reached at 25°C. After that, the gels were transferred and kept immersed in a phosphate buffer solution containing a desired amount of AFP at 25°C. The swelling ratio of the gels was determined from the ratio of their diameters by Eq. 1:

[1]

The diameters of the gels swollen in a phosphate buffer solution (d₀) and a buffer solution containing a free AFP (d) were measured with an optical microscope. To minimize experimental error, the diameters of the gels were measured five times for each sample and then averaged. Furthermore, every measurement was repeated at least three times to confirm the reproducibility.

Measurement of Compressive Modulus. Compression modulus tests were performed on gels with a tensile tester (EZ Test/CE, Shimadzu). The swollen gel was compressed by the crosshead of the apparatus, and then the relationship between the compressive stress and strain of the gel was recorded. The compressive modulus can be obtained by Eq. 2 from the compressive stress and strain of the gels. Moreover, the cross-linking density of the gel can be calculated by Eq. 3:

[2]

[3]

where τ is the compressive stress (Pa), G is the compressive modulus (Pa), R is the gas constant, T is the absolute temperature (K), α is the ratio of the thickness of the gel before and after the compression, ν_e is the effective cross-linking density (mol/liter), V₀ is the volume fraction of the polymer during network formation, and V₂ is the volume fraction of polymer in the gel, which is obtained by the swelling measurements by using Eq. 4:

[4]

where ρ_g and ρ_s are the density of the dried gel and the solvent, respectively.