Chapter 1 Introduction
1.2 Biochemical properties of β-LG
Due to the thermally unstable and molten-globule nature, β-LG has been studied extensively for its physical and biochemical properties (Qi et al, 1995;
Sawyer and Kontopidis, 2000; Kontopidis et al, 2004). Spectroscopic studies have demonstrated that irreversible modification of the β-LG structure occurs upon thermal treatment above 65-70 °C. Thermodynamic analysis of the calorimetric signal reveals that there are two domains unfolding independently while heating (Fessas et al, 2001). Heat treatment of β-LG at neutral pH causes the dimeric native protein to dissociate, partially unfold, denature, and aggregate; the rates and pathways are known to be dependent on the protein concentration, pH, temperature (Havea et al, 2004). Two major aggregating features are related to hydrophobic association and disulfide-bond interchange reactions (Havea et al, 2004). When a -LG is heated at neutral pH, the equilibrium between the native β-LG dimers and monomers is shifted toward the monomers. At temperatures above 60 °C, the β-LG monomer partially unfolds, with the loss in the helical structure allowing Cys121 to interact with the Cys106-Cys119 disulfide bond, to give a non-native
Cys160 is also likely to become involved in interprotein disulfide bonding in whey protein aggregates (Creamer et al, 2004). Thus, the resultant activated monomers lead to the formation of various intermediate aggregation products. Involvement of the free thiol group in aggregation has been shown in a number of studies, most notably by site-directed mutagenesis (Cho et al., 1994) where thermal stability was enhanced by incorporating the free thiol into a third disulphide bridge to the neighbouring helix. Further, the porcine protein has no free thiol and does not form gels under the same conditions as the cow (Gallagher et al., 1996). However, the exact regions involved in the thermal denaturation are still unclear. Whether the subtle unfolding changes can be detected by an immunochemical approach remains at question.
1.3 Physicochemical Properties of β-LG
A remarkable physicochemical property of the β-LG is its ability to bind in vitro small hydrophobic molecules such as retinol, fatty acids, vitamin D, and cholesterol (Narayan and Berliner, 1997; Qin et al, 1998; Wu et al, 1999;
Kontopidis et al, 2002). β-LG has been postulated to serve as a carrier for retinol in neonates (Sawyer et al, 1985), and it has been proposed that the binding of fatty
acids to β-LG at neutral pH is related to a biological function of this protein in bovine milk (Perez et al, 1992; Perez and Calvo, 1995). The binding of fatty acids to β-LG has been reported to increase the resistance of the protein to proteolytic degradation (Puyol et al, 1993), thermal degradation (Puyol et al, 1994), and unfolding in urea solution (Creamer, 1995).
Regardless of intensive research, biological function of this protein has not yet been satisfactorily resolved. Recently, we immunized the mice with commercially prepared dry milk and produced a panel of monoclonal antibodies (mAb). From 900 hybridomas screened a clone specific to dry milk, but not to raw milk, has been selected. Characterization of this dry milk specific mAb reveals that this antibody recognizes thermally denatured β-LG (Chen et al, 2004). It suggests that a new antigenic epitope in β-LG is being exposed by a heating process used in the preparation of dry milk. In the present study, we defined the immunoreactive site that was recognized by this specific mAb and attempted to relate it to the thermal denaturation properties of β-LG. The strategy for epitope mapping combined several approaches including tryptic and CNBr fragments, chemical modifications
peptides (with overlapped regions), and a synthetic peptide in solution for immunoassays. We demonstrate that the epitope was located exactly within the D strand of β-LG (residues 66-76). The immunoreactivity as recognized by this mAb was correlated to the thermal denaturation and conversion of β-sheet to a disordered structure of β-LG. Interestingly, the D strand is associated with the A-C strands forming one domain at the opening of the calyx (Fig. 1). For this reason, we also studied the effect of heating and pH on β-LG binding to retinol and palmitic acid. Further epitope mapping shows that conversion of Glu-74 into either Ala or negatively charged Asp totally abolished its immunoreactivity. A similar result was seen in Lys-69, but not the other Lys residues. Finally, we propose that strand D plays a provocative role in the molten-globule state of β-LG as probed by our mAb.
MATERIALS and METHODS
2.1 Materials
β-LG was purified from fresh raw milk using 30% saturated
ammonium-sulfate top fraction followed by a G-150 column chromatography as described previously (Chen et al, 2004; McCreath et al, 1997).
2.2 Preparation of Monoclonal Antibody Specific to Dry Milk
Monoclonal antibodies were produced according to the standard procedures
previously described by us (Mao et al, 1988; Mao et al, 1990), in which dry milk (Nestle Australia Ltd, Sidney, Australia) was used for immunization (Chen et al, 2004). In brief, the myeloma cell line (FO) was fused with spleen cells from immunized Balb/c mice at a ratio of 1:5. The culture medium (between days 14 and 21 after fusion) was assayed for the production of specific antibodies by a solid-phase ELISA using both raw and dry milk as a respective antigen. Each monoclonal was established by limiting dilutions at least 2 x (Mao et al, 1988; Mao et al, 1990).
2.3 Trypsin and CNBr Fragmentation
For trypsin treatment, 50 µg of β-LG in 100 µl phosphate buffered saline (PBS) containing 0.02 M phosphate and 0.12 M NaCl, pH 7.4, were preheated at 100 °C for 10 min. After which time, 1 µl trypsin (0.1 mg/ml) was added and incubated at room temperature for 4 h (Chen et al, 2004). Trypsinized LG was analyzed on a SDS-PAGE (18% polyacrylamide) followed by a Western blot. For CNBr fragmentation (Mao et al, 1975; Mao et al, 1977), 5 mg of β-LG were first dissolved in 70% (v/v) TFA with the addition of 10 mg CNBr in dark for 24 h at room temperature. After 3 x evaporation in Speed Vac (CVE 200D, ELELA, Japan) with the addition of 5 x volume of de-ionized water, the dry material was dissolved in the 10 mM phosphate buffer (PB), pH 7.0. The immunoreactivity of CNBr fragments were then analyzed on an 18% SDS-PAGE, followed by a Western blot.
2.4 Acetylation and Carboxymethylation of β-LG
Chemical modification of β-LG by acetylation was conducted by a modification of the procedure previously described by us (Mao et al, 1980). To 5 mg of β-LG in 2 ml 50 mM sodium bicarbonate (pH 8.0) containing 6 M urea, 5 µl of acetic anhydride were slowly added into the reaction mixture step by step, while maintaining the pH at 8.0 using 0.1 M NaOH. After 3 h incubation at room temperature, the acetylated protein was desalted on Bio-gel P-2 column eluted by 0.05 M ammonium bicarbonate and lyophilized. For carboxymethylation (Mao et al, 1980; Tseng et al, 2004), 5 mg of β-LG were first dissolved in 5 ml of 0.1 M Tris-HCl buffer (pH 8.6) containing 6 M ultra pure urea and 0.02 M dithiothreitol.
Following flushing with nitrogen, 20 mg of iodoacetic acid were added into the reaction mixture, while maintaining the pH at 8.6 by the addition of 0.1 M NaOH and incubation for another 3 h. Finally, carboxymethylated (CM) β-LG was desalted on a Bio-Gel P2 column eluted by 0.05 M ammonium bicarbonate and lyophilized. By amino acid analysis, the CM-β-LG contained 4.98 residues of CM-cysteine per mole of β-LG.
2.5 CD Spectrum
The secondary structure of native, heated or chemically modified β-LG was determined using a computerized Jasco J-715 circular dichroic (CD) spectropolarimeter. Each protein sample was dissolved in 10 mM phosphate buffer at pH 7 with a final concentration of 0.2 mg/ml. About 300 µl of the protein solution were used for analyzation within a cuvette of 1-mm path length.
The obtained spectra were accumulated for 25 times at a scanning rate of 50 nm/min. All the data were shown as the mean residue molar ellipticity [θ]MRW
(Tseng et al, 2004; Chen et al, 1994).
2.6 Peptide Array
Twelve synthetic peptides in one nitrocellulose-array, each containing 15
amino-acid residues, were designed corresponding to the residues 25-107 of β-LG or to the residues 67-75 within strand D (Fig. 1). The synthetic peptides were prepared under a contract with a local biotechnology company (Genesis Biotech Inc., Taipei, ROC). Briefly, the peptides were directly synthesized in situ on a nitrocellulose (NC) paper according to the method described (Frank, 1992). The NC membrane in 0.01 M Tris buffered saline containing 0.05% (v/v) Tween-20
(TBST), was blocked with 5% (w/v) gelatin in TBST for 2 h at room temperature followed by 3 x washes. After incubation with mAb for 2 h and 3 x washes, goat anti-mouse IgG conjugated with horseradish peroxidase (HRP) in 5% gelatin/TBST was added and incubated. Finally, following the washes, chemiluminescent substrate (ECL™ Western Blotting System, Amersham) was added, washed, and immediately developed by exposing onto a film.
2.7 Competitive ELISA
In brief, heated β-LG (1 µg in 50 µl of PBS) was first immobilized onto
microtiter wells followed by 3 x washes to remove unbound β-LG (Chen et al, 2004;
Mao et al, 1989). The wells were then blocked by 3% gelatin in PBS. After 3 x washes, 50 µl of the competitive protein (β-LG, heated β-LG, acetylated or carboxymethylated β-LG, or synthetic peptide residues 67-76) in PBS containing 3% gelatin were mixed with 50 µl of mAb and incubated at room temperature for 1 h. Following washes and secondary antibody (goat anti-mouse IgG conjugated with HRP) incubation, the microtiter plate was developed with 2, 2-Azino-bis (3-ethylbenz-thiazoline-6-sulfonic acid) and read at 415 nm.
2.8 Effect of pH on β-LG Binding to mAb
Heated β-LG (1 µg/well) was immobilized onto a microtiter plate followed by
blocking and washing at neutral pH. The immobilized β-LG was then incubated with mAb at various pH for 2 h at room temperature. After removing the unbound mAb, the plate was developed according to the standard ELISA procedures at neutral pH (Chen et al, 2004; Mao et al, 1989). Since pH itself can affect the antigen-antibody binding, a control experiment using mouse IgG as immobilized antigen was also conducted at various pH for a parallel comparison.
2.9 Retinol and Palmitic Acid Binding to β-LG
β-LG was reported to be a 1-to-1 binding ratio with retinol or palmitic acid as measured by fluorescence emission techniques (Yang et al, 2002). In general, binding of retinol to β-LG was measured by extrinsic fluorescence emission of retinol molecule at 470 nm using excitation at 287nm. Whereas, binding of palmitic acid to β-LG was measured by the fluorescence enhancement of Trp residues of β-LG at 332 nm using excitation at 287 nm. For the effect of the pH experiment, 5 uM or 20 uM of native β-LG was instantly incubated with 5 uM of retinol or 20 uM of palmitic acid, respectively, at various pH at 24 °C. For the
effect of the heat experiment, β-LG was preheated at 80 °C or 100 °C for 5 min and then incubated with retinol or palmitic acid at pH 8. Fluorescence spectra were recorded at 24 °C with a fluorescence Spectrophotometer F-4500 (Hitachi High-Tech.Cor., Tokyo, Japan).
2.10 Three Dimensional Analysis of β-LG Structure
3D structure of β-LG used in this context was provided by Protein Data Bank
(PDB, http://www.rcsb.org/pdb/), code 1CJ5 (Kuwata et al, 1999), with the diagram created by PyMOL (DeLano, 2002).
2.11 Three Dimensional Analysis of β-LG and Palmitic Acid Complex Structure
3D structure of Bovine β-LG complexed with palmitic acid used in this
context was provided by Protein Data Bank (PDB, http://www.rcsb.org/pdb/), code 1GXA (Kontopidis et al, 2002), with the diagram created by Rasmol (Sayle et al, 1995).
RESULTS
3.1 Characterization of The Monoclonal Antibody Specific to Dry Milk
Previous studies show that the monoclonal antibody (mAb) used in this report is specific to processed dry milk, but not to raw milk (Chen et al, 2004). It only recognizes β-LG, one of the major milk proteins. The mAb is apparently able to discriminate the denatured β-LG from a given milk product (Chen et al, 2004).
Since heating procedures are used to process the dry milk, the finding indicates that β-LG undergoes a conformational rearrangement, which facilitates the binding of this mAb. In the present study, we show a dramatic and sharp increase in β-LG immunoreactivity when raw milk was heated between 70 °C and 80 °C over time (Fig. 3). It was of interest that the increase in immunoreactivity was concomitant with the reported transition temperature for converting native to denatured β-LG (de Wit et al, 1980). The finding suggests that the immunoreactive site recognized by this mAb lied in the thermal denatured region of β-LG.
3.2 Mapping of Antigenic Determinant of Denatured β-LG Utilizing Tryptic Digestion and Acetyl-modification
To initially map out the specific immunoreactive region, heated β-LG was limitedly digested by trypsin. On Western blot, we demonstrated that the immunoreactivity was totally abolished after the trypsin treatment (Chen et al, 2004) suggesting that Lys, Arg, or both residue(s) were probably involved in maintaining the antigenic structure for β-LG (Fig 4). Chemical modification using acetylation (pH 8.0) on positively charged residues (mostly Lys) attenuated the immunoreactivity of β-LG on a competitive ELISA (Fig. 5). These two experiments support the notion that positively charged amino acids of β-LG attributed for the mAb recognition.
3.3 Immunoreactivity of CNBr Fragments of β-LG
To further delineate the major antigenic domain, CNBr cleavage on β-LG was conducted. Western blot analysis shows that there was a major immunoreactive fragment corresponding to a molecular weight about 9 kDa (Fig. 6). As estimated from its Met cleavage site, this fragment was presumed a peptide containing
was determined. The first six amino-acid residues (AASDIS) confirmed that the immunoreactive site was located between the residues 25-107 of β-LG (Fig. 1).
3.4 Final Antigenic Mapping Using a Solid-phase Peptide Array
As described above, Lys enriched-areas were assumed to participate in maintaining the antigenic structure. Pro residues are also considered to be involved as they are located at or near the antigenic determinant by forming a loop at the surface of a given protein. Using an EMBOSS program for searching a possible antigenic determinant within residues 25-107 of β-LG (Fig. 7), we predicted that two domains, namely residues 42-56 and 67-81, were most likely to be immunoreactive. Accordingly, a solid-phase peptide array containing the above predicted regions and ten other overlapped synthetic peptides (each with 15 residues) was prepared. These peptides were directly synthesized on a nitrocellulose membrane (Fig. 7). After binding of mAb and HRP-conjugated secondary antibody, the array was developed by a chemiluminescent agent. We show that only peptides 4 (residues 70-84), 8 (residues 61-75), and 12 (residue 67-81) were immunoreactive (Fig. 7). Among which peptide 4 gave a partial immunoreactivity suggesting that residues 67-70 were essentially involved in the
reactive site (Fig. 7). Since the size of an epitope is relatively small usually containing 6-9 amino-acid residues (Mao et al, 1990; Mao et al, 1989; Bhatnagar et al, 1983; Davies and Cohen, 1996; Atassi, 1984), it was possible to narrow down the reactive site from the immunoreactivity in overlapped peptides. We proposed that the reactive site was closely associated with AQKKIIAEK (or nine residues 67-75) (Fig. 7). Notably, this region is highly positive in charges. Observing from the high resolution crystal structure of β-LG (Kuwata et al, 1999), it is fascinating that this proposed region is exactly located within the D strand of surfaced β sheet (residues 66-76) (Fig. 1 ). We therefore defined this immunoreactive site as an epitope. Another noticed point is that there is a disulfide linkage between strand D (Cys-66) and carboxyl terminus (Cys-160).
This disulfide linkage plays an important role to stabilize the β-structure by forming antiparallel sheets of β-LG. The proposed epitope (67-75) in its native state is rather ordered with a β-sheet span about 28 Å in length. As such, the orientation in native β-LG may prohibit the binding of our specific mAb. We hypothesized when the D strain underwent disordered structure, it would then allow the “denatured” mAb binding.
3.5 Carboxymethylated β-LG and CD Spectrum
To test the above hypothesis by which the conformational change of the D strand would enhance the binding of our mAb, we chemically modified all the Cys residues to irreversibly block the disulfide linkages within the native β-LG (Fig. 8).
Using a competitive ELISA with heated β-LG as a positive control, it revealed that carboxymethylation on β-LG resulted in a striking increase in its immunoreactivity.
The increase was about 10 x greater than that of heated β-LG (Fig. 9). Meanwhile, analysis of CD spectra on carboxymethylated β-LG further confirmed a significant conformational change by converting β sheet (typically at 215 nm) to a more disordered structure than that of heated β-LG (Fig. 10).
3.6 Immunoreactivity of a Soluble Synthetic Peptide
Finally, a soluble peptide corresponding to the linear sequence of strand D (residues 67-76 or AQKKIIAEKT) was synthesized. Figure 11 shows that this linear sequence was able to completely inhibit mAb binding to heated β-LG on a competitive ELISA. Furthermore, this synthetic peptide exhibited a typical disordered structure rather than a β configuration (Fig. 10).
3.7 Role of Charged Residues in Epitope Specificity
To determine which Lys(s) were responsible for the mAb recognition, mutation on each Lys (Lys-69, Lys-70, and Lys-75) with Ala was conducted. As shown in Fig. 8, only Lys-69 was extremely specific for the mAb binding.
Replacement with positively charged Arg did not salvage the immunoreactivity.
Glu-74 played a similar role, replacement with Ala or negatively charged Asp failed to show any immunoreactivity. Ile-71 and Ile-72 also played an essential hydrophobic role, although the exact residue has not been identified. Meanwhile, negative control peptides (peptides 11-12, Fig. 12) retaining all the Lys residues did not show any binding.
3.8 Effect of pH on β-LG Binding to mAb
Because the structural stability of β-LG is pH dependent (Casal et al, 1988), we tested whether changes of pH could also induce an increase in β-LG immunoreactivity. Fig. 13A shows that the CD structure of β-LG was stable at pH 2 with some changes between 3 and 7, while a transition to disorder was seen from 8 to 10. However, such disordered structure did not facilitate the mAb binding
this residue started to become neutralized under pH 8-10, the immunoreactivity was decreased (Fig. 13B). A control experiment showing a typical pH dependent antigen-antibody reaction was performed (Fig. 13C), and there was a slight decrease in immunoreactivity at pH 9-10.
3.9 Effect of pH and Heating on β-LG Binding to Retinol and Palmitic Acid
To explore the correlation between the structural change of β-LG (at various pH) and its retinol binding, we monitored the extrinsic fluorescent change of retinol upon the binding to β-LG. The optimal binding for retinol appeared to be at pH between 8 and 10 (Fig. 14A). Heating β-LG at temperatures greater than 80 °C almost completely abolished its binding for retinol (Fig. 14B). The data support the notion that the striking increase in immunoreactivity of the D strand at this temperature (Fig. 3) was negatively correlated to the retinol binding, which requires the integrity of a β-sheet structure of β-LG.Interestingly, the binding to palmitic acid was decreased in some extent at pH 9-10 (Fig. 14C) correlated with the binding to mAb (Fig. 13B). Since Lys-69 played an extremely essential role in the antigenic site (Fig. 12), such correlation
suggests that the protonated state of this residue might be involved in stabilizing both the mAb binding and β-LG-palmate complex formation (please see more detail in Discussion). Similarly, heating on β-LG substantially reduced the binding for palmitic acid.
DISCUSSION
Molten globules are thought to be general intermediates in protein folding and unfolding (Ptitsyn et al, 1990; Chang and Li, 2001). β-LG, a major moiety of bovine whey proteins, is one of the most investigated models for understanding the mechanism involved in protein stability upon heating. Although the 3D crystal structure of β-LG has been elucidated, the area involved in thermal denaturation remains unclear. On the other hand the region responsible for Tanford transition (Qin et al, 1998), occurring at pH from 6.5 to 8.0, is known to be within the residues 85-90 (E-F loop). This region opens or blocks the entrance of the calyx (Ragona et al, 2003).
The present study demonstrates that denatured strand D of β-LG was responsible for the binding of our thermally sensitive mAb (Chen et al, 2004).
Several unique features of the binding are identified. First, heating on native β-LG resulted in a loss of β-sheet to more disordered structure (Fig. 10) in which the immunoreactivity was concomitantly increased (Fig. 9). Second, blocking the disulfide linkage between the D strand (Cys-66) and C-terminus (Cys-160) of β-LG by carboxymethylation not only produced a disordered structure (Fig. 10) but also
markedly enhanced the mAb binding. Further heating on carboxymethylated β-LG did not give more binding. Such enhancement was even greater than heated β-LG (10 x) on competitive ELISA (Fig. 9). Presumably, this was due to the augmented degree of freedom of the D strand without the disulfide linkage rendering more antibody binding. It is of interest to point out that the secondary structure of strand D alone, without including strand C, is predicted as 50% random coiled (residues 66-70) and 50% helical (residues 71-76) using the parameters from 3D-PSSM: the folding recognition server at the ICRF (www.sbg.bio.ic.ac.uk/~3dpssm/). However, in the presence of strand C as an anti-parallel orientation, the predicted structure of strand D becomes β-configuration. Obviously, the formation of an anti-parallel β-structure in native β-LG molecule is stabilized through the help of a disulfide-linkage (Cys-66 and Cys-160) between strand D and the helical domain at COOH-terminus (Fig. 2).
Thus, conformational change on strand D played a vital role for the mAb recognition. It is worth mentioning that although severe heating might break the disulfide linkage at Cys-66, it would be immediately “stabilized” via re-oxidation by forming high molecular or self-associated polymers as shown in our previous
than that of carboxymethylated β-LG. Third, the soluble synthetic-peptide (residues 67-76) corresponding to strand D (without Cys-66) was able to completely inhibit the binding of mAb to β-LG. Fourth, the D strand is topographically located at the surface of β-LG (Fig. 2), which is agreeable with the general concept of a given antigenic epitope (Atassi, 1980). Fifth, the buried side chain of Lys-69 was exposed upon the heating and then recognized by the mAb (described below).
With respect to the exact size of the epitope that was recognized by our mAb, we excluded the possibility of Cys-66 as part of the epitope from the D strand.
First, peptide 67-81 without Cys-66 gave an almost equal immunoreactivity to that of peptide 61-75 with Cys-66 in a peptide array assay (Fig. 7) suggesting that Cys-66 might not be located in antigenic determinant. Second, carboxymethylation on whole β-LG molecules with Cys-66 included in the modification markedly increased its immunoreactivity. If Cys-66 was involved in the antigenic site, introduction of such a bulky group (carboxymethyl) on this residue would have resulted in a significant loss of immunoreactivity (Atassi, 1984).
It should be noted here that Cys-66 was only responsible for the conformational