β-lactoglobulin (LG) is the major protein component of mammalian milk. Among the LGs, bovine LG (BLG) has been investigated most extensively.1 BLG contains 162 amino acid residues, including five cysteines (Cys66, Cys106, Cys119, Cys121, and Cys160), four of which form two disulfide bonds (Cys66-Cys160 and Cys106-Cys119), and Cys121 retains a free thiol group. Although BLG is dimeric at neutral pH, it dissociates into monomers at low pH and very low ionic strength condition because of intermolecular electrostatic repulsions.2, 3 However, the monomers retain their tertiary structures at acidic pH. Both crystal4–9 and solution10, 11 structures of BLG show that the monomer contains nine β-strands (strands A–I), an α-helix, and three short 310 helices. Eight of the β-strands (A–H) form an up-and-down β-barrel, and the two I-strands form an intermolecular β-sheet containing four intermolecular hydrogen bonds. A mutational study has shown that salt bridges between Asp33 and Arg40 of the opposing AB-loops that are important to the stability of the dimer.3
BLG binds a variety of hydrophobic compounds including retinol, fatty acids, and vitamin D3 at its hydrophobic cavity inside the barrel.12–14 The loop between the E- and F-strands (EF-loop) obstructs the entrance to the β-barrel in the crystal structure at pH = 6.2, whereas the entrance is open at pH = 7.1.5 This conformational transition is called the Tanford transition, and NMR studies15–17 have found that the deprotonation of Glu89 (pKa = 6.9) occurs first and is followed by the unfolding of the D-strand, the EF-loop, and the GH-loop.
Equine LG (ELG) contains the same number of amino acids and the two disulfides at the same positions as BLG. The sequence identity between ELG and BLG is 57%. Although a three-dimensional structure of ELG is not available, our previous NMR study showed that the secondary and tertiary structures of ELG are very similar to those of BLG.18 However, there are some structural differences between BLG and ELG, including their propensity to dimerize. At neutral pH, BLG is dimeric, whereas ELG is monomeric.19 Another difference involves their structural behavior in an acidic environment. BLG maintains its tertiary structure, whereas ELG loses its tertiary structure and assumes a molten globule state with non-native α-helices, which mimics the early kinetic folding intermediate.19, 20
Previously, Kobayashi et al.21 constructed a chimeric LG (LPI) in which the residues that form the BLG dimer interface replaced the corresponding residues in ELG; that is, Ser34 and Glu35 in the AB-loop of ELG were replaced with Ala and Gln from BLG, respectively, and the sequence flanking (and including) the I-strand (Gly145 to Leu153) was replaced with the corresponding BLG sequence [Fig. 1(A)]. It was hypothesized that if LPI could assume the BLG conformation, then the dimer interface would be formed as a consequence of the BLG-specific AB-loop and I-strand sequences. However, LPI did not dimerize, and the authors concluded that the monomeric property of ELG cannot attribute solely to the difference in the amino acid sequence of the AB-loop and the I-strand.
For this study, we constructed a new chimeric LG, named Gyuba, which means cow and horse in Japanese. With the exceptions of positions 81 and 121, Gyuba was constructed by joining the secondary structures from BLG with those of the ELG loops. Residues 17–27 (A-strand), 42–49 (B-strand), 53–62 (C-strand), 66–73 (D-strand), 80–86 (E-strand), 89–95 (F-strand), 102–108 (G-strand), 117–124 (H-strand), 130–138 (α-helix), and 145–153 (I-strand and its flanking sequences) were regarded as the secondary structures, and the other sequences were regarded as loops [Fig. 1(A)]. Residue 81 is Val in BLG and Glu in ELG. Although this residue is in the E-strand and by the criteria described above should be a Val in Gyuba, it was replaced with Glu so that the EcoRI site in the gene remained intact. Residue 121 is Cys in BLG and Tyr in ELG. This residue is in the H-strand and was changed to Ala to prevent thiol-disulfide exchange. We hypothesized that all interstrand and sheet-helix interactions would be maintained in Gyuba and therefore expected it to dimerize. We thus characterized the Gyuba structure and investigated how specific residues dictate interactions between the various secondary/tertiary structures, and thus the overall conformation, of Gyuba.