1. Introduction

To elucidate the mechanism by which proteins fold into their unique three-dimensional structure, it is essential to examine the stabilization mechanism of the native and non-native states. Depending on the protein and/or the method of denaturation, the protein shows a variety of conformational states, i.e. the native state, partially folded states with several native-like features (e.g. the molten globule state) and a highly denatured state resembling a random coil (Ptitsyn, 1995; Kataoka & Goto, 1996; Kataoka et al., 1997). In some cases, it has been shown that some residual structures exist in several proteins even when they are highly denatured (Neri et al., 1992; Shortle, 1996). These remaining structures are suggested to play a key role as the potential initiation site at an early stage of protein folding kinetics (Wong et al., 1996; Englander, 2000).

The molten globule state is a partially folded conformation and is characterized by the presence of substantial secondary structure, a compact shape and the absence of extensive rigid side-chain packing interactions. It is observed as an equilibrium unfolding/refolding intermediate under mildly denaturing conditions for many globular proteins such as α-lactalbumin, cytochrome c, apomyoglobin, ribonuclease HI and β-lactoglobulin (Kuwajima et al., 1976; Griko et al., 1988; Kuwajima, 1989; Ptitsyn, 1995; Dabora et al., 1996). Studies on the above proteins by a stopped-flow circular dichroism (CD) technique and by a pulse-labeled hydrogen-exchange technique coupled with two-dimensional NMR spectroscopy have shown that a transient intermediate similar to the equilibrium molten globule state is formed at an early stage of protein folding. Therefore, the molten globule state at equilibrium is a good model system for the study of the stabilization mechanism of folding intermediates.

Equine β-lactoglobulin (ELG) is a major component of milk and is a member of the lipocalin family. The members of this family share an eight-stranded (labeled A-H) up-and-down β-barrel structure with an α-helix and an off-barrel strand I. ELG takes a stable molten globule state (A state) at acidic pH (Ikeguchi et al., 1997). The A state has a significant amount of secondary structure including non-native α-helices. Small-angle X-ray scattering (SAXS) and analytical ultracentrifugation have shown that the protein in this state is nearly as compact as that in the native state (Ikeguchi et al., 1997; Yamada et al., 2005). The stopped-flow CD experiment has shown that the kinetic folding intermediate is indistinguishable from the A state (Fujiwara et al., 1999). Recently, we have found that the molten globule state of ELG undergoes a conformational transition when the temperature decreases (Yamada et al., 2005). The cold-denatured (C) state has been characterized by SAXS as an expanded chain-like conformational state. Nevertheless, the CD spectrum has shown that the α-helix content in the C state is much larger than that in the A state. The C state is also thermodynamically distinguishable from the A state. The temperature-dependent conformational transition of ELG at acidic pH has been interpreted as follows. In the A state, the compact conformation is mainly stabilized by the non-local hydrophobic interactions. Lowering the temperature weakens the hydrophobic interaction and results in the expansion of the polypeptide chain. Because the non-local hydrophobic interactions are absent from the expanded conformation, local interactions become predominant and the amount of non-native α-helix increases.

The expansion of the polypeptide chain may occur at a low ionic strength and at acidic pH where the electrostatic repulsion is strong. It has actually been observed for the acid-denatured cytochrome c and apomyoglobin (Goto et al., 1990; Kataoka et al., 1993; Hamada et al., 1994). In the case of ELG, the chain expansion and concomitant increase of the helical content were expected at a low pH and low ionic strength. It has been found, however, that the helix formation does not occur even at a low ionic strength (Yamada et al., 2006). This may be due to the presence of disulfide bonds. ELG has two disulfide bonds, Cys66–Cys160 and Cys106–Cys119. Previously, we constructed a mutant in which Cys66 and Cys160 were replaced with alanine (C66A/C160A) and the chloride-ion concentration dependence of its CD spectrum was measured at acidic pH. The CD spectrum of C66A/C160A did show the increased helix content at a low ionic strength (Yamada et al., 2006). In this study, thus, we have investigated the chloride-ion concentration dependence of the molecular shape of C66A/C160A and wild-type ELG (WT) at acidic pH.

2. Materials and methods

WT was prepared from horse milk as previously described (Ikeguchi et al., 1997). C66A/C160A was expressed and purified as previously described (Yamada et al., 2006). Chemicals were of analytical grade from Nacalai Tesque or Wako Pure Chemical Industries (Osaka, Japan). The pH was adjusted to 1.7 with 100 mM phosphoric acid and the chloride-ion concentration was controlled with KCl.

3. Results

It is well known that long exposure probably causes some chemical damage in the protein so that the damaged protein becomes liable to aggregate. At acidic pH, ELG is also susceptible to radiation damage. This is apparent from Fig. 1, which shows the change in the integration of scattering intensity during the exposure to X-rays. Using a normal static cell (a mica-windowed cell with a 1 mm path length), the integral intensity increases with the exposure time. In this study, therefore, we used a static flow cell (Kihara & Nagamura, 1999). The flow rate was adjusted in such a way that protein molecules would be exposed to the X-ray beam for less than 1.5 s. As a result, using a static flow cell, the integral scattering intensity does not change during the long exposure (Fig. 1). Therefore, it is possible to average the scattering data over a long time.

Figure 1 Change of integrated scattering intensity during the exposure of the acid-denatured ELG at pH 1.5 to X-rays. Integration was done for entire signals from the CCD detector. Using a normal static cell (squares), the integrated scattering intensity gradually increases with the exposure time. Using a static flow cell (circles), the integrated scattering intensity does not change. X-ray intensities as ion-chamber current values for the measurements using a normal static cell and a static flow cell are 69 and 87 nA, respectively.

Fig. 2 shows the Guinier plot of WT and C66A/C160A at pH 1.7. The protein concentration is as low as 0.3–0.5 mg ml⁻¹, which is comparable to the protein concentration used for the previous CD study (Yamada et al., 2006). The Guinier plot of WT is nearly linear both in the absence and presence of chloride ion, although the deviation from linearity is remarkable in the absence of chloride ion. The scattering curves of C66A/C160A are different in the presence and absence of chloride ion. In the presence of 80 mM chloride ion, C66A/C160A gives a nearly linear Guinier plot. In the absence of chloride ion, however, the Guinier plot shows a curvature like that observed for unfolded proteins.

Figure 2 The normalized Guinier plots of WT (a) and C66A/C160A (b) in the acid-denatured state. The open circles show data in the absence of chloride ion. The squares show data in the presence of 80 mM chloride ion. The solid and broken lines show best-fit curves of the polynomial fitting function.

Fig. 3 shows the chloride-ion concentration dependence of R_g,app. The R_g,app of C66A/C160A increases with decreasing chloride-ion concentration. Although a similar tendency was observed for WT, its change is much smaller than that observed for C66A/C160A. These results indicate that C66A/C160A undergoes a conformational transition from a compact form to an expanded form with decreasing chloride-ion concentration.

Figure 3 The chloride-ion concentration dependence of Rg,app in the acid-denatured states of C66A/C160A (squares) and of WT (circles). The Rg of the native ELG (solid line), of the unfolded ELG in 8 M urea (pH 8.7) (dotted line) and of the unfolded C66A/C160A in 8 M urea (pH 8.7) (broken line) are shown.

Fig. 4 shows Kratky plots of WT and C66A/C160A in the presence and absence of chloride ion. In the presence of 80 mM chloride ion, Kratky plots of both C66A/C160A and WT have a clear peak around Q = 0.09 Å⁻¹, indicating the globular shape. In the absence of chloride ion, the Kratky plot of WT still has a peak around Q = 0.09 Å⁻¹, although the peak height is lower than that in the presence of chloride ion. On the other hand, the Kratky plot of C66A/C160A has no peaks in the absence of chloride ion, suggesting that the molecule has lost its globularity. This pattern of Kratky plot is similar to that in the cold-denatured or urea-unfolded states of ELG.

Figure 4 The normalized Kratky plots in the acid-denatured states of WT (open symbols) and C66A/C160A (filled symbols). The triangles and circles show a Kratky plot in the presence and absence of 80 mM chloride ion, respectively. The open diamonds shows a Kratky plot of WT in the unfolded state at pH 4.0 in the presence of 8 M urea.

4. Discussion

The secondary structures in the molten globule state are stabilized not only by short-range interactions but also by non-local interactions (Arai & Kuwajima, 2000). It has been shown for apomyoglobin, cytochrome c and α-lactalbumin that the helices in the molten globule state are destabilized when the molecule is expanded at low temperatures, where the hydrophobic contraction is weak, or at low ionic strength, where the electrostatic repulsion is strong. We have shown that the cold-denatured ELG takes an expanded chain-like conformation at lower temperatures (Yamada et al., 2005). In this study, we have also confirmed that C66A/C160A is expanded at low anion concentration. As expected, the molecular dimension is well correlated with the helix content. Previously, we have shown that the cold-denatured ELG shows a CD spectrum indicating a significant amount of α-helix (Yamada et al., 2005) and that C66A/C160A shows a helical CD spectrum at low ionic strength (Yamada et al., 2006). Thus, the helix content in the acid-denatured state increases when the molecule is expanded. In contrast to helices in the molten globule states of other proteins, the helix content in the A state increases when the molecule is expanded. This is explained as follows. In the A state, the secondary structure is mainly determined by non-local interactions. When they are absent in the expanded conformation, local interactions becomes predominant and the amount of non-native α-helix increases.

It is interesting that C66A/C160A is significantly expanded in the absence of chloride ion while WT is not. The Cys66–Cys160 disulfide bond would suppress the expansion of the polypeptide chain so that WT is not expanded even in the absence of chloride ion. At low temperatures, however, WT is expanded although it has intact disulfide bonds. This result indicates that the electrostatic repulsion is not sufficient for expanding the disulfide-containing molecule and the low temperature is effective for expanding it. In fact, C66A/C160A is not expanded to the same degree as the cold-denatured state. Even in the absence of chloride ion, R_g,app of C66A/C160A (32 Å) is smaller than that in the cold-denatured ELG (37 Å). Unfortunately, the poor signal-to-noise ratio (Fig. 4) does not allow detailed comparison between the scattering curves of C66A/C160A in the absence of chloride ion and of the cold-denatured ELG.