- Top of page
- Materials and methods
The stability toward thermal and urea denaturation was measured for HAMLET (human α-lactalbumin made lethal to tumor cells) and α-lactalbumin, using circular dichroism and fluorescence spectroscopy as well as differential scanning calorimetry. Under all conditions examined, HAMLET appears to have the same or lower stability than α-lactalbumin. The largest difference is seen for thermal denaturation of the calcium free (apo) forms, where the temperature at the transition midpoint is 15°C lower for apo HAMLET than for apo α-lactalbumin. The difference becomes progressively smaller as the calcium concentration increases. Denaturation of HAMLET was found to be irreversible. Samples of HAMLET that have been renatured after denaturation have lost the specific biological activity toward tumor cells. Three lines of evidence indicate that HAMLET is a kinetic trap: (1) It has lower stability than α-lactalbumin, although it is a complex of α-lactalbumin and oleic acid; (2) its denaturation is irreversible and HAMLET is lost after denaturation; (3) formation of HAMLET requires a specific conversion protocol.
In the classical picture, each gene encodes for one polypeptide chain that folds into a unique protein structure, representing the thermodynamically most stable state, with one biological function. For some proteins, the native state is not the most stable structure, but is a kinetically trapped state in a local free energy minimum. Examples of proteins with a kinetically trapped native state are α-lytic protease (Jaswal et al. 2002), serine protease inhibitors (serpins) (Huber and Carrell 1989), and viral membrane fusion proteins (Carr and Kim 1993). A well-ordered native state is not required for a protein to be functional (Wright and Dyson 1999; Uversky 2002a,b). Many proteins involved in cell cycle regulation, DNA and RNA recognition, and signal transduction have an intrinsically disordered structure and adopt a structured state first upon binding their target (Dyson and Wright 2002). The free energy penalty associated with the conformational change will decrease the affinity while the specificity can be maintained, which is feasible in transient cellular events. It is also suggested that the disordered state could be a way of achieving larger intermolecular interfaces without an increase of the crowding effect or larger cell size (Gunasekaran et al. 2003). Sequence analysis of eukaryotic genomes suggests that 35%–51% of the genes code for proteins that are disordered (Dunker et al. 2001).
HAMLET (human α-lactalbumin made lethal to tumor cells) is a folding variant of human α-lactalbumin that induces apoptosis in tumor cells while sparing healthy cells, both in cell culture (Håkansson et al. 1995, 1999; Svanborg et al. 2003) and in tissue (Fischer et al. 2004; Gustafsson et al. 2004). Regular α-lactalbumin is a globular high-affinity Ca2+-binding protein present in the milks of most mammals. It binds to galactosyl transferase in the lactose synthase complex that catalyzes the formation of lactose in the mammary gland (Brew and Hill 1975; Ramakrishnan et al. 2001, 2002). The 123 amino acids of α-lactalbumin are organized into a smaller β-subdomain and a larger α-subdomain separated by a cleft and interconnected by four disulfide bonds (Fig. 1; Acharya et al. 1989, 1991; Chrysina et al. 2000). HAMLET is a partially unfolded conformer that is stabilized by a specific fatty acid cofactor (Fig. 1B). HAMLET binds to the surface of tumor cells, enters these cells, and eventually accumulates in the cell nucleus where it interacts with histones (Düringer et al. 2003). HAMLET has retained the high-affinity Ca2+-binding site of α-lactalbumin, but mutational studies have shown that a functional Ca2+-site is not required for the apoptosis-inducing effect (Svensson et al. 2003a).
Previous studies have shown that HAMLET has unusual thermodynamic properties. The complex is formed from calcium-free (apo) α-lactalbumin by ion exchange chromatography on a resin that is conditioned with oleic acid (Svensson et al. 2000, 2003b). It is not possible to produce active HAMLET by just mixing apo α-lactalbumin and oleic acid/oleate in solution (Svensson et al. 2000). Although HAMLET binds calcium with retained activity (Svensson et al. 2003a), HAMLET is not obtained by the chromatographic procedure if α-lactalbumin is applied in its calcium-bound form (Svensson et al. 2000). These findings show that the three-component system α-lactalbumin/oleic acid/calcium does not rapidly reach its thermodynamic equilibrium, suggesting high kinetic barriers between states. Compared to α-lactalbumin HAMLET may have lower, similar, or higher free energy. To address this question, we here investigate the relative stabilities of different folding states of α-lactalbumin including native, apo and molten globule forms, as well as HAMLET in its calcium-free and calcium-bound states. The stability toward urea and thermal denaturation is studied using far- and near-UV circular dichroism (CD), and tryptophan fluorescence spectroscopy, as well as differential scanning calorimetry (DSC). The data suggest that denaturation of HAMLET is an irreversible process, thus our experiments yield apparent stabilities. Nonetheless, our data support that HAMLET is less stable than α-lactalbumin. This means that HAMLET is a kinetically trapped, tertiary disordered fatty acid–protein complex with a novel and biologically beneficial function.
- Top of page
- Materials and methods
The stability of α-lactalbumin and HAMLET toward thermal and urea denaturation was studied under a range of conditions as regards salt and calcium concentration. Under all conditions examined, the stability of HAMLET was found to be comparable to or lower than that of α-lactalbumin. This is a remarkable result given that HAMLET is a complex between α-lactalbumin and oleic acid.
Ligand binding is often seen to enhance the stability of a protein toward denaturation, because the free energy of ligand binding factors into the observed unfolding equilibrium. The unfolding process that is observed in the presence of ligand represents both the release of ligand and denaturation of the protein (Shrake and Ross 1990). The apparent stability will therefore be higher the higher the concentration of ligand. The effect is seen at free ligand concentrations well above the dissociation constant of the protein–ligand complex as measured in the absence of denaturant. The two equilibria are linked, and the apparent ligand binding affinity becomes lower as conditions approach the denaturation transition.
For the high-affinity Ca2+-binding protein α-lactalbumin, it is well established that denaturation occurs at significantly higher temperature or denaturant concentration in the presence of calcium (Figs. 2, 5; Permyakov et al. 1985; Ikegushi et al. 1986; Kuwajima et al. 1986; Griko et al. 1999). The effect is seen as a gradual increase in stability all the way up to 10 mM Ca2+, while the Ca2+-dissociation constant in the absence of denaturant is around 100 nM at physiological salt concentration (Kronman 1989; Svensson et al. 2003a). The apparent Ca2+ affinity decreases when the conditions are altered to favor the denatured state (elevated temperature or increased denaturant concentration). This explains why a further stabilization is observed between 1 and 10 mM Ca2+, which is 104–105-fold higher than the KD that was measured under nondenaturing conditions.
In line with this, one might expect that HAMLET, which is a protein-oleic acid complex, would exhibit higher stability than the protein in the absence of bound fatty acid. On the contrary, HAMLET exhibits lower stability than α-lactalbumin at several solution conditions, and denaturation processes followed. This is one strong indication that HAMLET is a kinetic trap. The largest difference is seen for the calcium-free forms when thermal denaturation is followed by near-UV CD spectroscopy that reports on confinement of aromatic side chains in the tertiary structure (Fig. 2A).
It was recently shown that BAMLET, a protein–oleic acid complex with apoptosis-inducing capacity, could be formed from bovine α-lactalbumin (Svensson et al. 2003a). In analogy with the human proteins, the tertiary structure of BAMLET appears less stable toward thermal denaturation than that of bovine α-lactalbumin.
α-Lactalbumin has been used extensively over the last 40 years in folding studies because it is known to adopt different folds as a response to the solution conditions (Kronman and Andreotti 1964). At low pH, the protein forms a stable state that is more loosely packed than the native form but has native-like secondary structure. This state has been named the molten globule state (Dolgikh et al. 1981; Ohgushi and Wada 1983), and it lacks a rigid tertiary structure (Dolgikh et al. 1985; Pfeil et al. 1986; Kuwajima 1996; Veprintsev et al. 1997). In more detail, NMR and hydrogen exchange studies have shown that the molten globule state has a structured β-subdomain and a mainly disrupted α-subdomain (Schulman et al. 1995, 1997; Wijesinha-Bettoni et al. 2001). Its unfolding is noncooperative and the molten globule form exhibits a diffuse thermal transition, but a two-state transition in the presence of denaturants (Kuwajima 1996). Other states resembling the molten globule state are achieved by removing the tightly bound Ca2+, upon heating, or through disulfide reduction. All these states show normal reversible equilibrium behavior and the protein returns to the native state when conditions are reversed by adjusting the pH to neutral, adding Ca2+, cooling the solution, or oxidizing the disulfide bonds. The Ca2+-free (apo) state differs from the low pH molten globule in that it is more compact in salt solution and shows distinct transitions both on heating and solvent denaturation (Figs. 2, Figure 3., Figure 4.–5; Griko et al. 1994; Vanderheeren et al. 1996; Veprintsev et al. 1997).
HAMLET has been described as molten globule-like (Svensson et al. 2000) because it has reduced tertiary structure compared to the native protein but retained secondary structure, and because it is clearly not the same state as the classical pH 2.0 molten globule as it exists at neutral pH and in the presence of Ca2+ and it has a bound cofactor (Svensson et al. 2003a). Denaturation of apo HAMLET resembles that of apo α-lactalbumin much more than that of the molten globule. Both HAMLET and α-lactalbumin show distinct thermal transitions under all conditions examined, which is not true for the molten globule state (Fig. 2). Also, the Ca2+-bound forms of both HAMLET and α-lactalbumin show distinct thermal transitions in clear contrast to the molten globule state.
The denaturation of HAMLET by heat or urea was found to be irreversible. After denaturation, the biological activity against tumor cells was lost (Fig. 7) and repeated denaturation experiments are not identical. This is a second strong indication that HAMLET is a kinetically trapped state.
The stability we measure for HAMLET can only be viewed as an apparent stability since the complex will depart during the denaturation process. α-Lactalbumin binds oleic acid in solution (Fig. 8; Cawthern et al. 1997), but the complex that is formed is not HAMLET and has no apop-tosis-inducing activity.
Although HAMLET appears significantly less stable than α-lactalbumin toward thermal denaturation, there is no or only a small difference in stability toward urea denaturation. This discrepancy most likely arises from two factors: (1) the quite different denaturation mechanisms by heat and urea with different rates of folding and unfolding, and (2) the irreversibility of HAMLET denaturation. The secondary structure is not unfolded at 90°C, but denatures in urea solution. Thermal denaturation of the tertiary structure was monitored during a stepwise temperature change of 0.25°–1.0°C/min, while the urea denaturation involved equilibration of separate samples. The incubation time (1–3 h) seems to allow for a sufficient number of unfolding–folding cycles for the urea samples to reach equilibrium. Oleic acid may dissociate from HAMLET upon denaturation, which results in free fatty acid and denatured α-lactalbumin (Fig. 9). The protein may then renature into folded α-lactalbumin and the equilibrium established at each urea concentration is not for HAMLET, but for the α-lactalbumin–oleic acid complex that forms in solution. It has been shown in earlier studies that HAMLET can not be produced by simply mixing the protein and fatty acid, but requires a specific conversion protocol involving an oleic acid-conditioned ion exchange matrix (Svensson et al. 2000). In this work we find that HAMLET activity is lost after denaturation, and the renatured protein hence behaves as a mixture of α-lactalbumin and oleic acid.
In summary, there are three strong indications that HAMLET is a kinetically trapped conformer: (1) the lower apparent stability of HAMLET compared to α-lactalbumin; (2) the lack of reversibility—the biological activity is lost after denaturation; (3) the inability to produce HAMLET by simply mixing α-lactalbumin and oleic acid (Svensson et al. 2000).
It is still possible that HAMLET is the thermodynamically stable state under the conditions of at least one of the steps in the conversion procedure, and that there are high enough kinetic barriers separating it from α-lactalbumin or α-lactalbumin-oleic acid complex to trap the system in the HAMLET state when these conditions are removed. High kinetic barriers may be a means to devise multiple functions for α-lactalbumin, and to keep it in a particular state for enough time to serve its biological function.
Kinetically controlled protein states have been appreciated over the last 15 years (Baker and Agard 1994) with a number of mechanisms for kinetic control. α-lytic protease (Jaswal et al. 2002) has a pro region of 166 residues that is needed for the protein to attain the native state. For plasminogen activator inhibitor 1 (PAI-1) (Huber and Carrell 1989), the active serine protease inhibitor state appears due to faster folding kinetics than an inactive state that is thermodynamically more stable. For luciferase, the active heterodimeric enzyme forms 10-fold faster than homodimers. However, homodimers dissociate very slowly and represent a kinetic trap (Baldwin et al. 1993). For the prion protein (PrP), the disease mechanism involves the conversion from the monomeric α-helical native form (PrPC) into the β-sheet form (PrPSC), which forms toxic polymeric deposits (Zahn 1999). The kinetic barrier for the conversion process is very high. This may explain the low incident rate of disease. Disease is often coupled to single point mutations that reduce the barrier by stabilizing the transition state and thus speeding up the conversion.
The present results show that fatty acids can trap a protein in a state with elevated free energy and impose high kinetic barriers on the route toward a state with lower free energy.