Langmuir 2004, 20, 6841-6846
6841
Influence of Calcium on the Self-Assembly of Partially Hydrolyzed r-Lactalbumin Johanna F. Graveland-Bikker,*,† Richard Ipsen,‡ Jeanette Otte,‡ and Cornelis G. de Kruif†,§ NIZO food research, P.O. Box 20, 6710 BA Ede, The Netherlands, Department of Dairy and Food Science, The Royal Veterinary and Agricultural University, Rolighedsvej 30, DK-1958 Frederiksberg C, Denmark, and Van’t Hoff Laboratory, Debeye Research Institute, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands Received February 18, 2004. In Final Form: May 4, 2004 Self-assembly of R-lactalbumin after partial hydrolysis by a protease from Bacillus licheniformis can result in nanotubular structures, which show many similarities to microtubules. Calcium plays a crucial role in this process. The objective of this investigation was to study the role of calcium in more detail. The kinetics of the hydrolysis step and the self-assembly step were monitored by respectively liquid chromatography-mass spectrometry and dynamic light scattering. The microstructure of the gels finally formed was investigated by transmission electron microscopy. This investigation demonstrates that calcium accelerated the kinetics of the self-assembly, but it had no effect on the hydrolysis kinetics. As a result of the accelerated self-assembly kinetics at a high calcium concentration, the time of gelation decreased as well. A minimum concentration of calcium needed to obtain the tubular R-lactalbumin structures was determined. Below R ) 1.5 (mole calcium/mole R-lactalbumin), turbid gels with randomlike structure were obtained. Between R ) 1.5 and R ) 6, translucent gels with a fine stranded network of tubules were formed, while higher calcium concentrations had a negative effect on the tubule formation, resulting in amorphous structures. The optimum calcium concentration for R-lactalbumin nanotube formation seemed to be around R ) 3.
* Corresponding author. Phone: +31 (0)318 659511. Fax: +31 (0)318 650400. E-mail:
[email protected]. † NIZO food research. ‡ The Royal Veterinary and Agricultural University. § Utrecht University.
was accompanied by a dramatic increase in the β-structure content. Ipsen et al. found that R-lactalbumin after partial hydrolysis by a protease from Bacillus licheniformis formed nonbranching, apparently hollow strands with a diameter of 20 nm.5 In both studies, calcium played an important role; it is needed to form the rodlike assemblies. Ipsen et al. proposed that the calcium molecule forms a salt bridge between two negatively charged carboxyl groups, which are formed after hydrolysis.5 They studied the effect of calcium from R ) 3 up to R ) 18 (mole calcium/ mole R-lactalbumin). One of the conclusions of their study was that the addition of calcium made the strands become fuzzier. Although Ipsen et al. mentioned that at R ) 1 no gel was formed upon hydrolysis, the effect of calcium at concentrations lower than R ) 3 was not studied in detail.5 Therefore, in this study the influence of calcium on the self-assembly of partially hydrolyzed R-lactalbumin at a low calcium concentration was investigated. We showed that a minimum concentration of calcium of R ) 1.5 is needed to form the nanotubes, while a calcium concentration higher than R ) 6 had a negative effect on tube formation. Below R ) 1.5, a turbid white gel was formed, while at a higher calcium concentration the gel showed transparency, which increased at a higher calcium concentration, up to R ) 3. Further increasing of the calcium concentration, up to R ) 10, made the turbidity increase again. Furthermore, calcium was shown to accelerate tube formation. We demonstrated that this was purely an effect of self-assembly kinetics, because calcium did not affect the hydrolysis kinetics, as measured using dynamic light scattering (DLS) together with liquid chromatographymass spectrometry (LC-MS).
(1) Rochet, J. C.; Lansbury, P. T. J. Curr. Opin. Struct. Biol. 2000, 10, 60-68. (2) Scheibel, T.; Parthasarathy, R.; Sawicki, G.; Lin, X.; Jaeger, H.; Lindquist, S. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 4527-4532. (3) Reches, M.; Gazit, E. Science 2003, 300, 625-627.
(4) Goers, J.; Permyakov, S.; Permyakov, E.; Uversky, V.; Fink, A. Biochemistry 2002, 41, 12546-12551. (5) Ipsen, R.; Otte, J.; Qvist, K. B. J. Dairy Res. 2001, 68, 277-286.
Introduction Many proteins have the ability to form fibrillar aggregates. Some fibrillar structures, such as cytoskeleton components, are from natural sources, and others are made artificially. In medicine there is renewed interest in selfassembly of proteins and peptides in fibrillar structures, because a number of diseases, known as amyloidoses, originate from the deposition of stable aggregates, the so-called amyloid fibrils. Understanding the nature of selfassembled fibrillar structures in general promises many potential applications of these fibrillar self-assemblies. For example, it will contribute to the development of inhibitors for amyloid fibrils.1 Another field of application is nanotechnology. In recent papers, fibrillar protein structures served as a template for nanowires to be used in nanochip technology.2,3 Furthermore, because linear protein aggregates exhibit a high intrinsic viscosity, fibrils made of food proteins may be of interest as food thickeners. To develop all kinds of applications, an extensive knowledge of the assembly mechanisms is needed. The complex mechanism of fibrillogenesis is still not fully understood, and it requires further research. R-Lactalbumin, a 14.2-kDa Ca2+-binding whey protein, is one of the food proteins known to form fibrillar structures. Goers et al. showed that bovine R-lactalbumin formed amyloid fibrils at a low pH, where it adopts the classical molten globulelike conformation.4 Fibrillation
10.1021/la049579v CCC: $27.50 © 2004 American Chemical Society Published on Web 06/25/2004
6842
Langmuir, Vol. 20, No. 16, 2004
Graveland-Bikker et al.
Materials and Methods Materials. Bovine R-lactalbumin was purified at NIZO food research, by ion-exchange chromatography on a Staeck system (Pharmacia). The lyophilized powder consisted of 89.3% (w/w) dry matter, of which 97% (w/w) was R-lactalbumin and 0.26% (w/w) calcium, which corresponds to a molar ratio of R ) 0.9 (Ca2+/R-lactalbumin). A serine protease from Bacillus licheniformis (BLP), specific for Glu-X and Asp-X bonds,6 was kindly provided by Novozymes A/S (Bagsværd, Denmark). Tris, calcium chloride, and trifluoroacetic acid (TFA) were from Merck (Darmstadt, Germany), and highly purified water was used (Milli-Q Plus, Millipore Corp., Bedford, U.S.A.). For all experiments, 3% (w/w) R-lactalbumin powder was dissolved in 75 mM Tris-HCl buffer at pH 7.5 and the desired amount of CaCl2 was added. To start the hydrolysis, BLP was added to a level of 4% (w/w BLP/R-lactalbumin). After mixing, the solution was immediately filtered using a 0.1-µm filter (low protein binding, Millipore, France) and incubated at 50 °C in the DLS equipment, in a rheometer or in a waterbath (for LC-MS and transmission electron microscopy, TEM). Methods. DLS. DLS experiments were performed using a Malvern Autosizer IIC. The system consisted of a Malvern PCS41 optics unit with a 5-mW helium neon laser and a Malvern K7032ES correlator used in serial configuration. The scattering angle was fixed at 90°, and the wavelength of the laser beam was 632.8 nm, resulting in a wavevector of 0.0187 nm-1. From the fluctuation of the scattering intensity, the intensity autocorrelation function was obtained. Subsequently, the electric field autocorrelation function was fit with a second cumulant fit, from which the (translational) diffusion coefficient was obtained. Via the Stokes-Einstein relation, assuming a spherical object, an apparent hydrodynamic diameter was derived (the word apparent is used, because the particles were not spherical but linear). LC-MS. For LC-MS, hydrolysate samples were taken at various times during incubation and diluted eightfold with 0.1% TFA to lower the pH below 2.5 and stop the hydrolysis. The samples were stored at -20 °C until analysis. LC-MS analysis was performed using an Agilent 1100 LC-MSD trap (Agilent Technologies A/S, Naerum, Denmark) with LC/MSD Trap Control and Data Analysis Software, Version 4.0. An extended C18 column (2.1 × 150 mm, 5 µm) was used together with an extended guard column (2.1 × 12.5 mm), both from Agilent Technologies A/S. Buffer A was 0.1% TFA in water, and buffer B was 0.1% TFA in 90% acetonitrile. The gradient consisted of 8-55% B in 40 min, followed by 100% B for 6 min and re-equilibration with 8% B for 10 min. Ten microliters were injected, and the components were separated using a flow rate of 0.25 mL/min and detected at 210 nm. The trap was operated in the positive mode with a nebulizer pressure of 50 psi, a flow of nitrogen of 9 L/min, and a drying temperature of 300 °C. Scans from 50 to 2200 m/z were taken at the normal scan resolution, and the target mass was set to 1000 m/z. TEM. Samples for TEM were taken at different times during incubation and immediately diluted 1:1 with fixative (1.5% glutaraldehyde in 0.1 M sodium cacodylate buffer) and stored at 4 °C. TEM negative staining was performed as follows. A Formvar layer on a copper grid was coated with carbon. It was immersed in 5 µL of the (10-fold diluted) sample and blotted after 1 min. Then the copper grid was washed with water for 1 min and subsequently blotted, to remove excess protein material. To enhance the contrast, the copper grid was then immersed in 3% (w/w) uranyl acetate and blotted after 1 min. Examination was done using a Philips TEM 410 electron microscope, operated at 60 kV. BLP-induced gel of R-lactalbumin was made by incubating for 20 h at 50 °C followed by equilibration for 30 min at room temperature. Small cubes were cut from the gel and fixed in glutaraldehyde. Post-fixation was done using OsO4 (10 g/L in 0.1 M cacodylate buffer, pH 7.3). Dehydration, embedding, and cutting were performed as described previously.7 Examination was performed using a Philips CM-100 electron microscope PW (6) Breddam, K.; Meldal, M. J. Eur. Biochem. 1992, 206, 103-107. (7) Otte, J.; Ju, Z. Y.; Skriver, A.; Qvist, K. B. J. Dairy Sci. 1996, 79, 782-790.
Figure 1. (A) Gel after incubation of 3% (w/w) R-lactalbumin with 4% (w/w) BLP for 3 h at 50 °C (75 mM Tris-HCl, pH 7.5, R ) 3.0). The black horizontal lines are on the background paper. (B) Positively stained electron micrograph of the gel microstructure of A.
Figure 2. Negative staining electron micrograph of hydrolyzed R-lactalbumin nanotubes. 6020 (Philips Electron Optics, Eindhoven, The Netherlands) operated at 60 kV. Rheological Measurements. To determine the time of gelation of the BLP-induced gels at three different calcium concentrations, the incubation was carried out in the C-8 measuring system in a controlled-strain rheometer (Bohlin VOR, Bohlin, Ltd., Cirencester, U.K.). Immediately after mixing the R-lactalbumin and BLP, the solution was introduced into the measuring system, which was preheated at 50 °C. The solution was subjected to dynamic oscillation with a strain of 0.005 and a frequency of 0.5 Hz. Gelation was defined as the point at which the storage modulus (G′) was g 1 Pa.
Results and Discussion Incubation at 50 °C for several hours of 3% (w/w) R-lactalbumin at R ) 3 (mole calcium/mole R-lactalbumin) with 4% (w/w) BLP led to gel formation. The appearance of the gel was transparent (Figure 1A). In general, transparency of a gel can be attributed to the fine strandedness of the network. A transmission electron micrograph of a positively stained cross section of the gel formed after 20 h of incubation indeed showed a fibrillar network structure (Figure 1B). There are many open spaces in the gel, as a result of the effective connection of the long fibrils. Figure 2 shows a more detailed micrograph of the fibril structure. The structure clearly showed two
Tubular R-Lactalbumin Self-Assemblies
Figure 3. Influence of calcium on the kinetics of self-assembly [3% (w/w) R-lactalbumin, 4% (w/w) BLP, 75 mM Tris-HCl, pH 7.5, 50 °C]. R ) 0.9 (+), 1.2 (2), 1.5 (O), 1.7 (9), 2.0 (3), 2.5 (b), 3.0 (]), 4.0 (1), 6.0 (4), and 10 ([).
bright lines separated by an electron-dense line, indicating that it is a tubular structure. Ipsen et al. previously indicated this feature using positive staining electron microscopy.5 Here with another technique, namely, negative staining electron microscopy, the cavity of the tube is even clearer. The outer diameter was around 20 nm, while the inner diameter was about 7 nm. It has to be taken into account that these samples had been dried (vacuum), and, therefore, these values are not necessarily correct, but they give an indication. When the formation of the tube was monitored over time, first short tubes with a length of tens of nanometers were formed, followed by rapid growth into long tubes with a length of hundreds of nanometers (micrographs not shown). Furthermore, although no quantification of the electron micrographs was performed, it seemed that the number of tubular structures did not increase, which indicates that growth of the tube took place instead of formation of new tubes. This may imply that the mechanism is nucleation and growth, as has been observed for many fibril formation mechanisms.8 The aim of this work was to study the influence of calcium on the self-assembly of partially hydrolyzed R-lactalbumin, to enable determination of the minimum calcium concentration needed for tubule formation. Therefore, 3% (w/w) R-lactalbumin solutions were incubated at different calcium concentrations, ranging from R ) 0.9 to R ) 10. At all calcium concentrations, a gel was formed, except at R ) 10, where a viscous turbid solution was obtained. Figure 3 depicts the kinetics of self-assembly formation. Both the scattering intensity and the apparent hydrodynamic radius increased with time (Figure 3 only presents the scattering intensity). At all calcium concentrations studied, the scattering intensity first remained constant, then increased rapidly because of the formation of large and many self-assemblies. The length of this period of constant scattering intensity, or lag time, decreased with increasing calcium concentration. Furthermore, the increase in intensity was faster at higher calcium concentrations. Thus, calcium had a major influence on the rate of self-assembly formation. Increasing the calcium concentration threefold (from R ) 0.9 to R ) 3) led to the self-assembly rate increasing by almost 1 order of magnitude. At the lowest and the highest calcium concentrations measured, the scattering intensity increased to larger (8) Thirumalai, D.; Klimov, D. K.; Dima, R. I. Curr. Opin. Struct. Biol. 2003, 13, 146-159.
Langmuir, Vol. 20, No. 16, 2004 6843
values (up to about 900 au compared to about 400 au at intermediate calcium concentrations), before a gel was formed. This can be explained by the fact that in these cases random aggregates were formed (which will be shown later), and these scatter more light than linear structures. Figure 4 shows micrographs of the structures formed, together with the appearance of the gel expressed in words for the different calcium concentrations. At the lowest concentrations tested (R ) 0.9 and 1.2), a turbid gel was formed. This turbidity can be explained by the microstructure, which consisted of amorphous aggregates only. Thus, this calcium concentration was too low for the tubules to form. When the calcium concentration was raised to R ) 1.5, the gel formed was slightly whitish and transparent. This can be explained by the microstructure, because tubular structures were present. However, also some randomly aggregated material was visible. With a further increase in calcium concentration up to R ) 3, the tubular structures were clearer, and less random material was present. By eye it was possible to see a slight increase in transparency with higher calcium concentrations. This can be explained by the decreasing amount of amorphous aggregates with increasing calcium concentrations. A further increase in calcium concentration, however, led to an increase in the amount of unstructured material, as is visible for R ) 6 (Figure 4). This finally led to a turbid solution at R ) 10, with mainly contained random aggregates. From this calcium series, it can be concluded that the minimum calcium concentration needed for nanotube formation is 1.5 mole Ca2+/mole R-lactalbumin. Too high a calcium concentration, however, had an adverse effect on the tubule formation. An explanation for this could be the increase in ionic strength. At a calcium concentration of R ) 10, the ionic strength due to calcium chloride is 50 mM. This could be high enough to screen some important electrostatic charges needed to form the R-lactalbumin nanotubes. Ipsen et al. also found changes as a function of calcium concentration. They reported increased fuzziness of the strands upon increased calcium concentration, but this was at higher calcium concentrations (R ) 10 and higher).5 This shift could be due to the different R-lactalbumin concentrations they used, namely, 10% (w/w) instead of 3% (w/w) in this study. From the results in Figure 4, it can be concluded that there seems to be an optimal calcium concentration for R-lactalbumin nanotube formation. This concentration would be around R ) 3-4. The rest of the study was focused on a smaller window of calcium concentration, namely, R ) 0.9-3, including the minimum calcium concentration of R ) 1.5 and, hence, the transition between random and tubular structures. From the DLS results, it is clear that calcium had a major influence on the kinetics of assembly formation. But, because these kinetics were dependent on both the hydrolysis kinetics and the self-assembly kinetics, it was not possible to distinguish the influence of calcium on these two steps individually. Therefore, the influence of calcium on the hydrolysis kinetics was determined using LC-MS. Figure 5 shows the profile of hydrolysis products with time for a calcium concentration of R ) 2.0. The molar masses of these products as determined by MS are presented in the legend of this figure. Before hydrolysis, only R-lactalbumin was present in the sample, coinciding with peak A with a retention time of about 36 min. After 0.1 h, already 37% of the R-lactalbumin had been hydrolyzed into a component with a molar mass of 11 575 Da (peak B), which is a more hydrophobic component
6844
Langmuir, Vol. 20, No. 16, 2004
Graveland-Bikker et al.
Figure 5. Chromatograms of the hydrolysates of 3% (w/w) R-lactalbumin with 4% (w/w) BLP (75 mM Tris-HCl, pH 7.5, R ) 2.0, 50 °C) after different incubation times. Molar masses: (A) 14 176; (B) 11 577; (C) 10 119; (D) 10 277; (E) 9863; (F) 11 260; (G) 9802; (H) 8818; (I) 10 404; (J) 10 483; (K) 13 728; (L) 1864; (M) 1882; (N) 1508; (O) 1527; (P) 1504; (Q) 1476; (R) 1120; and (S) 550 Da.
Figure 6. Influence of calcium on the degradation of R-lactalbumin [3% (w/w) R-lactalbumin, 4% (w/w) BLP, 75 mM TrisHCl, pH 7.5, 50 °C; R (+) 0.9, (O) 1.5, (b) 2.0, (3) 2.5, and (]) 3.0].
Figure 4. Transmission electron micrographs of 3% (w/w) R-lactalbumin incubated with 4% (w/w) BLP at 50 °C (75 mM Tris-HCl, pH 7.5) at different calcium concentrations. Micrograph is 2100 nm × 2600 nm.
(retention time 37 min), and some small peptides. Further degradation resulted in two other components (peaks C and D) with molar masses of respectively 10 119 and 10 277 Da. After 1 h, all the R-lactalbumin had been hydrolyzed. By then, two new hydrolysis products had been formed, components E (9863 Da) and F (11 260 Da), and later on another two components (G and H after respectively 1.5 and 3.0 h). It appeared that the hydrolysis of R-lactalbumin by BLP resulted in a group of only partially hydrolyzed R-lactalbumin molecules with a molar mass around 11 kDa and just a few peptides. The probable reason that the hydrolyzed R-lactalbumin molecules remained rather intact is the combination of the limited amount of cleavage sites for BLP and the four disulfide bridges keeping the hydrolyzed parts of the molecule together. During the first 30 min, which is the period when no self-assembly took place, the degradation of R-lactalbumin seemed to be identical for the different calcium concentrations. This was visible from the respective chromatograms, which all showed the same curves (data not shown). Quantification of the influence of calcium on the primary hydrolysis of R-lactalbumin with time was done by calculating the area of the R-lactalbumin peak; the results are presented in Figure 6. From this figure, it is clear that the degradation of R-lactalbumin followed first-order kinetics as expected. Furthermore, calcium had no influence on the degradation kinetics. By averaging all data,
Tubular R-Lactalbumin Self-Assemblies
Langmuir, Vol. 20, No. 16, 2004 6845
Figure 7. Chromatograms of the hydrolysates of 3% (w/w) R-lactalbumin at different molar calcium ratios incubated with 4% (w/w) BLP for 90 min (75 mM Tris-HCl, pH 7.5, 50 °C).
Figure 8. Storage modulus of 3% (w/w) R-lactalbumin incubated with 4% (w/w) BLP at different molar calcium concentrations (75 mM Tris-HCl, pH 7.5, 50 °C).
the following kinetic relation was obtained: [R-lactalbumin]t ) [R-lactalbumin]0 exp(-4.4t). Svendsen and Breddam reported increased activity of the enzyme by adding CaCl2 (up to 30 µM), but this was only at lower concentrations than those used in this study.9 Figure 5 only shows the results of the hydrolysis at one of the calcium concentrations measured (R ) 2.0). As explained before, at all other calcium concentrations the hydrolysis profiles with time were identical, but only for the initial degradation, when no self-assembly had occurred. After more than 30 min of incubation, the hydrolysis products started to differ. For comparison, the hydrolysis products at the different molar calcium ratios after 90 min are presented in Figure 7. Here it is apparent that the profiles differ from each other; the profiles were similar up to a retention time of 35 min, but then the major peaks (retention time 35-38 min) were different for the different calcium concentrations. When these peaks are compared with the ones in Figure 5, it can be seen that at lower calcium concentrations the protein was more degraded. This difference in the degree of hydrolysis can be explained by the self-assembly of the hydrolysis products. At a high calcium concentration, many of the hydrolyzed R-lactalbumin molecules were already incorporated into the self-assemblies after 90 min (at R ) 3, the self-assembly of the building blocks already started after 35 min) and, thus, hardly accessible for the enzyme. At lower a calcium concentration, however, there was no self-assembly yet and the enzyme could continue further degradation of the hydrolysate. As a result, it seemed that the calcium concentration had an influence on the enzyme kinetics, but in fact this apparent influence was only the self-assembly of the hydrolyzed products, which made them no longer available for hydrolysis. Figure 7 also shows that the concentration of smaller hydrolysis products, namely, the peptides with a retention time around respectively 10 and 20 min, were the same for all calcium concentrations. This indicates that these peptides were not incorporated into the self-assembled structures. From the LC-MS together with the DLS results, it was concluded that the influence of calcium on the selfassembly formation, as observed by DLS, is completely due to self-assembly kinetics and not due to enzyme kinetics. This means that the building blocks (hydrolysate) of the self-assembled structures were produced equally fast at the different calcium concentrations, but selfassembly of these building blocks was highly accelerated by the presence of calcium. It could well be that calcium
facilitates the formation of a nucleus, which is needed to form the tubules. Presumably the Ca2+ ion acts as specific glue between two building blocks. Because many carboxyl groups are present in the building blocks, as a result of the presence of aspartic and glutamic acid and, moreover, the presence of carboxyl groups formed by hydrolysis of the peptide bonds, it is likely that the Ca2+ ion binds between two or more such carboxyl groups. It is known that calcium has a high affinity for carboxyl groups, because the calcium binding loop of R-lactalbumin mainly consists of aspartic acid residues.10 The binding of calcium between two specific amino acid residues could then determine the orientation of the building blocks with respect to one other, and as a result, they form a ring structure. The accelerated kinetics caused by increased calcium concentration could be explained by nucleation being faster the higher the calcium concentration. In Figure 8 the storage modulus G′ is plotted as a function of time, for three different calcium concentrations. This shows that at all calcium concentrations measured a gel was formed. Gelation occurred after respectively 1.3, 2.6, and 10 h for calcium concentrations R ) 3.0, 2.0, and 0.9. The higher the calcium concentration, the faster a gel was formed, as was expected from the light-scattering data (Figure 4), which indicated that the formation of self-assemblies was faster when the calcium concentration was higher. The difference in gelation times (respectively 1.3, 2.6, and 10 h) agrees well with the time at which slow relaxation times (which were translated to an apparent hydrodynamic diameter) were reached (for example, an apparent particle size of 1000 nm was reached at respectively 2, 3, and 10 h). Concerning the gel strength, it appeared that G′ was much higher at high calcium concentrations, but a longer measurement, to reach plateau values, is needed to draw definite conclusions on this. To check the specificity of calcium in the self-assembly process, other divalent and trivalent cations were used to produce the tubular structures consisting of partially hydrolyzed R-lactalbumin. It appeared that the divalent manganese and zinc and the trivalent aluminum could also induce R-lactalbumin nanotube formation. By using these ions instead of calcium at R ) 3, transparent gels as in Figure 1 were obtained. The structures formed appeared to be tubular structures similar to those formed with calcium. They had a diameter of about 20 nm, and the length of the tubules was up to micrometers long, as
(9) Svendsen, I.; Breddam, K. Eur. J. Biochem. 1992, 204, 165-171.
(10) Stuart, D. I.; Acharya, K. R.; Walker, N. P. C.; Smith, S. G.; Lewis, M.; Philips, D. C. Nature 1986, 324, 84-87.
6846
Langmuir, Vol. 20, No. 16, 2004
was visualized using TEM (micrographs not shown). These results show that the role of calcium in the tube formation is not a specific one. Apparently, the “glue” between two building blocks can be any ion that contains a sufficient positive charge to form a bridge between two or more specific carboxyl groups and fits in the orientation of these groups. Presumably there are also divalent ions that cannot replace calcium to form a specific salt bridge, because they do not fit in the geometry of the neighboring groups. This needs to be investigated more intensively, and in a following study this will be discussed in more detail. Conclusion This study shows that a minimum concentration of calcium was needed to form tubular structures and, subsequently, transparent gels, from partially hydrolyzed R-lactalbumin. Below a calcium concentration of R ) 1.5 (mole calcium/mole R-lactalbumin), random aggregates were formed, resulting in a white turbid gel. At a calcium ratio higher than R ) 1.5, tubular structures were formed, which led to a whitish transparent gel. Furthermore, the higher the calcium concentration (up to R ) 3), the more
Graveland-Bikker et al.
the tubular shape was favored. But further increase of calcium had a negative effect; finally, only random aggregates were formed (at R ) 10). Calcium did not influence the hydrolysis kinetics, but it had a major effect on the kinetics of self-assembly of partially hydrolyzed R-lactalbumin, possibly by facilitated nucleation. By increasing the calcium concentration, the rate of selfassembly was increased and, thus, a gel was formed faster. Acknowledgment. The authors thank the European Community - Access to Research Infrastructure action of the Improving Human Potential Programme, Friesland Coberco Dairy Foods, AVEBE, NIVE, and the Ministry of Economic Affairs through IOP for financial support. Furthermore, the authors thank Hans van der Meulen from the Center of Electron Microscopy of Leiden University Medical Center (Leiden, The Netherlands) for assistance in negative staining electron micrscopy, Pia Skjødt Pedersen for TEM sample preparation of the gel, and Mila Zakora for LC-MS measurements. Finally, the authors thank Novozymes A/S for supplying the enzyme. LA049579V
