Fibrillar Î²-Lactoglobulin Gels: Part 1. Fibril Formation and Structure

Biomacromolecules 2004, 5, 2408-2419

2408

Fibrillar β-Lactoglobulin Gels: Part 1. Fibril Formation and Structure Walraj S. Gosal,*,†,§ Allan H. Clark,‡,| and Simon B. Ross-Murphy† Department of Life Sciences, King’s College London, Franklin-Wilkins Building, 150 Stamford Street, London SE1 9NN, United Kingdom, and Unilever Research, Colworth House, Sharnbrook, Bedford MK44 1LQ, United Kingdom Received June 9, 2004; Revised Manuscript Received August 27, 2004

As a prelude to experimental and theoretical work on the mechanical properties of fibrillar β-lactoglobulin gels, this paper reports the structural characterization of β-lactoglobulin fibrils by electron and atomic force microscopy (AFM), infrared and Raman spectroscopy, and powder X-ray diffraction. Aggregates formed by incubation of β-lactoglobulin in various alcohol-water mixtures at pH 2, and in water-trifluoroethanol (TFE) at pH 7, were found to be wormlike (∼7 nm in width and 1 µm in length), smoother, and seemingly stiffer fibrils formed on heating aqueous β-lactoglobulin solutions at pH 2 and low ionic strength, although there was little evidence for the higher-order structures common in most amyloid-forming systems. Time-lapse AFM also revealed differences in the formation of these two fibril types: thermally induced aggregation occurring more cooperatively, in keeping with a nucleation and growth process. Only short stiff-rods (1 µm) become established, of average height 3.0 ( 0.7 nm (n was 57). Some of the “oligomeric” particles remained at this stage but seemed fewer in number. After 24 h, numerous, and seemingly entangled, fibrils (height 3.6 ( 0.6 nm, n was 73) were formed, and now the major globular species that remained had heights of ∼2 ( 0.4 nm (n was 55). In addition, some finer filaments (height ∼2 ( 0.3 nm, n was

Biomacromolecules, Vol. 5, No. 6, 2004 2411

Figure 2. Time-lapse AFM images of 4% w/w β-lactoglobulin aggregation at pH 2, induced by incubation at 80 °C for various times (10 min to 24 h). Samples were diluted 1 in 50 using deionized water, and applied onto mica by pipetting.

8) were also observed whose status was uncertain (i.e., independent species or protofibril components of the normal fibrils). These data reveal a highly cooperative process, and the only suggestion of an intermediate stage is the formation of rather diffuse aggregates/oligomers. It is not apparent from the data whether such aggregates nucleate fibrils, as has been suggested by others.42 Solvent-Induced Fibrils at pH 2. Numerous accounts of the unfolding/folding behavior of proteins in alcohol-water solutions, and in particular β-lactoglobulin, already exist in the literature.29,43-50 In the current work, such solutions were used as an alternative to thermal denaturation. At pH 2, clear gels, and solutions of aggregates, were formed in 50% alcohol-water mixtures. At this pH, the gelling propensity induced by the alcohols followed the pattern methanol > ethanol > propan-2-ol > TFE, both in terms of tgel and C0, the induction (gel) time and critical concentration for gelation, respectively. For instance, gels formed at concentrations of ∼4% w/w in 50% v/v water-methanol, whereas for water-TFE mixtures, the critical concentration for gelation was ∼10% w/w. These differences most likely relate to variations in the denaturing capacities of each solvent, these being known to differ.48 Electron micrographs from negative-staining experiments showed that well-defined fibrils formed in all four solvents (Figure 3A-D). In contrast to the thermally induced aggregates, these fibril types were more wormlike in appear-

2412

Biomacromolecules, Vol. 5, No. 6, 2004

Gosal et al.

Figure 5. AFM images of short rodlike aggregates induced by incubating 4% w/w β-lactoglobulin pH 7 at 80 °C for 24 h. Samples were diluted 1 in 50 in deionized water and applied onto mica by pipetting. Figure 3. Negative-staining EM of β-lactoglobulin fibrils formed in 50% v/v water-solvent mixtures: (A) methanol, (B) ethanol, (C) propan-2-ol, and (D) TFE, at pH 2, on incubating for 90 days.

Figure 4. AFM images of fibrils induced by incubation of 3.5% w/v β-lactoglobulin at pH 2 in 50% v/v TFE-water for 40 days at ambient. Samples were diluted (1 in 100-200) using deionized water, or using 50% v/v TFE-water pH 2 and applied onto mica by spraying under nitrogen gas.

ance, and shorter on average, being only ∼100 nm to 500 nm in length. Their widths were typically ∼7.2 ( 0.8 nm (n was 34). The wormlike fibrillar structure is particularly clear for the TFE/water member of the group (Figure 3D), and this formed the subject of a complementary AFM study. For this system, the AFM investigation confirmed the presence of wormlike fibrils (Figure 4) their average height being ∼ 2.7 ( 0.5 nm (n was 148). Evidently, the measured height of the fibrils is again in disagreement with the width determined from EM (∼7 nm). No suggestion of lateral aggregation of smaller species was detected, although numerous globular species were found that had heights of ∼1.2 ( 0.3 nm (n was 67). Some of the higher magnification AFM images of the TFE-induced fibrils clearly suggest a “beaded” structure (Figure 4) and this morphology is reminiscent of Aβ preliminary aggregates (or protofibrils)23 and β2-microglobulin fibrils formed in vitro at low pH and high ionic strength.21 The beaded appearance of the fibrils in Figure 4 suggests a periodic variation in height along the fibril length. In fact, the peak-to-peak distance of this fluctuation was found to be ∼34.3 ( 7.4 nm (n was 24), and the peak-totrough height was ∼0.9 ( 0.2 nm (n was 7). Temperature-Induced Aggregates at pH 7. β-Lactoglobulin is also prone to thermally induced aggregation at neutral pH. A recent EM study has indicated that, at pH 7,

short “wormlike” rods of width ∼3-5 nm (i.e., close to a single monomer in width) are formed28 in agreement with an earlier dynamic light-scattering analysis.51 Other microscopy data has suggested, however, that spherical particles, of diameter 35-70 nm, are created, although these structures are probably made up of rodlike particles, also with widths ∼3-5 nm.52 In agreement with this latter picture, neutron scattering results at high ionic strength suggest the formation of spherical structures ∼20 nm in diameter.53 In the present work, an AFM analysis of aggregates formed during incubation of a 4% w/w β-lactoglobulin, pH 7, solution at 80 °C, for 24 h, was carried out. The results are presented in Figure 5. In agreement with one of the previous EM investigations,28 short rodlike particles are indeed observed. However, these entities have a typical height ∼1.8 ( 0.3 nm (n was 79), and this again differs from previous EM estimates, where widths of ∼3-5 nm were found.28,52 Solvent-Induced Fibrils at pH 7. Solvent effects were also studied at pH 7 using the same solvents as at pH 2. At pH 7, the tendency to gel was found to be highest in propan2-ol, followed by ethanol > methanol ∼ TFE. This is in contrast (except for TFE) to the pattern in pH 2. Under negative-staining EM, the aggregates formed in 50% v/v TFE-water mixtures were revealed to be fibrillar (Figure 6D), much as they were at pH 2. Again, in contrast to the temperature-induced fibrils at pH 2, the TFE-induced fibrils at pH 7 were much more wormlike, and granular, in appearance. They were typically 150-500 nm in length, although much longer fibrils were also observed. These fibrils seemed to represent a single narrowly dispersed population, in terms of their width, the average being calculated to be 7.1 ( 1.6 nm (n was 51). Under metal shadowing (not shown), the TFE-water fibrils appeared to be smaller in length than was suggested by the negatively stained micrographs, although their wormlike appearance was retained. In contrast to these results for TFE-water mixtures, electron microscopy also revealed that, at pH 7, the other alcohols produced aggregates that were much less obviously fibrillar (Figure 6A-C). Evidently, under pH 7 conditions, the mechanism leading to aggregate formation in TFE-water solvents is different from that for the other alcohols. The origin of this difference was not investigated further, but it is likely to relate to protein unfolding. In this context, the exact pH of the final solution (see Materials and Methods)

Fibril Formation and Structure


Figure 6. Negative-staining EM of aggregates induced by incubation of 4% w/w β-lactoglobulin at pH 7 in 50% v/v 10 mM phosphate buffer pH 7, 50% v/v alcohol mixtures, for 3 days. Results are for (A) methanol, (B) ethanol, (C) propan-2-ol, and (D) TFE. Samples were diluted ∼1 in 400 in deionized water and sprayed onto the support grid under a stream of nitrogen gas.

Figure 7. AFM images of fibrils induced by incubation of 2.5% w/v β-lactoglobulin at pH 7 in 50% v/v 10 mM phosphate buffer pH 7, 50% v/v TFE, for 2 days at ambient. Samples were diluted 1 in 50 using deionized water and applied onto mica by spraying under nitrogen gas.

could well be important, since β-lactoglobulin is known to undergo alkaline denaturation at pH ∼ 8.5,44 as well as another conformational change involving a loop structure at neutral pH.54 In addition, variation in protein charge can modify the self-assembly process.55 Corresponding AFM work was restricted to studies of the 50% v/v TFE-water mixtures at pH 7 and confirmed the presence of wormlike fibrils (Figure 7). These fibrils had an average height of ∼2.3 ( 0.4 nm (n was 90) and displayed a similar granular appearance and axial periodicity to those formed at pH 2. In addition, globular species were also evident, with an average height of ∼1.3 ( 0.3 nm (n was 10). Fibrils, under these conditions, seemed to form more rapidly (at a comparative concentration) than at pH 2, a finding that correlates well with the propensity of TFEwater mixtures to form a gel at pH 7. “Ex situ time-lapse” AFM microscopy was also attempted for the TFE-water system (Figure 8). Here, samples were analyzed at various time intervals (1 min to 24 h) of incubation of 4% w/v β-lactoglobulin, pH 7, in 50% v/v water-TFE. After 1 min, a granular appearance suggested that the mica was saturated with β-lactoglobulin monomers, of height ∼ 1.2 nm, although some wider, and taller, spherical aggregates (presumably oligomeric) were also

Figure 8. Time-lapse AFM images of 4% w/v β-lactoglobulin aggregation at pH 7, induced by incubation in 50% v/v 10 mM phosphate buffer pH 7, 50% v/v TFE for various times (1 min to 24 h). Samples were diluted 1 in 100 using deionized water and applied onto mica by pipetting.

present that had heights of between 1.8 and 4 nm. After 10 min, these oligomeric species seemed to increase in number. Short rodlike species that had an average height of ∼2.7 ( 0.5 nm (n was 38) then formed after 20-40 min (i.e., identical to the mature fibrils in height, within experimental error). At this point, a significant amount of spherical species, of height ∼1 nm, remained. After 3 h, wormlike fibrils formed, with an average height of 3.0 ( 0.5 nm (n was 47), presumably due to elongation of the rods. After 24 h (at which point a “pre-gellike state” occurred), the wormlike fibrils further elongated (average heights ∼3.5 ( 0.5 nm, n was 91). Between 3 and 24 h, a significant amount of globular species remained, with an average height of 1.3 ( 0.3 nm (n was 60). From these data, it seems that, for the TFE/water system at pH 7, the aggregation process is less cooperative than for the thermally produced aggregates at pH 2 (Figure 2). This seems to imply an absence of nucleated aggregation and might be consistent with the random linear polymerization of critically sized oligomers that has recently been suggested for other wormlike aggregates.56 A much more detailed kinetic study would be needed to substantiate this, however. Insulin Temperature-Induced Fibrils at pH 2. Throughout the current study, it was noticed that width analysis from EM, did not agree with height analysis of AFM data. To

2414


Figure 9. AFM images of fibrils induced by incubating 4% w/w insulin at pH 2 and 80 °C for 24 h. Samples were diluted (1 in 100-200) using deionized water and applied onto mica by pipetting.

address this in further detail, and to compare the behavior of β-lactoglobulin with that of another fibril-forming protein, AFM data was collected for the thermally induced aggregation of the small protein hormone insulin at pH 2. Accompanying electron microscope work was omitted, as a considerable body of EM data is already available in the literature for this system. In the present work, 4% w/w insulin pH 2 solutions were heated and formed gray pasty fragile “gels” after 24 h at 80 °C, and AFM images of the fibrillar structures underlying these materials are presented in Figure 9. These data highlight a considerable polydispersity in fibril thickness. Thus, although a smooth, and seemingly flexible, “protofibril” is found, with an average height of ∼3.6 ( 0.5 nm (n was 55), this occurs together with other thicker fibrils. Common among these is a regular fibril having an average height of ∼5.6 ( 0.8 nm (n was 68), a left-handed twist, a regular peak-to-peak axial periodicity of ∼53 ( 8 nm (n was 36) and a peak-to-trough height of ∼1.2 ( 0.4 nm (n was 45). Other less common species are also present, including one with an average height of ∼9.7 ( 2 nm (n was 21), a peak-to-peak periodicity of ∼71 ( 9 nm (n was 11), and a peak-to-trough height of 2.9 ( 0.8 nm (n was 16). Although these morphologies determined for insulin by AFM are in agreement with previous EM analyses,15,57,58 the actual heights of the fibrils differ substantially. For example, from EM, the protofibril is considered to have a width of ∼5-10 nm, whereas “mature” twisted fibrils have a width ∼15-20 nm and an axial periodicity of 60-100 nm. This confirms that, as for β-lactoglobulin, the present AFM analysis derives dimensions for fibrils that differ by a factor ∼2 from previous EM work. Factors that influence the results and resolution achievable with AFM largely depend on the nature of the sample and the mode of AFM imaging used. These factors are usually the size of the tip-sample contact area, the magnitude of the substrate-sample interaction, factors influencing the contrast mechanism such as differential hydration of the sample and the substrate, the magnitude and effect of lateral and normal forces, water and salt contamination layers, and the optimization of scan-speeds and associated noise. Under intermittent-contact mode, height discrepancies between AFM data and predicted dimensions of biomolecules (i.e., from cryo-EM, X-ray crystal diffraction etc.) are commonly reported for some samples. A well-documented case is that of double-stranded DNA. Here, heights of ∼0.6-

Gosal et al.

0.8 nm are usually reported, whereas the actual height of these molecules is nearer ∼2.4 nm. Some of this discrepancy is due to compression by high normal forces, as can be demonstrated by the increased heights of these molecules when normal forces are apparently reduced (for example ∼1.6-1.7 nm using “transverse dynamic force” AFM 59 and “jumping-mode” AFM 60). Moreover, imaging under either the “attractive” or “repulsive” regimes in intermittent contact mode, can lead to a significant decrease in the measured height of DNA (from ∼1.2 to ∼0.4 nm).61 The presence of salt contamination layers, which aid the absorption of DNA to the substrate, under ambient conditions can also decrease the apparent height of DNA molecules.62 Poor samplesubstrate interaction under fluid can also lead to poor resolution and decreased heights, as is the case for pairedhelical filaments.60 Given the diversity of biological samples, there appears to be no unique AFM imaging mode and protocol that can produce optimum results, for any given sample. Optimization of AFM results presented here for amyloid fibrils is clearly required, and requires further work and an examination of the various factors highlighted above. 2. Raman and FTIR Spectroscopy. Raman Spectroscopy of Native β-Lactoglobulin. Raman spectroscopy was undertaken in an attempt to follow secondary structure changes upon self-assembly, and to investigate whether morphological differences between the thermally induced and solvent-induced fibrils at pH 2, would be reflected in the analyses. We analyzed the Amide I, III and optical skeletal vibrational modes, which are particularly sensitive to secondary structure. The spectrum of powdered native β-lactoglobulin is shown in Figure 10 and compared with corresponding spectra for the various aggregated samples at pH 2. Here, the major amide I peak at 1662 cm-1 agrees well with previous Raman studies of β-lactoglobulin39,63 and is made up of a number of components. This is consistent with the levels of native β-sheet, disordered, and R-helical, secondary structure suggested by X-ray crystallography.64 In the amide III region, two peaks at 1238 and 1246 cm-1 are found. The 1238 cm-1 peak is universally assigned to β-sheet,65 and it is probable that the sharper peak at 1246 cm-1 can also be assigned to β-sheet,66 this being superimposed upon an underlying broad contribution from random coil. Another much broader peak at ∼1253 cm-1 can be assigned exclusively to random coil, given that spectral analysis of alkaline-denatured β-lactoglobulin shows a peak shift from 1242 to 1257 cm-1.63 We are currently unable to assign any peaks exclusively to R-helical structure in the amide III region, although helical poly-L-alanine does show helical signatures in this region up to ∼1295 cm-1.67 Peaks corresponding to the so-called “optical skeletal modes” (∼901, 935, 953, 986, and 991 cm-1) also appear in the spectrum and have been identified with the presence of specific secondary structure components (e.g., R-helix is usually associated with a band around 940 cm-1). In summary, the current spectrum of native β-lactoglobulin is similar to those obtained in previous studies (including those from solutions) and are consistent with the latest X-ray crystallographic structure for the protein.64


Figure 10. Raman spectra (950-1050, 1225-1325, and 1625-1725 cm-1) of (a) lyophilised (native) β-lactoglobulin; powdered β-lactoglobulin pH 2 aggregates incubated at 80 °C for 14 h at a concentration of (b) 4% and (c) 8% w/w; powdered aggregates from solutions of (d) 2% w/v (e) 3.5 w/v (f) 7% w/v in 50% v/v methanol; (g) 2% w/v (h) 3.5% w/v (i) 7% w/v in 50% v/v ethanol; (j) 3.5% w/v (k) 7% w/v in 50% v/v propan-2-ol; (l) 3.5% w/v (m) 7% w/v in 50% v/v TFE. The 2% and 3.5% w/v alcohol-induced samples were incubated for 40 days, and 7% w/w alcohol-induced samples were incubated for 85 days.

Raman Spectroscopy of pH 2 Fibrillar Aggregates. The Raman spectra for β-lactoglobulin (pH 2) fibrils, in powdered form, and induced under various conditions also appear in Figure 10. Here the Raman intensities and frequencies were all normalized at 1002 cm-1, which is a conformationally insensitive band originating from a phenylalanine “ringbreathing mode”.65 From these data, changes in the amide I and III and the “optical skeletal” vibrational modes should be sensitive to secondary structure changes accompanying aggregation. The first observation, on detailed examination of the spectra, is that the band assigned to the Amide I vibrational


mode for the native protein shifts significantly in the spectra of the fibrils, from ∼1662 toward ∼1670 cm-1, to an extent that is sample dependent. In particular, this shift seems to correlate with the degree of aggregation which has occurred, since, for example, it was minimal in 50% v/v TFE (only to 1664 cm-1), which failed to gel at a concentration of 7% w/w but much higher in the 50% v/v methanol sample (to 1670 cm-1), which exhibited the most extensive propensity to gel. Furthermore, a 3.5% w/w sol in 50% v/v propan-2ol had a lower band shift than the corresponding 7% w/w gel. We interpret these shifts as indicating a change in the β-sheet present: either a significant change in the existing native sheet structure, a net increase in β-sheet, or perhaps both. The two major peaks at 1238 and 1245 cm-1 in the amide III region of the native protein Raman spectrum persist in the fibril spectra and confirm the presence of substantial amounts of β-sheet structure in the aggregates. It is difficult to interpret these as providing evidence of an altering sheet content in relation to the native protein, however, as the changes observed in this region relative to the native spectrum are slight. There may be some small differences in the peak frequencies involved, but proof of this would require a band deconvolution exercise not attempted in the present work. The slight shoulder at ∼1254 cm-1 could be diagnostic of randomly coiled structure also being present in the fibrils, and various peaks in the region between 1260 and 1295 cm-1 and the persistence of a peak at 1317 cm-1 suggest the existence of R-helical and β-turn structures as well. There were very few obvious changes in the optical skeletal region suggesting that within the accuracy of measurement there was little indication of significant changes in helix and coil content during any of the aggregation events studied. Changes in R-helix content have, in the past, been followed by observation of Raman spectral bands at ∼940 cm-1, for a number of largely R-helical proteins including insulin,27 bovine serum albumin,68 lysozyme,69 and R-lactalbumin.39 In the present work, the band assigned to R-helix (938 cm-1), as well as another peak in this region, at 988 cm-1, which probably relates to random coil content, both seem to be largely unaffected by the changes taking place. In summary, as for some other globular, largely β-sheeted proteins, loss of native secondary structure is apparently only partial under the conditions specified. FTIR Spectroscopy of pH 2 Fibrillar Aggregates. Corresponding FTIR spectra were obtained for the powders studied by Raman spectroscopy, attention focusing only on the amide I band. FTIR spectra for both native and aggregated β-lactoglobulin have previously been reported.28,33,70,71 The FTIR amide I bands of the native and aggregated β-lactoglobulin samples, obtained here without resolution enhancement, are shown in Figure 11. As expected from previous FTIR work on native β-lactoglobulin, the amide I peak (1632 cm-1) is largely dominated by the strong β-sheet band at ∼1628 cm-1, although some weak signal in the region 1640-1652 cm-1 also points to the presence of R-helix and disordered structures. The FTIR spectra of both the temperature and alcoholinduced fibrils show a definite shift in the amide I peak center

2416


Figure 11. FTIR spectra (1550 cm-1 to 1750 cm-1, amides I and II) of (a) lyophilised (native) β-lactoglobulin; powdered β-lactoglobulin pH 2 aggregates incubated at 80 °C for 14 h at a concentration of (b) 4% and (c) 10% w/w; powered aggregates from solutions of (d) 2% w/v (e) 3.5 w/v (f) 7% w/v in 50% v/v methanol; (g) 2% w/v (h) 3.5% w/v (i) 7% w/v in 50% v/v ethanol; (j) 3.5% w/v (k) 7% w/v in 50% v/v propan-2-ol; (l) 3.5% w/v (m) 7% w/v in 50% v/v TFE. The 2% and 3.5% w/v alcohol-induced samples were incubated for 40 days, and 7% w/w alcohol-induced samples were incubated for 85 days.

to a lower frequency, compared to that for the native protein. This shift is clear in each case, its greatest value being ∼15 cm-1 (to 1617 cm-1) for 7% w/w aggregates formed in both 50% v/v methanol and ethanol. The resulting amide I shifts also demonstrate that, for a given solvent, a correlation exists between the amide I peak shift and concentration. Hence, these data suggest that, just as for the Raman amide I peak, the FTIR amide I peak shift is an indication of the secondary structural changes that occur on aggregation. The pattern of the shifts found here is essentially the same as found in the Raman study, and again the indication is that aggregation is accompanied by changes in β-sheet secondary structure (new intermolecular β-sheet). It is worth noting that for most helical proteins, particularly insulin, previous studies27 by both Raman and infrared spectroscopy have revealed similar,

Gosal et al.

Figure 12. Difference FTIR spectra (each spectrum is intensitynormalized and the spectrum for the native subtracted from it) for (A) powders prepared from β-lactoglobulin pH 2 solutions heated at 80 °C for 24 h at a concentration of 4% w/w (broken line) and 10% w/w (solid line), (B) powders prepared from solutions of 3.5% w/w β-lactoglobulin pH 2 incubated for 40 days in 50% w/w water-alcohol mixtures (the symbols M, E, T, and P denote the alcohols used), and (C) powders prepared from solutions of 7% w/w β-lactoglobulin pH 2 incubated for 85 days in 50% w/w water-alcohol mixtures (symbols as before).

but much more prominent, changes toward the formation of new intermolecular β-sheet. Finally, the data are also suggestive of an increased, though comparatively small, amount of helical structure for the solvent-induced systems. To investigate this further, the difference spectra (after intensity normalization and subtraction of the native spectrum) are presented both for the thermally induced fibrils (Figure 12A) and for the solventinduced fibrils at protein concentrations of 3.5% w/w (Figure 12B) and 7% w/w (Figure 12C). These data clearly demonstrate that, although changes in β-sheet predominate in all systems and correlate well with the propensity for gelformation, some de novo helical structure is also present in solvent-induced fibrils. It is worth noting that previous work



Figure 13. Wide-angle powder wide-angle X-ray diffraction data (solid lines represent FFT filter smoothing) for (A) native (lyophilised) β-lactoglobulin; 7% w/v β-lactoglobulin pH 2 incubated for 85 days in (B) 50% v/v water-TFE, and (C) 50% v/v water-ethanol.

Figure 14. Wide-angle powder X-ray diffraction data (solid lines represent FFT filter smoothing) for powders prepared from β-lactoglobulin pH 2 solutions heated at 80 °C for 24 h at a concentration of (A) 2.5% w/w, (B) 5% w/w, and (C) 10% w/w.

has suggested that the partially unfolded helical conformation of β-lactoglobulin (known as the “H-state”) is fully reversible upon removal of the solvent,50 and hence, this result was unexpected for the powdered samples. The lack of a clear isodichroic point in these data complicates analysis, since this precludes a simple two-state event that could be used to explain the data. We thus can only speculate that either this de novo helix formation arises through participation of such a structure in the fibril aggregates induced in water-solvent mixtures, or that a monomeric “misfolded” state exists in the powders (i.e., helical induction by alcohol solutions is not totally reversible). It would be interesting to discriminate between these two possibilities and, in the case of the former explanation, see whether this de novo helical structure is an indication of an altered ‘building block’, which may account for the morphological differences between thermally induced and solvent-induced fibrils. De novo helical formation in fibril structure, if this is the case here, has not been previously observed for other systems. 3. X-ray Diffraction. Wide-Angle Powder X-ray Diffraction. X-ray diffraction patterns obtained from native β-lactoglobulin in lyophilised powder form (Figure 13) reveal two diffuse halos occurring at the angles 2θ ∼19 ( 1° and ∼8.6 ( 0.5° (i.e., Bragg spacings of ∼0.48 ( 0.02 nm and ∼ 1 ( 0.1 nm, respectively) and these are attributable to the amorphous character of the sample. Such a noncrystalline diffraction pattern is typical for native globular protein powders, the diffuse peaks indicating frequently occurring

distances between atoms72 such as occur in specific types of secondary structure (e.g., limited regions of β-sheet). Despite some variations in overall intensity, the diffraction patterns for the alcohol-induced (Figure 13) and the temperature-induced (Figure 14) pH 2 fibril aggregate powders are all very similar in shape to those obtained for native β-lactoglobulin and provide little indication of the emergence of a narrower and more intense diffraction maximum such as can be demonstrated for dried acid insulin gels (at 2θ ∼ 19°). In the present study, although fibrils are formed in the various β-lactoglobulin systems, which bear a superficial resemblance to their acid insulin counterparts, they are evidently much less ordered internally (less long-ranging regularly repeating intermolecular β-sheet present). This is consistent with the limited changes in secondary structure, on forming fibrils, indicated for β-lactoglobulin by the current Raman and FTIR work. Conclusion The results from electron microscopy have demonstrated that the temperature-induced fibrils formed on prolonged heating at pH 2 are more extended and smoother in appearance than most of the others and much greater in length. In comparison, fibrils formed in water-alcohol mixtures at pH 2, and TFE-water solutions at pH 7, have a more wormlike and flexible appearance, and are somewhat shorter. Width measurements from electron microscopy

2418


suggest that most of these fibrils are ∼8 ( 2 nm in diameter, regardless of the conditions used to induce them. Only the rods formed by heating at pH 7 seem to be thinner. In general, for the various β-lactoglobulin aggregates studied, single populations of fibril sizes are found, the widths of which follow normal distributions. The AFM experiments largely confirmed the various morphologies suggested by EM, there being a suggestion of organized internal repeating structure even for the smooth stiffened thermally induced pH 2 aggregates. However, a significant inconsistency was found between size estimates based on the two types of recorded image and requires further investigation and optimization of the AFM technique as applied to the current samples (as discussed above). Nevertheless, the present results are in keeping with two other reports that have recently been published22,73 on the application of AFM to β-lactoglobulin fibrils. Interestingly, both groups concluded that the heat-set fibrils had a repeating structure, either some form of helix (or pair of intertwined helices) involving protein subunits22 or a “string of beads” morphology.73 The latter appearance was particularly evident in images obtained for fibrils made at higher ionic strength than considered here, the result being similar to what was observed in the current work for the flexible fibrils formed in TFE/water. It may be possible that screening of electrostatic effects can induce a transition from a stiffened rodlike, to a more flexible wormlike, morphology (i.e., these two fibrillar types may not be intrinsically different); or morelikely that at higher ionic strength the assembly mechanism is altered, as has been noted with other systems.42 The absence of a clear left-handed twist for β-lactoglobulin thermally induced fibrils, as is seen here for the insulin system, suggests that these fibrils are unlikely to have formed through the helical winding of protofibrils. Some ex situ time-lapse AFM experiments were also conducted. These data showed that oligomeric species were formed prior to fibril formation. For temperature-induced β-lactoglobulin fibril formation at pH 2, for example, a polydisperse population of oligomers initially formed, ranging from ∼2-8 nm in height. The mechanism of transformation of these to long fibrils is still obscure, but it seems to be highly cooperative: stiffened extended fibrils appearing quite suddenly after a substantial time period. For TFEinduced fibril formation at pH 7, on the other hand, while oligomers ∼2-4 nm (i.e., similar in height to the normal fibrils) were again initially produced, these seemed to randomly associate (or grow progressively) with time, producing increasingly longer fibrils, with a final “beaded” appearance. This is in keeping with recent light scattering data on the formation of similar wormlike fibrils by PGK protein. In this latter work two stages are suggested, one of which involves the cooperative building of oligomers of a certain size, and the second stage, the assembly of such oligomers into fibrils.56 The fact that the overall scattering intensity scales linearly with concentration is taken to indicate a nonnucleated process for fibril growth. Thus, the formation of the two fibril types seen in the present study seem to be contrasting processes, with different levels of cooperativity, but it should be added that recent

Gosal et al.

evidence22 for the heat-set pH 2 fibrils cautions against too facile an interpretation. This work suggests that, on heating at pH 2, initial aggregation is reversible on cooling and that the long aggregates persist only when they undergo an eventual transition to a more stable structural form. This warns against too simple an interpretation of the time-lapse data for the two systems studied here and emphasizes how primitive current understanding is of the detailed mechanisms by which proteins form linear aggregates. Raman and FTIR amide I bands have been shown to be sensitive to β-lactoglobulin aggregation for all of the fibril types studied at acid pH and suggest either a net increase in β-sheet content through formation of new intermolecular secondary structure or conversion of some of the existing native sheet content to this intermolecular form. In the absence of a quantitative analysis of the amide I band profiles (not attempted in the present work), it is impossible to distinguish these possibilities convincingly, but it does seem that any overall increase in sheet content is small. This is in contrast to insulin where the change to new β-sheet conformation dominates spectra.27 In terms of the other forms of secondary structure involved, the presence of weak Raman bands at 938, 1265-1295, and 1656 cm-1, suggested the persistence of R-helical structure in the aggregates at something like the same level as in the native protein. However, there was some evidence from the FTIR amide I band that some, though small, amount of de novo helix formed in the alcohol-induced fibrillar samples. The fact that this occurred in dry samples, in the absence of solvent, was particularly surprising, and warrants further investigation. Overall, the picture provided was one of retention of significant amounts of helix and coil secondary structure in the aggregates. The X-ray results for all fibrils at pH 2 are also consistent with a limited change in native secondary structure during aggregation, and exclude the existence of extended regularly repeating “crystalline” regions of β-sheet along the fibril length as has been found for insulin. Overall, the present work suggests a picture of “protofibril” formation from β-lactoglobulin under a whole range of conditions, in which only partially unfolded, and still globular, particles selfassemble into organized linear structures of only limited internal order at the peptide level. Acknowledgment. The authors thank Dr. John Pacey (King’s College London) and Mr. Anthony Weaver (Unilever R&D Colworth) for assistance with electron microscopy, Dr. Neil Thomson (University of Leeds) for fruitful discussions, Dr. Paul Pudney and Mr. Dale G. Cunningham (Unilever R&D Colworth) for experimental FTIR/Raman data, and Mr. Martin Vickers and Dr. Jeremy Karl Crockcroft (Birbeck College, London) for access to, and help with, the X-ray diffractometer. W.S.G. thanks the BBSRC and Unilever Research for the award of a CASE studentship. References and Notes (1) Gosal, W. S.; Ross-Murphy, S. B. Curr. Opin. Colloid Interface Sci. 2000, 5, 188-194. (2) Ferry, J. D. AdV. Protein Chem. 1948, 4, 1-78. (3) Barbu, E.; Joly, M. Faraday Discuss. Chem. Soc. 1953, 13, 77-93.

Fibril Formation and Structure (4) Tombs, M. P. Faraday Discuss. Chem. Soc. 1974, 158-164. (5) Clark, A. H. In Functional Properties of Food Macromolecules, 2nd ed.; Hill, S. E., Ledward, D. A., Mitchell, J. R., Eds.; Aspen Publishers: Gaithersburg, MD, 1998; pp 77-142. (6) Clark, A. H. Curr. Opin. Colloid Interface Sci. 1996, 1, 712-717. (7) Sipe, J. D. Annu. ReV. Biochem. 1992, 61, 947-975. (8) Selkoe, D. J. Nature 2003, 426, 900-904. (9) Kelly, J. W.; Balch, W. E. J. Cell Biol. 2003, 161, 461-462. (10) Cohen, A. S.; Calkins, E. Nature 1959, 183, 1202-1203. (11) Cohen, A. S.; Shirahama, T.; Skinner, M. In Electron Microscopy of Proteins; Harris, J. R., Ed.; Academic Press: New York, 1982; pp165-205. (12) Burke, M. J.; Rougvie, M. A. Biochemistry 1972, 11, 2435-2439. (13) Beaven, G. H.; Gratzer, W. B.; Davies, H. G. Eur. J. Biochem. 1969, 11, 37-42. (14) Davidson, B.; Fasman, G. D. Biochemistry 1967, 6, 1616-1629. (15) Clark, A. H.; Judge, F. J.; Richards, J. B.; Stubbs, J. M.; Suggett, A. Int. J. Peptide Protein Res. 1981, 17, 380-392. (16) Ding, T. T.; Harper, J. D. Methods Enzymol. 1999, 309, 510-525. (17) Hatters, D. M.; MacRaild, C. A.; Daniels, R.; Gosal, W. S.; Thomson, N. H.; Jones, J. A.; Davis, J. J.; MacPhee, C. E.; Dobson, C. M.; Howlett, G. J. Biophys. J. 2003, 85, 3979-3990. (18) Goldsbury, C.; Kistler, J.; Aebi, U.; Arvinte, T.; Cooper, G. J. S. J. Mol. Biol. 1999, 285, 33-39. (19) Blackley, H. K.; Sanders, G. H.; Davies, M. C.; Roberts, C. J.; Tendler, S. J.; Wilkinson, M. J. J. Mol. Biol. 2000, 298, 833-840. (20) Kowalewski, T.; Holtzman, D. M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 3688-3693. (21) Kad, N. M.; Thomson, N. H.; Smith, D. P.; Smith, D. A.; Radford, S. E. J. Mol. Biol. 2001, 313, 559-571. (22) Arnaudov, L. N.; de Vries, R.; Ippel, H.; van Mierlo, C. P. M. Biomacromolecules 2003, 4, 1614-1622. (23) Harper, J. D.; Wong, S. S.; Lieber, C. M.; Lansbury, P. T. Biochemistry 1999, 38, 8972-8980. (24) Astbury, W. T.; Dickinson, S.; Bailey, K. Biochem. J. 1935, 19, 2354-2365. (25) Eanes, E. D.; Glenner, G. G. J. Histochem. Cytochem. 1968, 16, 673677. (26) Sunde, M.; Serpell, L. C.; Bartlam, M.; Fraser, P. E.; Pepys, M. B.; Blake, C. C. F. J. Mol. Biol. 1997, 273, 729-739. (27) Clark, A. H.; Saunderson, D. H. P.; Suggett, A. Int. J. Peptide Protein Res. 1981, 17, 353-364. (28) Kavanagh, G. M.; Clark, A. H.; Ross-Murphy, S. B. Int. J. Biol. Macromol. 2000, 28, 41-50. (29) Renard, D.; Lefebvre, J.; Robert, P.; Llamas, G.; Dufour, E. Int. J. Biol. Macromol. 1999, 26, 35-44. (30) Oberg, K.; Chrunyk, B. A.; Wetzel, R.; Fink, A. L. Biochemistry 1994, 33, 2628-2634. (31) Seshadri, S.; Khurana, R.; Fink, A. L. Methods Enzymol. 1999, 309, 559-576. (32) Fink, A. L. Folding Des. 1998, 3, R9-R23. (33) Renard, D.; Robert, P.; Garnier, C.; Dufour, E.; Lefebvre, J. J. Biotechnol. 2000, 79, 231-244. (34) Bauer, H. H.; Muller, M.; Goette, J.; Merkle, H. P.; Fringeli, U. P. Biochemistry 1994, 33, 12276-12282. (35) Fraser, P. E.; Nguyen, J. T.; Surewicz, W. K.; Kirschner, D. A. Biophys. J. 1991, 60, 1190-1201. (36) Conway, K. A.; Harper, J. D.; Lansbury, P. T. Biochemistry 2000, 39, 2552-2563. (37) Booth, D. R.; Sunde, M.; Bellotti, V.; Robinson, C. V.; Hutchinson, W. L.; Fraser, P. E.; Hawkins, P. N.; Dobson, C. M.; Radford, S. E.; Blake C. C. F.; Pepys, M. B. Nature 1997, 385, 787-793. (38) Caughey, B. W.; Dong, A.; Bhat, K. S.; Ernst, D.; Hayes, S. F.; Caughey, W. S. Biochemistry 1991, 30, 7672-7680.

Biomacromolecules, Vol. 5, No. 6, 2004 2419 (39) Nonaka, M.; Lichan, E.; Nakai, S. J. Agric. Food Chem. 1993, 41, 1176-1181. (40) Gosal, W. S.; Clark, A. H.; Pudney, P. D. A.; Ross-Murphy, S. B. Langmuir 2002, 18, 7174-7181. (41) Conway, K. A.; Harper, J. D.; Lansbury, P. T. Nat. Med. 1998, 4, 1318-1320. (42) Kad, N. M.; Myers, S. L.; Smith, D. P.; Smith, D. A.; Radford, S. E.; Thomson, N. H. J. Mol. Biol. 2003, 330, 785-797. (43) Uversky, V. N.; Narizhneva, N. V.; Kirschstein, S. O.; Winter, S.; Lober, G. Folding Des. 1997, 2, 163-172. (44) Townend, R.; Kumosinski, T. F.; Timasheff, S. N. J. Biol. Chem. 1967, 242, 4538-4545. (45) Tanford, C.; De, P. K.; Taggart, V. G. J. Am. Chem. Soc. 1960, 82, 6028-6034. (46) Inoue, H.; Timasheff, S. N. J. Am. Chem. Soc. 1968, 90, 18901898. (47) Hirota, N.; Mizuno, K.; Goto, Y. Protein Sci. 1997, 6, 416-421. (48) Hirota-Nakaoka, N.; Goto, Y. Bioorg. Med. Chem. 1999, 7, 67-73. (49) Dufour, E.; Haertle, T. Int. J. Biol. Macromol. 1993, 15, 293-297. (50) Dufour, E.; Bertrandharb, C.; Haertle, T. Biopolymers 1993, 33, 589598. (51) Griffin, W. G.; Griffin, M. C. A.; Martin, S. R.; Price, J. J. Chem. Soc., Faraday Trans. 1 1993, 89, 3395-3406. (52) Carrotta, R.; Bauer, R.; Waninge, R.; Rischel, C. Protein Sci. 2001, 10, 1312-1318. (53) Aymard, P.; Gimel, J. C.; Nicolai, T.; Durand, D. J. Chim. Phys. Phys.-Chim. Biol. 1996, 93, 987-997. (54) Qin, B. Y.; Bewley, M. C.; Creamer, L. K.; Baker, H. M.; Baker, E. N.; Jameson, G. B. Biochemistry 1998, 37, 14014-14023. (55) Lashuel, H. A.; LaBrenz, S. R.; Woo, L.; Serpell, L. C.; Kelly, J. W. J. Am. Chem. Soc. 2000, 122, 5262-5277. (56) Modler, A. J.; Gast, K.; Lutsch, G.; Damaschun, G. J. Mol. Biol. 2003, 325, 135-148. (57) Clark, A. H.; Kavanagh, G. M.; Ross-Murphy, S. B. Food Hydrocolloids 2001, 15, 383-400. (58) Nielsen, L.; Frokjaer, S.; Carpenter, J. F.; Brange, J. J. Pharm. Sci. 2001, 90, 29-37. (59) Antognozzi, M.; Szczelkun, M. D.; Round, A. N.; Miles, M. J. Single Mol. 2002, 3, 105-110. (60) Moreno-Herrero, F.; Colchero, J.; Gomez-Herrero, J.; Baro, A. M. Phys. ReV. E 2004, 69, 031915. (61) Round, A. N.; Miles, M. J. Nanotechnology 2004, 15, S176-S183. (62) Moreno-Herrero, F.; Colchero, J.; Baro, A. M. Ultramicroscopy 2003, 96, 167-174. (63) Frushour, B. G.; Koenig, J. L. Biopolymers 1975, 14, 649-662. (64) Brownlow, S.; Cabral, J. H. M.; Cooper, R.; Flower, D. R.; Yewdall, S. J.; Polikarpov, I.; North, A. C. T.; Sawyer, L. Structure 1997, 5, 481-495. (65) Peticolas, W. L. Methods Enzymol. 1995, 246, 389-416. (66) Benaki, D. C.; Aggeli, A.; Chryssikos, G. D.; Yiannopoulos, Y. D.; Kamitsos, E. I.; Brumley, E.; Case, S. T.; Boden, N.; Hamodrakas, S. J. Int. J. Biol. Macromol. 1998, 23, 49-59. (67) Chen, M. C.; Lord, R. C. J. Am. Chem. Soc. 1974, 96, 4750-4752. (68) Lin, V. J.; Koenig, J. L. Biopolymers 1976, 15, 203-218. (69) Lichan, E.; Nakai, S. J. Agric. Food Chem. 1991, 39, 1238-1245. (70) Casal, H. L.; Kohler, U.; Mantsch, H. H. Biochim. Biophys. Acta 1988, 957, 11-20. (71) Lefevre, T.; Subirade, M. Food Hydrocolloids 2001, 15, 365-376. (72) Hukins, D. W. L. X-ray Diffraction by Ordered and Disordered Systems; Pergamon Press Inc.: Oxford, U.K., 1981. (73) Ikeda, S.; Morris, V. J. Biomacromolecules 2002, 3, 382-389.

BM049659D

Fibrillar Î²-Lactoglobulin Gels: Part 1. Fibril Formation and Structure

Recommend Documents