Biomacromolecules 2005, 6, 255-261
255
Changes in Protein Secondary Structure during Gluten Deformation Studied by Dynamic Fourier Transform Infrared Spectroscopy Nikolaus Wellner,*,† E. N. Clare Mills,† Geoff Brownsey,† Reginald H. Wilson,† Neil Brown,‡ Jacqueline Freeman,‡ Nigel G. Halford,‡ Peter R. Shewry,‡ and Peter S. Belton§ Institute of Food Research, Norwich Research Park, Colney, Norwich NR4 7UA, United Kingdom, Rothamsted Research, Harpenden, Hertfordshire AL5 2JQ, United Kingdom, and School of Chemical Sciences and Pharmacy, University of East Anglia, Norwich NR4 7TJ, United Kingdom Received July 20, 2004; Revised Manuscript Received October 4, 2004
Fourier transform infrared (FT-IR) spectroscopy was used to monitor changes in the secondary structure of wheat prolamins, the main components of gluten, during mechanical deformation in a series of cycles of extension and relaxation. A sample derived from protein bodies isolated from developing grain showed a buildup of persistent β-sheet structure. In gluten, the ratio of β-sheet to random and β-turn structures changed on extension. After the applied force was released, the sample recovered some of its original shape and structure, but the material became stiffer in consecutive extension cycles. The relationship between gluten structure and mechanical properties is discussed in terms of a model in which conversion of β-turn to β-sheet structure is a response to extension and a means by which elastic energy is stored in the system. Introduction The prolamin storage proteins of wheat are characterized by their solubility in aqueous alcohols and high contents of proline and glutamine residues. They can be isolated as a cohesive mass, known as gluten, by washing wheat dough under stirring. The gluten fraction confers cohesive and viscoelastic properties to dough and is largely responsible for the ability to process wheat into a range of food products including bread, pasta, and noodles.1 Gluten comprises over 50 individual protein components that can be classified into two broad groups: monomeric gliadins and polymeric glutenins. One subgroup of the glutenin proteins, called the high molecular weight (HMW) subunits of glutenin, appears to be particularly important in determining gluten and dough viscoelasticity.2 These proteins comprise about 650-820 residues and contain an extensive repetitive domain (480-680 residues), which in solution appears to form a loose spiral structure based on β-reverse turns (a β-spiral). This domain is flanked by shorter nonrepetitive domains that are rich in R-helix.1 HMW subunits account for about 8-10% of the gluten protein with most of the remainder being low molecular weight (LMW) subunits and R-, β-, and γ-gliadins. These proteins are also rich in proline and glutamine and comprise a short unique N-terminal sequence followed by repetitive domain and nonrepetitive C-terminal domains of about equal length.2 It is thought that the glutenin polymers are stabilized by interchain disulfide bonds formed between cysteine residues * Corresponding author: Tel 0044 (0)1603 255012; fax 0044 (0)1603 507723; e-mail
[email protected]. † Institute of Food Research. ‡ Rothamsted Research. § University of East Anglia.
located predominantly in the nonrepetitive domains of the LMW and HMW subunits and that the individual polymers interact with other glutenin polymers and gliadin proteins3 by noncovalent interactions. Several models have been proposed to account for the viscoelastic behavior of gluten. As a consequence of its insolubility in water, it was originally thought that the elasticity of glutenin resembled that of the hydrophobic animal protein elastin.4 However, NMR experiments have shown different (i.e., hydrophilic) hydration behavior5 to elastin. Alternatively, the β-spiral structure of the repetitive domains in prolamins may be inherently elastic, like the spring it resembles,6 and atomic force microscopy (AFM) of individual HMW subunits of gluten has demonstrated their intrinsic elasticity at the molecular level7 (T. McMaster, A. Humphries, M. Miles, R. Tatham, and P. Shewry, unpublished results). Since prolamins can associate to form fibrils and sheets7 and Fourier transform infrared (FT-IR) studies have clearly shown the presence of intermolecular β-sheet structures in gluten,8,9 a model involving noncovalent interactions of molecules by hydrogen bonds (the “loop and train” model10) has been proposed. This model postulates that the glutenin repetitive domains exist either as hydrated chains that do not interact with other chains (these are the loop regions) or as regions where two or more molecules interact by the formation of intermolecular β-sheets (the train regions). On extension of the network, the loops form an extended structure that can interact with other chains to form a type of “interchain” β-sheet structure, with further extension disrupting the sheet regions. The force restoring the system to its original dimensions after extension therefore arises from the free energy recovered by restoration of the original
10.1021/bm049584d CCC: $30.25 © 2005 American Chemical Society Published on Web 11/23/2004
256
Biomacromolecules, Vol. 6, No. 1, 2005
equilibrium balance of loops and trains,11 and the creation of β-sheet structure is a means of storing elastic energy in the system. Gluten structure is therefore a dynamic, not a static, entity. Mechanical work input changes the balance of structures of the protein network, and if the rate of mechanical work input is faster than the rate of the reestablishment of molecular equilibrium, the mechanical properties will be expected to change. It is well-known that the mechanical properties of dough change with the input of mechanical work,12 and changes also occur in the size of the gluten polymers isolated from the worked dough.13 These polymers have very high molecular weight3 and it has been suggested that they form interacting particles,14 but details of their structure on the molecular scale await elucidation. Despite extensive rheological studies of dough and gluten and attempts to relate polymer size to dough rheology, experiments that provide information relevant to the molecular structures under deformation are scarce. Stretched dough exhibits birefringence due to molecular orientation,15 while NMR studies of sheared gluten16 demonstrated a change in the mobility of glutamine sidechain groups. FT-IR spectroscopy has also indicated that the secondary structure varies with protein aggregation depending on external factors such as hydration or solvent,9,11,17 and recent results have suggested differences in protein structure in undeveloped, developed, and stretched dough.18 However, these differences have not previously been related to extensibility. In this paper we report the use of dynamic infrared spectroscopy to determine secondary structure changes during gluten protein deformation. Small-scale extensions of the kind often employed in infrared spectroscopy allied to mechanical testing are of little value, as it has been shown that the small deformation behavior of dough or gluten did not show good correlations with baking quality.19 We have therefore analyzed gluten proteins during large-scale biaxial extension in order to mimic the kinds of extension encountered in dough mixing and in commercial dough test methods. Materials and Methods Samples: (A) Gluten. Gluten samples were obtained by washing Bu¨hler-milled wheat flour (cv. Hereward), provided by Dr. J. M. Field (Advanta Seeds). Flour (300 g) and H2O (186 mL) were mixed in a commercial breadmaker (Morphy Richards, model 48220) for 30 min. The dough was washed with distilled water until the water ran clear of starch. Fresh gluten was inhomogeneous on a small scale, and more importantly, separating small lumps induced different prestretching histories. To eliminate these, the gluten was frozen in water at -20 °C and freeze-dried. Then the dry gluten mass was ground to a powder in a mortar. Powder samples (10 mg) were immersed in distilled water without stirring and then left to equilibrate at 4 °C for at least 24 h. Gluten is insoluble in water but can absorb up to 1.5 g of water/g of protein. The rehydrated freeze-dried gluten behaved similarly to fresh gluten but was more reproducible in smallscale experiments.
Wellner et al.
(B) Protein from Protein Bodies. Protein bodies were isolated from developing wheat endosperms essentially as described by Davy et al.20 Wheat kernels from 10 ears of bread wheat cv. Cadenza at about 21 days after flowering (i.e., middevelopment stage) were dehusked and the pericarp was removed to leave only the endosperm (approximately 20 g total weight). The endosperms were chopped with a razor blade in 20 mL of buffer A [20 mM N-(2-hydroxyethyl)piperazine-N′-2-ethanesulfonic acid (HEPES), pH 7.6, 100 mM NaOAc, and 5 mM MgCl2] to allow the protein bodies to float out. The sample was filtered through four layers of prewetted muslin, carefully layered onto 2 × 5 mL 1.75 M sucrose cushions (made up in buffer A) in two Corex tubes, and spun in a centrifuge (500g, 2 min, 10 °C). The material from the top of the cushion (2 mL volume) was collected and resuspended in buffer B [20 mM HEPES, pH 7.6, 100 mM NaOAc, 5 mM ethylenediaminetetraacetic acid (EDTA), and 0.25 mM sucrose] to give a total volume of 10 mL. Two Percoll step gradients (2 mL of 1.13 g cm-3 and then 2 mL of 1.08 g cm-3, in 0.25 M sucrose) were prepared in 15 mL Corex tubes. Half of the suspension was layered onto each of the Percoll step gradients and then centrifuged (9000 rpm, 60 min, 10 °C in a JA25.50 rotor). The protein bodies were collected from the surface of the 1.13 g cm-3 Percoll layer (2 × 500 µL). The protein bodies were washed twice: 500 µL of water was added, the sample was centrifuged for 2 min at 13 000 rpm, and then the supernatant was removed. Protein bodies from three preparations were pooled and the pellet was stirred gently into 300 µL of chloroform/methanol, 2:1, to remove membranes. After centrifugation (13000g, 1 min) and removal of the chloroform/ methanol, the cohesive, elastic protein mass was stored at -20 °C. Rheo-FT-IR Measurements. Spectra were recorded on a Bio-Rad FTS175 FT-IR spectrometer with a mercurycadmium-telluride detector (Bio-Rad) and a single-reflection diamond attenuated total reflection (ATR) accessory (Graseby SPECAC, Orpington, U.K.). First the background spectrum of the ATR cell with water was recorded (32 scans at 8 cm-1 resolution). Then the hydrated gluten samples (in excess water) were placed on the ATR crystal and covered with a piece of Teflon foil and a microscope slide. This both prevented evaporation and allowed application of pressure to the sample. Biaxial extension of the sample was achieved either by placement of a fixed weight on top of the microscope slide or with a TA-XT2 texture analyzer (Stable Microsystems) attached to the spectrometer, which allowed better control. The texture analyzer was programmed to apply cycles of 2 N force (5 min, followed by 3 min intervals without force), recording applied force and sample thickness with a data rate of 10 s-1. Infrared spectra were recorded continuously with a time resolution of 0.5 s-1 and a spectral resolution of 8 cm-1 (averaging 10 scans/spectrum). The deformation was repeated several times with each sample, and the whole experiment was repeated three times with fresh samples. The experimental setup was limited by the fact that the sample was submerged in water and only in good contact with the ATR crystal while force was applied. During the
Protein Structure Changes during Gluten Deformation
Figure 1. Characterization of protein bodies prepared from developing wheat grain. (A) Light micrograph of protein bodies prepared from developing wheat endosperms (21 days after flowering). (B) Cohesive, elastic protein mass obtained by treating a protein body preparation with chloroform/methanol. (C) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of proteins from a protein body preparation from developing wheat endosperms (left) and gluten prepared from wheat flour (right).
relaxation phase the sample tended to peel away from the crystal surface while it contracted, leaving a thin layer of water between the crystal and the protein sample. Therefore, data points from the intervals where no force was applied were discarded. To determine the secondary structure changes in the protein, the infrared spectra were exported to Matlab (MathWorks Inc.). The spectra were baseline-corrected (absorbance at 1800 cm-1 set to zero) and normalized to the same height of the amide II band. The relative intensities at 1666, 1650, 1630, and 1620 cm-1 were calculated as the fraction of their sum. The significance of these values is discussed in the Results section. Structure contents obtained in this way were only estimates because of band overlaps but were highly reproducible (less than 0.1% standard deviation), and therefore small changes in structure could be reliably detected. Results The isolation of gluten by washing of dough involves considerable inputs of energy in mixing the dough to “develop” the gluten network and in the subsequent washing to remove starch and other components. It is therefore inevitable that the interactions of the individual glutenin polymers and gliadin monomers will have been considerably modified during these processes. To determine the properties of the same proteins without such work input, we decided to analyze proteins in the state in which they are deposited in the developing grain. We consequently isolated protein bodies from developing grain by a procedure that involved chopping to allow the bodies to float out of the ruptured cells followed by two centrifugation steps. The resulting fraction comprised small membrane-bound structures with diameters up to about 10 µm (Figure 1A), and electrophoretic analysis demonstrated that the protein composition was essentially identical to that of gluten isolated from flour of the same cultivar (Figure 1C). The isolated protein bodies did not form a cohesive mass, but removal of membranes by gentle washing with chloroform/methanol gave a cohesive mass that could be manipulated mechanically (Figure 1B). This mass was not subjected to significant work input during
Biomacromolecules, Vol. 6, No. 1, 2005 257
Figure 2. Strain curves for five cycles of deformation. (Left axis) Strain, calculated as deformation relative to the contact height of the sample in each cycle. (s) Protein body protein (cv. Cadenza); (- - -) gluten (cv. Hereward). (Right axis) Applied force (‚‚‚).
preparation and hence could correspond to the “unwashed” proteins present in gluten. However, it should also be noted that the protein body preparation had not undergone the dehydration which occurs during the latter stage of grain maturation and that this dehydration could also affect protein interactions. Measurements were carried out on “balls” of protein about 2-3 mm in diameter (10-15 mg) or on 10 mg portions of freeze-dried gluten that had been allowed to hydrate overnight. Initially the samples were only loosely cohesive. When force was applied, the sample was pressed into a flat disk about 0.2-0.3 mm thick that behaved like a viscoelastic material. Figure 2 shows strain curves for gluten and protein body protein during repeated deformation. Each time force was applied, the strain showed an initial steep increase, followed by a slow steady creep. These curves could be interpreted as an initial rapid elastic deformation of the gluten disk, followed by retarded elastic deformation, which finally turned into a steady slow deformation due to material flow under pressure. After the force was removed, the material relaxed but only partially recovered its original shape and hence the protein sample was thinner at the next deformation cycle. The measured strain also became smaller in consecutive deformation cycles. The overall extent of the plastic and elastic deformation was largest in the first cycle for both samples. The deformation of the protein body protein was larger than that observed with gluten in the first cycle, but became similar in later cycles. While a direct comparison is not possible because of the different origins of the samples, it would be reasonable to conclude that the protein body sample was more easily deformed because it had not previously been worked. Because of the small sizes of the samples and their illdefined dimensions, no attempts were made to calculate exact rheological parameters. However, the deformation experiments with the texture analyzer demonstrated that the mechanical properties of the samples did change during compression and extension. The work input (force × displacement per gram of dry material) was estimated to be of the same order of magnitude as that used in commercial dough mixing processes (40 J/g of dough in the Chorleywood breadmaking process). The observations are consistent with dough mixing data that show that the material properties of gluten change during extension.13 This process corresponds
258
Biomacromolecules, Vol. 6, No. 1, 2005
Figure 3. FT-IR spectra of (A) gluten and (B) protein body protein at end of cycles 1 (s), 2 (- - -), 3 (‚‚‚), 4 (- ‚ -), and 5 (- ‚‚ -). Water spectra were subtracted; the spectra were baseline-corrected and normalized.
to the storage of mechanical energy in the dough and in turn implies that the relaxation rate after the perturbation of the system by extension is slow compared to the rate at which the cycling process is carried out. Thus, it would be expected that a cycling regime used during the FT-IR experiments would result in cumulative behavior that mirrored that observed in the mechanical tests. The amide bands of the infrared spectrum reflect the secondary structure content of proteins.21,22 In the course of the deformation experiments, small but reproducible changes were observed in the amide bands of fully hydrated gluten, which indicated that the secondary structure changed during deformation. Band assignments used here are based on those given elsewhere and are considered appropriate for gluten proteins.17,21-23 Figure 3 shows the spectra obtained from samples that were subjected to five cycles of extension by application of a 2 N force. Toward the end of each cycle, the spectra did not vary with time and corresponded to the establishment of a steady state under pressure. In the gluten sample (Figure 3A) the amide I band shape changed only little between consecutive cycles, showing only a small decrease in the main peak of the amide I band at 1650 cm-1 and a slight increase in the amide II band shoulder around 1520 cm-1. In contrast, more marked changes were observed in the infrared spectra of the protein body samples (Figure 3B). The broad shoulder around 1620 cm-1 increased considerably over the first two cycles. The region between 1625 and 1600 cm-1 has been attributed to intermolecular β-sheet in prolamins.8,17 In contrast, the main peak of the amide I band
Wellner et al.
became less prominent and difference spectra (not shown) revealed the decrease of a component at around 1660 cm-1. Bands in the region of 1650-1660 cm-1 may be attributed to R-helix,21 while β-turns have a band in the region of 1660-1670 cm-1.23 The differences observed during the initial cycles of compression of the gluten and protein body protein were essentially similar to those observed between gluten proteins dissolved in 1% acetic acid and in their native hydrated solid state and to the changes observed on drying.8,17 In these cases the β-turn component at 1666 cm-1 was reduced, indicating a smaller proportion of β-spiral structure, and a more pronounced shoulder around 1618 cm-1 was observed, showing an increase in intermolecular β-sheet. This comparison with other studies of prolamins leads us to suggest that the decrease in the amide I maximum is due mainly to the loss of β-turn structures. The amide band shifts indicated that deformation of both samples caused an increase in β-sheet structures at the expense of β-turn and R-helical structures, and in both samples the change occurred mainly during the first two cycles. However, whereas the changes in the previously worked gluten samples were quite small, those in the protein body preparations were much greater, indicating that work input had a dramatic effect on the secondary structure. The changes in secondary structure during biaxial extension of gluten were analyzed by plotting the relative amounts of the main secondary structure elements as a function of time over several deformation cycles. Band fits to determine the absolute amounts of secondary structure content21 were not useful because strongly overlapping bands caused large variability in the results. Instead, the relative intensities at 1666 (β-turns), 1650 (R-helix and random), 1630 (β-sheet), and 1620 cm-1 (intermolecular β-sheet) were used to estimate secondary structure. The data from this analysis were comparable to previous structure estimates from HMW subunits, other prolamins, and prolamin-related peptides and were sufficiently reproducible to monitor small structure changes.24 Figure 4 shows the changes in the secondary structure of gluten during five cycles of deformation. When the sample was deformed with a 2 N force for the first time, there were apparent increases of ca. 0.5% in the levels of intermolecular and intramolecular β-sheet and a corresponding decrease of about 1% in β-turn structures, with marginally smaller changes in the amounts of unordered and R-helical structures. The changes in structure occurred within the first few seconds from the onset of the force, after which the system settled quickly, and no further changes were observed in the spectra. However, the rheological measurements showed that the flow process continued for much longer than this. Therefore, if material flow was associated with changes in molecular structure, this was not being detected. The spectra observed at the end of each subsequent compression cycle indicated that the proportion of intermolecular β-sheet (1620 cm-1) increased, while the proportions of β-turn and unordered structures decreased. However, the proportion of intramolecular β-sheet (1630 cm-1), which probably arose from globular domains of the gluten proteins, remained relatively unchanged.
Protein Structure Changes during Gluten Deformation
Biomacromolecules, Vol. 6, No. 1, 2005 259
Figure 4. Analysis of secondary structure changes in gluten (cv. Hereward) during successive deformation cycles with 2 N force: (s) intermolecular β-sheet (1620 cm-1); (- - -) intramolecular β-sheet (1630 cm-1); (‚‚‚) helix/random (1650 cm-1); (- ‚ - ) β-turn (1666 cm-1).
Whereas in the first cycle structure changes only occurred with the onset of force, in later cycles the level of intermolecular β-sheet structures jumped to a maximum just after the onset of the force and then slowly decreased, while the content of unordered structures increased. This implies that a relaxation process was occurring under stress, i.e., a change in the secondary structures in opposite direction to the initial change on application of force. The spectra showed that some of the β-sheet structure initially induced by the deformation in the stretched sample reverted to random structures. Also, the time dependence of this structure change appeared to be exponential, with more than one component probably being involved. An exponential fit of the ratio of β-turn to β-sheet (Figure 5) indicated that the initial fast relaxation (
