- Email: [email protected]
X-ray structural study of noncrystalline regenerated Bombyx mori silk fibroin
International Journal of Biological Macromolecules 34 (2004) 259–265
X-ray structural study of noncrystalline regenerated Bombyx mori silk fibroin Hiroyuki Saitoha,∗ , Ken-ichi Ohshimaa , Kozo Tsubouchib , Yoko Takasub , Hiromi Yamadab a
Institute of Materials Science, University of Tsukuba, Tsukuba 305-8573, Japan b National Institute of Agrobiological Sciences, Tsukuba 305-8634, Japan Accepted 20 September 2004
Abstract X-ray diffraction measurements of regenerated Bombyx mori silk fibroin were carried out to determine its structural characteristic from an analysis of differential radial distribution functions (DRDFs). The temperature dependence of X-ray diffraction patterns from noncrystalline and crystal structures of regenerated silk fibroin was investigated using a high temperature furnace. Time resolved X-ray diffraction profiles were also obtained to construct kinematical models of structural changes caused by the addition of water. DRDFs, calculated from the experimental data, were compared with the DRDFs simulated on the basis of the Monte Carlo method. In order to model the noncrystalline structures, structural units were assumed to be parts of the crystalline structure of silk and those with appropriate structural defects reported previously. From the comparison of experimental and simulated DRDFs, it was determined that noncrystalline regenerated silk consisted of locally ordered atomic sheets similar to the atomic arrangement in the silk I crystal (Type-I sheets), and the final state of the structural change was noncrystalline, consisting of small crystallites, the structure of which is similar to that of silk II (Type-II crystallites). Time resolved DRDFs were also qualitatively interpreted by both the ordering of Type-I sheets and structural changes from Type-I to Type-II. The formation of the small Type-II crystallites obtained in this study was consistent with the nucleation of silk II by birefringence measurements of silk glands and the spinneret of Bombyx mori silkworm reported previously. X-ray diffraction should be a useful technique to understand the structural characteristics of noncrystalline organic materials. © 2004 Elsevier B.V. All rights reserved. Keywords: Silk fibroin; X-ray diffraction; Noncrystalline structure; Differential radial distribution function
1. Introduction Silkworm silk produced by Bombyx mori has been used as a source of textile-grade fibers for at least 2000 years, and extensive research related to it has also been performed. This silk usually consists of a fibrous protein, fibroin, and a coating protein, sericin. The latter is removed in most applications of silk fibers by a process referred to as degumming. The silk fibroin molecule of Bombyx mori primarily consists of highly ∗ Corresponding author. Present address: Synchrotron Radiation Research Center, Japan Atomic Energy Research Institute, Mikazuki, Sayo, Hyogo 678-5148, Japan. Tel.: +81 791 58 2632; fax: +81 791 58 2740. E-mail address: [email protected] (H. Saitoh).
0141-8130/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ijbiomac.2004.09.003
repetitive regions, which are represented by the repetitions of six amino acid residues, Gly-Ala-Gly-Ala-Gly-Ser. Bombyx mori silk fibroin is biosynthesized in the silk gland cells and is secreted into the cavity as a concentrated solution in water. The liquid silk fibroin is transformed into a solid and insoluble silk fiber during the spinning process. The silk fiber primarily consists of a crystal region, the structure of which is referred to as silk II. Precise crystal structure analyzes of silk II were performed [1,2] and its crystal structure was understood to be an extended -sheet conformation. On the other hand, the metastable crystal structure of silk I can be obtained from the silk gland contents by quiescent drying without mechanical or chemical stress. Several researchers have investigated the crystal structure of silk I [3–13]; how-
260
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
ever this remains unclear because of the difficulty involved in preparing an oriented silk I specimen that is suitable for structural analysis. The structure of silk I is easily transformed into stable silk II by mechanical or chemical stress. Using detailed NMR studies, Asakura et al. recently proposed that the silk I crystal is a repeated -turn type II-like structure [14,15]. The repeated -turn type II-like structure contains hydrogen bonding not only along the intra-chain axis but also along the inter-chain axis. This type of structure appears to differ from conformations consisting of coiled or helical chains with hydrogen bonding only along the chain axis. Immediately after biosynthesis in silk glands, the liquid silk fibroin consists of random coil and/or silk I conformations. It is believed that the liquid silk fibroin changes into stable silk II during the spinning process of silkworms. Though the existence of the unstable random coil or silk I structure seems to play an important role in the silkworm’s spinning process, this process has been insufficiently understood until now because of the difficulties involved in measurements. Clarifying the silkworm’s spinning process is extremely important, from not only a biological but also a macromolecular point of view. One of the difficulties involved in measuring the spinning process is the incomplete structure of liquid silk. The diffraction technique is the most effective method to clarify a structure, but it is thought to be inefficient for noncrystalline organic materials because of its weak and broad diffraction profiles. However, the noncrystalline structures of inorganic glass materials in particular, have been successfully understood by means of X-ray diffraction. In this paper, we have attempted to apply diffraction techniques to analyze noncrystalline organic macromolecules and report the analysis by X-ray diffraction methods of the noncrystalline structures of a regenerated silk fibroin film and its structural change during wetting.
2. Materials and methods 2.1. Preparation of regenerated silk fibroin films Fresh (not dried) silkworm cocoon shells were used as the starting material. They were degummed by immersing in an 8 M urea solution containing 2% 2-mercaptoethanol at 80 ◦ C for 10 min to obtain refined fibroin, which is the subject for examination in this study. The degummed fibers were washed with water and dried properly. The weight of the fibers decreased to 75% of the starting material; this decrease was consistent with the composition of sericin. Degummed silk fibers were then dissolved in a saturated lithium thiocyanate (LiSCN) aqueous solution at room temperature [16,17]. After a small quantity of insoluble material was removed by centrifuging at 8000 rpm for 10 min, the fibroin solution was dialyzed for 2 h against deionized water that was renewed every 30 min. The resultant solution was again centrifuged to remove the aggregates that occurred during dialysis. The silk fibroin solution was then cast onto
a polyethylene film and dried. The obtained regenerated silk fibroin film was about 60 m in thickness. 2.2. X-ray diffraction measurement X-ray diffraction profiles were collected with a conventional two-axis diffractometer (Philips, PW3050 X’Pert) in order to obtain differential radial distribution functions (DRDFs), which are representative of the noncrystalline structures of materials. It is noteworthy that this system enables rapid measurement due to a compact solid-state array detector (Philips, X’Celerator). Although the diffracted beams from noncrystalline specimens are too weak, time resolved X-ray diffraction profiles using this system are taken on a time scale that is as short as 2 min. Ni filtered Cu K␣ radiation from a line focus X-ray generator (Philips, PW3040) was used as an incident beam. High temperature X-ray diffraction measurements were carried out using a high temperature attachment (Paar, HTK16). The attachment is well designed and it enables carrying out these measurements with ease. Owing to the direct heating system and effective cooling flow, over 1000 ◦ C measurements and quick thermal response are available. However, very sharp and strong peaks from the polycrystalline platinum heater were observed on the broad and weak halo pattern from the regenerated silk film. Therefore, crystalline platinum peaks were subtracted from (1 1 1) to (2 0 0) fundamental reflections. High temperature experiments were performed to (i) understand the kinetics of structural change when water was added, (ii) compare the resultant structure after the addition of water with the high temperature phase without water. The resultant structure after adding water was expected to be in the stable state and similar to that produced at high temperature. The computational routine to calculate DRDF is built following the procedure described below. DRDFs are, therefore, calculated from the measured X-ray diffraction profiles. The heating speed was fixed at 60 ◦ C/min for each measurement. Simultaneous measurements of differential thermal analysis (DTA) and X-ray diffraction have become a standard practice; in this study, however, we have particularly focused on the structure determined by X-ray diffraction. Therefore, thermal analysis such as DTA has not been carried out. 2.3. The calculation and simulation of DRDF To calculate DRDF, the measured intensity from the noncrystalline regenerated silk fibroin film should be normalized. As proposed by Norman [18], the normalization factor was calculated by integrating all measured intensity values. Using normalized intensity and calculated background intensity, which is the sum of self-correlation and incoherent intensity, the structural function i(k) is given as Ieu (k)/N − uc fm2 (k) i(k) = , (1) fe2 (k)
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
where Ieu (k)/N is the normalized intensity for one structural unit, uc fm 2 (k) the sum of the squared atomic scattering factor, which denotes the self-correlation component for one structural unit. It is convenient to introduce the abbreviation i(k), which is an experimentally observed quantity. fe (k) is the average scattering factor per electron. fm fe = uc , (2) Z uc m where fm is the atomic scattering factor of the mth atom and Zm the number of electrons in the atom. By introducing the average scattering factor fe , the angle dependence of atomic scattering factor fm for each atom is replaced by the averaged value over atoms in the structural unit, approximately. Using the structural function i(k), the DRDFs for noncrystalline regenerated silk fibroin are calculated from the following equation [19]: 2r ∞ 4πr2 Km (ρ(r) − ρ0 ) = ki(k) sin kr dk, (3) π 0 m where Km = fm /fe is the effective electron number, which is approximately equal to the atomic number Zm , ρ(r) and ρ0 represent the radial distribution function and averaged electron density, respectively. Km is the integral average of the effective electron numbers, kmax . 1 Km = Km (k) dk, (4) kmax . − kmin . kmin . m in which kmin. and kmax. are the lower and upper bounds of the range of k values restricted by the experimental condition. The mean effective electron number Km is introduced to describe DRDF by a form of the Fourier transformation of the structure factor i(k). However, information regarding RDF for each atom is lost by introducing this reduced expression. Besides the extremely weak intensities observed from the noncrystalline protein containing no heavy elements, peak broadening of DRDF occurred due to the limited observed range of reciprocal space, and the window function was employed to avoid the Fourier termination effect. Understanding the noncrystalline structures directly from calculated DRDFs is difficult; therefore, DRDFs for regenerated silk fibroin were simulated with structural models and compared with the calculated DRDFs. Structural units, which were randomly arranged in the noncrystalline state, constituted chains, hydrogen bonded sheets, and small crystallites, the atomic positions of which are determined after referring to the silk I [14,15] and silk II [2] crystal structures reported previously. Hereafter, structural units with the same conformation as silk I and II crystals are referred to as Type-I and Type II, respectively. For example, a chain structural unit with a conformation similar to silk I is denoted as a Type-I chain. These structural units with appropriate defects, which indicate structural faults when compared to the ideal crystalline form were also considered, though the fundamental conformation of the molecules
261
remained unaltered. The total number of structural units and the volume required for the simulation were determined based on the experimentally observed density. Structural units were then randomly arranged in the volume space on the basis of the Monte Carlo method. When these units have a completely random arrangement, the simulated DRDF became flat when compared with the observed DRDFs. Therefore, a locally oriented arrangement was introduced whose directions were distributed by the Gaussian function. The orientation pro˚ However, due to the posed here has a range of about 50 A. lack of a macroscopically preferred orientation in the noncrystalline sample, the shape of the DRDF would be spherical (isotropic). Finally, the DRDF corresponding to the modeled noncrystalline structure was simulated from the following equation: 4πr 2 Km (ρ(r) − ρ0 ) m
= fe2
Ki Kj A exp(−B(ri − rj )2 ) − 4πr a ρ0 ,
(5)
V
where A and B are the normalization factor for the Gaussian function and an artificial temperature factor, respectively. The single term Gaussian function that represents the FWHM of the DRDF peaks is empirically utilized for modeling the peak broadening. If the volume space is too large and the modeled noncrystalline structure has a complete spherical symmetry, the dimensional factor a in Eq. (5) must be 2. In the present model, a locally preferred orientation was introduced and the volume space was limited, so that the dimensional factor a was determined to be 1.8–1.9 by least squares fitting.
3. Results 3.1. DRDF for regenerated silk fibroin at room temperature The X-ray diffraction profile of regenerated silk fibroin measured at room temperature is shown in Fig. 1. In this figure, a weak and wide halo pattern only appeared due to
Fig. 1. The measured X-ray diffraction profile obtained from regenerated silk fibroin film at room temperature.
262
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
Fig. 2. The comparison between simulated and experimentally observed DRDFs of as-cast regenerated silk fibroin. The solid line represents simulated DRDF and open circles denote the observed DRDF.
Fig. 3. Temperature dependence of measured X-ray diffraction profiles of regenerated silk fibroin.
the noncrystalline form of the structure. In order to calculate DRDF from the observed diffraction profile, the intensity at small angle regions (0◦ ≤ 2θ ≤ 10◦ ) was extrapolated to the origin at the diffraction angle using the lower angle data (7◦ ≤ 2θ ≤ 10◦ ). The calculated DRDF is shown in Fig. 2. ˚ The most distinct positive peak in the DRDF appears at 4.8 A (arrow (a)). Furthermore, a characteristic flat region (b) also appears, whose absolute value of DRDF is nearly equal to ˚ in the ranges from 6 to 8 A. ˚ Since DRDF is 100 electrons/A not a radial distribution of electron density but a reduced expression of the difference between the electron density distribution and the average density, this flat region does not imply ˚ in the that there is no electron distribution around 6 < r < 8 A specimen. Interpreting the values of DRDF in the range of ˚ is difficult due the average interatomic distance r below 3 A to the termination effect of the Fourier transformation.
and it is certain that these structural changes occur at lower temperatures. At 270 ◦ C, the sharp peak disappears. This implies that thermal decomposition occurred and the specimen is carbonized at that temperature. Using the measured halo pattern at 210 ◦ C after eliminating the sharp peak, the DRDF corresponding to the high temperature noncrystalline structure was calculated. The obtained DRDF is shown in Fig. 4. Compared with the DRDF at room temperature, shown in ˚ (a) is also observed. On Fig. 2, a large peak around r = 4.8 A the other hand, a characteristic flat region observed at room temperature is not observed. A small positive peak in the ˚ (b) and a distinct negative peak at 8 A, ˚ (c) are vicinity of 6 A, ◦ observed at 210 C.
3.2. Structural change of regenerated silk fibroin on heating
In order to clarify the change in conformation and the crystalline structure of regenerated silk fibroin during wetting, time resolved X-ray diffraction measurements were carried out at 30 ◦ C. After the addition of water, the diffraction profiles were taken at every 2 min from 18 to 38 min. Water was added using a micro pipette and the volume was equal to that of the specimen. The DRDFs were then calculated from ex-
High temperature X-ray diffraction measurements were carried out to determine the structural changes of regenerated silk fibroin. The measured temperature ranged from room temperature to 300 ◦ C, and diffraction profiles were collected at every rise of 10 ◦ C. Parts of the measured diffraction profiles are shown in Fig. 3. When the temperature reaches about 190 ◦ C, the first halo peak at 2θ = 20◦ decreases. On the other hand, a sharp and strong peak appears at the same angle in the vicinity of 200 ◦ C. This sharp peak originates from the ˚ and it can crystalline state of silk II. The d-value is 4.22 A be indexed by overlapping (2 1 0) and (1 2 0) Bragg reflection. However, the broad and weak halo patterns that originated from the noncrystalline state remained. This implies that both noncrystalline and silk II crystalline regions coexist in the regenerated silk fibroin film. Motta et al. [20] reported that the random coil to -sheet transformation was observed at 213 ◦ C by means of DSC, and the result appears to be consistent with that of our study. However, the structures discussed in our study are expected to be in the metastable state,
3.3. Structural change of regenerated silk fibroin caused by the addition of water
Fig. 4. DRDF for regenerated silk fibroin at 210 ◦ C.
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
263
25 ◦ C and 33 ◦ C, the DRDFs exhibited unclear peaks in the ˚ (a). Above 40 ◦ C, a small positive peak range from 5 to 7 A ˚ (b) appeared and a negative peak around around r = 6 A, ˚ (c) became distinct; this is similar to the change in r = 7 A, time resolved DRDFs in Fig. 5. It is suggested that at a lower temperature the structural change in regenerated silk fibroin is obstructed from a kinematical point of view, and that the resultant substance is formed by a noncrystalline structure, which is formed at an early stage of structural change during wetting at higher temperatures.
4. Discussion 4.1. Modeling and analysis Fig. 5. Time resolved DRDF calculated from experimentally observed X-ray diffraction profiles of regenerated silk fibroin during wetting at 30 ◦ C. The top DRDF corresponds to the profile measured 18 min after the addition of water. The diffraction profiles were taken at intervals of 2 min until 38 min.
perimental data, as shown in Fig. 5. The top DRDF mainly originates from liquid–water. As the amount of water decreased, some peaks, the DRDF peak positions of which did not overlap with those of the resultant structure decreased ˚ (a)). On the other hand, DRDF peaks with a (e.g. r = 3 A, slight change in absolute value and shape also appeared (e.g. ˚ (b)). Since the simultaneous decrease in the abr = 4.8 A, solute value of DRDF originated from liquid–water and an increase in this value of the resultant substance occurred, the DRDFs apparently do not change significantly. The most dis˚ (c). tinct change was observed in the ranges from r = 5 to 7 A In this region, a negative peak of DRDF from liquid–water was observed; with an increase in time; however, this negative peak became flat and the absolute value became close to 0. The structural change in regenerated silk fibroin during wetting was also measured at several different temperatures. DRDFs of the resultant substance calculated from experimentally observed X-ray diffraction profiles at various temperatures are shown in Fig. 6. When the temperature was
Fig. 6. Comparison of DRDFs for the noncrystalline structure of resultant substances calculated from experimentally observed X-ray diffraction profiles at various temperatures.
In order to reproduce experimentally observed DRDFs of regenerated silk fibroin with the addition of water at 90 ◦ C (Fig. 7) and regenerated silk fibroin at 210 ◦ C (Fig. 4), both of which were similar, a small Type-II crystallite is assumed to exist in the specimen. Simulated and experimentally observed DRDFs are shown in Fig. 7. Not only the shapes and positions of peaks but also the absolute value were in good agreement. Although the least squares fitting procedure was only utilized to determine the average electron density ρ0 and the procedure was not performed to minimize the differences in the two patterns, a good agreement is evident. Consequently, it was determined that the structure of the specimens consists of randomly arranged small Type-II crystallites. The noncrystalline structure of as-cast regenerated silk fibroin was rather difficult to model, because of the existence of a characteristic flat region of DRDF shown in Fig. 2. The shape of this DRDF could not be reproduced by considering only the conformations reported previously in the crystalline state of silk I and II. In order to reproduce the DRDF of as-cast regenerated silk fibroin, the Monte Carlo simulation method was employed as described above. Only one model was successful in reproducing such DRDFs after calculating several
Fig. 7. The comparison between simulated and experimentally observed DRDFs of as-cast regenerated silk fibroin with added water at 90 ◦ C. Simulated DRDF was calculated by introducing a small crystallite of silk II. The solid line represents simulated DRDF and open circles denote the observed DRDF.
264
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
modeled noncrystalline structures. The simulated DRDF is shown in Fig. 2. It was derived from a noncrystalline structure consisting of slightly oriented Type-I sheets. The distribution of the direction of Type-I sheets was assumed to be a Gaussian distribution as described above, and the standard deviation of the Gaussian function was 10◦ . The Type-I sheet, which consisted of randomly arranged structural units for constructing a noncrystalline structure, is an isolated hydrogen-bonded Type-I chain, as reported by Asakura et al. [14,15]. Due to the poor resolution of the experimentally observed DRDFs, it was unclear if the conformation was correct. Since the TypeI crystallite model could not reproduce the observed DRDF, such a structure with hydrogen-bonded sheets seemed to be more favorable. It is, however, difficult to confirm whether such isolated and/or oriented sheets can exist by means of Xray diffraction techniques. Further experiments and theoretical discussions are required for understanding the structure. Compared with the reproduction of DRDFs of noncrystalline regenerated silk fibroin with added water at a high temperature (Type-II crystallite model), the Type-I sheet model could not reproduce experimentally observed DRDFs in sufficient detail. It is caused by the effect of the limited number of atoms available for simulation and the possibility of the coexistence of another conformation such as random coils. However, the as-cast regenerated silk fibroin does not, contain the TypeII conformation. If the model structure contains the Type-II conformation, the calculated DRDF does not show any flat ˚ On the other hand, a strucarea in the range from 6 to 8 A. tural model containing the random coil conformation was not calculated, since it is difficult to model these atomic arrangements. The as-cast regenerated silk fibroin seems to contain some random coil conformation. 4.2. Interpretation of time resolved DRDFs The kinematical model is introduced to interpret observed time resolved DRDFs. An intermediate state of regenerated silk fibroin with added water was a mixture of the specimen and a small amount of water. It is expected that water, which is a polar molecule, have a great influence on the structural change. To construct the model, it is more favorable to consider correlations between molecules of water and regenerated silk fibroin. However, only the structural change in regenerated silk fibroin conformation and the noncrystalline structure are discussed for the present calculation. There are several structural characteristics in the observed time resolved DRDFs that originate from conformational changes and/or noncrystalline structural changes of regenerated silk fibroin and a decrease in water, as described above. Changes in DRDFs related to the noncrystalline structure of regenerated silk fibroin cause an enhancement of the positive ˚ and the formation of a small positive peak around r = 4.8 A ˚ The peak and a distinct negative peak in the range r = 5 to 7 A. former is not clearly seen in Fig. 5 because of a simultaneous decrease in the DRDF peak from water and an increase in the DRDF peak from the regenerated silk fibroin. Kinemat-
Fig. 8. The simulated DRDFs corresponding to (a) ordering of silk I sheets (n /nall = 0), (b) structural change of silk I sheets to silk II sheets (σ = 10◦ ).
ical models to reproduce such features related to regenerate silk were investigated. These models contain the following parameters: standard deviation (σ) of the Gaussian distribution of directions of structural units and an existence ratio of different structural units (n /nall ). Here, n and nall represent the total numbers of Type-II structural units and all structural units, respectively. These parameters were varied with increasing time. The DRDFs thus obtained are shown in Fig. 8. In Fig. 8(a), it is observed that simulated DRDFs can reproduce an increase in the characteristic positive peak around ˚ (indicated by an arrow). In this model, Type-I sheets r = 4.8 A were used as the structural unit of noncrystalline regenerated silk fibroin in the same way as that at room temperature. Structural change was modeled by an ordering of the directions of Type-I sheets expressed by a decrease in the deviation of the Gaussian function. As σ decreases, which corresponds to an ordering of the directions of Type-I sheets, the increase in ˚ is reproduced. In Fig. 8(b), the positive peak around r = 4.8 A ˚ were also formation of DRDF peaks in the range r = 5–8 A reproduced with changing n /nall . In this model, both Type-I and Type-II sheets were present as noncrystalline structural units, and these structural units were randomly arranged by following the Gaussian function, the standard deviation σ of which is equal to 10◦ . As n /nall increases, which corresponds to the conformational change from Type-I to Type-II sheets, ˚ disapthe flat region of DRDFs in the ranges from r = 5 to 8 A pears and a small positive and distinct negative peak appears. As described above, this model was able to reproduce the observed DRDFs. The reproduction depends on the difference between conformations of Type-I and Type-II chains. The Type-II chain forms a rather straight and simple chain conformation compared with the Type-I chain that has complicated conformations; therefore, DRDF peaks that originate from the regular distribution of atomic distances appear with the formation of Type-II conformations. On the other hand, due to the irregular distribution of atomic distances, DRDF peaks do not appear in spite of the existence of Type-I chains. As Type-I chains are ordered and regularly arranged, hydrogen bonded Type-I sheets are formed and DRDF exhibits ˚ characteristic peak around r = 4.8 A.
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
Taking these results into consideration, the structural change of noncrystalline regenerated silk fibroin during wetting can be interpreted as follows. When water is added to regenerated silk fibroin, the chain conformations become random coil and/or isolated Type-I. As time passes and the amount of water decreases, chains form hydrogen-bonded Type-I sheets and an ordering of the formed Type-I sheets occurs. Subsequently, conformational transformation from Type-I to Type-II sheets occurs. Finally, the ordered Type-II sheets form small crystallites of Type-II, and the transformed structure of regenerated silk film consists of small Type-II crystallites. A small Type-II crystallite forming noncrystalline structure has been proposed in this study. This kind of nucleation has already been previously predicted by means of birefringence measurements of silk glands and a spinneret of the Bombyx mori silkworm [21]. It has been reported that the nuclei of Type-II was too small so as to sufficiently observe the intensity that originated from it on the X-ray diffraction pattern. Since no crystal structure exists in these specimens, it is natural that nucleation of small Type-II crystallites could not be detected by means of X-ray diffraction without a precise analysis of the noncrystalline structure, as carried out in this study. Therefore, a noncrystalline structural model consisting of small Type-II crystallites is consistent with the noncrystalline model predicted previously. The kinetics of the spinning process of natural silk remains unclear. Moreover, in the natural silk fiber, a sizeable crystal region induced by mechanical stress and/or elongational flow of Type-II exists. In order to clarify the kinetics of the spinning process, it is necessary to elucidate the crystallization process. The present noncrystalline structural analysis will facilitate understanding such a kinematical study.
5. Conclusion In order to clarify the noncrystalline structure of regenerated Bombyx mori silk fibroin, an X-ray diffraction measurement was carried out. DRDFs describing poorly defined noncrystalline structures of materials were calculated from experimentally observed diffraction profiles. The temperature dependence of the noncrystalline and crystalline structure of regenerated silk fibroin was investigated using a high temperature X-ray furnace. Time resolved X-ray diffraction profiles were also obtained to construct a kinematical model of structural transformation during wetting. Calculated DRDFs from experimental data were compared with simulated DRDFs on the basis of the Monte Carlo method.
265
Noncrystalline structures contain some structural units that are randomly arranged in the specimen. These structural units were similar to the previously reported ones in the crystalline state and those with appropriate structural defects. From the comparison of experimental DRDF and simulated DRDFs, it was determined that the regenerated silk consisted of locally ordered Type-I sheets, and that added water is noncrystalline state formed by small Type-II crystallites. Time resolved DRDFs were also qualitatively interpreted by the ordering of Type-I sheets and structural changes of Type-I into Type-II. The formation of small Type-II crystallites determined in this study was consistent with the nucleation of Type-II crystallites by birefringence measurements of silk glands and a spinneret of Bombyx mori silkworm predicted previously. The analysis of the noncrystalline structure and structural change performed in this study would be effective in clarifying the silkworm’s spinning process. References [1] Marsh RE, Corey RB, Pauling L. Biochim Biophys Acta 1955;16:1. [2] Takahashi Y, Gehoh M, Yuzuriha K. Int J Biol Macromol 1999;24:127. [3] Konishi T, Kurokawa M. Sen’i Gakkaishi 1968;24:550 (in Japanese). [4] Lotz B, Keith HD. J Mol Biol 1971;61:201. [5] Lotz B, Cesari FD. Biochimie 1979;61:205. [6] Saito H, Tabeta R, Asakura T, Iwanaga Y, Shogi A, Ozaki T, Ando I. Macromolecules 1984;17:1405. [7] Asakura T, Kuzuhara A, Tabeta R, Saito H. Macromolecules 1984;18:1841. [8] Asakura T, Yamaguchi T. J Seric Sci Jpn 1987;56:300 (in Japanese). [9] Anderson JP. Biopolymers 1998;45:307. [10] Okuyama K, Takahashi K, Nakajima Y, Hasegawa Y, Hirabayashi K, Nishi N. J Seric Sci Jpn 1988;57:23 (in Japanese). [11] Fossy SA, Nimithy G, Gibson KD, Scheraga HA. Biopolymers 1991;31:1529. [12] Asakura T, Demura M, Date T, Miyashita M, Ogawa K, Williamson MP. Biopolymers 1997;41:193. [13] Asakura T, Iwatdate M, Demura M, Williamson MP. Int J Biol Macromol 1999;24:167. [14] Asakura T, Yamane T, Nakazawa Y, Ando K. Biopolymers 2001;58:521. [15] Asakura T, Ashida J, Yamane T, Kameda T, Nakazawa Y, Ohgo K, Komatsu K. J Mol Biol 2001;306:291. [16] Takasu Y, Yamada H, Tsubouchi K. Biosci Biotechnol Biochem 2002;66:2715. [17] Yamada H, Nakao H, Takasu Y, Tsubouchi K. Mat Sci Eng C 2001;14:41. [18] Norman N. Acta Cryst 1957;10:370. [19] Warren BE. X-Ray Diffraction. New York: Dover publication, Inc.; 1969. [20] Motta A, Fambri L, Migliaresi C. Maclomol Chem Phys 2002; 203:1658. [21] Kataoka K. Sen’i Gakkaishi 1978;34:20 (in Japanese).
X-ray structural study of noncrystalline regenerated Bombyx mori silk fibroin Hiroyuki Saitoha,∗ , Ken-ichi Ohshimaa , Kozo Tsubouchib , Yoko Takasub , Hiromi Yamadab a
Institute of Materials Science, University of Tsukuba, Tsukuba 305-8573, Japan b National Institute of Agrobiological Sciences, Tsukuba 305-8634, Japan Accepted 20 September 2004
Abstract X-ray diffraction measurements of regenerated Bombyx mori silk fibroin were carried out to determine its structural characteristic from an analysis of differential radial distribution functions (DRDFs). The temperature dependence of X-ray diffraction patterns from noncrystalline and crystal structures of regenerated silk fibroin was investigated using a high temperature furnace. Time resolved X-ray diffraction profiles were also obtained to construct kinematical models of structural changes caused by the addition of water. DRDFs, calculated from the experimental data, were compared with the DRDFs simulated on the basis of the Monte Carlo method. In order to model the noncrystalline structures, structural units were assumed to be parts of the crystalline structure of silk and those with appropriate structural defects reported previously. From the comparison of experimental and simulated DRDFs, it was determined that noncrystalline regenerated silk consisted of locally ordered atomic sheets similar to the atomic arrangement in the silk I crystal (Type-I sheets), and the final state of the structural change was noncrystalline, consisting of small crystallites, the structure of which is similar to that of silk II (Type-II crystallites). Time resolved DRDFs were also qualitatively interpreted by both the ordering of Type-I sheets and structural changes from Type-I to Type-II. The formation of the small Type-II crystallites obtained in this study was consistent with the nucleation of silk II by birefringence measurements of silk glands and the spinneret of Bombyx mori silkworm reported previously. X-ray diffraction should be a useful technique to understand the structural characteristics of noncrystalline organic materials. © 2004 Elsevier B.V. All rights reserved. Keywords: Silk fibroin; X-ray diffraction; Noncrystalline structure; Differential radial distribution function
1. Introduction Silkworm silk produced by Bombyx mori has been used as a source of textile-grade fibers for at least 2000 years, and extensive research related to it has also been performed. This silk usually consists of a fibrous protein, fibroin, and a coating protein, sericin. The latter is removed in most applications of silk fibers by a process referred to as degumming. The silk fibroin molecule of Bombyx mori primarily consists of highly ∗ Corresponding author. Present address: Synchrotron Radiation Research Center, Japan Atomic Energy Research Institute, Mikazuki, Sayo, Hyogo 678-5148, Japan. Tel.: +81 791 58 2632; fax: +81 791 58 2740. E-mail address: [email protected] (H. Saitoh).
0141-8130/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ijbiomac.2004.09.003
repetitive regions, which are represented by the repetitions of six amino acid residues, Gly-Ala-Gly-Ala-Gly-Ser. Bombyx mori silk fibroin is biosynthesized in the silk gland cells and is secreted into the cavity as a concentrated solution in water. The liquid silk fibroin is transformed into a solid and insoluble silk fiber during the spinning process. The silk fiber primarily consists of a crystal region, the structure of which is referred to as silk II. Precise crystal structure analyzes of silk II were performed [1,2] and its crystal structure was understood to be an extended -sheet conformation. On the other hand, the metastable crystal structure of silk I can be obtained from the silk gland contents by quiescent drying without mechanical or chemical stress. Several researchers have investigated the crystal structure of silk I [3–13]; how-
260
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
ever this remains unclear because of the difficulty involved in preparing an oriented silk I specimen that is suitable for structural analysis. The structure of silk I is easily transformed into stable silk II by mechanical or chemical stress. Using detailed NMR studies, Asakura et al. recently proposed that the silk I crystal is a repeated -turn type II-like structure [14,15]. The repeated -turn type II-like structure contains hydrogen bonding not only along the intra-chain axis but also along the inter-chain axis. This type of structure appears to differ from conformations consisting of coiled or helical chains with hydrogen bonding only along the chain axis. Immediately after biosynthesis in silk glands, the liquid silk fibroin consists of random coil and/or silk I conformations. It is believed that the liquid silk fibroin changes into stable silk II during the spinning process of silkworms. Though the existence of the unstable random coil or silk I structure seems to play an important role in the silkworm’s spinning process, this process has been insufficiently understood until now because of the difficulties involved in measurements. Clarifying the silkworm’s spinning process is extremely important, from not only a biological but also a macromolecular point of view. One of the difficulties involved in measuring the spinning process is the incomplete structure of liquid silk. The diffraction technique is the most effective method to clarify a structure, but it is thought to be inefficient for noncrystalline organic materials because of its weak and broad diffraction profiles. However, the noncrystalline structures of inorganic glass materials in particular, have been successfully understood by means of X-ray diffraction. In this paper, we have attempted to apply diffraction techniques to analyze noncrystalline organic macromolecules and report the analysis by X-ray diffraction methods of the noncrystalline structures of a regenerated silk fibroin film and its structural change during wetting.
2. Materials and methods 2.1. Preparation of regenerated silk fibroin films Fresh (not dried) silkworm cocoon shells were used as the starting material. They were degummed by immersing in an 8 M urea solution containing 2% 2-mercaptoethanol at 80 ◦ C for 10 min to obtain refined fibroin, which is the subject for examination in this study. The degummed fibers were washed with water and dried properly. The weight of the fibers decreased to 75% of the starting material; this decrease was consistent with the composition of sericin. Degummed silk fibers were then dissolved in a saturated lithium thiocyanate (LiSCN) aqueous solution at room temperature [16,17]. After a small quantity of insoluble material was removed by centrifuging at 8000 rpm for 10 min, the fibroin solution was dialyzed for 2 h against deionized water that was renewed every 30 min. The resultant solution was again centrifuged to remove the aggregates that occurred during dialysis. The silk fibroin solution was then cast onto
a polyethylene film and dried. The obtained regenerated silk fibroin film was about 60 m in thickness. 2.2. X-ray diffraction measurement X-ray diffraction profiles were collected with a conventional two-axis diffractometer (Philips, PW3050 X’Pert) in order to obtain differential radial distribution functions (DRDFs), which are representative of the noncrystalline structures of materials. It is noteworthy that this system enables rapid measurement due to a compact solid-state array detector (Philips, X’Celerator). Although the diffracted beams from noncrystalline specimens are too weak, time resolved X-ray diffraction profiles using this system are taken on a time scale that is as short as 2 min. Ni filtered Cu K␣ radiation from a line focus X-ray generator (Philips, PW3040) was used as an incident beam. High temperature X-ray diffraction measurements were carried out using a high temperature attachment (Paar, HTK16). The attachment is well designed and it enables carrying out these measurements with ease. Owing to the direct heating system and effective cooling flow, over 1000 ◦ C measurements and quick thermal response are available. However, very sharp and strong peaks from the polycrystalline platinum heater were observed on the broad and weak halo pattern from the regenerated silk film. Therefore, crystalline platinum peaks were subtracted from (1 1 1) to (2 0 0) fundamental reflections. High temperature experiments were performed to (i) understand the kinetics of structural change when water was added, (ii) compare the resultant structure after the addition of water with the high temperature phase without water. The resultant structure after adding water was expected to be in the stable state and similar to that produced at high temperature. The computational routine to calculate DRDF is built following the procedure described below. DRDFs are, therefore, calculated from the measured X-ray diffraction profiles. The heating speed was fixed at 60 ◦ C/min for each measurement. Simultaneous measurements of differential thermal analysis (DTA) and X-ray diffraction have become a standard practice; in this study, however, we have particularly focused on the structure determined by X-ray diffraction. Therefore, thermal analysis such as DTA has not been carried out. 2.3. The calculation and simulation of DRDF To calculate DRDF, the measured intensity from the noncrystalline regenerated silk fibroin film should be normalized. As proposed by Norman [18], the normalization factor was calculated by integrating all measured intensity values. Using normalized intensity and calculated background intensity, which is the sum of self-correlation and incoherent intensity, the structural function i(k) is given as Ieu (k)/N − uc fm2 (k) i(k) = , (1) fe2 (k)
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
where Ieu (k)/N is the normalized intensity for one structural unit, uc fm 2 (k) the sum of the squared atomic scattering factor, which denotes the self-correlation component for one structural unit. It is convenient to introduce the abbreviation i(k), which is an experimentally observed quantity. fe (k) is the average scattering factor per electron. fm fe = uc , (2) Z uc m where fm is the atomic scattering factor of the mth atom and Zm the number of electrons in the atom. By introducing the average scattering factor fe , the angle dependence of atomic scattering factor fm for each atom is replaced by the averaged value over atoms in the structural unit, approximately. Using the structural function i(k), the DRDFs for noncrystalline regenerated silk fibroin are calculated from the following equation [19]: 2r ∞ 4πr2 Km (ρ(r) − ρ0 ) = ki(k) sin kr dk, (3) π 0 m where Km = fm /fe is the effective electron number, which is approximately equal to the atomic number Zm , ρ(r) and ρ0 represent the radial distribution function and averaged electron density, respectively. Km is the integral average of the effective electron numbers, kmax . 1 Km = Km (k) dk, (4) kmax . − kmin . kmin . m in which kmin. and kmax. are the lower and upper bounds of the range of k values restricted by the experimental condition. The mean effective electron number Km is introduced to describe DRDF by a form of the Fourier transformation of the structure factor i(k). However, information regarding RDF for each atom is lost by introducing this reduced expression. Besides the extremely weak intensities observed from the noncrystalline protein containing no heavy elements, peak broadening of DRDF occurred due to the limited observed range of reciprocal space, and the window function was employed to avoid the Fourier termination effect. Understanding the noncrystalline structures directly from calculated DRDFs is difficult; therefore, DRDFs for regenerated silk fibroin were simulated with structural models and compared with the calculated DRDFs. Structural units, which were randomly arranged in the noncrystalline state, constituted chains, hydrogen bonded sheets, and small crystallites, the atomic positions of which are determined after referring to the silk I [14,15] and silk II [2] crystal structures reported previously. Hereafter, structural units with the same conformation as silk I and II crystals are referred to as Type-I and Type II, respectively. For example, a chain structural unit with a conformation similar to silk I is denoted as a Type-I chain. These structural units with appropriate defects, which indicate structural faults when compared to the ideal crystalline form were also considered, though the fundamental conformation of the molecules
261
remained unaltered. The total number of structural units and the volume required for the simulation were determined based on the experimentally observed density. Structural units were then randomly arranged in the volume space on the basis of the Monte Carlo method. When these units have a completely random arrangement, the simulated DRDF became flat when compared with the observed DRDFs. Therefore, a locally oriented arrangement was introduced whose directions were distributed by the Gaussian function. The orientation pro˚ However, due to the posed here has a range of about 50 A. lack of a macroscopically preferred orientation in the noncrystalline sample, the shape of the DRDF would be spherical (isotropic). Finally, the DRDF corresponding to the modeled noncrystalline structure was simulated from the following equation: 4πr 2 Km (ρ(r) − ρ0 ) m
= fe2
Ki Kj A exp(−B(ri − rj )2 ) − 4πr a ρ0 ,
(5)
V
where A and B are the normalization factor for the Gaussian function and an artificial temperature factor, respectively. The single term Gaussian function that represents the FWHM of the DRDF peaks is empirically utilized for modeling the peak broadening. If the volume space is too large and the modeled noncrystalline structure has a complete spherical symmetry, the dimensional factor a in Eq. (5) must be 2. In the present model, a locally preferred orientation was introduced and the volume space was limited, so that the dimensional factor a was determined to be 1.8–1.9 by least squares fitting.
3. Results 3.1. DRDF for regenerated silk fibroin at room temperature The X-ray diffraction profile of regenerated silk fibroin measured at room temperature is shown in Fig. 1. In this figure, a weak and wide halo pattern only appeared due to
Fig. 1. The measured X-ray diffraction profile obtained from regenerated silk fibroin film at room temperature.
262
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
Fig. 2. The comparison between simulated and experimentally observed DRDFs of as-cast regenerated silk fibroin. The solid line represents simulated DRDF and open circles denote the observed DRDF.
Fig. 3. Temperature dependence of measured X-ray diffraction profiles of regenerated silk fibroin.
the noncrystalline form of the structure. In order to calculate DRDF from the observed diffraction profile, the intensity at small angle regions (0◦ ≤ 2θ ≤ 10◦ ) was extrapolated to the origin at the diffraction angle using the lower angle data (7◦ ≤ 2θ ≤ 10◦ ). The calculated DRDF is shown in Fig. 2. ˚ The most distinct positive peak in the DRDF appears at 4.8 A (arrow (a)). Furthermore, a characteristic flat region (b) also appears, whose absolute value of DRDF is nearly equal to ˚ in the ranges from 6 to 8 A. ˚ Since DRDF is 100 electrons/A not a radial distribution of electron density but a reduced expression of the difference between the electron density distribution and the average density, this flat region does not imply ˚ in the that there is no electron distribution around 6 < r < 8 A specimen. Interpreting the values of DRDF in the range of ˚ is difficult due the average interatomic distance r below 3 A to the termination effect of the Fourier transformation.
and it is certain that these structural changes occur at lower temperatures. At 270 ◦ C, the sharp peak disappears. This implies that thermal decomposition occurred and the specimen is carbonized at that temperature. Using the measured halo pattern at 210 ◦ C after eliminating the sharp peak, the DRDF corresponding to the high temperature noncrystalline structure was calculated. The obtained DRDF is shown in Fig. 4. Compared with the DRDF at room temperature, shown in ˚ (a) is also observed. On Fig. 2, a large peak around r = 4.8 A the other hand, a characteristic flat region observed at room temperature is not observed. A small positive peak in the ˚ (b) and a distinct negative peak at 8 A, ˚ (c) are vicinity of 6 A, ◦ observed at 210 C.
3.2. Structural change of regenerated silk fibroin on heating
In order to clarify the change in conformation and the crystalline structure of regenerated silk fibroin during wetting, time resolved X-ray diffraction measurements were carried out at 30 ◦ C. After the addition of water, the diffraction profiles were taken at every 2 min from 18 to 38 min. Water was added using a micro pipette and the volume was equal to that of the specimen. The DRDFs were then calculated from ex-
High temperature X-ray diffraction measurements were carried out to determine the structural changes of regenerated silk fibroin. The measured temperature ranged from room temperature to 300 ◦ C, and diffraction profiles were collected at every rise of 10 ◦ C. Parts of the measured diffraction profiles are shown in Fig. 3. When the temperature reaches about 190 ◦ C, the first halo peak at 2θ = 20◦ decreases. On the other hand, a sharp and strong peak appears at the same angle in the vicinity of 200 ◦ C. This sharp peak originates from the ˚ and it can crystalline state of silk II. The d-value is 4.22 A be indexed by overlapping (2 1 0) and (1 2 0) Bragg reflection. However, the broad and weak halo patterns that originated from the noncrystalline state remained. This implies that both noncrystalline and silk II crystalline regions coexist in the regenerated silk fibroin film. Motta et al. [20] reported that the random coil to -sheet transformation was observed at 213 ◦ C by means of DSC, and the result appears to be consistent with that of our study. However, the structures discussed in our study are expected to be in the metastable state,
3.3. Structural change of regenerated silk fibroin caused by the addition of water
Fig. 4. DRDF for regenerated silk fibroin at 210 ◦ C.
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
263
25 ◦ C and 33 ◦ C, the DRDFs exhibited unclear peaks in the ˚ (a). Above 40 ◦ C, a small positive peak range from 5 to 7 A ˚ (b) appeared and a negative peak around around r = 6 A, ˚ (c) became distinct; this is similar to the change in r = 7 A, time resolved DRDFs in Fig. 5. It is suggested that at a lower temperature the structural change in regenerated silk fibroin is obstructed from a kinematical point of view, and that the resultant substance is formed by a noncrystalline structure, which is formed at an early stage of structural change during wetting at higher temperatures.
4. Discussion 4.1. Modeling and analysis Fig. 5. Time resolved DRDF calculated from experimentally observed X-ray diffraction profiles of regenerated silk fibroin during wetting at 30 ◦ C. The top DRDF corresponds to the profile measured 18 min after the addition of water. The diffraction profiles were taken at intervals of 2 min until 38 min.
perimental data, as shown in Fig. 5. The top DRDF mainly originates from liquid–water. As the amount of water decreased, some peaks, the DRDF peak positions of which did not overlap with those of the resultant structure decreased ˚ (a)). On the other hand, DRDF peaks with a (e.g. r = 3 A, slight change in absolute value and shape also appeared (e.g. ˚ (b)). Since the simultaneous decrease in the abr = 4.8 A, solute value of DRDF originated from liquid–water and an increase in this value of the resultant substance occurred, the DRDFs apparently do not change significantly. The most dis˚ (c). tinct change was observed in the ranges from r = 5 to 7 A In this region, a negative peak of DRDF from liquid–water was observed; with an increase in time; however, this negative peak became flat and the absolute value became close to 0. The structural change in regenerated silk fibroin during wetting was also measured at several different temperatures. DRDFs of the resultant substance calculated from experimentally observed X-ray diffraction profiles at various temperatures are shown in Fig. 6. When the temperature was
Fig. 6. Comparison of DRDFs for the noncrystalline structure of resultant substances calculated from experimentally observed X-ray diffraction profiles at various temperatures.
In order to reproduce experimentally observed DRDFs of regenerated silk fibroin with the addition of water at 90 ◦ C (Fig. 7) and regenerated silk fibroin at 210 ◦ C (Fig. 4), both of which were similar, a small Type-II crystallite is assumed to exist in the specimen. Simulated and experimentally observed DRDFs are shown in Fig. 7. Not only the shapes and positions of peaks but also the absolute value were in good agreement. Although the least squares fitting procedure was only utilized to determine the average electron density ρ0 and the procedure was not performed to minimize the differences in the two patterns, a good agreement is evident. Consequently, it was determined that the structure of the specimens consists of randomly arranged small Type-II crystallites. The noncrystalline structure of as-cast regenerated silk fibroin was rather difficult to model, because of the existence of a characteristic flat region of DRDF shown in Fig. 2. The shape of this DRDF could not be reproduced by considering only the conformations reported previously in the crystalline state of silk I and II. In order to reproduce the DRDF of as-cast regenerated silk fibroin, the Monte Carlo simulation method was employed as described above. Only one model was successful in reproducing such DRDFs after calculating several
Fig. 7. The comparison between simulated and experimentally observed DRDFs of as-cast regenerated silk fibroin with added water at 90 ◦ C. Simulated DRDF was calculated by introducing a small crystallite of silk II. The solid line represents simulated DRDF and open circles denote the observed DRDF.
264
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
modeled noncrystalline structures. The simulated DRDF is shown in Fig. 2. It was derived from a noncrystalline structure consisting of slightly oriented Type-I sheets. The distribution of the direction of Type-I sheets was assumed to be a Gaussian distribution as described above, and the standard deviation of the Gaussian function was 10◦ . The Type-I sheet, which consisted of randomly arranged structural units for constructing a noncrystalline structure, is an isolated hydrogen-bonded Type-I chain, as reported by Asakura et al. [14,15]. Due to the poor resolution of the experimentally observed DRDFs, it was unclear if the conformation was correct. Since the TypeI crystallite model could not reproduce the observed DRDF, such a structure with hydrogen-bonded sheets seemed to be more favorable. It is, however, difficult to confirm whether such isolated and/or oriented sheets can exist by means of Xray diffraction techniques. Further experiments and theoretical discussions are required for understanding the structure. Compared with the reproduction of DRDFs of noncrystalline regenerated silk fibroin with added water at a high temperature (Type-II crystallite model), the Type-I sheet model could not reproduce experimentally observed DRDFs in sufficient detail. It is caused by the effect of the limited number of atoms available for simulation and the possibility of the coexistence of another conformation such as random coils. However, the as-cast regenerated silk fibroin does not, contain the TypeII conformation. If the model structure contains the Type-II conformation, the calculated DRDF does not show any flat ˚ On the other hand, a strucarea in the range from 6 to 8 A. tural model containing the random coil conformation was not calculated, since it is difficult to model these atomic arrangements. The as-cast regenerated silk fibroin seems to contain some random coil conformation. 4.2. Interpretation of time resolved DRDFs The kinematical model is introduced to interpret observed time resolved DRDFs. An intermediate state of regenerated silk fibroin with added water was a mixture of the specimen and a small amount of water. It is expected that water, which is a polar molecule, have a great influence on the structural change. To construct the model, it is more favorable to consider correlations between molecules of water and regenerated silk fibroin. However, only the structural change in regenerated silk fibroin conformation and the noncrystalline structure are discussed for the present calculation. There are several structural characteristics in the observed time resolved DRDFs that originate from conformational changes and/or noncrystalline structural changes of regenerated silk fibroin and a decrease in water, as described above. Changes in DRDFs related to the noncrystalline structure of regenerated silk fibroin cause an enhancement of the positive ˚ and the formation of a small positive peak around r = 4.8 A ˚ The peak and a distinct negative peak in the range r = 5 to 7 A. former is not clearly seen in Fig. 5 because of a simultaneous decrease in the DRDF peak from water and an increase in the DRDF peak from the regenerated silk fibroin. Kinemat-
Fig. 8. The simulated DRDFs corresponding to (a) ordering of silk I sheets (n /nall = 0), (b) structural change of silk I sheets to silk II sheets (σ = 10◦ ).
ical models to reproduce such features related to regenerate silk were investigated. These models contain the following parameters: standard deviation (σ) of the Gaussian distribution of directions of structural units and an existence ratio of different structural units (n /nall ). Here, n and nall represent the total numbers of Type-II structural units and all structural units, respectively. These parameters were varied with increasing time. The DRDFs thus obtained are shown in Fig. 8. In Fig. 8(a), it is observed that simulated DRDFs can reproduce an increase in the characteristic positive peak around ˚ (indicated by an arrow). In this model, Type-I sheets r = 4.8 A were used as the structural unit of noncrystalline regenerated silk fibroin in the same way as that at room temperature. Structural change was modeled by an ordering of the directions of Type-I sheets expressed by a decrease in the deviation of the Gaussian function. As σ decreases, which corresponds to an ordering of the directions of Type-I sheets, the increase in ˚ is reproduced. In Fig. 8(b), the positive peak around r = 4.8 A ˚ were also formation of DRDF peaks in the range r = 5–8 A reproduced with changing n /nall . In this model, both Type-I and Type-II sheets were present as noncrystalline structural units, and these structural units were randomly arranged by following the Gaussian function, the standard deviation σ of which is equal to 10◦ . As n /nall increases, which corresponds to the conformational change from Type-I to Type-II sheets, ˚ disapthe flat region of DRDFs in the ranges from r = 5 to 8 A pears and a small positive and distinct negative peak appears. As described above, this model was able to reproduce the observed DRDFs. The reproduction depends on the difference between conformations of Type-I and Type-II chains. The Type-II chain forms a rather straight and simple chain conformation compared with the Type-I chain that has complicated conformations; therefore, DRDF peaks that originate from the regular distribution of atomic distances appear with the formation of Type-II conformations. On the other hand, due to the irregular distribution of atomic distances, DRDF peaks do not appear in spite of the existence of Type-I chains. As Type-I chains are ordered and regularly arranged, hydrogen bonded Type-I sheets are formed and DRDF exhibits ˚ characteristic peak around r = 4.8 A.
H. Saitoh et al. / International Journal of Biological Macromolecules 34 (2004) 259–265
Taking these results into consideration, the structural change of noncrystalline regenerated silk fibroin during wetting can be interpreted as follows. When water is added to regenerated silk fibroin, the chain conformations become random coil and/or isolated Type-I. As time passes and the amount of water decreases, chains form hydrogen-bonded Type-I sheets and an ordering of the formed Type-I sheets occurs. Subsequently, conformational transformation from Type-I to Type-II sheets occurs. Finally, the ordered Type-II sheets form small crystallites of Type-II, and the transformed structure of regenerated silk film consists of small Type-II crystallites. A small Type-II crystallite forming noncrystalline structure has been proposed in this study. This kind of nucleation has already been previously predicted by means of birefringence measurements of silk glands and a spinneret of the Bombyx mori silkworm [21]. It has been reported that the nuclei of Type-II was too small so as to sufficiently observe the intensity that originated from it on the X-ray diffraction pattern. Since no crystal structure exists in these specimens, it is natural that nucleation of small Type-II crystallites could not be detected by means of X-ray diffraction without a precise analysis of the noncrystalline structure, as carried out in this study. Therefore, a noncrystalline structural model consisting of small Type-II crystallites is consistent with the noncrystalline model predicted previously. The kinetics of the spinning process of natural silk remains unclear. Moreover, in the natural silk fiber, a sizeable crystal region induced by mechanical stress and/or elongational flow of Type-II exists. In order to clarify the kinetics of the spinning process, it is necessary to elucidate the crystallization process. The present noncrystalline structural analysis will facilitate understanding such a kinematical study.
5. Conclusion In order to clarify the noncrystalline structure of regenerated Bombyx mori silk fibroin, an X-ray diffraction measurement was carried out. DRDFs describing poorly defined noncrystalline structures of materials were calculated from experimentally observed diffraction profiles. The temperature dependence of the noncrystalline and crystalline structure of regenerated silk fibroin was investigated using a high temperature X-ray furnace. Time resolved X-ray diffraction profiles were also obtained to construct a kinematical model of structural transformation during wetting. Calculated DRDFs from experimental data were compared with simulated DRDFs on the basis of the Monte Carlo method.
265
Noncrystalline structures contain some structural units that are randomly arranged in the specimen. These structural units were similar to the previously reported ones in the crystalline state and those with appropriate structural defects. From the comparison of experimental DRDF and simulated DRDFs, it was determined that the regenerated silk consisted of locally ordered Type-I sheets, and that added water is noncrystalline state formed by small Type-II crystallites. Time resolved DRDFs were also qualitatively interpreted by the ordering of Type-I sheets and structural changes of Type-I into Type-II. The formation of small Type-II crystallites determined in this study was consistent with the nucleation of Type-II crystallites by birefringence measurements of silk glands and a spinneret of Bombyx mori silkworm predicted previously. The analysis of the noncrystalline structure and structural change performed in this study would be effective in clarifying the silkworm’s spinning process. References [1] Marsh RE, Corey RB, Pauling L. Biochim Biophys Acta 1955;16:1. [2] Takahashi Y, Gehoh M, Yuzuriha K. Int J Biol Macromol 1999;24:127. [3] Konishi T, Kurokawa M. Sen’i Gakkaishi 1968;24:550 (in Japanese). [4] Lotz B, Keith HD. J Mol Biol 1971;61:201. [5] Lotz B, Cesari FD. Biochimie 1979;61:205. [6] Saito H, Tabeta R, Asakura T, Iwanaga Y, Shogi A, Ozaki T, Ando I. Macromolecules 1984;17:1405. [7] Asakura T, Kuzuhara A, Tabeta R, Saito H. Macromolecules 1984;18:1841. [8] Asakura T, Yamaguchi T. J Seric Sci Jpn 1987;56:300 (in Japanese). [9] Anderson JP. Biopolymers 1998;45:307. [10] Okuyama K, Takahashi K, Nakajima Y, Hasegawa Y, Hirabayashi K, Nishi N. J Seric Sci Jpn 1988;57:23 (in Japanese). [11] Fossy SA, Nimithy G, Gibson KD, Scheraga HA. Biopolymers 1991;31:1529. [12] Asakura T, Demura M, Date T, Miyashita M, Ogawa K, Williamson MP. Biopolymers 1997;41:193. [13] Asakura T, Iwatdate M, Demura M, Williamson MP. Int J Biol Macromol 1999;24:167. [14] Asakura T, Yamane T, Nakazawa Y, Ando K. Biopolymers 2001;58:521. [15] Asakura T, Ashida J, Yamane T, Kameda T, Nakazawa Y, Ohgo K, Komatsu K. J Mol Biol 2001;306:291. [16] Takasu Y, Yamada H, Tsubouchi K. Biosci Biotechnol Biochem 2002;66:2715. [17] Yamada H, Nakao H, Takasu Y, Tsubouchi K. Mat Sci Eng C 2001;14:41. [18] Norman N. Acta Cryst 1957;10:370. [19] Warren BE. X-Ray Diffraction. New York: Dover publication, Inc.; 1969. [20] Motta A, Fambri L, Migliaresi C. Maclomol Chem Phys 2002; 203:1658. [21] Kataoka K. Sen’i Gakkaishi 1978;34:20 (in Japanese).