The role of irregular unit, GAAS, on the secondary structure of Bombyx mori silk fibroin studied with ¹³C CP/MAS NMR and wide-angle X-ray scattering

Abstract

Bombyx mori silk fibroin is a fibrous protein whose fiber is extremely strong and tough, although it is produced by the silkworm at room temperature and from an aqueous solution. The primary structure is mainly Ala-Gly alternative copolypeptide, but Gly-Ala-Ala-Ser units appear frequently and periodically. Thus, this study aims at elucidating the role of such Gly-Ala-Ala-Ser units on the secondary structure. The sequential model peptides containing Gly-Ala-Ala-Ser units selected from the primary structure of B. mori silk fibroin were synthesized, and their secondary structure was studied with ¹³C CP/MAS NMR and wide-angle X-ray scattering. The ¹³C isotope labeling of the peptides and the ¹³C conformation-dependent chemical shifts were used for the purpose. The Ala-Ala units take antiparallel β-sheet structure locally, and the introduction of one Ala-Ala unit in (Ala-Gly)₁₅ chain promotes dramatical structural changes from silk I (repeated β-turn type II structure) to silk II (antiparallel β-sheet structure). Thus, the presence of Ala-Ala units in B. mori silk fibroin chain will be one of the inducing factors of the structural transition for silk fiber formation. The role of Tyr residue in the peptide chain was also studied and clarified to induce "locally nonordered structure."

Silkworms produce strong fibers with high toughness at room temperature and from an aqueous solution of silk fibroin. For example, the strength of the silk fibroin is superior to that of collagen, wool, and bone, and the toughness is also superior to that of collagen, wool, bone, Kevlar, carbon fiber, and high-tensile steel (Gosline et al. 1999). Many superiorities of the silk fibroin, making it an excellent natural fiber, basically originate from the combination of its unique amino acid sequences and their translation of them into a higher order of structures (Asakura and Kaplan 1994).

Mita et al. (1994) and later Zhou et al. (2000), employing shotgun sequencing strategy combined with traditional physical map-directed sequencing of the fibroin gene of the heavy chain of Bombyx mori (B. mori) silk fibroin, predicted the presence of unusual repeats in the sequence of the silk fibroin gene. The repetitive core is composed of alternate arrays of 12 repetitive (R01–R12) and 11 amorphous (A01–A11) regions (Zhou et al. 2000). The highly repetitive GAGAGS sequences, constituting the crystalline region, have been studied frequently as a typical structural model of B. mori silk fibroin. Actually, Strydom et al. (1977) reported the sequence of the fraction (about 55%) of the silk fibroin precipitated after chymotrypsin enzymatic cleavage (Cp fraction) to be GAGAGSGAAG[SG(AG)n]₈Y, where n is usually 2. The primary structure is essentially in agreement with the sequences reported by Zhou et al., except for the deletion of the G residue following the AA sequence. For this crystalline region, two crystalline forms, silk I and silk II, have been reported as the dimorphs of the silk fibroin. Silk I is the silk fibroin structure before spinning, and the sample is obtained as silk film from an aqueous solution of silk fibroin stored in the silkgland of the silkworm or an aqueous solution of the regenerated silk fibroin, whereas the silk II form is the structure after spinning and obtained as silk fibers from the cocoon. Numerous X-ray diffraction studies have been reported on the determination of the structure of silk II (Marsh et al. 1955; Fraser and MacRae 1973).

Despite a long history of interest in the "less stable" silk I form, its structure has remained largely poorly understood because attempts to induce orientation of the silk I form, for X-ray and electron diffractions studies, easily causes its conversion into the more stable silk II form. Nevertheless, we recently characterized the molecular structure of silk I by using a synthetic peptide: (AG)₁₅ as the model for the highly repetitive crystalline domain employing several solid-state NMR techniques and quantitative use of the ¹³C CP-MAS NMR chemical shifts, in conjunction with the results obtained by molecular simulation studies (Asakura et al. 2001a). The existence of a repeated β-turn type II structure stabilized by a classical 4 → 1 intramolecular hydrogen bond, was proposed.

Zhou et al. (2001) recently reported that each crystalline domain is made up of subdomains of ∼70 residues, which in most cases begin with repeats of the GAGAGS hexapeptide and terminate with the GAAS tetrapeptide. They assumed that the GAAS tetrapeptide might take a β-turn structure, which connected one β-strand to the next β-strand. However, the local structure of the GAAS tetrapeptide in typical silk fibroin sequences such as alternative AG and also the influence of the presence of such a peptide on the structure of alternative AG sequences should be examined experimentally. In general, it has been difficult to examine such effects because of the lack of suitable analytical techniques required. The combined use of solid-state NMR and "selected" model peptides labeled suitably for NMR studies may be a powerful approach to investigate their structure.

In this study, we synthesized several isotope labeled and unlabeled model peptides that contain GAAS units selected from the real primary structures reported by Zhou et al. (2000), together with stable isotope-labeled model peptides with slightly different sequences. The local structure of these peptides was studied with ¹³C CP/MAS NMR spectroscopy, especially the ¹³C NMR chemical shifts and the line shape of Ala C_β carbons. On the other hand, the secondary structures of these peptide chains were determined with WAXS pattern observation. Through these experiments, the local structure of the GAAS tetrapeptide unit in the secondary structure of B. mori silk fibroin and the role of the isolated AAS unit on the secondary structure, that is, alternate AG sequences, were clarified. The role of Tyr and Ser residues on the structure of the real sequences was also studied.

Results and Discussion

(AG)₁₅ with silk I and silk II forms (peptide 1)

The ¹³C CP/MAS NMR spectra of the peptide (AG)₁₅ with silk I and silk II forms have been reported previously (Fig. 1a,b, respectively) (Asakura et al. 2001a). When (AG)₁₅ was dissolved in 9 M LiBr aqueous solution and dialyzed against distilled water, the resulting structure was silk I. The (AG)₁₅ with silk I form was then treated with formic acid and dried; the structure was silk II. These are ascertained primarily employing ¹³C NMR spectroscopy by analyzing the peak shape and chemical shifts values of the Ala C_α, Ala C_β, Ala C=O and Gly C_α resonances. The WAXS patterns also show typical patterns of silk I and silk II as shown in Figure 2a and b ▶, respectively (Marsh et al. 1955; Fraser and MacRae 1973; Asakura et al. 2001b). These ¹³C CP/MAS and WAXS pattern data will give references for typical silk I and silk II patterns.

Fig. 2.

Wide-angle X-ray scattering patterns of B. mori silk fibroin model peptides. (a) (AG)₁₅ with silk I form, (b) (AG)₁₅ with silk II form, (c) peptide 2, (d) peptide 3, and (e) peptide 4. Peptides 2,3, and 4 were dissolved in 9 M LiBr and then dialyzed against distilled water, which was the same treatment of (AG)₁₅ with silk I form.

AGAGAGAGAGAGAGAGAG[3-¹³C]A₁₉[1-¹³C]A₂₀AGAG AGAGAG (peptide 2)

The Gly residue was changed to Ala residue at the 20th residue in (AG)₁₅. Both the role of isolated AAA units on the structure of alternative AG sequences and the structure of the AAA units were examined. For detailed investigation on the local structure, the C_β carbon at the 19th Ala and carbonyl carbon at the 20th Ala were¹³C isotope labeled. Figure 1c ▶ shows the ¹³C CP/MAS NMR spectrum of peptide 2. Both the 19th and 20th Ala residues take silk II structure locally after 9 M LiBr/dialysis treatment judging from the carbonyl carbon chemical shift (172.2 ppm) and Ala C_β chemical shift (mainly 20.2 ppm), respectively. This is in contrast to the case of (AG)₁₅, which takes silk I form after the same treatment (Asakura et al. 2001a). The relative intensity of the center peak of the Ala C_β triplet is slightly stronger compared with that of (AG)₁₅ with silk II form. This seems reasonable because the additional steric hindrance between the methyl group of the 20th Ala residue and carbonyl oxygen atom of the 19th Ala residue prevents from formation of the type-II β-turn structure. However, it is noted that both Ala and Gly C_α carbons give a single peak, and the chemical shifts were 48.8 and 42.5 ppm, respectively. This indicates that all of the residues take a β-sheet structure, although the detailed discussion on the conformation-dependent Ala C_α and C_β chemical shifts with the ¹³C chemical shift contour plots (Asakura et al. 1999) will be performed later. The WAXS pattern of peptide 2 shown in Figure 2c ▶ supports the β-sheet structure. Thus, the isolated AAA units induce the change of the structure of alternative AG sequence from silk I form to silk II form. As shown in Figure 3a and 3a` ▶, the β-sheet structure of peptide 2 is stable because formic acid treatment does not change the line shape of the Ala C_β carbon.

Fig. 1.

Solid-state ¹³C CP/MAS NMR spectra (10–70 ppm and 160–190 ppm) of B. mori silk fibroin model peptides. (a) (AG)₁₅ with silk I form, (b) (AG)₁₅ with silk II form, (c) peptide 2, (d) peptide 3, (e) peptide 3`, and (f) peptide 4. Peptides 2–4 were dissolved in 9 M LiBr and then dialyzed against distilled water, which was the same treatment of (AG)₁₅ with silk I form.

Fig. 3.

Expanded Ala C_β region in ¹³C CP/MAS NMR spectra of the samples after 9 M LiBr /dialysis treatment, (a–d) (left) and formic acid treatment, (a`–d`) (right). (a) and (a`); peptide 2, (b) and (b`); peptide 3, (c) and (c`); peptide 3`, (d) and (d`); peptide 4.

AGYGAGAGAGYGAGAGSG[3-¹³C]A₁₉[1-¹³C]A₂₀SGAG AGAGAG (peptide 3)

We selected the peptides containing the isolated GAAS unit from the real primary structure of B. mori silk fibroin. The peptide, AGYGAGAGAGYGAGAGSG[3-¹³C]A₁₉-[1-¹³C]A₂₀SGAGAGAGAG was synthesized together with nonlabeled peptide with the same sequence. The spectral pattern of peptide 3 shown in Figure 1d ▶ might indicate that the peptide takes the silk II form. However, the WAXS pattern of peptide 3 clearly shows that the peptide is amorphous, although there is a trace of the silk II pattern (Fig. 2d ▶). This is due to the difference in the basic structural information between solid-state NMR and X-ray diffraction methods. Large stretches of β-sheets structure in the sample will be recognized as the crystalline part in the X-ray diffraction pattern, but NMR gives structural information even on the local structure. However, there is an additional possibility. The peak pattern of the Ala C_α carbon is similar between peptides 2 and 3, but that of the Ala C_β carbon is quite different. This is due to the different conformation-dependent chemical shift contour plots between C_α and C_β carbons of Ala residue reported previously (Asakura et al. 1999). When the local structure of the Ala residue deviates from the typical β-sheet (φ, ψ) angles gradually, the C_β chemical shift decreases monotonously and the peak shifts to higher field. However, the chemical shift contour line is quite different in the case of the Ala C_α carbon, and very shallow over the wide range of the β-sheet region; namely, relatively insensitive to the deviation from typical β-sheet (φ, ψ) angles. If we compare the spectra between peptides 2 (Fig. 1c ▶) and 3 (Fig. 1d ▶) carefully, it is noted that Ala C=O and C_β peaks become broader, and the peak intensity at 16.7 ppm in the Ala C_β region increases. The observation of the broad peak at 16.7 ppm has been assigned to the peak with increased randomness (Asakura et al. 1985). Thus, increase in the randomness of the sample is clear in peptide 3, which is due to the incorporation of two Tyr residues in peptide 3. The ¹³C CP/MAS spectrum of the same sequence was also compared between isotope-labeled peptide 3 and nonlabeled peptide 3`. The difference was clearly observed in the carbonyl carbon region, but this is due to an increase of the peak intensity of the Gly <sup>13</sup>C-labeled carbonyl carbon in peptide 3. The most interesting point is the same peak pattern of the Ala C<sub>β</sub> carbon between these two spectra (Fig. 1d,e ▶). This means that the local conformation is almost the same, although the local environment along the chain is quite different among the Ala residues. This indicates that the local structure of each residue was controlled by the intermolecular interaction among the peptides rather than the intramolecular interaction along one peptide chain (Asakura et al. 2002). The formic acid treatment of peptides 3 and 3` gives slightly different spectral change in the Ala C_β region, although the structural change should be the same. The peak intensity at 16.7 ppm relative to that at 20.2 ppm is almost the same in the spectra (Fig. 3b,b` ▶), but the relative peak intensity at 16.7 ppm decreases clearly in the spectrum of peptide 3` after formic acid treatment (Fig. 3c,c` ▶). This means that the β-sheet structure of the peptide increases by formic acid treatment, but the 19th Ala residue is relatively insensitive to the formic acid treatment.

AGAGAGAGAGAGAGAGSG[3-¹³C]A₁₉[1-¹³C]A₂₀SGAG AGAGAG (peptide 4)

To clarify the effect of two Tyr residues on the structure, two Tyr residues are substituted to two Ala residues in peptide 4, AGAGAGAGAGAGAGA-GSG[3-¹³C]A₁₉[1-¹³C]A₂₀SGAGAGAGAG. From the comparison with the spectrum of peptide 2, it is also possible to know the effect of two Ser residues on the structure containing the AA units. As shown in Figure 2e ▶, the WAXS pattern indicates that peptide 4 takes a typical silk II structure and, therefore, it is clear that the randomness of peptides 3 and 3` comes from the presence of two Tyr residues. The peaks of Gly C_α, Gly CO, and Ala C_β carbons in peptide 4 are clearly sharper than those in peptide 3 (Fig. 1f ▶). In addition, the relative intensity of the peak at 16.7 ppm decreases slightly compared with that of peptide 3, but increases considerably compared with that of peptide 2. Basically, the replacement of Ala by Ser promotes silk II formation through the intermolecular hydrogen bond formation through the side chain OH group (Kameda et al. 1999). However, formic acid treatment of peptide 4 induces a change from a β-sheet to a random coil rather than the change to a β-sheet. When the position of AA units selected from the primary structure reported by Zhou et al. (2001) was changed such as AGAGSGAASGAGAGSGAGAGSGAGAGSGAG (peptide 5), the structure took a typical β-sheet structure (data not shown).

Material and methods

The following six model peptides were synthesized by the solid phase method.

AGAGAGAGAGAGAGAGAGAGAGAGAGAGAG 1

AGAGAGAGAGAGAGAGAG[3-¹³C]A₁₉[1-¹³C]A₂₀AGAG AGAGAG 2

AGYGAGAGAGYGAGAGSG[3-¹³C]A₁₉[1-¹³C]A₂₀SGAGA GAGAG 3

AGYGAGAGAGYGAGAGSGAASGAGAGAGAG 3`

AGAGAGAGAGAGAGAGSG[3-¹³C]A₁₉[1-¹³C]A₂₀SGAGA GAGAG 4

AGAGSGAASGAGAGSGAGAGSGAGAGSGAG 5

¹³C CP/MAS NMR measurements were conducted on a Chemagnetics CMX-400 spectrometer operating at 100 MHz for ¹³C. The chemical shifts were represented in ppm with respect to external reference adamantine and converted to TMS chemical shift reference.