Experimental methods for characterizing the secondary structure and thermal properties of silk proteins

Abstract

Silk proteins are biopolymers produced by spinning organisms that have been studied extensively for applications in materials engineering, regenerative medicine, and devices due to their high tensile strength and extensibility. This remarkable combination of mechanical properties arises from their unique semi-crystalline secondary structure and block copolymer features. The secondary structure of silks is highly sensitive to processing, and can be manipulated to achieve a wide array of material profiles. Studying the secondary structure of silks is therefore critical to understanding the relationship between structure and function, the strength and stability of silk-based materials, and the natural fiber synthesis process employed by spinning organisms. However, silks present unique challenges to structural characterization due to high molecular weight protein chains, repetitive sequences, and heterogeneity in intra- and interchain domain sizes. Here we review experimental techniques used to study the secondary structure of silks, the information attainable from these techniques, and the limitations associated with them. Ultimately, the appropriate utilization of a suite of techniques discussed here will enable detailed characterization of silk-based materials, from studying fundamental processing-structure-function relationships to developing commercially useful quality control assessments.

Graphical Abstract

The secondary structure of silk proteins is a critical feature that determines the mechanical properties and stability of silk-based materials. Here we review experimental techniques - both established methods and recent advancements - used for studying silk secondary structure, the information attainable from these techniques, and their limitations. Ultimately, successful characterization will require a combination of the methods reviewed here.

1. Introduction

Silk proteins are a family of naturally occurring polymers with exceptional mechanical properties. The high tensile strength and extensibility of silks is particularly remarkable because the native fibers are formed under aqueous conditions at close to ambient temperature and pressure.^[1,2] These properties can be attributed to the secondary structure and solid-state packing of silks, which arise from the unique primary sequence resembling a block copolymer (Figure 1).^[3–5] Consequentially, studying the secondary structure of silks and silk analogs is of great interest due to its implications on material strength and stability. Crystalline β-sheet formations dominate the secondary structure of silkworm fibroin and spider dragline silks, imparting mechanical strength to the fibers.^[3,5] These crystalline domains are interspaced with amorphous regions, yielding a semi-crystalline fiber that exhibits both toughness and extensibility ^[3]. Silk fibroin from the cocoons of Bombyx mori (B. mori) silkworms and dragline silk from Nephila clavipes (N. clavipes) spiders have been studied extensively due to their outstanding mechanical properties and promising applications in regenerative medicine, drug delivery, devices, optics, and materials engineering.^[6–11]

Figure 1:

The structure of silk proteins from B. mori silkworms and N. clavipes spiders. The primary structures contain both repetitive hydrophobic regions (crystalline) and hydrophilic regions (amorphous), giving rise to a semi-crystalline conformation and tough mechanical properties. Secondary structure can be studied by a variety of techniques, including those covered in this review: ssNMR, solution NMR, CD, X-ray scattering, Raman and FTIR spectroscopy, TGA, DSC, and FSC. Primary structure of B. mori adapted,^[17] primary structure of N. clavipes adapted. ^[37,120]

Studying the secondary structure of silks is essential for engineering silk-based materials and understanding how spinning organisms synthesize fibers in nature. However, silk proteins present unique challenges to structural analysis. Compared with other commonly studied proteins and small peptides, silks have relatively high molecular weights, contain long repeat regions, and have variable domain sizes.^[4,5] Additionally, the secondary structure of silk proteins is highly sensitive to processing conditions; the β-sheet and bound water contents in silks can be modulated by exposure to solvents (e.g., methanol), physical agitation (e.g., sonication, shear stress), temperature, pH changes, or humidity-controlled water annealing.^[2,12–14] Therefore, sample preparation must be carefully considered, as handling may alter the material properties or create artifacts.^[15] This review covers spectroscopic and thermal analysis techniques used for studying the secondary structure and thermal properties of silks that have been reported in the recent literature, and highlights the advantages and limitations of each technique as they relate specifically to silk proteins.

1.1. Silkworm silk fibroin

Silk fibroin extracted from the cocoons of B. mori silkworms is a ubiquitous protein in the textile industry and widely studied for biomedical and materials engineering applications.^[3,8,16] The primary structure of B. mori contains twelve highly repetitive hydrophobic regions (average of 413 residues) interspaced with eleven amorphous regions (42-43 residues), and hydrophilic N- (151 residues) and C- (50 residues) terminal domains.^[17,18] The hydrophobic regions are dominated by repeats of the hexamer Gly-Ala-Gly-Ala-Gly-Ser (GAGAGS), and terminated with GAAS tetramers. These hexamers are periodically disrupted by Tyr (Y) and Val (V) substitutions for Ser, resulting in GAGAGY and GAGAGV motifs.^[19] These highly ordered G-X repeats can self-assemble into antiparallel β-sheet conformations. The amorphous regions are non-repetitive and contain polar, charged, aromatic, and hydrophobic amino acids not found in the crystalline region, including a 25-residue sequence conserved between segments.^[17,18] Overall, the 390 kDa heavy chain of silk fibroin contains 45.9% Gly, 30.3% Ala, 12.1% Ser, 5.3% Tyr, and 1.8% Val; other amino acids found in the amorphous and terminal regions account for the remaining 4.7%.^[17] Silk fibers generated by B. mori are up to 55% crystalline ^[20], while dragline silk from N. clavipes have been reported to be 28% crystalline based on X-ray diffraction measurements.^[21]

The unique primary structure of B. mori has been reported to occur as three different polymorphs, termed silk I, silk II, and silk III. Silk I is a metastable, soluble form found in the glands of the silkworm;^[22,23] a similar structure can be obtained when reconstituted silk fibroin is cast into films.^[24] Silk I is transformed into silk II when it passes through the gland of the silkworm, where it is exposed to shear stress and changes in the biochemical environment.^[16,24,25] Silk II morphology is found in both native spun fibers as well as reconstituted silk that has undergone treatment to render it insoluble (e.g., methanol treatment, water annealing).^[26] A third, distinct polymorph, silk III, can occur at the air/water interface and reflects the amphilicity of silk fibroin.^[27]

1.2. Spider dragline silk

Spiders produce up to seven types of silk from different spinning glands, each with its own distinct amino acid sequence, mechanical properties, and use.^[28,29] Dragline silk, which is the dominant form used in web spinning, is primarily composed of spidroin proteins from the major (Ma) ampullate glands,^[30] and exhibits high tensile strength and extensibility.^[29] MaSp1 and MaSp2 silk proteins are composed of 30-40 amino acid sequences that are repeated approximately 100 times, and N- (155 residues) and C- (100 residues) terminal domains.^[13] The amino acid sequence is primarily composed of Gly, 42%, Ala, 25%, Glu, 10%, Leu, 4%, Arg, 4%, Tyr 3%, and Ser, 3% for the most commonly studied N. clavipes species.^[31] However, the amino acid composition can vary significantly depending on the spider species particularly when it comes to proline (Pro), which greatly impacts the mechanical properties of the various dragline fibers.^[32] Aqueous silk solution is found in high concentrations in the spinning glands (25-50 wt%), and is dominated by random coil with some α-helix structures.^[33–36] The solution then passes through a narrow duct and undergoes a conformational change to primarily β-sheet structure, in response to changes in the biochemical environment (pH and ion concentration) and shear forces to yield insoluble fibers.^[1,12,13] The self-assembly of β-sheet structures is made possible by the poly(Ala) repeats and flanking poly(Gly-X) units where X = Gly, Ala, or Ser.^[37,38]

2. Spectroscopic Techniques

2.1. Solid state Nuclear Magnetic Resonance (ssNMR)

ssNMR has been used to study silk proteins, providing high resolution analysis of the local conformations at the atomic level. NMR is a good alternative to x-ray diffraction (XRD), which only probes the crystalline domains in silks; the non-crystalline (amorphous) regions can comprise nearly half to three quarters of silkworm silk and spider dragline silk, respectively.^[39,40] Significant progress has been made in the field of ssNMR to study silk proteins in the fiber, typically combining stable isotopic (¹³C, ¹⁵N, ²H) labelling of silk samples with magic angle spinning (MAS). The incorporation of stable isotopes, and particularly selective isotope labels, allows multinuclear and multidimensional ssNMR to determine silk structure through a combination of inter-atomic distance and torsional angle measurements and isotropic chemical shift perturbations.^[40–43]

Stable isotopic labeling of ²H, ¹³C, and ¹⁵N in silk proteins for NMR analysis has been achieved by both synthesizing model peptides with isotopic amino acids,^[19,44,45] as well as introducing enriched amino acids into the diet of the spinning organism.^[38,46,47] In early work, cross polarization/magic angle spinning (CP/MAS) NMR and tracking of ¹³C chemical shifts was proven to be useful for studying conformational changes in both silk fibroin ^[48] and dragline silk.^[31] In an important step towards understanding silk I and silk II polymorphs in silk fibroin, ¹³C-CP/MAS NMR was used to study model peptides; a type II β-turn structure stabilized by intramolecular H-bonds structure was proposed for silk I,^[49] and a heterogeneous structure with distorted β-turns, distorted β-sheet, and antiparallel β-sheet was proposed for silk II.^[4,20] More recently, ¹³C-CP/MAS NMR was used to study the effect of plasticizers on silk, demonstrating that the addition of 30% glycerol to silk fibroin films resulted in a β-sheet fraction between that of untreated silk films and methanol treated silk films.^[50]

¹³C Direct Detection (DD)-MAS NMR has been used to probe protein-water interactions in silk samples by monitoring changes in both the mobile and immobile domains of hydrated silks.^{[46,47,51–54]} Water in the form of freezing bound water, nonfreezing bound water, and freezing free water has a significant influence on the secondary structure, thermal properties, and mechanical properties of silks.^[55,56] However, these interactions can be challenging to study because water overlaps with the amide I band in infrared analysis, and thus DD-MAS NMR is a powerful tool for studying this fundamental relationship. In one example, Ala, Ser, and Tyr residues in silk fibroin were isotopically labeled (¹³C) and studied by DD-MAS NMR in both the dry and hydrated state, revealing site specific effects of water on silk (Figure 2).^[52] In another case, the amount of Ala, Gly, and Ser in β-sheet and disordered, helical and/or turn-like regions in dragline silk was studied by ¹³C DD-MAS NMR, and quantitatively related to the protein primary sequence.^[57,58] In another example, 2D ¹³C-¹³C correlation spectra with dipolar-assisted rotational resonance (DARR) recoupling and ¹³C DD/MAS was used to study aromatic residues in dragline silks; results showed that Phe and Tyr residues in the repetitive domains are in disordered helical conformations, and are not incorporated into β-sheet crystals.^[59]

Figure 2:

¹³C DD/MAS NMR spectra of Ala Cβ in enriched B. mori fibers in the dry and wet state. The raw spectra are shown along with fitted peaks, (1) β-sheet B, (2) β-sheet A, (3) random coil, and (4) hydrated random coil. This data indicates that Ala residues in the random coil regions interact with water, while those present in β-sheet regions do not. X-axis represents the chemical shift in parts per million (ppm) from the tetramethylsilane (TMS) reference standard, and the y-axis is signal intensity. Reproduced with permission.^[52] Copyright 2015, American Chemical Society.

In a recent study on spider dragline silk from black widow spiders (Lactrodectus hesperus), the ¹³C DD-MAS method was used to track the conversion of spider silk Ma gland fluid (primarily random coil conformation) to β-sheet rich spider silk fibrils in vitro.^[60] The advantage of this approach is that poly(Ala) in both the fluid phase and the solid fibril can be observed simultaneously (Figure 3). This allows for the monitoring of spider silk assembly kinetics as a function of biochemical conditions; in this case, the kinetics of nucleation and fiber growth were tracked as a function of pH and time. The resulting kinetic profiles illustrated that silk fiber formation is nucleation dependent, and the rate of nucleation reaches a maximum when the pH value is close to the isoelectric point (pI) of the N-terminal domain and the silk protein. In addition, decreasing the pH accelerated the rate of elongation, while a higher percentage of the silk protein exhibited β-sheet conformation in the silk fiber at less acidic conditions. Therefore, the gradual pH decrease along the duct can be used to optimize the speed of silk production as well as achieve higher crystalline compositions in the spun fibers. We anticipate that this in vitro ¹³C ssNMR approach will be a valuable tool moving forward to investigate a range of biochemical conditions on silk assembly including the effect of various salts and solvents.

Figure 3:

(A) ¹³C DD-MAS ssNMR spectra of U-¹³C-L-alanine labelled black widow Ma silk gland fluid at pH = 4 before (black) and after (red) spider silk assembly. The Ala C_β region is framed by dashed lines and was analyzed to monitor assembly kinetics. (B) Ala C_β regions from the ¹³C DD-MAS spectra collected at different time points are overlaid to show that the decrease in random coil resonance and increase in β-sheet resonance can be observed simultaneously as a function of time. Reproduced with permission.^[60] Copyright 2015, American Chemical Society.

ssNMR has also been used to explore site specific dynamics by incorporating ²H/¹³C isotope labels and utilizing ²H-¹³C heternonucelar correlation (HETCOR) MAS methods. This approach was used to interrogate the dynamics of spider dragline silks in the dry and water-hydrated, supercontracted states. In one study, the Ala residues were ²H/¹³C-labelled and the HETCOR ssNMR data was used to extract dynamic information from ²H MAS lineshape and T₁ analysis.^[61] Two dynamic Ala sub-regions were detected in hydrated silk where one region exhibited slower dynamics, attributed to the β-sheet core structures (poly(Ala)), while a second region exhibited faster dynamics, attributed to the β-sheet interphase (poly(Gly-Ala)). Ala confined to disordered 3₁-helical domains (non-β-sheet) exhibited the most rapid dynamics as expected in wet, supercontracted silk. In a related study, ²H/¹³C isotopes were incorporated into Pro to monitor the supercontraction process for the elastic-like type II β-turn in MaSp2 from A. aurantia spiders. This study elucidated the Gly-Pro-Gly-XX backbone and side chain molecular dynamics with ²H-¹³C HETCOR MAS, yielding a model of what happens to this Pro-rich motif when dragline silk is water-wetted and supercontracted.^[62]

One of the main advantages of ssNMR is that it does not rely on long range periodicity like XRD, and therefore can readily be used to determine the secondary structure of non-crystalline, disordered regions. In a recent investigation of Ma dragline silk fiber from the N. clavipes species, the X amino acids in the Gly-Gly-X repeats were isotopically enriched.^{[63] 13}C chemical shifts extracted for X in the Gly-Gly-X repeat stretch where X = Leu, Tyr, and Gln were a bit ambiguous, however, it was determined that they were not β-sheet or α-helical. A combination of ssNMR and molecular dynamics (MD) simulations were then used to illuminate the conformational structure of poly(Gly-Gly-X) and results gave new insight into the secondary structure of these segments, providing further support that these regions are disordered and primarily non-β-sheet. Further, the combined ssNMR-MD approach revealed the possibility for several structural elements in the poly(Gly-Gly-X) region, including β-turns, 3₁₀-helicies, and coil structures with an insignificant population of α-helix, highlighting the complexity of this common spider silk protein motif. It is envisioned that this combined approach will be powerful moving forward for elucidating the conformational structure and hierarchical organization of other silk motifs that remain undetermined. Indeed very recently, a combined ssNMR-MD study on Gly-rich spider silk peptide mimics with selective isotopes continued to shed light on the complex Gly-rich regions of spider silk proteins.^[43]

ssNMR is a relatively expensive technique that requires considerable knowledge and skill to analyze, as well as substantial sample preparation where isotopic labeling is used. Silk proteins can be particularly challenging to study since certain amino acids exist in multiple conformations within a single protein.^[38] While ssNMR is high resolution and can reveal local molecular structure and conformation, silks are large and contain a multitude of conformations and assemblies, and therefore the ability to represent the protein structure more broadly remains challenging. For these reasons, NMR is most useful in addition to spectroscopic techniques such as FTIR, Raman, CD, or WAXS that provide complimentary information on the bulk material. A number of additional techniques have been applied to the study of silks by NMR, and more detailed reviews have been dedicated to this topic.^[40,64]

2.2. Solution NMR

Solution NMR techniques have been used to characterize the structure of silk protein contained in the gland of silkworms and spiders prior to spinning. Early solution NMR studies have exhibited that the silk protein contained within the silkworm glands of B. mori ^[65] and the Ma glands of N. clavipes, ^[35,36] N. edulis, ^[34] A. aurantia, ^[35] and L. hesperus ^[33] are present primarily as unstructured random coils. Model peptide mimics for the MaSp1 spider silk protein have also been investigated with similar results observed compared to the native Ma silk dope, where the peptide mimic appeared random coil based on solution NMR conformational chemical shifts.^[66]

More recently, by utilizing a combination of solution NMR and computational conformation analysis, it was illustrated that the native silkworm silk dope is present in a disordered structure with some evidence for a repeated type II β-turn structure similar to silk I.^[67] Very recently, solution NMR was used to investigate the conformational dynamics of the repetitive domain of N. clavipes spider dragline silk where two major populations were observed: ~65% random coil and ~24% polyproline type II helix (PPII helix). The authors proposed that the PPII helix conformation in the Gly-rich region is the soluble prefibrillar domain that undergoes intramolecular interactions, and could provide the initial steps to β-sheet formation.^[68]

Solution NMR is a powerful approach for determining silk protein backbone dynamics. The protein molecular dynamics are likely of equal, if not greater importance to the silk spinning process in both silkworms and spiders. Recently, the conformational structure and backbone dynamics of native black widow Ma spider silk proteins was explored with solution NMR (Figure 4).^{[33] 15}N NMR relaxation parameters T₁, T₂, and heteronuclear nuclear Overhauser effect (NOE) of the spider silk proteins were measured at two spectrometer frequencies (500 and 800 MHz). Chemical shift analysis and dynamics extracted from ¹⁵N relaxation data revealed that the repetitive core of the spider silk proteins within the Ma glands is unfolded, with random coils exhibiting rapid backbone dynamics on a sub-nanosecond timescale. This is indicative of highly flexible, unstructured proteins. The results suggest that future solution NMR studies utilizing different isotope labeling schemes may shed light on some long unresolved questions such as the silk folding pathway, intermediate states, and biochemical factors important for protein assembly. These include pH effects, salt concentration gradients and protein-protein interactions between MaSp1 and MaSp2.

Figure 4:

2D ¹H-¹⁵N HSQC spectrum of (A) ¹⁵N-labeled Black Widow spider MA glands in 90:10 H₂O:D₂O, pH 7.0, collected at 298 K, 800 MHz; (B) strip plots of the 3D triple resonance HNCACB spectrum of U- [¹³C, ¹⁵N]-L-alanine labeled Black Widow spider MA glands in 90:10 H₂O:D₂O, pH 7.0, collected at 298 K, 500 MHz. The 12 resolved backbone cross peaks (A) were labeled with one-letter amino acid codes (bold) next to the symbols of their previous residues. Assignment is based on the (B) HNCACB experiment. (B) Selected ¹⁵N planes with unique ¹H-¹³C correlations were displayed with antiphased ¹³Cα (black), ¹³Cβ (red) cross peaks from the corresponding amino acids with relative higher intensities and their previous amino acids with lower intensities. Reproduced with permission.^[33] Copyright 2014, Elsevier.

Solution NMR of silks has similar challenges to ssNMR, namely that the proteins are large, highly repetitive, and analysis requires significant skill to carry out. Solution NMR is particularly useful in probing the molecular-level conformations of silk proteins in the solution state, such as that found in the glands of spinning organisms, and for studying protein backbone dynamics. It is best used in addition to other techniques that provide broad information on protein structure, such as FTIR, Raman, or CD.

2.3. Circular Dichroism (CD)

In CD analysis, the difference in the absorption of left- and right-handed polarized light by silk proteins is measured to characterize secondary structure.^[69] Protein conformations have distinctive CD spectra in the far-UV range due to excitation of the amide group; α-helical conformations have negative bands at 222 and 208 nm and a positive band at 193 nm, antiparallel β-sheets have a negative band at 218 nm and a positive band at 195 nm, and random coil shows low ellipticity above 210 nm and a negative band near 195 nm.^[70] These characteristic CD spectra based on model proteins can be used to estimate the relative abundance of secondary structures in silks, and to study conformational stability over time.^[69]

The study of silks by CD was reported as early as 1966, when it was used to track an increase in β-sheet structures in silk fibroin samples with the addition of methanol.^[71] Few advancements have been made to the experimental method since, although different data processing algorithms to predict secondary structure have been developed.^[72] In one example, Ma silk extracted from two morphologically distinct regions of a spider gland exhibited distinctly different CD spectra; one region contained β-sheet-poor silk while the other contained silk with predominately β-sheet conformation.^[73] CD has also been used to estimate the relative fractions of random coil and β-sheet in silk fibroin after exposure to two different proteases. Digestion by protease XIV resulted in a product with 37% β -sheet and 50% random coil, while digestion by alpha-chymotrypsin resulted in 5% β-sheet and 57% random coil.^[74] More recently, a spontaneous shift from random coil to β-sheet conformation was studied by CD in reconstituted silk fibroin solution over 10 days, and it was postulated that tyrosine templating was a driving factor in the conformation change (Figure 5).^[75]

Figure 5:

CD spectra showing a shift from predominately random coil conformation (minima at 195 nm), to β-sheet (negative band at 218 nm), in silk fibroin hydrogels over a period of ten days. The x-axis represents the wavelength in nanometers, shown here in the far-UV range. The y-axis is the signal intensity, where positive intensity indicates greater absorbance of left-handed than right-handed light, and negative intensity indicates lower absorption of left-handed than right-handed light. Reproduced with permission.^[75] Copyright 2016, American Chemical Society.

Vibrational CD (VCD), an extension of CD into the IR range, has also been used to study secondary structures in silk proteins, particularly in the amide I region. In one example, the temporal change in liquid silk extracted from the middle gland of B. mori was studied by VCD. Initially, the liquid silk sample showed a major right handed band at 1650 cm⁻¹ and a minor band at 1680 cm⁻¹, indicative of random coil structure. Over time the major band decreased, the minor band increased, and two right handed bands appeared at 1620 and 1660 cm⁻¹. The bands at 1680 and 1660 cm⁻¹ were attributed to intermediate structures, while the band at 1620 cm⁻¹ indicated β-sheet structures. These intermediate structures may provide more information on the temporal changes in silk as it transitions from the liquid state to solid fibers.^[76]

CD analysis is useful for estimating the relative amounts of secondary structures, or for studying stability over time or under different conditions. However, β-sheet structures tend to have low signals and can be obscured by large signals from helical structures, making analysis of silk samples challenging.^[71,72] Additionally, CD peak assignments in deconvolution analysis programs are based on fundamental studies with globular proteins, and are not necessarily suitable for fibrous proteins.^[70,77] For these reasons, CD analysis of silks is best used qualitatively. CD also requires dilute liquid samples, approximately 0.005-5 mg/ml,^[69] which is irrelevant to the high protein concentrations found in native silk spinning dopes, and excludes solid samples. CD is useful in that it is rapid and requires a relatively small sample of less than 1 mL, but is best used in addition to quantitative or higher resolution analysis techniques, such as solution NMR.

2.4. X-ray Scattering

X-ray scattering is a high throughput method requiring minimal sample preparation that can be used to assess the shape (molecular envelope), orientation, and degree of crystallinity in silks.^[78] It is possible to determine the atomic coordinates and complete 3D structure if a single crystal of large enough size is obtainable for a given sample. Unfortunately for silks, samples are typically semi-crystalline fibers, films, or powders that contain significant amorphous domains, making the determination of atomic coordinates impossible from X-ray diffraction (XRD) patterns.^[5,38,49] WAXS can provide information on the degree of crystallinity, the relative order or disorder of structures, and the size of crystals in solids, while small angle x-ray scattering (SAXS) is useful for estimating the size (radius of gyration), molecular weight, and oligomerization state of proteins in solution.^[78,79]

SAXS methods have been applied to samples in both the solution state (dope)^[80] as well as solid state (fibers)^[81–83]. In one example, the naturally spinning process of silk fibroin was simulated in vitro in a microfluidic cell under low pH conditions, and the silk solution was studied by synchrotron SAXS. The radius of gyration was calculated from the SAXS data to study fibroin packing as it aggregated, providing insight into the conformational changes in a regenerated silk solution during the spinning process.^[80]

WAXS can be used to characterize silk fibers in their solid form. Crystalline solids have a characteristic spacing between planes, known as d-spacings, which can be measured to study the orientation and structures in silk samples.^[25] Crystal length in the direction of the diffraction plane, L, has been calculated from WAXS data by applying the Scherrer equation: L = 0.9λ/FWHM cos θ, where λ is the x-ray wavelength, FWHM is the full width at half maximum of the diffraction peak in a 1D radial intensity scan, and θ is the scattering angle.^[83–85] Using this technique, crystal size in degummed B. mori fibers has been estimated to be between approximately 1 and 11 nm, and it was hypothesized that the length of GAGAGS hexamer repeats in the primary sequence imparted an upper limit on crystal size.^[83]

WAXS data can also be used to calculate orientation, based on Herman’s orientation parameter: f = [3 cos²ψ) − 1]/2, where ψ is the FWHM of the curve-fitted peak, and a value of 1 would indicate perfect orientation.^[21,85,86] This analysis has been employed to demonstrate that drawing silk fibers - previously shown to enhance mechanical properties - increased molecular alignment.^[86] In another example, Ma fibers from Peucetia viridans (Green Lynx) spider dragline silk were found to have an orientation parameter of f=0.98 in the crystalline regions and f=0.89 in amorphous region (Figure 6).^[85] WAXS has also been employed to study the differences in orientation between Ma and minor (Mi) spidroin from N. clavipes and Argiope aurantia spiders. In both species, the Mi silk was found to have a higher crystalline content than the Ma silk, and the crystalline regions had high orientation with respect to the fiber axis. Interestingly, the order of the oriented-amorphous region varied considerably between silk samples, and it was hypothesized that this may contribute to differences in mechanical properties.^[21] Further work utilizing WAXS analysis revealed that the crystalline regions of dragline silk can regain orientation upon stretching after undergoing supercontraction, while the oriented-amorphous regions have limited ability to recover their orientation.^[87] This work suggests that not only crystalline content but also orientation may be a key determinant of fiber mechanical properties, and highlights the utility of WAXS analysis in studying these parameters.

Figure 6:

(A) 2D WAXS pattern of a MaSp dragline silk fiber bundle. (B) 1D radial intensity as a function of scattering angle, 2θ, and Gaussian curve fits to (200) and (120) peaks. The full width half max (FWHM) of the curve-fitted peaks was used to calculate crystal size by Scherrer’s equation. (C) Azimuthal integration and Gaussian curve fitting to the (200) reflection peak. FWHM of the curve-fitted peaks was used to calculate Herman’s orientation parameter, f. Reproduced with permission.^[85] Copyright 2015, Elsevier.

WAXS is a useful technique for studying crystallinity and orientation in silks, particularly between different sample groups, or to track changes over time or under different processing conditions. It requires relatively little sample prep, and is amenable to a number of sample formats including powders and fibers. However, it cannot provide information on the spatial location of crystals,^[83] or local structural changes.^[80] Further, estimating the degree of crystallinity from WAXS peaks tends to be inaccurate at low crystal densities or with small or irregular crystals.^[88] For these reasons, WAXS analysis is most useful in combination with other techniques.^[80,83] Lastly, WAXS also has the inherent disadvantage of providing no information about the structure of the non-crystalline domains which in spider silk comprises ~50-75% of the material.

2.5. Raman Spectroscopy

Raman spectroscopy has been used to study secondary structures and chain orientation in silk samples in a multitude of forms, including films, aqueous solutions, and single fibers.^[37,39] This technique is based on the inelastic scattering of near-ultraviolet (UV), visible, or near-infrared (IR) light that arises in molecules with a polarizable electron density.^[89] Raman spectroscopy can give complimentary information to Fourier transform infrared spectroscopy (FTIR) analysis. Centrosymmetric molecules can either be Raman active or IR active in a mutually exclusive manner;^[90] in silk analysis this is particularly important for samples containing water, which shows a strong IR signal but is Raman inactive.^[89] Peak shifts in the amide I (1750-1570 cm⁻¹), amide III (1350-1190 cm⁻¹), and C-C backbone (1150-970 cm⁻¹) regions are studied to indicate secondary structure, particularly β-sheet peaks at 1665, 1232, and 1269 cm⁻¹ and tyrosine peaks at 850 and 830 cm⁻¹. However, peak assignments vary between silk species due to differences in amino acid sequence.^[37,39,91]

Polarized Raman and microspectroscopy are two techniques that have been adopted to study silk proteins by Raman spectroscopy. Polarized Raman provides information on the molecular orientation of the sample (in addition to the chemical makeup), and microspectroscopy, in which the spectrometer is integrated with an optical microscope, allows for spectra to be collected on small samples or in discrete areas, including single fibers.^[39,92] In one example, β-sheets, β-turns, α-helices, and unordered structures in films and fibers from silk fibroin and dragline silk were studied using polarized Raman microspectroscopy; results showed that silk fibroin contained 50% β-sheet content, while the dragline silk contained 36-37% β-sheet. Interestingly, the percent of β-sheet is close to the percent of GAGAGS amino acids in silk fibroin (53%), however, the percent of β-sheet is higher than the proportion of Ala repeats in the dragline silk (18%). This suggests that the AG and GAA repeats are likely involved in the β-sheet crystals in addition to the Ala repeats (31% total) in dragline silk.^[37] In another instance, a device that applies strain to silk fibroin films was coupled with Raman to study the effect of mechanical deformation on the structure and orientation of silks (Figure 7).^[93] In another case, the ratio of the intensities of the double at 850 and 830 cm⁻¹ (I₈₅₀/I₈₃₀), which has been shown to be sensitive to the hydrogen bonding environment of tyrosine groups,^[94] was used to study the role of structural water in silk gels.^[95]

Figure 7:

Polarized Raman spectra of silk fibroin films measured perpendicular (XX, blue and orange) or parallel (ZZ, green and red) to the direction in which films were drawn. The draw ratio, λ, is the ratio of the final to the initial sample length. The XX and ZZ spectra are comparable before drawing, λ=1.0 (blue and green), while the spectra after drawing, λ=4.7 (orange and red) are different, indicating that mechanical deformation induces changes in molecular orientation. The scattering intensity (y-axis) is plotted as a function of the Raman shift, reported in wavenumber (cm⁻¹) (x-axis). Reproduced with permission.^[93] Copyright SAGE Publications, 2015.

Raman spectroscopy is a high throughput technique that requires minimal sample preparation, and is compatible with many sample formats including samples containing water and single fibers. It is also amenable to experimental modifications, such as controlling relative humidity or adding devices to strain the sample.^[93] However, Raman scattering is a relatively low probability event, and therefore low signal intensity can be problematic. Additionally, background fluorescence can be quite strong and interfere with the already low Raman signals.^[89] It should also be noted that the intensity of Raman bands is influenced by the molarity of the sample, and it is therefore challenging to compare between samples with different concentrations or amino acid sequences. In these cases, the ratio of the intensities of different peaks within the same spectra can be calculated in order to compare between samples.^[91]

2.6. Fourier Transform Infrared Spectroscopy (FTIR)

FTIR analysis is a rapid technique for studying silk secondary structure that can be used to study samples of a range of sizes, concentrations, and formats.^[96] It has been applied to the study of proteins based on the principle that hydrogen bonding associated with secondary structure elements have characteristic IR absorption frequencies.^[97] The amide I (1600-1700 cm⁻¹) and amide II (1600-1500 cm⁻¹) regions of proteins are affected by vibrational C=O stretching (80% contribution to amide I) and N-H bending (60% contribution to amide II), respectively.^[97]

In practice, bands associated with secondary structural elements often overlap, resulting in broad peaks from which individual structural features cannot be resolved.^[98] Fourier self-deconvolution (FSD) is a strategy to separate individual peaks from within the broad absorption bands. In FSD, Gaussian curves are fit to the second derivative of the FTIR spectra, and peaks are assigned based on standard spectra of proteins with known secondary structures.^[98] A FSD method was proposed for silk fibroin in which the amide I region was fit to 11 peaks, based on previously reported work with membrane proteins and globular proteins (Figure 8).^[88] The areas of the individual peaks can be compared to calculate the relative contribution of each structural element to the overall sample structure. Comparative data to x-ray diffraction was provided to confirm the FSD peak assignments, and relative fractions were shown to correlate well with DSC analysis.^[88] FTIR analysis has been used extensively to study silk proteins, by both qualitative observation of shifts in the raw spectra, as well quantitative analysis of secondary structure elements using FSD.

Figure 8:

Fourier self-deconvolution (FSD) curve fitting of the amide I region of silk fibroin films FTIR spectra. Peak assignments are labeled as follows: side chain (S), β-sheet (B), random coil (R), α-helix (A). Silk fibroin films were water annealed at (a) 4°C and (b) 95°C. The absorbance (y-axis) is proportional to the concentration, and is reported with respect to wavenumber (cm⁻¹) (x-axis), which is proportional to frequency. Reproduced with permission.^[14] Copyright 2011, American Chemical Society.

As an alternative to traditional transmission IR, spectrometers equipped with an Attenuated Total Reflection (ATR) element have been used to study silk constructs, based on the attenuation of an IR beam that has contacted the sample and been reflected in the ATR crystal.^[99] This technique has been used to study orientation and conformation in silk samples, from films and solutions to individual fibers,^[100] and even fibers spun directly onto the instrument.^[101] Silks from different species have also been classified based on multivariate analysis of FTIR-ATR spectra.^[102]

Another variation of traditional FTIR is synchrotron FTIR (S-FTIR) microspectroscopy. S-FTIR is particularly well suited to studying samples in the micrometer scale, such as silk fibers, as it combines a synchrotron IR beam with a microscope. This technique has been used to determine the β-sheet content in silk fibers obtained from different species. Semi-quantitative results were obtained by assigning peaks in the amide III region, based on a combination of traditional FTIR spectra on regenerated silk samples and literature values. Results of this study indicated 28% β-sheet content in B. mori silk fibroin, and 23 and 17% in Antheraea pernyi and N. edulis dragline silk, respectively.^[103]

More recently, a new approach to FTIR peak fitting was proposed which aimed to automate the fitting processing without deconvolution and minimize error contributed by user manipulation.^[104] Fitting parameters including peak position, width, and shape were optimized from FTIR-ATR spectra of silk fibroin films as cast (predominately random coil) and after water annealing (β-sheet). Model proteins (bovine serum albumin, α lactalbumin, and γ globulin) with known structural conformations, and in particular different β-sheet contents, were also studied to validate the peak fitting method. The further development of techniques to analyze FTIR spectra is expected to move the field towards the ultimate goal of producing reliable and consistent semi-quantitative information.

FTIR provides rapid analysis of secondary structures in silk samples, and is versatile with regards to sample form, concentration, protein size, and sample format.^[98] However, water absorbs strongly in the amide I region, and therefore can obscure the resulting spectra.^[96,98] Additionally, it is important to understand the assumptions that are made in FSD data processing, namely: 1) assigning 11 peaks to fit the amide I region assumes that all of the structural elements represented by the peaks are present in the sample, 2) fitting the peaks assumes that they are Gaussian in shape, and that all secondary elements have the same peak shape, and 3) using the area of each peak to calculate a relative contribution to the amide I region assumes that the molar absorptivity of each structural element is the same.^[97] FTIR is best suited for estimating the relative contribution of each structural element to a silk sample, and for understanding the broad structural makeup of silk proteins. However, there is no guarantee about the band assignment when applying this method to silk structural analysis as the restraints are typically based on previous assignments for other non-silk proteins.^[105] Additional techniques such as ssNMR with selective isotope labeling should be used to probe the local amino-acid specific structure. Raman and/or CD should also be used to confirm FTIR results and provide complimentary structural information.

3. Thermal Analysis

3.1. Differential Scanning Calorimetry (DSC)

DSC measures the amount of heat flow required to increase the temperature of an experimental sample against a reference sample. Based on changes in the required heat flow, exo- and endothermic processes including phase transitions, glass transition temperature (T_g), crystallization, and degradation can be detected in silk proteins. DSC does not give a direct measurement of secondary structure, however, information on protein conformation can be inferred from changes in heat capacity.^[55,106,107] In silk fibroin containing bound water, an early T_g is observed around 80°C, and an endothermic peak around 100°C can be attributed to the evaporation of water from the silk matrix. In pure silk fibroin, the T_g occurs at 178°C,^[107–109] a crystallization peak is observed at 213°C,^[55,88,109] and thermal degradation starts around 230°C.^[55,107,109]

Temperature-modulated DSC (TMDSC) has been reported to be particularly useful in the study of silk-water interactions. TMDSC can be used to study both forward and reverse heat flow, thus elucidating the effects of non-reversible events, such as water removal from silk samples. In one example, a cyclic TMDSC procedure carried out below the T_g, effectively removed all bound water from a silk fibroin sample without inducing crystallization or degradation, thus enabling pure silk fibroin to be subsequently studied.^[106] The presence of bound water has been shown to act as a plasticizer and depress the T_g of silk fibroin,^[106,108] and therefore the ability to control bound water content is essential to obtain meaningful results from thermal analysis. Further work correlated trends in heat flow from DSC to secondary structure in silk fibroin via FTIR analysis. No β-sheet formation reportedly occurred below the T_g or during water removal, however, changes in the amide II region indicated a shift in the microenvironment of the protein chains. Above the T_g, shifts in the amide I region indicated the formation of β-sheets (Figure 9).^[55] Through this comparison of DSC and FTIR analysis, the T_g has been (negatively) correlated to the β-sheet fraction, and can therefore be used as a proxy for FTIR for analysis of silk fibroin crystallinity.^[88]

Figure 9:

DSC scans from −65°C to 280°C of B. mori silk fibroin films (a-d) with water (5.5%) and (e) without water. Scans were run at rates of (a) 20 K/min, (b) 10 K/min, (c) 5 K/min, and (d-e) 2 K/min. Sample (e) shows a T_g (2) of typical pure silk fibroin of 178°C and a crystallization peak, (2), at 230°C. A lower glass transition, T_g (1), is observed in samples containing water. The y-axis is the normalized heat flow, measured with respect to a reference sample and normalized by sample mass. Positive and negative peaks indicate exothermic and endothermic events, respectively. Reproduced with permission.^[55] Copyright 2008, American Chemical Society.

DSC is a rapid technique that requires minimal sample preparation and a small amount (<30 mg) of solid analyte. A limitation of using DSC for silk characterization is that protein degradation takes place over a wide temperature range, obscuring the melting temperature (T_m) peak. This is due to the fact that the intramolecular bonds in the protein start to degrade before the intermolecular hydrogen bonds in the β-sheet crystals, and consequentially the melting of the crystals is obscured by the degradation endotherm.^[106,110] Fast scanning calorimetry (FSC) can be used to overcome these limitations.

3.2. Fast Scanning Calorimetry (FSC)

FSC is a relatively new technique, and utilizes fast scanning chip calorimeters to heat samples at rates up to 1 MK/s, significantly faster than traditional DSC rates of 1 to 20 K/min.^[111] While β-sheet crystal melt cannot be distinguished from protein degradation in DSC analysis, FSC run at a rate of 2000 K/s has been shown to overcome this limitation. The relatively short resonance time at which the high temperatures are held in FSC limits protein degradation, allowing crystal melting and glass transition to be observed.^[110] Through use of this technique, it was demonstrated that β-sheet crystals can melt and then rearrange into random coil, helices, and turns from heat input alone.^[110] In another example, silk I and silk II crystals in silk fibroin films were studied by FSC, and results revealed that the crystals had distinct melting temperatures (T_m(Silk I) = 292 ± 3.8°C, T_m(Silk II) = 351 ± 2.6°C) (Figure 10).^[111] FSC requires thin (<10 µm) film samples to work with the fast heating rates.^[112] Limitations associated with FSC include significant data processing, and an open experimental setup where heat is lost throughout the process.

Figure 10:

FSC scans at 2000 K/min of silk films (a) annealed at 25°C for 12 hr (silk I), (b) annealed at 37°C for 12 hr, (c) exposed to 50% MeOH for 15 hr, and (d) autoclaved for 1 hr (silk II). Samples (b) and (c) are expected to have mixed silk I and silk II form. The melting point of silk I crystals (a) is lower than that of silk II crystals (b). The heat flow is normalized by the sample mass (y-axis) and reported with respect to temperature (x-axis). Positive and negative peaks indicate exothermic and endothermic events, respectively. Reproduced with permission.^[111] Copyright 2017, Elsevier.

3.3. Thermogravimetric Analysis (TGA)

In TGA, the mass of a sample is measured continuously as the temperature is increased at a constant rate. In studying silk proteins, this technique has been useful to quantitatively evaluate changes in water content as temperature is increased. While not a direct measurement of secondary structural elements, differences in thermogravimetric behavior can indicate differences in the secondary structure of silk proteins.^{[106,107,113,114]} Water-protein interactions play an integral role in the conformational mobility of silk, since bound water can act as a plasticizer,^[55,106,108] and therefore measuring water content through TGA is often of interest when characterizing silks.

In one example, three different silk fibroin films were analyzed by TGA: (1) as cast at room temperature, (2) cast at room temperature and methanol treated, and (3) cast at 80°C. By FTIR analysis, the as cast samples contained no β-sheet, while the 80°C and methanol treated samples had a higher proportion of β-sheet. TGA curves showed that the as cast films had a faster rate of water loss and greater overall water loss than both 80°C films and methanol treated films, due to the fact that the as cast films had lower crystallinity, greater mobility, and higher water content.^[107] In another example, TGA was used to study the interactions between water, silk, and glycerol plasticizer in silk films. Results revealed that the silk films caused a shift in the water and glycerol evaporation peaks, suggesting that the silk provides some stabilization (Figure 11).^[50] TGA analysis has also been used in the study of spider silk supercontraction, a phenomena where fibers contract up to 50% of their length upon exposure to water.^[115,116] Dragline silk from N. clavipes spiders was tested both in the native form and after supercontraction in water; based on TGA results, it was concluded that supercontraction altered the molecular organization of the fibers.^[117,118]

Figure 11:

TGA study of the effect of glycerol plasticizer on silk films. Mass loss (solid lines) and derivative mass change (dashed lines) are reported (y-axes) as silk film samples were heated at a rate of 10°C/min from 25 to 800°C (x-axis). Methanol post-treatment appears to have removed glycerol from the films, as indicated by the similar curves (blue and green). Silk films showed two peaks (water evaporation and degradation), while silk/glycerol films showed three phases (water evaporation, glycerol evaporation, and degradation). The evaporation peak was shifted in the silk/glycerol samples compared with glycerol only, likely due to increased stability of water and glycerol in the silk matrix. Reproduced with permission.^[50] Copyright 2016, American Chemical Society.

TGA analysis is useful for quantifying water content in silk proteins, a parameter that has significant implications on structure, and for probing thermal stability. It is a relatively rapid method that requires small amounts of solid sample (<30 mg). However, it cannot be used to identify or quantify secondary structures, and is therefore best used as a compliment to additional analysis techniques.

4. Conclusions and Future Directions

A combination of analytical techniques can be used to study the secondary structure of silk proteins, a critical feature that defines the stability and strength of silk-based materials. High resolution techniques like ssNMR and solution NMR can provide information on local structural changes at the atomic level, while FTIR can be used to quantify the relative abundance of different structural elements (α-helix, β-sheet, turns) in the bulk material. Similarly, Raman and CD can also be used to determine shifts in the secondary structure of the bulk protein, such as a change from random coil to β-sheet as self-assembly occurs. X-ray scattering can be used to estimate crystal size, crystallinity, and molecular alignment in solid silk samples. Glass transition, crystallization, and degradation temperature can be determined by DSC, and used as a proxy to estimate crystallinity and structure. TGA has been reported as a complimentary technique to other analysis tools for quantifying water content and degradation temperature, which has a significant influence on the material properties of silks. With a fundamental understanding of the assumptions associated with each technique, the type of data that is attainable (qualitative vs. quantitative, local vs. bulk), and consideration of the sample conditions (solution, solid state, dilute, concentrated), meaningful conclusions can be drawn about the secondary structure of silk proteins by applying a suite of the techniques discussed here.

As demonstrated by advancements like the development of FSC to overcome limitations of DSC,^[110] the addition of polarization and microscopy to Raman to collect orientation data and improve sample resolution, and the development of an in vitro method to monitor the conversion from dope to fiber by NMR,^[33] we believe that this field will continue to grow.

Improvements to deconvolution analysis techniques and peak assignments for FTIR, Raman, and CD could be facilitated in the future by performing fundamental studies on model fibrous proteins, eliminating the need to rely on data from globular proteins. It is anticipated that a number of additional techniques will be applied to investigate the silk proteins in solution (dope) and in solid fibers, and the conformational conversions that occur between the two states. In solution, the advent and arrival of cryo-electron microscopy (cryo-EM) should be particularly advantageous to investigate silk protein assemblies and higher order structure in vitro.^[119] Recent results with magnetic resonance imaging (MRI) indicate that monitoring the silk spinning process in vivo is indeed possible on live organisms, although considerable development of this method is still required before this becomes reality.^[42]

Continued advancements to silk protein analysis and effective utilization of a suite of analysis techniques will be essential to both early stage research as well as the development of commercially viable products as the field of advanced, silk-based materials continues to grow.

Figure 12:

Summary of experimental techniques for studying the secondary structure conformations of silk proteins, grouped by results output (crystallinity, local structure, bulk structure, and thermal analysis) and sample type (solid state, dark purple, and liquid state, light purple).

Table 1:

A combination of techniques can be applied to analyze silk samples by taking into consideration the sample format, intended outputs, and the advantages and limitations associated with each technique

Technique	Sample requirements	Output	Advantages	Limitations	References
Solid state NMR (ssNMR)	Solid state sample; Istopically labeled for certain techniques	Local conformation, atomic-scale structure and dynamics	High resolution; study both crystalline and amorphous regions; study individual amino acids	Technically difficult analysis; significant sample prep for isotopic labeling and large sample size; local but not broad information	[20,38,40,46,47,51,57,59,121]
Solution NMR	Solution state sample; Istopically labeled for certain techniques	Local conformation, atomic scale structures and dynamics	High resolution site-specific structure; study protein backbone dynamics	Technically difficult and time consuming data analysis; significant sample prep for isotopic labeling	[33–36,65,67]
Circular dichroism (CD)	Solution state sample	Relative fraction of structural elements (α-helix, β-sheet, β-turn, random coil) based on peak shifts	Rapid analysis; small sample size required	Limited to dilute, solution; qualitative data; low resolution of β-form structures	[71,73–75]
X-ray scattering	Solid state sample (WAXS)	Crystallinity; calculated crystal size and molecular orientation	Rapid analysis; assess crystalline region	Limited to analysis of crystalline structures; low resolution of small/irregular crystals	[83–86]
Raman spectroscopy	Solid or solution state sample	Secondary structure elements (α-helix, β-sheet, β-turn, random coil) based on peak shifts; chain orientation (polarized Raman)	Small sample requirement; flexible sample format; measure aqueous samples without interference	Background interference; low signal; accurate peak fitting and quantification is still challenging	[37,39,90,92,93,95,122]
Fourier transform infrared spectroscopy (FTIR)	Solid or solution (deuterated water) state sample	Secondary structure elements (α-helix, β-sheet, β-turn, random coil) based on peaks shifts; calculate quantitative fractions (with FSD)	Small sample requirement; flexible sample format	Deuterated samples required for solution state analysis; cannot discern silk I crystals from silk II crystals; accurate peak fitting and quantification is still challenging	[88,90,100–102]
Differential scanning calorimetry (DSC)	Solid state sample	Glass transition temperature (Tg), crystallization peak, degradation temperature, heat capacity	Measure thermal events (Tg, phase transitions), thermal stability	Cannot directly measure or identify secondary structure elements; mass loss due to water removal affects results	[55,106,107]
Fast scanning calorimetry (FSC)	Solid state sample, thin film (<10 µm)	Same as DSC, also melting temperature (Tm)	Measure Tm, (unattainable by DSC)	Cannot directly measure or identify secondary structure elements; significant data analysis	[110–112]
Thermo-gravimetric analysis (TGA)	Solid state sample	Mass loss with increased temperature; thermal stability	Quantify water content in silk samples	Cannot directly measure or identify secondary structure elements	[106,107,113,114,117]