Tyrosine Templating in the Self-Assembly and Crystallization of Silk

Oct 13, 2016 - Native silk fibers exhibit strength and toughness that rival those of the best synthetic fibers. Despite significant research, further ...
15 downloads 7 Views 4MB Size
Download PDF
Article pubs.acs.org/Biomac

Tyrosine Templating in the Self-Assembly and Crystallization of Silk Fibroin Benjamin P. Partlow,†,§ Mehran Bagheri,‡,§ James L. Harden,*,‡ and David L. Kaplan*,† †

Department of Biomedical Engineering, Tufts University, 4 Colby Street Medford, Massachusetts 02155, United States Department of Physics, University of Ottawa, 338L MacDonald Hall, 150 Louis Pasteur Ottawa, Ontario K1N 6N5, Canada

‡

ABSTRACT: Native silk fibers exhibit strength and toughness that rival those of the best synthetic fibers. Despite significant research, further insight is still needed to understand the mechanisms by which silkworms are capable of spinning such tough fibers. Here we propose that π−π and π−OH group interactions of tyrosine side chains provide templating effects, such that the crystal-forming domains are in registration, thereby fostering the self-assembly of the spinning dope. Intrinsic fluorescence measurements, in conjunction with circular dichroism, showed that during self-assembly of regenerated silk solutions, the tyrosine residues were localized in a more hydrophobic local environment, suggesting preferential assembly. In situ Fourier transform infrared spectroscopy indicated that cross-linking of the tyrosine residues resulted in the development of extended β-sheet structure. Additionally, control of cross-link density directly influenced the degree of crystallinity upon drying. Molecular dynamics simulations were performed on silk mimetic peptides in order to more thoroughly understand the role of tyrosines. The results indicated that tyrosine residues tended to transiently colocate in solution due to π−π interactions and hydrogen bonds with adjacent residues and with the peptide backbone. These more stable tyrosine interactions resulted in reduced lateral chain fluctuations and increased incidence of coordinated intrachain association, while introduction of a dityrosine bond directly promoted the formation of β-sheet structures. In total, the experimental and modeling data support a critical role for tyrosine-tyrosine interactions as a key early feature in the self-assembly of regenerated silk protein chains and therefore are important in the robust and unusual mechanical properties ultimately achieved in the process.

■

to form globules and are elongated and form fibrils with the application of shear or elongational flow.6 Atomic force microscopy measurements showed that modulating concentration and temperature during self-assembly of regenerated silk solutions resulted in the development of protofibrils at low concentration while forming larger micelle-like structures at high concentration.7,8 Other studies have focused on individual repetitive motifs found in the protein sequence. The crystalline domain consists of hexameric repeats of GAGAGS, and the resultant β-sheet crystals have been studied extensively through X-ray and NMR methods to obtain insights into the spinning process.9−12 The highly conserved, hydrophilic linker region has been synthetically produced and analyzed as well as modeled and has been found to result in a 180° turn in the backbone of the silk molecule.13 The GY ∼ GY motif has also been studied both experimentally and empirically, but a definitive role has not been determined.14−18 In addition to protein sequence, silkworms and spiders employ a variety of physical and chemical means to convert the random coil silk fibroin spinning dope into a semicrystalline, highly oriented fiber. Included in the process are applied shear, physical drawing, dehydration, changes in salt concentrations,

INTRODUCTION Silk fibers from silkworms have been utilized for thousands of years due to their combination of strength, extensibility and toughness.1 Despite being spun by a caterpillar, with mulberry leaves as the feedstock, they are able to rival high performance synthetic fibers in their mechanical properties. While a significant amount of research has been conducted to determine the mechanisms of spinning, further insight is needed.2,3 Due to incomplete knowledge of the complex spinning mechanisms, the toughness achieved in naturally spun silk fibers can still not be fully explained, nor has the highly developed evolutionary process been successfully exploited to improve manmade fibers derived from these proteins. One area that has been incompletely explained is the interaction of individual amino acid residues in the protein sequence and how these may enhance the spinning process. The role of the protein sequence has been explored for its importance in the spinning process. The silk fibroin sequence consists of three distinct motifs: 11 hydrophilic linker domains, crystalline β-sheet domains comprised of glycine, alanine and serine, and regions rich in tyrosine residues. One hypothesis that has been proposed is that the hydropathy of the fibroin protein results in micelle formation.2,4,5 Specifically, the hydrophilic linker regions and N- and C- termini migrate to the outside, in contact with water, shielding the hydrophobic regions inside. It was proposed that these micelles then interact © XXXX American Chemical Society

Received: July 19, 2016 Revised: October 11, 2016

A

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules decreases in pH, and the introduction of metal ions.19 The shear sensitivity and rheological properties of silk fibroin solutions have been assessed, and significant differences have been found between native spinning dopes and those regenerated from cocoons.20−24 However, the degree of degradation during the sericin removal and solubilization process were not addressed, which have been shown to be important considerations.25 The species and concentrations of different salts and metal ions have also been shown to vary significantly along the length of the storage glands and spinning ducts in the silkworm. These factors influenced the transition from random coil protein chains to β-sheet crystalline forms of silk. The conformational transitions induced by the addition of salts also significantly changed the rheological properties of the solutions, suggesting they are critical to efficient spinning.26−29 The enzyme alkaline phosphatase was present in the lining of the spinning duct and results in a pH drop from the storage gland to the spinneret of 8.2 to 6.2, respectively.30 This pH drop has been shown previously to impact the secondary structure and rheology of the silk solutions, as is also the case with changes in salt concentration.31−33 While the role of the GY ∼ GY domain in Bombyx mori silk has not been definitively determined, the chemical reactivity of the phenolic side chain of tyrosine has been shown to be important in the spinning of other silks. The presence of covalent dityrosine bonds was been found in Tussah silk fibroin as well as marine silks from caddisfly larvae and sandcastle worms.34−36 These marine organisms also utilize an intermediate form, 3,4-dihydroxyphenylalanine (DOPA), as an adhesive site to bind dissimilar materials, which is capable of curing under aqueous conditions.37,38 Recently, dragline silks from several species of spiders were found to contain appreciable quantities of dityrosine.39 Peroxidases, necessary for the formation of dityrosine and DOPA, have also been reported in the spinneret of the Nephila clavipes spider, and it is suggested that they may result in cross-linking during the spinning process.40 Recently, silk hydrogels have been formed by covalently cross-linking tyrosine residues, in regenerated silk solutions, via an enzymatic reaction of horseradish peroxidase with hydrogen peroxide as the activator. Dityrosine formation was confirmed by the appearance of a strong fluorescence at excitation and emission wavelengths of 315 and 415 nm, respectively. In the hydrated state, the cross-linked silk proteins exhibited elasticity and variable mechanics based on the cross-link density and solution molecular weight.41 However, when dried, the hydrogels spontaneously crystallized, resulting in significant βsheet content without post treatments or additives normally required to crystallize other silk formulations. This phenomenon suggested that the tyrosine residues were playing a key role in the crystallization of the fibroin molecules, perhaps through a preorganization step facilitated by the enzymatic cross-linking process. Thus, the objective of the present study was to focus on the role of the tyrosines in the regenerated silk fibroin chains related to registry and control of the assembly process. While covalently bound dityrosine has not been isolated from B. mori silk fibers, aromatic residues are known to form noncovalent π−π interactions. These interactions play important functional roles in numerous materials including porphyrin aggregation,42 the strength of Kevlar,43 and the stability of the DNA double helix.44 In proteins, interactions between aromatic amino acids play a role in stabilizing multiple structural elements,45

including the hydrophobic cores of globular proteins,46,47 helical bundles,48 and β-hairpins.49 The interactions of aromatic residues are specific and analysis of proteins found in the protein data bank shows that aromatic residues often occur in clusters, with more than two interacting aromatic groups.50 This finding may point to the significance of the role of aromatic amino acids in protein structure, as clusters of aromatic groups have a much higher capacity to stabilize structures.51 Therefore, clarification of the role of the aromatic side chains of tyrosine residues in the silk fibroin sequence and during processing, an explanation for their repetitive spacing, and a greater understanding of the spinning process will serve several critical purposes.

■

EXPERIMENTAL SECTION Silk Solution Processing. Silk cocoons (Tajima Shoji, Yokohama, Japan) from the B. mori silkworm were regenerated into an aqueous fibroin solution as described previously.52 Briefly, the cocoons (Tajima Shoji, Yokohama, Japan) were cut into pieces, and the silkworm was removed. Five grams of cocoon pieces were added to 2 L of boiling deionized water with 0.02 M sodium carbonate (Sigma-Aldrich, St. Louis, MO) for 30 min to remove the sericin protein. After extraction, the fibers were rinsed with deionized (DI) water and allowed to dry overnight. The remaining pure fibroin fibers were then dissolved in a 9.3 M solution of lithium bromide (SigmaAldrich, St. Louis, MO) at 15 wt % and placed in a 60 °C oven for 4 h. The resultant solution was injected into 3500 MW cutoff dialysis cassette (ThermoFisher Scientific, Waltham, MA) and dialyzed against DI water for 48 h to remove the lithium and bromide from solution. Following dialysis, the solution was centrifuged for 20 min at 12 500g to remove any precipitates or debris. The final solutions were adjusted to a concentration of 50 mg/mL, frozen at −80 °C, and lyophilized. The dried silk fibroin (hereafter referred to as silk) was dissolved overnight in deuterated water at a concentration of 50 mg/mL and centrifuged to remove any residual fragments. Solutions were diluted in deuterated water to 20 and 40 mg/ mL for circular dichroism and Fourier transform infrared spectroscopy (FTIR) analyses, respectively. Tyrosine Cross-Linking. Horseradish peroxidase (HRP), type VI (Sigma-Aldrich, St. Louis, MO) lyophilized powder was mixed with deionized water to form a stock solution with a concentration of 1000 U/mL. The HRP solution was added to the silk solution in a ratio of 10 units of HRP to 1 mL of silk solution. To initiate gelation, 10 μL of hydrogen peroxide (Sigma-Aldrich, St. Louis, MO) solution were added per milliliter of silk solution. Stock hydrogen peroxide concentration was prepared at 100 times of that desired for a final concentration.41 Circular Dichroism (CD). CD spectra were collected to determine the conformation of the silk as a function of sample age, as defined as the time elapsed since solution preparation. Spectra were collected using an AVIV Biomedical Model 410 CD spectrometer (Lakewood, NJ) as described previously.41 Solutions were loaded into a 10 μm path-length demountable cuvette and placed in a temperature controlled cell holder set at 25 °C. Three wavelength scans were collected between 180 and 260 nm using 0.5 nm steps. Signal intensity was averaged and smoothed using a 9 point Savitzky-Golay smoothing algorithm. Fluorescence Spectroscopy. The intrinsic fluorescence of the native tyrosine and tryptophan residues were monitored as the silk solutions were aged, to detect differences in the local B

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules environment of the fluorescent amino acids.53 Fluorescence spectra were obtained using a Hitachi fluorescence spectrophotometer (F4500, 450 W xenon lamp; San Jose, CA). Solutions were loaded in a 1 cm path length quartz cuvette. The solutions were excited at 280 nm and the emission was recorded from 295 to 400 nm range in 1 nm increments. The scan speed was 1200 nm/min with excitation and emission slits set to yield a 2.5 nm resolution. Measurement of fluorescence intensity as an assessment of dityrosine cross-link density was performed using a VersaMax microplate reader (Molecular Devices, Sunnyvale, CA). Aliquots (100 μL), of silk solution, with cross-linking agents mixed in, were pipetted into an opaque 96 well plate and allowed to gel for 12 h. Evaporation was minimized by filling unused wells with deionized water and tightly sealing the plate with parafilm. End point fluorescent measurements were taken at an excitation-emission of 315 and 415 nm, and background fluorescence was subtracted. Fourier Transform Infrared Spectroscopy Analysis. Conformational content of dried hydrogels was analyzed via a JASCO FTIR 6200 spectrometer (JASCO, Tokyo, Japan) combined with a MIRacle attenuated total reflection (ATR) germanium crystal. Hydrogels were dried in a laminar flow hood for 3 days in order to fully dehydrate the samples. For each sample, 64 scans were coadded with a resolution of 4 cm−1, at wave numbers between 600 and 4000 cm−1. The background spectra were collected under the same conditions and subtracted from the scan for each sample. Fourier selfdeconvolution (FSD) of the infrared spectra covering the Amide I region (1595−1705 cm−1) was performed using Opus 5.0 software (Bruker Optics Corp., Billerica, MA), as described previously.54 The deconvoluted amide I spectra were areanormalized, and the relative contributions of the individual bands were used to determine the content off the secondary structures. Development of secondary structures immediately following gelation were determined by performing in situ FTIR of deuterated silk solutions. The gel precursor was gently mixed and pipetted onto the FTIR crystal using a liquid retainer. Low viscosity mineral oil was added on top of the solution to prevent evaporation during the course of the experiment. The background spectra was collected prior to loading the sample and subtracted from each measurement. For each time point, 32 scans were coadded with a resolution of 4 cm−1, at wave numbers between 600 and 4000 cm−1. After collection, each time point was peak normalized, using the amide I band, to compare relative intensities as a function of time post crosslinking. Molecular Dynamics Simulation. To assess the role of tyrosines at the molecular level, simulations were conducted on two sets of short silk-mimetic peptide chains, one set to assess the propensity of tyrosines for self-assembly of transient interchain contacts, and a second set to investigate the role of tyrosines for stabilizing the formation of β-sheet secondary structure. Both sets of peptide sequences were selected based on the highly repeated motifs in silk, [GAGAGS] and [GAGY] and included control sequences that lack tyrosines. The peptide sequences for interchain tyrosine contact studies were:

Note that in the second, control sequence, serine residues replace the tyrosine residues at the terminal ends and center of the sequence. The propensity for aromatic and peptide− peptide interactions was assessed and compared as a function of concentration for the Silk-Y and the Silk-S variants. To do so, initial conformations for both sequences were constructed in disordered, fully extended structures, using a custom designed Python code. In order to ascertain appropriate initial conditions for multichain studies, initially generated sequences were individually simulated for 100 ns, and multiple conformations from the last 50 ns of the simulation were selected as initial chain conformations for subsequent simulations. Three sets of multichain simulations of the mimetic peptides were then conducted at concentrations of 3 and 9 wt % for 200 ns. In the multichain simulations, an interchain association event between selected residues was defined by spatial separation within a chosen cutoff distance. Specifically, a tyrosine-tyrosine association event was noted when the distance between aromatic ring centers was within 6 Å, whereas a serine−serine association required a distance between side chain centers of mass that was less than 5 Å. In these simulations, tyrosine−tyrosine and serine−serine association behavior was characterized by the frequencies and lifetimes of these association events. Association lifetimes were calculated using a transition threshold time of 0.1 ns and a total sampling time of 50 ns. Next, the influence of the tyrosine residues on the stability of β-sheet secondary structure was assessed using two complementary peptide sequences with tyrosines located exclusively in the middle of the peptide (in order to monitor the propagation of β-sheet structure from the central core regions). The two complementary sequences utilized were as follows: Silk-mY1: GAGAGSGAG Y GAG Y GAGAGSGAG Silk-mY2: GAGAGSG Y GAG Y GAGAGAGSGAG These peptides were used to construct antiparallel β-strands with the tyrosine from one strand adjacent to those in the other strand. Two of these antiparallel β-structures were then positioned with the glycine residues facing each other to form an initial four-chain β-sheet bundle conformation for further simulation. To investigate the effects of tyrosine crosslinks, a second version of this four-chain β-sheet bundle was prepared in which nearest-neighbor tyrosine residues were permanently connected to form intermolecular dityrosine moieties. Finally, as a control, an analogous four-chain βsheet bundle was constructed using a modified complementary peptide sequence pair (Silk-mS1and Silk-mS2) that had all tyrosine residues in Silk-mY1 and Silk-mY2 replaced by serines. Starting from these three types of initial four-chain β-sheet bundle conformations, six sets of simulations were run for 25 ns each. A small tension of 0.1 KT/Å was applied to the chain ends during the simulation, in order to mimic the existence of a longer protein sequence in which this repetitive motif is embedded (i.e., the ends of the short peptides sequences are not true chain ends). The root mean squared displacements (RMSD) of the peptide backbone atoms, relative to those in an aligned ideal β-sheet bundle, were monitored using the VMD package to determine steady-state conditions. The β-sheet content was measured in verified steady-state conditions, over a 15 ns sample time, using the STRIDE algorithm55 implemented in VMD,56 with the criteria that residues with β-compatible dihedral angles and H-bonds in either extended or isolated bridge conformations count in the β-structure content calculation.

Silk-Y: Y GAGAGSGAGAG Y GAGAGSGAGAG Y Silk-S: S GAGAGSGAGAG S GAGAGSGAGAG S C

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 1. Fluorescence and structure during solution aging. (A) Intrinsic fluorescence of the silk solution over time indicates that tyrosine residues, with an emission of ∼305 nm are dominant immediately after dissolution. However, during the course of self-assembly, tryptophan fluorescence overwhelms that of the tyrosine. (B) CD spectra reveal a predominantly random coil structure in fresh solution that slowly transitions to a β-sheet rich gel over time.

Figure 2. Time-dependent conformational changes following tyrosine cross-linking. Immediately following and up to 70 min after cross-linking of the tyrosine residues, significant changes are seen in the secondary structure of the silk. (A) Changes in the amide I region indicate an increase in bands associated with β-sheet and crystalline silk I conformations. (B) Intensity changes in the amide II region suggest similar movement toward a crystalline conformation, containing elements of both crystalline silk I as well as β-sheet (silk II).

around ∼305 nm at early time points, while a shoulder centered on ∼340 nm progressively increased, until overwhelming the 305 nm emission (Figure 1a). The CD spectra of fresh solutions revealed random coil conformations with a characteristic minima near 195 nm. As expected, the β-sheet content increased with increasing aging time, as signified by the magnitude of the depression at 218 nm (Figure 1b). The correlation between secondary structure and the fluorescence of silk was previously reported in the fresh solution state and the β-sheet-rich gelled state.62 These results show similar starting and ending values, but show that both develop gradually as the silk molecules self-assembled. The fluorescence emission at 305 nm is characteristic of tyrosine, while that at 345 nm is from tryptophan residues. As the relative concentration of these amino acids is constant, the changes in intensity suggested either an energy transfer phenomenon or, more likely, a change in local environment of the tyrosine residues. In a recent study,63 Raman analysis of regenerated silk solutions during assembly and aging, following a sharp reduction in pH, revealed significant changes in the local environment. The tyrosine residues were found in a hydrophilic environment in the random coil regions present in

In all simulations, peptide chains were solvated within a water box, using periodic boundary conditions. The simulation box was extended to maintain a minimum of 15 Å between the peptide atoms and the boundary of the simulation box at all times. All MD simulations employed a Langevin thermostat and barostat and were conducted in pure water under NPT conditions at T = 300 K and P = 1 atm. The CHARMM36 force field57 with the protein backbone improvement58 and dihedral angle optimization59 was adopted for the peptides and the TIP3P model was used for the water molecules.60 The van der Waals (vdW) interaction cutoff was 12 Å and electrostatic interactions were computed with the particle-mesh Ewald (PME) algorithm. The time step was set at 2 fs, and all atom simulations were performed using NAMD (release 2.10).61

■

RESULTS AND DISCUSSION Secondary Structure and Tyrosine Fluorescence during Self-Assembly. In order to assess the interactions of the tyrosine residues during self-assembly, the intrinsic fluorescence and secondary structure of a regenerated solution were recorded at days 0, 3, 6, and 10 postdialysis. The shape of the fluorescence emission spectra changed with a maxima D

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 3. Effect of cross-link density on crystallinity. Degree of tyrosine cross-linking effects the β-sheet content following dehydration. (A) Intensity of dityrosine fluorescence, a measure of the number of residues involved in bonding, is a function of the amount of peroxide used in the reaction. (B) FTIR spectra of the dried hydrogel show a relationship between peroxide concentration and therefore a cross-link density and crystallinity. (C) Representative deconvolution of an FTIR spectra to quantify the crystalline content. (D) Relationship between the relative fluorescence and β-sheet content clear show that maximizing cross-linking results in higher crystallinity.

centered on 1652 cm−1 to a crystalline silk I conformation.64 Given that cross-linking should induce a greater degree of order, we attribute this to silk I here. There is also a discrepancy in the designation in the bands at 1618, 1627, and 1695 cm−1. These bands are traditionally identified as β-sheet; however, evaluation of a degraded fraction of silk fibroin, known to not contain any β-sheet content, had significant contributions at these wavenumbers. Based on this data, it was presumed that these bands may also be associated with short, irregular segments of β-turns, but lack the extended structure of a βsheet.65 Based on CD data, the development of β-sheet following cross-linking requires a longer duration than the current FTIR experiment.41 This suggests that the development of bands in these wavenumbers is characteristic of shorter fragments of silk I that coalesce to eventually transition to a silk II secondary structure. Similar conformational changes are seen in the amide II band with development of peaks at 1555 and 1543 cm−1, these have been assigned to helical and turn structures and silk I, respectively.66 The development of these conformations is in general agreement with the known transitions from silk I to silk II during silk spinning. While the exact crystal structure of silk I is inconclusive, the data here show that selective cross-linking of the tyrosine residues can facilitate these transitions.

the initial stages of assembly. As the content of β-structure increased, the local environment became more hydrophobic, suggesting the formation of specific quaternary structures that result in colocalization of the tyrosine residues. The specificity of the local environment surrounding tyrosine residues, their distribution throughout the sequence, and adjacent glycine residues, suggests that their position is important in the coordinated assembly of extended β-sheet structure and is not random. In Situ Development of Secondary Structure after Cross-Linking. To validate the hypothesis that tyrosine residues play a direct role in guiding the development of secondary (and higher order) structures in silk proteins, these residues were enzymatically cross-linked and the resulting material monitored for conformational changes. In order to actively monitor the development of higher order structure, the silk was cross-linked in situ on the FTIR-ATR crystal, and spectra were collected every 5 min. The amide I and amide II changes are shown in Figure 2a,b, respectively. Based on an analysis of peptides with different substitutions using IR spectroscopy, specific bands have been identified to distinguish between random, helical, and the crystalline conformations of silk I and silk II. In the amide I band, the region between 1637 and 1655 cm−1 is normally assigned to a random coil conformation.54 However, others have assigned the peak E

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules Cross-Link-Dependent Crystallinity. The rapid formation of secondary structures following cross-linking of the tyrosine residues suggested that selectively controlling the degree of cross-linking may result in different degrees of crystallinity following dehydration. The degree of cross-linking can be controlled by adjusting the hydrogen peroxide added to initiate the enzymatic reaction. This cross-linking efficiency can then be determined by measuring the intensity of the dityrosine fluorescence. As seen in Figure 3a, the fluorescence intensity was directly related to the peroxide concentration, with a maximum occurring at a final peroxide concentration of around 2.0 mM. This peak cross-linking efficiency is consistent with literature that has shown that excess hydrogen peroxide can have inhibitory effects on the reaction.67 As with the fluorescence intensity, following dehydration, the FTIR spectra were related to the concentration of peroxide added to the reaction (Figure 3b). At the extremes of peroxide concentration, the spectra indicated negligible crystalline content, similar to the control dried silk solution. However, as the peroxide levels approach the optimum value and the associated cross-linking increased toward a maximum, the β-sheet content also increased. Deconvolution of the spectra allowed for quantitative analysis of the extended β sheet content by comparing the area under the curve of the spectral components; a representative deconvolution for a 1.64 mM sample is shown in Figure 3c. Normalization of the fluorescence intensity allowed for a comparison of the crosslink density versus crystallinity. As indicated in Figure 3d, dehydration-induced crystallinity corresponded to the initial cross-link density, with higher cross-linking resulting in higher β-sheet content following dehydration. Numerous methods for inducing β-sheet in silk have been explored, including treatment with alcohols, plasticization through water annealing, autoclaving52 and isothermal heating above the Tg, among others.68 Additionally, reduction of pH, addition of ions or small molecules and the application of energy (shear, vortex, sonication) are known to accelerate the self-assembly process, leading to a more rapid transition to βsheet.52 While each of these methods has been helpful in determining the susceptibility to different perturbations, none of them provide information on the specific sites within the protein chain that are responsible for the structural changes. Residue specific cross-linking has been an important tool in the identification of the functional domains of active proteins.69 Tyrosine residues have been found at the edges of microcrystalline regions in B. mori silk,17 and this supports the fact that tyrosine colocation is critical in the development of increasingly ordered secondary structure. Evidence of a “super”secondary structure70 and an average crystallite size incorporating 35 residues71 also necessitates the tyrosine residues be consistently located at the edges. The bulky tyrosine side chain is incompatible with the tightly packed β-sheet domains and an analysis of the protein sequence shows that the large crystal size can only occur if the tyrosine residues are localized at domain boundaries and not randomly distributed. This suggests that these residues are active participants, most likely through aromatic interactions, in the self-assembly of the fibroin protein. Tyrosine Colocation in Solution. Empirical evidence indicates that tyrosine residues are involved in the self-assembly of reconstituted fibroin chains and that forced coupling (via dityrosine bonding) results in changes in secondary structure and dehydration induced β-sheet formation in fibroin solutions. However, the ability to probe the self-assembly process at the

atomistic level is limited experimentally. Molecular dynamics simulation allows for a high resolution evaluation of how substitutions in protein sequence can stabilize or destabilize protein structures. Simulation of silk-like peptides with and without tyrosine residues indicated that the presence of tyrosine residues plays a stabilizing role in transient chain assembly in a concentration-dependent manner. As shown in Table 1, at a weight fraction of 3%, the presence of tyrosines in Table 1. Silk Mimetic Peptide Association in Solutiona low concentration percentage of residues associated total association time per selected residue number of association longer than 0.5 ns

high concentration

Silk-Y

Silk-S

Silk-Y

Silk-S

22 ± 10%

8 ± 2%

44 ± 7%

12 ± 4%

2.1 ± 1.2 ns 7±6

0.5 ± 0.3 ns 1±1

2.3 ± 0.3 ns 43 ± 11

0.8 ± 0.3 ns 12 ± 4

a

Results were obtained from data in the last 50 ns of simulation for low and high concentration solutions (3 and 9 wt %). For each configuration, the average and standard deviation is given for results of three independent trials.

the peptides slightly increased the percentage of total associated residues, defined as the ratio of residues that were within a cutoff separation (6 Å for tyrosines or 5 Å for serines in the control sequence) to the total number of residues, from 8 ± 2% for peptides with serines replacing tyrosines to 22 ± 10% for peptides that included tyrosine. Moreover, the average transient association time per residue pair also increased from an average of 0.5 ± 0.3 ns to 2.1 ± 1.2 ns for serine and tyrosine, respectively. Perhaps a more relevant indicator of transient association is the total number of residues that remain in the associated state for an extended time period. In the presence of tyrosines, there were 8 ± 1 tyrosine pairs that were associated for more than 0.5 ns in the 50 ns analysis period, whereas for the control sequence there were only 1 ± 1 serine pairs that were associated for more than 0.5 ns. These modest effects observed at low concentration were significantly enhanced in simulations conducted at a weight fraction of 9%: the increase in concentration resulted in a significantly higher percentage of transiently associated tyrosine residues, 44 ± 7%, but not a significantly longer average association time (2.3 ± 0.3 ns). There was also a substantial increase in residues with association lifetimes greater than 0.5 ns, with 43 ± 11 of such tyrosine pairs associated in the 50 ns analysis period. By contrast, the peptides incorporating serine replacements had a significantly lower propensity for association of 12 ± 4%, a lower average association time of 0.8 ± 0.3 ns, and a relatively smaller number of long-lived pairs, 12 ± 4. The enhanced transient association events in the tyrosine-containing silk-like peptides is due to favorable interactions between aromatic side chains (e.g., parallel π-stacking and T-stacking conformations) and hydrogen bonding between tyrosine side chains and the peptide backbone.72 Figure 4 illustrates several of these prototypical interactions. A representative snapshot of the final distribution of chains incorporating tyrosine residues is shown in Figure 4a, while specific instances of aromatic stacking and hydrogen bond formations involving tyrosine residues are shown in Figure 4b. Such interactions do not occur in the absence of tyrosine, and serve to facilitate tyrosine-mediated interactions between different chains and thereby act to enhance intrachain association and subsequent intrachain HF

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 5. Representative conformations of silk mimetic peptide bundles. Initial β-conformation and conformations taken near the end of simulation of four-chain silk mimetic peptide bundles. (A) Initial conformation of an ordered beta-sheet bundle composed of two antiparallel beta strands (yellow arrows) facing each other from the glycine side (white), in which tyrosine residues (green) are adjacent to each other. (B) Postsimulation conformation of a bundle with crosslinked tyrosine residues, showing stabilized beta-sheet structure (yellow arrows) in the central portion of the bundle. (C) Postsimulation conformation of a bundle with free tyrosine residues, showing relatively weaker, fluctuating beta-sheet structure (yellow arrows) in the central portion of the bundle.

Figure 4. Representative transient network structure of a 9 wt % solution of silk mimetic peptide containing tyrosine. Following simulation for 200 ns, (A) the tyrosine residues in the silk mimetic peptide chains tend to self-assemble into metastable clusters. (B) In addition to tyrosine π−π interactions, these residues are able to interact with each other (B1) or with the chain backbone (B2) via their −OH groups. Synergistic effects of multiple types of interactions could further enhance the stability of these associations.

bond formation. There also was a clear concentrationdependent increase in association for peptides with phenolic side chains that was not seen in the serine substituted group. This was consistent with findings with other proteins, where increasing aromatic groups resulted in a nonlinear increase in stability.50 β Sheet Formation in Cross-Linked Silk Peptides. There is disagreement as to the propensity for the tyrosine residues to stabilize or destabilize the formation of β-sheet secondary structure.73 To determine whether the incorporation of tyrosine in the peptides studied here was compatible with the hydrogen bonding necessary for crystallization of extended βsheet structure, several configurations were simulated. Complementary pairs of silk mimetic peptides (silk-mY series), with two tyrosines in the middle, were initially configured in an extended and disordered antiparallel conformation with tyrosines intermolecularly cross-linked and simulated for 100 ns. The cross-linked residues were found to facilitate coorientation of the two linked chains and to nucleate transient βstructure in the central portion of the antiparallel strand (data not shown). These results motivated us to conduct extensive simulation studies of the effects of tyrosines on stability of βsheet structure in these silk mimetic peptides. To further investigate the ability of tyrosines in the sequence (free or cross-linked) to stabilize β-sheet structure, simulations were conducted on four-chain β-sheet bundles comprised of two pairs of antiparallel β-strands positioned with the glycine residues facing each other (Figure 5a). Both sequences with free and cross-linked tyrosines were studied, as well as a control bundle with tyrosines replaced by serines (silk-mS series). In each case (cross-linked tyrosines, free tyrosines and the tyrosines replaced by serines), six sets of such assemblies were simulated for 25 ns to allow for equilibration. Representative conformations after simulation of bundles with cross-linked and free tyrosines are shown in Figure 5b,c, respectively. The time-dependent RMSD of the peptide backbone atoms relative to those in an aligned, ideal antiparallel β-sheet bundle converged to steady-state conditions within 10 ns of simulation time in all cases. Thus, β-content during the last 15 ns of simulation, when backbone atom RMSDs were in steady-state conditions, was calculated to characterize the

stability of the core structures. Quantitative results are given in Table 2, and a pictorial β-strand heat map, which demonstrates Table 2. Propensity for β-Sheet Formationa chain average core average

cross-linked Tyr

free Tyr

Ser substitution

37 ± 8% 72 ± 16%

20 ± 4% 35 ± 12%

17 ± 4% 22 ± 3%

a

Average beta content is given for all residues (chain average) and for the central five residues (core average) as determined from simulation data during last 15 ns of simulation for six independent trails of configurations with cross linked tyrosine, free tyrosine, and with serines substituted for tyrosines.

the propensity of individual amino acids to maintain stable βsheet conformations, is shown for each system in Figure 6. Assemblies with cross-linked tyrosine showed substantially more chain-averaged β content (37 ± 8%) than either the free tyrosine (20 ± 4%) or serine substituted (17 ± 4%) configurations. Moreover, this effect is significantly pronounced in the core region, comprised of the central five residues: Core residue average β content was found to be 72 ± 16% for configurations with cross-linked tyrosines, 35 ± 12% for free tyrosine configurations, and 22 ± 3% for the serine control configurations (with p = 0.069 for the comparison of the free Tyr and Ser cases). The central tyrosines, when intermolecularly cross-linked, strongly reduce lateral chain fluctuations, thereby promoting extended hydrogen bonding between strands and stabilization of β-sheet content (Figure 6a). Even without cross-linking (Figure 6b), the central tyrosines, by virtue of their transient association, act to reduce lateral chain fluctuations relative to the control serine configuration (Figure 6c), thereby locally facilitating hydrogen bonding between strands and stabilizing β-sheet content in the core region in comparison to the serine control. Thus, we may conclude that even non-cross-linked tyrosine residues, as a result of their transient associations, are able to reduce chain fluctuations locally, and facilitate transient β-structure in their vicinity, even with the large end fluctuations inherent with a short peptide. While the stability of the aromatic interactions of free tyrosine G

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 6. Average propensity of individual residues for β-sheet conformation over six trials. Beta content for individual residues averaged over two beta-sheets in the bundle during the last 15 ns of simulation for six independent trials and presented as a color map. (A) cross-linked tyrosine residues form beta structure for residues in the core region which are highly stable (shown in dark red) and promote metastable beta content for the rest of the chain. (B) Non-cross-linked tyrosine residues fail to maintain initial beta structure but preserve transient beta structure in the core region near the tyrosine residues. (C) When tyrosine is replaced with serine, the initial beta structure is disrupted completely and does not show any tendency for beta structure to reform.

and the resulting β-content are lower than for cross-linked dityrosine configurations, we expect that these effects would be enhanced in full-length silk fibroin solutions, where chain fluctuations are less dominant, and also by other factors such as shear and dehydration that occur in the native spinning process. This is consistent with other work assessing how tyrosine interacts with adjacent and long-range β-type structures.74 The implication of the hydration state of tyrosine in the transition from the crystalline silk I state to the β-sheet silk II state also suggests that the tyrosine side chains do not specifically inhibit crystal formation.17 This work suggests that the amorphous regions of the B. mori silk fibroin, and more specifically the tyrosine residues in those regions, are partially responsible for the self-assembly of the chains. This directed assembly, driven by aromatic and other side chain interactions, provides the registration required to efficiently crystallize the protein upon dehydration. A cursory review of the N. clavipes spidroin sequence reveals a similar pattern, with alternating crystalline and amorphous domains. Analogous to the fibroin, the aromatic amino acids are found in the amorphous regions, suggesting that they may play a similar role in self-assembly and registration of the β-sheet forming regions. These noncrystalline domains may serve dual roles of enhancing the ductility and toughness of the silks, while also providing a template for the crystallization. Unlike synthetic polymers, with homogeneous repeating units, it is important that the crystallizable segments of structural proteins be appropriately positioned. Thus, inclusion of patterning sites in these regions is an effective way to align the molecules, while

also providing the local mobility required for tough, ductile fibers.

■

CONCLUSIONS Using a combination of correlational measurements, forced association and molecular modeling, we showed that tyrosine residues in regenerated silk fibroin play an active role in selfassembly and the transition to β-sheet structure. The intrinsic fluorescence of the silk protein was found to be directly related to the secondary structure. Further, the covalent coupling of tyrosine residues resulted in the rapid development of higher order structures, ultimately leading to cross-link-dependent βsheet content. Molecular simulations of silk-mimetic peptides confirmed the propensity for the clustering and transient association of the aromatic side chains of tyrosines and showed that their interactions enhanced the fraction of amino acids involved in hydrogen bonding and β-sheet formation. Together, these experimental and modeling results provide substantive support for a key role of tyrosines in silk protein chains in guiding chain associations (intra- and intermolecular) associated with β sheet formation. Such insights offer implications in modulating silk protein processing into new materials, via mimicking these natural associations.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. H

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules Author Contributions

(26) Sashina, E. S.; Bochek, A. M.; Novoselov, N. P.; Kirichenko, D. A. Russ. J. Appl. Chem. 2006, 79 (6), 869−876. (27) Zong, X.-H.; Zhou, P.; Shao, Z.-Z.; Chen, S.-M.; Chen, X.; Hu, B.-W.; Deng, F.; Yao, W.-H. Biochemistry 2004, 43 (38), 11932− 11941. (28) Matsumoto, A.; Lindsay, A.; Abedian, B.; Kaplan, D. L. Macromol. Biosci. 2008, 8 (11), 1006−1018. (29) Ruan, Q.-X.; Zhou, P.; Hu, B.-W.; Ji, D. FEBS J. 2008, 275 (2), 219−232. (30) Domigan, L. J.; Andersson, M.; Alberti, K. A.; Chesler, M.; Xu, Q.; Johansson, J.; Rising, A.; Kaplan, D. L. Insect Biochem. Mol. Biol. 2015, 65, 100−106. (31) Terry, A. E.; Knight, D. P.; Porter, D.; Vollrath, F. Biomacromolecules 2004, 5 (3), 768−772. (32) Foo, C. W. P.; Bini, E.; Hensman, J.; Knight, D. P.; Lewis, R. V.; Kaplan, D. L. Appl. Phys. A: Mater. Sci. Process. 2006, 82 (2), 223−233. (33) Kim, U.-J.; Park, J.; Li, C.; Jin, H.-J.; Valluzzi, R.; Kaplan, D. L. Biomacromolecules 2004, 5 (3), 786−792. (34) Raven, D. J.; Earland, C.; Little, M. Biochim. Biophys. Acta, Protein Struct. 1971, 251 (1), 96−99. (35) Wang, C.-S.; Ashton, N. N.; Weiss, R. B.; Stewart, R. J. Insect Biochem. Mol. Biol. 2014, 54, 69−79. (36) Balasubramanian, D.; Kanwar, R. Mol. Cell. Biochem. 2002, 234− 235 (1), 27−38. (37) Jensen, R. A.; Morse, D. E. J. Comp. Physiol., B 1988, 158 (3), 317−324. (38) Stewart, R. J.; Wang, C. S.; Shao, H. Adv. Colloid Interface Sci. 2011, 167 (1−2), 85−93. (39) dos Santos-Pinto, J. R. A.; Lamprecht, G.; Chen, W.-Q.; Heo, S.; Hardy, J. G.; Priewalder, H.; Scheibel, T. R.; Palma, M. S.; Lubec, G. J. Proteomics 2014, 105, 174−185. (40) Pouchkina, N. N.; Stanchev, B. S.; McQueen-Mason, S. J. Insect Biochem. Mol. Biol. 2003, 33 (2), 229−238. (41) Partlow, B. P.; Hanna, C. W.; Rnjak-Kovacina, J.; Moreau, J. E.; Applegate, M. B.; Burke, K. A.; Marelli, B.; Mitropoulos, A. N.; Omenetto, F. G.; Kaplan, D. L. Adv. Funct. Mater. 2014, 24 (29), 4615−4624. (42) Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990, 112 (14), 5525−5534. (43) Tanner, D.; Fitzgerald, J. A.; Phillips, B. R. Adv. Mater. 1989, 1 (5), 151−156. (44) Kool, E. T. Annu. Rev. Biophys. Biomol. Struct. 2001, 30, 1−22. (45) Pace, C. N.; Horn, G.; Hebert, E. J.; Bechert, J.; Shaw, K.; Urbanikova, L.; Scholtz, J. M.; Sevcik, J. J. Mol. Biol. 2001, 312 (2), 393−404. (46) Mulkerrin, M. G.; Wetzel, R. Biochemistry 1989, 28 (16), 6556− 6561. (47) Tsai, A. M.; van Zanten, J. H.; Betenbaugh, M. J. Biotechnol. Bioeng. 1998, 59 (3), 281−285. (48) Tsai, A. M.; van Zanten, J. H.; Betenbaugh, M. J. Biotechnol. Bioeng. 1998, 59 (3), 273−280. (49) Fatouros, A.; Ö sterberg, T.; Mikaelsson, M. Int. J. Pharm. 1997, 155 (1), 121−131. (50) Lanzarotti, E.; Biekofsky, R. R.; Estrin, D. A.; Marti, M. A.; Turjanski, A. G. J. Chem. Inf. Model. 2011, 51 (7), 1623−1633. (51) Chi, E. Y.; Krishnan, S.; Randolph, T. W.; Carpenter, J. F. Pharm. Res. 2003, 20 (9), 1325−1336. (52) Rockwood, D. N.; Preda, R. C.; Yücel, T.; Wang, X.; Lovett, M. L.; Kaplan, D. L. Nat. Protoc. 2011, 6 (10), 1612−1631. (53) Georgakoudi, I.; Tsai, I.; Greiner, C.; Wong, C.; DeFelice, J.; Kaplan, D. Opt. Express 2007, 15 (3), 1043. (54) Hu, X.; Kaplan, D.; Cebe, P. Macromolecules 2006, 39 (18), 6161−6170. (55) Frishman, D.; Argos, P. Proteins: Struct., Funct., Genet. 1995, 23 (4), 566−579. (56) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graphics 1996, 14 (1), 33−38. (57) MacKerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.;

§

Contributed equally to this work.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS B.P.P. was supported by the Department of Defense (DoD) through the National Defense Science & Engineering Graduate Fellowship (NDSEG) Program. We thank the AFOSR (FA9550-14-1-0015) and the NSERC CREATE and Discovery grant programs for support of this work.

■

REFERENCES

(1) Fu, C.; Shao, Z.; Fritz, V. Chem. Commun. (Cambridge, U. K.) 2009, No. 43, 6515−6529. (2) Walker, A. A.; Holland, C.; Sutherland, T. D. Proc. R. Soc. London, Ser. B 2015, 282 (1809), 20150259. (3) Holland, C.; Vollrath, F.; Ryan, A. J.; Mykhaylyk, O. O. Adv. Mater. 2012, 24 (1), 104−104. (4) Rabotyagova, O. S.; Cebe, P.; Kaplan, D. L. Biomacromolecules 2011, 12 (2), 269−289. (5) Lu, Q.; Zhu, H.; Zhang, C.; Zhang, F.; Zhang, B.; Kaplan, D. L. Biomacromolecules 2012, 13 (3), 826−832. (6) Jin, H.-J.; Kaplan, D. L. Nature 2003, 424 (6952), 1057−1061. (7) Zhong, J.; Ma, M.; Li, W.; Zhou, J.; Yan, Z.; He, D. Biopolymers 2014, 101 (12), 1181−1192. (8) Zhong, J.; Liu, X.; Wei, D.; Yan, J.; Wang, P.; Sun, G.; He, D. Int. J. Biol. Macromol. 2015, 76, 195−202. (9) Warwicker, J. O. Acta Crystallogr. 1954, 7 (8), 565−573. (10) Marsh, R. E.; Corey, R. B.; Pauling, L. Biochim. Biophys. Acta 1955, 16, 1−34. (11) Fossey, S. A.; Némethy, G.; Gibson, K. D.; Scheraga, H. A. Biopolymers 1991, 31 (13), 1529−1541. (12) Asakura, T.; Ashida, J.; Yamane, T.; Kameda, T.; Nakazawa, Y.; Ohgo, K.; Komatsu, K. J. Mol. Biol. 2001, 306 (2), 291−305. (13) Ha, S.-W.; Gracz, H. S.; Tonelli, A. E.; Hudson, S. M. Biomacromolecules 2005, 6 (5), 2563−2569. (14) Kameda, T.; Ohkawa, Y.; Yoshizawa, K.; Nakano, E.; Hiraoki, T.; Ulrich, A. S.; Asakura, T. Macromolecules 1999, 32 (25), 8491− 8495. (15) Asakura, T.; Ohgo, K.; Ishida, T.; Taddei, P.; Monti, P.; Kishore, R. Biomacromolecules 2005, 6 (1), 468−474. (16) Saito, H.; Ishida, M.; Yokoi, M.; Asakura, T. Macromolecules 1990, 23 (1), 83−88. (17) Asakura, T.; Okushita, K.; Williamson, M. P. Macromolecules 2015, 48 (8), 2345−2357. (18) Ma, M.; Zhong, J.; Li, W.; Zhou, J.; Yan, Z.; Ding, J.; He, D. Soft Matter 2013, 9 (47), 11325−11333. (19) Asakura, T.; Umemura, K.; Nakazawa, Y.; Hirose, H.; Higham, J.; Knight, D. Biomacromolecules 2007, 8 (1), 175−181. (20) Zhou, L.; Chen, X.; Shao, Z.; Huang, Y.; Knight, D. P. J. Phys. Chem. B 2005, 109 (35), 16937−16945. (21) Holland, C.; Terry, A. E.; Porter, D.; Vollrath, F. Polymer 2007, 48 (12), 3388−3392. (22) Holland, C.; Terry, A. E.; Porter, D.; Vollrath, F. Nat. Mater. 2006, 5 (11), 870−874. (23) Mo, C.; Holland, C.; Porter, D.; Shao, Z.; Vollrath, F. Biomacromolecules 2009, 10 (10), 2724−2728. (24) Holland, C.; Urbach, J. S.; Blair, D. L. Soft Matter 2012, 8 (9), 2590−2594. (25) Wray, L. S.; Hu, X.; Gallego, J.; Georgakoudi, I.; Omenetto, F. G.; Schmidt, D.; Kaplan, D. L. J. Biomed. Mater. Res., Part B 2011, 99 (1), 89−101. I

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules Joseph-McCarthy, D.; Kuchnir, L.; Kuczera, K.; Lau, F. T.; Mattos, C.; Michnick, S.; Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, W. E.; Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.; Wiórkiewicz-Kuczera, J.; Yin, D.; Karplus, M. J. Phys. Chem. B 1998, 102 (18), 3586−3616. (58) MacKerell, A. D.; Feig, M.; Brooks, C. L. J. Am. Chem. Soc. 2004, 126 (3), 698−699. (59) Best, R. B.; Zhu, X.; Shim, J.; Lopes, P. E. M.; Mittal, J.; Feig, M.; Mackerell, A. D. J. Chem. Theory Comput. 2012, 8 (9), 3257−3273. (60) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79 (2), 926. (61) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kalé, L.; Schulten, K. J. Comput. Chem. 2005, 26 (16), 1781−1802. (62) Georgakoudi, I.; Tsai, I.; Greiner, C.; Wong, C.; Defelice, J.; Kaplan, D. Opt. Express 2007, 15 (3), 1043−1053. (63) Martel, A.; Burghammer, M.; Davies, R. J.; Di Cola, E.; Vendrely, C.; Riekel, C. J. Am. Chem. Soc. 2008, 130 (50), 17070− 17074. (64) Lu, Q.; Hu, X.; Wang, X.; Kluge, J. A.; Lu, S.; Cebe, P.; Kaplan, D. L. Acta Biomater. 2010, 6 (4), 1380−1387. (65) Taddei, P.; Monti, P. Biopolymers 2005, 78 (5), 249−258. (66) Wilson, D.; Valluzzi, R.; Kaplan, D. Biophys. J. 2000, 78 (5), 2690−2701. (67) Nicell, J. A.; Wright, H. Enzyme Microb. Technol. 1997, 21 (4), 302−310. (68) Hu, X.; Kaplan, D.; Cebe, P. Macromolecules 2008, 41 (11), 3939−3948. (69) Sinz, A. Mass Spectrom. Rev. 2006, 25 (4), 663−682. (70) Valluzzi, R.; Jin, H.-J. Biomacromolecules 2004, 5 (3), 696−703. (71) Drummy, L. F.; Farmer, B. L.; Naik, R. R. Soft Matter 2007, 3 (7), 877−882. (72) McGaughey, G. B.; Gagné, M.; Rappé, A. K. J. Biol. Chem. 1998, 273 (25), 15458−15463. (73) Asakura, T.; Suita, K.; Kameda, T.; Afonin, S.; Ulrich, A. S. Magn. Reson. Chem. 2004, 42 (2), 258−266. (74) Taddei, P.; Asakura, T.; Yao, J.; Monti, P. Biopolymers 2004, 75 (4), 314−324.

J

DOI: 10.1021/acs.biomac.6b01086 Biomacromolecules XXXX, XXX, XXX−XXX