Article pubs.acs.org/Biomac
Self-Assembly of Recombinant Hagfish Thread Keratins Amenable to a Strain-Induced α‑Helix to β‑Sheet Transition Jing Fu,† Paul A. Guerette,*,†,‡ and Ali Miserez*,†,§ †
School of Materials Science and Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798 Energy Research Institute at Nanyang Technological University (ERI@N), 50 Nanyang Drive, Singapore, 637553 § School of Biological Sciences, Nanyang Technological University, 60 Nanyang Drive Singapore 637551 ‡
ABSTRACT: Hagfish slime threads are assembled from protein-based bundles of intermediate filaments (IFs) that undergo a strain-induced α-helical coiled-coil to β-sheet transition. Draw processing of native fibers enables the creation of mechanically tuned materials, and under optimized conditions this process results in mechanical properties similar to spider dragline silk. In this study, we develop the foundation for the engineering of biomimetic recombinant hagfish thread keratin (TK)-based materials. The two protein constituents from the hagfish Eptatretus stoutii thread, named EsTKα and EsTKγ, were expressed in Escherichia coli and purified. Individual (rec)EsTKs and mixtures thereof were subjected to stepwise dialysis to evaluate their protein solubility, folding, and self-assembly propensities. Conditions were identified that resulted in the self-assembly of coiledcoil rich IF-like filaments, as determined by circular dichroism (CD) and transmission electron microscopy (TEM). Rheology experiments indicated that the concentrated filaments assembled into gel-like networks exhibiting a rheological response reminiscent to that of IFs. Notably, the self-assembled filaments underwent an α-helical coiled-coil to β-sheet transition when subjected to oscillatory shear, thus mimicking the critical characteristic responsible for mechanical strengthening of native hagfish threads. We propose that our data establish the foundation to create robust and tunable recombinant TK-based materials whose mechanical properties are controlled by a strain-induced α-helical coiled-coil to β-sheet transition.
■
INTRODUCTION Hagfish are benthic marine scavengers that have the remarkable ability to secrete large volumes of slime, which is produced from dozens of specialized, ventro-laterally positioned, epithelial slime glands.1,2 Slime production is achieved with a combination of gland mucous cells (GMCs) and gland thread cells (GTCs) that are released in concert by holocrine secretion.1,2 GTCs contain ∼15 cm long protein-based fibers approximately 1−3 μm in diameter,3−5 which are intricately wound intracellularly into prolate ellipsoid shaped bundles ca. 65 μm in diameter and ca.125 μm in length that occupy up to 70% of the GCTs volume.1,3,6 During secretion and exposure to seawater, the cell membranes of both cell types are disrupted such that GTC-based thread bundles interact with GMC-based mucin granules, which facilitates the unwinding of the threads.7 Together, these materials mobilize water to create an ultrasoft (storage modulus G′ ∼ 0.02 Pa),8 sieve-like material.5 While a complete understanding of slime assembly and its functional design is still emerging,5,7,9−11 this material is known to clog fish gills and has been suggested to act as a deterrent to gill breathing predators.3,12,13 © XXXX American Chemical Society
Individual hagfish slime threads exhibit multiple regimes of deformation during uniaxial loading. In seawater they exhibit an initial elastic modulus of about 6 MPa and a yield stress around 3 MPa. The yield is followed by a nearly constant-stress plateau up to 70% extension, followed by extensive strain hardening.14 The ultimate strength of seawater-hydrated fibers is around 180 MPa, and thread failure occurs at 220% elongation. The mechanical properties of hagfish threads also depend intimately on deformation history. For example, the yield of the hagfish thread is reversible only up to 35% strain, as demonstrated by unloading cycles,14 and draw-processing is known to dramatically alter the stress−strain behavior. Notably, the tensile response of hagfish threads can be tailored by imposing specific draw-processing conditions. Failure stress, postyield strain plateau and failure strains can be tuned by drawing native hagfish threads in the hydrated state, followed by a drying step. The mechanical properties of these fibers can further be Received: April 24, 2015 Revised: June 22, 2015
A
DOI: 10.1021/acs.biomac.5b00552 Biomacromolecules XXXX, XXX, XXX−XXX
Article
Biomacromolecules enhanced by postdraw chemical cross-linking.15 Thus, optimally processed hagfish threads exhibit mechanical properties that approach those of spider dragline silks, including high initial elastic modulus (8 GPa), ultimate tensile strength (800 MPa) and strain energy to failure of about 200 MJ/m3. Hagfish threads are built from α-helical coiled-coil proteins that self-assemble into intermediate filaments (IFs), and these IFs are bundled intracellularly with a high degree of axial alignment into macroscopic threads.4,16−18 The mechanical response of drawn hagfish threads is the consequence of a strain-induced conformational transition of their α-helical coiled-coil proteins into β-sheet rich fibers (hereafter referred to as the α → β transition). This transition has been detected by Congo red staining combined with polarized light microscopy, and with wide-angle X-ray diffraction (WAXS),14 which indicate that the high stiffness and failure stresses associated with draw-processed fibers are associated with the presence of a silk-like, β-sheet reinforced supra-molecular network. The genes encoding the two major protein constituents of hagfish threads, namely Eptatretus stoutii thread keratin α and γ (hereafter referred to as EsTKα and EsTKγ), have been cloned and sequenced,17,18 and the encoded proteins have molecular weights (MW) of 66.7 and 62.8 kDa, respectively. While they exhibit relatively low sequence identity with most vertebrate keratins, they have recently been shown to share a high degree of primary amino acid sequence identity with thread keratins (TKs) that were identified by mining genomic data from the lamprey, teleosts, and amphibians.19 Analysis of the intron-exon organizations of the TK-genes, combined with phylogenetic analysis, have indicated that the EsTKα and γ proteins are related to vertebrate Type I and Type II keratins, respectively, and that they may represent an important link in the evolution of vertebrate keratins from Type III IF-proteins. Like other IFproteins, the TKs exhibit tripartite domain architectures, where amorphous/nonhelical N- and C-terminal domains flank a central α-helical rod domain of similar size to those observed in IF proteins. In addition, linker domains (L1, L12, L2) within the rod domain are evident, along with clearly distinguishable heptad repeats in domains 1A, 1B, 2A, and 2B that are characteristic of α-helical coiled-coil forming IF proteins. Furthermore, in both EsTKα and γ, subdomain 2B contains an interruptionor stutterin the heptad repeat which may play an important role in the unfolding/unwinding of the coiled-coils when they are subjected to strain.20−23 Based on the relative simplicity and versatility of their primary amino acid sequence designs, coiled-coil proteins and mimetic peptides thereof have recently been used to develop a variety of technologies with applications in biosensing,24−27 drug delivery,28−30 and proto-cell engineering.31 In addition, several groups have undertaken the engineering of materials and molecular devices based on the coiled-coil proteins from mantis egg case,32 honey bee silk,33 and the shock absorbing marine snail egg case.34,35 The remarkableand tailorable mechanical properties of coiled-coil based native hagfish threads clearly render them a rational target for biomimetic engineering.15 Initial attempts to generate biomimetic hagfish threads have involved the solubilization of native threads in concentrated formic acid, and the subsequent draw-spinning of concentrated EsTK-based dopes into fibers whose assembly and mechanical properties were influenced by protein concentration, Mg2+ cross-linking and postspin draw processing.36 However, the failure stress of these reconstituted fibers
(150 MPa) was inferior to draw processed native threads (800 MPa),14 supporting the view that proper coiled-coil folding, IF bundling, and IF alignment are required to mimic the native thread’s properties. Further attempts to mimic the hagfish thread system have been undertaken by draw-spinning concentrated Type III Vimentin IF-based hydrogels in the presence of Mg2+ into fibers that were subsequently drawprocessed to induce the α → β transition.37 In this case, reasonable mechanical properties were also achieved (for example ∼175 MPa breaking stress), but they did not yet match those of the native hagfish fiber, suggesting an important role for the primary amino acid sequence, assembly, and bundling of thread keratins. The production of biomimetic EsTK-based materials will require several milestones to be met. First, while individual hagfish generate large quantities of native fibers that are suitable for laboratory scale experimentation,1,2,16 their implementation in engineering applications will require efficient recombinant protein production and purification. Second, the biomimetics of hagfish threads will require the development of conditions that facilitate the folding and self-assembly of recombinant EsTKs into coiled-coil-based IF-like filaments. The third stage of the process will involve the use of IF-like filaments as building blocks for the assembly of mesoscale and macroscopic scale materials such as films, gels, and tissue scaffolds, or their alignment/bundling and spinning into macroscopic fibers. Finally, these materials will require draw processing to impose the α → β conformational transition that is essential for the tailoring of their mechanical performance. In this study, we address several of these critical elements to establish the foundation for the production of TK-based materials. (rec)EsTKα and γ proteins were expressed in Escherichia coli and streamlined methods for their purification were established. The solubility and folding behavior of (rec)EsTKα and γ were then investigated by evaluating a range of stepwise dialysis conditions that were based on the published in vitro self-assembly behavior of keratins, and by varying parameters including buffer type, pH, and temperature. A microenvironmental window was identified, which maintained the solubility of the individual (rec)EsTKs and facilitated the folding of α-helix enriched (rec)EsTKγ samples, as detected by circular dichroism (CD). Favorable conditions were used to self-assemble 1:1 (w/w) mixtures of (rec)EsTKα and γ. CD measurements and transmission electron microscopy (TEM) observations indicated the formation of α-helix rich IF-like filaments. The thermal stability of individual and self-assembled (rec)EsTKs was investigated by CD, which revealed distinct heat-induced conformational transitions for individual proteins compared to self-assembled filaments. Concentrated preparations of (rec)EsTK-based filaments were then measured by rheometry. The filaments exhibited gel like properties similar to those observed in IF-based gels. Notably, following imposed oscillatory strain cycling, the α → β conformational transition was detected by CD. We also tested the effect of Mg2+ ions on (rec)EsTK-gels. Contrary to the cross-linking effects that Mg2+ exerts on lab-spun solubilized hagfish threads and other IFbased solutions, Mg2+ was found to disrupt (rec)EsTK-based filament network integrity. Taken together, our data establishes the foundation to prepare α → β transformable TK-based materials, which can be engineered into mechanically tunable, high-performance materials. We also suggest that recombinant TK engineering complements and refines our understanding of B
DOI: 10.1021/acs.biomac.5b00552 Biomacromolecules XXXX, XXX, XXX−XXX
Article
Biomacromolecules
However, years of research into the self-assembly pathways of keratins and IFs provide an additional foundation for the evaluation of EsTK self-assembly.38−45 The weak to moderate primary amino acid sequence identity of EsTKα and γ with Type I and II keratins proteins suggested that their self-assembly pathways may be similar. We therefore developed a panel of conditions to test the solubility, folding, and self-assembly of (rec)EsTKα and γ and 1:1 (w/w) mixtures thereof. 1:1 (w/w) ratios were chosen because (i) native hagfish threads are considered to contain equivalent ratios of these two proteins;46 (ii) the proteins have been reported to assemble into heterodimers similar to those obtained with 1:1 (w/w) mixtures of keratins, and (iii) based on the results of Downing and Spitzer who reported successful self-assembly of purified EsTKα and γ with these ratios.46 Self-assembly conditions were also designed to closely resemble those known to facilitate the helical folding, coiled-coil formation, and IF assembly of Type I and Type II keratins in vitro.47 Briefly, lyophilized proteins were dissolved in 8 M urea in the desired buffers with 1 mM DTT (dithiothreitol, Bio-Rad) at specific pHs and temperatures using 0.2 mg/mL protein samples, whose concentrations were measured with a Bradford protein assay kit (Thermo Scientific). The proteins were then subjected to stepwise dialysis against 4 M, 2 M and finally 0 M urea with the same buffer and pH conditions, where each dialysis step was carried out for 24 h. This generalized protocol was used to conduct a panel of experiments with a set of buffers including 2 mM Sodium Citrate, 2 mM phosphate buffered saline (PBS), 2 mM 2-amino-2-hydroxymethylpropane-1,3-diol (Tris) and 2 mM 2-[N-morpholino]ethanesulfonic acid (MES). The final dialysis step was into 5 mM of the assigned buffers. A range of pH values (pH 4, 5, 6, 7, 8.4, and 9) and temperatures (4, 22, and 37 °C) were tested to identify favorable conditions for solubility, folding and potential assembly of the (rec)EsTKs. pH 8.4 and a temperature of 4 °C were considered to be conditions that resemble those that occur in the native system, while the additional conditions provided the opportunity to assess the sensitivity of these proteins to these conditions, an approach that has also been used to gain insights into the folding and self-assembly pathways of other IF-proteins.47 In each case, the solubility of the final dialysis products was evaluated visually to gauge the presence or absence of a pellet following centrifugation at 13 000g. The stepwise dialysis products of (rec)EsTKα and (rec)EsTKγ were concentrated using Vivaspin 2 (10 kDa molecular weight cut off, GE Healthcare). The stepwise dialysis products of 1:1 (w/w) mixtures were concentrated using Vivaspin 2 columns (100 kDa molecular weight cut off, GE Healthcare). All products were stored immediately at 4 °C, and their subsequent characterizations by CD, TEM, cryo-TEM and rheometry were conducted within 3 days of each dialysis experiment. The protease inhibitor phenylmethylsulfonyl fluoride (PMSF) was found to significantly affect CD spectra. Therefore, it was necessary to acquire our CD measurements in the absence of PMSF. We also conducted parallel self-assembly and TEM observations with PMSF (1 mM) and without this compound. Our TEM images indicated no differences in the fibrillar morphologies of self-assembled products between the two treatments, where they were evaluated after storage at 4 °C for 1 week and 1 month (data not shown). Biophysical Characterization of (rec)EsTK Dialysis Products. The secondary structures of the stepwise dialysis products of (rec)EsTKα, (rec)EsTKγ, and 1:1 (w/w) mixtures thereof were characterized by CD using a Chriascan spectropolarimeter (Model 420, AVIV Biomedical Inc.). The concentrated proteins were diluted to 1 mg/mL prior to each set of CD measurements. Corresponding dialysis buffers were used as calibrants in each case. Measurements were obtained in triplicate across wavelengths ranging from 190 to 260 nm, using a 1 nm step size and 1 nm bandwidth. Spectra were smoothed using the Savitzky-Golay method with a second order polynomial. Characterization of (rec)EsTK Dialysis Products by TEM and Cryo-TEM. Aliquots of 4 μL of 100 μg/mL (rec)EsTK dialysis products were applied to glowed, carbon-coated grids (Ted Pella), adsorbed for 30 s, and then fixed with 2% glutaraldehyde (Grade I, Sigma). Excess liquid was removed using filter paper, and the samples
the assembly, structure, functional mechanics, and evolution of the keratin gene/protein family.
■
EXPERIMENTAL SECTION
Expression and Purification of (rec)EsTKs. Codon-optimized genes encoding the full length of EsTKα and γ proteins were purchased from DNA 2.0 (Menlo Park, California). The genes were obtained as expression ready inserts cloned in the pJ414 expression vector. The plasmids were transformed into E. coli TOP10 cells and stocks were established in 20% glycerol and banked at −80 °C. TOP10 cultures were then grown overnight, and the plasmids were purified using a Qiaquick DNA purification kit (Qiagen) and stored in autoclaved milli-Q purified water. Approximately 10 ng of each plasmid was transformed into BL21 (DE3) cells by heat shock and stocks were prepared with 20% glycerol. Aliquots of 10 μL of these cell line stocks were used to inoculate 4 mL of LB media (BD Bioscience) containing ampicillin (50 μg/mL, Sigma) and chloramphenicol (34 μg/mL, Sigma) in 14 mL Falcon tubes. For each EsTK cell line, precultures were incubated overnight at 37 °C with a shaking speed of 250 rpm. One milliliter from each of the precultures was then transferred into 600 mL of fresh autoclaved LB medium in conical flasks containing the same Ampicillin and Chloramphenicol concentrations described above. The cultures were incubated at 37 °C with a shaking speed of 250 rpm until an OD600 of ∼0.6−0.8 was obtained. Protein expression was induced with 1 mM IPTG (isopropyl-β-D-thiogalactopyranoside, Sigma) with incubation at 37 °C and a shaking speed of 250 rpm for an additional 3 h. Cells were harvested by centrifugation at 15 700 g for 10 min at 4 °C. Pellets from 1 L cultures were resuspended in 50 mL of lysis buffer (50 mM Tris, 200 mM NaCl, 1 mM PMSF, pH 7.4) and lysed by six rounds of microfluidization using a Microfluidics M-110P apparatus. Cell lysates were then subjected to centrifugation at 52 000 g for 1 h at 4 °C and the resulting pellets were washed twice with 20 mL wash buffer 1 (100 mM Tris, 5 mM EDTA, 2 M urea, 2% (v/v) Triton X-100, 5 mM DTT, pH 7.4) followed by two washes of wash buffer 2 (100 mM Tris, 5 mM EDTA, 5 mM DTT, pH 7.4). Inclusion bodies were retrieved by low speed centrifugation (5000g for 15 min at 4 °C) between wash steps. Pooled sets for each protein were denatured and resolubilized in 10 mL of 8 M urea with 0.02 M NaH2PO4 and 0.5 M NaCl at pH 7.8. Solubilized inclusion body purification products were evaluated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDSPAGE) with Commassie Blue 250 (CBR-250) staining. Size-exclusion chromatography (SEC) was used to further purify the (rec)EsTKs. SEC experiments were performed using an AKTApurifier FPLC system (Fast Pressure Liquid Chromatography, GE Healthcare, Life Sciences) equipped with a UV detector (λ = 280 nm) using a Superose 6 10/300 gel filtration column at a flow rate of 0.4 mL/min with the running buffer (8 M urea, 0.02 M NaH2PO4, 0.5 M NaCl, pH 7.8). Each purification run involved the injection of 2 mL urea solubilized proteins (1 mg/mL) obtained by microfluidization. In each case, 30 1 mL fractions were obtained with an in-line fraction collector module. Ten microliters from selected fractions was evaluated by SDS-PAGE to gauge protein content and identity. Fractions containing predicted target proteins were dialyzed against 5% acetic acid. The molecular weights and purities of these target proteins were further confirmed using matrix assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF) using an AXIMA-TOF2 equipped with an N2 laser (337 nm, 4 ns pulse width). Briefly, 2 μL of 1 mg/mL predicted target proteins were mixed with 2 μL sinapinic acid matrix (10 mg/mL) (Sigma), which was dissolved in a 50/50 milli-Q purified water/acetonitrile mixture and applied to a MALDI-TOF plate. A 20 kV accelerating voltage was used, and spectra were recorded in linear mode by averaging at least 100 laser shots at a power setting of 120 system units. The target proteins were freezedried and stored at −80 °C for subsequent dialysis/self-assembly experiments. Folding and Self-Assembly of (rec)EsTKs. Purified native EsTKα and γ have previously been self-assembled into IF-like filaments,16 although those experiments have not been replicated.37 C
DOI: 10.1021/acs.biomac.5b00552 Biomacromolecules XXXX, XXX, XXX−XXX
Article
Biomacromolecules
Figure 1. Purification of (rec)EsTKα (left) and (rec)EsTKγ (right). (a,d) FPLC/SEC absorption spectra at 280 nm. (b,e) SDS-PAGE analysis of SEC purification fractions. L: Molecular weight ladder; 8−20: SEC fractions. (c,f) MALDI-TOF spectra from SEC fractions 13−15. Time of recovery for G′ experiments were conducted on the 1:1 self-assembled mixture (1 mg/mL). The strain was applied from 0.01% to 1000% until network rupture (Step 1), followed by a rapid return ( 1 Pa for 0.2 mg/mL), we conclude that a similar behavior explains the measured weak concentration dependence. We then evaluated the effect of strain sweeps at a frequency of 1.0 Hz on 1.0 mg/mL samples of (rec)EsTK-filament gels (Figure 5e). The gels exhibited a plateau at ∼1.8 Pa from 0.4 to 10% strain, followed by a strong decay to 0.2 Pa from 10 to 1000% strain (Step 1 in Figure 5e) due to rupture of the network. This behavior is similar, for example, to that observed for uncross-linked 1 mg/mL preparations of human HK5/14based IFs50 and K8/K18 keratin filaments.60 The (rec)EsTK network was then quickly (
