pH-Dependent Locking of Giant Mesogens in Fibers Drawn from

There has been some success with this approach in the cases of recombinant spider silk(4) and electrospun collagen fibers.(5) Here, fibers hand-drawn ...
0 downloads 0 Views 7MB Size
1480

Biomacromolecules 2008, 9, 1480–1486

pH-Dependent Locking of Giant Mesogens in Fibers Drawn from Mussel Byssal Collagens Matthew J. Harrington* and J. Herbert Waite Dept. of Molecular, Cellular, and Developmental Biology, University of California at Santa Barbara (UCSB), Santa Barbara, California 93106 Received January 25, 2008; Revised Manuscript Received February 29, 2008

Byssal threads are tough collagenous fibers that mussels use to secure themselves against dislodgement by waves in the marine intertidal zone. Here, preCol, a family of hybrid collagens comprising up to 96% of the protein content in certain regions of byssal threads, was purified in mg amounts from mussel foot tissue for the first time. Conditions for drawing preCols into quality fibers ex vivo were investigated. The most important factor affecting fiber formation was the pH of the drawing solution. The morphology and tensile properties of drawn fibers were also characterized and suggest that a liquid crystal mesophase combined with cross-linking by His-metal coordination plays a role in the assembly/mechanics of drawn fibers and likely in native byssal threads as well.

Introduction Proteins that self-assemble into fibrous structures in vivo play many important roles in biological systems,1 particularly as loadbearing structures such as spider silk2 and tendon collagen.3 Such biofibers have several desirable properties over man-made polymers including multifunctionality, exemplary mechanical function at low material densities, nanoscale hierarchical organization, and self-assembly under benign conditions. Understanding the design principles underlying these properties is a major focus in the field of biomaterials and biomimetics. One tactic in this line of research is a bottom-up approach in which the fiber-forming protein is isolated and its assembly is studied by replicating fiber formation in vitro. There has been some success with this approach in the cases of recombinant spider silk4 and electrospun collagen fibers.5 Here, fibers hand-drawn from a purified collagen-containing protein found in the byssal threads of the mussel Mytilus californianus are described for the first time. Byssal threads are protein fibers secreted externally by the mussel that secure the organism to hard surfaces in the intertidal zone as well as dissipate the energy of wave-induced lift and drag.6,7 Byssal threads are 100–200 µm in diameter, 2–5 cm in length, and have many desirable mechanical properties, including a combination of high ultimate strain (>100%), high ultimate stress (∼75–200 MPa), and high stiffness (∼900 MPa) surpassed only by spider silks among common biological fibers.8,9 Threads also have a toughness comparable to Kevlar9 and the ability to self-heal following yield.10 By dry weight, threads are 95% protein, with the remainder consisting of inorganic molecules, sugars, and divalent metal ions.11 In the distal region of the thread (furthest from the organism), up to 96% of the protein component consists of a family of collagen variants known as preCols.11 There are three known variants of preCol (D, NG, and P), all having a similar modular structure.12 The modules for each consist of a central kinked collagen domain, two variable flanking domains resembling load-bearing protein archetypes at either end of the collagen domain, and two terminal histidine-rich regions (Figure * To whom correspondence should be addressed. E-mail: harringt@ lifesci.ucsb.edu.

Figure 1. PreCol schematic. (A) PreCols are modular proteins with a central collagen domain, variable flanking domains at both ends of the collagen domain, and histidine-rich domains at the two termini. Flanking domains resemble known structural proteins including elastin and silk. Histidine residues are known to bind transition metals at basic pH. The collagen domain directs preCol peptide chains to form triple helices. (B) In the preCol synthesis gland of the mussel, preCol further assembles into bundles of seven helices. Typical length and diameters of helices and bundles are given.

1).12 The remaining protein component consists of two putative matrix proteins TMP-1 and PTMP-1.13,14 Thread formation requires less than five minutes and begins when soluble protein precursors are secreted into the ventral groove of the mussel foot.15 Muscular contractions in the foot shape the thread as it is solidified by cross-linking. Threads are divided into two mechanically distinct regions known as the distal and proximal regions.8 Prior to secretion, preCol mesogens are stored in granules within the mussel foot where they appear to be organized in a smectic liquid crystal phase, as evidenced by TEM images showing alternating rows of rod-like bundles and amorphous domains.16 AFM imaging of threads suggests that the smectic organization is maintained after secretion17 and possibly locked in place by metal coordination cross-links between histidine clusters at the ends of adjacent molecules.11,18

10.1021/bm8000827 CCC: $40.75  2008 American Chemical Society Published on Web 04/11/2008

pH-Dependent Locking of Giant Mesogens

The goal of this study was to further probe the byssal thread assembly process and the molecular and biochemical factors controlling it by making hand-drawn fibers from purified preCols. Factors influencing the drawing process were investigated and the morphological and mechanical properties of fibers were characterized. Our results demonstrate that purified preCols are readily drawn into anisotropic fibers in a pHdependent manner, which apparently involves self-assembly from a transient liquid crystalline phase. Mechanical testing of drawn fibers reveals they have some important similarities to mussel byssal threads and may provide a platform for studying the relationship between molecular structure and mechanical properties in threads.

Materials and Methods Protein Extraction. Mussels were harvested from Goleta Pier (Goleta, CA) and maintained in mariculture at 12–15 °C. Whole feet were dissected from mussels and frozen on glass plates at –80 °C. While frozen, the dark outer epithelium and the tissue above the distal depression were removed with a scalpel. The remaining foot tissue was homogenized on ice in a glass tissue grinder (Kontes Glass Co.) in 5% acetic acid and 10 µM leupeptin and pepstatin (Sigma) in a ratio of 5 mL/g of feet. The pink-brown homogenate was spun at 15000 × g for 30 min at 4 °C. The supernatant (S1) was decanted from the pellet, and its pH was raised to 5.5 by slowly adding solid sodium borate while stirring. S1 (pH 5.5) was then centrifuged at 15000 × g for 30 min at 4 °C. After harvesting the supernatant (S2), solid urea was added to a concentration of 8 M, while stirring, and the pH was titrated to 7.5 with NaOH. This was regarded as the foot extract. Chromatography. Approximately 10 mL of foot extract was injected onto an immobilized metal affinity chromatography (IMAC) HiTrap chelating HP column (GE healthcare) charged with Zn2+. Using an AKTA FPLC system (Amersham Pharmacia), the protein captured by the column was eluted with a steep gradient of imidazole, increasing from 0 mM to 200 mM in a 20 mM phosphate buffer, pH 7.5, with 500 mM NaCl and 8 M urea. PreCol began eluting at about 40 mM imidazole with several coeluting, contaminating proteins. PreCol containing fractions from IMAC (which elute over 3–4 mL) were pooled and 2 mL was injected onto a HiTrap desalting column (GE healthcare) and eluted with a 50 mM acetate buffer, pH 5.0, containing 3 M NaCl. For unknown reasons, the high salt concentration of the elution buffer conveniently enables the separation of preCol from the contaminants. NaCl was then removed from the purified preCol by desalting on the same column using 50 mM acetate buffer, pH 5.5, or Q-H2O. Confirming Purity. Aliquots of pertinent stages of the purification process were run on 15% SDS polyacrylamide gel eletrophoresis (SDSPAGE) and 7.5% acid-urea polyacrylamide gel electrophoresis (AUPAGE) and stained with Serva blue R-250.19 Aliquots of purified preCol were hydrolyzed in 6 N HCl and phenol for 24 h in vacuo at 110 °C and run on a postcolumn ninhydrin-based amino acid analyzer (Beckman 6300 Autoanalyzer) to determine composition and concentration of purified preCol. Purified preCol was run on an automated Edman sequencer (Porton) to confirm purity and determine the relative proportions of the three preCol variants. Due to their reliability in quantitative Edman sequencing, leucine and isoleucine in position 3 of preCol-P and -NG, respectively, and arginine in position 4 of preCol-D were used as indicators of relative quantity (Supporting Information). A 50 µL droplet of purified byssal collagen was placed on a mica surface and incubated for 1 min at room temperature. The mica was then washed in Q-H2O and air-dried. The mica surface was imaged with atomic force microscopy (AFM) on a DI3000 (Veeco). Height and length data were analyzed with nanoscope software version 7.10 (Veeco). Fiber Formation. A 5 µL droplet of freshly purified preCol in acetate buffer was placed on a clean polyacrylic microscope stage. The

Biomacromolecules, Vol. 9, No. 5, 2008

1481

pH was raised from 5.5 to ∼8.0 by adding 100 mM sodium phosphate buffer, pH 8.0, in a ratio of 1:1. Fibers were drawn by touching a metal dissecting probe to the surface of the droplet and slowly pulling away while monitoring fiber integrity through a Nikon SMZ1000 dissecting microscope. Fibers formed as the meniscus between the probe and the solution evaporated and were further drawn from solution by gently pulling away. Once drawn, fibers exhibited supercontraction-like behavior in which they curled up slightly when rehydrated. When the fiber ends were secured, rehydrating the fibers resulted in loss of slack. To determine the dependence of fiber formation on the pH of the pulling solution, purified preCol desalted into Q-H2O was mixed 1:1 with the following buffered solutions: citrate-phosphate buffer (pH 3.0, 4.0, 5.0, 6.0, 6.5, 7.0) and phosphate buffer (pH 8.0). Buffers were prepared according to Gomori.20 The pH of the mixture was recorded with a microelectrode (Mettler-Toledo) attached to a pH meter (Radiometer). Attempts to draw fibers were made from each mixture while under continuous observation under a dissecting microscope, and each was replicated five times. As a control, the pH experiment was performed using a solution of type I collagen from white rabbit skin (Sigma) at a concentration of 0.3 mg/mL. Structural and Mechanical Characterization. Drawn fibers were carefully loaded onto cardstock across a 3.85 mm notch, with ends adhered to the surface using double-sided tape. Fibers were examined with polarized light microscopy (PLM) to investigate birefringence. A first-order red plate at 550 nm (Lomo) was used to determine the sign of birefringence. Several fibers were sputter-coated with gold and imaged with a Tescan Vega TS 5130MM thermionic emission scanning electron microscope (SEM). Fibers used for transmission electron microscopy (TEM) were fixed in 1% glutaraldehyde and 1% paraformaldehyde in a 0.086 M phosphate buffer, pH 7.2, for 1 h, followed by a series of washes in 0.137 M phosphate buffer, pH 7.2. Secondary fixation with 1% osmium tetroxide in the same buffer was performed for two hours. Fibers were dehydrated stepwise in ethanol and embedded in Spurr’s epoxy (Polysciences Inc.). Thin sections of fibers (50–80 nm) were cut with a microtome (Leica), transferred to copper grids, stained with 1% phosphotungstic acid in 1 N HCl for 10 min, and then washed briefly in distilled water.16 Stained sections were imaged on a JEOL 2000FX TEM operated at 80 kV. A small drop of glue (Elmer’s) was placed at each end of the fibers used for mechanical testing to further secure them to the cardstock. Fibers were tested using a Nanobionix tensile tester (MTS) equipped with a homemade humidity chamber. Fiber diameter was determined by photographing each fiber with a microscope. Measurements made in five different regions along the length were in agreement with average diameters determined by SEM. Fibers were loaded securely in the grips and hydrated immediately prior to testing by gently spraying them with a fine mist of water from a nebulizer (Invacare). It is easy to tell when fibers are hydrated because they are very hygroscopic and water beads on them. The humidity of the chamber was raised to 99.9 ( 5% to ensure fiber hydration during testing. Fibers were loaded to failure at a strain rate of 0.01 s-1 while measuring the extension and force on the fiber, which were later converted to stress and strain using the measured fiber diameter to calculate cross-sectional area.

Results Purification. PreCols from M. californianus were purified from acid-extracted foot tissue in two steps: an IMAC column, which exploited the presence of the naturally occurring Hisrich domains, and a desalting column. Notably, the IMAC column alone was not adequate for reproducible purification of preCols. The desalting column removed the remaining contaminants clustered at about 40 kD. Aliquots from the important steps subjected to SDS- or AU-PAGE and protein staining are shown in Figure 2. As in previous studies using other mussel species, there were three preCols (D, P, and NG) that ran anomalously on SDS-PAGE with apparent masses of ∼100

1482

Biomacromolecules, Vol. 9, No. 5, 2008

Harrington and Waite

Figure 3. AFM image of preCols dried on a mica surface. Different stages of preCol assembly can be seen in this image, including a preCol triple helix (A), a bundle of several preCol helices (B), and end-to-end alignment of helices (C).

Figure 2. Purification of preCol from mussel feet. (A) Lanes 1–3 show aliquots from various steps of the purification procedure run on 15% SDS-PAGE. Lane 1 shows foot extract, lane 2 shows IMAC purified fractions, and lane 3 shows the final purified product used for fiber formation. Precol D and NG comigrate in the denser slower running band and preCol P migrates in the lighter faster running band. (B) Lanes 1 and 2 show the foot extract and purified preCol, respectively, run on 7.5% AU-PAGE. Table 1. Percent Amino Acid Composition of cDNA Deduced preCol Sequence vs Purified preCol

Ala Arg Asx Cys Glx Gly His Ile Leu Lys Met Phe Pro+Hyp Ser Thr Trp Tyr+DOPA Val

Da

NGa

Pa

predictedb

purifiedc

19.2 4.2 3.1 0.1 5.8 36.4 1.8 0.9 2.2 1.9 0.7 1.7 13.0 3.5 2.1 0.0 0.7 2.4

14.2 2.5 4.9 0.1 6.7 38.5 2.6 1.8 3.9 2.2 0.2 1.6 10.9 4.7 1.2 0.0 0.7 3.4

9.4 3.5 3.2 0.0 4.2 39.0 3.9 2.5 2.4 1.8 0.0 2.2 14.1 9.8 1.8 0.1 0.5 1.6

16.0 3.7 3.5 0.1 5.7 37.4 2.4 1.4 2.6 1.9 0.4 1.8 12.7 5.1 1.8 0.0 0.7 2.5

14.3 3.3 4.2 0.0 6.5 36.5 3.1 1.3 2.9 2.9 0.4 1.9 12.0 5.3 2.1 0.0 1.1 2.3

a cDNA deduced composition determined from ref 18. b Predicted composition of purified preCol based on cDNA composition and Edmandetermined percent composition of each variant. c Purified values represent the average of three runs and overall have a standard error of 24 h preCol solutions lost the ability to form fibers. This may be due to preCol aggregation based on evidence from SDS-PAGE of aged protein solution, which showed a diminution of the preCol bands and an increase in bands that do not travel past the well or stacking gel when compared with fresh protein (data not shown). Table 2 summarizes the effect that preCol solution pH had on fiber formation. From pH 3.0–5.0 fibers did not form. In the range of pH 5.5–6.1, a small meniscus formed at the probe tip and, in some cases, a tiny, unstable fiber briefly appeared but immediately receded into the solution. At pH 6.4, stable, but short and thin fibers formed, which remained attached to the probe after they broke free from solution. In the range of pH 6.9–8.0, long fibers were freely pulled from the protein solution with lengths exceeding 50 mm and diameters averaging ∼5 µm;

pH-Dependent Locking of Giant Mesogens

Biomacromolecules, Vol. 9, No. 5, 2008

1483

Figure 4. Reconstituted drawn fibers from purified preCol solution. (A) Image of the thick meniscus that forms at the interface of the metal probe and the preCol solution. (B) As the probe is pulled away, the solvent in the meniscus begins to evaporate, and eventually a stable fiber appears in its place (C). All scale bars represent 100 µm. Table 2. Effect of pH on Fiber Formation pH range

fiber formation

3.0–5.0 5.5–6.1 6.4 6.9–8.0

no fibers no fibers or unstable fibers; L < 1 mm short, thin fibers; L < 5 mm, D < 3 µm long fibers; L > 30 mm, D g 3 µm

a draw ratio of over 10000. Attempts to pull hand-drawn fibers from solutions of type I collagen at the same concentration and pH range were unsuccessful. As the SEM images in Figure 5 indicate, typical fiber diameter was in the range of 3–6 µm. SEM imaging of fibers also revealed that fiber diameter varied by less than 5% over the whole length. Fibers showed a strong birefringence that peaked when the fiber was oriented 45° to the analyzer (Figure 6B) and showed complete extinction when oriented 0 or 90° to the analyzer. The sign of birefringence was determined to be positive, indicating that the slow axis runs parallel to the fiber axis. TEM of longitudinal sections of fibers showed lateral alignment of filaments along the axis of the fiber (Figure 6C). The anisotropic morphology appeared in many regions along the length of a single TEM section and continuous filaments as long as several microns could be distinguished. Typical lateral center-to-center spacing of filaments appeared to be 20–25 nm, which is consistent with the morphology of the distal region of native threads.17,24 Figure 7A shows a representative stress–strain curve of a fiber pulled to break, emphasizing the presence of a yield point. Figure 7B shows the stress–strain curve of the same fiber compared to the distal and proximal regions of a native byssal thread. The mean mechanical properties of the fibers tested (N

Figure 5. SEM images of drawn fibers. (A) Two drawn fibers, as seen with SEM. Fiber diameter varies by less than 5% along the length of a single fiber. (B) A magnified image of the fiber in (A), as indicated by the white dashed box.

) 5) alongside those measured previously for native threads from M. californianus are listed in Table 3.8 As Figure 7B reveals, the ultimate stress for the distal region of native threads obtained in this study are consistently higher than the values previously cited in the literature (∼200 vs ∼75 MPa).8 Both values are given in Table 3 for purposes of comparison to drawn fibers. Fibers showed substantial differences from distal byssal threads despite the similar ultrastructural appearance seen with TEM. The modulus, ultimate stress, and energy to break were appreciably lower in fibers than in distal threads. Still, there were several important similarities, including the ultimate strain (∼100%) and the presence of a yield point ∼19% strain (Table 3). However, yield in drawn fibers was more subtle, with a reduction in modulus from 70 to 34 MPa (∼50% loss in modulus) when compared with the yield in the distal portion of native threads that drop from 900 to 0 MPa during yield.

Discussion Mussel byssal threads are assembled in a matter of minutes by secretion of soluble precursor proteins into a groove in the mussel foot. To investigate the biochemical nature of assembly, we have isolated the major protein component of threads and produced stable hand-drawn fibers with large draw ratios. This has not been done previously and is not possible with purified type I collagen within the same concentration and pH range. The purification scheme outlined above yields high purity preCols as evidenced by PAGE, amino acid analysis, and N-terminal Edman sequencing (Figure 2, Table 1). The typical length (∼200 nm) and height (∼1.5 nm) of preCols imaged by AFM are very consistent with the predicted size of a preCol triple helix (Figure 1). Filamentous material significantly longer than 200 nm likely represents end to end cross-linking of individual preCol bundles by chemically interactive residues at the termini. These include histidine, which forms coordination complexes with divalent transition metal ions18,25,26 and 3,4dihydroxyphenylalanine (DOPA), which has been shown to form both coordinate and covalent cross-links within the threads.27,28 Similarly, the presence of filaments with a diameter larger than 1.5 nm is not unexpected because in the preCol secretory granules16 and the mature thread,17,24 preCol helices are believed to form bundles with a diameter up to 9 nm (Figure 1b). The presence of these different forms suggests that purified preCol spontaneously reassembles into higher order arrangements. Furthermore, once purified and brought to an appropriate pH, preCols are capable of assembling into cohesive fibers through

1484

Biomacromolecules, Vol. 9, No. 5, 2008

Harrington and Waite

Figure 6. Anisotropy of drawn fibers. (A) Hand-drawn fiber, as seen under light microscopy. (B) Same fiber seen with polarized light microscopy at peak birefringence (fiber is oriented 45° to the analyzer). Complete extinction of birefringence is seen when the fiber axis is oriented 0 or 90° to the analyzer, suggesting a highly anisotropic orientation of preCols along the fiber axis. (C) TEM of a longitudinal section of a fiber stained with phosphotungstic acid. Small white arrows indicate distinct filaments of 20–25 nm width running along the axis of the fiber. Table 3. Mechanical Comparison of Hand-Drawn Fibers of Purified preCol and Native Byssal Threads

Young’s modulus (MPa) ult stress (MPa) ult strain (mm/mm) strain energy (MJ/m3) yield stress (MPa) yield strain (mm/mm) diameter (µm)

fibers

distala

proximala

70.8 ( 20.9 40.2 ( 9.8 1.07 ( 0.32 26.12 ( 11.2 10.9 ( 5.0 0.19 ( 0.09 5.05 ( 0.87 N)5

868.6 73.3, ∼200b 1.09 nd 63.2 0.16 ∼200

15.6 34.5 1.97 nd na na ∼200

a Mechanical properties for native threads from M. californianus are taken from ref 8. b Ultimate stress values for the distal region observed in this study.

Figure 7. Mechanical performance of hand-drawn fibers of purified preCol. (A) Representative stress–strain curve of a drawn-fiber pulled to break. Dashed lines drawn tangent to the curve clearly indicate the presence of a yield point ∼15% strain. (B) Stress–strain curve from (A) plotted against curves from the distal and proximal regions of a native byssal thread from M. californianus. Although drawn fibers appear ultrastructurally similar to the distal region of byssal threads and break around the same strain, the overall mechanical performance is quite different. Drawn fibers are also mechanically distinct from the morphologically dissimilar proximal region of the thread.

hand-drawing. Notably, the diameters of a single fiber and a group of fibers are remarkably uniform considering their hand drawn production. Longitudinal TEM sections reveal that fiber birefringence is due to 20–25 nm filaments running parallel to the axis (Figure 6). This morphology is consistent with that seen in TEM sections and AFM images of native byssal threads.17,24 The hand-drawing process is unlike the injection-molding

process the mussel uses to form threads.15 Thus, the fact that the ultrastructural morphologies of fibers and threads are so similar leads us to reason that the instructions for assembly may be programmed into the biochemical structure of the preCols rather than due to any particular aspect of the manufacturing process itself. This inevitably begs the question of how assembly is self-regulated; what external conditions or properties inherent to the preCols favor anisotropic ordering along the fiber axis? A plausible explanation may be a predisposition of preCols to form a liquid crystal phase (also known as a mesophase). Prior to the formation of certain fibrous biomaterials, including tendon and silk, the protein precursors are stored as lyotropic mesophases.29,30 In lyotropic liquid crystals, as the concentration of mesogen (liquid crystal forming molecule), increases past a critical point, it becomes more energetically favorable for the elongated molecules to align anisotropically than isotropically.31 The propensity of a protein to form a lyotropic mesophase is dependent on several factors, including molecular aspect ratio, molecular rigidity, and intermolecular interactions.31,32 Aspect ratio is particularly important and is inversely related to the critical concentration; rigid molecules with higher aspect ratios tend to form mesophases at lower concentrations.31 The increased level of order in a mesophase benefits processing of fibers because it reduces the bulk viscosity of the drawing solution and leads to a more aligned and stronger final product.33,34 TEM images of preCols in the granules of the collagen gland exhibit order reminiscent of a smectic liquid crystal phase.16 This degree of alignment is retained during thread formation, suggesting that the smectic mesophase is secreted into the mussel foot groove and locked in place.17 It seems likely that a preCol mesophase also plays an integral role during the drawing of reconstituted fibers because purified

pH-Dependent Locking of Giant Mesogens

preCol solution appears isotropic under cross polarizers, yet produces highly anisotropic fibers similar in ultrastructure to native threads.24 With a steady hand, it is possible to observe the transition from solution to semicrystalline fiber as the air-evaporated solvent at the meniscus recedes (Figure 4). Because lyotropic mesophases are concentration dependent, and the evaporation at the meniscus may provide a focal point for transition from disorder to order in the form of a transient mesophase. PreCols meet all the requirements for a mesogen set forth by Collings, that is, elongated shape, rigid center, floppy ends.32 It is interesting to note that bundling of preCols reduces the axial ratio from ∼130 to ∼20, which approaches the lower limit at which a calamitic molecule will form a liquid crystal phase.31 It is unclear why bundling would be selected because a lower axial ratio requires a higher critical concentration for mesophase formation. One intuitive explanation is that bundling provides a compromise between the required high axial ratio and the rigidity to form a calamitic mesogen, however, at this time, there is no clear evidence for this. Apparently then, the only thing preventing purified preCol from forming a mesophase is the relatively dilute concentration (∼0.2 mg/ml). Studies have shown that in vitro liquid crystal formation in type I collagen requires concentrations greater than 50 mg/mL,35 and evaporation is utilized to achieve this. We conjecture that, at the meniscus, as the solvent evaporates and the protein becomes more concentrated, it becomes more energetically favorable for the long fibrous bundles of preCols to be aligned in the direction that the fiber is being pulled (parallel to the fiber axis). Studies are underway to determine if the transient preCol mesophase can be isolated through intensive evaporation and characterized. Processing of tensile fibers in this manner requires a mechanism for securing the free-flowing liquid crystal phase once drawn, for example, shear-induced β-sheet formation is used during extrusion of spider silks.29,36 We propose that, in the case of the drawn fibers, as the preCols align end-to-end during evaporation, the histidine-rich termini of nearby preCols crosslink with one another through mutual coordinate interactions with divalent metal ions. Experimental evidence suggests such cross-links exist in native threads and that they play an integral role in initial modulus, yield, and self-healing.18,37 This model of metal-mediated locking of the mesophase is consistent with the results in Table 2, which reveal that stable fibers form only from solutions with a pH above that of the pKa of histidine (pH ∼ 6.5). This is significant because His-metal cross-links can only form when histidine is in the deprotonated state.26 As mentioned earlier, there is AFM evidence of end-to-end crosslinking of purified preCols (Figure 3). These findings add weight to the proposition that mussels utilize His-metal coordination cross-links to stabilize the preCol mesophase during byssal thread formation. If this is the case, the mussel has an integrated pH switch as the preCols move from the granules (pH ∼ 5.5) to seawater (pH 8.2) during thread secretion that would prove ideal as a mesophase locking switch.17 Changes in the pH of the drawing solution and the presence of transition metals are also essential to proper silk processing; however, the mechanisms by which they contribute to assembly appear to be very different from that in preCol fiber formation.38,39 The reproducibility of mechanical properties in preCol fibers is surprising considering that they were drawn by hand. This consistency may reflect the high degree of alignment induced by the preCol mesophase. Despite morphological similarities to native byssal threads, reconstituted drawn fibers are largely mechanically dissimilar from them with a stiffness almost 10-

Biomacromolecules, Vol. 9, No. 5, 2008

1485

fold lower and an ultimate stress 4-fold lower (Figure 7b, Table 3). Fibers are also lacking the post-yield stiffening characteristic of the distal region. However, fibers do retain some of the mechanical features distinctive of native threads. Notably, fibers yield and break at strains indistinguishable from the distal region. Yield, however, is much less dramatic in fibers and resembles the yield seen in threads that have been leached of metals or treated at low pH, both of which serve to disrupt His-metal cross-links.18,37 Reduced yield and Young’s modulus in drawn preCol fibers may suggest that both covalent cross-links such as those involving DOPA and sacrificial cross-links such as Hismetal bonds are not forming in fibers to the same degree they do in native threads. As mentioned earlier, there is evidence for the presence of both in byssal threads,18,27 and it seems likely that drawn fibers are deficient in comparison because we make no effort here to incorporate them. No metals are added to the drawing solution except those that make it through the purification process from the IMAC step, however, only a modicum of Zn2+ would be necessary for the minimum degree of crosslinking. Future experiments will attempt to regulate the level of these cross-links in drawn fibers and correlate changes in mechanical performance with particular cross-links. A similar approach has been successful with identifying functional roles of type I collagen cross-links.40

Conclusions This study supports the notion that the biochemistry and morphology of preCols guide their assembly into byssal threads by means of a transient mesophase. It also supports the proposed role of histidine residues as ligands in pH-triggered coordinate cross-links that stabilize the mesophase during thread formation. While the formation of reconstituted fibers from purified preCol is fascinating of its own accord, the fibers are also valuable as a stripped-down version of the thread composed entirely of preCol. With them we can investigate many of the hypotheses concerning the molecular origin of the distinctive mechanical properties of byssal threads, including yield, hysteresis, and selfhealing. Along these lines, current research is attempting to use reconstituted preCol fibers to explore the origin of the mechanical gradient from the distal to proximal regions of byssal threads and the role of His-metal cross-links in mechanical function. Acknowledgment. The authors thank Cheryl Hayashi and Matt Collin for the use of and help with the MTS Nanobionix. The authors thank James Weaver for help with SEM imaging of fibers. The authors thank an anonymous reviewer for penetrating suggestions for improvement. This work made use of MRL Core Facilities supported by the MRSEC Program of the National Science Foundation under Award No. DMR0520415. This work was supported in part by the NASA University Research, Engineering and Technology Institutes on Bioinspired Materials under Award No. NCC-1-02037 and by the National Institutes of Health Grant R01 DE 014672. Supporting Information Available. Determination of percent composition of preCol variants in purified preCol solution with quantitative Edman sequencing. This material is available free of charge via the Internet at http://pubs.acs.org.

References and Notes (1) Scheibel, T. Curr. Opin. Biotechnol. 2005, 16, 427–433. (2) Gosline, J. M.; Guerette, P. A.; Ortlepp, C. S.; Savage, K. N. J. Exp. Biol. 1999, 202, 3295–3303. (3) Wang, J. H.-C. J. Biomech. 2006, 39, 1563–1582.

1486

Biomacromolecules, Vol. 9, No. 5, 2008

(4) Lazaris, A.; Arcidiacono, S.; Huang, Y.; Zhou, J. F.; Duguay, F.; Chretien, N.; Welsh, E. A.; Soares, J. W.; Karatzas, C. N. Science 2002, 295, 472–476. (5) Matthews, J. A.; Wnek, G. E.; Simpson, D. G.; Bowlin, G. L. Biomacromolecules 2002, 3, 232–238. (6) Yonge, C. M. J. Mar. Biol. 1962, 42, 113–125. (7) Denny, M.; Gaylord, B.; Helmuth, B.; Danial, T. Limnol. Oceanogr. 1998, 43, 955–968. (8) Bell, E.; Gosline, J. J. Exp. Biol. 1996, 199, 1005–1017. (9) Gosline, J.; Lillie, M.; Carrington, E.; Guerette, P.; Ortlepp, C.; Savage, K. Philos. Trans. R. Soc. London, Ser. B 2002, 357, 121–132. (10) Carrington, E.; Gosline, J. Am. Malacol. Bull. 2004, 18, 135–142. (11) Waite, J.; Vaccaro, E.; Sun, C.; Lucas, J. Philos. Trans. R. Soc. London, Ser. B 2002, 357, 143–153. (12) Waite, J. H.; Qin, X.-X.; Coyne, K. J. Matrix Biol. 1998, 17, 93–106. (13) Sun, C.; Lucas, J. M.; Waite, J. H. Biomacromolecules 2002, 3, 1240– 1248. (14) Sagert, J.; Sun, C. J.; Waite, J. H. In Biological AdhesiVes; Smith, A. M., Callow, J. A., Eds.; Springer-Verlag: Berlin, Heidelberg, 2006; pp 125-140. (15) Waite, J. H. Results Probl. Cell. Differ. 1992, 19, 27–54. (16) Zuccarello, L. V. J. Ultrastruct. Res. 1980, 73, 135–147. (17) Hassenkam, T.; Gutsmann, T.; Hansma, P.; Sagert, J.; Waite, J. H. Biomacromolecules 2004, 5, 1351–1354. (18) Harrington, M. J.; Waite, J. H. J. Exp. Biol. 2007, 210, 4307–4318. (19) Waite, J. H.; Benedict, C. V. Methods Enzymol. 1984, 107, 397–413. (20) Gomori, G. Methods Enzymol. 1955, 1, 138–146. (21) Qin, X. X.; Waite, J. H. J. Exp. Biol. 1995, 198, 633–644. (22) Maeda, H. Langmuir 1999, 15, 8505–8513. (23) Kotch, F. W.; Raines, R. T. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 3028–3033.

Harrington and Waite (24) Bairati, A.; Zuccarello, L. V. Cell Tissue Res. 1976, 166, 219–234. (25) Schmitt, L.; Ludwig, M.; Gaub, H. E.; Tampe, R. Biophys. J. 2000, 78, 3275–3285. (26) Sundberg, R. J.; Martin, R. B. Chem. ReV. 1974, 74, 471–517. (27) McDowell, L. M.; Burzio, L. A.; Waite, J. H.; Schaefer, J. J. Biol. Chem. 1999, 274, 20293–20295. (28) Lee, H.; Scherer, N. F.; Messersmith, P. B. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 12999–13003. (29) Vollrath, F.; Knight, D. P. Nature 2001, 410, 541–548. (30) Belamie, E.; Mosser, G.; Gobeaux, F.; Giraud-Guille, M. M. J. Phys.: Condens. Matter 2006, 18, 115–129. (31) Flory, P. J. J. Polym. Sci. 1961, 69, 105–128. (32) Collings, P. J. Liquid Crystals: Nature’s Delicate Phase of Matter; Princeton University Press: Princeton, 2002. (33) Magat, E. E. Philos. Trans. R. Soc. London, Ser. A 1980, 294, 463– 472. (34) Wang, X.-J.; Zhou, Q.-F. Liquid Crystalline Polymers; World Scientific Publishing Co.: Singapore, 2004. (35) Gobeaux, F.; Belamie, E.; Mosser, G.; Davidson, P.; Panine, P.; Giraud-Guille, M. M. Langmuir 2007, 23, 6411–6417. (36) Knight, D. P.; Vollrath, F. Philos. Trans. R. Soc. London, Ser. B 2002, 357, 155–163. (37) Vaccaro, E.; Waite, J. H. Biomacromolecules 2001, 2, 906–911. (38) Zhou, L.; Chen, X.; Shao, Z. Z.; Huang, Y. F.; Knight, D. P. J. Phys. Chem. B 2005, 109, 16937–16945. (39) Wong Po Foo, C.; Bini, E.; Hensman, J.; Knight, D. P.; Lewis, R. V.; Kaplan, D. L. Appl. Phys. A: Mater. Sci. Process. 2006, 82, 223–233. (40) Cornwell, K. G.; Lei, P.; Andreadis, S. T.; Pins, G. D. J. Biomed. Mater. Res. 2007, 80A, 362–371.

BM8000827