Reversibly labile, sclerotization-induced elastic properties in a keratin analog from marine snails: whelk egg capsule biopolymer (WECB)

We investigated the mechanical properties of an unusual protein polymer,made by marine whelks, in order to understand the relationship between the biogenesis of a complex biomaterial and its mechanical properties. Whelks are a species-rich assemblage of carnivorous marine snails known especially for their elaborate egg capsules, the subject of this study. The whelk egg capsule biopolymer (WECB) is a proteinaceous composite elastomer possessing alphahelical structural motifs (determined from X-ray diffraction studies)(Flower et al., 1969) and hierarchical layers of ordered fibrous constituents(Rawlings, 1999), suggestive of a structure analogous to an intermediate filament (IF)-based material such as that of hard alpha-keratin (Whitely and Kaplin, 1977).

Despite apparent structural similarities at the IF level, there are important structural and biogenic differences between WECB and hard alpha-keratin. For example, WECB is composed of a plywood-type arrangement of discrete fibrous sheets (foils) arranged at various angles such that isotropic mechanical behavior is observed in the plane formed by the transverse and longitudinal axes of the capsule (H.S.R. and R.E.S., unpublished data). In contrast, hard alpha-keratin is a complex cylindrically arranged composite with concentric hierarchical outer layers around a major axis of fibrillar orientation (Hearle,2000).

Whelk egg capsule biopolymer (WECB) possesses unusual tensile properties(Rapoport and Shadwick, 2002). Quasi-static mechanical tests revealed the following stress-strain features:an initial region of low strain (>3-5%) distinguished by relatively high modulus of elasticity and low hysteresis, followed by an apparent yield to a plateau-like region with relatively low modulus and high hysteresis that persists up to strains of at least 50% (see Fig. 1). Although this transition has the appearance of tensile failure, it is transient only, and complete recovery occurs during unloading, as evidenced by stable stress-strain curves from repeated cycles to high strains(Rapoport and Shadwick, 2002). For simplicity in referring to these material properties, the initial stiff portion of the stress-strain curve is termed the `Hookean' region, due to its linear elastic response. Following the Hookean region is the plateau-like, or`yield' region, so called due to the zone of apparent yield that precedes it,a convention used previously (Feughelman,1964) to describe the similar response to strain found in hard alpha-keratin (Fig. 2). Although similar responses to strain are generated in both materials,significant differences in mechanical parameters exist between WECB and hard alpha-keratin. For example, Denny (Denny,1988) reported that keratin possesses an approximate tensile strength, breaking strain, toughness and maximum elastic modulus of 150 MPa,50%, 20 MJ m^-3 and 4000 MPa, respectively. We have found WECB to posses a similar order of magnitude breaking strain of 95%, but all other parameters are greatly reduced: 5 MPa tensile strength, 3 MJ m^-3toughness, and a 100 MPa maximum elastic modulus(Rapoport and Shadwick,2002).

Since hard alpha-keratin is based on hierarchical arrangements of IF structures, and WECB is putatively similarly organized, it is reasonable to consider differences in higher order structure, accessory proteins, matrix and stabilization as causative factors in resultant mechanical properties. Some insight into the mechanics of hard alphakeratin has been obtained in the studies of hagfish slime threads, which are matrix-free IF analogues that are free from complicating modifications listed above(Fudge and Gosline, 2004). These IF analogues demonstrate mechanical properties that are different from those of either WECB or hard alphakeratin. For example, hagfish slime threads in saline possess an approximate tensile strength, breaking strain, toughness and maximum elastic modulus of 180 MPa, 220%, 130 MJ m^-3 and 220 MPa, respectively (Fudge et al.,2003).

It is clear in the deviation of IF-based materials from simple IF fiber mechanics that successive levels of processing in hard alpha-keratin and WECB are responsible for the development of end-stage mechanical properties, and that the respective formation of these materials is very different. Hard alphakeratin is derived from epithelial cells and possesses cellular remnants from this complex biogenic process (Van Steensel et al., 2000), whereas WECB precursor is stored within glandular vesicles and formed into a functional material in the following two-step process. The raw capsule containing embryos is first formed in the nidamental (capsule) gland (Fig. 3), passed outside the body through the reproductive tract onto the substratum, and then secondarily covered by the ventral pedal gland (VPG)located on the ventral surface of the foot(Fig. 4). Treatment over the course of many minutes to an hour in the VPG renders the newly formed capsule rigid and insoluble (Ankel,1929). Because a clear formation process exists that can be interrupted at finite stages, we are able to form the hypothesis that treatment in the VPG involves a chemical process that operates on IFs already present to provide the critical end-stage mechanical properties in WECB. Moreover, we hypothesize that harsh treatments involving desiccation, thermal treatment and change in pH, will show that the chemical treatment in the VPG involves the addition of covalent cross-linking to WECB.

Therefore, the present studies had the following objectives: (1) To investigate structural differences in WECB pre- and post-VPG processing; (2)to measure mechanical differences in WECB pre- and post-VPG processing; (3) to examine changes in mechanics effected in WECB through the application of denaturing treatments (pH and temperature); and (4) to discuss potential relationships among IF materials: hard alpha-keratin, hagfish slime threads and WECB.

Previous work revealed similar properties in WECB from seven species representing five genera: Busycon carica, Busycon canaliculatum, Kelletia kelletii, Chorus giganteus, Trophon geversianus, Acanthia spirata and Ocenebra poulsoni (Rapoport,2003). In this study we focus on the WECB of two species, Busycon canaliculatum (Linnaeus 1758) and Kelletia kelletii(Forbes 1852). Because significant similarities in fundamental mechanical properties of WECB exist among whelk species, in particular the shape of the stress-strain curve, we feel the results of the current study are representative of WECB across the order Caenogastropoda.

Fresh egg capsules from Busycon canaliculatum were obtained in seawater from The Marine Biological Laboratory, 7 MBL Street, Woods Hole, MA 02543, USA. Capsules were subsequently washed with deionized water and frozen at -20°C until ready for use. Defrosted capsules were washed exhaustively in deionized water. The capsule was then dissected along the medial ridge connecting both sides of the capsules, and the inner contents discarded. Whole thickness strips for testing were formed by using pairs of standard razor blades at a fixed width of 6 mm. Strips were on average 10-12 mm in length, by repeatedly selecting the same cut orientation. Thickness of egg capsule strips varied between 100 μm and 500 μm.

Capsule properties change as they are manufactured and processed by the snail in a sequence of stages. Two designations are used in describing capsules based on processing that occurs in the ventral pedal gland. Mature capsules are those that have been fully completed by the snail and released from contact with the animal. Immature capsules are those that have been intercepted before the last stage of manipulation occurs in the ventral pedal gland. In order to obtain immature capsules one must wait for the period of reproduction and intercept capsules as they pass out of the nidamental gland en route to the ventral pedal gland. Another, more controllable technique for isolating immature egg capsules has been developed(Ram, 1977). In this technique, a neural extract is created by homogenizing the esophageal ganglion of either male or female snails in 0.2 μm-filtered seawater on ice and then centrifuging at >10 000 g for 30 s. The supernatant is subsequently injected into the sinus cavity of the fecund female snail. Extract injection reliably induced the production of single immature egg capsules from gravid B. canaliculatum specimens maintained in aquaria at The Marine Biological Laboratory, Woods Hole, MD, USA. J. Ram (Wayne State University) generously donated an immature capsule and a mature capsule from C. giganteus for comparison. In all cases, capsules were stored and prepared for testing as previously described.

Specimens of B. canaliculatum egg capsule material were air-dried and sputter-coated with gold prior to viewing at 20 kV and a magnification of up to ×15000 on a Cambridge S-360 scanning electron microscope.

Kelletia kellettii is a species endemic to the coast of Southern California and can be readily collected from the subtidal zone. Specimens of K. kellettii were maintained in an indoor aquarium with ambient flow-through seawater. This laboratory population was observed to continue normal reproductive behavior in synchrony with free-living cohorts in the ocean. During a reproductive cycle following mating, a female will begin the process of egg capsule deposition on available substrate, which in this case was an aquarium glass wall (see Fig. 4). It is possible to watch capsule formation in progress and intercept capsules en route to the ventral pedal gland(Fig. 5). In this fashion an experiment was conducted in which snail capsules (N=3) were intercepted at different stages in processing and tested by quasi-static methodology (Rapoport and Shadwick,2002). Briefly, tensile tests on capsule specimens were performed on an MTS material testing apparatus; model 858 Mini Bionix, MTS Systems Corporation, Eden Prairie, MN, USA.

Most biological materials possess viscoelastic properties(Vogel, 1988). Since these materials present time-dependent as well as temperature-dependent properties,a standardized method for characterizing viscoelastic behavior was utilized(i.e. forced-oscillation). Specifically, we wanted to make separate tests of the dynamic response of the two distinct parts of the stress-strain curve,i.e. the Hookean and the yield regions (see Figs 1 and 2). This technique might present a method for teasing apart contributors to bulk mechanical behavior in WECB and also addressing observations from stress-relaxation experiments made previously (Rapoport and Shadwick,2002). These experiments showed that when WECB was stretched into the yield region and held at constant length, the force relaxed, and the high stiffness (Hookean) behavior reappeared when stretch was reapplied.

Egg capsule strips 6 mm wide were clamped into a dynamic testing device. Specimen thickness was measured with a micrometer. The initial length of the specimen, L_o, was measured using a Vernier caliper. The testing rig consisted of an electromagnetic vibrator (model 400, Ling Dynamic Systems, Don Mills, Ontario, Canada) mounted vertically on the adjustable crosshead of a test frame, which allowed different levels of pre-stretch to be applied. Sinusoidal oscillations were imposed by a function generator (BK Precision 3011B, Chicago, IL, USA) driving the vibrator. Specimens were tested in two modes relating to the shape of the quasi-static stress-strain curve:Hookean (N=5) and yield (N=6), and a third mode, vibe-ramp(N=5). For the Hookean test mode, specimens were extended to a mean strain of 2% by adjusting the position of the crosshead to place the material's behavior well within the Hookean region (e.g. <∼5% strain). Since dynamic testing is done at small strains (on the order of ±0.05 mm), the material remained in that region during the extent of the dynamic testing. In the yield test mode, pre-strain was applied as described above. Finally, vibe-ramp denotes a test conducted in the yield region where, during the dynamic test, a constant extension rate was applied to the material(∼0.03 mm s^-1) through motion of the crosshead. This had the effect of overlaying a ramp displacement onto a sinusoidal displacement. Displacement was measured at the vibrator through a fabricated cantilever transducer comprising strain gauges in a Wheatstone bridge. Force was measured through another fabricated cantilever transducer also comprised a similar strain gauge-based Wheatstone bridge. The resonant frequency of the force transducer was 200 Hz, sufficiently higher than the maximal testing frequencies that approached 15 Hz. Force and displacement signals were conditioned by matched amplifiers before being recorded digitally using a TL-2 Interface board (Axon Instruments, Burlingame, CA, USA) and Axotape 1.3.0.09 software on a 486-PC.

In addition to dynamic and quasi-static mechanical tests of native B. canaliculatum egg capsule strips, tests were also conducted following treatment of specimens with formic acid. Formic acid is a common organic acid used in a variety of protein applications. Formic acid irreversibly destabilizes protein secondary structure when its aldehydic functions form esters with the aminohydroxy group of the protein backbone(Josefsson, 1962; Josefsson, 1966). In this fashion, we would expect dissolution of the labile or noncovalent associations within WECB (e.g. hydrogen bonds) due to the low pH formic acid environment. As normal protein conformation is disrupted, progressive attachment of aldehydic functions will proceed. In destabilizing the protein, it might be possible to study the purely covalent interactions present in the material that are consistent with the different stages of processing.

Specimens were first measured using a micrometer to determine thickness,and then mounted between the clamps of a MTS material testing apparatus (model 858 Mini Bionix, MTS Systems Corporation, Eden Prairie, MN, USA). All strips used were pre-tested before treatment to ensure normal behavior. Formic acid was employed as follows: strips of WECB were allowed to equilibrate for 30 min in the following concentrations of formic acid prior to testing: 0% (control; N=4), 11% (N=4), 22% (N=4), 44% (N=3) and 88% (N=4). We noticed that for incubation times approaching 1 h,there was a prominent alteration to mechanics that was only partially reversible. We mitigated any damage-history effect that might also be present at 30 min incubation times by using multiple samples per concentration, and no cross-concentration treatment of samples. All concentrations of formic acid used in this study possess a pH <2.0.

In this study, H is the hysteresis of the stress-strain loop derived from quasi-static mechanical testing. Hysteresis is calculated numerically in spreadsheet form by subtracting the area under the return curve from the area under the extension curve and dividing the result by the area under the extension curve. Resilience can be thought of as the elastic efficiency of the material in question. Synthetic rubbers, for example,approach a resilience of 96-97% (Gosline,1980).

Finally, an additional strip that was allowed to equilibrate for 1 h in 88%(pH <2) formic acid was tested dynamically for comparison to the native state using the same protocol mentioned for the native specimens in the preceding section. The longer incubation time as mentioned earlier was found to irreversibly affect the mechanics of WECB.

where |E| is the complex modulus.

The storage and loss moduli are indexes of elastic and viscous contributions to stiffness, respectively. A common convention in studying viscoelastic materials is to plot both Tanδ and E′ versusthe logarithm of the excitation frequency generated from dynamic behavior at various temperatures (e.g. Vincent,1990). These resultant master curves are useful for understanding the range of viscoelastic behavior of a material and facilitate inter-material comparisons.

To further explore the effects that a wide pH range has on WECB protein mechanics, strips (N=5) were in turn equilibrated for 30 min in citrate/phosphate buffer (Stoll and Blanchard, 1990) at the following pH values: 2, 5, 7 and 10. To eliminate a possible effect of history by proceeding sequentially through the pH range, the following order was chosen to create maximum shift in pH during the course of the experiment: 10, 2, 7 and 5. Following equilibration, B. canaliculatum egg capsule strips were tested at 19°C as described previously for the formic acid-treated specimens, with the exception that hydration was now maintained through the use of an environmental chamber. Use of the chamber allowed samples to be equilibrated at various pH values without removing them from the testing clamps- a potential source of variation in the mechanical response of the material. Elastic modulus of the Hookean and the yield regions, yield stress and resilience were measured as described above.

Thermal studies present an alternative to acidification that similarly affects stabilization in proteins, primarily through the disruption of hydrophobic interactions, electrostatic interactions and hydrogen bonding(von Hippel and Wong,1965).

For this study strips were placed in water in a temperature-controlled environmental chamber. Specimens tested in this fashion were clamped into the tensile testing device throughout the experiment while being subjected to temperatures ranging from 5°C to 92°C. Ten stress-strain cycles were performed after the specimen reached the target temperatures of 7°C(N=2), 20°C (N=2), 40°C (N=2), 65°C(N=2), 78°C (N=2) and 92°C (N=5). Elastic modulus of the Hookean and the yield regions, yield stress and resilience were measured as described previously. Following testing, strips were allowed to equilibrate for 24 h at 19°C in water prior to exploratory testing to determine the extent of recovery from thermal perturbation in the specimen.

Chemical treatments (formic acid, buffers) have been used to alter the behavior of WECB. Hydration is also a related concern in testing IF materials(Fudge and Gosline, 2004). Therefore, the quasi-static properties of WECB as determined through toughness calculations were determined by air-drying five strips of the material. These strips were subsequently tested as described previously with the exception that instead of repeated cycling, these strips were pulled to failure. The area under the curve to failure is defined as a materials' toughness - the ability to absorb energy - and a useful index for comparing properties among materials (Rapoport and Shadwick,2002). Initial modulus was calculated as the initial slope of the stress-strain curve. Yield strain, yield stress and breaking strain were obtained directly from the stress-strain curves.

For the dynamic testing data set, statistical differences among treatments were investigated. Statistical regression analyses, ANOVA comparisons and post-analysis Tukey HSD or Holm-Sidak tests were conducted using both STATGRAPHICS Version 5.1 and MINITAB version 13.32 software (Statistical Graphics Corp., Herndon, VA, USA and Minitab Inc., State College, PA, USA,respectively). For dynamic data, regression lines were fit to pooled storage modulus/frequency and Tanδ/frequency plots for each test mode (Hookean,yield and vibe-ramp) and each parameter, Tanδ and storage modulus. Pair-wise comparisons of regression coefficients were made. For quasi-static data, one-way ANOVAs were carried out for the following parameters calculated for each test mode: Hookean modulus, yield region modulus, yield stress and resilience. Sample size is reported elsewhere. Statistical significance was calculated using a 95% confidence level.

The egg capsules experience a time-dependent change in mechanical properties during and following processing in the ventral pedal gland. This still unknown sclerotizing process stabilizes the material at two levels: (1)in a chemically, thermally and mechanically reversibly labile Hookean region;(2) in a non-labile yield region. The interactions of these two levels of stabilization in the material allow for a deviation from expected dynamic mechanical behavior seen in rubbery polymers.

Immature capsules roughly approximate the shape of the fully processed mature capsule as illustrated for three species in Fig. 5. Additionally the immature capsules can be readily solubilized whereas mature capsules are resistant to degradation (H.S.R. and R.E.S., unpublished). Attempts to mechanically test immature capsules either quasi-statically or dynamically were unsuccessful due to a lack of material cohesiveness when subjected to even low-level deformation. SEM studies of immature and mature B. canaliculatum (Fig. 6)revealed structured lamina with distinct fibrous presence (foils) in differing orientations. Capsules isolated from K. kelletii during various stages of processing and subsequently quasi-statically tested revealed that the characteristic bimodal stress-strain behavior was not evident until some time had been spent in the ventral pedal gland(Fig. 7). Specifically, these capsules possess long-range elasticity consistent with the yield region of the bimodal stress-strain diagram for mature capsule material, but lack the high-stiffness Hookean region found at lower strains. The Hookean region develops only after longer exposure times in the ventral pedal gland.

Mechanical properties: formic acid concentration series The formic acid treatments and the buffered pH treatment elicited seemingly similar results in that there was an apparent change in the shape of the stress-strain curve (Figs 8, 9). Specifically, the Hookean region is diminished with increasing concentrations of acid while the yield region seemed unaffected. We found that by increasing the formic acid incubation time from 30 min to 1 h at higher acid concentrations, the effect became pronounced enough to completely eliminate the Hookean region. Since this behavior was partly irreversible(Fig. 10), and probably represented damage in the material through excessive aldehydic attachment, we only analyzed the data from the 30-min incubations. In contrast, with shorter time intervals, changes to stress-strain behavior were apparently fully reversible upon returning the specimen to native conditions. In sum, the formic acid effect acts on a labile structure that provides the high-stiffness(Hookean) behavior.

The resilience showed no statistically significant differences across formic acid concentration (P=0.111; Fig. 11). If the specimens with longer incubation times were included, an increase in resilience would probably be present due to changes in the hysteresis of the stress-strain diagram based on the diminishment of the Hookean region. The yield stress did change significantly at higher formic acid concentrations (P=0.0; Fig. 12). At a concentration of 88% the yield stress was the lowest of the series at ∼1.7 MPa. In experiments with formic acid that persisted for longer incubation times(∼1 h), the yield region was removed completely. Lastly, the elastic modulus of the yield region was not altered significantly with formic acid treatment (P=0.136; Fig. 13). However, the Hookean region modulus did vary significantly with formic acid (P=0.002) concentrations above 11%, all varying with the 11% concentration and ranging from ∼30 MPa to ∼10 MPa. The overall formic acid effect (especially at longer incubation times) appears to be consistent with the preliminary heating results previously discussed(Rapoport and Shadwick, 2002)as well as the heating results from this study that follow.

Replicates were cycled ten times each at pH of 2, 5, 7 and 10. The tenth cycle from each sample was used for comparison in order to ensure stability of the stress-strain response. Visual inspection of the results of each piece of WECB in buffer of differing pH suggests that only pH 2 has any noticeable effect (Fig. 9). Statistically,resilience and elastic modulus of the yield region did not vary(P=0.319) and (P=0.986), respectively, across pH (Figs 11, 12 and 13). Yield stress did vary significantly across treatments (P=0.003) with a Tukey HSD test (95%confidence limit) showing that pH=10 and pH=7 were both significantly different from the pH=2 treatment. The pH=5 treatment was not significantly different from the pH=2 treatment although qualitatively this appeared to be the case (Fig. 12). The Hookean region modulus did not vary significantly across treatments based on a Tukey HSD test (95% confidence limit), but differences were suggested through an ANOVA (P=0.046) (Fig. 13).

Fig. 14 shows the extension results from a representative specimen showing the transition that occurs as the yield stress decreases with increasing temperature. At 92°C the Hookean behavior disappears completely. Not shown is the recovery curve following the thermal treatment that illustrates, as in the other treatments,the ability of WECB to apparently return to its initial properties following perturbation with harsh treatments that normally cause protein denaturation or solubilization. There are qualitative behavioral similarities between the 92°C treatment, and the 1-h incubation time in 88% formic acid, both of which appear similar in mechanical behavior to WECB retrieved from the VPG before processing is complete (Fig. 7II). The Hookean region modulus decreases steadily with increasing temperature, but due to high variability these changes are not statistically significant (P=0.702; Fig. 13). Since there was no Hookean region in the 92°C treatment, it was not considered in this comparison. Similarly, there was no yield stress for statistical comparison from the 92°C treatment. For the remaining temperature regimes, yield stress decreased steadily but, like the Hookean modulus, these changes were not statistically significant (P=0.702; Fig. 12). Elastic modulus in the yield region was unaffected statistically by the temperature treatments(P=0.695; Fig. 13). Lastly, resilience values did show a significant difference(P=0.043), but only in a comparison between resilience values from the 92°C and 7°C treatments. Qualitatively, differences could be seen,but variation in the experiment failed to reveal strong statistical differences.

Results from dynamic testing of native B. canaliculatum specimens are expressed in two graphs consisting of frequency as the independent axis and two respective dynamic mechanical parameters as the dependent axes. Tanδ, an index of phase shift between the force generated in a specimen and its excitation displacement (viscous damping) is one parameter. Tanδprovides information about mechanical hysteresis as well as resilience in a material. The relationship between Tanδ and hysteresis is directly proportional, while the relationship of Tanδ with resilience is inversely proportional. The other parameter that can be realized from dynamic testing is the storage modulus (E′), an index of elastic energy storage. Results from the three test modes (Hookean, yield and vibe-ramp) were plotted together to assess differences. Fig. 15 shows Tanδ and storage modulus (E′) results from the aforementioned experiments plotted against frequency. In all cases, a weak regression coefficient is indicated with r² values<0.5, and in several cases, r² is <0.1: (Tanδresults: Hookean, N=5, r²=0.0149; yield, N=6, r²=0.0625; vibe-ramp, N=5, r²=0.4077; E′ results: Hookean, N=4, r²=0.30; yield, N=6, r²=0.27; vibe-ramp, N=5, r²=0.535). Prior to cross-test mode statistical comparisons, within-test mode statistical comparisons were made and from these it was determined that within-test mode replicates were not significantly different from each other at a 95%confidence level (Tanδ results: Hookean, N=5, P=0.593;yield, N=6, P=0.448; vibe-ramp, N=5, P=0.815; E′ results: Hookean, N=4, P=0.584;yield, N=6, P=0.367; vibe-ramp, N=5, P=0.464). The storage modulus (E′), of the Hookean test mode was significantly greater than E′ both for the yield (P=0.001)and vibe-ramp test modes (P=0.20), which were not themselves statistically different from each other (P=0.116). Tanδ results show that the Hookean test mode is significantly different from the vibe-ramp test mode (P=0.0). The Hookean test mode, however, is not statistically different from the yield test mode (P=0.743).

These dynamic tests were repeated following the treating of specimens with 88% formic acid for 1 h (significant differences were noted for the longer incubation time warranting further exploration). Fig. 16 shows results from the 1-h 88% formic acid treated WECB specimen. Based on the shape of the extension curve for 88% formic acid treated specimens for this time duration, yield and vibe-ramp test modes were not possible. Tanδ results indicated that the formic acid treatment is significantly different from the other tests(w/Hookean, P=0.0015; w/yield, P=0.0; w/vibe-ramp, P=0.0). Similarly, the storage modulus is significantly lower than Hookean and vibe-ramp tests (P=0.0129; and P=0.0008,respectively), but not the yield mode (P=0.0610).

To summarize these results, storage modulus and Tanδ results for the three test modes all possessed weak regressions as a function of frequency,with Hookean and yield Tanδ possessing extremely weak regressions indicative of virtually no frequency dependence. Within-test mode statistics failed to show differences, thus indicating fidelity among the replicates. Variability among the test modes is summarized as follows. The Hookean storage modulus is different from the storage modulus of yield and vibe-ramp, which are not different from each other. The Tanδ for the Hookean test mode is different from vibe-ramp, but not yield. Lastly, the formic acid Tanδ is different from all other experiments, and the storage modulus from formic acid treatment is different from Hookean and vibe-ramp, but not yield. Qualitatively, the storage modulus for the formic acid-treated specimen was lower than either yield or vibe-ramp, as expected, but these differences failed to materialize in the statistics.

Fig. 17 illustrates the results from five dehydrated specimens extended to failure. Failure ranged from approximately 40% to 140% strain. There is an order of magnitude increase in yield stress [mean value (± s.e.m.) = 37.0±5.56 MPa, N=5], and little to no change in yield strain (mean yield strain was 3.5±1.0%, N=5), when compared with hydrated tested specimens. Initial modulus was also an order of magnitude greater than hydrated specimens(mean value = 900.96±58.91 MPa, N=5). Mean toughness of these specimens was 72.57±16.67 GJ m^-3 (N=5). Mean breaking strain was 84.9±19.5% (N=5).

There have been numerous competing attempts over the past 30+ years to construct a model that describes the important contributing factors to the mechanical properties of hard alpha-keratin(Feughelman, 1959; Chapman, 1969; Feughelman, 1979; Wortmann and Zahn, 1994; Hearle, 2000). One important question is how matrix and IFs each contribute to the mechanical properties. These models were revisited (Fudge and Gosline, 2004) and the compelling argument, based on examination of a matrix-free IF analog (hagfish slime threads), was presented that the matrix limits hydration of IFs, thus regulating water's role as a plasticizer;this in turn leads directly to the observed stress-strain responses. Furthermore, this `matrix squeeze' hypothesis(Fudge and Gosline, 2004) is based on the presence of covalent disulfide crosslinks that appear to function in IF-IF, matrix-IF and inter-IF roles. It is thought that crosslinks, in binding matrix together with IFs, restrict the hydration of IFs, which in turn directly impacts mechanical performance through alterations in elastic modulus(i.e. stiffness).

WECB is hydrated in its native state, and therefore matrix-squeeze modeling is not entirely appropriate for this material. However, there are aspects to the model that apply to WECB. Observations on how IFs can or should behave with various modes of covalent crosslinking and differing hydration levels may be relevant to WECB. In particular, simple crosslinking between adjacent coiled-coils and possibly combined with crosslinking between non-coiled-coil terminal regions in IFs should lead to high initial modulus (stiffness) and relatively low yield stress, as seen in WECB. The lack of these crosslinks and matrix in hagfish slime threads means that there should be low initial modulus and low yield stress, which is indeed the case. However, by dehydrating hagfish slime threads, modulus and yield stress can be substantially increased because the lack of a plasticizer allows H-bonding to dominate. In hard alpha-keratin, a very stiff and brittle material, the addition of water acts as a plasticizer allowing a decrease in stiffness, but not much due to the`matrix squeeze model'. WECB fits somewhere in between these two materials. It is normally hydrated and thus possesses lower stiffness and strength, but covalently crosslinked, and when dry, exhibits marked increases in stiffness.

WECB's mechanical behavior is interesting because it can be tied to a discrete manufacturing process terminating in the snail's ventral pedal gland. Based on SEM comparisons of immature capsules with mature capsules, we now know that the mechanical integrity of WECB does not arise from any gross structural changes resulting from manipulations occurring in the pedal gland. Rather, the capsule structure is completed in the nidamental gland and is both molded and somehow stabilized to a mechanical cohesiveness by activity of the ventral pedal gland. We conclude that since the hierarchical fibrous structure is already present in the immature capsules, it is highly likely that treatment by the ventral pedal gland induces formation of crosslinks and possibly adds matrix, ultimately leading to the creation of an insoluble protein polymer. Furthermore, qualitative observations of the mechanics of these capsules as they are processed prior to complete stabilization suggest that the complex bimodal mechanical behavior develops sequentially. The preliminary mechanical properties seem likely to be related to the long-range elasticity seen in the yield region of the native egg capsules, or alternatively, a much simpler rubber-like behavior existing in the nascent capsule prior to its final crosslinking. Irrespective of the nature of the initial properties at the early stages, the Hookean region appears to develop later in processing and heralds the completion of the capsule maturation process.

Based on our observations, and observations for alpha-keratin and hagfish slime threads, we advance the following model on the structural stability and mechanics of WECB (Fig. 18).(A) In general WECB consists of sheets (foils) of 1 μm diameter macrofibrils consisting of hierarchical assemblages of intermediate filament(IF)-type structures arranged in parallel (cylindrical structures). SEM images(Fig. 6) suggest that the macrofibrils may not be discrete structures, and that there may be connections among macrofibrils. There is presumably matrix associated with the IFs, but it is not well characterized at this time. Successive layers are arranged in varying orientations throughout the thickness of the material and are presumably responsible for bulk orthotropic mechanical behavior in the plane formed by foils. As the inset illustrates, each macrofibril comprises the staggered head to tail arrangement of coiled-coil molecules at its smallest level of organization (Gathercole,1969; Flower et al.,1969). Alignments of coiled-coils favored by charge-based self-assembly are responsible for repeat striation patterns seen in EM sections. For simplicity, IF structures, the next hierarchical level of organization, are not shown in the model. (B) Tensile loading results in shearing and sliding of successive layers, leading to material failure. Immediately following formation in the nidamental gland (NG), WECB is white in color and still soluble, indicating lack of sclerotization. Bulk mechanical measurements at this stage are hampered by a lack of cohesiveness in the material, but noncovalent interactions (further supported by the ease at which material at this stage can be rendered soluble; H.S.R., unpublished observation) are probably responsible for maintaining its structure and rudimentary cohesiveness. Here we conjecture that the application of tensile force allows the foils to slide apart fairly easily. Since the individual foils appear to be discontinuous and interdigitated with other layers (foils appeared to be laid down in a fashion analogous to disordered strokes of a broad painting brush; thus, the foils do not necessarily span the entire length of the egg capsule), the loose material associations resulting from self-assembly are not sufficient to hold the material together and difficult to measure at the bulk material scale. (C) Shearing of successive sheets with restoring force provided by sporadic linkages, primarily among macrofibrils(idealized crosslinks, illustrated as black lines in inset). During treatment in the ventral pedal gland (VPG) crosslinks begin to stabilize WECB by linking successive foils as the muscular action of the VPG brings them closer together. These crosslinks are hierarchically ubiquitous, stabilizing IFs,macrofibrils and matrix. At early stages of crosslinking, the density of the crosslinking is minor (see inset as well), and the material behaves much like a pliant rubber with a restoring force provided by bulk deformations of matrix. Mechanics are dominated by features on the bulk material scale (i.e. changes in foil position and perhaps some warping of foils). (D) Tightly crosslinked material with stresses transferred down to level of coiled-coils. Following processing in the VPG, the crosslink density is now sufficient to transfer mechanical stress down to the smallest hierarchical level of the material, thus adding a level of complexity to the witnessed mechanical response. Here, we see the development of the Hookean region that results from strain directed to a network of stiff coils. Unraveling of coiled-coils into random coiled configurations (see inset) begins at the transition between the Hookean and yield regions. In the yield region, extensive unraveling of coiled-coils occurs. Under physiological conditions this is not a stable conformation, so recovery of the initial state occurs when the strain is removed; the restoring force is provided by a combination of matrix contributions and entropic mechanisms, but not necessarily due to the IFs themselves (Kreplak et al.,2005). The organization that allows the stiffer Hookean region to occur at initial strain application is an in-parallel loading of both coiled-coils and matrix.

We previously found that the Hookean region of the stress-strain behaviour could be reversibly eliminated from the material behavior with a thermal treatment pointing to the fact that the Hookean region is heat labile and the yield region is not. The present study determined that high concentrations of formic acid as well as buffers with low pH could also mimic aspects of the thermal effect. In sum, the results of these treatments suggest a crosslinked polymer with the ability to maintain long-range elasticity in conditions that might otherwise be destructive to proteins.

There is utility in exploring the conditions necessary for diminishment of the Hookean region. A synthesis of the salient features of alpha-keratin mechanical models suggest that the Hookean region represents coiled-coil structural motifs, and that by applying strain to the material, sufficient energy is applied to begin unraveling these domains at the transition between the Hookean and yield regions. Additional unraveling events can accommodate further extension within the yield region(Kreplak et al., 2004).

How does the behavior of WECB resemble this? First, support exists in the observation that there is a recovery feature to the mechanics of WECB at extensions past the Hookean region that allow Hookean-like stiffness to reappear in the yield region. For example, consider a stress relaxation experiment conducted earlier (Rapoport and Shadwick, 2002), during which the WECB specimen was stretched into the yield region and subsequently held at constant strain. Stresses dropped almost immediately. When a strain was applied to the specimen, stresses rose in a fashion consistent with the Hookean region (high slope=high modulus=high stiffness), instead of being consistent with the yield region where the specimen was originally resting at constant strain. This suggests that during the stress-relaxation experiment, holding the specimen at constant strain allowed some sort of molecular reorganization as stresses fell that imparted behavior upon the material commensurate with the Hookean region. This molecular reorganization is probably due to the reconstitution of coiled-coils. Second, with thermal/chemical treatments we are probably seeing a decrease in the strain energy required to unravel these coiled-coils,particularly in the case of nonreversible formic acid treatment in which aldehydic functionalities are indiscriminately attaching to WECB in a fashion that impedes the reformation of higher order structure. The general mechanism of action of acid/heat treatments is the dissolution of non-covalent stabilization, probably hydrogen bonds. Thus, the yield point can be shifted to lower energy levels.

The data also support this sequence of events. There is a clear decrease in yield stress and elastic modulus of the Hookean region with all three denaturing treatments (Figs 10, 11). Changes in resilience(Fig. 8) are not as clear,except in the case of temperature treatment. In this particular case, high temperature was the only treatment that completely removed the Hookean region. Later explorations with longer incubation times at higher concentrations of formic acid also completely removed the Hookean region, and should be expected to exhibit the same resilience increase as ordered structure is replaced with random coils. Complete elimination of the Hookean region leads to changes in the overall shape of the stress-strain diagram and concomitant changes in resilience.

Dynamic testing afforded a better opportunity to look at the time course of coiled-coil reconstitution following perturbation. The `vibe-ramp' test mode consisted of extending the specimen into the yield region, but instead of holding at constant strain (and allowing Hookean behavior to re-develop) a small-value constant strain rate was maintained, which we believed would be sufficient to prevent molecular reorganizations from developing during the dynamic testing.

The storage modulus resulting from the Hookean test mode (85±2.3 MPa, N=48) was similar to the elastic modulus measured in quasi-static tests, but higher than the storage modulus resulting from either the yield (26±0.83 MPa, N=72) or vibe-ramp (22±0.83 MPa, N=67) test modes. This is supporting evidence that WECB possesses two distinct mechanical behaviors derived from coiled coil domains and matrix, with very different (yet not necessarily independent) properties. Curiously, results suggest that the vibe-ramp test mode did not have the desired effect of eliminating any possible Hookean contribution. That is,there are no significant differences between the behaviors of WECB under the vibe-ramp or yield test modes. Indeed, storage moduli in both these tests were higher than expected based on the elastic moduli from the quasi-static results(which range from approximately 2 to 10 MPa). This implies that the coiled-coil motifs consistent with the Hookean region (and, the likely hydrogen bonding that both stabilizes within and among these structures) are reforming or engaging even under continuous extension. This agrees with keratin models suggesting that in the yield region there are still coiled-coil motifs unraveling, and that the witnessed strain-relaxation effect during quasi-static testing must only be a partial reforming of the coiled-coil regions.

Tanδ (an indicator of viscous loss, related to both hysteresis and resilience) results suggest that there is no real difference in resilience among the three test modes when data spread is considered. This is consistent with the idea that strain energy alone cannot remove coiled-coil structure like the harsher of the acid/thermal treatments is capable of accomplishing. Additionally, in dynamic testing where small oscillatory strains are applied and material recovery is demonstrated, it suggests that not passing through a yield zone allows a convergence of hysteresis/resilience values. This is a departure from expected results. The Hookean region should have a larger resilience (i.e. lower Tanδ) than either the yield or the vibe-ramp test modes because of its high stiffness and low hysteresis. If we now consider the dynamic mechanical results of the longer, 1 h incubation time of the 88%formic acid treated specimen (which is effective at eliminating the Hookean region behavior from the material), the mean storage modulus (2.4±0.09 MPa, N=49) is much less than for any of the other dynamic tests, but similar to the elastic modulus of the yield region based on quasi-static tests. The mean Tanδ for the formic acid treatment (0.25±0.01, N=41) is in the same range as for the other tests (Hookean,0.28±0.01, N=61; yield, 0.28±0.006, N=72;vibe-ramp, 0.31±0.01, N=54), but slightly lower, which is consistent with the disruption of the non-covalent interactions and coiled-coil structures that generates the Hookean region.

It is the interplay of hierarchical levels of organization that allows WECB to present interesting mechanical properties. As an extracellular keratin-like material from the marine environment, WECB still displays characteristic behavior that can be defined by keratin models and elucidated by matrix-free modeling of hagfish slime threads (Fudge et al., 2003). We have shown that WECB possesses lower stiffness(∼100 MPa) when compared to hydrated alpha-keratin (∼2-6 GPa). This is likely explained by the matrix-squeeze model(Fudge and Gosline, 2004),where WECB's native hydrated state requires less covalent crosslinking for its function when compared to stiffer keratins such as horn, hair and hoof. Additionally, we see that we can generate a stiffer, stronger material by dehydrating WECB, at which point both its yield stress and initial modulus rise by an order of magnitude (37.0±5.56 MPa, N=5) and(900.9±58.91 MPa, N=5), respectively. This is analogous to the changes seen between wet and dry alpha-keratin, but differing in magnitude. Yield strain in dehydrated WECB remains essentially unchanged from the hydrated form and breaking strain decreases from 95% to 85%, although the dehydrated breaking strain had a degree of variation, probably due to incomplete drying among samples. In comparison with hydrated keratin, it is interesting to note that breaking strains for these two materials are similar:50% (hydrated alpha-keratin) to 85±19.5% for WECB. Drying WECB appears to make it more similar to hydrated alpha-keratin, but because crosslink and/or matrix density is comparatively low, WECB never truly attains a modulus on par with keratin. Again, this is probably due to the environment in which these materials are expected to operate. A further consideration is that WECB functionally acts as a protective, yet diffusive barrier for developing embryos. Covalent crosslinking of the same magnitude seen in keratin (based on comparison of detectable cys residues as a proxy for disulfide interactions)might impede this function by creating a much more compacted and water resistant material.

Lastly, harsh denaturants appear to cause WECB to behave in a fashion analogous to wet hagfish slime threads, although the utility in comparing these two materials in this fashion is minor because WECB presumably has matrix present irrespective of treatment, whereas slime threads do not. Based on this model, we predict that in the case of large strains, sufficient change to coiled-coil conformation and resultant fiber banding periodicity should be observable as a change in birefringence (or through the use of X-ray diffraction). Relaxation of the capsule that occurs during a constant strain experiment in the yield region will probably also present a change in conformation detectable by birefringence microscopy or X-ray diffraction as coiled-coils reform.

We have already almost completely identified a gene for WECB. With complete gene information, we hope to undertake expression experiments in which a more detailed understanding of structural relationships to mechanical properties[see other work (Keeley et al.,2002)] can be determined. Moreover, complete gene information should provide additional corroboration of coiled-coil structure and enhance WECB's characterization as a keratin analog. Finally, we intend to obtain protein sequence information from the primary protein(s) present in the ventral pedal gland during the egglaying portion of the reproductive season. It is our hope that the isolation of temporally expressed gene products in the ventral pedal gland will lead to an understanding of the biochemistry responsible for the sclerotization of the capsules.

We thank Dr Herb Waite for technical collaboration and useful discussions on structural proteins. We thank Eddie Kisfaludy (and Ron McConnaughey) for providing marine material support. We thank Jeff Ram for providing capsule specimens. This manuscript benefited greatly from the careful reading of an anonymous reviewer and of Drs Jon Gosline and Doug Fudge. We thank Russ Moll for his unwavering support. This project was supported by funding from California Sea Grant, R/MP-91, and NSERC.