Extensible Collagen in Mussel Byssus: A Natural Block Copolymer

See allHide authors and affiliations

Science  19 Sep 1997:
Vol. 277, Issue 5333, pp. 1830-1832
DOI: 10.1126/science.277.5333.1830


To adhere to solid surfaces, marine mussels produce byssal threads, each of which is a stiff tether at one end and a shock absorber with 160 percent extensibility at the other end. The elastic extensibility of proximal byssus is extraordinary given its construction of collagen and the limited extension (less than 10 percent) of most collagenous materials. From the complementary DNA, we deduced that the primary structure of a collagenous protein (preCol-P) predominating in the extensible proximal portion of the threads encodes an unprecedented natural block copolymer with three major domain types: a central collagen domain, flanking elastic domains, and histidine-rich terminal domains. The elastic domains have sequence motifs that strongly resemble those of elastin and the amorphous glycine-rich regions of spider silk fibroins. Byssal thread extensibility may be imparted by the elastic domains of preCol-P.

Mussel byssal threads are undoubtedly among nature's most peculiar tendons. One end of each thread inserts into the byssal retractor muscles at the base of the foot; the other end is disposed outside the animal and attached to a hard surface by an adhesive plaque. In a process that resembles injection molding, the foot of the mussel is able to make a new thread in 5 min or less (1). Despite their rapid production and vulnerable location, byssal threads are durable and exquisitely engineered fibers (Fig. 1A). They contain a graded distribution of tensile molecular elements that result in a material that is strong and stiff at one end and pliably elastic at the other (2, 3). Overall, byssal threads are five times tougher than Achilles tendon (4).

Figure 1

(A) Schematic representation of M. edulis byssus subdivided into four morphologically distinct regions: adhesive plaque, stiff distal thread, elastic proximal thread, and stem. For clarity, all but one thread were cut from the stem. Insertion denotes point of attachment to mussel. Typically, threads are 2 to 4 cm long with diameters of 0.10 to 0.15 mm. (B) Model of the block-like structure of domains in a homotrimer of α-preCol-P from the proximal portion of a byssal thread. Mass estimates of domains are based on the molecular sequence indicated in Fig. 2 and analyses of preCol-P by matrix-assisted laser desorption–ionization mass spectrometry (17).

The collagen content of mussel byssal threads has been well established by wide-angle fiber x-ray diffraction, the presence oftrans-4-hydroxyproline, and byssus-derived pepsin-resistant peptides with Gly-X-Y repeats (5, 6). Unlike tendon, however, byssus has a nonperiodic microstructure and shrinkage and melting temperatures in excess of 90°C (5). Any attempt to reconcile these unusual biochemical and mechanical features with what is known about collagen provokes the following molecular questions: How is byssal collagen different from tendon collagen, and what role does this difference play in the material performance of byssus?

Because of its highly cross-linked structure, there has been only one recourse for recovering collagen from byssus: acid extraction coupled with extensive treatment with pepsin. We (6) used this approach to characterize two collagen fragments, Col-P and Col-D (both apparently homotrimers), with apparent α-chain masses of 50 and 60 kD, respectively (7). These fragments coexist in complementary gradients along the length of each byssal thread with Col-D predominating at the distal end and Col-P predominating at the proximal end. Similar gradients were found to exist in the mussel foot, which fabricates the byssus one thread at a time. Precursors to Col-P and Col-D (preCol-P and -D) were identified in foot extracts with specific polyclonal antibodies. Apparent masses of 95 and 97 kD were determined for α-chain preCol-P and preCol-D (α-preCol-P and α-preCol-D), respectively. The amino acid compositions for the NH2- and COOH-terminal extensions consist of Gly plus Ala concentrations of 50 (preCol-P) and 60 (preCol-D) mole percent (mol%) (6). Although it was inferred that the extensions or flanking domains might resemble structural proteins such as silk fibroin or elastin or resilin, confirmation of such homology has awaited determination of the complete primary structure. Here we report the primary structure of α-preCol-P, a natural block copolymer with putative cross-linking, elastic, and collagen domains (8).

α-PreCol-P may be divided into seven domains based on its amino acid sequence (Fig. 2). The most prominent feature is a central region consisting of a 435-residue collagenous sequence identified by alignment with amino acid sequences reported previously for peptides derived from the pepsin-resistant fragment Col-P (6). The collagen domain consists of 146 Gly-X-Y repeats in which X and Y are frequently Pro. Only the Pro residues at position Y appear to be converted totrans-4-hydroxyproline (6). Gly also occurs in the X (10 times) and Y (once) positions. The continuity of Gly-X-Y repeats is interrupted after the eleventh repeat (Gly-Ser-Thr) by a single missing Gly. The effect of this on the triple-helical structure of collagen is unclear at this time; however, it is noteworthy that this region is not susceptible to cleavage after extensive treatment of the native homotrimer with pepsin (6). It is tempting to suppose that omission of one Gly in each α chain induces a kink or bend in the triple helix. However, the best studied kinked collagen, complement Clq, is a heterotrimer in which one chain has no triplet discontinuity, another has an Ala for Gly substitution, and the third is missing Gly-X of the triplet (9). The collagen domain of α-preCol-P is followed by an acidic patch about 15 residues long that is enriched in Glu and Asp.

Figure 2

Predicted amino acid sequence (34) of α-preCol-P. Full sequence from cDNA is arranged to exhibit the block copolymer structure of preCol-P and to align the characteristic repeat motifs of each domain. Thus, elastin-like pentapeptides (boldface) are aligned at P to distinguish them from intervening GGX clusters and Ala-rich runs (double underline), and collagen is arrayed as a series of tripeptides. The nucleotide sequence has been deposited in the GenBank database (accession number AF015539). Lowercase, signal peptide; bold italics, histidines in His domain; •, missing G; single underline, acidic patch; and dashed underline, partial peptide sequences from (6).

The elastic domains flank the collagen domain on both sides (Fig.2) and exhibit sequence and solubility similarities with elastin (10). Both are dominated by Gly, Pro, and bulky hydrophobic residues and have Pro-containing pentapeptide repeats (that is, X1-X2-X3-Pro4-X5). The Pro is not hydroxylated (6). In the elastic domains of preCol-P, X is usually Gly but can be Phe (nine cases), Ala (four cases), Ile (two cases), or Val (two cases) (Fig.3). These resemble the pentapeptide repeats of elastin, the most common of which is Val-Gly-Val-Pro-Gly (11). Gly, Pro, and the hydrophobic residues of elastin are critical to the elastic recoil, which is entropically driven (12). The entropy is derived from dynamic hydrophobic interactions or “librations,” which are maximized in unstretched protein and accommodated by the flexible structure imposed by Gly and Pro. Similar librations are predicted for the elastic domains of preCol-P. All in all, there are 6 pentapeptide repeats in the NH2-terminal side and 12 in the COOH-terminal side of the flanking elastic domains. Both elastin and preCol-P also have poly(Gly) clusters peppered with hydrophobic residues (Leu, Ile, Val, and Phe) and Ala-rich regions (Figs. 2 and 3). Many of these features are evident in lamprin (13) and spider silk fibroins (14) as well. The Ala-rich regions of α-preCol-P are more substituted than the poly(Ala) tracts of elastin, in which substitutions are limited to Lys, and spider silk fibroins, which lack substitutions except for occasional Ser residues (Fig. 3). Curiously, in spider silk fibroins, poly(Ala) runs evidently form nanocrystalline β sheets (15), whereas in elastin they form α helices (16).

Figure 3

Sequence alignment of repeated motifs in the elastic flanking domains of α-preCol-P with consensus repeats of the known extensible proteins bovine elastin (10) and spider dragline protein ADF-3 (12). X denotes hydrophobic residues such as F, V, L, I, and A; Z denotes polar residues such as S, N, and R. Number of repeats in each protein is indicated in parentheses after the sequence. Shaded areas represent conserved sequences.

His-rich regions are located at the NH2- and COOH-termini. The only Tyr and noncollagenous Lys residues of preCol-P are also found there. The presence of trace 3,4-dihydroxyphenyl-l-alanine (dopa) in hydrolyzed preCol-P suggests that, as in other byssal precursors, Tyr may be posttranslationally hydroxylated to dopa in the mature protein (17). The His-rich domains comprise about 80 and 50 residues at the NH2- and COOH-termini, respectively. At both ends, His represents 20 mol% of the composition or one in five residues. Other prominent residues are Gly, Ala, and Ser (Fig. 2). His-rich domains occur in several other proteins including high molecular weight kininogen (18), plasmodial His-rich protein HRP (19), blood His-rich glycoprotein HRG (20), and trematode eggshell precursors (21). Sequence homology is limited to short repeats—for example, Ala-His compared with Ala-His-His repeats in plasmodial His-rich protein, and His-Gly-Gly or Gly-Gly-His compared with Gly-His repeats in the kininogen and eggshell precursors. The function of the His-rich tracts in almost all known cases is connected with metal binding such as zinc in kininogen and HRG and heme iron in HRP. His-rich nereid jaws are high in complexed copper or zinc (22), and poly(His) tracts are routinely added to recombinant proteins for facile recovery by metalloprecipitation or metalloligand affinity chromatography (23). Of the transition metals detected in Mytilus edulis byssus, zinc levels are consistently the highest at about 1 mg per gram of dry weight (24). Studies are under way to determine whether the recovery of initial stiffness in distal byssal threads after yield to 0.5 extension (25) is related to the breakage and re-formation of protein-zinc complexes.

The results reported here are unusual in several respects. PreCol-P is the first known protein to contain both collagenous and elastin-like domains. Indeed, elastins were thought to be absent in invertebrates (26). In addition, the tandem array of collagenous and elastic domains is remarkably reminiscent of synthetic block copolymer designs with their hard and soft segments (27) (Fig. 1B). In this case, “blocks” of elastin (breaking strain, ≤1.6; tensile stress, ≤5 × 106Pa) would improve the extensibility of the collagen (breaking strain, ≤0.1; tensile stress, ≤1 × 108 Pa) (28). This results in a material with breaking strength lower than that of tendon but with significantly greater toughness (3, 4). Block copolymer design has also been detected in another byssal precursor, preCol-D, which has silk fibroin-like instead of elastic flanking domains in a sequence that is otherwise homologous with that of preCol-P (29). Future studies would profit from a closer scrutiny of the mechanism of preCol fibrillogenesis as well as the interplay of hard (collagen) and soft (elastin) segments in determining the peculiar mechanical properties of byssal threads.


View Abstract

Stay Connected to Science

Navigate This Article