Collagens chiefly fromscaffolding oknown metazconstruct materials outside the confines of the living organism(Adams, 1978; Gross, 1963). Gooseberry sawfly silk (Spiro

etal. 1971), selachian egg capsules (Knight et al. 1993), wormcuticle collagens (Kramer et al. 1982; Mann et al. 1992),spongin (Garrone, 1978) and mussel byssus are but a fewexamples of such exoskeletal collagenous materials. With thepossible exception of the worm cuticle collagens, none of thesehas been particularly well studied. Indeed, in most, theexistence of collagen is based on little beyond amino acidanalyses that have included the detection of hydroxyproline,staining of specimens for transmission electron microscopyand, occasionally, wide-angle fiber X-ray diffraction.

The byssus of the common mussel Mytilus edulis hasintrigued biologists since the beginning of the century (Seydel,1909). This can be at least in part attributed to the silken byssal

(diamdly fo other nous’

1955; Merceramino acid composition (Benedict and Waite, 1986; DeVoreet al. 1984; Pujol et al. 1976) and shrinkage temperatures(Pikkarainen et al. 1968). Although the collective evidencepresents a strong case, some of the extraordinary properties ofbyssus are hard to reconcile with ‘typical’ collagens. Theseinclude the high melting temperature, lack of banding patternsand resistance to denaturants, acids and proteases (Brown,1950; Pikkarainen et al. 1968). Moreover, according to severallines of evidence, byssal threads are not uniformly collagenousthroughout, i.e. the diffractive and mechanical propertiestypical of fibrous collagen are best reflected at the stiff distalend of the thread, whilst the more extensible or elastic proximalend seems to be disordered (Gathercole and Keller, 1975;Price, 1981; Smeathers and Vincent, 1979; Vitellaro-

The Journal of Exp

Printed in Great Br

Byssal th

contain colresistant fraThese show of the threadand the othethree identi

50 kDa (Colcollagenous glycine anhydroxyprolabsent. Col-consisting ofX and Y genoccurs, it is hthe Y positiodetected in

tomu

and c (13 %

r prect. Thed silkagenoogoldted wal portraighuter

rds: byuence.

E BY

WColle mi

*Author for correspondence.

eter 0.1 mm; length 2–4 cm) which a mussel canrm at a few minutes notice for attachment to asolid surface (Fig. 1). These threads were dubbed on the basis of fiber X-ray diffraction (Rudall,, 1952), hydroxyproline content (Melnick, 1958),

633

isolate the precursors of Col-P and Col-Dssel foot. PreCol-P has a molecular mass ofontains 36 % glycine but a lower imino acid). It has a complementary distribution with

ursor (preCol-D, 97 kDa) along the length of two precursor compositions suggest resilin--fibroin-like structures, respectively, in theus domains of preCol-P and preCol-D. labelling studies indicate that Col-P isith the coiled fibers of the inner core in thetion of the thread, whereas Col-D is localizedt fiber bundles of the distal thread as well as

core of the proximal thread.

ssus, collagen, gradient, Mytilus edulis, mussel, amino

SSUS OF THE MUSSEL

AITE*stry, University of Delaware, Newark,

erimental Biology 198, 633–644 (1995)itain © The Company of Biologists Limited 1995

reads of the common mussel Mytilus edulislagenous molecules from which two pepsin-

were usedfrom the

Summary

XOTIC COLLAGEN GRADIENTS IN THE MYTILUS EDULIS

XIAOXIA QIN AND J. HERBERTge of Marine Studies and Department of Chemistry/Bioche

DE 19716, USA

Accepted 4 November 1994

are highly diversified structural proteins derived mesenchymal cells and employed in the

f connective and interstitial tissues. In some lesseroans, collagens of ectodermal origin are used to

threadsrepeaterock or‘collage

gments have been isolated and characterized.a complementary distribution along the length, such that one predominates distally (Col-D)r proximally (Col-P). Both fragments containcal a-like chains with molecular masses of-P) and 60 kDa (Col-D) and have typicallyamino acid compositions; for example, 35 %d almost 20 % proline plus 4-trans-ine. Hydroxylysine and 3-hydroxyproline wereP sequences are also typical of collagen in tandem repeats of the triplet Gly-X-Y in whicherally represent any amino acid. When prolineydroxylated to 4-trans-hydroxyproline only in

n. Seven instances where X is glycine have beenCol-P. Specific polyclonal anti-Col antibodies

95 kDacontentanothethe foolike annoncollImmunassociaproximto the sto the o

Key woacid seq

Introduction

634

Zuccareprotein glength oMascolohow suformatio

Byssagranularinjectionbe madholocrinventral they coorientedfrom th(Waite, precursoexamplecloser sc

Our aof collacollagentechniquthe foot

byssal threads. One of these, Col-D, prevailson of each thread, while the other, Col-P,e proximal thread. Viewed over the entire

bution of the two collagens is roughly

Materials and methodsus edulis L.) were collected locally fromar the mouth of the Broadkill River, DE,

ned in an aquarium with running sea water byssal threads were harvested every otherdistilled water at 220 ˚C prior to use.

n of collagens from byssal threads

were homogenized to a purée in 5 % acetic

X

Fig. 1. A

collagens exist in in the distal portipredominates in ththread, the districomplementary.

Mussels (MytilRoosevelt Inlet neUSA, and maintaiat 15–20 ˚C. Freshday and stored in

Extractio

Byssal threads

. QIN AND J. H. WAITE

e

g

e

llo et al. 1983). This has led to the suggestion thatradients including a collagen gradient occur along thef each byssal thread (Gathercole and Keller, 1975; and Waite, 1986). Nothing is known at present aboutch gradients might be contrived during threadn.l threads are assembled by the mussel foot from

acid with 4 mol l21 urea using frosted-glass tissue grinders(Kontes, Vineland, NJ, USA) and centrifuged at 31 000 g(average force) for 30 min at 4 ˚C (Miller and Rhodes, 1982;Benedict and Waite, 1986). Following a brief rinse, the pelletwas resuspended in 0.5 mol l21 acetic acid, pH 2.6, containingbetween 1:10 and 1:30 pepsin:insoluble protein (Boehringer-Mannheim, Indianapolis, USA) and digested for 48 h at 4–7 ˚C,whereupon it was again centrifuged at 17 000 g for 30 min. Thesupernatant was collected and dialyzed against 0.1 mol l−1

sodium borate, pH 8.0, to precipitate collagens preferentiallywhilst inactivating pepsin. Precipitates were harvested bycentrifugation (31 000 g for 60 min) and redissolved in0.5 mol l21 acetic acid. Residual insoluble substances wereremoved by centrifugation. Collagens were again precipitated,this time by adjusting pH to 7.0 with NaOH and centrifugationat 17000 g for 30min. The pellet was redissolved in 0.5 mol l21

acetic acid with 4 mol l21 urea and prepared for high-performance liquid chromatography (HPLC). Since the pelletis only partially soluble in acetic acid with urea, it typicallyrequired centrifugation on an Eppendorf microcentrifuge(10 min at 15 000 revs min21) prior to application to a C-8column for HPLC. The 250 mm34.6 mm column consisted of

mussel dangling by its byssal threads.

precursors in a rapid process resembling polymer-molding (Waite, 1992). When a new thread needs to, fiber-containing granules are released from large

e glands and conducted through ciliated ducts into theroove of the foot, which serves as the mold. There

alesce into a thread that is shaped, and possibly, by peristalsis in the groove. Before being disengaged groove, the thread is sized by a protective coating1992). Such processing of a diverse assortment ofrs into a natural polymeric material is an exquisite of microengineering and, in our opinion, worthy ofrutiny.

ims in the present study were to establish the presencegen in the byssus, to isolate and characterize the and its precursors, and to use immunochemicales to localize these in the byssal threads as well as inof Mytilus edulis. Our findings suggest that two unique

RP-300 Aquapore (Applied Biosciences Inc.) and was elutedwith a linear gradient (15 % to 28 %) of aqueous acetonitrilewith 0.1 % trifluoroacetic acid (Waite et al. 1985). Eluent wasmonitored continuously at 230 nm, and 1 ml fractions werecollected and assayed by amino acid analysis andelectrophoresis following freeze-drying.

To determine the distribution of pepsin-resistant collagensin different parts of the byssal thread, about 300 threadsmeasuring 2–4 cm were carefully collected from musselsmaintained at 4–8˚C. The threads were then sectioned into fivesegments: the plaque and the cuff, the distal (rigid) andproximal portions of the thread and the transitional regionbetween the distal and proximal regions. Similar segmentswere pooled, homogenized and treated with pepsin, as above,to release sufficient protein for visualization followingpolyacrylamide gel electrophoresis in the presence of sodiumdodecylsulfate (see below).

Trypdigestioreactioof 50 mpH 7.5 constanin eachintervaelectrothe adfreezinwas dispeptidecalf skincubat70 % tto medilutioat 281965)electrowas 250 m(Rainiwas a 0.1 %

Pep10 % for 20acids excha(Beckprograderivasame peptida PortophenylchromaprogramknownPROT for therecommthousanCol-D

where and theThis wvalues a fracti

635Exotic collagen gradients in the mussel

Preparation and purification of peptides

sin (Boehringer-Mannheim, Indianapolis, USA)n of byssal collagen was performed using the following

n conditions. About 1 mg of Col-P was dissolved in 1 mlmol l21 Tris–HCl with 10 mmol l21 CaCl2 buffered toat a protease:protein mass ratio of about 1:10, witht stirring at 22–24 ˚C for 18 h. The progress of digestion case was monitored by removing 5 ml samples at 3 hls for inspection by sodium dodecylsulfate gelphoresis (see below). The digestion was terminated bydition of 0.3 ml of glacial acetic acid, followed byg (280 ˚C) and lyophilization. The freeze-dried residuesolved in 0.5 ml of 5 % acetic acid. Cyanogen bromides were generated from purified Col-P and types I and IIin collagen (Ethicon, Morristown, NJ, USA) byion (4 h for types I and II collagen; 48 h for Col-P) in

Electrophoresis

Routine electrophoresis was carried out on polyacrylamidegels (10 % acrylamide and 0.2 % N,N9-methylenebisacrylamide) in the presence of sodiumdodecylsulfate and discontinuous Tris–glycine (HoeferScientific Catalogue, 1989). Proteins were stained with ServaBlue R-250 (Serva Fine Chemicals, Westbury, NY, USA),with which collagens stain metachromatically, showing abathochromic shift of 30–40 nm (McCormick et al. 1979).Apparent molecular mass was determined by polyacrylamidegel (10 %) electrophoresis of samples and standards (lowmolecular mass range, BioRad, Richmond, CA). Isoelectricfocusing was carried out on PAGE in 8 mol l21 urea in the pHrange 3–10.

Western blotting

rifluoroacetic acid containing a 100-fold excess (relativethionine) of CNBr at 30 ˚C. Cleavage was terminated byn with a 10-fold excess of cold distilled water, freezing0 ˚C and lyophilization to dryness (Bornstein and Piez,. CNBr peptides were comparatively mapped byphoresis (see below). Resolution of the tryptic peptidesachieved by reversed-phase HPLC using a

m34.6 mm Phenomenex or Microsorb C-18 columnn Instruments, Woburn, MA, USA). The eluting solventlinear gradient of aqueous acetonitrile (0 % to 25 %) withtrifluoroacetic acid.

Amino acid analysis and sequencing of peptides

tides and proteins were hydrolyzed in 6 mol l21 HCl withphenol and 10 % trifluoroacetic acid in vacuo at 150 ˚C and 40 min to correct for the losses of certain amino(Tsugita et al. 1987). Amino acid analysis was by ion-nge HPLC and a ninhydrin-based detection systemman System 6300 autoanalyzer) using a gradientm described previously (Waite, 1991). Hydroxylatedtives of Pro and Lys were separately quantified by theprogram. The amino acid sequence of protein andes was derived by automated Edman degradation using

Proteins separated by SDS–PAGE were horizontallytransferred to Immobilon P (Millipore Corp., Bedford, MA,USA) by electrophoresis at 200 mA in transfer buffer(22.5 mmol l21 Tris borate, pH 8.5, in 25 % aqueous methanolwith 1 % sodium dodecylsulfate) for 2 h using a Genie transferunit with platinized electrode plates (Idea Scientific,Minneappolis, MN, USA). One panel of transferred proteinswas stained with 0.1 % Serva Blue G-250 in 40 % aqueousmethanol with 7 % acetic acid and destained with the samesolvent minus the stain. Protein bands (usually inquintuplicate) were excised with a clean single-edged razor andprepared as below for hydrolysis and amino acid composition(Tous et al. 1989). Specific polyclonal antisera against purifiedCol-P were raised under contract (Cambridge ResearchBiochemicals, Wilmington, DE, USA) in three female NewZealand rabbits according to standard immunization protocols.Antibody titer was tested by dot blots, and specificity wasdetermined by immunoblotting following electrophoretictransfer and visualized by a secondary goat anti-rabbit Fcantiserum coupled to alkaline phosphatase (Blake et al. 1984).Another polyclonal antiserum ostensibly against byssalcollagen was donated by Dr C. V. Benedict (Farmington, CT,USA). It was found to be preferentially reactive towards Col-

n Instruments microsequencer (Porton, Tarzana, CA).thiohydantoin (PTH) derivatives of amino acids weretographically separated according to a gradient specified by Waite (1991). Sequence homologies with

proteins were sought using the FASTA and SWISSdata banks (Lipman and Pearson, 1985). Calculations amino acid compositions of DD and DP were asended by Segel (1976). Briefly, the ‘residues perd’ value (RPT) for each amino acid of Col-P and

was converted to residues per molecule (R/M):

R/M = (RPT/1000)(Mcol/120) ,

Mcol and 120 denote the molecular mass of the protein average molecular mass of amino acids, respectively.as also done for preCol-P and preCol-D. The formerwere then subtracted from the latter and normalized toon of 1000.

D and was used at titers of 1:105 in immunoblots.

Identification of precursors in the foot tissue

Mussel foot tissues were examined for the presence ofcollagen precursors and prefabricated collagen gradients.Freshly amputated mussel feet were frozen on glass plates at280 ˚C. Pigmented epithelial and muscular layers were flayedfrom the frozen feet with a scalpel. Each foot was then seriallysectioned into 1–2 mm slices (as in Fig. 7, for example). Eachslice was individually homogenized in 20 volumes of 5 %acetic acid, microfuged (5 min) to remove insoluble material,and digested with pepsin (1:20 enzyme to tissue wet mass) for2 h at 4 ˚C. Pepsin digestion was terminated by adjusting thepH to 7.0. Parallel samples were homogenized in 5 % aceticacid with inhibitors (0.01 % pepstatin, 0.01 % leupeptin,1 mmol l21 iodoacetate and 0.1 mmol l21 phenylmethyl-

636

sulfonylacid-sol(5 min aof cold redissolseparateidentifietransferaside foabove.

Smallthreads These 0.1 mol lin 1 % osamples50 nm wWilminUltrathi10 % halbuminand incaccordinsectionsimmunoWashindomain dense trsectionsunder a

Solubbyssal tpepsin dtypicallynotable collagencollagenidentica60 kDa, extractathe sea examplewhereasthe loware less To imprresolvedextractemusselsextractioanalyzed

X

fluoride) at a 1:20 wet mass to volume ratio. Theuble materials were clarified by microcentrifugationt 14 000 revs min21) and precipitated by the additionperchloric acid (to 1 %). Precipitated material was

ved in cold 5 % acetic acid. Protein precursors wered by SDS polyacrylamide gel electrophoresis andd by immunoblotting following electrophoretic to Immobilon P. Parallel Immobilon panels were setr band excision and amino acid analysis, as described

Immunogold electron microscopy

pieces of the proximal and distal portions of byssalwere dissected from mussels maintained at 4–6 ˚C.samples were fixed in 2.5 % glutaraldehyde in21 cacodylate/HCl, pH 7.3, followed by post-fixation

. QIN AND J. H. WAITE

91

67

45

31

10−3×Mr

Col-P

have the compositions shownrk of collagen types I–III: one-ycine, and the proline plushes 20 %. Notably different isproline and hydroxylysine inains of denatured Col-P wereated by C-18 HPLC (Fig. 4).terminus of Col-P reflects thet lower pepsin concentrations,bserved; upon raising pepsinto the bare collagen domaind from Col-P conform to the

a byssal collagen (Col-P) by SDSs. Lane 1, low molecular masssin-solubilized fraction of byssus;neutralization precipitate; lane 5, stained with Serva Blue R-250.

3 4 5

smium tetroxide in the same buffer. After dehydration, were embedded in Epon-812 resin and sectioned atith a diamond knife (DuPont Biomedical Products,

gon, DE, USA) on a Reichert model E ultramicrotome.n sections mounted on nickel grids were treated withydrogen peroxide, blocked with 2 % bovine saline and 1 % cold fish gelatin in phosphate-buffered serum,

ubated with specific polyclonal antisera diluted 1:100g to procedures recommended by Hyatt (1984). The were then washed and incubated with 15 nmgold particles (Electron Microscopy Sciences, Fort

gton, PA, USA). The gold particles bind to the Fcof the primary antibody, thereby creating an electron-acker for its presence. After extensive washing, the were counterstained with uranyl acetate and examined Zeiss model 902 transmission electron microscope.

Resultsle proteins are at best poorly extractable from maturehreads. To remedy this, we attempted an extendedigestion of mechanically disrupted threads. Pepsin is indiscriminate in its digestion of proteins, with the

pepsin-resistant byssal collagensin Table 1. Both bear the hallmathird of the residues are glhydroxyproline content approacthe absence of 3-trans-hydroxythe byssal collagens. The a-chdigested with trypsin and separThe partial sequence of the N extent of digestion as follows: aan alanine-rich N terminus is olevels, this is trimmed down (Fig. 5). Tryptic peptides derive

Fig. 2. Tracking the purification of polyacrylamide gel electrophoresistandards (5mg of each); lane 2, peplane 3, borate precipitate; lane 4, Col-P as purified by HPLC. All are

1 2

exception of the triple helical domain of native. Prolonged pepsin digestion releases at least twoous fragments, Col-P and Col-D, each with threel a-chains with molecular masses of 50 kDa and

respectively. It is noteworthy that the ratio ofble Col-P:Col-D varies according to the temperature ofwater to which the mussels were introduced; for, at warm temperatures (18–20 ˚C), the ratio was 1:0.5, at 4–8 ˚C it was 1:2. Our experience suggests that, ater temperatures, Col-D-containing byssal precursorseffectively cross-linked and thus more easily extracted.ove the yield of pure Col-P, which is normally poorly by HPLC from Col-D (see Fig. 3), the former wasd only from threads of the warm-water-acclimated. Col-P was purified to homogeneity using preferentialn and reversed-phase HPLC (Figs 2, 3). Col-D was from protein electroblotted to Immobilon. The

classical collagen prototype, with glycine at every third residue(Fig. 5). However, without breaking stride from the repeatingtriplet, at least seven occurrences of double Gly are evident.Imino acids occur in either the X or Y position of Gly-X-Y, butonly those in the Y position are hydroxylated. As a result ofpoor peptide binding to the polybrene sample holder duringEdman sequencing, complete sequence to the C terminus wasrare. CNBr digestion of Col-P yielded two peptides of 36 and14 kDa, which is consistent with the deduced presence of onemethionine in Col-P (results not shown). The number andapparent size of the CNBr peptides from Col-P did not matchany of those from the bovine types I and II collagen controls.

Specific polyclonal antibodies were successfully raisedagainst Col-P in rabbits. In dot blot tests, these reacted strongly(titer 1:105 dilution) with Col-P but not perceptibly with Col-D, bovine collagen types I and II or with any of the standardproteins. The availability of an anti-Col-P antibody enabled us

637Exotic collagen gradients in the mussel

to address two questions concerning the distribution of Col-Pwithin the byssal thread and the distribution of its precursor,preCol-P, within the foot. Byssal threads (secreted at 4–8 ˚C)2–4 cm in length were cut into five segments, and segmentsfrom equivalent positions in about 300 threads were pooled,ground up and digested with pepsin. The pepsin-resistantproteins from each segment are shown in Fig. 6 afterSDS–PAGE. The distribution of Col-D and Col-P resembles apair of complementarymoves longitudinally whilst Col-P content inDo these gradients pre-they ‘manufactured’ dthis, a frozen mussel fsectioned transversely of equal length. Proteinwithout pepsin treatmeby electrophoresis andFig. 7 make three signantiserum recognizes

Table 1. Amino acid composition of byssal thread, Col-P, preCol-P, C

Amino acid Thread* Col-P PreCol-P DP Co

3-Hyp 1.9 0 0 04-Hyp 37.6 117.1 60.3 0 10Asx 82.7 52.9 50.7 50.7 4Thr 41.8 31.6 19.1 5.6 2Ser 66.3 50.8 74.4 105.1 2Glx 63.3 102.8 66.0 25.4 11Pro 64.3 83.2 73.9 62.9 8Gly 266.5 346.4 362.5 395.4 35Ala 100.2 65.3 97.6 130.8 8Cys/2 2.5 0 0 0Val 35.0 13.3 25.3 40.5 2Met 7.0 2.9 2.2 2.0Ile 41.1 13.8 20.5 29.4 1Leu 37.9 13.8 20.7 30.4 1Dopa 2.0 0 0 0Tyr 16.1 1.4 2.3 2.0Phe 19.0 3.4 23.3 46.6His 19.7 9.8 25.7 45.6Hyl 0.8 0 0 0Lys 43.0 36.4 26.0 16.2 2Arg 43.4 57.9 43.9 30.4 5

DD and DP represent deduced differences between each precursor and its pepsin-resistant fragmAmino acids are listed in order of elution time, and all values are reported in residues per 1000 *Benedict and Waite (1986).

B

A

0.10

0.10

Abs

orba

nce

at 2

30nm

% A

ceto

nitr

ile

80

60

40

20

80

60

40

20

0 10 20 30 40 50 60 70 80

Col-P

Col-D

Col-P

Col-D

Volume eluted (ml)

Fig. 3. HPLC profiles of pepsin-solubilized byssus on a C-8 reversed-phase column. (A) Extract of byssal threads formed at 18–20 ˚C.(B) Extract of threads formed at 4–8 ˚C. Peaks containing Col-P andCol-D were identified by SDS polyacrylamide gel electrophoresis ofthe fractions under the peak. Differences in the elution times probablyreflect the degree of pepsinization into the collagenous core.

gradients: the former decreasing as onefrom the adhesive plaque to the cuff,creases over the same course (Fig. 6).

exist in the secretory foot tissues, or areuring thread formation? To determineoot approximately 2 cm in length waswith a single-edged razor into segmentss extracted from each segment with ornt were analyzed following separation electroblotting. The results shown inificant points. First, that the anti-Col-Pa preCol-P weighing approximately

ol-D and preCol-D

l-D PreCol-D DD

0 0 02.1 60.2 04.4 51.0 60.06.9 23.9 20.04.6 30.9 40.07.6 78.5 20.04.3 55.8 12.5

6.4 345.1 327.57.1 173.0 302.50 0 05.2 24.1 22.59.6 7.4 2.51.2 8.5 5.04.9 30.5 55.00 0.3 2.52.7 8.2 15.03.7 10.7 22.57.5 23.5 47.50 0 09.7 21.1 7.52.2 46.4 37.5

ent.and have a standard error of about 5 %.

638

95 kDa; second, that Col-P can be generated from preCol-P bypepsin treatment, and third, that Col-P or preCol-P shows agradient in the foot that mirrors its gradient in the threads. Alsoevident, but without the immunoblot illustration, is thecomplementary gradient of preCol-D and pepsin-derived Col-D. Anti-Col-D antiserum was of similarly high titer(approximately 105), showing a strong preferential binding toCol-D over Col-P. PreCol-D has an apparent molecular massof 97 kDa (Fig. 7). It appears that three of these chains arecombined in the native precursor.

Amino aciperformed onelectrophoreticresemble thosethe former haserine than itsof glycine, hyprecursor. Thcollagenous anbest assessed

X. QIN AND J. H. WAITE

0.60

0.40

0.20

Abs

orba

nce

at 2

14nm

% A

ceto

nitr

ile (

––

––)

60

40

20

0 10 20 30 40 50 6

A′

B′

A

B

C DE

F

G

H I

Elution volum

Fig. 4. Elution profile ofCol-P tryptic peptides(A–M; see Fig. 5) by C-18 reversed-phaseHPLC.

Fig. 5. Amino acid sequence of Col-P Ntermini and tryptic peptides (see Fig. 4). Prefers to 4-trans-hydroxyproline.

Col-P N terminus

Partial pepsin cleavage

:A-A-R-A-N-A

Complete pepsin cleavage:G-G-P-G-N-P-G-N-S-G-S-P-G-

Col-P tryptic peptide sequences

Fraction SequenceA′ P-G-A-P-G-Q-P-G-Q-P-G-N-PB′ P-G-N-P-G-N-P-G-Q-P-G-N-PA G-Q-P-G-Q-P-G-G-P-G-Q-PB G-A-P-G-T-P-G-G-P-G-P-P-GC G-P-A-G-S-Q-G-P-S-G-G.....D G-K-P-G-P-Q-G-P-P-G-N-P-E A-G-P-A-G-P-Q-G-E-T-G-P-P-F G-K-P-G-L-A-G-K-P-G......G G-N-P-G-E-P-G-N-P-G-A-P-H G-P-Q-G-P-Q-G-A-E-G-P-AI G-P-P-G-A-Q-G-P-A-G-P-A-J G-P-I- G-P-S-G-P-S-G-A-P-GL G-M-P-G-T-P-G-T-P-G-Q-P-M G-L-Q-G-S-N-G-E-V-G-P-Q

JK

LM

reCol-P werens followingcompositionsects, although, alanine and higher levelsmate than itsbetween thee preCols are

Col-P or Col-

110

d analyses of preCol-D and p hydrolysates of these protei transfer to Immobilon. Their of Col-P and Col-D in some resp

s notably lower levels of glycine precursor; Col-D, in contrast, hasdroxyproline and glutamine/glutae differences in composition d non-collagenous domains of th

by subtracting the composition of

0 70 80 90 100e (ml)

Q-P-G-P-P-G-Q-P-G-S-P-

-G.........-G-T-P-G-K-G-G-P-G-A-G-G-Q-P-G-K......

......G......G-P-K

G-G-P-G-S-T-G-P-T-G-P-Q-G-P-Q-G-P-P-G.....-G-P-A-G......G-P-P-G-E-Q-G-P-P-G-E......-D-N-G-P-N-G......

G-I-P-G-T-V-G......-G-P-S-G-P-A-G-P-Q-G......

639 gradients in the mussel

D fromonly thousathe cathe difrom tand alaccou

Ultcarriethreadcorruga smtransmelectrocore can am(Fig. 8and indiscon(Fig. 8(Fig. 8antibo(Fig. 8straighof thobserv

. Neither of the antisera bound to the

Discussionudy indicate unequivocally that thecontains collagenous proteins. This

made earlier on the basis of wide- and detection of hydroxyproline-psinized byssus (Benedict and Waite,sal collagen is, however, by no meanslagen, as DeVore et al. (1984) haveistant domain is only half the size ofollagen types I–III and lacks 3-trans-

ylysine and, according to anotherl-derived cross-links (van Ness et al.l in containing several instances of

A

Plaque

Fig. 6.byssal distribpepsintransitiand cusampleblottin

Exotic collagen

bound in the distal threadcuticular sheath.

The results of this stbyssus of Mytilus eduliscorroborates suggestionsangle X-ray diffractioncontaining proteins in pe1986; Rudall, 1955). Bysidentical with type I colconcluded. Its pepsin-respepsin-resistant fibrillar chydroxyproline, hydroxstudy, aldimine and aldo1988). It is also unusua

B C D E

Distal Proximal Cuff

A B C D E Pur

e

b

sf

anrd

o

o

n

e

ui

g

that of its respective precursor. This calculation cane made following the conversion of ‘residues per

nd’ to ‘residues per molecule’ or part of a molecule ase may be (see Materials and methods). Table 1 showsference compositions for DD and DP. DD is predictedhis to consist of more than 60 mol% equimolar glycinenine. Serine, alanine and glycine (approximately 40 %)t for more than 60 mol% of DP.astructural and immunogold labelling studies were out on the proximal and distal portions of byssal

tandem Gly–Gly sequences, although these have also beenreported in invertebrate collagens from Nereis virens and Riftiapachyptila cuticles (Sharma and Tanzer, 1983; Gaill et al.1991; Mann et al. 1992) and in sea urchins (Exposito et al.1992). Apart from the obvious homology of having a glycineat every third residue and frequent imino acids at positions Xand Y, byssal collagens have no match with any knowncollagen sequence.

In contrast to the secretion of interstitial procollagens byfibroblasts, byssal collagens are stored in and secreted fromlarge (approximately 1–2 mm diameter) granules produced bythe ‘collagen gland’, a large holocrine gland in the interior ofthe mussel foot (Pujol et al. 1972; Vitellaro-Zuccarello, 1980).Whether the 95 kDa preCol-P subunit isolated from foot tissuerepresents a comparable byssal procollagen is not known, butthis seems unlikely in view of its composition. More probably,preCol-P is used without much proteolytic processing. Thespeed with which new threads must be made (2–5 min) maypreclude the trimming necessary for the conversion of otherprocollagens to tropocollagen. Our results, however, dosuggest that, whatever happens during its conversion frompreCol-P to proximal byssal thread, there remains a tidy central

Col-DCol-P

SDS polyacrylamide gel electrophoresis of the pepsin-resistantcollagens Col-P and Col-D showing their differentialtion over the length of a typical thread. Lanes A–E containzed protein derived from plaques (A), rigid portions (B), theon from rigid to extensible portion (C), extensible portions (D)ffs (E). The lane marked ‘pure’ was loaded with a purified of Col-P. Both collagens were identified in parallel by western with their specific respective antisera and amino acid analysis.

s. Proximally, the thread is elastic and flattened with aated surface, while distally it is rigid and cylindrical withoth outer surface. Sagittal sections examined byission electron microscopy show a homogeneousn-dense outer sheath surrounding an inner compositeonsisting of fiber bundles anisotropically distributed inrphous matrix, with fibers greatly outnumbering matrixE). In proximal thread, the core consists of distinct outerer morphologies. In the outer core, as in distal portions,

tinuous straight fibers are embedded in a matrixA,C). In contrast, the inner core contains coiled fibersB,D). Immunogold studies suggest that anti-Col-Pdies are localized with a high density on the coiled fibersB), whereas anti-Col-D is associated chiefly with thet fibers of the outer core (Fig. 8C). In the distal portion thread, however, only anti-Col-D antibodies areed bound to bundled fibers (Fig. 8F). Anti-Col-P is not

Col-P domain that is unmodified by cross-links and excisablewith pepsin from byssus (see Fig. 9). The same cannot be saidof Col-D which, except in conjunction with cold shock, isextracted at very low yields. Only a little can be said about thedomains flanking Col-P and Col-D. First, they must be cross-linked; second, the deduced amino acid composition of DP is40 % Gly, 13 % Ala, 11 % Ser and little to no Hyp. In the caseof DD, Gly and Ala each contribute more than 30 %, whereasPro content is low and Hyp is absent (Table 1). It is noteworthythat DP and DD compositions are reminiscent of resilin (Seifterand Gallop, 1966) and silk fibroin (Rudall, 1962), which areelastic and stiff proteins, respectively (Table 2). Quite possiblythen, the pepsin-labile flanking domains of preCol-D may befibroin-like; by the same token, the flanking domains ofpreCol-P may resemble an elastic protein such as resilin. Atthis stage, such associations must be viewed as tentative atbest, since composition does not reveal protein structure.

640

Curiousbyssus wX-ray dthey ‘suthe collastudies elaborat

The pin the coquestion

X

a b c d e fPure

91

67

45

Pepsinized series

Base

Foot sections

Fig. 7. DfollowedCol-P an(lower pwestern-bpurified C

. QIN AND J. H. WAITE

B

A

Col-D

Col-P

Tip

c′ d d′ e e′ Col-P10−3×Mr

91

67

45

WB

Intact series

B C D E

a b b′ c

A

ly, the existence of a major b-protein motif in distalas predicted by Rudall (1955) using wide-angle fiber

iffraction. In presenting his results, he mentions thatggest that there is another protein (or another phase ofgen structure) in the byssus threads’. Future molecularof the complete preCol sequences should be able toe on the validity of this hypothesis.ossible inclusion of silk-like and resilin-like domainsllagenous tensile element of byssus begs the following. Why do mussels use collagen at all? We can think of

one reason. Since true underwater silks are not known to existin nature, a collagenous domain may be necessary to endowbyssal precursors with the capacity for fibrillogenesis. Thereason for amalgamating collagen with silk- and resilin-likedomains is more subtle and is probably rooted in the conceptof composite materials; for example, ‘it is a common principlethat two or more components may be profitably combined toform a composite material so as to make best use of the morefavorable properties of the components while simultaneouslymitigating the effects of some of their less desirable

istribution of Col-P and Col-D and their precursors in transverse foot sections. (A) Col-P and Col-D from pepsinized foot sections by SDS polyacrylamide gel electrophoresis (upper panel, lanes a–e), followed by pure Col-P and Col-D (lane f). (B) Precursors ofd Col-D identified immunochemically from extracts of nonpepsinized foot sections followed by SDS polyacrylamide gel electrophoresisanel). Lanes a–e stained for protein with Serva Blue R-250; lanes b9–e9 contain parallel samples, transferred to nitrocellulose andlotted with anti-Col-P. The fast-moving species in e9 are degradation products of preCol-P. The last two lanes contain, respectively,ol-P stained with Serva Blue R-250 (left) and western-blotted Col-P (WB) using anti-Col-P antisera.

641Exotic collagen gradients in the mussel

characunder under tmateriahigher doubt,

oc

v

ic

A B

Fig. 8. longitudistal abyssal throughshowin(v) andregionslabelled(14 700of proxanti-Coboxed (C) Oproximanti-Cocore relabelled(130 00distal immunoantiColarea isshows portionanti-CoincubatP and separate

v

C D

E F

f

f

.f

r

Electron micrographs odinal sections through thend proximal portions othreads. (A) Section

proximal threadg the corrugated varnish outer and inner core

(oc, ic), immunogold with anti-Col-P3). (B) Inner core regionimal thread labelled withl-P (14 5003). Theregion is enlarged in Duter core region o

al thread labelled withl-D, 133 0003. (D) Innegion of proximal thread with anti-Col-P03). (E) Section through

teristics’ (Harris, 1980). Each byssal thread functionsthe somewhat paradoxical expectation that it be strongension and that it absorb shock. Collagen is as strong al as silk, but silk is 30–40 times tougher, i.e. has abreaking strain energy density (Denny, 1988). No

in distal thread the presence of silk would toughen the

collagen. In proximal thread, extensions of more than 100 %(Smeathers and Vincent, 1979) are carried by the elasticdomains whereas, at much greater strains, the collagen andsilk-like network would limit the extensibility of thecomposite. Indeed, the breaking energy of byssal threads isreported to be 12.503106 J m23, which is intermediate

portion of thread,gold labelled with

-P (24 0003). The boxed enlarged in F, whichthe core region of distal of thread labelled withl-D (70 0003). Allions with antisera to Col-Col-D were performedly and in parallel.

642

between53106 J1988).

The rbyssal respectivof collagexperimefaithfullyP and Cdistributnot appedetectablP does nof argumthe moleshould Cimmunogthis. Colfibers frorespectivcarried oto the pcuticularmight bevarnish arisk of cushrinkag

X

Table 2and

Amino ac

3-Hyp4-HypAsxThrSerGlxProGlyAlaCys/2ValMetIleLeuDopaTyrPheHisHylLysArg

Values

Varnish

PreCol-P

Col-P

Thread

Proximal

B

. QIN AND J. H. WAITE

. Amino acid compositions of collagen, silk fibroin resilin (adapted from Seifter and Gallop, 1966)

Collagen type I Silk fibroin Resilinid (rat tail) (Braura truncata) (Locust)

4.1 0 093.0 0 047.0 98.4 102.219.7 10.1 29.641.0 85.8 78.674.0 11.4 50.4

129.0 0 79.4329.0 347.0 376.0110.0 293.0 111.0

0 0 019.2 5.1 25.69.0 0 06.3 10.0 20.4

PreCol-D

Col-D

Plaque

Distal

C

A

tic modulus intermediate between that of thee elastic thread components. This would thenond gradient perpendicular to the first, in whichs as one moves from the cuticular varnish to theximal thread (Fig. 9). In contrast to the above, has proposed an alternative model of the

Collagendomains

Elasticdomains

Cross-links

Pepsinsites

ical model of preCol-P and preCol-D distribution inindows of molecular structure include a section of

thread (A), extensible proximal thread (B) and theoximal thread just inside the cuticular varnish (C).

cross-links is largely arbitrary. The role of matrix is this model.

the breaking energies of tendon (23106 tom23) and silk (503106 to 1803106 J m23) (Denny,

ough correlation of Col-P and Col-D gradients inthreads with the extensible and stiff portions,ely, is intriguing. Fig. 9 presents a simplistic modelen distribution in byssus that reconciles many of thental observations. It is simple because it does not reflect the extent of overlap in the distribution of Col-ol-D that is apparent from Fig. 6. Although the

ion of the two is complementary to some extent, it does

D has an elasvarnish and threpresent a secCol-D decreasecore of the proBairati (1991)

18.1 10.0 25.60 − −3.7 36.6 29.2

11.9 8.8 27.41.5 0 6.54.8 0 0

30.1 18.9 4.949.0 79.8 33.6

given in residues per 1000.

Silk-likedomains

Fig. 9. Hypothetbyssal threads. Wthe rigid distal outer layer of prThe position of not explained in

ar to be symmetrically complementary, i.e. Col-D ise in all portions of the thread except the cuff, and Col-ot appear to be present in rigid thread. If, for the sakeent, the parent molecule of Col-P serves to providecular extensibility of proximal thread, then whyol-D continue to persist proximally? The results ofold labelling studies appear to shed some light on

-D and Col-P are associated with straight and coiledm the distal and proximal portions of the thread,

ely. However, some straight fibers, i.e. Col-D, arever into the proximal thread, where they are limitederipheral layer of the thread core just inside the varnish (Fig. 8C). One reason for this carry-over to provide shear stability between the rigid cuticularnd the extensible thread core. Without this, there is aticular delamination during cyclic thread elongation/

e, were it applied directly to the core. Perhaps preCol-

proximal thread based on electron microscopical analysis thatplaces the elastic elements just inside the varnish and the morerigid tensile fibers in the thread core.

The observation that matching complementary distributionsof two byssal collagens were found along the long axis ofbyssal threads and the foot suggests that a compositionalgradient has been predetermined and prefabricated in thecollagen gland. Although it is premature to say how this hasbeen achieved, there are two intriguing possibilities: (1) thatcollagen granules are titrated with different proportions of thetwo collagens along the length of the foot, and (2) that thereare two different types of granules, i.e. one for preCol-D andthe other for preCol-P, whose relative abundance variesaccording to their location in the foot. Ultrastructural studiesof the collagen gland of M. galloprovincialis (Vitellaro-Zuccarello, 1980) support the presence of two major granuletypes; however, unpublished immunogold labelling studies of

collag(X. Qgranuspecu

WeElectrBiostrHealththe OHealth

ADAMBAIRA

molultraLan

BENEDbyss

BLAKEGOTof aAna

BORNSbase135

BROWdeteanim

DENNYEnv

DEVOSAUprodPhy

EXPOS(199vert155

GAILLENGinteven

GARROof M

GATHEwavIn SKel

GROSSBioc

HARRISym

HYATTImmM.

KNIGHin th

643Exotic collagen gradients in the mussel

en granules in our laboratory tend to support the formerin and J. H. Waite, unpublished observations). Until thele content has been determined, this will remainlative.

thank Professor Roger Wagner of the Universityon Microscope Facility and Dr Mark Trumbore of theucture Analysis Center at the University of Connecticut Center. This work was supported in part by grants from

ffice of Naval Research and the National Institutes of.

ReferencesS, E. (1978). Invertebrate collagens. Science 202, 591–597.TI, A. (1991). The byssus of the mussel Mytilus from theecules to the organ: Functional performance resides in the

secretion in the nidamental gland and formation of the egg capsulewall in the dogfish Scyliorhinus canicula. Phil. Trans. R. Soc. Lond.B 341, 419–436.

KRAMER, J. M., COX, G. N. AND HIRSCH, D. (1982). Comparisons ofthe complete sequences of two collagen genes from Caenorhabditiselegans. Cell 30, 599–606.

LIPMAN, D. J. AND PEARSON, W. R. (1985). Rapid and sensitive proteinsimilarity searches. Science 227, 1435–1441.

MANN, K. H., GAILL, F. AND TIMPL, R. (1992). Amino acid sequenceand cell adhesion activity of fibril-forming collagen from the tubeworm Riftia pachyptila living at deep sea hydrothermal vents. Eur.J. Biochem. 210, 839–847.

MASCOLO, J. M. AND WAITE, J. H. (1986). Protein gradients in thebyssal threads of some marine bivalve molluscs. J. exp. Zool. 240,1–7.

MCCORMICK, P. J., CHANDRASEKHAR, S. AND MILLIS, A. J. T. (1979).Direct visualization of collagens and procollagens inpolyacrylamide gels. Analyt. Biochem. 97, 359–366.

3N

R

5,

rt

l

S

V

structural assembly. In Form and Function in Zoology (ed. G.zavecchia and G. Valvassori), pp. 163–177. Modena: Mucchi.ICT, C. V. AND WAITE, J. H. (1986). Location and analysis ofal structural proteins. J. Morph. 189, 171–186., M. S., JOHNSTON, K. H., RUSSELL-JONES, G. J. ANDSCHLICH, E. C. (1984). A rapid sensitive method for detectionlkaline phosphatase-conjugated anti-antibody on western blots.lyt. Biochem. 136, 175–179.TEIN, P. AND PIEZ, K. A. (1965). Collagen: Structure studiesd on the cleavage of methionyl bonds. Science 148,–1355., C. H. (1950). A review of the methods available for the

rmination of the forces stabilizing structural proteins inals. Q. J. microsc. Sci. 91, 331–339., M. W. (1988). Biology and the Mechanics of the Wave Sweptironment. Princeton: Princeton University Press.

E, D. P. G. H., ENGEBRETSON, G. H., SCHACTELE, C. F. ANDK, J. J. (1984). Identification of collagen from byssus threadsuced by the sea mussel Mytilus edulis. Comp. Biochem.

siol. 77B, 529–531.ITO, J. Y., D’ALESSIO, M., SOLURSH, M. AND RAMIREZ, F.2). Sea urchin collagen evolutionarily homologous to

ebrate Pro-a2 (I) collagen. J. biol. Chem. 267,9–15562.

F., WIEDEMANN, H., MANN, K., KÜHN, K., TIMPL, R. AND

MELNICK, S. C. (1958). Occurrence of collagen in the phylumMollusca. Nature 181, 1483.

MERCER, E. H. (1952). Observations on the molecular structure ofbyssus fibres. Austr. J. mar. Freshwater Res. 3, 199–204.

MILLER, E. AND RHODES, R. K. (1982). Preparation andcharacterization of different types of collagen. Methods Enzymol.82, 33–64.

PIKKARAINEN, J., RANTANEN, J., VASTAMÄKI, M., LAMPIAHO, K., KARI,A. AND KULONEN, E. (1968). On collagens of invertebrateswith special reference to Mytilus edulis. Eur. J. Biochem. 4,555–560.

PRICE, H. A. (1981). Byssus thread strength in the mussel, Mytilusedulis. J. Zool., Lond. 194, 245–255.

PUJOL, M. M., BOCQUET, J. AND BOREL, J.-P. (1976). Le byssus deMytilus: étude électrophorétique de fractions protéiques riches enhydroxyproline extraites de la ‘glande du collagène’. C.R. hebd.Séanc. Acad. Sci. Paris 283, 555–558.

PUJOL, M. M., HOUVENAGEL, G. AND BOUILLON, J. (1972). Lecollagène du byssus de Mytilus edulis L. I. Ultrastructure descellules sécrétrices. Archs Zool. exp. gen. 113, 251–264.

RUDALL, K. M. (1955). The distribution of collagen and chitin. Symp.Soc. exp. Biol. 9, 49–71.

RUDALL, K. M. (1962). Silk and other cocoon proteins. Comp.Biochem. 4, 397–433.

EL, J. (1991). Molecular characterization of cuticle andstitial collagen from worms collected at deep sea hydrothermals. J. molec. Biol. 221, 209–223.NE, R. (1978). Phylogenesis of connective tissue. In Frontiersatrix Biology, vol. 5 (ed. L. Robert), pp. 1–250. Basel: Karger.RCOLE, L. J. AND KELLER, A. (1975). Light microscopiceforms in collagenous tissues and their structural implications.tructure of Fibrous Biopolymers (ed. E. D. T. Atkins and A.er), pp. 153–167. London: Butterworths., J. (1963). Comparative biochemistry of collagen. Comp.hem. 5, 307–345., B. (1980). The mechanical behaviour of composite materials.

p. Soc. exp. Biol. 34, 37–74., A. D. (1984). Immunogold labelling techniques. Inunolabelling for Electron Microscopy (ed. J. M. Polak and I.arndell), pp. 59–75. Amsterdam: Elsevier.

T, D. P., FENG, D., STEWART, M. AND KING, E. (1993). Changese macromolecular organization in collagen assemblies during

SEGEL, I. H. (1976). Biochemical Calculations. New York: JohnWiley and Sons.

SEIFTER, S. AND GALLOP, P. M. (1966). The structure proteins. In TheProteins (ed. H. Neurath), pp. 155–458. New York: AcademicPress.

SEYDEL, E. (1909). Untersuchungen über der Byssusapparat derLamellibranchiaten. Zool. Jb. 27, 465–582.

SHARMA, Y. D. AND TANZER, M. L. (1984). HPLC separation ofglycopeptides from Nereis cuticle collagen. Analyt. Biochem. 141,205–212.

SMEATHERS, J. E. AND VINCENT, J. F. V. (1979). Mechanical propertiesof mussel byssus threads. J. Moll. Stud. 45, 219–230.

SPIRO, R., LUCAS, F. AND RUDALL, K. M. (1971). Glycosylation ofhydroxylysine in collagens. Nature 231, 54–55.

TOUS, G. I., FAUSNAUGH, J. L., AKINYOSOYE, O., LACKLAND, H.,WINTER-CASH, P., VITORICA, F. J. AND STEIN, S. (1989). Amino acidanalysis on polyvinylidene difluoride membranes. Analyt. Biochem.179, 50–55.

644

TSUGITA, A., UCHIDA, T., MEWES, H. W. AND ATAKA, T. (1987). Rapidvapor phase acid hydrolysis of peptide and protein. J. Biochem.,Tokyo 102, 1593–1597.

VAN NESS, K. P., KOOB, T. J. AND EYRE, D. R. (1988). Collagen cross-linking: distribution of hydroxypyridinium cross-links amonginvertebrate phyla and tissues. Comp. Biochem. Physiol. 91B,531–534.

VITELLARO-ZUCCARELLO, L. (1980). The collagen gland of Mytilusgalloprovincialis: An ultrastructural and cytochemical study onsecretory granules. J. Ultrastruct. Res. 73, 135–147.

VITELLARO-ZUCCARELLO, L., DE BIASI, S. AND BAIRATI, A. (1983).The ultrastructure of the byssal apparatus of a mussel. V.

Localization of the collagenic and elastic components in thethreads. Tissue & Cell 15, 547–554.

WAITE, J. H. (1991). Detection of peptidyl-3,4-dihydroxyphenylalanine by amino acid analysis andmicrosequencing techniques. Analyt. Biochem. 192, 429–433.

WAITE, J. H. (1992). The formation of mussel byssus: Anatomy of anatural manufacturing process. In Results and Problems in CellDifferentiation, vol. 19, Biopolymers (ed. S. T. Case), pp. 27–54.Berlin: Springer-Verlag.

WAITE, J. H., HOUSLEY, T. J. AND TANZER, M. L. (1985). Peptiderepeats in a mussel glue protein: Theme and variations.Biochemistry, N.Y. 24, 5010–5014.

X. QIN AND J. H. WAITE