Collagen-Binding Matrix Proteins from Elastomeric Extraorganismic

1240

Biomacromolecules 2002, 3, 1240-1248

Collagen-Binding Matrix Proteins from Elastomeric Extraorganismic Byssal Fibers Chengjun Sun, Jared M. Lucas, and J. Herbert Waite* MCDB Department and Marine Science Institute, University of California at Santa Barbara, Santa Barbara, California 93106 Received June 7, 2002; Revised Manuscript Received August 22, 2002

The byssal threads of marine mussels represent a peculiar case of extraorganismic extracellular material. The threads consist of fibrous chimeric collagens such as preCol-P (with collagenous, elastin-like and histidinerich domains) embedded in a microfibrillar matrix. We report here on the extraction, purification, and characterization of water-soluble proximal thread matrix protein 1 (PTMP1), which is preferentially located in the proximal portion of each byssal thread and decreases in a proximal to distal direction. PTMP1 has a mass of about 50 kDa as determined by matrix-assisted laser desorption-ionization with time-of-flight (MALDI-TOF) mass spectrometry. Glycine is the most common residue at 12.2 mol %, followed by asparagine/aspartic acid and glutamine/glutamic acid at 11.4 and 9.9 mol %, respectively. Glycosylation has been detected by Western blotting with biotinylated concanavalin A and neutral sugar analysis. With degenerate primers designed from the N-terminal sequence and an additional internal peptide derived by Lys-C endopeptidase digestion, a complete cDNA sequence for this protein was obtained by polymerase chain reaction (PCR) amplification of a Mytilus edulis foot cDNA library. Two variants with minor sequence differences limited to the N-terminus were found. The cDNA-deduced protein sequence reveals two symmetric internal repeats that together account for >85% of the protein. Sequence and epitope similarity of PTMP1 to the A domains of von Willebrand factor and integrin R1I suggest a capacity for collagen binding. Enzymelinked immunosorbent assay (ELISA)-based measurement of PTMP1 binding to immobilized type I collagen shows high affinity (apparent KD ) 0.25 µM), but the binding exhibits no dependence on metals. Using primers designed from M. edulis, we also found a PTMP1-like cDNA in a related species, M. galloproVincialis, with a deduced protein sequence having 97% identity with one M. edulis variant and 99% identity with the other. The corresponding cDNA sequences have 94% and 96% identity, respectively. Mussel byssal threads possess a strength and suppleness that once inspired a flourishing textile industry in the Mediterranean region.1,2 This industry has disappeared probably because of depletion of mussel (Pinna) stocks, but byssal threads, like spider silks, continue to be an important resource for biomimetic engineering.3 Indeed, these two fibers have much in common. Both are formed from liquid crystalline protein precursors aligned and denatured by shearing forces,4,5 both contain similar repeating sequence modules such as polyalanines, glycine-rich stretches, and elastin-like pentapeptides,6,7 and both display an outstanding capacity to dissipate energy. For example, resilience is about 35% and 30% for dragline silk and byssus, respectively.8 Spider dragline silk is certainly superior to mussel byssus in terms of strength and toughness, but byssal threads also have unusual characteristics that continue to captivate. Unlike strands of spider silk, which are chemically and mechanically uniform from end to end, the byssal threads of Mytilus have significantly different properties at their ends. The proximal end is one-tenth as stiff and twice as extensible as the distal end.7,8 The change in mechanical properties is not sudden but is a gradual one and is correlated with an axial gradient * To whom correspondence should be addressed. Tel: 805-893-2817. Fax: 805-893-7998. E-mail: [email protected].

of strongly aligned thread proteins called preCols. PreCols are essentially block copolymers in which a dominant central collagenous domain is flanked by either two silk-like or two elastin-like domains and terminated by histidine-rich repeat sequences.7,9-11 The collagens with silk-like flanking domains (prepepsinized distal collagen, preCol-D)9 predominate in the stiffer distal portion of the thread, while those with elastinlike flanks (preCol-P) prevail in the proximal portion.10 It is tempting to attribute the stiffness of the distal and proximal portions of the thread to the reported stiffness of spider dragline silk and elastin, respectively,7 but the situation is unlikely to be as simple as that. Unlike the distal portion, which has densely packed bundles of collagen, the volume fraction of collagen in the core sections of the proximal portion is estimated at about 0.3 with the balance referred to collectively as “matrix protein”.12 Moreover, when subjected to cyclic stress at a fixed strain, the proximal portion shows a prominent strain-stiffening effect not associated with collagens, silks, or elastin,13 while the distal portion undergoes stress-softening.14 We report here on the isolation and characterization of a noncollagenous protein from the proximal portion of the thread. This protein has sequence and antigenic resemblance to the A domains of the von Willebrand factor (vWF), a

10.1021/bm0255903 CCC: $22.00 © 2002 American Chemical Society Published on Web 10/08/2002

Von Willebrand Factor A Domains in Byssal Protein

protein reported to bind to various ligands including interstitial type I and III collagens and to be critically involved in a stiffening phenomenon known as shear-induced hemostasis.15 Experimental Procedures Mussel Maintenance and Thread Collection. Mytilus edulis was collected from the breakwater at Lewes, DE, or obtained from the Marine Biological Laboratory, Woods Hole. After collection, the mussels were transferred to an aquarium with running seawater thermostated at 6-10 °C and fed twice weekly with Isochrysis galbana cultures. Mussels were tethered by rubber bands to a nylon fishing line wrapped around Plexiglas plates. These plates were draped from a thicker fishing line running across the top of the aquarium tank. To maximize collection of fresh threads, threads were harvested once every 2 days. The threads were cut a couple millimeters away from the plaque at one end and from the mussel at the other end without the attached stem. The collected threads were put into a chilled beaker of distilled, deionized water (ddH2O). All collected threads, except those used for dissection, were rinsed twice with cold ddH2O and stored in a -80 °C freezer. Freshly collected threads were dissected under a light microscope into four distinct parts: stem, proximal thread (corrugated), proximal/distal transitional portion, and distal thread (stiff and smooth). Each part was washed with ddH2O, matted on paper towel and stored at -80 °C. Protein Extraction. Whole threads or thread portions were first minced on ice using a pair of scissors or a razor blade. The cut threads were then put into a 50 cm3 tissue grinder (Kontes, Vineland, NJ) and homogenized with cold 5% acetic acid/8 M urea on ice at a ratio of 10 mL extraction buffer to 1 g of thread. The homogenized threads were centrifuged at 15 000 g for 15 min to pellet the insoluble material. Then to 1 mL of supernatant, 12 µl of 60% (v/v) perchloric acid and 20 µl of 18 M sulfuric acid were added to a final concentration of 0.7% and 0.36 M, respectively. This was followed by the addition of 1 volume of chilled acetone (-20 °C) per volume of sample supernatant. After incubation on ice for 15 min, the sample was centrifuged at 15 000 g for 15 min to pellet the precipitated material. The pellet was discarded, and another volume of cold acetone was added to the supernatant (S1). Precipitated protein pellet (P2) was harvested by centrifugation (as above) within 15 min. Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS PAGE). A small amount of pellet P2 obtained above was redissolved in ddH2O and run on a 10% miniSDS PAGE denaturing gel using the Laemmli tris-glycinate electrode buffer system. The gel was run at a constant current of 20 mA and was stained with Coomassie Blue R-250 as previously described.16 Reversed Phase High Performance Liquid Chromatography (HPLC) Purification of PTMP. P2 was redissolved in ddH2O. The insoluble material was removed by centrifugation and subjected to another extraction with 5% acetic acid. The water- and acetic acid-soluble supernatants were then pooled and run on reversed phase HPLC using a C-8 column and continuous detection at a wavelength of 280

Biomacromolecules, Vol. 3, No. 6, 2002 1241

nm. Linear gradients of stationary (aqueous 0.1% trifluoroacetic acid [TFA]) and mobile phases (acetonitrile with 0.1% TFA) were used for protein elution. Specific programmed events were as follows: acetonitrile 0-30% in 10 min, 3060% from 10 to 60 min, 60-100% from 60 to 70 min, and 100-0% from 70 to 80 min, all at a flow rate of 1 mL/min. Column eluate was monitored at 215 nm using a Gilson diode array detector. Collected fraction volumes were 1 mL. The peak fractions were frozen at -80 °C and lyophilized. MALDI TOF Mass Spectrometry. Matrix-assisted laser desorption ionization mass spectrometry with time-of-flight (MALDI-TOF) was done on a PerSeptives Voyager DE instrument (Perkin-Elmer) with either purified protein or byssal threads as the analyte. A fresh solution of sinapinic acid (Aldrich) in aqueous 30% acetonitrile and 0.1% trifluoroacetic acid was prepared daily for use as matrix. Sample protein was diluted 1:25 with matrix solution, and 1-2 mL was spotted onto the gold-plated sample tray. Typical operating conditions in positive ion mode included an accelerating voltage of 25 000 V, grid voltage at 96%, delay time of 500 ns, guide wire voltage at 0.10%, and N2 laser power at 1900-2500 (arbitrary units). Internal calibrant was bovine serum albumin with [M + H]+ at 66 431 and [M + 2H]2+ at 33 216. Direct laser irradiation of byssal threads was also done during MALDI-TOF mass spectrometry. In this case, the threads were first partitioned into distal, transitional, and proximal portions and then soaked in 5% acetic acid for 30 min. Thread sections were blotted dry and carefully slit open using stereomicroscopic surgery. The dissected sections were mounted on MALDI-TOF sample plates using double-stick tape (optional) and, following application of matrix solution (see above) and drying, irradiated at a laser power of 2100 with an accelerating voltage of 25 000 in positive ion mode. Typical spectra represent the average of 256 scans. Amino Acid Analysis. Amino acid compositions of proteins/peptides were determined on a 6300 Autoanalyzer (Beckman Instruments, Fullerton, CA) following hydrolysis for 1 h (6 M HCl with 10% TFA and 5% phenol in vacuo at 155 °C) or 24 h (6 M HCl with 5% phenol in vacuo at 110 °C). N-terminal gas-phase sequencing was done on a Porton 2020 protein sequencer (Beckman-Coulter) with a dedicated in-line HPLC (model 2090) for separating the PTH amino acid derivatives. Protease Digestion. Lys-C endoproteinase was used to digest the protein at a ratio of 1:500 (enzyme/protein) in 1mM tris-HCl (pH 7.4) buffer with stirring for 24 h at room temperature. The digested solution was fractionated by reversed phase HPLC with a C-18 column using an elution program suitable for peptide separation. Peak fractions were collected, lyophilized, and subjected to N-terminal sequencing. Carbohydrate Detection. Purified PTMP was electroblotted following 10% SDS PAGE onto a poly(vinylidene difluoride) (PVDF) membrane in transfer buffer (22.5 mM tris-borate, pH 8.5, in 25% methanol with 1% SDS) at 200 mA for 75 min using a Genie transfer unit (Idea Scientific, Minneapolis, MN). The membrane was then blocked for 1 h in TBS (tris buffered saline, pH 7.5) with 3% BSA

1242

Biomacromolecules, Vol. 3, No. 6, 2002

followed by a 1 h exposure to biotinylated concanavalin A (1 mg/mL; Pierce Chemical) in TBS buffer with 0.3% BSA, 0.05% Tween 20. After three changes of washing buffer (TBS with 0.3% BSA, 0.05% Tween 20), the membrane was incubated in washing buffer with 1 mg/mL avidin-horseradish peroxidase for 1 h. Following three more changes of washing buffer, labeled glycoproteins were detected using 4-chloro-1-naphthol and H2O2. The high molecular weight standards served as negative controls and ovalbumin served as a positive control. Sugar Analysis. A protocol for proximal thread matrix protein 1 (PTMP1) sugar analysis by high-performance anion-exchange chromatography and pulsed amperometric detection (HPLC-PAD) was adapted from Borch and Kirchman.17 Protein samples were first hydrolyzed in 0.85 M H2SO4 for 24 h at 100 °C followed by neutralization and partial desalting by adding precombusted CaCO3. After neutralization, CaCO3 was centrifuged down, and the supernatant was desalted through desalting resin column. Desalted sample was tested on a Dionex HPLC-PAD (Dionex Corp., Sunnyvale, CA). Standards used for calibration curve were hydrolyzed and desalted the same way as the sample. Preparation of Antiserum against PTMP1. Antiserum was prepared in New Zealand white male rabbits. Two milligrams of PTMP1 were dissolved in 250 mL of PBS and emulsified in an equal volume of Freund’s complete adjuvant before being injected subcutaneously. A second and third injection using PTMP1 emulsified in Freund’s incomplete adjuvant was done after 4 and 8 weeks, respectively. Serum was harvested by exsanguination of the rabbit by heart puncture under anesthesia two weeks after the third injection. Serum collected from the rabbit prior to injection of PTMP1 was used as preimmune serum in controls for western blotting. National Institutes of Health guidelines for laboratory animal care (NIH publication no. 86-23, revised 1985; National Institute of Health, Bethesda, Maryland) were followed in all procedures associated with the production of the antiserum. Western Blotting. Antisera were tested against various antigens that were electrotransferred from SDS PAGE gels to PVDF membranes using the Genie transfer unit as described earlier. The PTMP1 antiserum was used at a 1:1000 dilution in conjunction with goat anti-rabbit peroxidaselabeled secondary antibody and 4-chloro-1-naphthol and H2O2 as substrate. Preimmune serum was used as a negative control under identical conditions. To explore the similarity of PTMP1 to von Willebrand Factor (vWF), western blotting was also performed with polyclonal anti-human vWF antibody (dilution 1:500; Dako, Denmark), monoclonal antihuman vWF domain A3 antibody RU5 (gift of Dr. Jan Sixma, Utrecht University18), and recombinant vWF domain A3 (also from Dr. Sixma). Immunohistochemical Localization. Proximal and distal thread portions were collected, separately fixed in 4% (v/v) paraformadehyde, dehydrated, and embedded in an 80/20% n-butyl/methyl methacrylate blend according to a standard protocol.19 Sections (2 mm thick) were cut with a standard lab microtome and transferred to uncoated slides. Sections were de-embedded with xylene and rehydrated with a graded

Sun et al.

series of ethanol. Rehydrated sections were equilibrated with digestion buffer (10 mM tris at pH 8.0 with 10 mM KCl, 10 mM NaCl, and 1 mM dithiothreitol) and digested with chymotrypsin (1 mg/mL) for 5 min. Nonspecific interactions were screened by a 30 min incubation in blocking buffer (10 mM phosphate at pH 7.5, 150 mN NaCl, 0.8% bovine serum albumin, 0.1% gelatin, and 2 mM sodium azide); incubation with PTMP1 antiserum (1:100 dilution) was carried out overnight. Detection was done using a colloidal gold-labeled secondary antibody (Auroprobe One, Amersham Pharmacia) that was enhanced for 10 min with an IntenseSEM kit (Amersham Pharmacia), followed by a ddH2O water rinse before viewing.20 Collagen Binding Assay. The collagen binding activity of PTMP1 was tested using kangaroo type I collagen (Sigma) following the modified protocol from Fischer et al.21 Wells in a Microtest III (Becton Dickinson) microtiter plate were first incubated with 125 µL of kangaroo collagen (0.10 mg/ mL) dissolved in 50 mM sodium acetate buffer, pH 5.0, for 30 min in the dark. Unbound collagen was rinsed away by three washes with 10 mM tris/HCl buffer, pH 7.5 (TBS). The reactive groups were blocked with 1% bovine serum albumin in TBS for 1 h at room temperature. Aliquots of PTMP1 (stock solution 0.28 mg/mL) diluted in TBS buffer with 0.1% BSA (total volume 100 µL) were added to the collagen-coated microtiter plate and incubated at room temperature overnight. The plate was then subjected to three changes of TBS buffer before adding 100 µL of rabbit antiPTMP1 IgG (1:500 dilution in TBS + 1% BSA) and was incubated for 1 h at room temperature. After three washes with TBS-1% (v/v) Tween 20, 100 µL of goat anti-rabbit horseradish peroxidase-conjugated antibody (1:1000 dilution in TBS) was added to the wells and incubated for 1 h. Unbound antibody was removed by three washes with TBS1% Tween 20. Color development proceeded with BM chemiluminescence enzyme-linked immunosorbent assay (ELISA) substrate (Roche, Indianapolis, IN). Color intensity was measured at 405 nm after a 30 min incubation using an automated plate reader. Controls include collagen with primary and secondary antibodies, collagen with secondary antibody, 3% BSA in TBS buffer, and color substrate. Cloning the cDNA of PTMP1. A M. edulis foot cDNA library made in rZAPII EcoR I/Xho I vector (Stratagene) by Coyne22 was used as the template for the polymerase chain reaction (PCR) experiments. Forward degenerate primer P2L2f with a sequence of ATGGGNCAYCAYGGNGTNATGCC (N ) A + G + T + C; Y ) C + T) was designed according to the N-terminal amino acid sequence (MGHHGVMP) of PTMP1. Reverse primer P2L-3r with a sequence of GCYTGYTGNCCDATDAT (corresponding to IIGQTA of one Lys-C peptide, D ) G + A + T) and P2L-4r with a sequence of CCRTCNCCDATNGCYTGYTG (corresponding to QTAIGDG of the peptide, R ) A + G) were designed according to an internal peptide sequence obtained from Lys-C digestion and Edman sequencing. These primers were used in the polymerase chain reaction (PCR) to amplify the cDNA of PTMP1 from the M. edulis cDNA library. Direct PCR of a M. edulis cDNA library at temperature gradients ranging from 37 to 65 °C in 25 µl aliquots was


carried out to determine the most appropriate annealing temperature. The template was the M. edulis foot cDNA library. PCRs were done under the standard conditions (1mM of each primer, 1X PCR buffer [Fisher Scientific], 1mM of each dNTP, 1mM of MgCl2 and 50 units of Taq polymerase). Denaturation was achieved at 95 °C for 30 s, and elongation was done at 72 °C for 2 min. The annealing temperature was determined by the temperature gradient and was set at 50 °C for 1.5 min. Negative controls were done in reaction buffer without template. To clone the cDNA, the PCR products from the above reactions were run on 1% agarose gels and stained with ethidium bromide. Amplified bands of the predicted size were excised and gel purified using the Qiagen Gel Purification Kit (Qiagen Inc, Valencia, CA). The purified PCR product(s) was ligated into a pGEM-T easy cloning vector (Promega). Competent JM109 cells (Promega) were used for transformation, and positive transformed clones were selected via blue/white screening. To obtain the cDNA sequence of the protein, individual colonies were cultured in LB buffer. The plasmids were purified using the Plasmid Purification Kit (Qiagen) and sequenced at UCSB’s Advanced Instrumentation Center using vector specific primers M13 and SP6. cDNA and polypeptide sequences were searched against the SwissProt and Expasy databases using the BLAST program. 5′ RACE. To obtain the complete untranslated sequence of PTMP1, the 5′ rapid amplification of cDNA ends (RACE) GeneRacer Kit (Invitrogen) was used as follows: Fresh amputated M. edulis feet (usually two) were depigmented and quickly frozen with liquid nitrogen. Total RNA was extracted using the Plant RNeasy RNA extraction Kit (Qiagen) and treated with calf intestinal phosphatase (CIF) to dephosphorylate non-mRNA and truncated RNA. The mRNA was purified, decapped, ligated to GeneRacer RNA oligo with a sequence of 5′-CGACUGGAGCACGAGGACACUGACAUGGACUGAAGGAGUAGAAA-3′ (italic is the GeneRacer 5′ primer), and reverse-transcribed with the gene specific primer P2l-180r with a sequence of 5′ATTCTCCTTGTTATCATATTGGA-3′. PCR was performed using GeneRacer 5′ primer and P2l-180r and the resulting product was gel purified, cloned into pGEM-T easy cloning vector, amplified, and sequenced as described above. Identical cloning, RACE, and sequencing protocols were used to characterize a PTMP cDNA from a M. galloproVincialis foot cDNA library, as well as freshly extracted mRNA. Results Purification and Characterization of PTMP1. The fibrous proteins of byssal threads including the collagens (preCols) are thought to be stabilized by extensive crosslinking that prevents their solubilization during extraction.7 With this in mind, byssal threads were subjected to several treatments to determine whether more soluble constituents are also present. Direct laser irradiation of matrix-treated thread sections by MALDI-TOF showed 40 and 50 kDa proteins to be the primary desorbed and ionized proteins (Figure 1). The latter appeared to be preferentially localized in the proximal portion of the thread, while the former


Figure 1. Direct MALDI-TOF mass spectrometry of different thread portions showing the graded distribution of PTMP1. Distal, transitional, and proximal thread sections were swelled in 5% HAC for 30 min before adding matrix (sinapinic acid) and irradiating with an accelerating voltage of 25 000 and arbitrary laser (337 nm) intensity of 2100 in a positive ion mode and delayed extraction.

predominated distally. To identify the 50 kDa protein, gram amounts of the proximal portions of byssal threads were extracted by trituration in 5% acetic acid with 8 M urea. Solubilized proteins as separated by SDS PAGE are shown in Figure 2A (lanes 2-4) for the distal, transitional, and proximal portions. These showed a similar trend in mass, shifting from 40 to 50 kDa in a distal to proximal direction, but the resolution is less than that shown by MALDI-TOF analysis. SDS PAGE of crude byssal samples is often compromised by smearing due to the tendency of some of the proteins to form insoluble complexes with dodecyl sulfate.23 The 50 kDa protein was further purified from acidextracted proximal sections by precipitation with acetone. The second acetone cut (P2) was resuspended in 5% acetic acid and subjected to C-8 HPLC (Figure 3). To find the 50 kDa protein, sample fractions were collected at 1 min intervals, freeze-dried, and rerun on a 10% SDS gel alongside the crude extracts from different parts of the threads. SDS PAGE of the HPLC fractions located the 50 kDa proximal thread specific protein in the 55-min HPLC peak (Figure 3A). This protein was named “proximal thread matrix protein 1” or PTMP1, and its migration on SDS PAGE is shown in Figure 2A (lane 5). The actual mass of HPLC-purified PTMP1 was determined to be 49 997 Da using MALDITOF mass spectrometry (not shown). Amino acid analysis of purified PTMP1 following acid hydrolysis detected no DOPA, hydroxyproline, or other posttranslational modifications typical of previously characterized byssal proteins (Table 1). The amino acid composition of PTMP was unremarkable with 12.2 mol % glycine; 11.4 mol % asparagine/aspartic acid and 9.9 mol % glutamine/glutamic acid are the next most abundant residues. Cystine was detected at levels of 0.7%, which corresponds to about three disulfides per 50 kDa (Table 1). The cDNAdeduced sequence of PTMP1 revealed six cysteine residues, which would appear to be oxidized in the mature protein. Slight discrepancies between the composition of purified PTMP1 and that calculated from the cDNA-deduced sequence of variant a (Table 1) were likely due to the presence of multiple variants in the purified protein.

1244


Sun et al.

Figure 2. SDS PAGE and Western Blots of PTMP1. Panel A shows 0.5 mL of high molecular weight standards (ovalbumin, bovine serum albumin, phosphorylase, galactosidase, myosin (BioRad) (lane 1), crude distal thread extract (lane 2), transition thread extract (lane 3), proximal thread extract (lane 4), and HPLC-purified PTMP1 (lane 5). Panel B shows the Western blotted version of panel A (lanes 2 and 4) using polyclonal anti-PTMP1 antiserum. Approximately equal amounts of proximal (lane 1) and distal (lane 2) proteins were applied. Panel C shows a Western blot of purified PTMP1 reacted with commercial polyclonal anti-vWF antiserum (1:1000 dilution) (lane 1) and high molecular weight standards (as negative controls). Panel D shows recombinant vWF A3 domain reacted with anti-PTMP1 (1:1000 dilution) (lane 1). Because of the small size of the vWF A3 domain, the gel acrylamide concentration was increased to 16% (w/v). Molecular weight markers are as indicated at right. Table 1. Amino Acid Compositions of PTMP1 in mol % (residues/ 100 residues) or mol/mola

Figure 3. Reversed phase HPLC purification of PTMP1 and Lys-C digested PTMP1 peptides: (A) elution profile of PTMP1 on a C-8 column. Pure PTMP1 eluted at an acetonitrile concentration of 60% (55 min). Panel B shows the elution profile on a C-18 column of PTMP1-derived peptides produced by Lys-C digestion. The peptide at 44 min was subjected to Edman sequencing. Absorbance was monitored at 215 nm. Flow rate was 1 mL/min.

Direct peptide sequencing of PTMP1 procured two partial sequences. The N-terminus of intact PTMP1 revealed 15 residues: MGHHGVMPYKAVPYE. Another internal peptide, produced by Lys-C endoproteinase digestion of PTMP1 (no. 44; Figure 3B), yielded 25 residues of additional sequence: VTPSIIGQTAIGDGLENARLEVFPN. The binding of Con-A by electroblotted PTMP1 suggested glycosylation by R-D-manno- or R-D-glucopyranosyl residues or both (not shown). This was confirmed by HPLC-PAD analysis for neutral sugars following PTMP1 hydrolysis:

residue

PTMP1 (mol %)

PTMP1 (mol/mol)

PTMP1 cDNA (mol/mol)

Asxb Thr Ser Glxb Pro Gly Ala Cys/2 Val Met Ile Leu Tyr Phe His Lys Arg Trp

11.4 4.6 8.8 9.9 4.5 12.2 7.1 0.7 7.8 2.1 5.5 5.7 2.5 4.3 2.5 7.6 3.0 nd

51 20 38 44 18 53 31 3 34 9 24 25 11 19 11 34 13

51 26 32 45 16 45 29 3 39 11 28 25 9 21 9 38 11 0

a Mol % composition for PTMP1 represents the average of three different analyses of purified protein. The conversion of mol % to mol/mol assumes a molecular weight of 50 kDa. The cDNA-deduced composition is for M. edulis variant a. nd ) not detected. b Asx ) asparagine/aspartic acid; Glx ) glutamine/glutamic acid.

each mole of PTMP1 contained 4 equiv of mannose, 3 equiv of glucose, 2 equiv of fructose and 1 equiv each of fucose and arabinose, as well as miscellaneous trace sugars. The specificity of the polyclonal anti-PTMP1 antiserum (Figure 2B) was determined by Western blotting SDS PAGEseparated proteins that were extracted by acid-urea from separate lots of proximal and distal thread portions. A 50 kDa protein in the proximal extract was recognized; nothing reacted in the distal portion. Commercial antiserum against human vWF (Figure 2C) also cross-reacted with PTMP1. Given the presence of four different domains (A, B, C, and D) in vWF, we were intrigued to ascertain whether antiPTMP1 recognized a recombinant version of human vWF


Figure 4. Immunohistochemical detection of PTMP1 in byssal threads using colloidal gold labeling. Tissue sections were 2 mm thick and cut at oblique to perpendicular angles to the axis of the thread. Sections were exposed first to the primary anti-PTMP1 antibody followed by a secondary colloidal gold-labeled (antiFc) antibody. Panel A shows the proximal thread section exposed to anti-PTMP1; Panel B shows the proximal thread with preimmune serum; Panel C shows the distal thread treated with anti-PTMP1; Panel D shows the distal thread with preimmune serum. Scale bars are as shown.

domain A3. It did and strongly so at antiserum to buffer dilutions of 1:1000; however, monoclonal anti-vWF domain A3 (RU518) did not recognize PTMP1. This outcome was not unexpected given that this monoclonal antibody required a specific epitope consisting of three nonlinear sequences, the third of which (MHGAR) was missing in PTMP1 (see below). Colloidal gold-based immunohistochemical detection of PTMP1 was performed with anti-PTMP1 antiserum on sections of proximal and distal threads and examined by light microscopy. PTMP1 was limited to the proximal portion, where it was distributed as islets or “freckles” in the core of the thread (Figure 4A). Transmission electron microscopy will be required to explore the spatial relationship of PTMP1 with the preCols. A collagen-binding ELISA assay using microtiter plates indicated that PTMP bound type I collagen despite the presence of high concentrations of BSA in the medium (Figure 5). It would have been preferable to use byssal preCols instead of type I collagen in the binding assay, but purified native preCols are not yet available. The apparent binding affinity for collagen based on a Lineweaver-Burk plot of the data was about 0.25 µM PTMP1. PTMP1 binding to collagen was not perturbed by addition of EDTA or magnesium to the assay medium (Figure 5, inset). Cloning the PTMP1 cDNA. The PCR amplification of mussel foot cDNA using two sequence-based degenerate oligonocleotides (P2L-2f and P2L-4r) gave a single product by agarose gel electrophoresis. This gel band was purified, cloned, and sequenced (∼640 bp). From this known piece of cDNA sequence, another two forward primers were designed and the rest of the cDNA sequence for the protein was obtained by repeating the procedure described above. To amplify the entire PTMP1 transcript into a suitable cloning vector, one forward primer, P2L-f, with a sequence


Figure 5. Dose-dependent binding of PTMP1 to immobilized type I collagen as measured by ELISAs. Increasing concentrations of PTMP1 were incubated with immobilized type I kangaroo collagen. The amount of bound protein was determined by an ELISA as described in Experimental Procedures. The inset shows the effect of EDTA (100 mM) and different MgCl2 (10, 50, 100 mM) concentrations on PTMP1 (0.12 mM) binding to collagen (type I kangaroo tail). Bars represent mean of three plus or minus SD readings.

of ATGGGGCATCATGGGGTAATGC (N-terminus) and one reverse primer, P2L-r, with a sequence of TTATCCAATGGCTCCTGATCCT (C-terminus) were designed and used for the PCRs on the cDNA library. The whole cDNA was cloned and resequenced to ensure that it was the same one as the aligned sequence obtained above. Sequencing of M. edulis PTMP1 cDNA detected two variants. Variant a was 1332 bp long and encoded a polypeptide of 444 amino acids (Figure 6); variant b was 1323 bp long and encoded a polypeptide 441 amino acids in length. Application of REPRO (http://mathbio.nimr.mrc.ac.uk) to PTMP1 (variant a) revealed the presence of two internal repeats, each about 195 amino acids in length: repeat I spanned residues 39-235 and repeat II residues 238-431. These repeats are 44% identical and together represent more than 85% of the protein. A database search for possible posttranslational modifications (www.expasy.ch, posttranslational modification prediction) suggests one O-glycosylation (T319) and two N-glycosylation sites (N88 and N271) in variant a. The molecular mass of PTMP1 deduced from its cDNA sequence was 47 595 Da, which is 2.4 kDa less than the mass measured by the MALDI-TOF. Given the results of neutral sugar analysis and positive ConA-binding, glycosylation is expected to account for some of the remaining difference of 2.4 kDa. Almost 15 hexosyl groups could be accommodated in mature PTMP1; at least 11 sugar equivalents were detected per PTMP1. It is also possible that part of the mass discrepancy could be due to other posttranslational modifications not yet detected. For example, several serine, threonine, and tyrosine residues in PTMP1 have high probabilities as phosphorylation sites, for example, >0.9 according to NetPhos2.0 (www.expasy.ch). Such modifications, however, have not been detected in purified protein. A database search for sequence homology revealed that both repeat domains of PTMP1 contain significant similarity to A domains, whether they are from human von Willebrand Factor or integrin R1I. For purposes of example, a sequence

1246


Sun et al.

Figure 6. cDNA and translated amino acid sequence of PTMP1. In panel A, the underlined amino acids indicate the sequenced peptides. The v symbol indicates the signal peptide cleavage site the first residue of which is +1 of the mature sequence; b denotes the stop codon. Panel B shows a schematic illustration of the repeat structure of PTMP1.

beginning toward the end of the first repeat (residue 167) and extending beyond the middle of the second repeat (residue 361) has been aligned with A domain sequences from von Willebrand Factor and integrin R1I (Figure 7). Notably, only 20 of 200 residues are identical to all three proteins. Identity with domain A of R1I was 26%, while identity with A3 of vWF was about 28%; homology with both is considerably higher (>50%) when similar chargefor-charge or hydrophobic substitutions are considered. While

these identities are not particularly high, it must be pointed out that even the identity between the R1I A domain and the vWF A domain (both of which are collagen binding and have Rossmann folds) is only about 30%. There are two putative metal ion-dependent adhesion sites (MIDAS) for collagen binding having signature sequences related to DXSXS. One such sequence is DDSSS (variant a residues 51-55; variant b residues 48-52); the other one is DASSS (variant a residues 250-254; variant b residues 247-251). A Val-,



Table 2. Comparison of Sequence Differences in the Two PTMP1 Variants (a and b) from M. edulis and PTMP1 (variant b gallo) from M. galloprovincialisa

a

PTMP1

AAb

N- terminal sequence

C- terminal sequence

variant a (edulis) variant b (edulis) variant b (gallo)

441 444 444

MGHHGVMPYKAVP--YESP-VATSPTKYKP MGHHGVMPYKAVPVSYDPPVVAVDPPPYQP MGHHGVMPYKAVPVSYDPPVVAVDPPPYQP

ENVVKLACMSCKPRAHKRGSGAIG ENVVKLACMSCKPRAHKRGSGAIG EDVVKLACMSCKPRAHKRGSGAIG

Bold letters are used to highlight sequence differences. b AA refers to the total number of residues in each protein.

Figure 7. Multiple sequence alignment of PTMP1 (ptmp) from M. edulis with the A domain region of human vWF (vwfa) and integrin RI (int1). PTMP1 sequence (residues 167-361) is mostly from the second repeat region (middle). Sequence of von Willebrand factor (residues 1609-1809, top) corresponds roughly to domain A3. Domain A of human integrin RI (residues 74-257) is shown at bottom. Identity in all three is denoted by bold, while identity of ptmp1 with either int1 or vwfa is with shaded italics. Nonlinear sequences that serve as epitopes for monoclonal RU5 are underlined. MIDAS sequence of integrin RI is denoted by wavy line. Sequence alignment was obtained using LALIGN (www.ch.embnet.org/cgi-bin/LALIGN).

Leu-, and Asp-rich C-terminal sequence is also shared by all three (Figure 7). PTMP1 sequence of M. galloproVincialis has a 97% identity with M. edulis variant a and a 99% identity with variant b (Table 2). The corresponding cDNA sequences have 94% and 96% identity, respectively. Codon usage is essentially identical. Discussion Despite the biological peculiarity of byssal threads, the ingenious fashion in which fiber stiffness is controlled by a graded shuffling of collagen, silklike and elastin-like tensile elements along the longitudinal axis of each thread represents an important biomolecular paradigm.3,7 Elastin and the proximal portion of byssus share several attributes including an extensibility to about 200%, recovery of initial length, and a primary structure that includes tandem repeats of an elastic pentapeptide sequence.7,10 However, proximal byssus is unlike vertebrate elastin or invertebrate resilin and abductin in other properties such as low resilience, that is 50% as compared with 95% in elastin,7,8 high collagen content,7 and a very high matrix/fiber ratio.12 The identification of matrix components and their interaction with fibers is thus crucial to understanding the formation and complex mechanical behavior of these composite elastomers. PTMP1 is the first matrix-associated protein to be characterized from byssus. Direct MALDI mass spectrometry, Western blotting of proteins from byssal threads, and immunohistochemistry suggest that PTMP1 is preferentially

localized in and an important component of the elastic proximal portion. The most apparent feature of PTMP1 is the presence of vWF-like A domains that comprise more than 80% of the primary structure. There are few other reports of extraorganismic proteins with vWF domains; one is spiggin, a protein with vWF D domains that is secreted and used by male stickleback fish for nest building.24 Another protein (PbWARP) with A domains is secreted from the microneme of Plasmodium in the ookinete stage.25 Interestingly, all three proteins appear to be associated with fibers: PTMP1 and the microneme-derived protein bind collagens, and spiggin binds plant fibers in the fish nests. The original vWF is a large multimeric glycoprotein present in blood plasma, endothelial cells, and the subendothelial matrix of the vessel wall.14 It is critical to both hemostasis and thrombosis by mediating the attachment of platelets to exposed tissues and subsequent platelet aggregation. vWF contains four different types of repeated domains: A, B, C, and D. Of these, type A domains are widely distributed in other extracellular protein superfamilies26 including integrins,27 complement,28 nonfibrillar collagens,29 some microfibrillar, anchoring, and fibrilassociated collagens,30 and cartilage matrix proteins, that is, matrilins.31-33 Most type A domains are known to be involved in binding various macromolecular ligands.26 These include collagens,33-35 laminin,36 and glycosaminoglycans such as heparin and hyaluronan.37 One of the vWF A domains, A3 in particular, interacts specifically with fibrillar collagen types I and III.33,38 A domains also play a role in the self-assembly of some matrix proteins into microfilamentous networks.30 On the strength of the sequence homologies of PTMP1 to integrin RI and vWF A domains (Figure 7) and the cross-reactivity of anti-PTMP1 with recombinant vWF-A3 and of anti-vWF with PTMP1, it is reasonable to conclude that PTMP1 contains recognizable features of A domains. This is further supported by the PTMP1 binding affinity for collagen. The apparent dissociation constant of 0.25 µM is tighter than that for recombinant vWF domain A3 and type I collagen (2 µM34), but less specific than the interaction between the integrin R2I A domain and collagen (10 nM33). The mechanism of collagen binding by A domains and their homologues has recently come under intense scrutiny. The A domain of integrin RI, which has a 30% identity with vWF A3 and binds to collagen, was crystallized with a bound collagen peptide.38 On the integrin side, the binding locus involves metals, such as Mg, which is octahedrally complexed to a DXSXS MIDAS sequence. Upon binding of the RI to collagen, one coordination site of Mg is replaced by a Glu side chain from the collagen.27;38 Although vWFassociated A3 domain also contains a vestigial DXSXS

1248


sequence and binds collagens, recent crystallographic studies using recombinant A3 domain and a specific collagenblocking antibody fragment (RU5) indicate that the binding site is not near the DXSXS sequence nor is it metaldependent like integrin RI.18,39 As with vWF A3, collagen binding by PTMP1 is also metal-independent. The mechanism of binding, however, is not likely to resemble that of vWF because PTMP1 lacks His in the sequence MHGAR (human vWF residues 1785-1789), which was shown to be essential for collagen binding by vWF.18 The presence of binding domains in PTMP1 likens it to a multifunctional cross-bridging macromolecule, an excellent property for a matrix that is capable of holding together fibers in a composite material. It is instructive to note that the fibercross-bridging ability of proteins such as R-actinin and vWF, for example, is not a static one but rather is shear-dependent. Thus, both R-actinin and vWF A domain exhibit increased adhesion to actin and collagen, respectively, during increased shear.40,41 Could a similar dynamic cross-bridging of byssal collagens by PTMP1 account for the strain-induced stiffening observed in proximal threads subjected to repetitive loading?12 Future work will explore the correlation between mechanical properties and the dynamic molecular relationships between byssal collagens and matrix proteins. Acknowledgment. We thank Loren Knapp of USCColumbia for successful preparation of specific anti-PTMP antibodies. Jan Sixma (University Medical Center of Utrecht) generously provided the recombinant vWF A3 domain and monoclonal anti-A3 antibody. Craig Carlson and Stuart Goldberg helped with the neutral sugar analysis, and Kathryn Coyne contributed her cDNA library of the Mytilus edulis foot. This work was supported by a grant from the Biomaterials Program at NIDCR (J.H.W.). GenBank accession numbers for PTMP1 cDNA are nos. AF414454, AY053390, and AY053391. References and Notes (1) Turner, R. D.; Rosewater, J. Johnsonia 1958, 3, 285-326. (2) Yonge, C. M.; Thompson, T. E. LiVing Marine Molluscs; Collins: London, 1976; pp 171-172. (3) Deming, T. J. Curr. Opin. Chem. Biol. 1999, 3, 100-105. (4) Vollrath, F.; Knight, D. P. Nature 1999, 410, 541-548. (5) Vitellaro-Zuccarello, L. J. Ultrastruct. Res. 1980, 73, 135-147. (6) Hayashi, C.; Shipley, N. H.; Lewis, R. V. Int. J. Biol. Macromol. 1999, 24, 271-275. (7) Waite, J. H.; Vaccaro, E.; Sun, C.-J.; Lucas, J. M. Philos. Trans. R. Soc. London 2002, 357, 143-153. (8) Gosline, J., Lillie, M.; Carrington, E.; Guerette, P.; Ortlepp, C.; Savage, K. Philos. Trans. R. Soc. London 2002, 357, 121-132. (9) Qin, X. X.; Coyne, K. J.; Waite, J. H. J. Biol. Chem. 1997, 272, 32623-32627.

Sun et al. (10) Coyne, K. J.; Qin, X. X.; Waite, J. H. Science 1997, 277, 18301832. (11) Waite, J. H.; Qin, X. X.; Coyne, K. J. Matrix Biol. 1998, 17, 93106. (12) Vitellaro-Zuccarello, L.; De Biasi, S.; Bairati, A. Tissue Cell 1983, 15, 547-554. (13) Sun, C. J.; Vaccaro, E. V.; Waite, J. H. Biophys. J. 2001, 81, 35903596. (14) Vaccaro, E. V.; Waite, J. H. Biomacromolecules 2001, 2, 906-911. (15) Ruggeri, Z. M.; Ware, J. FASEB J. 1993, 7, 308-316. (16) Qin, X.; Waite, J. H. J. Exp. Biol. 1995, 198, 633-644. (17) Borch, N. H.; Kirchman, D. L. Mar. Chem. 1997, 57, 85-95. (18) Romijn, R. A.; Bouma, B.; Wuyster, W.; Gros, P.; Kroon, J.; Sixma, J. J.; Huizinga, E. G. J. Biol. Chem. 2001, 276, 9985-9991. (19) Glauert, A. M. In Practical Methods in Electron Microscopy: Fixation, Dehydration and Embedding of Biological Specimens; North-Holland Publishing: Amsterdam, 1975; pp 123-170. (20) Sun, C. J. Matrix protein PTMP1 and its possible role in the biomechanics of mussel byssal thread. PhD Dissertation, University of California, Santa Barbara, CA, 2002. (21) Fischer, B. E.; Thomas, K. B.; Dornen, F. Ann. Hematol. 1998, 76, 159-166. (22) Coyne, K. J. Cloning and characterization of the byssal collagen, preCol-P, from the marine mussel Mytilus edulis. PhD Dissertation, University of Delaware, 1997. (23) Waite, J. H. J. Biol. Chem. 1983, 258, 2911-2915. (24) Jones, I.: Lindberg, C.; Jakobsson, S.; Hellqvist, A.; Helman, U.; Borg, B.; Olsson, P.-E. J. Biol. Chem. 2001, 276, 17857-17863. (25) Yuda, M.; Yano, K.; Tsuboi, T.; Torii, M.; Chinzei, Y. Mol. Biochem. Parasitol. 2001, 116, 65-72. (26) Colombatti, A.; Bonaldo, P.; Doliana, R. Matrix Biol. 1993, 13, 297306. (27) Bella, J.; Berman, H. M. Structure 2000, 8, R121-R126. (28) Perkins, S. J.; Smith, K. F.; Williams, S. C.; Haris, P. I.; Chapman, D.; Sim, R. B. J. Mol. Biol. 1994, 238, 104-119. (29) Yamagata, M.; Yamada, K. M.; Yamada, S. S.; Shinomura, T.; Tanaka, H.; Nishida, Y.; Obara, M.; Kimata, K. J. Cell Biol. 1991, 115, 209-221. (30) Chen, Q.; Zhang, Y.; Johnson, D. M.; Goetinck, P. F. Mol. Biol. Cell 1999, 10, 2149-2162. (31) Dea´k, F.; Wagener, R.; Kiss, I.; Paulsson, M. Matrix Biol. 1999, 18, 55-64. (32) Argraves, W. S.; Dea´k, F.; Sparks, K. J.; Kiss, I.; Goetinck, P. F. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 464-468. (33) Xu, Y.; Gurusiddappa, S.; Rich, R. L.; Owens, R. T.; Keene, D. R.; Mayne, R.; Ho¨o¨k, A.; Ho¨o¨k, M. J. Biol. Chem. 2000, 275, 3898138989. (34) Pareti, F. I.; Niiya, K.; McPherson, J. M.; Ruggeri, Z. M. J. Biol. Chem. 1987, 262, 13835-13841. (35) Dickeson, S. K.; Walsh, J. J.; Santoro, S. A. J. Biol. Chem. 1997, 272, 7661-7668. (36) Calderwood, D. A.; Tuckwell, D. S.; Eble, J.; Ku¨hn, K.; Humphries, M. J. J. Biol. Chem. 1997, 272, 12311-12317. (37) Kielty, C. M.; Whittaker, S. P.; Grant, M. E.; Shuttleworth, C. A. J. Cell Biol. 1992, 118, 979-990. (38) Emsley, J.; Knight, C. G.; Farndale, R. W.; Barnes, M. J.; Liddington, R. C. Cell 2000, 101, 47-56. (39) Bienkowska, J.; Cruz, M.; Atiemo, A.; Handin, R.; Liddington, R. J. Biol. Chem. 1997, 272, 25162-25167. (40) Xu, J.; Tseng, Y.; Wirtz, D. J. Biol. Chem. 2000, 275, 35886-35892. (41) Smith, C.; Estavillo, D.; Emsley, J.; Bankston, L. A.; Liddington, R. C.; Cruz, M. A. J. Biol. Chem. 2000, 275, 4205-4209.

BM0255903

Collagen-Binding Matrix Proteins from Elastomeric Extraorganismic

Recommend Documents