Báo cáo khoa học: Comprehensive sequence analysis of horseshoe crab cuticular proteins and their involvement in transglutaminase-dependent cross-linking potx

Proteins appearing on the gel were designa-ted high molecular mass chitin-binding proteins, and these proteins were then grouped into classes based on their approximate isoelectric point

cuticular proteins and their involvement

in transglutaminase-dependent cross-linking

Manabu Iijima*, Tomonori Hashimoto*, Yasuyuki Matsuda, Taku Nagai, Yuichiro Yamano,

Tomohiko Ichi, Tsukasa Osaki and Shun-ichiro Kawabata

Department of Biology, Faculty of Sciences, Kyushu University, Fukuoka, Japan

The arthropod cuticle functions principally as an

exo-skeleton covering the total body surface, and is a

highly organized structure produced by extracellular

secretion from the epidermis It is constructed as a

composite consisting of chitin filaments (a

homopoly-mer of N-acetyl glucosamines conjugated by b-1,4

linkages), structural proteins, lipids, catecholamine

derivatives, and minerals Its structural properties,

however, vary among species and also according to surface location and developmental stage in individuals [1–3] The mechanical properties of the cuticle depend

on the content of chitin, the microarchitecture of chitin filaments, and the interaction between the chitin-filament system and cuticular proteins Furthermore, the cuticle can be modified by sclerotization, namely the oxidative incorporation of o-diphenols into cuticular

Keywords

chitin-binding proteins; exoskeleton;

horseshoe crab; innate immunity;

transglutaminase

Correspondence

Shun-ichiro Kawabata, Department of

Biology, Faculty of Sciences, Kyushu

University, Fukuoka 812–8581, Japan

Tel & Fax: +81 92 6422632

E-mail: skawascb@mbox.nc.kyushu-u.ac.jp

Note

nucleotide sequence data are available in

the DDBJ databases under the accession

numbers AB201765, AB201766, AB201767,

AB201768, AB201769, AB201770,

AB201771, AB201772, AB201773,

AB201774, AB201775, AB201776,

AB201777, AB201778 and AB201779.

*These authors contributed equally to this

work.

(Received 14 June 2005, revised 25 July

2005, accepted 29 July 2005)

doi:10.1111/j.1742-4658.2005.04891.x

Arthropod cuticles play an important role as the first barrier against inva-ding pathogens We extensively determined the sequences of horseshoe crab cuticular proteins Proteins extracted from a part of the ventral side of the cuticle were purified by chitin-affinity chromatography, and separated by two-dimensional SDS⁄ PAGE Proteins appearing on the gel were designa-ted high molecular mass chitin-binding proteins, and these proteins were then grouped into classes based on their approximate isoelectric points and predominant amino acid compositions Members of groups designated basic G, basic Y, and acidic S groups contained a so-called Rebers and Riddiford consensus found in arthropod cuticular proteins Proteins desig-nated acidic DE25 and DE29 each contained a Cys-rich domain with sequences similar to those of insect peritrophic matrix proteins and chitin-ases In contrast, basic QH4 and QH10 contained no consensus sequences found in known chitin-binding proteins Alternatively, a low molecular mass chitin-binding fraction was prepared by size exclusion chromatogra-phy, and 15 low molecular mass chitin-binding proteins, named P1 through P15, were isolated With the exception of P9 and P15, all were found to be identical to known antimicrobial peptides P9 consisted of a Kunitz-type chymotrypsin inhibitor sequence, and P15 contained a Cys-rich motif found in insulin-like growth factor-binding proteins Interestingly, we observed transglutaminase-dependent polymerization of nearly all high molecular mass chitin-binding proteins, a finding suggests that transgluta-minase-dependent cross-linking plays an important role in host defense in the arthropod cuticle, analogous to that observed in the epidermal cornified cell envelope in mammals

Abbreviations

DCA, monodansylcadaverine; IGFBP, insulin-like growth factor-binding protein; HMM, high molecular mass; LMM, low molecular mass;

R & R, Rebers and Riddiford; RACE, rapid amplification of cDNA ends; TGase, transglutaminase.

matrix [4–7] For insect cuticular proteins sufficient

sequence information is available to allow recognition

of consensus sequences The motif first identified is

the so-called Rebers and Riddiford (R & R)

consen-sus: -G-X(8)-G-X(6)-Y-X-A-X-G-X-G-Y-X(7)-P-X(2)-P-,

where X represents any amino acid, and the values in

parentheses indicate the number of intervening residues

[8–10] A slightly modified R & R consensus has also

been reported, -G-X(7)-(D, E, or N)-G-X(6)-(F or

Y)-X-A-(D, G, or N)-X(2 or 3)-G-(F or Y)-X-(A or

P)-X(6)-P [3] The region flanking the N-terminus of the

R & R consensus is enriched in hydrophilic amino acid

residues [11,12], and a conserved stretch of

approxi-mately 68 amino acid residues is referred as the

exten-ded R & R consensus [13] The extenexten-ded R & R

consensus has no sequence similarity to other known

cysteine-containing chitin-binding domains [14], such

as plant chitin-binding proteins and horseshoe crab

antimicrobial peptides with chitin-binding affinity

[15–17] Rebers and Willis reported that the extended

R & R consensus is a chitin-binding domain and

pos-tulated that the conserved aromatic residues in this

domain may play an important role in chitin binding

[18] Short consensus repeats with the sequences

-A-A-P-(A or V)- and -G-Y-G-G-L- have also been

identi-fied in insect cuticular proteins [4]

Although cuticular proteins from other arthropods

have not yet been characterized to the same extent as

those from insects, a sequence motif similar to R & R

consensus has been identified in proteins from an

arachnid, the spider Araneus diadematus [19], as well

as in those from a crustacean, the lobster Homarus

americanus [20,21] Several cuticular proteins from

H americanus and the crab Cancer pagurus contain a

repeating consensus sequence that is 18-residues in

length and contains Gly residues at positions 4, 7 and

13: -X-u-u-G-P-S-G-u-u-X-X-(D⁄

N)-G-X-X-X-Q-u-(where u represents a hydrophobic residue) Gly

resi-dues within this motif may play an important role in

maintaining a conformation required for the essential

interaction with calcium ions in the exoskeletal matrix

[20–22] Reports in the literature regarding the amino

acid composition and sequence of horseshoe crab

pro-teins have been sparse, and have primarily dealt with

the American horseshoe crab Limulus polyphemus [23–

25] A cuticular extract from the silkworm Bombyx

mori has been shown to contain prophenoloxidase

[26,27], an enzyme that typifies arthropod innate

immune response and which is activated during the

related processes of sclerotization, exoskeletal wound

healing, and host defense response to microorganisms

[28,29] In order to identify novel cuticular

chitin-bind-ing proteins and cuticular proteins involved in innate

immunity, we undertook an extensive examination of cuticle proteins from the horseshoe crab Tachypleus tridentatus This investigation involved fractionation

by chitin-affinity chromatography, two-dimensional SDS⁄ PAGE (2D SDS⁄ PAGE) and reverse-phase HPLC, and culminated in the determination of numer-ous cuticle protein sequences

Results and Discussion

Separation of cuticular chitin-binding proteins The cuticular proteins of T tridentatus were extracted with acetic acid, and the resulting extract was subjec-ted to chitin-affinity column chromatography to obtain chitin-binding proteins Proteins bound to chitin were eluted with acetic acid and lyophilized Based on an extinction coefficient of 10 at 280 nm for a 1% protein solution, we estimate that approximately 20 mg of chi-tin-binding proteins were reproducibly obtained from

5 g of carapace fragments The cuticular chitin-binding proteins were separated by 2D SDS⁄ PAGE over a range of isoelectric points from 3 to 10, and were dis-tributed into two (acidic and basic) clusters on the gel (Fig 1A) A similar pattern on 2D SDS⁄ PAGE was observed for the cuticular chitin-binding proteins extracted with 8 m urea (data not shown) Proteins extracted with 10% acetic acid were used in all subse-quent experiments The majority spots corresponded to proteins with apparent molecular masses of 16, 20 and

25 kDa, with most of these spots clustered in the basic region In contrast, spots with higher molecular masses (ranging from 67 to 94 kDa) were observed in the acidic region of the gel In the neutral region, only two major spots (32 and 38 kDa) were present Thirty-seven spots on the gel were subjected for amino acid composition and sequence analyses

Cuticular proteins extracted from cuticle of the American horseshoe crab L polyphemus with 6 m urea containing 0.1% trifluoroacetic acid have been separ-ated by 2D SDS⁄ PAGE over a rage of isoelectric points from 3.5 to 10 [25] Their isoelectric points range from 6.5 to 9.2, and the 2D gel shows no protein spots in the acidic range from 3.5 to 5.5, corresponding to the members of acidic S and acidic DE in T tridentatus,

as described below This discrepancy in protein spots

on the 2D gels between L polyphemus and T tridenta-tus may be due to a discrepancy in age of the samples, juvenile (L polyphemus) and adult (T tridentatus) Using a complementary approach, we also analyzed low molecular mass (LMM) chitin-binding proteins The eluate from the chitin affinity column was fractionated

by gel filtration on a Sephadex G-50 column, and

fractions containing proteins determined by SDS⁄

PAGE analysis to be less than 10 kDa were further

fractionated by reverse-phase HPLC to obtain LMM

chitin-binding proteins (Fig 1B) A single fraction from

the gel filtration step contained a protein that appeared

as a single band on SDS⁄ PAGE with an apparent

molecular weight of 10 kDa This protein, which we

designated P1, was identified as big defensin [30] by

amino acid sequence analysis Proteins isolated by

reverse-phase HPLC were designated P2 through P15

Previous reports have demonstrated that arthropod

cuticular proteins may be resistant to extraction In the

beetle Agrianome spinnicollis, for example, more than

50% of total cuticular protein is retained following

extraction [31] It is therefore possible that the proteins

obtained here may not be representative of all

cuticu-lar chitin-binding proteins in T tridentatus

Amino acid compositions of high molecular mass

(HMM) chitin-binding proteins

HMM chitin-binding proteins resolved by 2D

SDS⁄ PAGE were categorized into seven groups based

on amino acid composition: basic G, basic Y, basic

QH, neutral, acidic S, acidic DE, and others (Table 1)

The proteins in the basic G group had a disproportion-ately high content of Gly (19–35%) Those in the basic

Y group were characterized by a high content of Tyr (10–15%), Gly (16–19%), and Asp (10–12%) Proteins

in the basic QH group were abundant in Glu (13–16%) and His (7%) as determined by amino acid analysis, but their partial amino acid sequences indicated a high content of Gln rather than Glu Proteins in the neutral group were abundant in Ala (15%), Pro (12%) and Gly (11–12%), and proteins in the acidic S group, while similar otherwise to those in the neutral group, were additionally characterized by an abundance of Ser (9– 12%) The members of acidic DE group had a high content of Asp (14%) and Glu (10–14%)

N-Terminal amino acid sequence analysis of HMM chitin-binding proteins

Eleven of 37 spots observed on 2D SDS⁄ PAGE were resistant to sequence analysis by Edman degradation, presumably due to N-terminal blocking, whereas the remaining spots yielded sequences with lengths between five and 78 residues (Table 1) The N-terminal sequence of neutral 2 was identical to that of acidic S37 despite a significant difference in their apparent

B A

Fig 1 (A) 2D SDS ⁄ PAGE of chitin-binding proteins For experimental details see text The members of the basic Y, basic G, basic

QH, neutral, acidic S and acidic DE groups are bounded, and numerical designations used in the text are indicated (B) Reverse-phase HPLC chromatograph of LMM binding proteins LMM cuticular chitin-binding proteins were resolved by reverse-phase HPLC using an acetonitrile gradient.

Table 1 N-Terminal amino acid sequences of cuticular chitin-binding proteins ND, not detectable.

YFHGGHAGAF AGGVGGGLAR GYYGQQPV 78

isoelectric points and molecular masses (33 and

70 kDa, respectively), indicating that neutral 2 may be

a proteolytic product corresponding to the N-terminus

of acidic S37 Alternatively, neutral 2 and acidic S37

may be isoforms of one another, or result from

differ-ential mRNA splicing Certain members of the basic G

group, such as G13, G14, G15 and G16 had similar

apparent molecular mass of about 25 kDa, and were

characterized by an abundance of Gly near the amino

terminus The N-terminal 78 residues of basic G13, for

example, contained 34 Gly residues The four proteins

of this group were identical within at least the first 15

residues, whereas their isoelectric points differed,

sug-gesting that they are either isoforms of one another or

are differentially post-translationally modified

Simi-larly, within the acidic DE group, acidic DE25, DE26

and DE27 were identical throughout the first 18

resi-dues and had similar apparent molecular masses

(20 kDa), though they differed in apparent isoelectric

points In contrast, while the N-terminal sequences of

acidic S33 and acidic S36 were identical, these proteins

differed from one another both in isoelectric point and

in apparent molecular mass (25 and 40 kDa,

respect-ively), suggesting that acidic S33 may be a proteolytic

fragment of acidic S36 In addition, the N-terminal

sequence of acidic S37 was highly similar to those of

acidic S33 and S36, indicating that the acidic S may

contain several protein isoforms Finally, Tyr residues

occurred repeatedly within the N-terminal sequences

of all basic Y proteins (basic Y6 through Y9), and

repeats of the di- and tri-peptide sequences QQ and

QQQ were observed in the amino terminal sequences

of basic QH proteins 4 and 10

Nucleotide sequences of HMM chitin-binding

proteins

cDNA fragments of HMM chitin-binding proteins

were amplified by PCR using degenerate

oligonucleo-tide primers based on amino acid sequences derived

from intact proteins or from proteolytic fragments

thereof Sense and antisense primers based on the

resulting cDNA sequences were selected and used to

amplify full-length cDNAs by RACE PCR, resulting

in 11 full-length and two partial cDNA clones for

HMM chitin-binding proteins

The cDNA of basic G13 encoded a 206 residue

pro-tein with a 16-residue signal peptide Three types of

cDNA clones, designated basic G13A (accession

num-ber AB201771), basic G13B (AB201772) and basic

G13C (AB201773), were identified Gly8 in G13A was

replaced by Glu in G3B, and Leu7 in G13A was

replaced by Val in G13C The cDNA of basic G19

encoded a 161-residue protein with a 16-residue signal peptide (AB201774) A homologous cDNA, designated G19 h (AB201775) encoded a 141-residue protein with

a 16-residue signal peptide The basic G13 variants, G19 and G19 h all contained an R & R consensus sequence, and they exhibited significant sequence simi-larity (G19 and G13, 58% identity; G19 h and G13, 48%; G19 and G19 h, 64%)

A partial basic Y6 cDNA lacked the 5¢-region, but overlapped with the region determined by Edman degra-dation of the intact protein, thereby allowing deduction

of the sequence of a mature protein consisting of 143 residues (AB201768) The basic Y7 cDNA encoded a mature protein of 131 residues (AB201769) At the pro-tein level, basic Y6 and Y7 showed 50% sequence iden-tity overall, and both possessed an R & R consensus sequence A blast homology search using basic Y6 and Y7 revealed significant sequence similarity between these proteins and Ad-ACP15.7, a cuticular protein from the spider A diadematus [19] (Y6 vs Ad-ACP15.7, 33% identity; Y7 vs Ad-ACP15.7, 26% identity) as well

as between these proteins and LpCP14b, a cuticular pro-tein from L polyphemus [25] (Y6 vs LpCP14b, 78% identity; Y7 vs LpCP14b, 50% identity)

The acidic S37 cDNA encoded a 608-residue protein with a 16-amino acid signal peptide (AB201765) The deduced protein sequence contained four tandem repeats of a 68-residue extended R & R consensus sequence, with sequence identity among the four repeats ranging from 66 to 94% In addition the cDNA encoded seven copies of the pentapeptide sequence -A-A-P-A⁄ V-, a short consensus sequence found in insect cuticular proteins [4] A blast homol-ogy search revealed no other regions of similarity between acidic S37 and other known proteins

The members of basic Y and G, and acidic S37 all contain the extended R & R motif commonly found in insect cuticular proteins (Fig 2) The motif found in the horseshoe crab proteins shows the highest sequence similarity to RR-2, one of the three variants of the consensus [9,10] A recombinant protein containing the extended R & R consensus of a putative cuticular pro-tein from the mosquito Anopheles gambiae has been shown to be necessary and sufficient for chitin binding [18] A secondary structure prediction and homology modeling of the extended R & R consensus suggest an antiparallel b-sheet structure [13,32] Interestingly, aci-dic S37 contained four tandem extended R & R con-sensus repeats This tandem R & R repeat structure has not previously been identified in arthropod cuticu-lar proteins, and suggests that acidic S37 may interact polyvalently with chitin fibers to form a stable three-dimensional network In addition to the R & R motif,

basic Y6 and Y7 contain an 18-residue motif found in

the cuticular proteins isolated from calcified regions of

crustacean exoskeletons [20–22], a finding suggests that

basic Y6 and Y7 may play a role in the deposition of

calcium ions required to maintain the mechanical

properties of cuticles (Fig 3)

Basic QH4 cDNAs were identified and shown to

encode a 135-residue protein with a 16-residue acid

sig-nal peptide Two cDNA variants were isolated and

designated basic QH4A (Pro36 and His83) and QH4B

(Leu36 and Tyr8) (AB201766 and AB201767) The

cDNA of basic QH10 encoded a 110-residue protein

with a 16-amino acid signal peptide (AB201770) Basic

QH4 and QH10 showed significant sequence similarity

to one another (53% identity), and neither contained

the R & R consensus As basic QH4 and QH10 do not

contain the R & R consensus, they must have an

unknown a chitin-binding motif A homology search

revealed high sequence similarity (84% identity)

between basic QH 10 and the cuticular protein

LpCP13 from L polyphemus [25], a degree of similarity

that is particularly notable given that the clottable

pro-tein coagulogen shows 70% identity between the two

species [33,34] The sequences of basic QH4 and QH10

can be divided into two regions, the Gln-rich

N-ter-minal half and the C-terN-ter-minal half in which Tyr and

His are abundant It has been proposed that

cross-linking between proteins and chitin fibers in insect

cuticles is mediated by His-catechol-chitin linkages, the

formation of which involves the oxidation of

catechol-amines to quinonoid sclerotizing agents with subse-quent nucleophilic addition to certain His residues within cuticular proteins [35,36] The abundance of His residues in the basic QH proteins therefore raises the possibility that these proteins play an important role in maintenance of the integrity of the exoskeleton

A cDNA of acidic DE25 encoded a 137-amino acid protein and a 22-amino acid signal peptide (AB201776) A partial cDNA of acidic DE29 lacked the 5¢-region and its N-terminal sequence determined

by Edman degradation overlapped to the deduced sequence to obtain the sequence of a mature protein of

120 residues (AB201777) Acidic DE25 and DE29 had

an overall sequence identity of 46%, and showed no sequence similarity to other known cuticular proteins Acidic DE25 and DE29 also lack an R & R consen-sus Rather, they contain six Cys residues within their central region in positions similar those of a Cys-rich motif found in insect chitinases and peritrophic mem-brane proteins [14,37–40] (Fig 4) Peritrophin-44, a major peritrophic membrane protein identified in the larvae of the fly Lucilia cuprina, contains five tandem repeats of the Cys-rich motif as well as several conserved aromatic residues within the proposed domain bound-ary The peritrophic membrane is a semipermeable chi-tinous matrix lining the gut of most insects and is thought to play an important role in the maintenance of insect gut structure, the facilitation of digestion, and the protection from invasion by microorganisms and para-sites [37] The C-terminal three Cys residues of the Cys-rich motif in acidic DE25 and DE29 can be aligned with the C-terminal domain of tachycitin (Cys40 to Cys61), a horseshoe crab chitin-binding protein [15], as well as with the chitin-binding domain of hevein (Cys12 to Cys33), a plant chitin-binding protein [41,42], as shown

in Fig 4 This segment of tachycitin forms an antiparal-lel b-sheet and aligns with the known chitin-binding region of hevein [17] It is therefore likely that the cor-responding region of the Cys-rich motif in acidic DE25

Fig 2 Alignment of the extended R & R consensus regions of cuticular proteins of T tridentatus showing the R & R domains of basic Y6, basic Y7, basic G13, basic G19, basic G19 h, and the four contiguous domains from acidic S37 Highly conserved residues are designated with black boxes Numbers on the right indicate amino acid residue numbers.

Fig 3 Alignment of basic Y6, basic Y7 and an 18-residue motif

identified in Cancer pagurus (motif) [22] Three conserved glycine

residues are indicated by an asterisk u indicates a hydrophobic

resi-due Numbers on the right indicate amino acid residue numbers.

and DE29 is involved in chitin binding In insects,

cuticular proteins containing cysteine residues have not

been reported, but analyses of total cuticles following

performic acid oxidation have demonstrated the

pres-ence of minor amounts of cysteic acid, suggesting the

presence of disulfide bond-containing proteins in insect

cuticles [43,44]

Nucleotide sequences of LMM chitin-binding

proteins

All of the LMM chitin-binding proteins identified,

with the exception of P9 and P15, were determined

to be known antimicrobial peptides, such as tachy-plesin, tachystatin, and their isoforms, or proteolytic fragments thereof (Table 2) In vertebrates, antimi-crobial peptides are expressed on epithelial cell surfa-ces and have been proposed to play a role in innate immunity by acting as ‘natural antibiotics’ [45–47]

In horseshoe crabs, antimicrobial peptides have been shown to be able to induce the intrinsic phenoloxi-dase activity of hemocyanin [48] The localization of antimicrobial peptides in the cuticle therefore sug-gests that these peptides may facilitate wound heal-ing in the exoskeleton in addition to actheal-ing as antimicrobial substances

Fig 4 Alignment of cysteine-rich domains of acidic DE25, acidic DE29, peritrophic membrane protein, chitinase, and antimicrobial peptides The cysteine-rich domains of acidic DE25, acidic DE29, the first of five domains found in peritrophin-44 from the fly L cuprina (Peritrophin) [37], chitinase from the parasitic nematode Brugia malayi (Chitinase) [38], tachycitin from the horseshoe crab T tridentatus (Tachycitin) [15] and hevein from rubber tree (Hevein) [41,42] were aligned The conserved Cys residues designated with black boxes, and the conserved aro-matic amino acids are indicated with asterisks Numbers on the right indicate amino acid residue numbers.

Table 2 Features of cuticular chitin-binding proteins.

Number

Protein

Name

Residue number

Chitin-binding motif

DCA incorporation

D melanogaster homolog

A gambiae homolog

P3 Tachyplesin II or III fragment Antibacterial

P10 Tachystatin B2 fragments Antibacterial

P14 Tachystatin A fragment Antibacterial

The cDNA for P9 encoded a 56-residue protein

with a 17-residue signal peptide (AB201778) The

cDNA for P15 encoded a 47-residue protein with a

29-residue signal peptide (AB201779) P9 shows

signi-ficant sequence similarity (52% identity) to the

Kunitz-type protease inhibitor from T tridentatus

hemocytes [49] Based on sequence homology, the

reactive site of P9 can be predicted to be at the

Tyr18–Ala19 bond, suggesting that it is a Kunitz-type

inhibitor of chymotrypsin-like activity (Fig 5) P15

contains eight Cys residues in positions similar to

those observed in an insulin-like growth factor

bind-ing motif (IGFBP motif) [50] (Fig 6) In mammals,

insulin-like growth factor-binding proteins, which

con-tain the IGFBP motif, modulate the actions of the

insulin-like growth factors in endocrine, paracrine,

and autocrine systems [51] Insulin-like growth factors

are essential for growth and development [52], and

the presence of the IGFBP motif in P15 raises the

possibility that it might play an analogous role in

the exoskeleton

Tissue-specific expression of HMM and LMM

chitin-binding proteins

Basic G13 and G19, acidic DE25 and DE29 and S37

were shown by RT-PCR to be expressed predominantly

in epidermis (Fig 7) In contrast, basic Y6, Y7 and

QH4 were broadly expressed, and basic QH10 was

highly expressed predominantly in muscle, heart and exoskeleton Tachyplesin and big defensin were highly expressed in all tissues, and P9 and P15 were expressed

in all tissues except for the intestine In plants, many protease inhibitors perform a protective function against insect infestation through the inhibition of insect pro-teases [53] In a similar way the Kunitz-type chymotryp-sin inhibitor bikunin is expressed on the keratinocyte cell membrane in human skin, and has been suggested

to play a regulatory role [54] The presence of a Kunitz-type chymotrypsin inhibitor sequence in P9 suggests that it may regulate endogenous proteases within the exoskeleton or inactivate those of invading pathogens

TGase-dependent cross-linking of HMM chitin-binding proteins

TGases catalyze the formation of isopeptide bonds between Gln and Lys residues and play an important role during the final stage of blood coagulation in mammals and crustaceans [55,56] In T tridentatus, TGase promotes the cross-linking of coagulin with hemocyte surface antigens called proxins and may faci-litate the formation of physiological barrier to invading pathogens [57] In mammals, TGase-catalyzed forma-tion of e-(c-glutamyl)-lysine bonds is involved in the formation of the cornified cell envelope of the skin, which serves as a frontline barrier against invading pathogens [58,59] Horseshoe crab TGase was expressed predominantly in hemocytes, and expression in epider-mis was not significant (Fig 7) Horseshoe crab TGase

Fig 5 Alignment of LMM-P9 and horseshoe crab kunitz-type

tryp-sin inhibitor (Trp inh) [49] Conserved cysteine residues are

designa-ted with black boxes Numbers on the right indicate amino acid

residue numbers The arrow indicates the predicted reactive site.

Fig 6 Comparison of LMM-P15 and the N-terminal regions of

IGFBP family members Amino acid sequences of LMM-P15,

mac25, IGFBP-1, -3, -4, -5, and -7 were aligned [50] Conserved

cysteine residues are designated with black boxes The

characteris-tic IGFBP motif (GCGCCXXC) is boxed by a solid line.

Fig 7 Expression patterns of cuticular chitin-binding proteins and TGase Relative mRNA levels were investigated by RT-PCR as des-cribed in ‘Experimental procedures’ Lane 1, hemocytes; lane 2, heart; lane 3, stomach; lane 4, intestine; lane 5, hepatopancreas; lane 6, epidermis; lane 7, skeletal muscle.

is released from hemocytes into the extracellular fluid in

response to external stimuli, such as bacterial

lipopoly-saccharides [57] Recently, an epidermal barrier wound

repair pathway has been shown to be evolutionally

con-served between Drosophila and mice In Drosophila, the

transcription factor grainy head regulates production of

the enzymes dopa decarboxylase and tyrosine

hydroxy-lase, which are required for covalent cross-linking of

cuticular structural components [60] Mice lacking a

homologue of Drosophila grainy head display defective

skin barrier function and deficient wound repair,

accompanied by reduced expression of TGase 1, the

key enzyme involved in protein cross-linking in

main-tenance of the stratum corneum [61]

To determine whether potential TGase substrates

are present in the arthropod exoskeleton, the cuticular

chitin-binding proteins of T tridentatus were incubated

with TGase, and subjected to SDS⁄ PAGE TGase

induced the formation of SDS-insoluble aggregates of

HMM chitin-binding proteins, which were incapable

of migrating into the gel Upon TGase treatment, the

major HMM chitin-binding proteins (16, 20 and

25 kDa) were no longer visible on SDS⁄ PAGE,

indica-ting that these proteins were cross-linked to form

higher molecular weight polymers (Fig 8, lane 2)

Monodansylcadaverine (DCA), a synthetic fluorescent substrate for TGase, competitively inhibited TGase-dependent polymerization (Fig 8, lane 3), and was incorporated into the major HMM chitin-binding proteins (Fig 8, lane 5) When analyzed by 2D SDS⁄ PAGE, DCA was incorporated into nearly all groups of HMM chitin-binding proteins including basic G, basic Y, basic QH, and acidic DE (Fig 9), and the identity of most DCA-labeled proteins were confirmed by amino acid sequence analysis (numbered spots in figure) These finding indicate that cuticular proteins in arthropods are capable of acting as sub-strates for TGase and may be involved in a TGase-dependent cross-linking system analogous to that observed in the epidermal cornified cell envelope in mammals

Experimental procedures

Protein extraction

Cuticles were obtained from the horseshoe crab T tridenta-tus, which had died of natural causes while in captivity, and stored at )80
C until use A part of cuticle from the ventral side of a single specimen, called the doublure, was excised, and epidermal cells were stripped from the cuticle with a sterilized spatula and used subsequently for the pre-paration of mRNA The remaining cuticle fragments were washed with distilled water, cut into small pieces with steril-ized scissors and homogensteril-ized in ice-cold homogenization buffer (50 mm Tris⁄ HCl, pH 7.5, 0.1 m NaCl) using a Poly-tron (Central Scientific Commerce Inc., Tokyo, Japan) at

15 000 r.p.m for 1 min The insoluble material was recov-ered by centrifugation at 3200 g for 30 min at 4
C, and subjected to a second round of homogenization and clarifi-cation The resulting precipitate was mixed with 10% acetic acid or 8 m urea for 16 h at 4
C with gentle agitation, and centrifuged at 4500 g) for 20 min at 4
C The resulting supernatant constituted the cuticular extract

Purification of chitin-binding proteins

The lyophilized extract was dissolved in a buffer consisting

of 50 mm Tris⁄ HCl, pH 7.5, 0.1 m NaCl, and applied to a chitin (Seikagaku Corp., Tokyo, Japan) affinity column (2.7· 17.5 cm) equilibrated with the same buffer After washing with the equilibration buffer, chitin-binding proteins were eluted with 10% (v⁄ v) acetic acid or 8 m urea For isola-tion of low molecular mass chitin-binding proteins, the cuticular proteins fractionated by chitin-affinity chromatog-raphy, and the resulting eluate was lyophilized, dissolved in 10% (v⁄ v) acetic acid, and applied to a Sephadex G-50 (Pharmacia Fine Chemicals, Uppsala, Sweden) column (2.7· 105 cm) equilibrated with 10% (v ⁄ v) acetic acid

Fig 8 TGase-dependent protein cross-linking of cuticular

chitin-binding proteins Lane 1, nontreated cuticular proteins; lane 2,

cuticular proteins + TGase; lane 3, cuticular proteins + TGase +

DCA; lane 4, cuticular proteins + TGase + EDTA; lane 5, otherwise

identical to lane 3 but illuminated by UV light.

Eluted fractions were analyzed by 15% SDS⁄ PAGE, and the

fractions containing proteins with molecular masses of less

than 10 kDa were lyophilized, and subjected to reverse-phase

HPLC

2D SDS/PAGE

The chitin-binding proteins purified by chitin-affinity

chro-matography were reduced, S-alkylated with iodoacetamide,

and an aliquot was precipitated with trichloroacetic acid

The precipitates were dissolved in 350 lL of 2% IPG buffer

(pH 3–10) (Amersham Pharmacia Biotech, Uppsala,

Swe-den) containing 8 m urea, 2% Chaps, 65 mm dithiothreitol,

and a trace of bromophenol blue, and then applied to the

IPG strip (18 cm, pH 3–10NL) The strip was covered with

silicone oil and rehydrated overnight The proteins were

focused at 20
C, according to the following voltage

gradi-ent program: 500 V, 2 h; 700 V, 1 h; 1000 V, 1 h; 1500 V,

1 h; 2000 V, 1 h; 2500 V, 1 h; 3000 V, 1 h; 3500 V, 10 h,

using a Multi Drive XL electrophoresis power supply The

strip was then equilibrated for 15 min in 50 mm Tris⁄ HCl,

pH 6.8, 6 m urea, 30% (v⁄ v) glycerol, 2% (w ⁄ v) SDS, and

10 mgÆmL)1 dithiothreitol, then treated with 25 mgÆmL)1

iodoacetamide for 15 min to alkylate free cysteine residues

Resolution of proteins in the second dimension was

per-formed by SDS⁄ PAGE (8–18%), according to the

manu-facturer’s instructions

Proteolytic digestion and reverse-phase HPLC

High molecular mass (HMM) chitin-binding proteins

separ-ated by 2D SDS⁄ PAGE were transferred to polyvinylidene

difluoride membranes overnight at 20 V using an

electro-blotting apparatus (Bio-Rad Laboratories, Hercules, CA,

USA) The membrane was stained with Coomassie Brilliant

Blue R-250, and major spots were excised Proteins were digested on the membrane with TPCK-trypsin (Worthing-ton Biochemical Corporation, Freehold, NJ, USA) in a buffer consisting of 100 mm NH4HCO3, pH 7.8, 10 mm CaCl2 and 10% acetonitrile at 25
C for 16 h Peptides in digested samples and LMM chitin-binding proteins were resolved by reverse-phase HPLC, using a Cosmosil 5C18

-MS column (2.0· 150 mm, Nacalai Tesque Inc., Kyoto) Peptides were eluted from the column with a linear gradient

of 0–72% acetonitrile in 0.1% trifluoroacetic acid for

120 min at a flow rate of 0.2 mLÆmin)1with effluent monit-oring at 210 nm

Amino acid composition and sequence analyses

Amino acid analysis was analyzed using the AccQ-Tag system (Waters Corp., Milford, MA, USA) Amino acid sequence analysis was performed using an Applied Biosys-tems 491 protein sequencer

Purification of mRNA and cDNA synthesis

Purification of mRNA derived from cuticular epidermal cells was performed using a QuickPrep mRNA Purification Kit (Amersham Pharmacia Biotech) The synthesis of double-stranded cDNA was performed using a SuperScriptTM III RNase H–reverse transcriptase kit (Invitrogen Corp., Carls-bad, CA, USA), according to the manufacturer’s instruc-tions

Amplification of cDNA fragments of the chitin-binding proteins

The sequences of degenerate oligonucleotide primers used for RT-PCR were based on amino acid sequences identified

Fig 9 TGase-dependent incorporation of DCA into chitin-binding proteins on 2D-SDS gel DCA incorporation was examined in the presence

of 10 m M CaCl 2 and 10 m M dithiothreitol Samples were subjected to 2D SDS ⁄ PAGE after incubation at 37
C for 1 h, and illuminated by UV light.