Báo cáo khoa học: Identification of an osteopontin-like protein in fish associated with mineral formation pot

The recently developed Sparus aurata gilthead seabream osteoblast-like cell line VSa16 was used to construct a cDNA subtractive library aimed at the identification of genes associated wit

associated with mineral formation

2Vera G Fonseca*, Vincent Laize´*, Marta S Valente and M Leonor Cancela

Centro de Cieˆncias do Mar (CCMAR), Universidade do Algarve, Faro, Portugal

Fish, by sharing with mammals a large number of

impor-tant characteristics (e.g organ systems, developmental

and physiological mechanisms), has become a suitable

model organism to study vertebrate physiological

pro-cesses, particularly skeletal development and tissue

min-eralization [1–3] While intensively studied in mammals

for decades, mechanisms of bone formation and

skeleto-genesis have been under scrutiny in fish only in the past

few years Consequently, genetic resources, tools and

methods that may be used towards the study of tissue

mineralization are limited in fish In an effort to develop

that aspect, numerous studies have been carried out

recently in a number of fish species, including several freshwater fish (e.g zebrafish, goldfish, Nile tilapia and common carp) and one marine fish (gilthead seabream) The latter is among the most important marine species grown in European farms, and because hatchery-reared seabream larvae develop high levels of skeletal malfor-mations [4–6], it has become the focus of recent studies related to skeletogenesis As a result,

involved in seabream ossification have been cloned (e.g osteocalcin, matrix Gla protein, osteonectin, bone morphogenetic protein 2, alkaline phosphatase) [7–11] (V Laize´ & M Leonor Cancela, unpublished results)

Keywords

bone-derived cell line; gilthead seabream

Sparus aurata (Teleostei); osteopontin;

subtractive library; tissue mineralization

Correspondence

M Leonor Cancela, Centro de Cieˆncias do

Mar (CCMAR), Universidade do Algarve,

Campus de Gambelas, 8005-139 Faro,

Portugal

Fax: +351 289800069

Tel: +351 289800971

E-mail: lcancela@ualg.pt

Website: http://www.ualg.pt/fcma/edge/

web/

*These authors contributed equally to this

work

(Received 11 April 2007, revised 21 June

2007, accepted 2 July 2007)

doi:10.1111/j.1742-4658.2007.05972.x

Fish has been recently recognized as a suitable vertebrate model and repre-sents a promising alternative to mammals for studying mechanisms of tis-sue mineralization and unravelling specific questions related to vertebrate bone formation The recently developed Sparus aurata (gilthead seabream) osteoblast-like cell line VSa16 was used to construct a cDNA subtractive library aimed at the identification of genes associated with fish tissue min-eralization Suppression subtractive hybridization, combined with mirror orientation selection, identified 194 cDNA clones representing 20 different genes up-regulated during the mineralization of the VSa16 extracellular matrix One of these genes accounted for 69% of the total number of clones obtained and was later identified as the

gene The 2138-bp full-length S aurata osteopontin-like cDNA was shown

to encode a 374 amino-acid protein containing domains and motifs charac-teristic of osteopontins, such as an integrin receptor-binding RGD motif, a negatively charged domain and numerous post-translational modifications (e.g phosphorylations and glycosylations) The common origin of mamma-lian osteopontin and fish osteopontin-like proteins was indicated through

an in silico analysis of available sequences showing similar gene and protein structures and was further demonstrated by their specific expression in min-eralized tissues and cell cultures Accordingly, and given its proven associa-tion with mineral formaassocia-tion and its characteristic protein domains, we propose that the fish osteopontin-like protein may play a role in hard tissue mineralization, in a manner similar to osteopontin in higher vertebrates

Abbreviations

1 Asp, aspartic acid; ECM, extracellular matrix; EST, expressed sequence tag; Gly, glycine; Glu, glutamic acid; MOS, mirror orientation selection; OP-L, osteopontin-like; SaOP-L, Sparus aurata osteopontin-like; Ser, serine; SSH, suppression subtractive hybridization.

(b) cell lines representing different bone-related cell types

have been obtained [12], (c) expressed sequence tag

(EST)

6 collections have been developed [13,14], (d) DNA

microarrays have been built [13,14] and (e) a radiation

hybrid panel has been developed for seabream [15]

The present study aimed to identify fish genes

involved in the mineralization of the extracellular matrix

of a seabream osteoblast-like cell line [12] using a

subtractive cloning approach A cDNA subtractive

library was first constructed using the suppression

sub-tractive hybridization technique (SSH) [16], then

impro-ved using the mirror orientation selection (MOS) in

order to eliminate false positives [17] This approach

allowed the identification of 194 cDNA clones

represent-ing 20 different up-regulated osteoblast-related genes

Results

Up-regulated genes during mineralization of

VSa16 osteoblast-like cells

VSa16 cells were cultured for 3 weeks under control

or mineralizing conditions then stained using the von

Kossa method to demonstrate extracellular matrix

(ECM)

7 mineralization of treated cells (results not

shown) Two cDNA libraries were constructed using

RNA extracted from control or treated cells, then

sub-tracted (mineralization minus control), enriched and

normalized according to the MOS method A total of

1600 bacterial clones containing fragments of up-regu-lated cDNAs inserted into pGEM-T Easy were screened

in situ From these, 194 were confirmed to be differen-tially expressed

each clone were sequenced and identified by similarity search using blast facilities at the National Center for Biotechnology Information Sequence analysis identi-fied 20 different cDNAs (Table 1), encoding different classes of proteins involved in a wide range of cell mechanisms, including regulation of ECM mineraliza-tion (n¼ 1), cellular metabolism (n ¼ 8) and cell orga-nization and biogenesis (n ¼ 2) The remaining genes (n¼ 9) were found to encode proteins with unknown function Up-regulated expression of these 20 genes was confirmed by reverse northern analysis (results not shown) Interestingly, the most up-regulated gene was also the most represented (69% of all occurrences, rep-resented by three different fragments later shown to be part of the same cDNA) This gene (i.e the fragments obtained from SSH) exhibited the highest similarity with fish osteopontin-like (OP-L)

rainbow and brook trouts) and, to a lesser extent, with mammalian osteopontin genes, and was consequently termed Sparus aurata osteopontin-like (SaOP-L) gene

Cloning and reconstruction of OP-L sequences Specific PCR primers (Table 2) were designed according

to the three nonoverlapping cDNA fragments obtained

Table 1 Genes up-regulated during the mineralization of VSa16 cells.

29

BLASTX

Development

Cellular process

Cellular metabolism

Cell organization and biogenesis

Unknown

a According to the definition of the Gene Ontology database at http://www.geneontology.org b Three different fragments corresponding to different regions of osteopontin-like cDNA were identified (occurrence ¼ 2, 8 and 123 clones) c

Two different fragments corresponding to different regions of short-chain dehydrogenase cDNA were identified (occurrence ¼ 20 and 11 clones).

from SSH (a, b and c in Fig 1A) and used in a

combi-nation of RACE and standard PCR amplifications to

amplify overlapping fragments (d, e and f in Fig 1A)

A 2138-bp sequence corresponding to the full-length

cDNA of SaOP protein (GenBank accession number

AY651247) was finally reconstructed (Fig 1A,B) An

ATG initiation codon was found at position 130 with

an in-frame stop codon at position 1254, generating a

1125-bp open reading frame that encoded a 374

amino-acid peptide Analysis of the primary sequence of the

protein demonstrated various domains, motifs and

post-translational modifications, including (a) a 16

amino-acid transmembrane signal peptide at the N-terminus

for protein secretion, (b) an integrin receptor-binding

RGD motif [arginine (Arg) 178–glycine(Gly)179–

aspartic acid(Asp)180], suggesting a role of SaOP-L in

cell adhesion, (c) a negatively charged domain rich in

Asp and glutamic acid (Glu) residues (Asp109–Glu136)

and (d) 64 putative serine (Ser) and threonine

phos-phorylated residues located in the target sequence of

mammary gland casein kinase [S⁄ T-X-E ⁄ S(P) ⁄ D] and

casein kinase II (S-X-X-E⁄ S(P) ⁄ D), two enzymes

responsible for most phosphorylations in human

osteo-pontin [18] Searching online public databases (e.g

GenBank at http://www.ncbi.nlm.nih.gov and Ensembl

at http://www.ensembl.org) using blast revealed

numerous ESTs or genomic clones with high similarity

to OP-L proteins The analysis, clustering and

assem-bly of these sequences permitted the reconstruction of

three new OP-L sequences (two cDNAs and one gene;

see supplementary Fig S1), all of fish origin

A total of seven complete OP-L sequences (three

previously annotated, three reconstructed and one

cloned) have been collected for this study (Fig 2) Interestingly, searching GenBank and Ensembl sequence databases using OP-L sequences identified only bony fish sequences (Osteichthyes) and none from mammals, birds or amphibians Similarly, searching sequence databases using annotated osteopontin sequences identified only sequences from mammals, birds and amphibians, and none from fish The pair-wise per cent identities among mammalian osteopontin and fish OP-L protein sequences were  60% and 40%, respectively, whereas the identity between fish and mammalian sequences was only 14% (Table 3), further confirming the weak similarity existing between the two proteins at the amino acid level

Comparison of osteopontin and OP-L sequences Despite their weak sequence similarity, we hypothe-sized that fish OP-L protein could be orthologous to mammalian osteopontin To test this hypothesis and

to determine whether osteopontin and OP-L protein have retained the same function in the course of evolu-tion, the gene and peptide structures of both proteins were investigated Annotated osteopontin sequences were collected from GenBank and Ensembl sequence databases (seven sequences from mammals and two from birds; Fig 2) and compared with OP-L sequences

similar, exhibiting the same pattern of exon distribu-tion (four to five small exons at the 5¢ end and two lar-ger

12 exons at the 3¢ end; note that fish exons 3 and 4 have probably merged to generate exon 3 in birds and mammals) and an identical pattern of intron insertion: all occurred between two different codons (phase 0) Analysis of the protein primary structure (Fig 3B and Table 4) identified several conserved features, including (a) similar size, molecular weight, isoelectric point and hydropathicity, (b)

repre-sented residues (Asp, Glu and Ser), (c) the presence of numerous, possibly phosphorylated, O- and N-glycosy-lated residues throughout the protein (d) and similar domains (N-terminal negatively charged domain), motifs (integrin receptor-binding RGD motif) and pro-teolytic cleavage sites (thrombin) Altogether, these observations point

of osteopontin and OP-L genes and proteins

Table 2 PCR primers used to clone Sparus aurata osteopontin-like

(SaOP) full-length cDNA and analyze its gene expression.

Primer Sequence (5¢ ) to 3¢)

Fig 1 SaOP-L full-length cDNA (A) cDNA fragments obtained from the subtractive library (a, b and c) and amplified by PCR (d, e and f) (B) SaOP-L reconstructed cDNA and deduced amino acid sequences (bold) The light grey box indicates the N-terminal signal peptide, the dark grey box indicates the negatively charged region and the black box indicates the RGD motif Putative phosphorylation located in the target sequence of the mammary gland casein kinase and casein kinase II are indicated by

cDNA sequence can be retrieved from GenBank using accession number AY651247.

B

Expression patterns of the SaOP gene

Expression of the SaOP-L gene was strongly induced

in osteoblast-like VSa16 and chondrocyte-like VSa13

cells after 4 weeks of mineralization (Fig 4),

suggest-ing a role of OP-L protein

mineralization In addition, this observation indicated

that OP-L gene expression was not limited to

osteo-blasts but was associated, in both cases, with the

min-eralization process Expression of the SaOP-L gene

was further investigated in the course of VSa16 ECM

mineralization (data not shown) and shown to be

rapidly and strongly up-regulated during this period

while severely repressed in cells cultured under normal

conditions, thus providing additional evidence for a

role of OP-L protein

miner-alization

Expression of the SaOP-L gene was then

investi-gated during S aurata development using RNA

pre-pared from embryos, larvae and juvenile fish Gene

expression was detected early in development, with a

net increase observed

concomitant with the progressive increase known to

occur, both in number and size, of calcified skeletal

structures throughout fish development

Finally, distribution of the SaOP-L transcript was

investigated in a number of adult tissues, including

cal-cified, mixed (partially calcified) and noncalcified

tis-sues The SaOP-L gene was expressed in all calcified and partially calcified tissues, with the highest levels detected in teeth, bone-dentary and branchial arches (Fig 6), while absent or barely detectable in soft tis-sues (i.e brain, skeletal muscle, heart, aorta, adipose tissue, intestine, kidney, ovary, testis, pancreas, spleen, stomach, liver, gills, urinary bladder, gall bladder, swim bladder; results not shown) This result demon-strated the specific expression of the SaOP-L gene

in mineralized tissues, and further confirmed data obtained in vitro with S aurata bone-derived cell lines

Discussion This work identified, through a subtractive cloning approach, 20 different transcripts up-regulated in min-eralized cultures of seabream bone-derived cells Even though almost all genes obtained were new with respect to the seabream gene pool, some have already been given specific functions in other vertebrates, par-ticularly in mammals However, genes usually associ-ated with osteoblast function (e.g tissue nonspecific alkaline phosphatase, type I collagen, osteonectin, osteocalcin, etc.) have not been identified through this subtractive approach The reasons why these genes were not uncovered during our study could be that (a) the screening of 1600 bacterial clones was insufficient (therefore more clones should be screened) or (b) the

Ostariophysi

Tetraodontiformes Acanthopterygii

Protacanthopterygii Osteichthyes

Perciformes

Mammalia

Rodentia

Lagomorpha

Bovinae

Caprinae Bovidae

Suidae Actiodactyla

Primates

Aves

Vertebrates

Gasterosteiformes

Scientific name (common name) Acronym Accession

Rattus norvegicus (Norway rat) RnOP AAA41765

Mus musculus (house mouse) MmOP AAM53974

Ovis aries (sheep) OaOP AAD38388

Bos taurus (domestic cattle) BtOP AAX62809

Sus scrofa (pig) SsOP CAA34594

Homo sapiens (human) HsOP AAA86886

Oryctolagus cuniculus (European rabbit) OcOP BAA03980

Gallus gallus (chicken) GgOP AAA18584

Coturnix japonica (Japanese quail) CjOP AAF63330

Danio rerio (zebrafish) DrOP-L AAT39545

Pimephales promelas (fathead minnow) PpOP-L Reconstructed a

Oncorhynchus mykiss (rainbow trout) OmOP-L AAG35656

Salvelinus fontinalis (brook trout) SfOP-L AAG49534

Tetraodon nigroviridis (green pufferfish) TnOP-L Reconstructed a

Sparus aurata (gilthead seabream) SaOP-L AAV65951

Gasterosteus aculeatus (three spined stickleback) GaOP-L Reconstructed a

Fig 2 Osteopontin and osteopontin-like sequences used in this study and taxonomy of represented species Taxonomic data were retrieved December 20, 2005 from the Integrated Taxonomic Information System at http://www.itis.usda.gov a, see the supplementary Fig S1.

MOS technique, used to reduce the number of

back-ground clones, might have decreased cDNA species

diversity in the subtracted library, as already seen in

other studies (Ricardo

of Algarve, Portugal, personal communication)

Alter-natively, some of these genes may be represented by

cDNA fragments whose identity was not unraveled

through sequence comparison From the analysis of

clone abundance in the SSH library, the SaOP-L gene

was clearly the most highly expressed and was the

focus of this work

Fish OP-L protein is probably orthologous to

mammalian osteopontin

The most abundant and up-regulated gene obtained

from the subtractive approach was termed OP-L, in

agreement to its similarity with annotated trout [19]

sequences and the proposed affiliation of these

sequences with mammalian osteopontin The in silico

analysis of available sequences (annotated, cloned

and reconstructed) clearly demonstrated the overall

conservation of both gene (i.e similar pattern for

exon size and identical phase of intron insertion) and

protein (i.e an acidic Asp-rich domain, an RGD motif, a thrombin cleavage site and numerous puta-tive phosphorylated residues) structures between fish OP-L and mammalian osteopontin proteins; we there-fore concluded that the fish protein is probably the ortholog

19 of mammalian osteopontin The weak simi-larity observed at the amino acid level indicates

both proteins have diverged significantly during evo-lution and might have developed distinct functions

By using similar evidence

compari-son) but a more restricted set of sequences, Kawasaki and colleagues have drawn a similar conclusion concerning zebrafish NOP⁄ OP-L and mammalian SPP1⁄ OP proteins [20] and have proposed that OP-L protein

22 and osteopontin may have a similar cellular role (i.e as a modulator of hydroxyapatite crystalliza-tion) but a distinct function because of the differences observed in their amino acid content in acidic clus-ters The gene expression pattern further supports the idea that OP-L protein

orthologs: they are both strongly expressed in calcified tissues (bone and calcified cartilage) and up-regulated during the mineralization process [21,22] These data, combined with the absence of data describing

Table 3 Pairwise per cent identities among osteopontin and osteopontin-like protein sequences Light grey, mammals; dark grey, fish; white, birds Bt, Bos taurus (domestic cattle); Cj, Coturnix japonica (Japanese quail); Dr, Danio rerio (zebrafish); Ga, Gasterosteus aculeatus (three spined stickleback); Gg, Gallus gallus (chicken); Hs, Homo sapiens (human); Mm, Mus musculus (house mouse); Oa, Ovis aries (sheep); Oc, Oryctolagus cuniculus (European rabbit); Om, Onchorynchus mykiss (rainbow trout); Pp, Pimephales promelas (fathead min-now); Rn, Rattus norvegicus (Norway rat); Sa, Sparus aurata (gilthead seabream); Sf, Salvenilus fontinalis (brook trout); Ss, Sus scrofa (pig);

Tn, Tetraodon nigroviridis (spotted green pufferfish) Diagonal values in black boxes represent the sequence length.

23 ± 01

91 ± 00

14 ± 02

10 ± 01

40 ± 16

Mammals Birds Fish

osteopontin in fish or an OP-L protein in mammals, favor the assumption that OP-L protein is indeed the fish equivalent of mammalian osteopontin

OP-L protein plays a role in the process

of mineralization Osteopontin is a multifaceted protein [23–25], which has been associated in mammals with multiple phy-siological and pathological processes, in particular mineralization [26–28], and is ubiquitously expressed

in adult mammalian organisms [29–32] In this study, OP-L gene expression was detected in calcified tis-sues or in tistis-sues showing some degree of mineral accumulation, but not in soft tissues Previous find-ings in adult brook trout [19] have shown some expression of an

in testis and ovulatory ovary, and to a lesser extent

in kidney, gills and skin), but this has not been 25

investigated in calcified tissues Differences in tissue distribution of OP-L gene expression observed when comparing trout and seabream results (a comparison limited to soft tissues) may be explained by the recent genome duplication event that specifically affected Salmonids [9,33] and probably not Sparids

If the brook trout genome has two copies of the OP-L gene, as seen for other mineralization-related genes (e.g osteonectin), it would be expected that the two isoforms show different patterns of tissue distribution and⁄ or regulation, a common feature associated with gene duplication However, we can-not rule out the fact that differences in gene expres-sion in soft tissues could be related to different

Table 4 Selected features of osteopontin (mammals) and

osteo-pontin-like (fish) proteins CKII, casein kinase II; MGCK, mammary

gland casein kinase Asn, asparginine; Asp, aspartic acid

glutamic acid; Thr, threonine.

Features

Mammals (n ¼ 7)

Fish (n ¼ 7)

MGCK ⁄ CKII phosphorylation sites 40 ± 2 53 ± 7

O-glycosylated residues (Thr)d 7 ± 3 23 ± 6

a

Predicted using PROTPARAM at http://www.expasy.org. bGRAVY

(grand average of hydropathicity). cPredicted using NETNGLYC at

http://www.cbs.dtu.dk d Predicted using NETOGLYC at http://www.

cbs.dtu.dk e Predicted using PEPTIDECUTTER at http://www.expasy.

org.fExcept in the predicted sequence of Gasterosteus aculeatus

osteopontin-like (GaOP-L)

Control

SaOP-L

SaRPL27a

Mineral Control Mineral.

Control Mineralization

1 0

2 3 4 5

Fig 4 Relative SaOP-L gene expression in VSa16 and VSa13 cells cultured under control or mineralizing conditions Top panel, SaOP-L and SaRPL27a signals after autoradiography; bottom panel, SaOP-L relative gene expression normalized with RPL27a ND, not detected.

Human

Chicken

Bovine

Mouse

Zebrafish

A

B

SP

RGD

E-, D-rich

MP

SP MP D-rich

OP-like

Fish

OP

Mammals

Thrombin cleavage

N

O

Negatively-charged cluster

O N

N

0 0

0 0

0 0

0 0 0

0 0

0 0 0

0 0

0 0 0

0

0 0

0 0

0 0

0 0

Fig 3 Osteopontin and osteopontin-like gene and protein

struc-ture (A) Structural organization of osteopontin-like (tetraodon and

zebrafish) and osteopontin (chicken, bovine, mouse and human)

coding sequences at the gene level Grey boxes indicate exons (or

part of exons) representing the coding sequence, starting from the

translation initiation codon and ending at the translation termination

codon The phase of intron insertion is indicated in

and is defined according to Patthy [50] (B) Structural organization

of osteopontin-like (fish) and osteopontin (mammals) proteins MP,

mature peptide; N and O, predicted N- and O-linked glycosylations,

respectively; SP, signal peptide.

developmental or physiological stages of the

speci-mens used in both studies, as evidenced by the

regu-lation of OP-L gene expression in brook trout ovary

during ovulation [19]

The comparison of fish OP-L protein with

mamma-lian osteopontin expression patterns indicates that both

proteins may play a similar role in calcified tissues and

gonads, for example in bone remodeling by mediating

osteoclast attachment to the mineralized bone matrix

during resorption [23,34–36] and⁄ or in matrix

minera-lization by regulating calcium phosphate crystal

deposition [26,37–39], and in gonads by preventing calcium-containing-crystal aggregation

restricted tissue distribution of the OP-L gene tran-script also indicates that fish protein may be less pleio-tropic than that from mammals Finally, the massive up-regulation of OP-L gene expression during in vitro mineralization of VSa16 (osteoblast-like) and VSa13 (chondrocyte-like) ECM is highly suggestive of a role

of the OP-L protein in mineralization, which is likely

to be relevant based on the highly significant induction observed, and further emphasizes the importance of fish as a model to understand osteopontin function The pattern of developmental expression found for osteopontin is consistent with its involvement in the early mechanisms of ossification, which start in

S aurata at early larval stages and are continuous until 70–90 days after hatching [42–44] In addition, the later OP-L gene expression detected in fish

27130 days after hatching could also be related to ongo-ing bone remodelongo-ing, which occurs at a later stage during skeletal development⁄ growth

In conclusion, our results indicate that OP-L protein

is probably the fish ortholog to mammalian osteopon-tin, and is likely to play a role in the mineralization process under physiological conditions

Experimental procedures Materials

Tissue culture medium (DMEM), fetal bovine serum, anti-biotics (penicillin and streptomycin), antimycotics (fungi-zone), trypsin-EDTA and l-glutamine were purchased from Invitrogen (Carlsbad, CA, USA) Tissue culture plates were purchased from Sarstedt (Nu¨mbrecht, Germany) All other reagents were purchased from Sigma-Aldrich (St Louis,

MO, USA), unless otherwise stated

0 1 2 3 4 5 6 7 8 9 10

U/E 8 512 10 14 18 24 3 6 10 20 37 48 61 67 75 82 130 Cells Hours post

fertilization

Days post hatching

Maternal De novo transcription

Fig 5 Relative SaOP-L gene expression

during development Values are the mean of

three independent real-time PCR

experi-ments SaOP-L relative gene expression

was normalized with RPL27a U ⁄ E,

unfertil-ized eggs.

0

1

2

3

4

5

6

7

8

9

10

Fig 6 Relative SaOP-L gene expression in adult tissues Values

are the mean of three independent real-time PCR experiments.

SaOP-L relative gene expression was normalized with RPL27a BA,

branchial arch; Bd, bone-dentary; Bo, bone-opercula; Bs, bone-skull;

Bv, bone-vertebra; CA, cartilage; E, eye; Fso, soft rays; Fsp,

fin-spiny rays; SC, scale; SK, skin; TE, teeth; TO, tongue.

Cell culture and extracellular matrix

mineralization

S aurata VSa16 and VSa13 bone-derived cells were

cul-tured and maintained as described by Pombinho and

col-leagues [12] Briefly, cells were routinely grown in DMEM

supplemented with 10% fetal bovine serum, 1%

penicil-lin⁄ streptomycin, 1% fungizone and 2 mm l-glutamine, and

incubated at 33C in a 10% CO2 humidified atmosphere

ECM mineralization was induced by supplementing the

cul-ture medium with 50 lgÆmL)1 of l-ascorbic acid, 10 mm

b-glycerophosphate and 4 mm CaCl2 Mineral deposition

was detected using the von Kossa staining method and

observed under an Axiovert 25 inverted microscope (Zeiss,

Go¨ttingen, Germany) equipped with phase contrast

Subtracted cDNA library construction

and cloning

Total RNA was isolated from cultured cells, as described

by Chomczynski & Sacchi [45], and poly(A+) RNA was

extracted using the Oligotex Mini kit (Qiagen, Hilden,

Ger-many) SSH was carried out using 2 lg of poly(A+) RNA

and the PCR-Select cDNA Subtraction kit (Clontech, Palo

Alto, CA, USA) following the manufacturer’s protocol

Subtraction was obtained using cDNAs prepared from

min-eralized cells as the tester sample and cDNAs prepared

from control cells as the driver sample To normalize and

eliminate false-positive cDNA clones, SSH was combined

with the MOS technique, as described by Rebrikov and

colleagues [17] Secondary PCR products obtained from the

forward subtracted SSH were inserted into the pGEM-T

Easy vector (Promega, Madison, WI, USA) and the

result-ing plasmids were transformed into DH5a competent cells

Positive bacterial clones were selected on Luria–Bertani

agar plates containing ampicillin (100 lgÆmL)1), X-Gal

(80 lgÆmL)1) and isopropyl thio-b-d-galactoside (IPTG)

(0.5 mm) then grown for 20 h at 37C in 96-well plates,

each well containing 100 lL of Luria–Bertani supplemented

with ampicillin

In situ differential screening

Adaptor-free cDNAs from forward and reverse

subtrac-tions were radiolabeled according to the Clontech protocol

PT1117-1 with [32P]dCTP[aP] (3000 CiÆmL)1; Amersham

Biosciences, Piscataway, NJ, USA) using the Rediprime II

kit (Amersham Biosciences) and purified from

unincorpo-rated radionucleotides using Microspin S-200 HR columns

(Amersham Biosciences) Bacterial clones were blotted onto

Hybond-XL nylon membranes (Amersham Biosciences), as

described by Fonseca and colleagues [46] Membranes were

hybridized overnight at 42C in ULTRAhyb solution

(Ambion, Austin, TX, USA) using probes prepared from

forward or reverse subtractions, and washed twice (5 min each wash) in low-stringency solution [2· NaCl ⁄ Cit, 0.1% SDS (1· NaCl ⁄ Cit is 0.15 m NaCl and 15 mm sodium citrate), pH 7.0] and 2· 15 min in high-stringency solution (0.1· NaCl ⁄ Cit, 0.1% SDS) at 55 C Membranes were then exposed to a Kodak XAR film (Amersham Biosciences)

DNA sequencing and identification

DNA from selected clones was sequenced (Macrogen, Seoul, South Korea) and compared with sequences in the GenBank database using blastx and tblastx facilities at the National Center for Biotechnology Information

Pike, Bethesda, MD, USA, http://www.ncbi.nlm.nih.gov)

Reverse northern blot analysis

DNA from selected clones was PCR amplified using NP1 and NP2R primers (Clontech) and blotted in quadruplicate onto Hybond-XL nylon membranes using the Multi-Print manual arrayer (V & P Scientific, San Diego, CA, USA) DNA was cross-linked to the membrane for 3–4 min under

UV and for 2 h at 80C Membranes were probed, as described above, with radiolabeled VSa16 poly(A+) RNA (from either control or mineralized samples) Signal inten-sity was estimated by densitometric methods using quan-tity one software (Bio-Rad, Hercules, CA, USA) The relative expression of each gene was normalized with

S aurata ribosomal protein L27a (SaRPL27a, GenBank accession number AY188520) signals

Northern blot analysis

Ten micrograms of total RNA was fractionated on a 1.2% formaldehyde-agarose gel and transferred onto a

Hybond-XL nylon membrane by capillary blotting using

10· NaCl ⁄ Cit Membranes were probed, as described above, using radiolabeled S aurata OP-L (GenBank acces-sion number AY651247) or SaRPL27a probes, and the sig-nal intensity was determined by densitometric methods using quantity one software (Bio-Rad) Relative OP-L gene expression was normalized with SaRPL27a signals

PCR, RACE-PCR and cDNA cloning

All PCRs were performed using a 1 : 50 dilution of the VSa16 library constructed from poly(A+) RNA (control and mineralized) using the Marathon cDNA Amplification kit (Clontech) Amplification of the 5¢- and 3¢-RACE-PCR products was performed using the Advantage cDNA poly-merase mix (Clontech) and AP1⁄ AP2 primers combined with specific primers designed according to S aurata OP-L cDNA fragments previously obtained (Table 2) PCR frag-ments were size-fractionated by agarose-gel electrophoresis,

purified and inserted into the pGEM-T Easy vector DNA

inserts were sequenced and identified as described above

Analysis of gene expression by quantitative

real-time PCR

Real-time PCR assays were performed using the iCycler

PCR system and software to quantify nucleic acids

(Bio-Rad) Total RNA (1 lg) was reverse-transcribed at 37C

for 1 h using the Moloney-murine leukemia virus

(M-MLV) reverse transcriptase (Invitrogen), RNase Out

(Invitrogen) and specific reverse primers SaOPreal-RV and

SaRPL27a-RV for OP-L and ribosomal protein L27a

cDNAs The reaction mixture, containing 1· iQ SYBR

Green I mix (Bio-Rad), 0.4 lm forward and reverse primers

and 100 ng of reverse-transcribed RNA, was subjected to

the following PCR conditions: 4 min at 95C, and 55

cycles of 30 s at 95C and 45 s at 68 C RPL27a relative

gene expression was used to normalize OP-L gene

expres-sion levels Fragments of 153 bp for OP-L cDNA and

160 bp for RPL27a cDNA were amplified using the primer

sets SaOPreal-FW⁄ SaOPreal-RV and SaRPL27a-FW⁄

SaRPL27a-RV, respectively

Protein sequence analysis

Signal peptide, and O- and N-linked glycosylation sites,

were predicted using signalp 3.0 [47], netnglyc 1.0 and

netoglyc 3.1 [48] facilities at http://www.cbs.dtu.dk

Protein domains were identified using InterProScan

facili-ties at http://www.ebi.ac.uk Percentage protein identity

was calculated using the Sequence Manipulation Suite [49]

available at http://www.bioinformatics.org

Acknowledgements

Authors thank Marta S Rafael from the CCMAR,

University of Algarve, Faro, Portugal, for her

techni-cal help in the course of gene identification The

authors are also grateful to Ricardo B Leite and

Dr Paulo J Gavaia for data on the MOS technique

and fish skeletogenesis, respectively VGF was partially

supported by CCMAR funding This work was

par-tially supported by grants POCTI⁄ BCI ⁄ 48748 ⁄ 2002

from the Portuguese Science and Technology

Founda-tion (FCT) and GOCE-CT-2004-505403 (Marine

Genomics Europe) from the European Commission

under the 6th Framework Program

References

1 Gavaia PJ, Sarasquete C & Cancela ML (2000)

Detection of mineralized structures in early stages of

development of marine Teleostei using a modified alcian

blue-alizarin red-double staining technique for bone and cartilage Biotech Histochem 75, 79–84

2 Nissen RM, Amsterdam A & Hopkins N (2006) A zebrafish screen for craniofacial mutants identifies wdr68

as a highly conserved gene required for endothelin-1 expression BMC Dev Biol 6, 28

3 Fisher S, Jagadeeswaran P & Halpern ME (2003) Radiographic analysis of zebrafish skeletal defects Dev Biol 264, 64–76

4 Andrades JA, Becerra J & Fernandez-Llebrez P (1996) Skeletal deformities in larval, juvenile and adult stages

of cultured gilthead sea bream (Sparus aurata L.) Aquaculture 141, 1–11

5 Carrillo J, Koumoundouros G, Divanach P & Martinez

J (2001) Morphological malformations of the lateral line

in reared gilthead sea bream (Sparus aurata L 1758) Aquaculture 192, 281–290

6 Koumoundouros G, Oran G, Divanach P, Stefanakis S

& Kentouri M (1997) The opercular complex deformity

in intensive gilthead sea bream (Sparus aurata L.) larvi-culture Moment of apparition and description Aqua-culture 156, 165–177

7 Pinto JP, Ohresser MCP & Cancela ML (2001) Cloning

of the bone Gla protein gene from the teleost fish Sparus aurata Evidence for overall conservation in gene organization and bone-specific expression from fish to man Gene 270, 77–91

8 Pinto JP, Conceic¸a˜o N, Gavaia PJ & Cancela ML (2003) Matrix Gla protein gene expression and protein accumulation colocalize with cartilage distribution during development of the teleost fish Sparus aurata Bone 32, 201–210

9 Laize´ V, Pombinho AR & Cancela ML (2005) Charac-terization of Sparus aurata osteonectin cDNA and

in silicoanalysis of protein conserved features: Evidence for more than one osteonectin in Salmonidae Biochimie

87, 411–420

10 Cancela ML, Williamson MK & Price PA (1995) Amino acid sequence of bone Gla protein from the african clawed toad Xenopus laevis and the fish Sparus aurata Int J Pept Protein Res 46, 419–423

11 Simes DC, Williamson MK, Ortiz-Delgado JB, Viegas CSB, Price PA & Cancela ML (2003) Purification of matrix Gla protein from a marine teleost fish, Argyroso-mus regius: Calcified cartilage and not bone as the pri-mary site of MGP accumulation in fish J Bone Miner Res 18, 244–259

12 Pombinho AR, Laize´ V, Molha DM, Marques SMP & Cancela ML (2004) Development of two bone-derived cell lines from the marine teleost Sparus aurata; evidence for extracellular matrix mineralization and cell-type-spe-cific expression of matrix Gla protein and osteocalcin Cell Tissue Res 315, 393–406

13 Sarropoulou E, Kotoulas G, Power DM & Geisler R (2005) Gene expression profiling of gilthead sea bream