Báo cáo khoa học: Hierarchical subfunctionalization of fabp1a, fabp1b and fabp10 tissue-specific expression may account for retention of these duplicated genes in the zebrafish (Danio rerio) genome docx

Identification of a cDNA sequence coding for a second zebrafish FABP1, FABP1b The cDNA sequence coded by a duplicated fabp1 gene in zebrafish was identified and determined using 3¢-RACE an

and fabp10 tissue-specific expression may account

for retention of these duplicated genes in the zebrafish (Danio rerio) genome

Mukesh K Sharma1, Rong-Zong Liu1, Christine Thisse2, Bernard Thisse2, Eileen M Denovan-Wright3and Jonathan M Wright1

1 Department of Biology, Dalhousie University, Halifax, NS, Canada

2 Institut de Ge´ne´tique et de Biologie Mole´culaire et Cellulaire, Department of Developmental Biology, Illkirch, France

3 Department of Pharmacology, Dalhousie University, Halifax, NS, Canada

The fatty acid-binding protein 1 (FABP1), commonly

termed liver-type fatty acid-binding protein (L-FABP),

is a low molecular mass (
14 kDa) polypeptide that

belongs to the multigene family of intracellular

lipid-binding proteins (iLBP) [1] At least 16 members of the iLBP multigene family have been described [2] Origin-ally, FABPs were named according to their initial site of isolation, e.g intestinal-type fatty acid-binding

Keywords

embryonic development; FABP1; gene

duplication; linkage mapping;

subfunctionalization

Correspondence

J M Wright, Department of Biology,

Dalhousie University, Halifax, Nova Scotia,

Canada, B3H 4J1

Fax: +1 902 494 3736

Tel: +1 902 494 6468

E-mail: jmwright@dal.ca

Database

The following sequences have been

submit-ted to the GenBank database under

acces-sion numbers DQ062095 (fabp1a cDNA),

DQ062096 (fabp1b cDNA, long transcript

variant) and DQ474062 (fabp1b cDNA, short

transcript variant).

(Received 30 January 2006, revised 10 April

2006, accepted 18 May 2006)

doi:10.1111/j.1742-4658.2006.05330.x

Fatty acid-binding protein type 1 (FABP1), commonly termed liver-type fatty acid-binding protein (L-FABP), is encoded by a single gene in mam-mals We cloned and sequenced cDNAs for two distinct FABP1s in zebra-fish coded by genes designated fabp1a and fabp1b The zebrazebra-fish proteins, FABP1a and FABP1b, show highest sequence identity and similarity to the human protein FABP1 Zebrafish fabp1a and fabp1b genes were assigned

to linkage groups 5 and 8, respectively Both linkage groups show con-served syntenies to a segment of mouse chromosome 6, rat chromosome 4 and human chromosome 2 harboring the FABP1 locus Phylogenetic analy-sis further suggests that zebrafish fabp1a and fabp1b genes are orthologs of mammalian FABP1 and most likely arose by a whole-genome duplication event in the ray-finned fish lineage, estimated to have occurred 200–450 million years ago The paralogous fabp10 gene encoding basic L-FABP, found to date in only nonmammalian vertebrates, was assigned to zebrafish linkage group 16 RT-PCR amplification of mRNA in adults, and in situ hybridization to whole-mount embryos to fabp1a, fabp1b and fapb10 mRNAs, revealed a distinct and differential pattern of expression for the fabp1a, fabp1b and fabp10 genes in zebrafish, suggesting a division of func-tion for these orthogolous and paralogous gene products following their duplication in the vertebrate genome The differential and complementary expression patterns of the zebrafish fabp1a, fapb1b and fabp10 genes imply

a hierarchical subfunctionalization that may account for the retention of both the duplicated fabp1a and fabp1b genes, and the fabp10 gene in the zebrafish genome

Abbreviations

EST, expressed sequence tag; FABP1, liver-type fatty acid-binding protein type 1; FABP10, basic liver-type fatty acid-binding protein; hpf, hours post fertilization; iLBP, intracellular lipid-binding proteins; mya, million years ago; 5¢-RLM-RACE, 5¢ RNA ligase-mediated-RACE; YSL, yolk syncytial layer.

protein (I-FABP), heart-type fatty acid-binding protein

(H-FABP), liver-type fatty acid-binding protein

(L-FABP), etc [3] Many studies have shown,

how-ever, that different types of FABP are present in the

same tissue [4] Moreover, some orthologous FABPs

from different species exhibit different tissue-specific

patterns of distribution Owing to the many names for

the same or orthologous proteins, Hetzel & Bernlohr

[5] have suggested an alternative nomenclature in

which each FABP is referred to only by its gene

desig-nation, presumably reflecting the chronological order

of their discovery, i.e FABP1 (liver-type), FABP2

(intestinal-type), FABP3 (heart⁄ muscle-type), etc

More recently, individual iLBPs have been classified

according to phylogenetic analyses based on the amino

acid sequences [6–8] We prefer the FABP

nomen-clature of Herzel & Bernlohr [5], and therefore use it

here However, where appropriate, we also include the

common name for FABPs Upper case letters are

used to designate the protein, e.g FABP1, and lower

case italic letters (e.g fabp1) and upper case italic

letters (FABP1) are used to designate a gene coding

for a particular FABP in zebrafish and mammals,

respectively

FABP1 is thought to be involved in the uptake of

fatty acids [5], the modulation of enzyme activity by

altering lipid levels [9], the sequestering of fatty acids

to protect cells against the harmful detergent effects of

excess free fatty acids [10], regulation of the expression

of specific genes, and the control of cell growth and

differentiation [11] In addition to long chain fatty

acids, FABP1 binds lysophospholipids, prostaglandins,

phytanic acid, eicosanoids, heme and acyl-CoAs

[4,9,12] Several studies have shown that FABP1 binds

two fatty acids per molecule [13–15], whereas other

iLBPs bind only a single ligand [15,16] Previously, we

reported the cDNA sequence for FABP10, the basic

liver-type fatty acid-binding protein, and the

distribu-tion of the fabp10 gene transcripts in adult zebrafish

[17] To date, the FABP10 protein, fabp10 gene and its

encoded mRNA have been found only in

nonmamma-lian vertebrates Like FABP1, FABP10 has the

capa-city to bind two ligand molecules, unlike all other

iLBPs that bind a single molecule [18], suggesting that

FABP1 and FABP10 are evolutionarily related,

possi-bly by a gene duplication event Phylogenetic analysis

of available iLBPs places FABP10 as a separate

branch of an iLBP subfamily that includes FABP1 and

FABP6, the ileal-type fatty acid-binding protein [19]

It is estimated that the fabp1, fabp6 and fabp10 genes

diverged from their last common ancestral gene
679

million years ago (mya) [2] We therefore anticipated

that the zebrafish genome may harbor a functional

fabp1 gene Moreover, as a whole-genome duplication event is thought to have occurred in the ray-finned fish lineage after their divergence from the lobe-finned fish lineage 200–450 mya [20–25], we further predicted that duplicate copies of the fabp1 gene may be present in the zebrafish genome

Here, we describe the cloning and sequence of cDNAs coded by the duplicated zebrafish genes for FABP1, hereafter referred to as cDNAs encoded by the fabp1a and fabp1b genes These duplicated fabp1 genes were assigned to different linkage groups using radiation hybrid mapping, and tissue-specific patterns

of distribution for the fabp1a and fabp1b mRNAs were determined during embryonic and larval development, and in adulthood

Results and Discussion

Sequence of a zebrafish FABP1a cDNA Searches of the National Centre for Biological Infor-mation (NCBI) DNA sequence database revealed an expressed sequence tag (EST; GenBank accession num-ber BI846703) coding for a protein described as being similar to human FABP1 3¢-RACE and 5¢ RNA ligase-mediated-RACE (5¢-RLM-RACE) using primers based on this EST were performed to obtain the cDNA sequence A single product of 759 bp, excluding the 20 bp adapter sequence and the poly(A) tail, was obtained in 3¢-RACE and a single major product of

91 bp (excluding the 38 bp 5¢-RLM-RACE adapter sequence) was obtained in 5¢-RLM-RACE (data not shown) Three clones for each 3¢-RACE and 5¢-RLM-RACE product were sequenced on both strands Sequence analysis showed that the zebrafish fabp1a cDNA was 827 bp, excluding the poly(A) tail (Fig 1A) An ORF of 384 bp from nucleotide 58 to

441, including the stop codon, was identified that codes for a polypeptide of 127 amino acids with a molecular mass of
14.1 kDa and a calculated isoelec-tric point of 4.97 The 5¢- and 3¢-UTR were 57 and

386 bp, respectively A polyadenylation signal (AAT AAA) was located from nucleotide 809 to 814 In one

of the 5¢-RLM-RACE clones, the nucleotide at posi-tion 70 was an adenine in place of guanine resulting in

a change in the encoded amino acid from Gly to Arg (Fig 1A) This nucleotide difference at position 70 in the 5¢-RLM-RACE clones for FABP1a may represent

a polymorphism or a PCR artifact

Analysis of zebrafish fabp1a cDNA using blastx revealed sequence similarity to FABP1 sequences from other species available in the NCBI database Align-ment of the human FABP1 sequence to the deduced

zebrafish FABP1a sequence (Fig 2) using bioedit

sequence alignment editor (blosum62 similarity matrix,

v 5.0.9) [26] revealed 64% sequence identity and 83%

sequence similarity to human FABP1, suggesting that

the cDNA clone codes for a FABP1 in zebrafish

Phy-logenetic analysis using clustalx [27] strongly

sup-ports the inclusion of zebrafish FABP1a in the same

clade as FABP1s from other species (boot strap

value¼ 1000; Fig 3) suggesting that the zebrafish

fabp1a gene is an ortholog of the mammalian FABP1

gene

Identification of a cDNA sequence coding for a second zebrafish FABP1, FABP1b

The cDNA sequence coded by a duplicated fabp1 gene

in zebrafish was identified and determined using 3¢-RACE and 5¢-RLM-RACE with mRNA-specific primers based on a zebrafish EST (GenBank accession number BQ075349) described as being similar to the human FABP1, hereafter referred to as the fabp1b cDNA A single product of 378 bp, excluding the

20 bp adapter sequence and the poly(A) tail, was

A

B

Fig 1 Sequences of cDNAs coding for zebrafish FABP1a and FABP1b (A) The 827 bp fabp1a cDNA sequence, excluding the poly(A) tail, was determined by cloning and sequencing of 3¢-RACE and 5¢-RLM-RACE products The cDNA sequence contained an ORF of 384 nucleo-tides coding for a polypeptide of 127 amino acids with the identity of the amino acid sequence shown below the nucleotide sequence The stop codon is underlined A variation between sequenced 5¢-RLM-RACE products is shown in bold with the variation indicated above A polyadenylation signal sequence, AATAAA, is in bold italic font The NCBI GenBank accession number for this sequence is DQ062095 (B) The 498 bp (long transcript variant) or 467 bp (short transcript variant) fabp1b cDNA sequences were determined by cloning and sequencing

of 3¢-RACE and 5¢-RLM-RACE products The ORF of 387 bp encodes a polypeptide of 128 amino acids with the amino acid sequence shown below the nucleotide sequence Variation between the 5¢-RLM-RACE products is shown in bold with the variation indicated above A poly-adenylation signal sequence, ATTAAA, is in bold italics The two transcription start sites determined from the mature and capped mRNA using 5¢-RLM-RACE are indicated by * The stop codon is underlined The NCBI GenBank accession number for these nucleotide sequences are DQ062096 (long transcript variant) and DQ474062 (short transcript variant).

obtained in 3¢-RACE Two products of 343 and

312 bp (excluding the 38 bp 5¢-RLM-RACE adapter

sequence) were obtained in 5¢-RLM-RACE for the

fabp1bmRNA owing to the presence of two

transcrip-tion start sites for the fabp1b gene (see Fig 4B and

dis-cussion below) Three clones were sequenced for each

3¢-RACE and 5¢-RLM-RACE product

Sequence analysis of the fabp1b 3¢-RACE and

5¢-RLM-RACE products revealed that the fabp1b

cDNA was either 498 bp (long transcript variant) or

467 bp (short transcript variant), excluding the poly(A)

tail, depending on the transcription initiation site

utili-zed (Fig 1B, see Fig 4B) All nucleotide positions for

fabp1bcDNA mentioned here refer to nucleotide

posi-tions in the fabp1a long transcript variant A putative

polyadenylation signal, ATTAAA, is located from

nuc-leotide 479 to 484 (Fig 1B) An ORF of 387 bp from

nucleotide 53 to 439, including the stop codon, was

identified that codes for a polypeptide of 128 amino

acids with a molecular mass of
14.1 kDa and a

cal-culated isoelectric point of 5.53 In one of the

5¢-RLM-RACE clones, the nucleotide at position 99

was thymine in place of adenine, resulting in a change

in the encoded amino acid from Glu to Val (Fig 1B)

The difference in sequence between the 5¢-RLM-RACE

clones for FABP1b may be a polymorphism or an

arti-fact of PCR Alignment of the human FABP1

sequence with the deduced zebrafish FABP1 sequence

using the bioedit sequence alignment editor

(blo-sum62 similarity matrix, v 5.0.9) [26] revealed 60%

identity and 78% similarity with human FABP1

(Fig 2), suggesting that the cDNA clone codes for a

second FABP1 in zebrafish, referred to here as

FABP1b

As with zebrafish FABP1a, phylogenetic analysis using clustalx [27] strongly supports the inclusion of zebrafish FABP1b in the same clade as FABP1s from other species (boot strap value¼ 1000; Fig 3) This suggests that zebrafish fabp1a and fabp1b genes are orthologs of the mammalian FABP1 gene, and the duplicated copies of the fapb1 genes in zebrafish most likely arose as a result of a whole-genome duplication event in the ray-finned fish lineage [20–25]

We noted that the amino acid sequences for zebra-fish FABP1a and FABP1b encoded by the sister dupli-cate genes, fabp1a and fabp1b, do not cluster as closely

as we observed for proteins encoded by other duplica-ted gene copies in this zebrafish multigene family, e.g FABP7a and FABP7b, or CRABP1a and CRABP1b [7,28] An explanation for this may be a reflection of the different rates of amino acid substitution in FABP1a compared with FABP1b (or vice versa) owing

to reduced selective pressure on one of the duplicated genes Alternatively, the topography of the tree may change with the addition of other fish FABP1s in a phylogenetic analysis when these sequences become available

Zebrafish FABP10 (basic liver-type FABP) formed

a separate clade, along with FABP10s of other non-mammalian vertebrates, suggesting that FABP10 and FABP1 are paralogs The presence of fabp1 and fabp10 genes in zebrafish and other nonmammalian verte-brates is likely due to a duplication event, presumably predating the divergences of mammals, birds, fishes, reptiles and amphibians [2] Absence of the FABP10 gene in mammals is likely due to loss of the FABP10 gene function in mammals, which was subsequently acquired by the FABP1 gene (see below)

Fig 2 Sequence alignment of zebrafish FABP1a and FABP1b with human FABP1 The amino acid sequences of the human FABP1 (Hu-FABP1; GenBank accession number P07148), zebrafish FABP1a (Zf-FABP1a) and zebrafish FABP1b (Zf-FABP1b) were aligned using the BLOSUM 62 similarity matrix in BIOEDIT sequence alignment editor (v 5.0.9) [25] Dots indicate identity and dashes were introduced to maxi-mize alignment The amino acid sequences that determine higher-order structure are shown in bold and labeled: aI, aII, a helix I and II; bA to

J, b sheet A to J The percentage sequence identity and percentage sequence similarity between the human FABP1 and zebrafish FABP1a, and FABP1b are shown at the end of each sequence.

Fig 3 Phylogenetic relationship of zebrafish FABP1a, FABP1b and FABP10 in the iLBP multigene family The bootstrap neighbor-joining phy-logenetic tree was constructed with CLUSTALX [26] using human Von Ebner’s gland protein (Hu-LCN1, GenBank accession number NP_002288) that belongs to the lipocalin family of the calycins as an out-group Bootstrap values supporting the branch points are shown as number per 1000 duplicates Branch points supported by a bootstrap value of at least 700 are indicated The inclusion of zebrafish FABP1a (Zf-FABP1a) and FABP1b (Zf-FABP1b) in the FABP1 clade is highly supported *Indicates the duplication event and subsequent divergence

of the fabp1, fabp6 and fabp10 genes 679 mya; ** indicates the whole-genome duplication event giving rise to the fabp1a and fabp1b in ze-brafish (ray-finned fishes)
200–450 mya and the subsequent subfunctionalization of gene function The sequences used in the analysis include: human FABP2 (Hu-I-FABP; GenBank accession number P12104), zebrafish FABP2 (Zf-FABP2; AF180921), cow FABP5 (Co-FABP5; P55052), human FABP5 (Hu-FABP5; Q01469), cow FABP4 (Co-FABP4; P48035), human FABP4 (Hu-FABP4; P15090), cow FABP3 (Co-FABP3; CAA31212), human FABP3 (Hu-FABP3; P05413), human FABP7 (Hu-FABP7; O15540), zebrafish FABP7a (Zf-FABP7a; AF237712), mouse FABP9 (Mo-FABP9; O08716), rat FABP9 (Ra-FABP9; P55054), human CRABP1 (Hu-CRABP1; NM_004378), human CRABP2 (Hu-CRABP2; M68867), human CRBP1 (Hu-CRBP1; NP_002890), human CRBP2 (Hu-CRBP2; P50120), pufferfish FABP1 (Pf-FABP1; AAC60290), orange-spotted grouper FABP1 (Og-FABP1; AAM22208), cow FABP1 (Co-FABP1; P80425), human FABP1 (Hu-FABP1; P07148), pig FABP1 (Pi-FABP1; P49924), rat FABP1 (Ra-FABP1; P02692), mouse FABP1 (Mo-FABP1; Y14660), chicken FABP1 (Ch-FABP1; AAK58095), salamander FABP (Sa-FABP1; P81399), pig FABP6 (Pi-FABP6; P10289), human FABP6 (Hu-FABP6; NP51161), mouse FABP6 (Mo-FABP6; NP51162), rat FABP6 (Ra-FABP6; P80020), lungfish FABP10 (Lf-FABP10; P82289), salamander FABP10 (Sa-FABP10; P81400), toad FABP10 (To-FABP10; P83409), iguana FABP10 (Ig-FABP10; U28756), chicken FABP10 (Ch-FABP10; P80226), zebrafish FABP10 (Zf-FABP10; AF254642), catfish FABP10 (Cf-FABP10; P80856) and shark FABP (Sh-FABP10; P81653) Scale bar ¼ 0.1 substitutions per site.

B

C

Fig 4 Transcription start sites for the zebrafish fabp1a, fabp1b and fabp10 genes (A) The 5¢ upstream sequence of the initiation codon for the zebrafish fabp1a gene was obtained from zebrafish genome database at the Wellcome Trust Sanger Institute (contig: ctg30243.1, assembly Zv2, http://www.sanger.ac.uk ⁄ Projects ⁄ D_rerio ⁄ ) The transcription start site was identified from the capped and mature fabp1a mRNA by 5¢-RLM-RACE and is marked by an arrow over the * sign An A-rich region that may serve a similar function as a TATA box is in bold font and underlined The initiation codon is boxed, the sequence of intron 0 is in and lower case italics, and the sequence 5¢ upstream

of the transcription start site is in lower case The six CA dinucleotide repeats not present in the genomic sequence, but present in the 5¢-RLM-RACE sequence, are underlined A single nucleotide variation between the 5¢-5¢-RLM-RACE sequence and the genomic sequence is in bold with the variation indicated above (B) The zebrafish fabp1b sequence, 5¢ upstream of the initiation codon, was obtained from the zebra-fish genome database at the Wellcome Trust Sanger Institute (scaffold 1725, assembly Zv4, http://www.sanger.ac.uk/Projects/D_rerio/) The two transcription start sites obtained from the sequences of 5¢-RLM-RACE using capped and mature fabp1b mRNA are marked by arrows over the * sign Putative TATA boxes are in bold and underlined The initiation codon is boxed and the sequence 5¢ upstream of the second transcription start site is in lower case A single nucleotide variation between the 5¢-RLM-RACE sequence and the genomic sequence is in bold above the sequence (C) The single transcription start site of the zebrafish fabp10 gene was determined by amplifying the capped and mature fabp10 mRNA by 5¢-RLM-RACE and aligning the sequence with the genomic sequence obtained from GenBank (accession number AF512998) The transcription start site is marked by an arrow over the * sign A putative TATA box is in bold and underlined The initiation codon is boxed and the sequence 5¢ upstream of the transcription start site is in lower case.

Amino acid substitutions involved in ligand

binding by FABP1

FABP1 has been reported to bind a broad range of

ligands including heme, bilirubin and certain

eicosa-noids FABP1 is the only iLBP that forms a complex

with two fatty acid molecules at the same time

[13,15,29] One ligand molecule, located at the bottom

of the protein cavity in a bent conformation, is

coordinated via an extensive hydrogen-bonding

net-work Residues S39, R122 and S124 have been shown

to be involved in the network [30–33] Based on

amino acid sequence alignment (Fig 2), the S39

resi-due is identical in human FABP1, zebrafish FABP1a

and FABP1b Residue R122 in human FABP1

corres-ponds to residue R121 in zebrafish FABP1a and

R122 in zebrafish FABP1b Residue S124 corresponds

to S123 in zebrafish FABP1a and S124 in zebrafish

FABP1b (Fig 2) Therefore, all three residues shown

to be involved in ligand binding to the primary site

(site 1) of FABP1 have been conserved in human and

zebrafish FABP1s

The second ligand molecule in FABP1 adopts a

rather linear shape, with the solvent-exposed

carboxy-late end sticking out of the fatty acid portal The

poorly delimited ‘portal’ region consists of a helix II

and the turns connecting b strands C and D, as well as

E and F (Fig 2) The hydrogen-bonding network

involved in ligand binding at site 2 involves K31, A54,

S56, D88¢ and bound water [34] Residue K31 is

identi-cal in human FABP1 and zebrafish FABP1a and

FABP1b Residue A54 corresponds to A54 in zebrafish

FABP1a and to T54 in zebrafish FABP1b The human

FABP1 residue S56 corresponds to T56 in zebrafish

FABP1a and S56 in FABP1b There are no residues in

zebrafish FABP1a and FABP1b that correspond to

residue D88¢ The interaction of amino acid D88¢ is

due to crystal packing from a symmetry-related

mole-cule and is of lesser importance [34] Variations in the

residues involved in the binding at site 2 of the human

FABP1, and possibly in zebrafish FABP1a and

zebra-fish FABP1b, may reflect differences in the binding

affinities of the zebrafish FABP1a and zebrafish

FABP1b to the ligand molecules

Transcription start sites for fabp1a, fabp1b and

fabp10 genes

Amplification of the capped and mature 5¢-ends of the

fabp1a, fabp1b and fabp10 mRNA transcripts from

adult zebrafish RNA by 5¢-RLM-RACE identified the

transcription start site(s) for the zebrafish fabp1a,

fabp1band fabp10 genes

One abundant product of 91 bp (excluding the

38 bp 5¢-RLM-RACE adapter sequence) was detected

in the reaction with CIP⁄ TAP-treated RNA amplified

in nested PCR with the adapter primer and the fabp1a mRNA-specific primer (data not shown) In order to determine the transcription start site of the zebrafish fabp1a gene, we searched the zebrafish genome data-base (assembly Zv2) at the Wellcome Trust Sanger Institute (http://www.sanger.ac.uk/Projects/D_rerio/) using the zebrafish fabp1a cDNA sequence and identi-fied a contig, ctg30243.1, containing sequence of the zebrafish fabp1a gene Alignment of the 5¢-RLM-RACE sequence with the zebrafish fabp1a gene sequence localized the transcription start site 407 bp upstream of the initiation codon (Fig 4A) An addi-tional intron was identified in the 5¢-UTR of the zebra-fish fabp1a gene by sequence alignment of the cDNA and genomic sequences, thereby generating a fifth exon, exon 0 A fifth exon is unprecedented in mem-bers of the vertebrate iLBP multigene family (Evi-dence for an additional intron in the zebrafish fabp1a gene is provided below.)

Alignment analysis also revealed that the 5¢-RLM-RACE product contained a string of 10 ‘CA’ repeats, whereas the genomic sequence had only four ‘CA’ dinucleotide repeats in exon 0 (Fig 4A) The difference

in the number of ‘CA’ repeats may represent a zebra-fish strain-specific or individual polymorphism Dinu-cleotide repeats are highly polymorphic that have been used to detect genetic variation in a number of fish species [35,36]

Two transcription start sites for the zebrafish fabp1b gene were mapped by 5¢-RLM-RACE Sequencing of 5¢-RLM-RACE products for the fabp1b gene revealed that the two products are 343 and 312 bp (excluding the 38 bp 5¢-RLM-RACE adapter sequence) Align-ment of the 5¢ upstream sequence of the zebrafish fabp1b gene (scaffold 1725, assembly Zv4, http:// www.sanger.ac.uk/Projects/D_rerio/) with the fabp1b 5¢-RLM-RACE sequences localized the two transcrip-tion start sites at 21 and 52 bp upstream of the initi-ation codon (Fig 4B)

A single band of
125 bp was produced by 5¢-RLM-RACE using primers specific for the fabp10 mRNA (data not shown) The band was isolated and purified following size-fractionation by agarose gel electrophoresis Sequencing of randomly selected clones revealed that the cDNA end was 90 bp, exclu-ding the 38 bp 5¢-RLM-RACE adapter sequence Alignment of the fabp10 5¢-RLM-RACE product and fabp10 genomic sequence reported by Her et al [37] (GenBank accession number AF512998) identified the transcription start site of the zebrafish fabp10

gene at 27 bp upstream of the initiation codon

(Fig 4C)

Putative TATA boxes were identified using

matin-spector [38] and visual inspection at position 24 to

30 bp 5¢ upstream of the first transcription start site

that is nearest to the initiation codon in fabp1b and at

22 to 29 bp upstream of the second transcription start

site that is located further from the initiation codon in

fabp1b (Fig 4B) A TATA box is present at position

25 to 31 bp 5¢ upstream of transcription start site in

fabp10 (Fig 4C) No TATA-like box was identified at

25 bp upstream of the transcription start site in

fabp1a (Fig 4A) The absence of a TATA box in the

promoter region of eukaryotic genes is not uncommon,

with
50% or more gene promoters lacking TATA

boxes [39] An A-rich region, however, is evident in

the fabp1a gene 25–31 bp 5¢ upstream of transcription

start site that may well serve a similar function as a

TATA box

Evidence for a fifth exon in the zebrafish fabp1a

gene

An additional intron of 351 bp was identified in the

5¢-UTR of the zebrafish fabp1a gene, thereby

gener-ating an unprecedented five exons in a vertebrate iLBP

gene, consisting of exons 1–4, plus exon 0 (Fig 4A)

Evidence to support this contention is: (a) none of the

expressed sequence tags coding for zebrafish FABP1a

in the NCBI database (GenBank accession numbers

DN899828, DN893893, DN893404, BI846703,

DN893348, DN893665 and CR929576) contains this

351 bp sequence; (b) the exon⁄ intron splice junctions

conform to the GT⁄ AG rule [40]; and (c) a

PCR-amplified DNA fragment of
650 bp was generated

using primers based on a sequence 5¢ upsteam of

exon 0 and a sequence within exon 1, and zebrafish

genomic DNA as template (see Experimental

proce-dures) A product of this size could only be generated

from genomic DNA if it contained an intron of

350 bp (data not shown)

Intron 0 identified in the 5¢-UTR of the zebrafish

fabp1a gene was not present in either the fabp1b or

fabp10 genes (Fig 4A–C) The rat FABP1 gene, the

only FABP1 for which a transcript start site has been

mapped to provide for a full cDNA and gene sequence

[30], does not contain an additional exon within the

5¢-UTR All vertebrate iLBP genes sequenced to date

contain four exons interrupted by three introns, with

each exon exhibiting a similar coding capacity [1,2]

No intron has been reported in the 5¢-UTR for any

other iLBP genes As an additional intron is not found

in the zebrafish fabp1b gene, a possible explanation for

the presence of intron 0 in the zebrafish fabp1a gene is

an insertional mutation that occurred after the whole-genome duplication event which followed the diver-gence of the ray- and lobe-finned fish lineages [19–24]

An alternative explanation is that the insertion in the 5¢-UTR of the fabp1a gene occurred in zebrafish inde-pendent of other fish lineages Analysis of fabp1a and fabp1b genes in various fish species would likely resolve the timing of the insertion of intron 0 in the fabp1agene

Linkage group assignment of zebrafish fabp1a, fabp1b and fabp10 genes using radiation hybrid mapping

The fabp1a gene was assigned to linkage group 5 at a distance of 10.2 cR from marker Z22208 with a LOD

of 14.5 using the LN54 panel of radiation hybrids [41] and primers specific to the fabp1a gene Using the same radiation hybrid panel and gene-specific primers, the fabp1b gene was assigned to linkage group 8 at a distance of 4.81 cR from marker Z10731 with a LOD

of 16.2, and the fabp10 gene was assigned to linkage group 16 at a distance of 8.23 cR from marker Z10256 with a LOD of 17.1 [mapping data available at ZFIN (http://zfin.org/) for fabp10 and upon request for fabp1aand fabp1b] Both zebrafish fabp1 genes showed conserved syntenies with the mouse FABP1 gene on chromosome 6 (position 30 cM), the rat FABP1 gene

on chromosome 4 (position 4q32) and the human FABP1 gene on chromosome 2 (position 2p11) [2] (Table 1) This provides further evidence that the duplicated zebrafish fabp1a and fabp1b genes are orthologs of the mammalian FABP1 gene

Based on sequence similarities, phylogenetic analysis, linkage group assignment and conserved syntenies, we can draw the following conclusions regarding the fabp1a, fabp1b and fabp10 genes: (a) the zebrafish fabp1a and fabp1b genes are orthologs of the mamma-lian FABP1 gene, and the two fabp1 genes arose as a result of a duplication event, most probably the whole-genome duplication that occurred in the ray-finned fish lineage after their divergence from lobe-finned fishes some 200–450 mya [20–25]; (b) the zebrafish fabp1 genes are the paralogs of the fabp10 gene and the proto-fabp1 gene diverged from the fabp10 gene before the fish-tetrapod split,
679 mya [2]

Tissue-specific distribution of the fabp1a, fabp1b and fabp10 gene transcripts in adult zebrafish

We analyzed the tissue distribution of the zebrafish fabp1a and fabp1b gene transcripts by RT-PCR using

total RNA extracted from various adult tissues as a

template A fabp1a- and a fabp1b-specific product were

generated by RT-PCR from total RNA extracted from

adult zebrafish intestine No fabp1a- or fabp1b-specific

RT-PCR product was generated from total RNA

extracted from liver, brain, testis, muscle, heart, skin

or ovary of the adult zebrafish (Fig 5) Product from

the constitutively expressed rack1 mRNA was

ampli-fied by RT-PCR and used as a positive control for

each RNA tissue sample assayed (Fig 5) Di Pietro

et al [42] studied the expression of the fabp1 gene in

catfish tissues by western blot analysis using antibodies

raised against rat FABP1 Catfish fabp1 expression is

restricted to the intestine as was observed by RT-PCR

analysis of fabp1a and fabp1b mRNA in the adult

zebrafish FABP1 mRNA is found in adult rat

intes-tine and liver [43], whereas in adult zebrafish, no

fabp1a or fabp1b gene transcripts were detected in the

liver (Fig 5)

A fabp10-specific product was generated from total

RNA extracted from liver, intestine and testis of adult

zebrafish (Fig 5) Total RNA extracted from brain,

muscle, heart, ovary and skin did not generate a

fabp10-specific product by RT-PCR (Fig 5) In an

earlier report, we described the detection of fabp10

mRNA in the liver of adult zebrafish using

tissue-section in situ hybridization [17] The highly sensitive

RT-PCR assay, however, also detected fabp10 gene

transcripts in intestine and testis fabp1 (FABP1), a

paralogous gene phylogenetically closely related to

fabp10, is believed to have arisen by duplication of an

ancestral gene,
679 mya [2], and is expressed in the intestine and liver of adult rats [43] The fabp10 gene

or its product has not been detected in mammals However, both fabp1 and fabp10 are expressed in nonmammalian vertebrates including fish, amphibians,

Table 1 Conserved synteny of zebrafish fabp1a and fabp1b genes with their mammalian ortholog, Fabp1 –, Data not available.

Chromosomal ⁄ linkage group assignment

8 (fabp1b)

Fig 5 RT-PCR analysis of tissue-specific distribution of the fabp1a, fabp1b and fabp10 mRNAs in adult zebrafish RT-PCR generated a fabp1a and fabp1b mRNA-specific product from total RNA extrac-ted from adult zebrafish intestine (I) No fabp1a or fabp1b mRNA-specific product was generated from adult zebrafish liver (L), heart (H), muscle (M), ovary (O), skin (S), brain (B), testes (T) or the neg-ative control (–) lacking total RNA derived from a whole zebrafish in the RT-PCR (1 and 2) RT-PCR generated a fabp10 mRNA-specific product from total RNA extracted from adult zebrafish liver (L), intestine (I) and testis (T) No fabp10 mRNA-specific product was generated from adult zebrafish heart (H), muscle (M), ovary (O), skin (S), brain (B), or the negative control (–) (3) A rack1 mRNA-specific product was generated from all the adult zebrafish tissues analyzed (4).

reptiles and birds Detection of fabp1a and fabp1b gene

transcripts in zebrafish intestine, and fabp10 gene

tran-scripts in zebrafish liver shows a division of

gene-expression patterns, or subfunctionalization [44,45], of

the gene-expression patterns exhibited by the

mamma-lian FABP1 gene As such, subfunctionalization of the

fabp1 and fabp10 genes may explain the retention of

these paralogous genes, at least in zebrafish, and

poss-ibly in many nonmammalian genomes

Developmental expression of fabp1b and fabp10

genes during zebrafish embryogenesis

Whole-mount in situ hybridization to zebrafish

embryos did not detect fabp1a mRNA in any of the

embryonic stages investigated (data not shown) fabp1b

gene transcripts were detected at the beginning of

somitogenesis, i.e 11 h post fertilization (hpf), in the

yolk syncytial layer (YSL) (Fig 6A) An increase in

the intensity of the hybridization signal suggested that

fabp1b transcripts were more abundant in the YSL

at 17 hpf (Fig 6B), at 24 hpf (Fig 6C) and then at

36 hpf (Fig 6D) At 48 hpf, fabp1b transcripts were detected in the intestinal bulb and the YSL (Fig 6E,F) In 5-day-old larvae, fabp1b transcripts continued to be abundant in the intestinal bulb (Fig 6G,H), but were also detected in the anterior part

of intestine (Fig 6G) No fabp1b transcripts were detected in the liver of the developing zebrafish

An antisense RNA probe prepared from zebrafish fabp10 cDNA did not detect fabp10 transcripts in the embryos from the gastrula (
6 hpf) up to 36 hpf (data not shown) This suggests that the fabp10 gene product, FABP10, most likely does not play an important role in the early stages of liver morphogen-esis, at least not for a few hours after the developing hepatocytes have aggregated (
28 hpf) [46] The weak detection of fabp10 gene transcripts in the ventral endoderm (near the heart chamber) of the 36 hpf zebrafish embryos [47] was not observed in this study The difference between the first detection of fabp10 mRNA reported here and that of Her et al [47] may

Fig 6 Detection of fabp1b and fabp10 mRNAs by whole-mount in situ hybridization during zebrafish embryogenesis and larval development fabp1b mRNA was detected in: (A) the YSL at the 3-somite stage (11 hpf); (B) 16-somite stage (17 hpf) embryo in the YSL; (C) in the YSL at

24 hpf embryo; increased levels in the YSL (D) at 36 hpf; and (E) at 48 hpf embryo compared with (C); (F) dorsal view of embryo shown in (E) showing expression in YSL and intestinal bulb (Ib); (G) larvae at 5 days of development showing fabp1b mRNA restricted to intestinal bulb and anterior part of intestine (Apoi); and (H) the dorsal view of embryo shown in (G) showing localization of fabp1b mRNA in intestinal bulb fabp10 mRNA was detected in (I) liver (L) of the 48 hpf zebrafish embryos (J) Dorsal view of 48 hpf embryos showing expression

in the liver (L); (K) in the liver (L) of 5-day-old larvae; and (L) dorsal view of larvae at 5 days of development showing fabp10 mRNA in the liver (L).