Duplication of the IGFBP-2 Gene in Teleost Fish: Protein Structure and Functionality Conservation and Gene Expression Divergence

Background Insulin-like growth factor binding protein-2 (IGFBP-2) is a secreted protein that binds and regulates IGF actions in controlling growth, development, reproduction, and aging. Elevated expression of IGFBP-2 is often associated with progression of many types of cancers. Methodology/Principal Findings We report the identification and characterization of two IGFBP-2 genes in zebrafish and four other teleost fish. Comparative genomics and structural analyses suggest that they are co-orthologs of the human IGFBP-2 gene. Biochemical assays show that both zebrafish igfbp-2a and -2b encode secreted proteins that bind IGFs. These two genes exhibit distinct spatiotemporal expression patterns. During embryogenesis, IGFBP-2a mRNA is initially detected in the lens, then in the brain boundary vasculature, and subsequently becomes highly expressed in the liver. In the adult stage, liver has the highest levels of IGFBP-2a mRNA, followed by the brain. Low levels of IGFBP-2a mRNA were detected in muscle and in the gonad in male adults only. IGFBP-2b mRNA is detected initially in all tissues at low levels, but later becomes abundant in the liver. In adult males, IGFBP-2b mRNA is only detected in the liver. In adult females, it is also found in the gut, kidney, ovary, and muscle. To gain insights into how the IGFBP-2 genes may have evolved through partitioning of ancestral functions, functional and mechanistic studies were carried out. Expression of zebrafish IGFBP-2a and -2b caused significant decreases in the growth and developmental rates and their effects are comparable to that of human IGFBP-2. IGFBP-2 mutants with altered IGF binding-, RGD-, and heparin-binding sites were generated and their actions examined. While mutating the RGD and heparin binding sites had little effect, altering the IGF binding site abolished its biological activity. Conclusions/Significance These results suggest that IGFBP-2 is a conserved regulatory protein and it inhibits growth and development primarily by binding to and inhibiting IGF actions in vivo. The duplicated IGFBP-2 genes may provide additional flexibility in the regulation of IGF activities.


Introduction
Peptides of the insulin-like growth factor (IGF) and insulin family control growth, metabolism, reproduction, and longevity in a wide variety of animals ranging from invertebrates to humans. In extracellular fluids, these peptides are bound to and regulated by a family of IGF binding proteins (IGFBPs) and IGFBP related proteins [1,2,3]. Most IGFBPs are present in circulation, where they act to prolong the half-life of circulating IGFs. IGFBPs are also expressed in many peripheral tissues and are present in various biological fluids. In these local environments, IGFBPs regulate the availability of IGFs to the cell surface IGF receptors, thereby regulating the net biological activity of the IGF signaling pathway. Indeed, studies using a wide variety of mammalian cell culture systems have shown that IGFBPs have the ability to inhibit and/or potentiate the biological actions of IGFs, and some IGFBPs even exhibit ligand-independent biological activities in vitro [2,4,5].
There are six distinct IGFBPs, designated as IGFBP-1 to -6, and each is coded by a single copy gene in humans and other mammals [2,4]. These IGFBPs all contain a highly conserved Nterminal domain, a conserved C-terminal domain, and a variable central linker (L) domain. The N-domain contains 12 conserved cysteine residues. C-domains have six conserved cysteine residues. These cysteine residues form intra-domain disulfide bonds [6,7] and help to shape the globular structure of the N-and C-domains. The divergent L-domain often contains sites for post-translational regulation, including glycosylation, phosphorylation, and proteolysis. Despite similar domain arrangements, each IGFBP has distinct structural features. For example, IGFBP-2 has an arginine-glycine-glutamate (RGD) sequence in its C-domain and a heparin binding domain/motif (HBD) in its L domain. Comparative studies have shown that the gene, protein primary structure, and biological actions of IGFBPs are conserved in non-mammalian vertebrates [8]. In agreement with these structural features, both IGF-dependent and independent actions of IGFBP-2 has been reported in various mammalian culture systems, and RGD and/ or heparin binding motifs have been implicated in the IGFindependent actions on cell growth and migration [9]. Ubiquitous expression of IGFBP-2 transgenes results in growth inhibition in mice [10,11]. Although IGFBP-2 is the most abundantly expressed IGFBP in fetal tissues, the IGFBP-2 knockout mice were viable and fertile, and their prenatal and postnatal body weights were similar to those of their wild-type litter mates [12]. However, these mice had reduced spleen and enlarged liver. A more recent study indicates that the IGFBP-2 knockout mice had reduced trabecular bone volume and thickness in the adult male, but not female mice [13]. The modest phenotypes may be due to the compensatory adjustments in the expression of other IGFBPs and/or by maternal factors delivered through the placenta [12].
The use of the zebrafish has provided new insights into our understanding of the physiological functions of IGFBP-2 in vertebrate development [14,15]. Targeted knockdown of zebrafish IGFBP-2a by antisense morpholino (MO) resulted in delayed development, reduced body growth, and disruptions to cardiovascular development [14].
Many teleost fish, including zebrafish, experienced an additional genome wide duplication event [16,17]. Since then, while many of the duplicated genes have been lost, a substantial percentage of the duplicates have been retained. As a result, fish often have two coorthologs in contrast to a single copy gene in humans and other mammals. One such co-orthologous pair is the zebrafish igf-1r genes [18,19]. The retention of a particular pair of genes provided the unique opportunity to gain insights into the impact of evolution in reshaping genes and their roles in physiology. In this study, we have identified two IGFBP-2 genes in the zebrafish, medaka, fugu, tetraodon, and the stickleback genomes using a comparative genomics approach. Our molecular and functional analysis on the two zebrafish IGFBP-2 genes suggests that they have undergone subfunctionalization partitioning by evolving distinct gene expression patterns, while their protein functionality remains largely unchanged. Furthermore, taking advantage of the amenability of the zebrafish model, we investigated the relative contributions of the IGF binding domain, RGD, and HBD in IGFBP-2 actions in vivo.

Zebrafish and other teleost fish have two IGFBP-2 genes
In addition to the previously reported IGFBP-2 gene [20], we identified another IGFBP-2 like gene (ENSDARP00000019643) by searching the zebrafish genome. Using 59-and 39-RACE, its full-length cDNA was determined (EF507265). The encoded protein has a putative signal peptide of 17 amino acids (aa) and a mature protein of 248 aa with a predicted molecular mass of 29.7 kDa. Structural analysis suggests that there are 12 cysteine residues in its N-terminal domain (residues 1-79) and 6 cysteine residues in its C-terminal domain (residues 152-231) (Fig. 1A). A typical IGFBP motif is present in the N-domain and there is a thyroglobulin-1 motif in the C-domain. Among various members of the IGFBP family, the new IGFBP shares higher identities with IGFBP-2 than with other known IGFBPs. The overall sequence identities of this protein with the six human IGFBPs are 33% (IGFBP-1), 44% (IGFBP-2), 32% (IGFBP-3), 37% (IGFBP-4), 33% (IGFBP-5), and 30% (IGFBP-6). Its identities with the known zebrafish IGFBPs are 33% (IGFBP-1), 62% (IGFBP-2), 29% (IGFBP-3), and 31% (IGFBP-5), respectively. Human and other vertebrate IGFBP-2s contain an Arg-Gly-Asp (RGD) sequence in the C-terminal domain [1,2]. An RGD motif is also present in the zebrafish IGFBP at a.a.position 226 to 228. The putative heparin binding motif (PKKXRP) located in the central domain in all the known mammalian IGFBP-2 is absent in the zebrafish protein.
Phylogenetic analysis using full-length amino acid sequences of known vertebrate IGFBPs grouped the newly identified IGFBP into the IGFBP-2 clade (Fig. 1B), indicating that it is a co-ortholog of human IGFBP-2. We therefore named the previously reported zebrafish IGFBP-2 gene [20] as igfbp-2a, and this new gene as igfbp-2b.
The genomic structure of igfbp-2a and igfbp-2b was determined by searching the zebrafish genome database. igfbp-2a is 30,665 bp long and contains 4 exons and 3 introns (Fig. 1C). The size and overall structure of igfbp-2b are similar (Fig. 1C). The chromosomal loci of igfbp-2a and igfbp-2b were also mapped. While igfbp-2a is located on linkage group LG 6, igfbp-2b is mapped to LG 9 ( Fig. 2A). The human IGFBP-2 gene resides on chromosome 2 adjacent to the IGFBP-5 gene in a tail-to-tail fashion [21]. In the zebrafish genome, we found that zebrafish igfbp-2a and igfbp-5a are located on LG 6, next to each other in a tail-to-tail fashion. Likewise, zebrafish igfbp-2b and igfbp-5b are next to each other in a tail-to-tail fashion. In the proximate regions, there are several other zebrafish genes (tns1, tmem169) whose orthologs are located on human chromosome 2 ( Fig. 2A). This conserved synteny strongly suggests that igfbp-2a and igfbp-2b are co-orthologs of the human IGFBP-2 gene.
To determine whether the duplication of the IGFBP-2 gene is unique to zebrafish or a more general phenomenon in teleost fish, we searched the available genome databases for medaka (Oryzias latipes), Fugu (Takifugu rubripes), Tetraodon (Takifugu nigroviridis), and the three-spine stickleback (Gasterosteus aculeatus) and investigated the evolutionary relationships within this subfamily. As in the case of zebrafish, there are two igfbp-2 genes in all four teleoest species and their primary structures are highly similar to that of IGFBP-2 ( Fig. 1A). Phylogenetic analysis using mammalian IGFBP-2 as an out-group indicated that the duplication of the igfbp-2a/b subfamily expanded during the ray-fin fish genome duplication (Fig. 2B).
Zebrafish igfbp-2b encodes a secreted protein that binds IGFs Previously, we have shown that zebrafish IGFBP-2a binds IGFs with high affinity and specificity [20]. To determine whether igfbp-2b also encodes a functional IGFBP, we constructed an expression plasmid by subcloning the zebrafish IGFBP-2b ORF into the pcDNA3.1/ Myc-His(-)A expression vector. As a positive control, a pcDNA3.1/Myc-His(-)A-human IGFBP-2 plasmid was also generated. After these plasmids were introduced into HEK293T cells by transient transfection, conditioned media were prepared and subjected to Western immunoblot and ligand blot analysis. Both human and zebrafish IGFBP-2b had apparent sizes between 36 and 50 kDa on SDS-PAGE, likely due to the Myc tagging and posttranslational modifications. Zebrafish IGFBP-2b had a slightly smaller apparent size than human IGFBP-2 (Fig. 3A). Ligand blotting showed that both zebrafish IGFBP-2b and human IGFBP-2 proteins bind human IGF-1 (Fig. 3B). These results suggest that zebrafish igfbp-2b also encodes a secreted protein that binds to IGFs.
Zebrafish igfbp-2a and -2b exhibit distinct spatiotemporal expression patterns The spatial and temporal expression patterns of igfbp-2a and -2b were determined by semi-quantitative RT-PCR and in situ hybridization. As shown in Fig. 4A, IGFBP-2a mRNA was not detectable until 10 hpf. The IGFBP-2a mRNA levels gradually increased from 10 to 36 hpf and was maintained at high levels thereafter. Likewise, IGFBP-2a mRNA was undetectable by in situ hybridization in early embryonic stages. At 24 hpf, IGFBP-2a mRNA was detected in the lens and cranial region (Fig. 4B). At 48 and 72 hpf, IGFBP-2a mRNA was detected in the brain boundary vasculature. Expression in these regions persisted throughout the hatching period (Fig. 4B). By 96 hpf, IGFBP-2a mRNA expression was most abundant in the liver. This dynamic spatiotemporal pattern of igfbp-2a is in good agreement with a previous study [14]. As shown in Fig. 4A, IGFBP-2b mRNA was easily detected by RT-PCR throughout embryogenesis. In agreement with the RT-PCR data, in situ hybridization analysis showed that IGFBP-2b mRNA was detectable at all stages examined. In early embryos, the IGFBP-2b signal was detected in most, if not all, tissues at low levels. At 96 hpf, IGFBP-2b mRNA expression was mostly detected in the liver (Fig. 4B).
In the adult stage, the highest levels of IGFBP-2a mRNA were found in the liver in both the male and the female (Fig. 4C). In addition, modest levels of IGFBP-2a mRNA were seen in the brain. Low levels of IGFBP-2a mRNA were also detected in the muscle and gonad in male, but not female adult fish. There appeared to be sexual dimorphism in the expression pattern of IGFBP-2b mRNA in the adult stage. In the male, it was exclusively expressed in the liver. In the female fish, it was not only detected in the liver, but also in a number of other tissues, including the gut, kidney, ovary, and muscle. These results indicate that the duplicated igfbp-2a and -2b have evolved distinct spatial and temporal expression patterns.
To determine if the two zebrafish IGFBP-2 genes have associated with divergent regulatory elements, we analyzed a 2 kb fragment upstream of the cap site in igfbp-2a and igfbp-2b. Transcription factor binding sites within the regions were predicted by the on-line program Match using a library of mononucleotide weight matrices from TRANSFAC 6.0 (www. gene-regulation.com) with both core similarity and matrix similarity of the transcription factor binding sites set higher than 0.85. The results revealed that the two promoter regions have different regulatory element sites ( Fig 4D).
Ectopic expression of IGFBP-2a and IGFBP-2b causes a similar degree of reduction in embryonic growth and development but has no effect on cell fate and patterning To determine and compare the biological activities of zebrafish IGFBP-2a and -2b, we performed in vivo functional studies. For this, capped mRNAs encoding zebrafish IGFBP-2a and -2b were generated and introduced into zebrafish embryos by microinjection. As shown in Fig. 5A, embryos injected with either zebrafish IGFBP-2a or -2b were morphologically normal, but they were smaller in size and developmentally delayed compared to GFP mRNA injected or wild type control embryos. The body lengths of IGFBP-2a and IGFBP-2b injected embryos were 1.9260.14 mm and 1.9960.17 mm at 24hpf, which are significantly smaller than the 2.3260.22 mm and 2.2660.15 mm of the WT and GFPinjected embryos (p,0.05). At 24 hpf, wild type embryos and GFP mRNA injected control embryos had 29.460.6 and 28.060.8 somites, respectively. In comparison, embryos injected with either zebrafish IGFBP-2a or IGFBP-2b had only 23.461.9 and 23.261.5 somites (p,0.05) (Fig 5B). They were developmentally equivalent to wild type or GFP-injected embryos at ,20-21 hpf. We next performed in situ hybridization to determine any potential defects in patterning. Overexpression of either IGFBP-2a or IGFBP-2b did not alter the brain and somite patterning, as indicated by mRNA expression patterns of emx1 (labeling Zebrafish igfbp-2a and igfbp-2b are located on linkage group (LG) 6 and 9, and human IGFBP-2 is located on chromosome 2. The human IGFBP-2 and IGFBP-5 are neighboring genes in a tail-to-tail orientation. Likewise, zebrafish igfbp-2a/igfbp-5a and igfbp-2b/igfbp5-b are also neighboring genes in a tail-to-tail orientation. B) Phylogenetic tree of IGFBP-2. Amino acid sequences of full-length IGFBP-2s were analyzed. Values on branches are percentages that the two clades branched as sisters (1000 run). The results indicate that teleost igfbp-2a and igfbp-2b are likely a result of a fishspecific genome duplication event. doi:10.1371/journal.pone.0003926.g002 forebrain), rx1 (retina), pax2a (optic stalk, mid-hindbrain boundary, hindbrain), egr2b (third and fifth rhombomeres), and myoD (somite) (Fig 5C). These data suggest that the two IGFBP-2 isoforms have comparable biological activities. When expressed, they both inhibited embryonic growth and developmental rates without notable effects on cell fate determination or patterning.
The conserved growth inhibitory action of IGFBP-2 is dependent on IGF binding but not its interaction with heparin or integrin in vivo To determine whether the biological action of IGFBP-2 is conserved between teleost and humans, capped mRNAs encoding human IGFBP-2 were generated and introduced into zebrafish embryos. Embryos injected with human IGFBP-2 mRNA phenocopied those of zebrafish IGFBP-2a and -2b mRNA injected embryos, i.e., they were smaller, developmentally delayed, but morphologically normal. Further quantitative analysis showed that ectopic expression of human IGFBP-2 caused a comparable decrease in growth and developmental rates (Fig. 6B).
In addition to its ability to bind IGFs, human IGFBP-2 was reported to also act independently of IGF via its RGD and/or heparin binding motifs in cultured mammalian cells [22,23]. The in vivo significance of these in vitro findings, however, has not been well studied. Because the biological action of IGFBP-2 is conserved, we tested the functional importance of these motifs in vivo by taking advantage of the zebrafish model. For this, human IGFBP-2 mutants with altered ligand binding site (LBDm), RGD (RGDm), and heparin-binding site (HBDm) were engineered (Fig. 6A). The effects of these mutations on IGF binding and cell surface association have been characterized in vitro by others [6,22,23]. The corresponding capped mRNA was synthesized and introduced into zebrafish embryos. While the RGDm and HBDm had similar effects in slowing down growth and developmental rates (Fig. 6B), embryos injected with LBDm mRNA were indistinguishable from those of GFP mRNA injected controls in morphology, body size, and somite number (Fig. 6B), suggesting that altering the IGF binding site abolishes its biological activity.
These results suggest that IGFBP-2 is a conserved regulatory protein and it acts by binding to IGF and inhibiting its actions in vivo.

Discussion
In this study, we have demonstrated that the ortholog of tetrapod IGFBP-2 has duplicated in the zebrafish and other teleost fish. The conclusion that there are two IGFBP-2 genes in these teleost genomes is supported by phylogenetic and detailed structural data. Fish IGFBP-2a and -2b share much higher identities with human IGFBP-2 than with other human IGFBPs. All known vertebrate IGFBP-2s contain an RGD sequence in the C-terminal domain [2]. An RGD motif is also present in the zebrafish IGFBP-2a and -2b. Phylogenetic analyses grouped the newly identified genes into the IGFBP-2 clade. The strongest evidence came from the conserved synteny analysis. In humans, IGFBP-2 and -5 genes are located next to each other on the same chromosome in a tail-to-tail orientation [1]. In this study, we found that igfbp-2a and igfbp-5a are neighboring genes located on LG6 in a tail-to-tail orientation, resembling the situation of human IGFBP-2 and -5 genes localized on chromosome 2. If igfbp-2a and igfbp-5a arose by a chromosomal duplication event, one would expect that igfbp-2b should also be located next to another igfbp-5b gene. Indeed, our analysis shows that igfbp-2b and igfbp-5b genes are neighboring genes on LG9. In addition to these IGFBP genes, we were able to identify other orthologous genes between zebrafish LG9 and human Chr.2. This conserved synteny strongly argues that zebrafish igfbp-2a and igfbp-2b are co-orthologs of the human IGFBP-2 gene.
The fact that there are also two IGFBP-2 genes in the medaka, fugu, tetraodon, and the stickleback implies that the origin of this gene duplication is ancient, likely before the divergence of ray-fin fish and tetrapods. This notion is also supported by the phylogenetic analysis data and is in good agreement with the notion that a genome duplication event occurred in the teleost lineage ,350 million years ago, prior to the beginning of the teleost radiation [16,17,24]. While the majority of duplicated genes have been lost since the genome duplication event in the teleost lineage, a substantial percentage of the duplicates have been retained. When the putative zebrafish orthologs of 74 human genes on chromosome 17 were examined, Postlethwait et al. found that 15 (20%) are present as duplicate genes [25]. Like the mammals, the IGF signaling systems in teleosts are composed of IGF ligands, receptors, and IGFBPs [8,15]. It has been shown previously that zebrafish contain two igf-1r genes and two insulin receptor genes [18]. Recent studies suggest that there are two igf-1 genes and two igf-2 genes in zebrafish (our unpublished data). Additionally, there is also evidence indicating that zebrafish have two igfbp-1 genes [26]. In this study, we have shown that there are two igfbp-2 and two igfbp-5 genes in zebrafish. Therefore, it appears that all major components in the fish IGF signaling pathway are retained as duplicates. This high rate of gene retention in genes belonging to the IGF signaling pathway in the ray-fin fish is interesting. Unlike mammals, most teleosts do not reach a static adult size but continue to grow indeterminantly well beyond puberty. Given its central importance in growth regulation, it is possible that the retention of many duplicates of major players in the IGF signaling system may play an important role in the continuous growth pattern of the teleost.
The retention of two functional igfbp-2 genes in zebrafish also provided a unique opportunity to gain insights into how the IGFBP genes may have evolved through partitioning of ancestral functions. Subfunction partitioning can involve protein structural  changes and has important functional and evolutionary consequences [17]. Taking advantage of the zebrafish model, we examined the spatial and temporal expression patterns and biological activities of the duplicated igfbp-2 genes. Analysis of expression patterns for zebrafish igfbp-2a and igfbp-2b clearly reveal spatial and temporal subfunction partitioning. While zebrafish igfbp-2b is expressed throughout the entire early developmental stages in a ubiquitous fashion, igfbp-2a mRNA is not detected until 10 hpf in a highly tissue specific and dynamic manner. In the adult stage, these two duplicates are expressed in overlapping yet distinct tissues. While both igfbp-2a and -2b are highly expressed in the liver in both the adult male and female fish, only IGFBP-2a mRNA was detected in the brain. There appears to be a sexual dimorphism in their expression patterns. IGFBP-2a mRNA was also detected in the muscle and gonad in the male at low levels but not the female adult fish. In contrast, IGFBP-2b mRNA was detected in the gut, kidney, ovary, and muscle in adult female fish only. These gender specific expression patterns are interesting in light of the recent findings that male mice had 66% higher circulating IGFBP-2 levels compared with females [13]. While the duplicated zebrafish IGFBP-2 genes exhibit clear temporal and spatial partitioning in their expression, our previous and current biochemical analysis suggest that the duplicated igfbp-2 genes have similar biological properties. Both zebrafish IGFBP-2a and -b are secreted proteins that are capable of IGF binding. Furthermore, overexpression of zebrafish IGFBP-2a and -2b in zebrafish embryos had nearly identical phenotypes, i.e., they caused a similar decrease in the growth and developmental rates. These data suggest that the duplicated igfbp-2 genes encode two functionally equivalent proteins but have evolved distinct spatial and temporal expression patterns.
A number of in vivo studies using transgenic mice have shown that IGFBP-2 has a predominantly inhibitory effect on IGF actions [4]. Our functional analysis using transgenic zebrafish suggests that the inhibitory action of IGFBP-2 is conserved across species. When overexpressed in developing zebrafish embryos, human IGFBP-2 had a growth inhibitory effect similar to that of zebrafish IGFBP-2. While the inhibitory effect of IGFBP-2 can be explained by its competition for IGF binding with the IGF-1R based on biochemical and cell culture studies [2,4,5], in vivo evidence supporting this theory is still scarce. In this study, we have tested this model using transgenic zebrafish embryos. Our data shows that mutation of the major IGF binding site of human IGFBP-2 completely abolished its biological activity in this in vivo animal Figure 6. The growth inhibitory effect of IGFBP-2 is evolutionarily conserved and is dependent on its IGF binding but not its interaction with heparin or integrin. A) A schematic diagram showing the native and mutant IGFBP-2 structure. The ligand binding mutant (LBDm) was engineered following Forbes et al. [6], heparin binding domain mutant (HBDm) following Russo et al. [22], and RGD mutant (RGDm) following Wang et al. [23].  . Ectopic expression of IGFBP-2a and IGFBP-2b causes a similar degree of reduction in embryonic growth and development but has no effect on cell fate and patterning. A) Representative images of wild-type, GFP mRNA (800 pg/embryo), IGFBP-2a mRNA (800 pg/ embryo), or IGFBP-2b mRNA 800 pg/embryo) injected embryos at 24 hpf. B) Body length and somite number of the groups indicated. Results are from three independent microinjection experiments. The total embryo number for each group is 79 (wt), 95(gfp), 79(igfbp-2a), and 100 (igfbp-2b). Values are represented as means6S.E. (n = 3). Groups with common letters are not significantly different from each other (p,0.05). C) Whole mount in situ hybridization analysis of various marker genes in gfp mRNA-injected control (left column), igfbp-2a mRNA-injected (central column), and igfbp-2b mRNA-injected (right column) embryos at 24 hpf: emx1 expression in the forebrain; pax2a expression in the optic stalk, mid-hindbrain boundary, and hindbrain; egr2b expression in the third and fifth rhombomeres of the hindbrain; rx1 expression in the retina; myoD expression in the somatic myotome. Scale bar = 100 mm. Similar patterns were observed in all embryos examined in each group (n = 8-14). doi:10.1371/journal.pone.0003926.g005 model. These results have provided direct evidence that IGFBP-2 is a conserved regulatory protein and it acts primarily by binding to and inhibiting IGF actions during early development in vivo.
IGF-independent actions of IGFBP-2 have been documented in cultured mammalian cells [2,4]. It has been suggested that an RGD motif in human IGFBP-2 may be partially responsible for its cell surface association. The RGD motif in human IGFBP-2 has been reported to be involved in binding to integrin and this interaction is important for IGFBP-2-induced SNB19 glioma cell migration [23]. Since all known vertebrate IGFBP-2s, including the zebrafish IGFBP-2b reported in this study, have an RGD motif in their C-domains, we tested the possible role of RGD in mediating any biological actions of IGFBP-2 in transgenic zebrafish in vivo. We discovered that mutation of the RGD motif in human IGFBP-2 did not cause any notable changes in the growth inhibitory effects of IGFBP-2 in this animal model. This is consistent with the report by [11], which showed that overexpression of an RGE IGFBP-2 mutant did not affect IGFBP-2 cell surface association in transgenic mice.
In addition to the RGD motif, human IGFBP-2 possesses a putative heparin-binding motif (HBD). Recent studies have suggested that the association of IGFBP-2 with the cell surface or the extracellular matrix is due to its ability to bind to proteoglycans [22], but the relative contributions of the HBD in vivo is unknown. Taking advantage of the zebrafish model and the fact that the biological action of human IGFBP-2 is conserved, we investigated the role of the HBD in vivo. Our results have shown that transgenic zebrafish overexpressing the HBDm were essentially indistinguishable from those overexpressing native human IGFBP-2, indicating that the mutation of the HBD does not alter the growth inhibitory action of IGFBP-2. This conclusion is further supported by the fact that the two zebrafish IGFBP-2s had similar biological effects even though the HBD motif is not present in these proteins [27].
In conclusion, we have shown that there are two functional IGFBP-2 genes in zebrafish and other teleosts, which likely resulted from a genome wide duplication event before the beginning of the teleost radiation. Both zebrafish igfbp-2a and -2b encode secreted proteins that bind IGFs. Gene expression and functional analysis suggest that the two fish IGFBP-2 genes have undergone subfunctionalization partitioning by evolving distinct gene expression patterns, while their protein functionality remains largely unchanged. We further show that IGFBP-2 is a conserved regulatory protein and it inhibits embryo growth and development primarily by binding to IGF and inhibiting its actions in vivo. This study has not only provided novel insights into the functional evolution of IGFBPs but also demonstrated the utility of the zebrafish model system in the study of IGFBP physiology.

Chemicals
Reagents and chemicals were purchased from Fischer Scientific (Pittsburgh, PA) unless noted otherwise. RNase-free DNase and restriction enzymes were obtained from Promega (Madison, WI) and New England Biolabs (Beverly, MA). Pfu Turbo DNA polymerase was from Stratagene (La Jolla, CA). Superscript II reverse transcriptase (RT) and oligonucleotide primers were obtained from Invitrogen Life Technologies, Inc. (Carlsbad, CA).

Experimental animals
Zebrafish (Danio rerio) were maintained on a 14 h light/10 h dark cycle. Fertilized eggs, obtained by natural cross, were raised in embryo medium at 28.5uC and staged according to Kimmel et al. [28]. For in situ hybridization analysis, embryo medium was supplemented with 0.003% (w/v) 2-phenylthiourea to inhibit embryo pigment formation. All experiments were conducted in accordance with the guidelines approved by the University Committee on the Use and Care of Animals at the University of Michigan.
For whole mount in situ hybridization analysis, plasmids encoding zebrafish IGFBP-2a 39 UTR and IGFBP-2b partial ORF with 39UTR cDNAs or other genes were linearized by restriction enzyme digestion, followed by in vitro transcription reactions with either T7 or SP6 RNA polymerase, and then generated as digoxigenin (DIG)-labeled RNA probes. The specificity of each of the IGFBP-2 riboprobe was verified by dotblot analysis and shown not to cross-react with each other's target. Hybridization was carried out as described previously [18]. Images were captured with a Nikon DC50NN camera mounted to a Nikon Eclipse E600 microscope (Melville, NY).

Expression of recombinant IGFBP-2 proteins
Recombinant c-Mycand 6xHistidine-tagged human and zebrafish IGFBP-2b were produced by transfecting the expression plasmids into human embryonic kidney (HEK) 293T cells. Cells were cultured in high glucose Dulbecco's modified Eagle's medium with 10% fetal bovine serum. The transfected cells were maintained with culture medium containing 10% FBS and antibiotics (100 units/ml Penicillin and 100 mg/mL streptomycin sulfate, Gibco BRL). After the cells reached confluence, they were incubated with serum free media (SFM) for another 48 h. The SFM containing recombinant proteins was harvested, and the protein expression was confirmed by Western immunoblotting and ligand blotting as described below.

Western immunoblotting and ligand blotting analysis
The SFM containing recombinant proteins was separated by SDS-PAGE and transferred to Immobilon P membranes (Millipore Corp., Bedford, MA) and subjected to Western blot using monoclonal anti-c-myc (clone 9E10/SC40; Santa Cruz, CA) as reported previously. Ligand blot analysis was performed using digoxigenin-labeled human IGF-I following published procedure [29].

Microinjection experiment
Capped mRNA synthesis was carried out using a commercial kit and linearized plasmid DNA as template (mMESSAGE mMA-CHINE kit; Ambion, Inc.). mRNA (800 pg per embryo) was microinjected into zebrafish embryos at the 1-2 cell stage as reported previously (14). GFP mRNA injected and wild type embryos were used as controls. After injection, embryos were placed in embryo rearing medium [30] and kept at 28.5uC. Body length and somite number were measured at 24 hpf as described previously [31].
Statistics-Values are means6S.E.M. Differences among groups were analyzed by one way ANOVA followed by Tukey's Multiple Comparison Test using GraphPad Prism version 5.00 (San Diego, CA). Significance was accepted at p,0.05.