Introduction

In mammals, mannose 6-phosphate receptors (MPRs) play an important role in lysosome biogenesis by sorting newly synthesised acid hydrolases at the trans-Golgi network (TGN) (reviewed in Dahms and Hancock 2002; Ghosh et al. 2003). Two structurally related MPRs have been described [referred to as the cation-dependent MPR (CD-MPR) and cation-independent MPR (CI-MPR)], and it has been suggested that they arose from a common ancestor, with the CI-MPR resulting from multiple duplications of a single ancestral gene (Kornfeld 1992). Both types of MPR are present in chicken, reptiles, amphibians and fish, and putative MPRs have been reported in invertebrates (Nadimpalli and Von Figura 2002). The invertebrate M6P-specific proteins are not well characterized, and thus, we do not yet know how the MPR genes evolved.

In addition to their involvement in lysosome formation, the MPRs have other biological activities. In particular, the CI-MPR is an endocytic receptor and interacts with M6P-containing ligands that do not have acid hydrolase activity (Dahms and Hancock 2002). These include many peptides that are important in embryonic development such as TGF-β, proliferin and leukaemia inhibitory factor (LIF). The CI-MPR interacts with several ligands through M6P-independent means. These include the foetal mitogen, insulin-like growth factor 2 (IGF2), which the receptor targets for degradation, and this activity is essential for normal mammalian development. The acquisition of an IGF2-binding site by the CI-MPR appears to have occurred after the divergence of marsupial and placental mammals from their common ancestor with egg-laying mammals, and it has been suggested that this acquisition was a major factor in driving the evolution of an imprinted CI-MPR in some mammals (Killian et al. 2000). However, a high-affinity IGF2 binding site has been described in the CI-MPR of a teleost fish (trout) (Mendez et al. 2001), raising the possibility that IGF2 binding was an ancestral property of the CI-MPR.

Recently, a Drosophila protein (lysosomal enzyme receptor protein, LERP) that is structurally and functionally related to the mammalian CI-MPR was identified (Dennes et al. 2005). LERP mediates lysosomal enzyme targeting and rescues the missorting of lysosomal enzymes that occurs in MPR-deficient mammalian cells. Interestingly, however, the residues that are involved in M6P recognition in mammalian MPRs are not conserved in the Drosophila protein, and LERP does not bind to the multimeric M6P ligand phosphomannan. The nature of the interaction between Drosophila LERP and mammalian lysosomal enzymes has not yet been elucidated. However, it may reflect an evolutionarily ancient aspect of lysosome biogenesis that predates the M6P-dependent trafficking of lysosomal enzymes seen in present-day mammals. It also raises the question of how, and when, the involvement of the M6P-specific system in lysosome biogenesis evolved.

Clearly, the structures and ligand-binding activities of the MPRs raise questions about their phylogenetic origin, evolutionary history and functional significance in vivo. These can only be answered by comparative studies that include non-mammalian vertebrates and invertebrates. We predict that the zebrafish, a genetically tractable non-mammalian vertebrate, will be pivotal in these comparisons. In this report, we compare the zebrafish CD-MPR and CI-MPR sequences with those of their mammalian orthologs. We also describe the expression patterns of the zebrafish MPRs during embryonic development.

Materials and methods

Genomic DNA was prepared from adult zebrafish by conventional protease K digestion and alcohol precipitation procedures. Total RNA was extracted from adult zebrafish by homogenization of approximately 100 mg of tissue in RNA-Stat 60 (Tel-Test, Friendswood, TX). cDNA was prepared by reverse transcription of 1–5 μg of DNase-treated total RNA (Superscript II, Life Technologies, Grand Island, NY). Polymerase chain reaction (PCR) amplification was performed with Platinum Taq DNA polymerase (Life Technologies) or with the Expand Long Template PCR system using buffer 3 (Roche, Alameda, CA) (primer sequences available upon request from authors). Amplified fragments were sequenced with an automated ABI 377 sequencer (PE Biosystems, Foster City, CA) using the manufacturer’s BigDye terminator cycle sequencing kit (Duke University’s DNA sequencing facility).

Identification of zebrafish MPR cDNA sequences

Zebrafish CD-MPR cDNA (NM_213205) was identified by searching the National Center for Biotechnology Information (NCBI) database with the mouse CD-MPR sequence (NM_010749). For zebrafish CI-MPR sequence, we initially accessed the publicly available results of a Sanger Center ‘Human BLAST against Zebrafish’ search to identify zebrafish genomic trace files that showed similarity to the human CI-MPR/IGF2R (located on human chromosome 6; accession no. NM_000876). These data are available at http://134.174.23.160/HumanblastZebrafish/. Sequences that produced significant alignments were used to design oligonucleotide primers for PCR assays with zebrafish cDNA as template, and cDNA fragments corresponding to the majority of the zebrafish CI-MPR were amplified. During this process, the sequence of a genomic contig in linkage group 20 that contained all the zebrafish CI-MPR cDNA sequence became available (accession no. AL929119), and using this, we completed the experimental determination of the CI-MPR coding sequence. The zebrafish CI-MPR cDNA sequence has been deposited in GenBank (accession no. AY570286). We found no evidence for duplicate copies of either zebrafish MPR.

Sequence analysis

Signal peptide sequences, transmembrane domains, cytoplasmic tails and predicted sites of N-linked glycosylation were identified using the predictor programmes available at http://www.cbs.dtu.dk. Multiple sequences were aligned using the ClustalW programme at http://www.ebi.ac.uk. Accession numbers of compared CD-MPR sequences are as follows: bovine, P11456; human, NP_002346; and mouse, NM_010749. Accession numbers of compared CI-MPR sequences are chicken, AAC59718; echidna, AAL23910; bovine, NP_776777; human, NP_000867; kangaroo, AAF19160; mouse, NP_034645; opossum, AAL23909; platypus, AAF68173; and Xiphophorus, CAB94817.

Whole-mount in situ hybridization

cDNA fragments of the zebrafish CI-MPR transcript (772 bp, spanning exons 43–47) and of the CD-MPR transcript (596 bp, spanning the last two exons) were amplified by reverse transcriptase (RT)-PCR and subcloned into a pGEM-T easy vector (Promega). Digoxigenin-labelled sense and antisense riboprobes were generated, and in situ hybridization was performed on AB wild-type zebrafish embryos as described (Eivers et al. 2004). At later developmental stages, 18, 24, 36, 48 and 72 h post-fertilization (HPF), fixed embryos were treated with proteinase K (10 μg/ml) for 4, 8, 15, 45 and 60 min, respectively.

Results and discussion

The zebrafish MPRs

The open reading frame of the zebrafish CI-MPR (AY570286) encodes a protein of 2,459 amino acids: a signal peptide of 29 amino acids, a 2,258-amino-acid extracytoplasmic domain, a 23-amino-acid transmembrane region and a 149-amino-acid cytoplasmic tail. The extracytoplasmic domain resembles those of mammalian and chicken CI-MPRs in consisting of 15 repeating units (average length 147 amino acids), and containing an insert in repeat 13, similar to the type II repeats found in the collagen-binding site of fibronectin (Dahms and Hancock 2002). The amino acid sequence of the mature zebrafish receptor is 51 and 53% identical to those of the human and chicken receptors, respectively, and the amino terminal portion is 61% identical to the partial CI-MPR sequence of the teleost Xiphophorus (Yerramalla et al. 2000). The zebrafish CI-MPR contains 118 cysteine residues in the extracytoplasmic domain, all of which are conserved in the human, bovine and chicken receptors (not shown). This region also contains 18 predicted sites for N-linked glycosylation, indicating that the molecular weight of the expressed protein is likely to be larger than predicted from the coding sequence (271 kDa). A search of the NCBI database revealed a zebrafish mRNA (accession no. NM_213205) encoding an open reading frame with striking similarity to mammalian CD-MPRs (overall amino acid identity 44–47%). The predicted zebrafish CD-MPR consists of a 20-residue signal peptide, an extracytoplasmic domain of 154 amino acids, a transmembrane domain of 23 residues and a 68-amino-acid cytoplasmic tail. Six conserved cysteine residues, essential for the formation of properly folded receptors in mammals, are present in the extracellular domain.

Carbohydrate recognition sites of the zebrafish MPRs

The amino acid residues that are critical for ligand binding by both MPRs have been identified (Dahms and Hancock 2002). Four amino acids are essential for carbohydrate recognition by the bovine CD-MPR (Q66, R111, E133 and Y143) (Sun et al. 2005), and all four are conserved in the zebrafish receptor (Fig. 1a). An aspartic acid that plays a key role in Mn2+ binding by the bovine CD-MPR (D103) (Sun et al. 2005) is also present in the zebrafish protein (Fig. 1a). The mammalian CI-MPR has two carbohydrate recognition domains located in repeats 3 and 9 (Dahms and Hancock 2002), and in zebrafish CI-MPR, these repeats are 53 and 45% identical to the corresponding bovine repeats, respectively. In addition, each repeat contains the five amino acids that are critical for M6P binding except for zebrafish T412, which is a conserved substitution for bovine S431 (Fig. 1b,c). The two carbohydrate recognition domains of the bovine CI-MPR have different binding properties: repeat 9 is highly specific for the phosphomonoester M6P, whereas repeat 3 is able to bind the M6P-OCH3 phosphodiester (Dahms and Hancock 2002). This difference has been explained by the presence of small residues in domain 3 as opposed to bulkier residues in domain 9 (Dahms and Hancock 2002). In the zebrafish receptor, the corresponding residues in repeat 3 are also small (S411, T412 and G413), while those in repeat 9 are bulky (H1299, K1300 and I1301) (Fig. 1b,c).

Fig. 1
figure 1

Conservation of carbohydrate recognition domains of zebrafish MPRs. a The extracytoplasmic domain of the zebrafish CD-MPR (NM_213205) was aligned with the indicated mammalian CD-MPRs using ClustalW analysis. The amino acid residues involved in recognition of M6P are indicated by the arrowheads above the alignment, and the Asp residue involved in coordination of Mn2+ is indicated by a filled circle. b, c Repeats 3 (b) and 9 (c) of the bovine and zebrafish (AY570286) CI-MPRs were aligned using Clustal W [the position of the repeats was established using the signal sequence cleavage site as the amino terminus of repeat 1 and the conserved XLXXL motif as the carboxy terminus (Yerramalla et al. 2000)]. The five amino acids that are critical for ligand binding in each repeat of the bovine CI-MPR (Dahms and Hancock 2002) are indicated by arrowheads. Three small residues in repeat 3 and three bulky residues in repeat 9, associated with M6P phosphomonoester and M6P-OCH3 phosphodiester binding, respectively (Dahms and Hancock 2002), are indicated in bold. (consensus line below the alignments: * indicates identical residues in all sequences; : indicates highly conserved residues; . indicates weakly conserved residues)

Full size image

Conserved targeting signals in the cytoplasmic tails of zebrafish MPRs

A key function of MPRs is to segregate newly synthesised acid hydrolases from secretory proteins at the TGN. In both MPRs this sorting signal is composed of a cluster of acidic residues followed by a dileucine motif at the carboxy terminus of the cytoplasmic domain [referred to as an acidic cluster–dileucine motif (AC-LL)] and also involves several binding sites for AP-1 (Ghosh et al. 2003 and references therein). Carboxy-terminal AC-LL motifs and conserved AP-1 binding sites are present in the cytoplasmic tails of zebrafish CD-MPR and CI-MPR (Fig. 2a,b). In the case of the zebrafish CI-MPR, the mammalian AP-1 binding site YSKV is represented by YSRV, and the ETEWLM site is represented by EMEWLM (Fig. 2b).

Fig. 2
figure 2

Conserved sorting signals in the cytoplasmic tails of zebrafish MPRs. The cytoplasmic tails of vertebrate MPRs were aligned by ClustalW (numbering starts after the transmembrane domains). a Conserved sorting signals in the zebrafish CD-MPR consist of the acidic-dileucine motif (AC-LL) at the carboxy terminus (residues 62–66), two AP-1 binding sites (residues 28–43 and 50–68), the FW motif at residues 18–19 and three internalization sequences (residues 10–18, 46–49 and the region containing the AC-LL motif). b The zebrafish CI-MPR contains an AC-LL motif at the carboxy terminus (region 4) and four AP-1 binding sites (YSRV in region 1, EMEWLM in region 2, and DSED and DDSDED in regions 3 and 4, respectively). The TIP47-interacting region is indicated by the filled circles above the alignment, with the PPAP sequence of the bovine receptor highlighted in bold. There is also a conserved internalization motif (YSRV, region 1). (consensus line as in Fig. 1)

Full size image

Following delivery of acid hydrolases, mammalian MPRs do not enter lysosomes and are instead recycled from late endosomes to the TGN in processes mediated by the protein TIP47 (Ghosh et al. 2003). TIP47 recognises a phenylalanine–tryptophan motif in the cytoplasmic tail of the bovine CD-MPR, and this motif is present in the zebrafish CD-MPR (residues 18–19 of the cytoplasmic tail, Fig. 2a). The TIP47 recognition site of the CI-MPR is located between residues 48 and 74 of the bovine cytoplasmic tail, and the tetrapeptide 49PPAP51 has been proposed to interact directly with TIP47 (Ghosh et al. 2003). This tetrapeptide is not highly conserved in vertebrate CI-MPRs and is not present in the zebrafish receptor (Fig. 2b). Recycling of the CI-MPR to the TGN also involves interaction of the cytoplasmic tail with other proteins. PACS-1 binds to the acidic amino acids clustered at the carboxyl terminus of the CI-MPR (Ghosh et al. 2003), while a complex of proteins called retromer interacts with the CI-MPR at two non-overlapping regions (amino acids 48–80 and 80–100, respectively, of the human receptor) (Arighi et al. 2004). One portion of the retromer-binding determinant thus overlaps with the TIP47 recognition site. It is probably significant that this region [QENGH(I/V)(T/A)(T/A)KXV] is highly conserved across all the vertebrates examined, including zebrafish (Fig. 2b).

Mammalian MPRs are internalized from the plasma membrane into endosomes. Internalization of the CD-MPR is mediated by three separate sequences (Ghosh et al. 2003), all of which are present in the zebrafish receptor: a phenylalanine-containing sequence, a tyrosine-based motif and the carboxy-terminal region that contains the acidic-dileucine cluster (Fig. 2a). Internalization of the mammalian CI-MPR is mediated by the tetrapeptide YSKV, with the tyrosine and valine residues being most important (Jadot et al. 1992). The zebrafish CI-MPR has a similar motif (YSRV) in its cytoplasmic tail, with arginine a conservative substitution for lysine (Fig. 2b).

The IGF2 binding region of the CI-MPR is not highly conserved in vertebrates

The primary determinants of IGF2 binding by mammalian CI-MPRs are in repeat 11, in particular, within residues 1,508–1,575 (human receptor numbering) (Dahms and Hancock 2002). A phylogenetic study suggested that high-affinity IGF2 binding could be restricted to residues 1,530–1,579 (human numbering) because these residues are highly divergent in non-IGF2 binding receptors (Killian et al. 2000). The availability of the zebrafish sequence confirms the lack of conservation in vertebrates in the IGF2 binding region and emphasizes the similarities in the flanking residues (Fig. 3). Interestingly, however, the zebrafish receptor has an isoleucine at position 1,560, corresponding to Ile1572 of the human CI-MPR, which is directly involved in ligand binding (Brown et al. 2002). Even more intriguingly, Dennes et al. (2005) suggest that many of the residues that contribute to the IGF2 binding pocket of the mammalian CI-MPR, including the equivalent of Ile1572, are present in the Drosophila LERP. In view of these observations, experiments that address the ability of zebrafish CI-MPR to bind IGF2 are currently underway.

Fig. 3
figure 3

IGF2 binding status of zebrafish CI-MPR is unclear. ClustalW alignment of the region of the human CI-MPR that contains determinants for IGF2 binding (residues 1,530–1,579, between the filled circles), with the corresponding region of the indicated CI-MPRs. Human, mouse, bovine, kangaroo and opossum CI-MPRs bind IGF2, while platypus, echidna and chicken receptors do not. The arrowhead indicates Ile1572 of the human receptor (Ile1560, zebrafish receptor). (consensus line as in Fig. 1)

Full size image

Expression of zebrafish MPRs during embryonic development

Zebrafish should be an excellent model organism for in vivo experiments designed to understand the reason(s) why animals have two types of MPR (and perhaps for examining IGF2-independent functions of the CI-MPR). As the first step in this approach, we used in situ hybridization to determine the expression patterns of the zebrafish receptors during embryonic development. Although there were individual differences in staining intensities, expression patterns of both receptors were generally similar. Thus, mRNAs for CD-MPR and CI-MPR were detected in the one-cell embryo and in all blastomeres prior to the midblastula transition (Fig. 4a–d), indicating that maternal mRNAs for both receptors are deposited in the developing oocyte. Expression of both receptors was detected in all cells of blastula-stage embryos (Fig. 4e–f), throughout gastrulation (50% epiboly and tail-bud stages) (Fig. 4g–j) and in the midsegmentation period (12 somites, data not shown).

Fig. 4
figure 4

Expression of zebrafish MPRs during early embryonic development. In situ hybridization was performed using CD-MPR- or CI-MPR-specific antisense riboprobes. The embryonic stages used were one-cell zygote (a, b), cleavage period (c, d), blastula (e, f), gastrula (g–j) and segmentation period (k, l). Embryos are shown in lateral view, photographed at ×30 magnification

Full size image

In later-stage embryos, expression of both receptors ceased to be uniform and became progressively more restricted to the anterior region of the embryo. By late segmentation (18 HPF), staining was intense and even throughout the anterior region of the embryo but was much weaker along the trunk and tail (Fig. 4k–l). During the early pharyngula stage (24 HPF), staining became concentrated in the anterior regions of the embryo, including the telencephalon, retina, tectum, midbrain and hindbrain (Fig. 5a,b). Although expression was detected along the spinal cord and somites at this stage, it was less intense and more diffuse than the anterior expression. By 36, 48 and 72 HPF, expression of both receptors was restricted to the anterior region of the developing embryo, covering the forebrain, midbrain, hindbrain and retina (Fig. 5c–h).

Fig. 5
figure 5

Expression of zebrafish MPRs during late embryonic development. In situ hybridization was performed using CD-MPR- or CI-MPR-specific antisense riboprobes. Embryos are in the pharyngula period at 24 (a, b) and 36 HPF (c, d), and in the hatching period at 48 (e, f) and 72 HPF (g, h). Embryos are shown in lateral view, photographed at ×30 magnification

Full size image

In 48-HPF embryos, an unusual staining pattern was observed for the CD-MPR. This consisted of two intensely stained stripes in the brain region (Fig. 5e, inset). In an attempt to identify the stained regions, we performed double in situ staining using a probe specific for Shh (sonic hedgehog), which marks periventricular cells in the diencephalon at this stage (Krauss et al. 1993). The CD-MPR stain was dorsal to the Shh-stained region (data not shown). The identity of the CD-MPR-expressing regions is still unclear, but the larger stripe (stripe 2) may represent the pretectum, and the narrower stripe (stripe 1) is slightly caudal to the zona limitans intrathalamica. Further analysis with additional informative markers will be required to clarify this.

Potential conservation of MPR function in neuronal development

Endosomes and lysosomes regulate the activity of critical signaling molecules in development and may play a role in morphogen generation (Piddini and Vincent 2003). The widespread expression of both zebrafish MPRs in very early embryos, followed by more restricted expression in later stages, suggests time- and tissue-specific functions for both receptors, with particular indication of potential roles in neural development. MPR expression patterns in chickens and mice suggest that such roles may be conserved. Thus, the CD-MPR is expressed uniformly in very early chicken embryos, and this uniformity is lost at later stages, when expression becomes most prominent in neuronal tissues (Matzner et al. 1996). Expression of the CD-MPR is also developmentally regulated in rat brain (Romano et al. 2005). CI-MPR expression is prominent in developing neural tissue of chickens (Matzner et al. 1996). In early rodent embryos the CI-MPR is widely expressed and, although the heart, dorsal aorta and somites are the major sites of expression in post-implantation mouse embryos, CI-MPR expression is also readily detectable in the brain (Matzner et al. 1992; Lerchner and Barlow 1997). Zebrafish embryos are more accessible than chicken or mice embryos, and the ability to manipulate gene expression in zebrafish, using loss-of-function and gain-of-function approaches, makes it possible to test our hypothesis of a role for MPRs in neural development.

Conclusion

Zebrafish MPRs are structurally similar to the mammalian and chicken orthologs, indicating that the duplications of the common ancestral gene that are proposed to have given rise to extant CI-MPR and CD-MPR (Kornfeld 1992) must have occurred prior to the evolution of bony fish. Zebrafish MPRs possess all the residues that are important in mammals for M6P recognition and for intracellular trafficking pathways, suggesting that targeting of lysosomal enzymes by MPRs is an ancient pathway that was a feature of the common ancestor of modern vertebrates. Further studies in non-vertebrate chordates and other invertebrates should clarify the origin of these two genes and their functional relationship to Drosophila LERP, as well as the origin of the multifunctional CI-MPR. Our study establishes Danio rerio as a valuable model vertebrate for functional studies that will complement the use of more traditional, but genetically less tractable, mammalian species. The comparative approach should help to uncover the individual contributions of the two receptors to normal physiology and to pathological processes.