Early evolutionary origin of the neurotrophin receptor family

Ronald E. van Kesteren, Michael Fainzilber, Garry Hauser, Jan van Minnen, Erno Vreugdenhil, August B. Smit, Carlos F. Ibáñez, Wijnand P.M. Geraerts, Andrew G.M. Bulloch

Author Affiliations

  1. Ronald E. van Kesteren*,1,
  2. Michael Fainzilber2,3,
  3. Garry Hauser4,
  4. Jan van Minnen1,
  5. Erno Vreugdenhil5,
  6. August B. Smit1,
  7. Carlos F. Ibáñez2,
  8. Wijnand P.M. Geraerts1 and
  9. Andrew G.M. Bulloch4
  1. 1 Graduate School of Neurosciences Amsterdam, Research Institute Neurosciences, Vrije Universiteit, De Boelelaan 1087, 1081, HV Amsterdam, The Netherlands
  2. 2 Division of Molecular Neurobiology, Department of Neuroscience, Karolinska Institute, Doktorsringen 12A, S 17177, Stockholm, Sweden
  3. 3 Present address: Department of Biological Chemistry, Weizmann Institute of Science, 76100, Rehovot, Israel
  4. 4 Neuroscience Research Group, Faculty of Medicine, University of Calgary, 3330 Hospital Drive NW, Calgary, AB, Canada, T2N 4N1
  5. 5 Leiden/Amsterdam Center for Drug Research, Division of Medical Pharmacology, PO Box 9503, 2300, RA Leiden, The Netherlands
  1. *Corresponding author. E-mail: revankes{at}
View Full Text


Neurotrophins and their Trk receptors play a crucial role in the development and maintenance of the vertebrate nervous system, but to date no component of this signalling system has been found in invertebrates. We describe a molluscan Trk receptor, designated Ltrk, from the snail Lymnaea stagnalis. The full‐length sequence of Ltrk reveals most of the characteristics typical of Trk receptors, including highly conserved transmembrane and intracellular tyrosine kinase domains, and a typical extracellular domain of leucine‐rich motifs flanked by cysteine clusters. In addition, Ltrk has a unique N‐terminal extension and lacks immunoglobulin‐like domains. Ltrk is expressed during development in a stage‐specific manner, and also in the adult, where its expression is confined to the central nervous system and its associated endocrine tissues. Ltrk has the highest sequence identity with the TrkC mammalian receptor and, when exogenously expressed in fibroblasts or COS cells, binds human NT‐3, but not NGF or BDNF, with an affinity of 2.5 nM. These findings support an early evolutionary origin of the Trk family as neuronal receptor tyrosine kinases and suggest that Trk signalling mechanisms may be highly conserved between vertebrates and invertebrates.


The survival and differentiation of vertebrate neurons, as well as various aspects of synaptic plasticity, are influenced by polypeptide neurotrophic factors, of which the best characterized are the neurotrophins, i.e. nerve growth factor (NGF), brain‐derived neurotrophic factor (BDNF), neurotrophin‐3 (NT‐3) and neurotrophin‐4/5 (NT‐4/5) (Korsching, 1993; Lindsay et al., 1994; Lewin and Barde, 1996). These factors interact with three structurally related receptor tyrosine kinases (RTKs), named Trk receptors (Meakin and Shooter, 1992; Barbacid, 1994). NGF specifically activates TrkA, whereas BDNF and NT‐4/5 specifically activate TrkB. NT‐3 primarily activates TrkC, but recognizes both TrkA and TrkB to a lesser extent. Trk receptors consist of an extracellular ligand‐binding domain, a single transmembrane region and an intracellular tyrosine kinase (TK) domain. Ligand binding to Trk receptors results in dimerization of receptor molecules followed by autophosphorylation of their cytoplasmic tyrosine residues (Jing et al., 1992). This initiates a cascade of signalling events that results in a cellular response (Schlessinger and Ullrich, 1992; Greene and Kaplan, 1995; Kaplan and Miller, 1997).

Over the past few decades, invertebrate nervous systems have provided useful models for diverse studies of neural development; however, the nature of neurotrophic mechanisms in invertebrates is poorly understood. The nervous systems of molluscs in particular offer advantageous model preparations for in‐depth studies of molecular and cellular mechanisms underlying neural development, regeneration and plasticity because (i) they contain large identifiable neurons whose functions in defined neuronal networks are well‐studied (reviewed in Geraerts et al., 1991; Weiss et al., 1992), and (ii) axotomized adult central neurons are able to regenerate and re‐establish functional synaptic contacts with identified target cells (reviewed in Bulloch and Syed, 1992). Significant time has been expended to identify neurotrophins and their receptors in invertebrates (see, for instance, Barde, 1994). Although these efforts have resulted in the identification of a number of candidate growth factors in molluscs (Smit et al., 1988; Fainzilber et al., 1996), and a series of neuronal RTKs in molluscs (Roovers et al., 1995; Jonas et al., 1996) and insects (Pulido et al., 1992; Wilson et al., 1993; Oishi et al., 1997), no neurotrophin‐like growth factor or Trk‐like RTK has been discovered to date in non‐vertebrate lineages.

In the freshwater snail Lymnaea stagnalis, a series of studies provided circumstantial evidence that an NGF‐like factor may be involved in neurite outgrowth (Ridgway et al., 1991), synapse formation (Syed et al., 1996), synaptic plasticity (Magoski and Bulloch, 1996) and the modulation of Ca2+ currents (Wildering et al., 1995). A protein purification approach that attempted to isolate putative neurotrophic factors in Lymnaea resulted in the cloning of a novel neurotrophic factor, designated CRNF, that is not sequence‐related to the vertebrate neurotrophins, although it is capable of binding to the shared mammalian p75 neurotrophin receptor (Fainzilber et al., 1996). This finding further sharpened the debate on the evolutionary origin of the neurotrophins and their receptors, and the nature of neurotrophic mechanisms in invertebrates. In this paper we report on the identification of Ltrk, an RTK from the central nervous system (CNS) of this snail that is most similar to TrkC, and specifically binds NT‐3, but not other neurotrophins. The expression profile of Ltrk suggests that it has a role during embryonic and sub‐adult development. Thus, Ltrk is the first Trk family member identified in an invertebrate.


Cloning of Ltrk

In order to identify Trk‐related receptors in L.stagnalis, we performed a homology‐based PCR screen on cDNA from freshly dissected CNS. Using primers TrkA1 and TrkS1, directed against conserved sequences in the TK domain of most RTKs, a single PCR product with the expected size of 200 bp was amplified. When cloned and sequenced, this product appeared to consist of multiple cDNA fragments, of which three encoded different putative RTKs. One of these had significantly higher sequence identity to Trk receptors than to other RTKs. Therefore, we tentatively named this clone Lymnaea Trk (Ltrk). The Ltrk PCR fragment was used to screen an organ‐cultured CNS‐specific λZAPII cDNA library for a full‐length clone. Ltrk clones were identified in three of the 29 library fractions, purified and rescued by in vivo excision. One plasmid, named pBSLtrk, has an insert of 2829 bp containing a single open reading frame that encodes a 794 amino acid protein (Figure 1; this sequence has been submitted to the DDBJ/EMBL/GenBank database under accession No. U61728). The open reading frame is preceded by several in‐frame stop codons, indicating that the coding region is complete at the 5′ end.

Figure 1.

cDNA sequence and deduced amino acid sequence of Ltrk. Nucleotide and amino acid positions are numbered at the right hand side. In‐frame stop codons within the 5′ untranslated leader sequence are underlined. The putative signal sequence cleavage site is indicated by an arrow. The transmembrane domain is shown in reverse contrast. The cysteine clusters and leucine‐rich motifs are underlined by solid and dotted lines, respectively, and the tyrosine kinase domain is boxed. The autophosphorylation sequence (D(L/I/V)Y(N)3YYR) is shown in bold, and the autophosphorylated tyrosine residue is indicated by a black dot (●). Triangles (▵) indicate putative N‐linked glycosylation sites (Asn‐X‐Ser/Thr) and * indicates the stop codon.

Structural characteristics of Ltrk

The predicted Ltrk protein contains a single hydrophobic region that serves as a putative membrane‐spanning domain, dividing the receptor into extracellular and cytoplasmic parts. A comparison of Ltrk with vertebrate Trk receptors and various RTKs from Drosophila reveals highest similarity with vertebrate Trk receptors (Figure 2A–D). In particular, Ltrk is most like TrkC (34% overall sequence identity, compared with 31% for TrkA and TrkB).

Figure 2.

Comparison of Ltrk with vertebrate Trk receptors and neuronal RTKs from Drosophila. (A) The structural organization of the various extracellular and intracellular domains of Ltrk, compared with vertebrate Trk, Dtrk and Dror and Dnrk. Although Ltrk shares an intracellular tyrosine kinase (TK) domain with all other receptors, it shares the typical arrangement of cysteine clusters (Cys‐clusters) and leucine‐rich (Leu‐rich) motifs in the extracellular domain with vertebrate Trk only. However, it lacks the immunoglobulin (IgG) domains and contains a unique N‐terminal extension. Dror, Dnrk and Dtrk, on the other hand, have different extracellular organizations, as indicated in the figure. (B) Amino acid sequence identities between the TK domains of various RTKs (i.e. amino acid residues 510–779 in Ltrk, and the corresponding parts of the other receptors). Amino acid identities of 55% and higher are printed in bold. (C) Amino acid sequence alignment of the intracellular parts of Ltrk, rat TrkA (Meakin et al., 1992), rat TrkB (Middlemas et al., 1991) and rat TrkC (Merlio et al., 1992). Note that sequence identity extends in front and beyond the TK domain (boxed), which is not the case for the Drosophila RTKs. Amino acid residues that are identical in Ltrk and either of the other receptors are shown in bold. (D) Amino acid sequence alignment of the cysteine clusters and the leucine‐rich motifs of Ltrk, TrkA, TrkB and TrkC. Conserved cysteine and leucine residues are indicated by (●) and (▵) respectively. (E) Amino acid sequence alignment of the second immunoglobulin domain of rat TrkC with the corresponding region in Ltrk. Note that, although functional immunoglobulin domains are lacking in Ltrk, sequence identity is retained in this part to some degree. (F) Radiolabelled in vitro translation products of the Ltrk cDNA separated on SDS–PAGE. Mr markers are indicated in kDa at the right hand side. Arrows indicate the two translation products of 89 and 78 kDa, respectively. (G) Phylogenetic tree based on a sequence alignment of the TK domains of various RTKs. The rat insulin receptor (RinsR) was used as an outgroup. Numbers at the forks are the result of bootstrap analysis, and represent confidence limits for grouping together the sequences on the right‐hand side of that fork.

The N‐terminal, extracellular domain has a putative signal sequence that is most likely cleaved after Ser32 (Von Heijne, 1983), and is further characterized by a tandem array of three leucine‐rich motifs (amino acids 158–228; all amino acid positions refer to the Ltrk sequence, unless otherwise stated) flanked by two cysteine clusters (amino acids 121–157 and 229–282), a specific combination of sequence motifs that is found exclusively in vertebrate Trk receptors. Interestingly, two immunoglobulin‐like domains that are found further downstream in the extracellular domains of all Trk receptors are lacking in Ltrk. In the corresponding region, Ltrk has some sequence identity with the C‐terminal half of the second immunoglobulin domain of TrkC (FNGQV), and the sequence immediately downstream of it (GQANQT) (Figure 2E), suggesting the previous existence of at least one immunoglobulin‐like domain at some earlier time during evolution. Another unique feature of the Ltrk extracellular domain is that the predicted translation product from the first in‐frame ATG in favourable Kozak consensus (Kozak, 1987) has an N‐terminal extension of ∼100 amino acids, when compared with other Trk sequences (Figure 2A). This 89 kDa product is the major one obtained in in vitro transcription/translation from an Ltrk plasmid (Figure 2F), demonstrating that the Met1 codon can be efficiently recognized by ribosomes as a translation initiation site. However, a minor product corresponding to a shorter protein (78 kDa) was also detected (Figure 2F), suggesting that Met105 can also be used to translate a protein that lacks the N‐terminal extension.

The cytoplasmic domain encompasses the TK domain of the receptor and contains a typical autophosphorylation site (DVYTTDYYR; amino acids 271–279) that characterizes most RTKs. Autophosphorylation of the second Tyr residue within this sequence (Tyr277) initiates intracellular signalling of mammalian Trk receptors. The cytoplasmic domain of Ltrk contains several amino acids and sequence motifs that are found in Trk receptors only. For example, Thr650, Trp691, Leu770 and Met772 are unique to Trk receptors, and replaced with Ala, Tyr, Pro and Phe, respectively, in most other RTKs. Also the sequences HGPDA (amino acids 599–603) and YIDIIA (amino acids 789–794; YLDILG in most Trk receptors) are Trk receptor‐specific. None of the above amino acids or sequence motifs are found at corresponding positions in non‐Trk RTKs, including Dtrk (Pulido et al., 1992), Dror (Wilson et al., 1993) and Dnrk (Oishi et al., 1997), with the exception of Thr650 in Dtrk and Dnrk, and Trp691 in Dnrk.

Phylogenetic relationships of Ltrk and related RTKs

From a sequence alignment of the TK domains of Ltrk, Dtrk (Pulido et al., 1992), Dror (Wilson et al., 1993), Dnrk (Oishi et al., 1997), rat TrkA, (Meakin et al., 1992), rat TrkB (Middlemas et al., 1991) and rat TrkC (Merlio et al., 1992) (amino acid residues 510–779 in Ltrk, and the corresponding parts of the other receptors), a phylogenetic tree was calculated using the Dayhoff PAM maximum likelihood method. This tree (Figure 2G) demonstrates that, in contrast to the Drosophila RTKs, Ltrk is clearly sequence‐related to the vertebrate Trk family of receptors. Bootstrap analysis shows that the assignment of Ltrk to the vertebrate Trk receptor clade is supported with 100% confidence. These findings strongly suggest that Ltrk and its vertebrate counterparts evolved from a common ancestral Trk‐like receptor.

Expression of Ltrk mRNA in developing and adult Lymnaea

To establish the expression profile of Ltrk in adult Lymnaea, we performed RNase protection assays (RPAs) of various tissues, as well as in situ hybridizations on sections of the CNS. RPA analysis revealed that the expression of Ltrk is confined to the CNS and its associated tissue (Figure 3A). In situ hybridizations revealed that Ltrk mRNA is primarily expressed in the centrally located, non‐neuronal, endocrine dorsal bodies, which are located on top of the cerebral ganglia (Figure 3B). In addition, various neurons express Ltrk mRNA, although at a lower level. These neurons are primarily located in the right parietal and visceral ganglia, and include the neuro‐endocrine yellow cells (Figure 3B and C).

Figure 3.

Ltrk expression profile in the adult snail. (A) RNase protection assay of various tissues from adult Lymnaea (shell height: 25 mm) using a 200 base pair Ltrk fragment as a probe. A 120 base pair fructose 1,6‐biphosphate aldolase fragment (Lald) was used as a positive control. Only in the CNS was the Ltrk fragment protected from degradation by RNase A, indicating the presence of Ltrk transcripts. (B–D) In situ hybridization of adult brain sections using full‐length antisense Ltrk cRNA as a probe. A strong hybridization signal is observed in the dorsal bodies (arrows in B), and a much weaker signal is present in neurons in the visceral and the right parietal ganglia (arrows in C), and in some neurons that are located within the intestinal nerve (arrow in D). This localization pattern is characteristic for the neuro‐endocrine yellow cells. LCG, left cerebral ganglion; VG, visceral ganglion; RPG, right parietal ganglion; IN, intestinal nerve.

In total Lymnaea embryonic tissue, high background interfered with the detection of Ltrk transcript using RPA. Therefore, RT–PCR analysis was performed on embryos from day 1 after deposition of the eggs, until day 13, prior to hatching of the snails, and on heads and central ganglia of sub‐adult snails. From adult animals, the central ganglia without the cerebral ganglia, the dorsal‐body‐containing cerebral ganglia, and the oesophagus were used as controls. RNA was isolated and checked for integrity prior to cDNA synthesis. After reverse transcription, Ltrk cDNA was amplified using primers against the second cysteine cluster and the transmembrane domain (Figure 4A). The quality of the cDNA was checked by PCR amplification of a Lymnaea fructose 1,6‐biphosphate aldolase cDNA fragment (Figure 4B). Expression of Ltrk was detected in day 1–3 embryos and in day 8–9 embryos, as well as in heads and central ganglia of sub‐adult snails. Because the central ganglia from animals 15 and 20 mm in size did not contain the dorsal bodies, the expression in sub‐adults is probably neuronal in origin. In adults, Ltrk mRNA expression in ganglia excluding the dorsal bodies could not be detected, whereas the dorsal‐body‐containing cerebral ganglia express Ltrk mRNA abundantly. No Ltrk mRNA was detected in the oesophagus. In all adult and developmental stages at which Ltrk expression was detected, only the expected 450 base pair‐product was amplified. This indicates that no alternative splice variants of the Ltrk mRNA are expressed that might encode a protein with functional immunoglobulin‐like domains, as this would significantly increase the size of the PCR product formed. Different primer combinations in various parts of the Ltrk cDNA repeatedly produced the same Ltrk expression pattern (data not shown).

Figure 4.

Ltrk and CRNF expression in the developing snail. (A) RT–PCR analysis using oligonucleotide primers against the second cysteine cluster and the transmembrane domain of Ltrk shows that there is a developmental stage‐specific expression of Ltrk. The Ltrk transcript is detected in day 1–3 embryos, day 8–9 embryos, and throughout sub‐adult development (i.e. in heads and brains of animals 5–20 mm in size). In the adult animal (25 mm), expression is only detected in the dorsal body‐containing cerebral ganglia. Note that only a PCR product with the expected size of 450 bp is generated, indicating that no alternative splice variants of the Ltrk transcript are expressed that might encode functional immunoglobulin‐like domains. (B) After reverse transcription of the RNA, oligonucleotide primers against fructose 1,6‐biphosphate aldolase generated a PCR product of the expected size in all tissues. PCR water controls are shown in the right hand lanes; size markers are indicated on the right hand side. (C) RNase protection assay showing CRNF expression patterns during development and in the adult. CRNF expression does not correspond with the expression pattern of Ltrk. Thus, it is unlikely that CRNF is the endogenous ligand for Ltrk. Note that, unlike Ltrk, CRNF expression peaks at the end of embryonic development (day 12), and is not detected during sub‐adult development. In the adult, CRNF is expressed only in the foot (see also Fainzilber et al., 1996), but Ltrk mRNA could not be detected in neurons that are known to innervate the foot (see Discussion).

Ltrk specifically binds neurotrophin‐3

In order to examine the ligand‐binding capabilities of Ltrk, the receptor was expressed by transient transfection of COS cells, as well as stable transfection of MG87 fibroblasts. Iodinated ligands for all three vertebrate Trks were tested for binding to Ltrk, and specific binding was observed only for NT‐3. Scatchard analyses of saturation curves for iodinated NT‐3 binding to the Ltrk fibroblast cell line revealed a Kd of 2.5 nM (Figure 5A). NT‐3 binding was not displaceable by NGF, BDNF or extracts of Lymnaea foot‐ or brain‐conditioned medium (CM), which are known to contain CRNF (Figure 5B). Chemical cross‐linking of 125I‐NT‐3 to cells transfected with Ltrk, followed by immunoprecipitation with Ltrk antibody, revealed a radioactive band of the expected Mr for an NT‐3–Ltrk cross‐linked complex (Figure 5C). Thus, Ltrk is capable of specifically binding NT‐3.

Figure 5.

Pharmacological characterization of Ltrk expressed in fibroblasts and COS cells. (A) Scatchard analyses of saturation curves for iodinated NT‐3 binding to the Ltrk fibroblast cell line. Ltrk binds NT‐3 with a Kd of 2.5 nM. (B) NT‐3 binding to Ltrk‐expressing fibroblasts was not displaceable by NGF, BDNF or protein extracts of Lymnaea foot or brain‐conditioned medium (CM) that are known to contain CRNF. (C) Chemical cross‐linking of 125I‐NT‐3 to fibroblasts (lanes 1–3) and COS cells (lanes 4–6) transfected with Ltrk, followed by immunoprecipitation with the anti‐Ltrk antibody, revealed a radioactive band of the expected Mr for an NT‐3–Ltrk cross‐linked complex (lanes 2 and 5; arrow). This band was not observed in mock‐transfected cells (lanes 1 and 4), or when binding was performed in the presence of excess cold NT‐3 (lanes 3 and 6). Mr markers are indicated in kDa at the right hand side.


Ltrk is an invertebrate Trk

A number of RTK families are widely conserved from Caenorhabditis elegans to mammals, most notably the insulin and ror receptor families (Wilson et al., 1993; Roovers et al., 1995; Jonas et al., 1996; Oishi et al., 1997). However, to date no true Trk family member has been described from an invertebrate. Some Drosophila RTKs (Dtrk, Dnrk and Dror), although Trk‐related in their intracellular sequences (Figure 2B), have very different extracellular domains (Figure 2A), and thus are not considered to be Trk receptors. In this paper, we set out to identify true Trk receptors from the snail, L.stagnalis.

Using a PCR amplification strategy, we identified several TK sequences from the Lymnaea CNS. One of the amplified fragments, named Ltrk, was characterized further. Many structural features uniquely identify Ltrk as a Trk receptor homologue (e.g. the extracellular cysteine clusters and leucine‐rich motifs and the high sequence identity with vertebrate Trk receptors in the TK domain; see Results section). Moreover, the conservation of several tyrosine residues that are potential targets for autophosphorylation indicates that activation of Ltrk may initiate similar intracellular signal transduction events as those known for vertebrate Trk receptors (Greene and Kaplan, 1995; Kaplan and Miller, 1997). The sequence NPXY (residues 480–483), for instance, is involved in a Ras‐dependent signalling cascade in vertebrates that involves binding of the Shc protein (Van Der Geer et al., 1995). Also, the phospholipase Cγ1 binding site YLD(I/V) (Songyang et al., 1993) is well‐conserved in Ltrk (residues 789–792). Shc and phospholipase Cγ1 are co‐operatively involved in neurite outgrowth of neuronal precursor cells (Stephens et al., 1994).

In contrast to vertebrate Trks, the extracellular part of Ltrk has an additional N‐terminal domain of unknown function, and lacks the immunoglobulin domains. Two different sized bands of Ltrk are observed upon in vitro translation, although the larger product is far more abundant. Differential usage of translation initiation sites is common in insects (Markussen et al., 1995), and may be used here to generate an N‐terminally truncated receptor isoform with biologically distinct functions. At present, we cannot test this possibility, nor do we know if truncated receptors are expressed in the Lymnaea brain. In the region that corresponds to the immunoglobulin‐like domains of vertebrate Trk receptors, Ltrk retains some sequence identity with the C‐terminal half of the second immunoglobulin domain (cf. Figure 2D), suggesting the earlier presence of at least one ancestral immunoglobulin‐like domain in Ltrk. It is unlikely that the lack of immunoglobulin‐like domains in Ltrk is due to alternative splicing events generating different Ltrk transcripts, because oligonucleotide primers spanning the immunoglobulin region of the mRNA, generated only the expected PCR product of 450 bp, both in developing and adult snails (cf. Figure 4C). The presence of an exon encoding functional immunoglobulin‐like domains would have resulted in the formation of a larger PCR product.

Ltrk is expressed in developing and in adult animals

Our RPA experiments indicate that in the adult animal, Ltrk mRNA is exclusively expressed in the brain. Within the brain, Ltrk mRNA is present in neurons, but it is most abundant in the dorsal bodies. The dorsal bodies are non‐neuronal, endocrine organs that synthesize and release female gonadotropic hormones that are involved in vitellogenesis and differentiation of the female sex organs (Geraerts and Algera, 1975; Geraerts and Joosse, 1975). In addition to their neuronal expression, vertebrate Trk receptors are also expressed in endocrine tissues (reviewed in Barbacid, 1994). Their functions within these tissues remain unresolved. Neuronal expression of Ltrk is only detectable in the neuro‐endocrine yellow cells. The yellow cells are located primarily in the visceral and parietal ganglia (Boer et al., 1994), and are involved in the control of salt metabolism (De With et al., 1993; Smit et al., 1993).

In order to test whether Ltrk might have a function in development, we performed a RT–PCR analysis of embryos and sub‐adult tissues. From this, there is clear evidence of regulated expression of Ltrk during embryonic and sub‐adult development. Expression levels peak during early (days 1–3) and mid‐ (days 8–9) embryonic development, and throughout sub‐adult development (in animals 5–20 mm in size). Post‐natal expression is probably neuronal in origin, because in animals 15 and 20 mm in size, it was detected in parts of the brain that exclude the dorsal bodies. It is known that the CNS undergoes dramatic developmental changes around this time (Croll and Chiasson, 1989). The entire neuronal network that is responsible for the initiation and execution of male reproductive activities, for instance, is formed when snails are 10–15 mm in size (Smit et al., 1992). Thus, it is likely that Ltrk has important functions during specific stages of the development of the CNS; however, functional roles in subsets of neurons in the adult brain are also possible.

Ltrk has binding capacity for neurotrophin‐3

Ltrk exogenously expressed in fibroblasts or COS cells is capable of binding NT‐3, but not NGF or BDNF, nor could Lymnaea extracts containing CRNF displace bound NT‐3 (Figure 5). The affinity of NT‐3 for Ltrk was measured as Kd = 2.5 nM, which is comparable with the previously reported low‐affinity component of NT‐3 binding to its cognate receptor TrkC (Kd = 4 nM; Lamballe et al., 1991). In addition, Lamballe et al. (1991) reported a high‐affinity binding of NT‐3 to TrkC with a Kd of 26 pM. A number of recent reports have implicated the immunoglobulin‐like domains in determining specificity and high‐affinity binding of neurotrophins to their cognate Trk receptors (Pérez et al., 1995; Urfer et al., 1995; MacDonald and Meakin, 1996; Holden et al., 1997). However, others have shown a contribution of the leucine‐rich motifs to low‐affinity (nM range) neurotrophin binding to Trks (Windisch et al., 1995a,b,c; MacDonald and Meakin, 1996). Finally, Ninkina et al. (1997) most recently described splice variants of TrkB which lacked the leucine‐rich motifs and were incapable of binding or responding to TrkB ligands. Considering the apparent lack of complete immunoglobulin‐like domains, it is likely that in Ltrk the leucine‐rich motifs are responsible for NT‐3 binding.

Despite the nM range affinity of NT‐3 for Ltrk, no functional responsiveness (Erk kinase phosphorylation or increased survival) could be measured in Ltrk‐expressing fibroblasts upon NT‐3 application, and Ltrk phosphorylation was not observed (data not shown). It is possible that NT‐3 binds to Ltrk in a conformation that does not favour or enable receptor dimerization. Thus, although Ltrk is pharmacologically TrkC‐like in its neurotrophin‐binding specificity, a signalling response by this receptor probably requires a cognate ligand of as yet unknown nature.

Trk ligands in Lymnaea

The data presented above firmly establish Ltrk as a member of the Trk family of RTKs. The identity of the endogenous ligand(s) for Ltrk is clearly a topic of much interest. On the one hand, NT‐3 binding to Ltrk (this paper), and NGF effects on Lymnaea neurons (Wildering et al., 1995; Magoski and Bulloch, 1996; Syed et al., 1996), provide some indications that endogenous Trk ligands might be neurotrophin‐related. On the other hand, the recently described CRNF from Lymnaea, identified on the basis of p75 binding, proved to be completely sequence‐unrelated to neurotrophins (Fainzilber et al., 1996). The unique N‐terminal domain of Ltrk might serve as a high‐affinity binding domain for a non‐neurotrophin ligand. CRNF, however, is not likely to be the endogenous ligand for Ltrk, since CRNF and Ltrk expression are not complementary either during development or in the adult (Figure 4).

The identification of a molluscan Trk homologue suggests an early evolutionary origin for the Trk family of neurotrophin receptors. Phosphotyrosine signalling has been suggested as one of the critical innovations in the evolution of multicellular organisms (Darnell Jr, 1997). Our data suggest that Trk signalling might be a requirement for the evolution of complex and flexible nervous systems. In this light, it is intriguing that the C.elegans genome sequence (completed for ∼75%; has not yet revealed Trk or neurotrophin homologues, suggesting that the simple nematode nervous system, with its predetermined cell number and connectivity pattern, does not require Trk signalling for its proper development and maintenance. Identification of the cognate Ltrk ligand in Lymnaea, and further elucidation of neurotrophic mechanisms in invertebrates, should shed light on the contribution of this signalling system to the evolution of complex brain structures.

Materials and methods

PCR amplification of Trk‐related sequences

Total RNA was isolated from acutely dissected Lymnaea CNS using the guanidine isothiocyanate method (Chomczynski and Sacchi, 1987). Poly(A)+ RNA was extracted either by oligo‐d(T) cellulose chromatography (Maniatis et al., 1982) or using the Dynabeads mRNA purification kit (Dynal), and reverse transcribed into oligo‐(dT17)‐primed cDNA using 200 units M‐MLV reverse transcriptase (Gibco‐BRL). Degenerate oligonucleotides were designed, based on the conserved sequences (V/I)HRDLA and DVW(A/S)(Y/F)G in the TK domain of most RTKs [TrkS1: 5′‐ATCGGATCCGTICA(CT)(AC)GIGA(CT)CTIGC‐3′ and TrkA1: 5′‐GCTGGAATTCC(AG)(AT)AIG(TA)CCAIAC(GA)TC‐3′; I: inosine]. The 5′ end the primers contained a recognition sequence for the restriction endonucleases BamHI and EcoRI, respectively. PCR was performed in a 50 μl solution containing 1 animal equivalent of cDNA, 200 μM of each of the four deoxynucleotides, 25 pmol of each of the primers TrkS1 and TrkA1, 20 mM Tris–HCl (pH 8.4), 50 mM KCl, 3 mM MgCl2 and 1 unit of Taq DNA polymerase (Gibco‐BRL). The PCR mixture was overlaid with 70 μl of mineral oil and incubated in a DNA thermal cycler (Perkin–Elmer Cetus) for 35 cycles of 30 s at 94°C, 2 min at 37°C and 1.5 min at 72°C. Of this PCR mixture, 1 μl was re‐amplified under the same conditions. Amplified cDNA fragments were digested with EcoRI and BamHI, separated on agarose gel, and fragments of the expected size of 200 bp were isolated, cloned into pBluescript SK (Stratagene) and sequenced using an ABI 373A automated sequencer and the ABI PRISM Dye Terminator Cycle Sequencing kit (Perkin–Elmer Cetus). Databases were searched for similar sequences using the BLAST program (Altschul et al., 1990).

Construction and screening of a cDNA library from organ cultured CNS

Dissected Lymnaea CNS were washed 4× for 10 min in antibiotic saline [51.3 mM NaCl, 1.7 mM KCl, 4.1 mM CaCl2, 1.5 mM MgCl2, 5 mM HEPES (pH 7.9) and 150 μg/ml gentamicin (Sigma)] and cultured for 3 days in defined medium [serum‐free 50% Liebowitz L‐15 medium (Gibco‐BRL) containing 40.0 mM NaCl, 1.7 mM KCl, 4.1 mM CaCl2, 1.5 mM MgCl2, 10.0 mM HEPES (pH 8.1), 150 μg/ml l‐glutamine, 54 μg/ml d‐glucose and 25 μg/ml gentamicin] (Ridgway et al., 1991). Cultured CNS were used to construct a λZAPII cDNA library according to the manufacturer's instructions (Stratagene). In short, 5 μg of mRNA was used to synthesize double‐stranded cDNA. After ligation of the linkers, cDNAs were size separated and those <1000 bp were discarded. Larger cDNAs were ligated into the vector arms and packaged, resulting in a primary library of 106 cDNA clones. The primary library was divided into 29 fractions which were amplified individually.

Library fractions containing cDNA clones corresponding to one of the Trk‐related PCR fragments were identified by PCR. Of the positive fractions, ∼100 000 clones were plated and absorbed to replica Hybond N filters (Amersham). After prehybridization for 16 h, the filters were hybridized at 65°C for 60 h to the 200 bp PCR fragment labelled with [α32P]dCTP using the random priming kit from Boehringer Mannheim. The (pre)hybridization solution contained 6×SSC (1×SSC = 150 mM NaCl, 15 mM sodium citrate; pH 7.4), 5×Denhardt's, 0.1% SDS and 100 μg/ml salmon sperm DNA. The filters were washed twice in 1×SSC/0.1% SDS at 65°C for 15 min and autoradiographed. Positive clones were isolated, replated at a lower plaque density and screened again. Of pure clones, the insert‐containing pBluescript phagemid (pBSLtrk) was rescued by in vivo excision and sequenced on both strands using vector‐based primers and internal primers.

Phylogenetic analysis

The TK domains of Ltrk, Dtrk, Dror, Dnrk, rat TrkA, rat TrkB and rat TrkC (amino acid residues 510–779 in Ltrk, and the corresponding parts of the other receptors) were aligned using the program Clustal V (Higgins et al., 1992). Phylogenetic analysis was performed using the PHYLIP software package (Felsenstein, 1995). A protein distance matrix was computed using the Dayhoff PAM maximum likelihood method. From this, phylogenies were estimated using the Fitch–Margoliash criterion. The input order of the sequences was randomized 10× using the jumble option, and the best tree was used. Bootstrap resampling was performed on the original data set in order to generate 100 data sets, from which confidence limits for the individual branches in the tree were calculated.

In vitro translation

In vitro transcription and translation were performed on Ltrk cDNA subcloned in the pCDNA3 mammalian expression vector, using the T7‐Coupled TNT reticulocyte lysate system (Promega), according to the manufacturer's protocols.

RNase protection assays

Ribonuclease protection assays (RPAs) were carried out on total RNA from various Lymnaea tissues, extracted as described above. For Ltrk, 32P‐labelled antisense RNA probes were generated by transcription from a 200 bp linearized subclone of Ltrk, corresponding to base pairs 2261–2460 of the cDNA sequence. CRNF probes were generated as described previously (Fainzilber et al., 1996). All reagents were from the RPA II kit (Ambion), and used according to the manufacturer's instructions.


Using the guanidine isothiocyanate method (Chomczynski and Sacchi, 1987), total RNA was isolated from 1–13‐day‐old Lymnaea embryos (∼200 embryos per isolation), from heads of animals 5 and 10 mm in size, from the central ganglia ring (without the cerebral ganglia and the dorsal bodies) of animals 15 and 20 mm in size and adults, and from the cerebral ganglia (including the dorsal bodies) of adults only. The oesophagus of adult animals was used as a negative control. Of all post‐embryonic tissues, 3 animal equivalents were used per isolation. RNA was reverse transcribed into oligo‐(dT17)‐primed cDNA using 200 units M‐MLV reverse transcriptase (Gibco‐BRL). PCR primers were designed against the second cysteine cluster (LtrkCys: 5′‐TTCAAGTGCGAACCATGTGG‐3′; sense) and against the transmembrane domain (LtrkTM: 5′‐CAACTGGCAGTATGACCTGC‐3′; antisense). As a positive control for the cDNA synthesis, Lymnaea fructose 1,6‐biphosphate aldolase cDNA (DDBJ/EMBL/Genbank accession No. U73114) was amplified using primers Lald‐6 (5′‐GCTGGTCAAGGATGCCCC‐3′; sense) and Lald‐4 (5′‐TAGCTTGTAGAGCTCGGCCAT‐3′; antisense). PCR reactions were performed as described. Forty cycles of 15 s at 94°C, 15 s at 58°C and 1.5 min at 72°C were performed. Of the PCR mixtures, 20 μl was separated on a 1.5% agarose gel and photographed.

In situ hybridization

Radiolabelled full‐length run‐off antisense RNA was synthesized from 5′‐end linearized plasmid pBSLtrk using 20 units T3 RNA polymerase and 60 μCi 35S‐UTP (NEN Dupont). Lymnaea brains were fixed at 4°C for 16 h in phosphate‐buffered saline (PBS) containing 1% paraformaldehyde and 1% acetic acid, and embedded in paraffin. Sections of 7 μm were collected on 0.5% gelatin/0.5% chrome‐alum coated microscope slides. Before hybridization, slides were pre‐treated with 0.005% pepsin in 0.2 N HCl at 37°C for 7 min, with 4% paraformaldehyde in PBS at 20°C for 20 min, and with 1% hydroxyl‐ammonium chloride in PBS at 20°C for 15 min. Slides were then washed in PBS at 20°C for 5 min, and dehydrated. In situ hybridization was carried out as described previously (Tensen et al., 1991), with the exception that the hybridization buffer contained 60% formamide and 106 c.p.m. of the cRNA probe per 100 μl hybridization buffer. Control slides were incubated with a sense RNA probe. After 16 h of hybridization at 50°C, slides were washed twice in 2×SSC at 20°C for 5 min, twice in 2×SSC/50% formamide at 50°C for 15 min, once in 2×SSC containing 0.1 mg/ml RNase A at 37°C for 25 min, three times in 2×SSC/50% formamide at 50°C for 15 min, and twice in 2×SSC at 20°C for 5 min. Slides were dehydrated, air dried and autoradiographed for 6 weeks in Kodak NTB 2 emulsion.

Antibody production

A synthetic peptide corresponding to the C‐terminal 15 amino acids of Ltrk (REEVSGDPVYIDIIA) was coupled to ovalbumin (Sigma) at a molar ratio of 10:1. After mixing with Freund's complete adjuvant (Sigma), conjugate (0.5 mg) was used to immunize rabbits. Boosters were carried out once every 2 weeks, using 0.3 mg conjugate mixed with Freund's incomplete adjuvant (Sigma). All positive bleeds were purified over a peptide affinity column obtained by immobilization of the peptide on CnBr‐activated Sepharose.

Generation of Ltrk cell line

The Ltrk cDNA was subcloned into pCDNA3 (Invitrogen) and introduced into MG87‐NIH 3T3 fibroblasts by lipofection. Transfected cells were selected with the neomycin analogue G‐418, isolated, amplified and subsequently analysed for Ltrk expression by RPA. The line MG87‐Ltrk/6 expressed Ltrk mRNA at levels comparable with those in Lymnaea CNS, and was used in subsequent experiments. Cells were cultured in Dulbecco's modified Eagle’s medium with 10% fetal calf serum (FCS), 1 mg/ml gentamicin and 100 μg/ml G‐418.

Neurotrophin binding and cross‐linking

Neurotrophins were iodinated by the lactoperoxidase method, and competitive binding assays were performed as described previously (Ryden et al., 1997). Briefly, hot and cold neurotrophins were diluted in binding buffer as required (phosphate buffer; pH 6.5, containing 2% BSA and 0.7 mM CaCl2) and added to the cells (2×106 cells/ml), in a final volume of 100 μl per replicate. Each assay point was in triplicate, and the experiments were set up in 96‐well filter plates. The plates were then incubated for 90 min at 4°C with vigorous shaking, followed by rapid filtration on a suction device. A multichannel pipette and suction were used to wash the filters once each with ice cold binding buffer, before counting. Affinity labelling and chemical cross‐linking of cell monolayers with NT‐3 was carried out in plates 10 cm in diameter. Cells were incubated with 3 ml binding buffer containing 2.5 nM 125I‐NT‐3. Control plates included also a 100‐fold excess of cold NT‐3. Binding was performed with gentle shaking for 90 min at 4°C, followed by the addition of the cross‐linking agents bis[sulfosuccinimidyl] suberate and disuccinimidyl suberate (Pierce), as recommended by the manufacturer. After 30 min incubation with the cross‐linkers, the monolayers were aspirated, and cells taken up in ice cold lysis buffer [1% NP‐40 (ICN Biomedicals), 20 mM Tris pH 8, 137 mM NaCl, 10% glycerol, 2 mM EDTA, 1 mM PMSF]. After a brief spin to remove any debris, the supernatants were processed for immunoprecipitation and SDS–PAGE or direct SDS–PAGE as described previously (Ryden et al., 1994).

Foot tissue was homogenized in 2 vol Lymnaea isotonic saline, and centrifuged for 10 min at 14 000 g. The supernatant was then loaded onto Supelclean C‐8 cartridges (Supelco Inc., Bellefonte), and peptides and small proteins were eluted with stepwise acetonitril/water washes (up to 80% ACN). All washes were combined and lyophilized. Brain‐conditioned medium (CM) was prepared as described previously (Ridgway et al., 1991). Twenty animal equivalents of foot extract and CM were used in displacement studies.


We thank Ms E.R.Van Kesteren and Ms A.‐S.Nilsson for excellent technical assistance. This work was supported by the Medical Research Council (Canada), the Neuroscience Network (Canadian National Centres of Excellence), the European Neuroscience Program (European Science Foundation), the Human Frontiers Science Program Organization (HFSPO; grant RG0045/1997B), the Netherlands Organization for Scientific Research (NWO) and the Alberta Heritage Foundation for Medical Research (AHFMR). R.E.van Kesteren is a Royal Dutch Academy of Sciences (KNAW) fellow. M.Fainzilber was an HFSPO long‐term fellow. A.G.M.Bulloch is an AHFMR scientist.


View Abstract