TrkB Variants with Deletions in the Leucine-rich Motifs of the Extracellular Domain*

We have isolated two novel variants involving the extracellular domain of TrkB from developing sensory neurons. These variants are generated by alternative splicing and lack two or all three of the leucine-rich motifs. Each of these variants is expressed as isoforms that possess or lack the intracellular tyrosine kinase domain. Fibroblast cell lines stably expressing these variants do not bind any of the TrkB ligands (brain-derived neurotrophic factor, neurotrophin-3, and neurotrophin-4/5) and neither survive nor undergo morphological transformation in response to neurotrophins. These results demonstrate that the leucine-rich motifs in TrkB are essential for ligand binding and signaling and indicate that the extracellular immunoglobulin-like domains alone are insufficient to confer neurotrophin binding to TrkB.

Trks are transmembrane glycoproteins that possess an intracellular region that contains the tyrosine kinase domain and an extracellular region with a complex subdomain organization (18) that is well conserved between Trk family members and in evolution. This region consists of a signal peptide, two cysteinerich domains, a cluster of three leucine-rich motifs (LRMs), and two immunoglobulin-like domains (see Fig. 2). Alternative splicing additionally generates receptor variants that differ from this basic plan in either their intracellular or extracellular regions.
TrkB and TrkC variants lacking the kinase domain (19 -22) are widely expressed by non-neuronal cells (19,(23)(24)(25)(26)(27)(28) and by some neurons (29,30). These noncatalytic receptors are thought to play a role in limiting the diffusion of their neurotrophin ligands (28), and there is evidence that kinase-deficient TrkB functions as a negative modulator of the BDNF survival response in developing neurons (30). Insertions within the kinase domain of TrkC have been described, and there is evidence that these insertions affect the signaling characteristics of the receptor following NT3 binding (21,22). Variants with in-frame deletions and insertions in the cytoplasmic juxtamembrane region of TrkB have also been identified (31), although the function of these variants is not known.
Trk variants with differences in extracellular domain are of particular interest as these may differ in their ligand binding characteristics. Variants with deletion of a short sequence between the transmembrane and second immunoglobulin-like domains have been described for all three Trks (31)(32)(33)(34)(35), and there is evidence that this short sequence plays an important role in ligand discrimination in TrkA and TrkB receptors (35,36). TrkB and TrkC variants lacking the first cysteine-rich domain have also been described (31,37), although the function of these variants is not known. The correspondence of each exon of chicken TrkB gene to one subdomain in the encoded protein (35) raises the possibility that alternative splicing generates additional natural Trk variants affecting the organization of the extracellular domains. Here we report two such isoforms of mouse TrkB with deletions of LRMs. Studies of these molecules ectopically expressed in fibroblasts show that the LRMs are essential for the interaction of BDNF, NT4/5 and NT3, with TrkB.

Cloning of Novel TrkB Variants That Differ in the Extracellular
Domain-Total RNA was extracted from E14 trigeminal ganglia using guanidinium isothiocyanate (38) and was purified using easiRNA (Nuncleon kit, Scotlab) after treatment with DNase I. cDNA was synthesized from 1 g of total RNA with 0.5 g of random hexamer primers using SuperScript reverse transcriptase according to the supplier's recommendations (Life Technologies, Inc.). The cDNA was amplified by PCR using primers that are specific for the extracellular domain of mouse TrkB (forward primer, 5Ј-TTCCGCTAGGATTTGGTGTAC-3Ј, nucleotides 628 -648 in the sequence deposited in GenBank under accession no. X17647; reverse primer, 5Ј-GAGCAGCCAGACGTGCA-GATG-3Ј, nucleotides 1101-1121). All amplifications were carried out in a volume of 50 l using Taq-polymerase in the supplied buffer (Promega); the primers were used at a final concentration of 1.5 M. PCR was carried for 35 cycles of 45 s at 95°C, 30 s at 62°C, and 60 s at 72°C. 20 l of these reactions were run out on agarose gels. Amplified products that differed from the expected size of 494 bp (39) (GenBank accession no. X17647) were recovered from agarose gel, cloned into the pGEM-T vector (Promega), and sequenced.
Reverse Transcriptase-PCR Detection of Specific TrkB Variants-To * This work was supported by a grant from the Wellcome Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The  1 The abbreviations used are: NGF, nerve growth factor; BDNF, brain-derived neurotrophic factor; NT3, neurotrophin-3; NT4/5, neurotrophin-4/5; LRM(s), leucine-rich motif(s); PCR, polymerase chain reaction; bp, base pair(s); DMEM, Dulbecco's modified Eagle's medium; MTT, 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyl tetrazolium bromide. determine if the L1 and L0 variants encode proteins that possess or lack the intracellular kinase domain, nested reverse transcriptase-PCR was carried out on newborn mouse brain polyadenylated RNA isolated using Poly(A)Tract mRNA isolation System (Promega). In the first PCR stage, single-stranded cDNA was amplified with forward primers specific for each variant and reverse primers specific for either the kinase domaincoding region or a COOH-terminal peptide coding region that is present only in the truncated isoform (39). PCR was carried out for 25 cycles of 45 s at 95°C, 45 s at 58°C, and 140 s at 72°C. The forward primers specific for the L0 and L1 variants spanned the upstream and downstream sequences flanking the deletions in these variants (Fig. 1) and only anneal under the PCR conditions employed to cDNAs coding for these variants. The L0-specific primer was 5Ј-GAACATCACGGAAAT-GATCC-3Ј, and the L1-specific primer was 5Ј-GGAGAACATCACG-GAAATAAAT-3Ј (Fig. 1). The kinase domain-specific primer for first stage PCR was 5Ј-TGCTCTGGGCAGAGGTTGT-3Ј (nucleotides 2181-2199 in the sequence deposited in GenBank under accession no. X17647), and the truncated peptide-specific primer for first stage PCR was 5Ј-CCTTTATCTCAGCTACCCATC-3Ј (nucleotides 1423-1443 in the sequence deposited in GenBank under accession no. M33385). 5-l aliquots from the first stage amplification reactions were used as a source of the DNA template for a second stage of PCR using the same L1-specific or L0-specific primer, and another primer from either kinase domain coding region or truncated peptide coding region was used for amplification of shorter fragments. The kinase domain-specific primer for the second stage PCR was 5Ј-AACTTTCCCGAAGGCTCCTT-3Ј (nucleotides 2145-2164 in the sequence deposited in GenBank under accession no. X17647), and the truncated peptide-specific primer for this stage was 5Ј-CCAGTGGGATCTTATGAAACA-3Ј (nucleotides 1403-1423 in the sequence deposited in GenBank under accession no. M33385). PCR was carried out for 25 cycles of 45 s at 95°C, 45 s at 57°C, and 140 s at 72°C.
Construction of Expression Plasmids-For expression of the fulllength TrkB that possesses a functional tyrosine kinase domain (designated L3-TK ϩ in this study) in fibroblast cell lines, full-length TrkB cDNA (gift of Rü diger Klein) was subcloned into the pMEX vector in which its expression is driven by the strong promoter of the long terminal repeat of Moloney murine sarcoma virus. To produce the L0-TK ϩ and L1-TK ϩ expression plasmids, the internal 475-bp MaeI-MaeII fragment of pL3-TK ϩ was replaced with the corresponding fragment from L0 and L1 variants cloned in the pGEM-T vector. These constructs were checked by sequencing before use.
DNA Transfections into NIH 3T3 Cells-NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% calf serum. 1 g of the expression plasmids (pL0-TK ϩ , pL1-TK ϩ , or pL3-TK ϩ ) and 10 ng of pSVneo were cotransfected into NIH 3T3 cells using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. Two weeks after transfection, G418-resistant clones were isolated by the cylinder cloning technique. TrkB expression in these clones was analyzed by immunoblotting, and at least two clones expressing each of the L0, L1, and L3 variants at high levels were used in binding and surviving assays.
Western Blotting-Western blotting was used to detect TrkB protein in lysates of NIH 3T3 cells transfected with expression plasmids for full-length TrkB and the novel TrkB variants. Confluent 35-mm Petri dish cultures of transfected NIH 3T3 cultures were lysed in 0.2 ml of sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer and were boiled for 5 min. The proteins were separated on 8% polyacrylamide gels and blotted onto Hybond-polyvinylidene difluoride membranes in Tris/glycine/methanol buffer according to the manufacturer's instructions (Amersham Corp.). After washing with phosphatebuffered saline, the membranes were blocked for 1 h at room temperature in 4% skimmed milk, phosphate-buffered saline, 0.05% Tween 20. The same buffer was used for incubations with primary and secondary antibodies and between washes. Either of two primary antibodies were used: anti-TrkB polyclonal antibody (Transduction Laboratories, diluted 1:250) and pan-Trk antibody diluted 1:2,000 (gift of David Kaplan). The secondary antibody was a horseradish peroxidase-linked donkey anti-rabbit IgG (Amersham, diluted 1:2,000). After washing the membranes three times in phosphate-buffered saline, 0.1% Tween-20, ECL detection was carried out according to the manufacturer's protocol (Amersham Corp.).
MTT Survival Assay-NIH 3T3 cell lines expressing L3-TK ϩ , L1-TK ϩ , or L0-TK ϩ TrkB receptor isoforms were plated at a density of 1,000 cells/well in 96-well plates in DMEM supplemented with 10% calf serum. After 24 h, the cultures were washed twice with serum-free DMEM and were cultured in DMEM supplemented with 0.5% calf serum in the absence or presence of various concentrations of neurotrophins. The medium was changed, and fresh factor was added every day. At the indicated times, the number of viable cells was assessed using the MTT assay according to the manufacturer's instructions (Promega). Colorimetry was performed using the MRX Microplate Reader (Dynatech).
Neurotrophin Binding Assay to TrkB-expressing Fibroblasts-Purified recombinant BDNF, NT3, and NT4/5 (gifts of John Winslow, Gene Burdon, and Arnon Rosenthal, Genentech Inc.) were radiolabeled (1 g of each) using 1 mCi of 125 I-Bolton-Hunter reagent (4,000 Ci/mmol; Amersham Corp.) according to the manufacturer's instructions. The labeled protein was separated from unbound reagent by gel filtration through a Sephadex G-50 column using 50 mM sodium phosphate, pH 7.5, containing 0.25% gelatin. Competition binding was carried out on cell suspensions as described previously (11,41). Briefly, cells were harvested using a rubber policeman and washed once with ice-cold binding medium (DMEM containing 2.5% calf serum and 10 mM HEPES, pH 7.0). 2 ϫ 10 5 cells were preincubated in 80 l of binding medium at 4°C for 1 h with or without unlabeled neurotrophin (10 nM for BDNF and NT4/5 and 100 nM for NT3). 20 l of ice-cold 125 I-labeled neurotrophins was added to each sample to reach a final concentration of 100 pM for BDNF and NT4/5 or 1 nM for NT3. After a 2-h incubation at 4°C, the cells were washed four times with ice-cold DMEM containing 0.1% bovine serum albumin, 10 mM HEPES, pH 7.0, and were lysed in 100 l of 1 M NaOH. The cell-associated radioactivity was determined on a G5500 ␥-counter.

Novel TrkB Variants with Deletions of LRMs-Transcripts
encoding two novel TrkB variants that differ in their extracellular domains were identified in the course of using reverse transcriptase-PCR to study the expression of TrkB mRNA in embryonic mouse trigeminal ganglia. RNA from E14 ganglia was reverse transcribed and amplified as described previously (30) using primers located in cysteine cluster I and the first immunoglobulin-like domain (Fig. 1). In addition to the expected 494-bp fragment based on the published sequence of TrkB (39), several additional minor fragments were obtained. The two most abundant of these fragments (347 and 275 bp) were cloned and sequenced. Densitometry of autoradiograms showed that the mRNA containing the 347-bp fragment comprises 3% of the TrkB transcripts in E14 trigeminal ganglia, and the mRNA containing the 275-bp fragment comprises 2% of the TrkB transcripts in E14 trigeminal ganglia. Sequence analysis revealed that each had internal, in-frame deletions. The encoded proteins lacked either the first two LRMs (L1) or all three LRMs (L0) in their extracellular domains (Figs. 1 and  2). In the L0 variant, there was an isoleucine to methionine substitution as the result of the splicing event and an additional 3-bp deletion 152 bp downstream of the main deletion (Fig. 1). This deletion leads to the absence of Gly 195 in cysteine cluster II.
Alternative splicing is known to generate TrkB variants that lack the intracellular, catalytic, tyrosine kinase domain (20). To determine if the L1 and L0 variants encode proteins that possess or lack the kinase domain, nested reverse transcriptase-PCR was carried out on polyadenylated RNA isolated from newborn mouse brain using forward primers specific for each of these variants and reverse primers specific for either the kinase domain coding region or the COOH-terminal peptide coding region that is present only in the truncated isoform. Fig. 3 shows that fragments of expected sizes were amplified in all four primer combinations, suggesting that mRNA for all four variants (L0-TK ϩ , L0-TK Ϫ , L1-TK ϩ , and L1-TK Ϫ ) is present in newborn mouse brain mRNA.
Ectopic Expression of TrkB Receptor Isoforms in NIH 3T3 Cells-To compare the neurotrophin binding and signaling characteristics of the L0 and L1 TrkB variants with full-length TrkB (designated L3-TK ϩ in this study), stable NIH 3T3 fibroblast cell lines expressing the tyrosine kinase domain containing forms of these variants (L0-TK ϩ and L1-TK ϩ ) were generated. Plasmids expressing these variants and the pSVneo plasmid were cotransfected into NIH 3T3 cells. After G418 selection, clones were checked for TrkB expression by Western blotting using two TrkB-specific antisera: one raised against the extracellular domain of TrkB (Transduction Laboratories) and a pan-Trk antibody (gift of Dr. David Kaplan) which rec-ognizes the cytoplasmic tyrosine kinase domain of all three Trk receptors. Clones that expressed a high level of TrkB protein (Fig. 4) were chosen for further experiments. The level of TrkB protein expressed in the L0-TK ϩ and L1-TK ϩ cell lines was similar to or greater than the amount of the wild type receptor in the L3-TK ϩ cell line.
To confirm that ectopically expressed TrkB receptor variants are present at the plasma membrane of 3T3 cells, binding of antibodies against the extracellular and cytoplasmic domains was performed on nonpermeabilized and detergent-permeabilized cells followed by detection of bound antibodies with 125 Iprotein A. Nonpermeabilized 3T3 cells expressing L0-TK ϩ , L1-TK ϩ , or L3-TK ϩ variants (but not control, neo-transformed cells) bound antibodies against the extracellular domain. In contrast, binding of antibodies against the cytoplasmic domain required cell permeabilization (data not shown). This suggests that the expressed receptor variants are targeted to and appropriately oriented at the cell membrane.
Ligand Binding Characteristics of Cell Lines Expressing TrkB Variants-Stable NIH 3T3 cell lines expressing the L0-TK ϩ , L1-TK ϩ , and L3-TK ϩ variants were used for competitive binding assays using 125 I-labeled neurotrophins as described previously (11,41). Fig. 5 illustrates the results of experiments using at least two independently generated 3T3 lines for each of the L0-TK ϩ , L1-TK ϩ , and L3-TK ϩ variants. In agreement with previously reported results (11), 3T3 cells expressing fulllength TrkB (L3-TK ϩ ) bind all three ligands, BDNF, NT3, and NT4/5. Excess corresponding unlabeled neurotrophin completely abolished binding. In contrast, neither the L0-TK ϩ nor the L1-TK ϩ variant was able to bind any of these three ligands.
Neurotrophin Survival Responses of TrkB-expressing Fibroblasts-Although we were not able to demonstrate neurotrophin binding to the L0-TK ϩ and L1-TK ϩ TrkB variants, the neurotrophin survival response of serum-depleted NIH 3T3 cell lines expressing these variants was used to investigate if neurotrophins could signal via these receptors at levels of binding which were undetectable in the above assays. In these experiments, the cell lines expressing TrkB variants were grown overnight in DMEM with 10% calf serum and switched to DMEM with 0.5% calf serum in the absence or presence of various concentrations of neurotrophins. The number of viable cells was then assessed at intervals using the MTT assay. Consistent with the binding experiments, BDNF, NT3, or NT4/5 did not enhance the survival of cell lines expressing the L0-TK ϩ and L1-TK ϩ TrkB variants. In contrast, cell lines expressing full-length TrkB (L3-TK ϩ ) showed a respond to all three neurotrophins (Fig. 5).
Morphological Transformation of Fibroblasts by Cotransfection of TrkB Variants and Neurotrophins-To investigate further potential interactions between neurotrophins and the L0-TK ϩ and L1-TK ϩ TrkB variants, NIH 3T3 fibroblasts were cotransfected with expression plasmids encoding the L0-TK ϩ , L1-TK ϩ , or L3-TK ϩ variant and expression plasmids encoding NGF, BDNF, NT3, or NT4/5 under the same promoter element. After 18 days, Giemsa staining was used to reveal transformation foci. In agreement with previous studies (16), full-length TrkB (L3-TK ϩ ) elicited transformation when coexpressed with BDNF, NT3, or NT4/5 (Fig. 6). The greatest number of transformation foci was seen in cells cotransfected with full-length TrkB and BDNF or NT4/5, a smaller number was seen in cells cotransfected with full-length TrkB and NT3, and no foci were seen with full-length TrkB and NGF. In contrast, the L0-TK ϩ and L1-TK ϩ TrkB variants were unable to transform NIH 3T3 cells either on their own or in cooperation with any of the neurotrophins (Fig. 6). DISCUSSION We have isolated two novel TrkB splice variants from the embryonic mouse trigeminal ganglion. Both variants encode receptors with deletions of the LRMs in the extracellular domain. In the L1 variant, the first and second LRMs are absent, and in the L0 variant, all three LRMs are missing (Figs. 1 and  2). Although information about the organization of the mouse TrkB genomic locus is not yet available, the recently published structure of the chicken TrkB locus (35) has allowed us to map the L1 and L0 deletions. The positions of the deletions in these variants directly correspond with the boundaries of exons 2, 3, and 4 in the chicken TrkB gene (Fig. 1). An additional feature of the L0 variant is a deletion of three nucleotides in the position corresponding to the junction between exons 5 and 6 of chicken TrkB. These findings indicate that the organization of the mouse and chicken TrkB genes is very similar and that the L1 and L0 variants are generated by alternative splicing. Using reverse transcriptase-PCR with upstream primers specific to the L1 and L0 variants and downstream primers specific to tyrosine kinase or truncated peptide coding regions, we have demonstrated the expression in newborn mouse brain of transcripts encoding each extracellular domain variant with or without a tyrosine kinase domain (Fig. 3).
We have shown that fibroblast cell lines expressing the L0 and L1 TrkB variants with intracellular kinase domains do not bind any neurotrophins and do not display any cellular responses to these ligands. However, cell lines expressing the full-length TrkB receptor (L3 variant, gp 145 ) bind the appropriate TrkB ligands (BDNF, NT4/5, and NT3, but not NGF) and survive and undergo morphological transformation in response to these ligands. These results clearly demonstrate that the LRMs in TrkB are essential for ligand binding and signal transduction. Because the L1 variant, which possesses the third LRM, does not bind neurotrophins, it is likely that this essential function resides within the first and second LRMs, or the integrity of the whole domain is necessary.
The relative importance of the different subdomains of the extracellular region of Trk receptor tyrosine kinases in mediating neurotrophin binding and discrimination is a highly controversial topic. The importance of the second LRM has been suggested from the results of binding studies using isolated LRMs. An immobilized peptide corresponding to the second LRM of TrkA specifically binds NGF, whereas a peptide corresponding to the second LRM of TrkB specifically binds BDNF, NT4/5, and NT3. The LRM2 peptide of TrkA specifically inhibits the binding of NGF to TrkA, whereas the LRM2 peptide of TrkB specifically inhibits the binding of BDNF, NT4/5, and NT3 to TrkB (42)(43)(44). Interestingly, neurotrophins fail to bind to immobilized peptides corresponding to the immunoglobulinlike domains in these assays (43,44). In contrast, studies of neurotrophin binding to chimeric Trk receptors suggest that the immunoglobulin-like domains contribute significantly to binding and ligand discrimination whereas the NH 2 -terminal sequences including the LRMs exert a minimal effect on binding. For example, NGF binds appropriately to a chimera consisting of the immunoglobulin-like domains from TrkA and the NH 2 -terminal sequences from TrkB but fails to bind to a chimera consisting of the immunoglobulin-like domains from TrkB and the NH 2 -terminal sequences from TrkA (45). Similarly, TrkC chimeras in which the second immunoglobulin-like domain is exchanged for the homologous sequences in TrkB or TrkA bind BDNF and NGF, respectively, with high affinity (46). Furthermore, in studies of TrkA and TrkC, the second immunoglobulin-like domain alone binds with similar affinity to neurotrophins as their respective full-length receptors, and deletion of this domain abolishes neurotrophin binding (46). An additional role for the second immunoglobulin-like domain of TrkB in ligand discrimination has come from the demonstration that a mutation (Cys 345 3 Ser 345 ) in this domain abolishes the ability of NT4/5 to transform morphologically 3T3 fibroblasts expressing TrkB, whereas it has no effect on the transforming activity of BDNF (16). A similar mutation in TrkA influences ligand (NGF) binding and converts a nontransforming receptor into an active Trk oncogene (47).
To make the story even more complicated, short deletions generated by alternative splicing in the region between the second immunoglobulin-like and transmembrane domains have been found in all three Trks (31,(33)(34)(35). Such deletions do not influence NGF binding to TrkA or BDNF binding to TrkB, but severely reduce binding of NT3 and NT4/5 (35), suggesting that regions downstream from the immunoglobulin-like domains are also involved in ligand discrimination.
It is possible that the controversy about the role of LRM and immunoglobulin-like domains in neurotrophin binding originates from the different experimental approaches used in these studies. The conclusion that neurotrophins bind to LRMs but not to immunoglobulin-like domains was based on experiments using synthetic LRM peptides and bacterially produced LRMs and immunoglobulin-like domains (42)(43)(44). In contrast, the conclusion that the immunoglobulin-like domains are important was based on experiments in which Trk receptors were produced by eukaryotic cells (35,40,(45)(46)(47). Because a number of post-translational modifications do not occur in bacterial expression systems, it is possible that their absence may account for the failure of neurotrophin binding to immunoglobulin-like domains produced in bacteria. However, our results clearly demonstrate in eukaryotic cells that LRMs are essential for ligands binding and that the immunoglobulin-like domains are not sufficient.
It is important to note that in all experiments demonstrating the importance of the immunoglobulin-like domains in ligand binding, chimeric (swap) receptors were generated (35,45,46). This left the general structure of a receptor molecule relatively intact and presumably able to form functional dimers, which is thought to be the key event in Trk receptor activation (48). Furthermore, direct binding of neurotrophins to eukaryotically expressed immunoglobulin-like domains was achieved when two separate immunoglobulin-like domains were brought together side by side by linking them to the Fc portion of a human antibody, generating an artificial dimer (46). Because LRMs are known to be involved in protein-protein interactions in a variety of proteins (49), it is possible that Trk dimerization brought about by the LRMs may be an early critical step in neurotrophin binding which facilitates specific interactions between neurotrophins and the immunoglobulin-like domains. The density of Trk receptors on neurons in vivo is much lower than on cells overexpressing recombinant Trk proteins in culture. Under these circumstances, specific ligand binding to LRMs could accelerate dimerization, resulting in the formation of binding sites on the immunoglobulin-like domains or in "handing over" the neurotrophin to the immunoglobulin-like domains as was proposed previously (43).
Increasing experimental data about neurotrophin-receptor interactions raise the possibility that it is a complex, multistage process. Further study of this process will lead to an understanding of neurotrophin signal transduction and the role of naturally occurring receptor variants.