Activation Loop Tyrosines Contribute Varying Roles to Trkb Autophosphorylation and Signal Transduction

Home , Tyrosine phosphorylation

Oncogene (1998) 16, 1691 ± 1700  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Activation loop tyrosines contribute varying roles to TrkB autophosphorylation and signal transduction

Joseph H McCarty and Stuart C Feinstein

Department of Molecular, Cellular and Developmental Biology and Neuroscience Research Institute, University of California, Santa Barbara, California 93106, USA

The TrkB receptor tyrosine kinase (RTK) is a high and synaptic eciency (for reviews, see Davies, 1994; anity receptor for the neurotrophins brain derived Snider, 1994). neurotrophic factor (BDNF) and neurotrophin-4/5 (NT- Neurotrophin action is mediated via high anity 4/5). Following exposure to BDNF or NT-4/5, TrkB is binding with the Trk family of transmembrane autophosphorylated on ®ve cytoplasmic tyrosines: Y484, receptors (for review, see Kaplan and Stephens, Y670, Y674, Y675, and Y785. Based on crystallographic 1994). TrkA receptors bind NGF, TrkB receptors analyses for others RTKs, TrkB tyrosines Y670, Y674, bind BDNF and NT-4/5 (and to a lesser extent NT-3) and Y675 are expected to lie within a putative kinase and TrkC binds NT-3. TrkA, TrkB and TrkC are activation loop. Phosphorylation of these activation loop encoded by separate genes and are each synthesized in tyrosines is postulated to be a conserved event required multiple isoforms as a result of alternative RNA for complete RTK activation. Here, we have assessed the splicing. In each case, there is at least one isoform importance these activation loop tyrosines play in that is a receptor tyrosine kinase (RTK) capable of regulating TrkB autophosphorylation, cytoplasmic signal mediating ligand-dependent signal transduction. In the transduction, and cell proliferation. We show that while case of TrkB and TrkC, there are also multiple tyrosine 670 is dispensable for BDNF-inducible TrkB isoforms lacking the intracellular tyrosine kinase autophosphorylation and the activation of certain signal domain (Barbacid, 1994). transduction events, it is required for complete TrkB- Based upon the many actions of the neurotrophins mediated cellular proliferation. Combinatorial mutagen- and their potential utility as therapeutic agents for esis of tyrosines 674 and 675 only moderately aects neurodegenerative conditions (Hefti, 1994), there has TrkB autophosphorylation, but signi®cantly impairs the been substantial interest in the molecular basis of Trk BDNF-inducible stimulation of cytoplasmic signaling receptor action. Most of this eort has focused upon events and cellular proliferation. The combined mutation the structure and function of receptor tyrosine kinase of all three activation loop tyrosines results in an inactive isoforms. In the case of the TrkB-RTK, herein receptor, which is unable to autophosphorylate, stimulate designated as simply TrkB, ligand-induced activation signaling events, or induce mitogenesis. The data high- occurs through a conserved mechanism used by most light the varying degrees of importance of the three RTKs (Schlessinger and Ullrich, 1992). Ligand binding activation loop tyrosines in TrkB mediated biological to the TrkB extracellular region promotes receptor responses. dimerization, which activates the intrinsic tyrosine kinase domain, leading to autophosphorylation on Keywords: Shc; phospholipase C; Erk; proliferation; speci®c receptor tyrosines. In addition to activating receptor tyrosine kinase the kinase domain, autophosphorylation serves to recruit various cytoplasmic molecules to the activated receptor complex. In many cases, these molecules contain src homology 2 (SH2) domains (Pawson, Introduction 1995) or phosphotyrosine binding (PTB) domains (Borg and Margolis, 1995), both of which bind to The proper development and maintenance of the receptor phosphotyrosine-containing sequences. The vertebrate nervous system requires the actions of the quantitative and qualitative nature of these interac- nerve growth factor (NGF) family of neurotrophic tions subsequently activate multiple downstream factors, which includes NGF, brain derived neuro- cytoplasmic signaling cascades leading to a cellular trophic factor (BDNF), neurotrophin-3 (NT-3), neuro- response. trophin-4/5 (NT-4/5) and neurotrophin-6 (NT-6). TrkB can elicit dierent cellular responses depending These neurotrophins were originally viewed solely as on the cellular context in which it is expressed. For target derived trophic factors, regulating naturally example, PC12 cells, which do not normally express occurring neuronal cell death during development and TrkB, dierentiate to a neuronal phenotype in response providing neuronal maintenance in maturity (for to BDNF following exogenous TrkB expression review, see Purves, 1988). However, more recent work (Squinto et al., 1991). Alternatively, exogenous has shown that neurotrophins also act during expression of TrkB in NIH3T3 ®broblasts, followed neurogenesis, neuronal dierentiation, path®nding by BDNF stimulation, produces a potent mitogenic eect (Glass et al., 1991; Klein et al., 1991). Decipher- ing the signaling pathways used by TrkB to elicit these Correspondence: SC Feinstein diering cellular responses is crucial to understanding Received 25 March 1997; revised 31 October 1997; accepted 31 the normal role of the Trks in dierentiation and October 1997 proliferation of cells within the nervous system. TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1692 Furthermore, de®ning the molecular events regulating TrkB-mediated mitogenesis may help in understanding a the potential oncogenic roles of the Trk receptors (Martin-Zanca et al., 1986; Nakagaware et al., 1994). To date, ®ve BDNF-inducible autophosphorylation sites have been identi®ed in the 368 amino acid intracellular region of TrkB (Guiton et al., 1994; Middlemas et al., 1994). Tyrosine 484 (Y484) lies within the juxtamembrane region of TrkB. Based on its neighboring N-terminal amino acid sequence (NPQY), Y484 is predicted to be the binding site for the PTB domain of the SHC adapter protein. Indeed, the corresponding phosphorylated tyrosine (pY490) in TrkA is responsible for the NGF-inducible association of TrkA with SHC (Obermeier et al., 1993b; Stephens et al., 1994; Dikic et al., 1995). Tyrosine 785 is in the b extreme C-terminal region of TrkB. Based on mutational analysis of the corresponding TrkA tyrosine (also pY785), this phosphotyrosine is likely to be the binding site for the SH2-containing enzyme PLCg (Obermeier et al., 1993a; Loeb et al., 1994). This interaction is believed to induce tyrosine phosphoryla-

tion of PLC-g as well, presumably via TrkB kinase kDa Non-transfected parental cells K540A Y670F YY674/675FF YYY670/674/675FFF WT TrkB activity (Middlemas et al., 1994). 205 — The remaining three tyrosine autophosphorylation TrkB sites, Y670, Y674 and Y675, lie within the TrkB 116 — catalytic domain (Figure 1a). Based on comparison 97.4 — with the recent crystal structure determinations for the insulin receptor (Hubbard et al., 1994) and FGF receptor kinase domains (Mohammadi et al., 1996b), TrkB residues Y670, Y674, and Y675 appear to lie Probe: Anti-TrkB(348-363) within a structurally ¯exible `activation loop'. Auto- Figure 1 Expression of wild type and mutant TrkB proteins. (a) phosphorylation of tyrosines within this region is Schematic representation of the ®ve known BDNF-inducible postulated to induce conformational changes within TrkB autophosphorylation sites (Middlemas et al., 1994; Guiton the kinase domain, allowing the active site greater et al., 1994) (b) Expression levels of wild type and mutant TrkB accessibility to substrates. Although it has not been proteins. cDNAs encoding wild type rat TrkB or rat TrkB containing the various tyrosine to phenylalanine substitutions demonstrated, it is possible that SH2 and/or PTB were co-transfected into NIH3T3 cells along with the pSV2neo domain-containing proteins interact with one or more plasmid. Stable transfectants were selected by growth in G418- of the activation loop phosphotyrosines. containing media. Positive clones were lysed and 25 mg/lane of To address the functional importance of the three soluble lysates were fractionated by SDS ± PAGE. Clones TrkB activation loop phosphorylation sites, we have expressing the TrkB wild type or mutant proteins were identified by immunoblotting with anti-TrkB348 ± 363 (an anity puri®ed assessed the ability of receptors containing various polyclonal antibody directed against the TrkB extracellular tyrosine to phenylalanine substitutions to stimulate domain) signaling events in TrkB expressing NIH3T3 cells. In concurrence with other reports (Guiton et al., 1994; Segal et al., 1996; Cunningham et al., 1997), we show partially dispensable for kinase activation and down- that receptors containing the YY674/675FF double stream signal transduction. substitution and the Y670F single substitution are capable of BDNF-inducible receptor tyrosine autophosphorylation. Extending these ®ndings, we show that Results TrkB YY674/675FF is greatly impaired in its ability to stimulate the phosphorylation of SHC and PLCg, Expression of wild type and mutant TrkB proteins activate ERK1 and ERK2, induce c-fos protein synthesis, and stimulate NIH3T3 mitogenesis. The The ®ve known TrkB tyrosine phosphorylation sites Y670F mutation, however, only partially aects the are diagrammed in Figure 1a. Various tyrosine to ability of the receptor to induce the tested signaling phenylalanine site-directed substitutions were intro- responses. Furthermore, receptors containing the duced into the rat TrkB cDNA, followed by Y670F mutation are capable of stimulating NIH3T3 sequencing to con®rm the presence of each mutation. proliferation, albeit to a lesser extent than wild type The mutations include a Y670F single substitution, a TrkB. Mutation of all three activation loop tyrosines, YY674/675FF double substitution, and a YYY670/ TrkB YYY670/674/675FFF, impairs both receptor 674/675FFF triple substitution. Additionally, we tyrosine autophosphorylation and the activation of all constructed a lysine to alanine substitution at amino tested downstream signaling events. These data suggest acid 540, which has been shown to abolish all TrkB that while the phosphorylation of all three TrkB kinase activity (Guiton et al., 1994). tyrosines within the activation loop are required for Stable NIH3T3 transfectants expressing wild type full signaling capabilities, individual tyrosines are TrkB or the various TrkB mutant receptors were TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1693 selected in G418. Clonal cell lines expressing each ligand. A similar, although somewhat reduced, level of receptor were identi®ed by Western analysis of NP40 receptor phosphorylation was detected in the Y670F soluble lysate using the TrkB-speci®c polyclonal and YY674/675FF expressing cells. These reduced antisera, anti-TrkB(348 ± 363) (Figure 1b). As shown, levels of phosphorylation may be due to reduced all cell lines express approximately equivalent levels of receptor catalytic activity. Alternatively, they might be the TrkB proteins. No detectable TrkB is present in the explained by fewer potential sites of tyrosine phos- parental NIH3T3 cells. The multiple bands detected by phorylation present in these mutant proteins. No the antibody most likely represent dierent TrkB detectable level of TrkB phosphorylation was ob- glycosylation variants (Barbacid, 1994) and/or proteo- served in either the K540A or YYY670/674/675FFF lytic fragments of the intact TrkB protein. expressing cells. Additionally, no signal was detected for non-transfected control cells (data not shown). A relatively equivalent amount of TrkB was present at BDNF-inducible tyrosine phosphorylation of TrkB and each of the timepoints for each cell line as judged by cellular proteins parallel anti-TrkB Western blotting of the pan-Trk To test the ability of the various TrkB proteins to immunoprecipitates (Figure 2a, bottom panel). autophosphorylate in response to BDNF, we ®rst Using an in vitro kinase assay, we next tested the performed a timecourse of BDNF treatment followed ability of the various TrkB mutant proteins to by TrkB immunoprecipitation and anti-phosphotyro- phosphorylate an exogenous substrate, myelin basic sine Western analysis. As seen in Figure 2a, BDNF- protein (MBP). MBP has been used as an eective in inducible tyrosine phosphorylation of wild type TrkB vitro substrate for various tyrosine kinases (Watts et was observed at both 5 and 15 min after addition of al., 1992; Longati et al., 1994). Wild type TrkB and the various mutants were immunoprecipitated and incubated with MBP in the presence of 32P-g-ATP. Very little MBP phosphorylation was detected by immunoprecipitated TrkB K540A. However, immunoprecipita- a tion of both wild type TrkB and TrkB Y670F markedly increased MBP phosphorylation. Unexpect- edly, we found that both TrkB YY674/675FF and TrkB YYY670/674/675FFF showed MBP phosphor-

K540A Y670F YY674/675FF YYY670/674/ 675FFF WT ylation above background levels. Although the TrkB triple mutant displayed only a modest increase, 0 5' 15' 0 5' 15' 0 5' 15' 0 5' 15' 0 5' 15' +BDNF immunoprecipitated TrkB YY674/675FF phosphorylated MBP at levels similar to that of the wild type and TrkB Y670F receptors. The enhanced activity of TrkB YY674/675FF correlated well with the ability of this Ip: Pan-Trk Probe: Anti-PY receptor to autophosphorylate (Figure 2a). Therefore, although TrkB YY674/675FF and YYY670/674/ 675FFF do not stimulate detectable in vivo increases in cellular protein tyrosine phosphorylation, the TrkB receptors appear to have some kinase activity using Ip: Pan-Trk Probe: Anti-TrkB in vitro conditions. During these experiments we were unable to detect a signi®cant BDNF-inducible in vitro b increase in TrkB kinase activity. This eect was not exclusive to MBP, since we observed a similar BDNF- independent activity towards the substrate enolase (data not shown). We believe this enhanced BDNF-

K540A Y670F YY674/675FF YYY670/674/ 675FFF WT independent activity is caused by the divalent nature of the pan-Trk antibody. The antibody recognition and MBP subsequent immunoprecipitation may cause receptor aggregation, thus stimulating kinase activity in a Figure 2 (a) BDNF-induced tyrosine autophosphorylation of wild type and mutant TrkB receptors. The various TrkB-3T3 cell ligand-independent manner. Indeed, a report by lines were serum starved and then treated with BDNF for the Longati et al. (1994) showed that c-Met immunopre- indicated times. 500 mg of NP-40 soluble lysates were immuno- cipitates possess HGF-independent kinase activity. precipitated with a polyclonal pan-Trk antibody. Tyrosine Additionally, recombinantly expressed TrkA is cap- phosphorylated TrkB proteins were then detected by immuno- able of NGF-independent in vitro autophosphorylation blotting with the anti-phosphotyrosine antibody PY20 conjugated to HRP (upper panel). A portion of the immunoprecipitates (5%) when immunoprecipitated from Sf9 cells (Stephens et were probed with anti-trkB348 ± 363 (lower panel). (b) Auto- al., 1994). Attempts using pan-Trk Fab fragments did radiogram of in vitro phosphorylation of the exogenous substrate not circumvent this problem (data not shown), likely myelin basic protein (MBP) by wild type TrkB and the TrkB due to a similar aggregation eect caused by the goat mutant proteins. The various TrkB-3T3 cell lines were serum starved, lysed and the wild type or mutant TrkB proteins were anti-rabbit-agarose used to precipitate the Fab frag- immunoprecipitated from the NP-40 soluble fraction with pan- ments. Additionally, a recent report by Iwasaki et al. Trk antisera and Protein A-Sepharose. The washed immunopre- (1997) using the recombinantly expressed intracellular cipitates were resuspended in kinase buer containing MBP and 32 portion of TrkB suggests that the receptor is capable of [ P]g-ATP, and the reactions was incubated at room temperature an initial burst of cis, intramolecular autophosphoryla- for 30 min. The samples were terminated by the addition of 26SDS sample buer, resolved by SDS ± PAGE, and the tion, which may account for at least some of the non- incorporation of 32P into MBP determined by autoradiography BDNF inducible TrkB kinase activity. TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1694 The ability of wild type TrkB and the various tion of the p21ras GTPase leads to ERK activation via mutants to stimulate the phosphorylation of cellular a series of phosphorylation events collectively termed proteins was determined by a timecourse of BDNF the MAP kinase cascade. The ability of the MAP treatment followed by anti-phosphotyrosine Western kinase cascade to elicit various cellular responses analysis of NP-40 soluble cell extracts. As shown in ranging from NIH3T3 transformation to PC12 Figure 3a, application of BDNF to cells expressing dierentiation has been reported (Marshall, 1994b), wild type TrkB stimulated the phosphorylation of although the breadth of these ®ndings have recently multiple cellular proteins (denoted by arrows). been reassessed (Vallaincourt et al., 1995). Never- Although a similar quantitative pattern of protein theless, the importance of ERK activation for TrkA- phosphorylation was observed in cells expressing TrkB mediated cellular responses has been well documented Y670F, a BDNF-inducible phosphoprotein of approxi- (Stephens et al., 1994; Obermeier et al., 1994). mately 50 kDa transiently induced in the wild type To test the ability of the various TrkB cell lines to TrkB lysates was absent in the Y670F lysates, stimulate the activation of ERK1 and ERK2, we indicating this mutation somehow qualitatively aects employed a phospho-speci®c ERK1/2 polyclonal the tyrosine phosphorylation of cellular substrates. No antiserum. This antibody recognizes only the phos- signi®cant level of phosphotyrosine increase in phorylated and thus catalytically active form of ERK1/ response to BDNF was seen in the YY674/675FF 2. Cell lines expressing wild type TrkB and the various expressing cells. Thus, while TrkB YY674/675FF mutants were stimulated with BDNF, lysed, and the receptors are capable of BDNF inducible autophos- NP-40 soluble lysates were blotted and probed with the phorylation and possess in vitro kinase activity (Figure phospho-speci®c ERK1/2 antibody. The lysates were 2b), they are not capable of phosphorylating detectable also probed with an ERK1/2 speci®c antibody to levels of any cellular proteins. Finally, consistent with assure that equivalent amounts of ERKs were present their inability to autophosphorylate, the K540A and in the various samples. As shown in Figure 4, wild type YYY670/674/675FFF receptors did not elicit a BDNF- and TrkB Y670F cells induced an activation of ERK1 inducible stimulation of cellular phosphotyrosine levels. and ERK2 within 5 min of BDNF treatment. Activation of ERK2 by wild type TrkB and TrkB Y670F was temporally indistinguishable and remained Eects of activation loop mutations on SHC and PLC detectable for at least 60 min post-BDNF stimulation tyrosine phosphorylation and TrkB association (data not shown). The K540A, YY674/675FF, and SHC and PLCg are two cytoplasmic signaling YYY670/674/675FFF cells did not stimulate ERK1/2 molecules involved in TrkA and TrkB mediated activation. Interestingly, the activation of ERK2 by signaling events (Obermeier et al., 1994; Stephens et both wild type and TrkB Y670F was signi®cantly more al., 1994; Middlemas et al., 1994). We determined robust than ERK1 activation. whether receptors harboring the various activation loop mutations were capable of stimulating PLCg TrkB stimulation of c-fos protein synthesis and SHC phosphorylation (Figure 3b and d). As shown, after 5 min of BDNF treatment, cells expres- It is well established that many TrkB-mediated cellular sing wild type TrkB were capable of stimulating the responses are accompanied by an increase in c-fos phosphorylation of PLCg (Figure 3b). TrkB Y670F protein levels within 30 ± 60 min following BDNF cells, however, were not capable of inducing detectable treatment (Squinto et al., 1991; Marsh et al., 1993). PLCg phosphorylation, indicating that the presence of Recent data suggest that induction of c-fos protein the Y670F substitution does partially impair TrkB synthesis is dependent on ERK1 and ERK2 activation mediated signaling events. Furthermore, only wild type (Karin, 1994; Treisman, 1994). To test the ability of the TrkB could be co-immunoprecipitated with anti-PLCg various TrkB mutant proteins to activate pathways antisera (Figure 3c). Although basal levels of SHC leading to gene regulation, we tested each for its ability phosphorylation varied in each cell line, only wild type to induce expression of the Fos protein. Each cell line TrkB and TrkB Y670F were capable of a BDNF- was stimulated with BDNF for 60 min and nuclei were inducible increase in the tyrosine phosphorylation of harvested, lysed, fractionated by SDS ± PAGE and p46SHC and p52SHC (Figure 3d). Consistent with immunoblotted with an anti-c-fos speci®c antibody. these results, wild type TrkB and TrkB Y670F could be As seen in Figure 5, wild type TrkB and TrkB Y670F co-immunoprecipitated with anti-SHC antisera (Figure receptors induce robust levels of c-fos protein. 3e). TrkB K540A, YY674/675FF, and YYY670/674/ Coincident with their inability to activate ERK1 and 675FFF were all incapable of stimulating PLCg or ERK2, cells expressing K540A, YYY670/674/675FFF SHC phosphorylation, coincident with the inability of or YY674/675FF do not induce detectable c-fos protein the receptors to be co-immunoprecipitated with these levels in response to BDNF. signaling molecules. Together, these data further support the notion that substrate association with BDNF-inducible viability and proliferation of TrkB is required for subsequent tyrosine phosphoryla- TrkB-3T3 cells tion of the interacting protein. An interesting characteristic of the Trk RTKs is their ability to elicit mitogenic or proliferative responses BDNF-inducible activation of ERK1 and ERK2 depending on the cellular context in which they are A well characterized signaling cascade activated by expressed. Previous reports have shown that expression many RTKs involves the dual speci®city extracellular of TrkA, TrkB or TrkC in NIH3T3 ®broblasts induces regulated kinases 1 and 2 (ERK1 and ERK2) (for a potent mitogenic response when stimulated with their review see Marshall, 1994a). RTK mediated stimula- respective ligands (Cordon-Cardo et al., 1991; Glass et TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1695 al., 1991; Lambelle et al., 1991). Indeed, TrkA was quiescent by serum starvation in de®ned media. Cells originally identi®ed as an oncogenic protein that led to were then exposed to BDNF for 5 days, harvested and the ligand-independent transformation of NIH3T3 cells the number of cell divisions determined by counting (Martin-Zanca et al., 1989). cells with a hemacytometer. As positive and negative To test the eects of the various mutations on controls for each cell line tested, parallel dishes of TrkB-NIH3T3 cell proliferation, we ®rst made cells cells were exposed to normal serum-containing

a b K540A Y670F YY674/675FF YYY670/674/ 675FFF WT K540A Y670F YY674/675FF YYY670/674/ 675FFF WT kDa 0 5' 0 5' 0 5' 0 5' 0 5' +BDNF 0 5' 15' 0 5' 15' 0 5' 15' 0 5' 15' 0 5' 15' +BDNF 205 — kDa PLCγ 205 — 116 — Ip: Anti-PLCγ Probe: Anti-PY 205 — 116 — PLCγ 97.4 — 116 — Ip: Anti-PLCγ Probe: Anti-PLCγ 66 — c

45 — K540A Y670F YY674/675FF YYY670/674/ 675FFF WT 0 5' 0 5' 0 5' 0 5' 0 5'

TrkB 29 —

γ Probe: Anti-PY Ip: Anti-PLC Probe: Anti-TrkB

PLCγ

Ip: Anti-PLCγ Probe: Anti-PLCγ

d e K540A Y670F YY674/ 675FF YYY670/674/ 675FFF WT K540A Y670F YY674/ 675FF YYY670/674/ 675FFF WT

0 5' 0 5' 0 5' 0 5' 0 5' +BDNF 0 5' 0 5' 0 5' 0 5' 0 5' +BDNF kDa 66 — TrkB p52SHC p46SHC 45 — Ip: Anti-SHC Probe: Anti-PY Ip: Anti-SHC Probe: Anti-TrkB

66 — p52SHC p52SHC p46SHC 45 — p46SHC

Ip: Anti-SHC Probe: Anti-SHC Ip: Anti-SHC Probe: Anti-SHC Figure 3 BDNF inducible tyrosine phosphorylation of cellular proteins and their association with TrkB. (a) BDNF-inducible tyrosine phosphorylation of cellular proteins. NIH3T3 cell lines expressing the various TrkB proteins were serum starved, BDNF stimulated for the indicated amounts of time, and then lysed as described in Materials and methods. Soluble lysates (50 mg/lane) were resolved by SDS ± PAGE and immunoblotted with PY20:HRP conjugated antibody. The arrows indicate prominent BDNF- inducible phosphoproteins. The asterisk indicates a 50 kDa phosphoprotein that is phosphorylated in the TrkB WT cells, but not in the TrkB Y670F cells. (b, c) The tyrosine phosphorylation of PLCg and the association of PLCg with TrkB is dependent on the phosphorylation of all three activation loop tyrosines. The various cell lines were treated as described in (a) and PLCg was immunoprecipitated. The BDNF-inducible PLCg phosphorylation state was determined by anti-phosphotyrosine Western blot analysis (b). The ability of PLCg to associate with TrkB was determined by probing the immunoprecipitates with anti-TrkB(348 ± 363) antisera (c). Parallel immunoprecipitates were also probed with anti-PLCg antisera (b, c bottom panels). (d, e) The tyrosine phosphorylation of SHC and its association with TrkB requires the autophosphorylation of activation loop tyrosines 674 and 675, but not 670. SHC was immunoprecipitated from BDNF stimulated TrkB lysates and immunoblotted with either anti- phosphotyrosine (d), anti-trkB(348 ± 363) (e), or anti-SHC (d, e bottom panels) TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1696 medium or de®ned media lacking BDNF, respectively. Discussion As shown on Figure 6, cells grown in de®ned media lacking added factors generally remained quiescent or This study was designed to investigate the role of were not viable at the end of the 5 day assay. All cell speci®c tyrosine residues of the TrkB RTK in BDNF- lines exposed to serum containing medium displayed induced signal transduction. More speci®cally, we have similar growth rates, doubling an average of six times coupled site directed mutagenesis and cDNA transfec- during the 5 day assay. Similarly, cells expressing wild tion strategies in order to assess the biochemical and type TrkB produced a potent mitogenic response following BDNF exposure, doubling ®ve times over the 5 day period. BDNF treatment of cells expressing the Y670F receptors exhibited a more modest level of proliferation, doubling an average of 2.5 ± 3 times during the assay. These data indicate that although TrkB Y670F can undergo BDNF-inducible autophosphorylation and stimulate ERK1/2 activation and c- K540A Y670F YY674/ 675FF YYY670/674/ 675FFF WT fos protein synthesis, its mitogenic signaling capabilities are somewhat impaired. This reduction in kDa 0 60' 0 60' 0 60' 0 60' 0 60' +BDNF proliferation may be due to the inability of TrkB 97.4 — Y670F to phosphorylate PLCg, raising the possibility 66 — that the PLCg signaling pathway may be critical for Fos TrkB mediated mitogenicity. BDNF treatment of the 45 — K540A, YY674/675FF, and YYY670/674/675FFF Probe: Anti-c-fos expressing cells induced no detectable increase in cell Figure 5 Eects of TrkB mutations on BDNF-induced c-fos number. In general, most of these cells were non- protein synthesis. The various TrkB-3T3 cells were serum starved viable by day three of the experiment. However, cells prior to the addition of BDNF for 1 h. Following cell lysis and expressing TrkB YY674/675FF, even in the absence of isolation of nuclei, equivalent amounts of nuclear lysates (100 mg/ BDNF, remained viable, but quiescent 5 ± 6 days post- lane) were fractionated by SDS ± PAGE. Fos protein levels were serum starvation. Since, TrkB YY674/675FF is detected by immunoblotting with the c-fos speci®c polyclonal antibody SC-4 (Santa Cruz Biotech). The 65 kDa band present in capable of in vivo autophosphorylation (Figure 2a), all lanes is a non-speci®c protein which cross-reacts with the c-fos and possesses in vitro kinase activity (Figure 2b), low antibody levels of intracellular signaling may occur in these cells, possibly accounting for their extended survival period. K540A Y670F YY674/ 675FF YYY670/674/ 675FFF WT

0 5'15' 0 5'15' 0 5'15' 0 5'15' 0 5'15' +BDNF 45 — ERK1 ERK2

Probe: Anti-phospho-ERK1/2 kDa 66 —

45 — ERK1 ERK2

29 — Probe: Anti-ERK1/2 Figure 4 Activation of ERK1 and ERK2 by wild type and mutant TrkB proteins. The various TrkB-3T3 cell lines were serum starved then treated with BDNF for varying times. NP-40 Figure 6 TrkB-3T3 cellular proliferation. The various TrkB-3T3 soluble lysates (25 mg/lane) were resolved by SDS ± PAGE and cell lines were seeded onto six well dishes pre-coated with immunoblotted with a phospho-speci®c ERK1/2 polyclonal ®bronectin and poly L-lysine, and serum starved in de®ned antisera (top panel), which recognizes only the kinase active media for 24 h prior to the addition of BDNF. 5 days later, cells forms of ERK1 and ERK2. Parallel lysates were immunoblotted were harvested and the number of cell doublings was determined with another anti-ERK1/2 polyclonal antibody which detects by counting with a hemacytometer. The experiment was both phospho- and non-phospho-ERK1/2 (bottom panel), to performed three separate times. These data are from one ensure equivalent amounts of ERKs were present in each lane representative experiment TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1697 cell biological consequences of tyrosine mutations dephosphorylation of these residues also seems to be within the putative TrkB kinase activation loop. regulated, with pY674 and pY675 being rapidly Sequence alignments of the tyrosine kinase domains dephosphorylated, and pY490 remaining phosphory- from various eukaryotic proteins suggest a number of lated for an extended time. distinct subdomains (designated I-XI). Within each Because of the high degree of sequence similarity subdomain, invariant amino acids are presumed to (50% identity) between the TrkA, TrkB, TrkC, and play critical structural/functional roles (for reviews, see insulin receptor kinase domains, it was expected that Morgan and De Bondt, 1994; Hanks and Hunter, Trk activation loop tyrosines would possess a similar 1995). Studies of invariant tyrosine 416 within regulatory role. Indeed, such a role was supported subdomain VII of the pp60-src kinase (corresponding initially by studies conducted by Mitra (1991). Using to Y674 in TrkB) identi®ed this residue as a major site the oncogenic TrkA protein, single phenylalanine of phosphorylation and crucial for ecient kinase substitutions at tyrosines 503 or 504 (corresponding activation (Cooper et al., 1986). Mutagenesis studies in to TrkB Y674 and Y675) resulted in a dramatic the human insulin receptor indicate that phosphoryla- decrease in both in vitro kinase activity and tion of three subdomain VII tyrosines (Y1158, Y1162, transformation potential. Stephens et al. (1994) Y1163; corresponding to Y670, Y674, and Y675 in showed that a proto-oncogene derived TrkA protein TrkB) is necessary for complete kinase activation, containing phenylalanine substitutions at both tyro- allowing the subsequent phosphorylation of other more sines 674 and 675 resulted in autophosphorylation distal tyrosines. The receptor phosphorylation events de®cient receptors. However, Segal et al. (1996) showed seem to be sequentially ordered and temporally that phosphorylation of TrkA Y490 occurs in receptors regulated. Single substitutions of phenylalanine for containing a YY674/675FF double substitution, in- tyrosine at the human insulin receptor tyrosines 1162 dicating the presence of catalytic activity in these or 1163 results in a signi®cant reduction in receptor receptors. More recently, Cunningham et al. (1997) catalytic activity, with combined mutations at any of have shown that PC12 cells expressing TrkA YY674/ the three sites leading to increasingly diminished 675FF, while not able to extend neurites, are capable cytoplasmic signaling responses (Flores-Riveros et al., of signi®cant levels of NGF-inducible TrkA autophos- 1989; Wilden et al., 1990; Zhang et al., 1991; phorylation. Similarly, Guiton et al. (1994) have shown Murakami and Rosen, 1991). More recently, Longati that TrkB receptors containing phenylalanine substitu- et al. (1994) and Mohammadi et al. (1996a) have tions at both Y674 and Y675, while not capable of reported that single or double tyrosine to phenylala- stimulating BDNF-inducible transcription events, are nine substitutions at tyrosines corresponding to 674 still somewhat capable of BDNF-inducible receptor and 675 in TrkB within the hepatocyte growth factor autophosphorylation. Interestingly, Guiton et al. (HGF) and ®broblast growth factor (FGF) receptor (1994), using phosphopeptide mapping strategies, subdomain VII, respectively, greatly impair ligand- detected a phosphopeptide with a mobility consistent stimulated RTK activity, as well as the activation of with that of the predicted pY785 peptide in TrkB subsequent signal transduction events. YY674/675FF receptors, but reported no phosphoryla- Recent crystallographic structures for the unphos- tion of the TrkB Y484 site (corresponding to TrkA phorylated insulin and FGF receptor tyrosine kinase Y490). domains suggest critical roles for subdomain VII Consistent with the work of Guiton et al. (1994); tyrosines (Hubbard et al., 1994; Mohammadi et al., Segal et al. (1996), and Cunningham et al. (1997), we 1996b). The subdomain VII invariant tyrosines are show that signi®cant levels of BDNF-inducible proposed to lie within a structurally ¯exible `kinase autophosphorylation occurs in receptors containing activation loop'. Phosphorylation of at least a subset of phenylalanine substitutions at both TrkB tyrosines activation loop tyrosines (especially Y1162 in the 674 and 675 (Figure 2a). We extend these ®ndings to human insulin receptor) is believed to facilitate the show that TrkB YY674/675FF expressing cells are accessibility of ATP for the active site, and may also greatly impaired in their abilities to activate down- enhance kinase activity by allowing the active site stream signaling events. Although these receptors can greater access to the polypeptide substrate. Indeed, a autophosphorylate, they are incapable of inducing crystal structure of the tyrosine phosphorylated insulin detectable phosphorylation of cytoplasmic proteins receptor kinase domain suggests that phosphorylation (Figure 3). Furthermore, TrkB YY674/675FF is structurally stabilizes the activation loop in a con- incapable of stimulating the phosphorylation and formatin that enhances active site accessibility to ATP activation of ERK1 or ERK2 and is unable to induce (S. Hubbard, personal communication). Consistent the expression of c-fos protein (Figures 4 and 5). with the mutagenesis data, phenylalanine substitutions Unlike NIH3T3 cells expressing wild type TrkB, TrkB at subdomain VII/activation loop tyrosines are YY674/675FF cells are impaired in their ability to predicted to inhibit the ligand-induced conformational proliferate in response to BDNF (Figure 6). Similarly, change, resulting in impaired kinase activation. PC12 cells expressing TrkB YY674/675FF also are Original work by White et al. (1988) also suggests incapable of BDNF-inducible neurite outgrowth that phosphorylation of the insulin receptor activation (McCarty and Feinstein, unpublished data). loop tyrosines occurs in a temporally ordered manner How might TrkB YY674/675FF be capable of and is necessary for the phosphorylation of tyrosines autophosphorylation yet be defective in the activation outside the kinase domain. More recent analyses by of signaling responses? One possibility is that this Segal et al. (1996) using TrkA phospho-speci®c mutant may be able to phosphorylate at appropriate antibodies shows that activation loop tyrosine phos- tyrosines, but with reduced eciency. Perhaps the loss phorylation precedes and enhances the phosphoryla- of signaling capability re¯ects the inability of the tion of the more distal tyrosines Y490 and Y785. The receptor to reach a requisite threshold level of TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1698 phosphorylation of itself and cytosolic substrates. this data, NIH3T3 cells expressing TrkB Y785F are Alternatively, perhaps dimerization induced kinase completely impaired in their ability to survive and activation proceeds, but the YY674/675FF mutation proliferate in response to BDNF, indicating that Y785 causes loss of the necessary temporal pattern of phosphorylation is required for a proper mitogenic phosphorylation among the dierent tyrosines. Final- response (McCarty and Feinstein, unpublished data). If ly, it is possible that phosphorylation of tyrosines 674 TrkB Y670F receptors were incapable of phosphor- and 675, although not essential for complete activation ylating Y785, a complete loss of proliferative ability of the TrkB kinase domain and the subsequent (rather than the observed 40 ± 50% decrease) would be phosphorylation of tyrosines 484 and 785, may be expected. Another possibility to account for the required for ecient association of the receptor with observed decrease in TrkB Y670F survival is that various signaling molecules (e.g., SHC and PLCg). pY670 serves as a docking site for a cytosolic protein Phosphorylation of Y674 and Y675 may contribute to involved in signaling. Consistent with this possibility, a establishing a proper structural conformation, allowing comparison of the BDNF-inducible phosphoproteins the TrkB intracellular region to interact productively reveals a protein of approximately 50 kDa that is with downstream signaling molecules. Therefore, while phosphorylated in TrkB WT cells but not in TrkB TrkB YY674/675FF is catalytically active, this mutant Y670F cells (Figure 3a). protein may be incapable of interacting with other The most extreme signaling impairment was ob- necessary cytoplasmic factors because of an aberrant served in cells expressing the TrkB YYY670/674/ structural conformation. Indeed, data in Figure 3 675FFF triple mutant. Although TrkB YYY670/674/ demonstrate that TrkB YY674/675FF is incapable of 675FFF exhibited low levels of in vitro kinase activity inducing the tyrosine phosphorylation of various (Figure 2b), a complete lack of receptor autophos- cellular proteins, including SHC and PLCg, consistent phorylation (Figure 2a) and stimulation of downstream with the inability to associate with those substrates. signaling responses (Figures 3, 4 and 5) was observed Site directed mutagenesis of tyrosine 670 also in cells expressing this receptor. TrkB YYY670/674/ exhibits a slightly reduced, but still very signi®cant 675FFF was also unable to stimulate NIH3T3 level of autophosphorylation. This result is consistent proliferation (Figure 6) and PC12 dierentiation with that of Longati et al. (1994), who detected no (McCarthy and Feinstein, unpublished data) in a signi®cant decrease in receptor autophosphorylation BDNF-inducible manner. These results point to the after mutating the corresponding tyrosine (Y1230) essential nature of TrkB activation loop tyrosine within the HGF receptor. Similarly, mutation of the phosphorylation for complete receptor autophosphor- homologous tyrosine within the insulin receptor ylation, and the activation of downstream signaling (Y1158) does not signi®cantly impair receptor auto- responses. phosphorylation in response to insulin (Zhang et al., Finally, the neurotrophins are known to elicit 1991); however, induction of DNA synthesis, glycogen numerous biological responses, ranging from neuro- synthesis, and glucose uptake are all compromised in blast proliferation in the early stages of neuronal cells expressing this receptor mutation (Wilden et al., development to promoting and maintaining the 1990). Interestingly, TrkB Y670F receptors remain dierentiated state in later stages of development and capable of stimulating the phosphorylation of multiple in maturity (Davies, 1994). The expression of TrkB and cytoplasmic proteins, including SHC (Figure 3d), BDNF also correlate well with higher levels of activation of ERK1/2 (Figure 4), and an increase in invasiveness in human neuroblastoma cells (Nakaga- c-fos protein levels (Figure 5). However, TrkB Y670F wara et al., 1994; Matsumoto et al., 1995). Consistent receptors are not capable of stimulating PLCg with these results, inappropriate expression of a phosphorylation, indicating a functional impairment. mutated trk gene, which encodes a constitutively A similar lack of PLCg phosphorylation in PC12 cells active kinase, was responsible for a human colon expressing TrkA Y670F was recently reported by carcinoma (Martin-Zanca et al., 1986), leading to the Cunningham et al. (1997). Further supporting this original identi®cation of the ®rst member of the Trk signaling de®ciency, NIH3T3 cells expressing TrkB family, TrkA (Martin-Zanca et al., 1989). This range of Y670F are compromised in their ability to proliferate physiological responses to the same ligand-receptor in response to BDNF. These cells repeatedly generate a pair highlights the importance of cellular context in mitogenic response that is 55 ± 65% of that observed determining the outcome of the signal transduction with cells expressing wild type TrkB (Figure 6), cascade. Thus, understanding the mechanism of Trk con®rming that proper phosphorylation of Y670 is receptor action in dierent cellular contexts, both necessary for complete BDNF-inducible proliferation. neuronal and non-neuronal, should lend insights into Phosphorylation of this tyrosine might aect prolifera- the varied eects of these versatile receptors. tion by imparting an important conformational change which in turn aects subsequent association with various signaling components, including PLCg. Alter- Materials and methods natively, and recently proposed by Cunningham et al. (1997), mutating Y670 may not allow the receptor to Antibodies and plasmids phosphorylate Y785, thus precluding association with Recombinant human BDNF and the pBJ5 plasmid (Elliot PLCg. However, based on the phosphopeptide map- et al., 1990) were kind gifts from Andy Welcher, AMGEN ping data of Guiton et al. (1994), this possibility seems (Thousand Oaks, CA), Phospho-speci®c ERK1/2 antibody unlikely. TrkB Y670F proteins showed a BDNF- was purchased from New England Biolabs (Beverly, MA). inducible phosphopeptide migration pattern similar to Anti-ERK1/2 (SC-4) and anti-c-fos antibody (SC-52) were that of wild type TrkB, suggesting that these receptors both purchased from Santa Cruz Labs (Santa Cruz, CA). indeed contain phosphorylated Y785. Consistent with Anti-SHC was purchased from Transduction Labs TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1699 (Lexington, KY) and anti-PLCg was purchased from UBI To assess the phosphorylation states of SHC and PLCg, (Lake Placid, NY). Anti-phosphotyrosine PY20:HRP similar amounts of NP-40 soluble lysate were used for antibody was purchased from Transduction Laboratories immunoprecipitations. SHC was immunoprecipitated using (Lexington, KY). Goat-anti rabbit IgG:HRP secondary 2 mg anity puri®ed antibody, and PLCg was immunopreci- antibody was purchased from Biorad (Hercules, CA). Pan- pitated using 0.3 ml of total antisera. Immunoblots using anti- Trk polyclonal antisera was generated using a peptide phosphotyrosine were performed as described above. To test antigen corresponding to the last 15 amino acids (775 ± the association between TrkB with SHC and PLCg, 790); QNLAKASPVYLDILG) of rat TrkB. Anti-trkB348 ± immunoprecipitations of SHC and PLCg were performed as 363 polyclonal antisera was generated and anity puri®ed described above and TrkB was detected by probing using an antigen peptide corresponding to amino acids immunoblots with anti-TrkB(348 ± 363) antisera. All detec- 348 ± 363 of rat TrkB (MGRPGVDYETNPNYPE). The tion of antibody binding was performed using the Supersignal 4.7 kbp rat TrkB cDNA was a kind gift from Tony Hunter chemiluminescence kit (Pierce). (The Salk Institute) and has been described elsewhere BDNF inducible cellular tyrosine phosphorylation and (Middlemas et al., 1991). ERK1/2 activation were determined by serum starving 2.56106 cells for 48 h followed by treatment (in suspension) for 0, 5, or 15 min with 50 ng/ml BDNF. Cells were Mutagenesis, cell culture, and TrkB expression lysed in NP-40 lysis buer and for each timepoint 25 mg soluble lysate was electrophoresed (8% SDS ± PAGE), Site-directed mutagenesis of the rat TrkB cDNA was blotted, and probed with PY20:HRP or phospho-speci®c performed using the Transformer Kit from Clontech ERK1/2 antibodies, respectively (both diluted 1 : 3000 in (Palo Alto, CA). The sequences of the oligonucleotides TBST, 3% BSA). used for mutagenesis are as follows: K540A, 5'- Induction of c-fos protein was determined in monolayer GTGGCCGTGGCGACGCTGAAG-3'; Y670F, 5'-CG- cultures by serum starving 16106 cells for 24 h followed by a GGATGTATTCAGCACCGAC-3'; YY674/675FF, 5'-TA- 1 h treatment with 50 ng/ml BDNF. Cells were lysed in CAGCACCGACTTCTTCCGGGTTGGTGGC - 3'; YYY- 20 mM Tris, pH=7.4, 150 mM NaCl, 0.1% Triton X-100, 670/74/75FFF, 5'-ATGTCCCGGGATGTATTCAGCAC- 5mM EDTA, 1 mM PMSF, 2.5 mg/ml pepstatin, 2.5 mg/ml CGACTTCTTCCGGGTTGGTGGCCAC-3'. All mutated leupeptin, and nuclei were isolated by a 48C centrifugation cDNAs were sequenced to con®rm the presence of the for 5 min at 1146 g (2500 r.p.m.) in a Sorvall GLC-2B desired point mutation(s). tabletop centrifuge. Following solubilizing in sample buer, NIH3T3 cells, which do not normally express TrkB, were crude nuclear proteins were resolved on a 10% SDS- co-transfected with 1 mg pSV2neo and 20 mg pBJ5 into which polyacrylamide gel, blotted and probed with anti-c-fos the rat TrkB cDNA (wild type or receptors containing the speci®c antisera followed by a goat anti-rabbit IgG:HRP 28 various point mutations) had been subcloned. Transfections antibody (both diluted 1 : 3000 in TBST, 5% Blotto). were performed using the calcium phosphate procedure (Wigler et al., 1979). Stable transfectants were selected in 400 mg/ml G418 (Sigma) and subsequently maintained in In vitro kinase assay 150 mg/ml G418. Resistant clones were isolated and tested for Semi-con¯uent 10 cm dishes of NIH3T3 cells expressing TrkB expression by immunoblotting 50 mg of NP-40 soluble wild type TrkB or the various TrkB mutant proteins were lysates (NP-40 lysis buer; 20 mM Tris, pH=7.4, 150 mM serum starved for 18 h using DMEM containing 0.1% NaCl, 5 mM EDTA, 10% glycerol, 1% Nonidet P-40, 1 mM BSA. The cells were harvested and lysed in 1 ml of NP-40 sodium orthovanadate, 50 mM sodium pyrophosphate, lysis buer. TrkB was immunoprecipitated from the soluble 50 mM sodium ¯uoride, 1 mM PMSF, 2.5 mg/ml pepstatin lysate fractions (2 mg/ml) with 25 mg pan-Trk (IgG and 2.5 mg/ml leupeptin) with the anti-trkB348 ± 363 poly- fraction) for 2 h at 48C, followed by a 1 h incubation clonal antibody. Unless otherwise noted, all cells were grown with 25 ml Protein A-Sepharose (1 : 3 slurry). The immu- in Dulbecco's modi®ed Eagle's medium (Gibco BRL) noprecipitates were washed three times with lysis buer supplemented with 5% horse serum and 5% calf serum and incubated for 30 min at room temperature in 20 ml (Gemini; Calabasas, CA). kinase buer (20 mM Tris, pH=7.4, 10 mM MnCl2,5mg MBP, 50 mCi [32P]g-ATP). The kinase reactions were terminated with 26SDS ± PAGE sample buer and the Immunoprecipitation and immunoblotting proteins were resolved on a 15% SDS reducing gel, To assess the BDNF-inducible tyrosine phosphorylation of followed by autoradiography. wild type and mutant TrkB receptors, cells were serum starved overnight in DMEM supplemented with 0.1% calf Growth assays serum. The next day, approximately 16107 cells (3.36106 cells/ml) in suspension were treated with 50 ng/ml BDNF To measure growth factor induced viability and prolifera- at 378C for various times. Cells were lysed in NP-40 lysis tion, cells were seeded in de®ned media (Zhan and buer, and the soluble fraction was clari®ed by centrifuga- Goldfarb, 1986) onto six-well dishes (16105 cells/well) tion for 10 min in a Beckman microfuge at 14 000 g at 48C. precoated with 50 mg/ml poly-L lysine and 2.5 mg/ml Following quantitation of protein concentrations using the ®bronectin (Sigma). The de®ned media consisted of a 3 : 1 Amido-Schwarz assay (Schaner and Weissman, 1973), mixture of DMEM and Ham's F12 medium supplemented TrkB was immunoprecipitated by mixing 500 mgoflysate with 15 mM HEPES, pH=7.4, 4 mM manganese chloride, (1 mg/ml) with 5 mg pan-Trk polyclonal antisera for 2 ± 4 h 3mM histidine, 10 mM ethanolamine, 16ITSE+1 (10 mg/l at 48C. The lysates were then incubated at 48Cfor1hwith insulin, 5.5 mg/l transferrin, 5 mg/l selenium, 0.5 mg/l BSA, 25 ml (1 : 3 slurry) of Protein A-Sepharose (Sigma). The 4.7 mg/l linoleic acid; Sigma, Inc.), 2 mM hydrocortisone immunoprecipitates were washed three times with NP-40 and 150 mg/ml G418. The following day (de®ned as day 0), lysis buer, boiled in 2X sample buer and electrophoresed fresh, de®ned media containing 50 ng/ml BDNF was added on an 8% SDS-polyacrylamide gel. Following electro- to the various cell lines. As a negative control, de®ned transfer to nitrocellulose (Schleicher and Scheull Inc.) and media containing no BDNF was added. To serve as a blocking in TBST (20 mM Tris, pH=7.4, 137 mM NaCl, positive control, DMEM media containing 5% horse serum 0.2% Tween-20) containing 3% bovine serum albumin and 5% supplemented calf serum was used. Media was (BSA), membranes were probed overnight at 48C with replaced every 24 h, and on day 5 cell lines were assessed PY20:HRP (1 : 3000) in TBST-BSA. for proliferation by counting with a hemacytometer. The TrkB phosphorylation and signal transduction JH McCarty and SC Feinstein 1700 ability to proliferate in response to the various media thank Monte Radeke for the anti-trkB antibodies and conditions was determined by counting triplicate samples Margie Jess of the Neuroscience Research Institute in three independent experiments. Computer Laboratory for assistance with computer graphics. We also acknowledge and appreciate the eorts of the UCSB Advanced Instrumentation Center (AIC), Acknowledgements which generated all of the oligonucleotides used in this We are grateful to Kathy Foltz and Beth Hinkle for critical study. The UCSB AIC is part of the Materials Research comments and suggestions on the manuscript, to Andy Lab Central Facilities, supported by the National Science Welcher and his colleagues at AmGen for BDNF and Foundation under Award #DMR-9123048. This work was pBJ5, and to David Middlemas and Tony Hunter for the supported by a pre-doctoral fellowship #94-421 from the trkB cDNA. We thank Stephen Hubbard (NYU Medical American Heart Association, California Aliate to JHM Center) for sharing data prior to publication. We also and a grant from the USPHS (EY10739) to SCF.

References

Barbacid M. (1994). J. Neurobiol., 25, 1386 ± 1403. Mitra G. (1991). Oncogene, 6, 2237 ± 2241. Borg P and Margolis B. (1995). Cell, 80, 693 ± 694. Mohammadi M, Dikic I, Sorokin A, Burgess WH, Jaye M Cooper JA, Gould KL, Cartwright CA and Hunter T. and Schlessinger J. (1996a). Molec. Cell. Biol., 16, 977 ± (1986). Science, 231, 1431 ± 1434. 989. Cordon-Cardo C, Tapley P, Jing S, Nanduri V, O'Rourke E, Mohammadi M, Schlessinger J and Hubbard SR. (1996b). Lambelle F, Kovary K, Klein R, Jones K. Reichardt LF Cell, 86, 577 ± 587. and Barbacid M. (1991). Cell, 66, 173 ± 183. Morgan DO and De Bondt HL. (1994). Curr. Opin. Cell Cunningham ME, Stephens RM, Kaplan DR and Greene Biol., 6, 239 ± 246. LA. (1997). J. Biol. Chem., 272, 10957 ± 10967. Murakami MS and Rosen OM. (1991). J. Biol. Chem., 266, Davies AM. (1994). J. Neurobiol., 25, 1334 ± 1338. 22653 ± 22660. Dikic I, Batzer AG, Blaikie P, Obermeier A, Ullrich A, Nakagawara A, Azar CG, Scavarda NJ and Brodeur GM. Schlessinger J and Margolis B. (1995). J. Biol. Chem., 270, (1994). Molec. Cell. Biol., 14, 759 ± 767. 15125 ± 15129. Obermeier A, Halfter H, Weismuller K-H, Jung G, Schles- Elliot JF, Albrecht GR, Gilladoga A, Handunetti SM, singer J and Ullrich A. (1993a). EMBO J., 12, 933 ± 941. Neequaye J, Lallinger G, Minjas J and Howard RJ. Obermeier A, Lammers R, Weismuller K-H, Jung G, (1990). Proc. Natl. Acad. Sci. USA, 87, 6363 ± 6367. Schlessinger J and Ullrich A. (1993b). J. Biol. Chem., Flores-Riveros JR, Sibley E, Kastelic T and Lane AD. 268, 22963 ± 22966. (1989). J. Biol. Chem., 264, 21557 ± 21572. Obermeier A, Bradshaw RA, Seerdorf K, Choidas A, Glass DJ, Nye SN, Hantzopoulos P, Macchi MJ, Squinto P, Schlessinger J and Ullrich A. (1994). EMBO J., 13, Goldfarb M and Yancopoulos GD. (1991). Cell, 66, 405 ± 1585 ± 1590. 413. Pawson T. (1995). Nature, 371, 573 ± 580. Guiton M, Gunn-Moore FJ, Stitt N, Yancopoulos GD and Purves D. (1988). Body and Brain., Harvard University Press: Tavare JM. (1994). J. Biol. Chem., 269, 30370 ± 30377. Cambridge. Hanks SK and Hunter T. (1995). FASEB J., 9, 576 ± 596. Schaner W and Weismann C. (1973). Anal. Biochem., 56, Hefti F. (1994). J. of Neurobiol., 25, 1418 ± 1435. 502 ± 514. Hubbard SR, Wei L, Ellis L and Hendrickson A. (1994). Schlessinger J and Ullrich A. (1992). Neuron, 9, 383 ± 391. Nature, 372, 746 ± 754. Segal RA, Bhattacharyya A, Rua LA, Alberts JA, Stephens Iwasaki Y, Nishiyama H, Suzuki K and Koizumi S. (1997). RM, Kaplan DR and Stiles CD. (1996). J. Biol. Chem., Biochemistry, 36, 2694 ± 2700. 271, 20175 ± 20181. Kaplan DR and Stephens RM. (1994). J. of Neurobiol., 25, Snider WD. (1994). Cell, 77, 627 ± 638. 1404 ± 1417. Squinto SP, Stitt TN, Aldrich TH, Davis S, Bianco SM, Karin M. (1994). Curr. Opin. Cell. Biol., 6, 415 ± 424. Radziejewski C, Glass DJ, Masiakowski P, Furth ME, Klein R, Nanduri V, Jing S, Lambelle F, Tapley P, Bryant S, Valenzuela DM, DiStefano PS and Yancopoulos GD. Cordon-Cardo C, Jones KR, Reichardt LF and Barbacid (1991). Cell, 65, 885 ± 893. M. (1991). Cell, 66, 395 ± 403. Stephens RM, Loeb DM, Copeland TD, Pawson T, Greene Lambelle F, Klein R and Barbacid M. (1991). Cell, 66, 967 ± LA and Kaplan DR. (1994). Neuron, 12, 691 ± 705. 979. Treisman R. (1994). Curr. Opin. Genet. Dev., 4, 96 ± 101. Loeb DM, Stephens RM, Copeland T, Kaplan DR and Vaillancourt RR, Heasley LE, Zamarripa J, Storey B, Valius Greene LA. (1994). J. Biol. Chem., 269, 8901 ± 8910. M, Kazlauskas A and Johnson GL. (1995). Molec. Cell. Longati P, Bardelli A, Ponzetto C, Naldini N and Comoglio Biol., 15, 3655 ± 3643. PM. (1994). Oncogene, 9, 49 ± 57. Watts JD, Wilson GM, Ettehadieh E, Clark-Lewis I, Marsh HN, Scholz WK, Lambelle F, Klein R, Nanduri V, Kubanek C-A, Astell CR, Marth JD and Aebersold R. Barbacid M and Palfrey HC. (1993). J. Neurosci., 13, (1992). J. Biol. Chem., 267, 901 ± 907. 4281 ± 4292. White MF, Shoelson SE, Keutmann H and Kahn CR. Marshall CJ. (1994a). Curr. Opin. Genet. Dev., 4, 82 ± 89. (1988). J. Biol. Chem., 263, 2969 ± 2980. Marshall CJ. (1994b). Cell, 80, 179 ± 185. Wigler MS, Silverstein S, Lee L-S, Pelicer A, Cheng Y-C and Martin-Zanca D, Hughes SH and Barbacid M. (1986). Axel R. (1979). Cell, 11, 223 ± 232. Nature, 319, 743 ± 748. Wilden PA, Backer JM, Kahn CR, Cahill DA, Schroeder GJ Martin-Zanca D, Oskam R, Copeland T and Barbacid M. and White MF. (1990). Proc. Natl. Acad. Sci. USA, 87, (1989). Molec. Cell. Biol., 9, 24 ± 33. 3358 ± 3362. Matsumoto K, Wada RK, Yamashiro JM, Kaplan DR and Zhan X and Goldfarb M. (1986). Molec. Cell. Biol., 6, 3541 ± Thiede CJ. (1995). Cancer Res., 55, 1798 ± 1806. 3544. Middlemas DS, Lindberg RA and Hunter T. (1991). Molec. Zhang B, Tavare JM, Ellis L and Roth RA. (1991). J. Biol. Cell. Biol., 11, 43 ± 153. Chem., 266, 990 ± 996. Middlemas DS, Meisenhelder J and Hunter T. (1994). J. Biol. Chem., 269, 5458 ± 5466.