High Affinity Nerve Growth Factor Binding Displays a Faster Rate of Association Than p140trk Binding IMPLICATIONS FOR MULTI-SUBUNIT POLYPEPTIDE RECEPTORS*

Nerve growth factor (NGF) binds to two cell surface receptors, p140trk and p75NGFR, which are both ex- pressed in responsive sensory, sympathetic, and basal forebrain cholinergic neurons. While pl4Otrk belongs to the family of receptor tyrosine kinases, p75NGFR is a member of the TNF/Fas/CD40/CD30 family of receptors. Current views of neurotrophin receptor function have tended to interpret p140trk as the high affinity NGF- binding site. To assess if the binding of NGF to p140frk was distinguishable from binding to high affinity sites on neuronal cells, PC12 cell sublines were generated which expressed p140trk alone, or coexpressed both and . Kinetic analysis of 12SI-NGF bind- ing indicates that it has an unusually slow rate of association with pl4Otrk (k+l = 8 x M - s-l). When both p140trk and p75NGFR receptors are coexpressed, the rate of association of NGF is increased 25-fold to produce a higher affinity binding site. An increase in the rate of internalization was also observed. Since high affinity binding and internalization are believed to be prerequi-site for the biological activities of NGF, these results suggest that the biological effects by NGF are derived from a novel kinetic binding site that requires the

The family of nerve growth factors represented by the neurotrophins is defined by multiple, structurally related factors. Although NGF has served as the prototypic example, the molecular cloning of brain-derived neurotrophic factor, BDNF (Leibock et al., 1989), lead to the subsequent identification of two additional neurotrophin factors, NT-3 and NT-4/5 (Maisonpierre et al., 1990;Hohn et al., 1990;Hallbook et al., 1991;Berkemeier et al., 1991). BDNF, NT-3, and NT4/5 have been found to promote survival and differentiation of many sensory neurons but have differential effects upon other target populations (Barde, 1989). The identification of multiple neurotrophin family members raises the important issue of how biological responsiveness is encoded for each neurotrophic factor.
NGF displays high and low affinity binding sites in sensory neurons and pheochromocytoma PC12 cells (Sutter et al., 1979;Landreth and Shooter, 1980;Schechter and Bothwell, 1981). Binding studies have indicated that 1251-NGF binds to p75 with low affinity (Chao et al., 1986;Radeke et al., 1987). Steady state binding experiments have also demonstrated that p140trk binds NGF with a Kd of lo-' M, a value reflecting low affinity binding, similar to the p75 interaction (Kaplan et al., 1991;Hempstead et al., 1991). Other measurements indicated that NGF can form a small percentage of high affinity sites with p140trk (Klein et a l . , . These results have lead to two different models of high affinity binding, one in which p140trk acts alone as the high affinity receptor for NGF, and another involving both p75 and ~1 4 0 "~ in high affinity binding (Bothwell, 1991;Chao, 1992).
Since the concentration of NGF in target tissues is in the subpicomolar range , the formation of a high affinity receptor site may act to discriminate among similar ligands and limit neuronal survival during development. With each NGF receptor individually displaying predominantly low affinity binding (Kaplan et al., 1991;Hempstead et al., 1991;Battleman et al., 1993), the analysis of NGF binding properties to cells which express both receptors and display both high and low affinity sites becomes a complex problem that requires additional biochemical and pharmacological analysis. Here we have investigated the kinetic properties of p75 and p140trk in intact cells and demon-

Kinetic Analyses
of NGF Receptor Binding 6885 strate that each receptor binds NGF with markedly different rates of association and dissociation. Moreover, when both receptors are coexpressed, a novel kinetic site is generated, resulting in high affinity binding.

EXPERIMENTAL PROCEDURES
Radiodination of NGF-Mouse submaxillary NGF (2.5s) was obtained from Bioproducts for Science and was radioiodinated using lactoperoxidase and hydrogen peroxide as described (Hempstead et al., 1989). The specific activity ranged from 2800-3500 countdmixdfmol, and the NGF was used within 2 weeks of labeling. The radiolabeled NGF did not display any proteolysis, or degradation, as assessed by SDS-polyacrylamide gel electrophoresis.
Cell Culture-The human melamona cell line A875 was obtained from the American Type Culture Collection, and maintained in Dulbeceo's modified Eagle's media (DMEM) supplemented with 10% vlv fetal calf serum. Fibroblast cells expressing the rat p14 (rrk (Kaplan et al., 1991) or human ~7 5~ were maintained in D~E M , 10% fetal calf serum plus 0.2 mg/m1 G418 (Life Technologies, Inc.). PC12 cells overexpressing trk (Hempstead et al., 1992) were maintained in DMEM plus 10% calf serum, 5% horse serum, and 0.2 mgtml G418. NR18 cells, a PC12 cell subline (Bothwell et al., 1980) lacking p75 but expressing low levels of endogenous rat p140frb were transfected with rat p140"* to generate a neural crest-derived cell line which expresses significant levels of p140, but no p75. These cells, a gift of Luis Parada, were maintained in DMEM plus 10% fetal bovine serum, 5% horse serum, and 0.2 mg/ml G418. All media was supplemented with glutamine and penicillin/ streptomycin. A875 cells overexpressing p140trk were generated after transfection with pDM115 (Hempstead et ul., 1991) and were identified by affinity cross4inking to 'T-NGF and immunoprecipitation with anti-trk antibodies.
~n t e r n a~~u t~o n A n u l y s i~-I n t e~~i z a t i o n assays were performed using a protocol modified from Haigler et al. 11980) Cells grown to 80% confluence in 60-mm2 plates were washed twice in serum-free DMEM and incubated in serum-free DMEM containing 0.5 n M lz5I-NGF, in the presence or absence of 500 rn unlabeled NGF. Following incubation at 37 or 4 "C, the media was removed, cells were washed twice with icecold PBS, and surface bound ligand was dissociated from cells using ice-cold 0.2 M acetic acid, pH 3.5, in 0.5 M NaCl for 5 min. The acid solution and a 1-ml wash were pooled and represented the cell surfacebound radioactivity. Internalized ligand was determined by lysis of the cells with 1 ml of 1 M NaOH for 30 min at 37 "C followed by a 1-ml wash.
Equilibrium Binding StudiesSteady state binding studies with IZ5I-NGF were performed using intact cells in a shaking ice water bath. Cells were harvested by trituration in PBS-2 m EDTA, washed twice in PBS-EDTA, and resuspended at a concentration of 0.5-1.0 x lo6 celldml in PBS containing 1 mg/ml BSA and 1 mg/ml glucose. Cells were incubated with increasing concentrations of lZ5I-NGF (0.001-5 nM) in the presence or absence of an excess of unlabeled NGF (0.8 PM) for 2 h. Aliquots were layered over a cushion of calf serum, and cell-bound radioactivity was separated from free radioactivity by centrifugation at 14,000 revolutiondmin for 2 min at 4 "C in a Beckman microfuge. Tubes were immediately frozen in a dry ice-ethanol bath, and tips containing the cell pellet were clipped and counted. Samples were assayed in triplicate, and specific binding (total minus nonspecific) was 60-85% of Kinetic Binding Assays-The rate of association of NGF to cells expressing different NGF receptors was determined using a ~o d i f i~t i o n of Sutter et al. (1979). Cell suspensions were incubated in PBS, 1 mg/ml BSA, and 1 mg/ml glucose with lZ5I-NGF (2.5 x lo-" M or as indicated) in a shaking ice water bath. At the indicated times, an aliquot was layered onto calf serum, and bound radioactivity separated from free by K. C. Hsu and M. V. Chao, unpublished results. centrifugation as above. Nonspecific binding of NGF to cells was determined in replicate tubes in which unlabeled NGF was present (2.5 X IO-'? M) and accounted for less than 40% of total counts. Triplicate or quadruplicate samples were analyzed at each time point.
To determine the association rate of NGF to the high affinity site on cells coexpressing both NGF receptors, excess unlabeled NGF was briefly added to the cell suspension, to promote rapid dissociation of NGF from low affinity p75-binding sites. Specifically, cells were incubated with W-NGF (2.5 x lo-" M), in the presence or absence of unlabeled NGF as above. At the indicated time, unlabeled NGF (2.8 X M) was added to each tube and incubation was continued for 30 s at 0.4 "C. A 0.1-ml aliquot was removed, and bound ligand was separated from free by centrifugation.
The rate of dissociation of NGF from each receptor was measured using two techniques, displacement and dilution. For each study, steady state binding of lz5I-NGF to cells was achieved at 0.4 "C for 2 h for cells expressing p14(Yp*, or 20 min for cells expressing only p75, as described above using replicate samples incubated in the presence or absence of 100-fold excess unlabeled NGF. To initiate displacement, unlabeled NCF was added to a final concentration of 200 nu, and bound radiolabeled NGF was separated from unbound NGF by centrifugation over a calf serum gradient. The zero time point was determined in samples to which an equivalent volume of diluent was added and was determined in sextuplet. For the dilution method, cells preincubated for 2 h (if expressing p140rrk, or coexpressing both receptors) or 20 min (if expressing only p75) at 0.4 "C with lz5I-NGF (5 x lo-" M, in the presence or absence of 5 x M of unlabeled NGF, in a total volume of 150 111) were diluted into 15 ml of PBS containing 1 mg/ml BSA and 1 mg/ml glucose. At the indicated times, a 1.3-ml aliquot was removed and directly centrifuged at 14,000 revolutiondmin in a Beckman Microfuge at 4 "C for 1 min. Tubes were either immediately frozen using dry ice ethanol (for cells expressing p75) or aspirated to dryness, and then tips were cut and counted.
Kinetic Data Analysis-For analysis of the association rate, the integrated rate equation for a reversible second-order reaction was used: (Eq. 1) with t = time calculated in seconds, L, = initial ligand concentration, B = ligand bound at time t , and B,,, = maximal binding determined in independent saturation experiments. This analysis assumes that the concentration of bound ligand is so low that the k., of the receptorligand complex is zero. In these experiments, the concentration of bound ligand at equilibrium represented less than 15% of the total ligand, validating that assumption.

Surface Expression and Internalization of
Receptors-Cultured cells expressing each receptor subunit, and coexpressing both receptors, were analyzed for steady state and kinetic binding properties. In these studies, we have utilized both fibroblasts and PC12-derived cell lines, which express p140trk, p75, or both receptors. Since primary neuronal cultures are heterogeneous in neurotrophin receptor expression, we have used transfected fibroblast and PC12 sublines.
To confirm cell surface expression of each receptor, affinity cross-linking o f lZ5I-NGF to intact cells was performed, followed by immunoprecipitation with anti-receptor antisera. As shown in Fig. 1, 3T3 fibroblasts expressing rat p140trk (3T3trk); 3T3 fibroblasts expressing human p75 (3T3-p75); rat pheochromocytoma cells transfected with human trkA (trk-PClZ, 615 subline); and the PC12 cell subline, NR18 (Bothwell et al., 1980), .overexpressing rat p140trk (referred to as neural crest fF&2-derived NC-trk) displayed significant levels of NGF-receptor complexes. These cross-linking experiments confirm the expression of only one receptor in 3T3-p75,3T3-trk, and NC-trk cells. Another neural erest-derived cell line, A875 melanoma, expressing comparable levels of p75 as 3T3-p75 cells, was utilized in some experiments. The trkA-overexpressing PC12 cell line (615) was used because it expresses both p75 and approximately 20-fold more ~1 4 0 ' '~ receptors than do parental PC12 cells ( Fig. 1) and more high affinity sites (Hempstead et al., 1992). The increase in the number of high affinity sites would enhance the ability to detect differences in kinetic rate constants in cells possessing both high and low affhity binding sites.
The time dependence of binding of NGF to receptors in intact cells has been previously examined a t elevated temperatures such as 25 or 37 "C (Klein et Weskamp and Reichardt, 1991;Ibanez et al., 1991), conditions which facilitate internalization of NGF (Bernd and Greene, 1984). Therefore, initial experiments were aimed a t identifying conditions which circumvent receptor-mediated endocytosis, synthesis of new re-

. Expression of NGF receptors in cultured cells assessed
ml) were incubated with 50 ng/ml I2"-NGF for 2 h a t 4 "C, and crosslinked with 25 p~ disuccinimidyl suberate and 1 mM ethyldimethylaminopropyl-carbodiimide. Following washing, cells were lysed in detergent, and '2"I-NGF-receptor complexes were immunoprecipitated with the antisera indicated. Autoradiography of SDS-polyacrylamide gel electrophoresis resolved proteins was performed for 3 days (anti-trk immunoprecipitates) or 18 hours (anti-p75 immunoprecipitates). Autoradiography of neural crest ( N C ) cells expressing trkA, NC-trkA(NR18-trkA cells) complexes was performed for 24 h. *O 1 ceptors, and receptor-ligand degradation, and would therefore minimize these secondary events following ligand-receptor binding. To distinguish between ligand that was bound to receptors at the cell surface and that which had been internalized, cells were briefly exposed to low pH to release surfacebound ligand, and then internalized ligand was quantitated following cell solubilization. As shown in Fig. 2, cells which express only the p75 receptor internalize NGF slowly, with a low ratio of interna1ized:external ligand ( V E ) of 0.37 after 90 min, consistent with previous results . The modest increase in surface-bound ligand following incubation a t 37 "C may reflect recycling of the p75 receptor.
In contrast, cells expressing only ~1 4 0 "~ or cells coexpressing both p75 and pl4Ofrk internalize NGF more rapidly, with an IIE of 1.5 and 1.7 after 90 min of incubation a t 37 "C, respectively. There was a small, but reproducible increase in the relative rate of internalization when NGF binding to p140Lrk was measured in the presence of p75. This increase in the rate of internalization can also be detected when p75 and 140frk are coexpressed in PC12 cell lines, compared with PC12 sublines expressing p140trk alone (data not shown). When the binding experiments were performed a t 4 "C, the ratio of internalized to surface-bound ligand was markedly reduced ( V E of 0.23, 0.09, and 0.15 for p75, p140frk and p75-pl4Otrk coexpressing cells, respectively). These results indicate that measurements of surface-bound NGF could be overestimated, if binding is carried out a t 37 "C (Klein et al., 1991), due to internalization of NGF.
Thus, all binding analysis was performed a t 0.4 "C to reduce internalization of receptors. p75 can be estimated as approximately 13.5 (trk-PC12 cells), or 1:3 (A875-trk cells). This was determined by incubation of equilibrium binding reaction with excess cold NGF immediately prior to centrifugation, a condition that displaces labeled NGF from p75 receptors. These binding results with whole cells are consistent with previous membrane reconstitution experiments (Hempstead et al., 1989(Hempstead et al., , 1991Battleman et al., 1993) carried out at 30 "C. Furthermore, these experiments demonstrate that p75 or ~1 4 0 "~ expressed independently in neural crest-derived cell lines are incapable of yielding a biphasic curve with significant (greater than 10%) numbers of high affinity sites. Rate of Dissociation-NGF binding has been characterized by fast and slow binding sites, based upon off-rate measurements (Schechter and Bothwell, 1991). To determine which site is represented by each receptor, the rate of dissociation of 1251-NGF from p75 and p140trk was determined. To calculate the off-rate of lZ5I-NGF from its receptors, intact cells were incubated with radiolabeled ligand for 2 h at 0.4 "C to assure that equilibrium binding had been reached. In displacement experiments, unlabeled ligand (2 x 10" M) was added, and 12'I-NGF remaining surface bound was determined at increasing times from the addition of cold ligand. The rate of dissociation of NGF from p75 is very rapid, with a tlt2 of less than 1 min ( Fig. 6.41, a result consistent with previous studies (Radeke et al., 1987). The dissociation of NGF from p14fyrk, in contrast, is much slower, with a tm of 70 min (Fig. 4A). The dissociation curve was nonlinear, and this complex curve was confirmed using a wide range of initial binding concentrations of '251-NGF from 0.01 to 1 x lo-' M (Fig. 4A, and data not shown). Such a dissociation profile has been observed for many receptor systems, including NGF .
When the rate of dissociation was determined by dilution of cells pre-equilibrated with 1251-NGF, the off-rate of NGF appears monophasic with fibroblast or neural crest cell lines expressing p140trk alone (Fig. 4 B ) or p75 alone (Fig. 4Cf. In addition, the dissociation rate from p75 expressing cells remains more rapid than from ~1 4 0 ' '~ expressing cells (t, = 12 and 160 min, respectively). Utilizing the equation k l = 0.693/t, the k l can be calculated for p75 expressing cells (1 x s-') and p140fr' expressing cells (7.2 x s-*). The dissociation rates of NGF from ~1 4 0~~~ were found to be identical between fibroblast and PC12 sublines (Fig. 4B).
In neural crest-derived cell lines co-expressing both NGF receptors and displaying both high and low affinity sites, the rate of dissociation of NGF appears biphasic (Fig. 401, with an initial fast component, and a slower component which is very similar to the rate obtained from cells expressing p140trk alone. This result was obtained using two independent cell lines coexpressing p75 and p140frk (the PC12 subline, trk-PC12 cells, and the melanoma cell line trk-A875). This dissociation profile was evaluated using cells preincubated with diff'erent concentrations of labeled NGF.
When trk-PC12 cells were preincubated with low concentra- were observed (70%, Fig. 3 0 ) . When cells were preincubated with higher concentrations of NGF (1 x lo-' M), predominantly fast sites (75%), and fewer slowly dissociating sites (25%) were observed (data not shown). The slope of the fast component in the dissociation plot from ~7 5 -p l 4 0 "~ coexpressing cells is consistent with the dissociation curve obtained with cells expressing p75 alone. Furthermore, the slope of the slow component in the dissociation plot from p75-pl4Otrk coexpressors (trk-PC12 and trk-A875 cells) is very similar to that obtained using cells expressing p140trk alone. This suggests that the biphasic curve obtained with cells coexpressing both receptors could result from the independent dissociation of NGF from p75 sites and pl4Otrk sites. Since the majority of sites present in coexpressing cells equilibrated with low concentrations of NGF (2.5 x 10"l M, Fig. 4 0 ) a r e of the slowly dissociating type, the measurements indicate that the rate of dissociation from the high affinity site is very similar to the rate of dissociation from ~1 4 0 "~.
Rate of Association-To determine the rate of association of NGF to each receptor species, and to the high M) affinity site, cells were incubated with labeled NGF and the bound ligand was determined as a function of time of incubation. The rate of association of NGF with p75 was extremely rapid (k+l= 8 x lo6 M -~ s-l; Fig. 5 A ) , suggesting that it may be diffusion limited (Sutter et al., 1979). The rate of association of NGF with p140trk is markedly slower with a calculated k+, = 8 x lo5 M -~ s-l (32'3-trU; Fig. 5B) and k+l = 6.4 x lo5 M -~ s-l (NC-trU, Fig. 5C). The rapid rates of association of NGF with p75 and the slow rate of association of NGF with p140trk were consistent whether the receptors were expressed in fibroblasts or PC12 sublines.
In order to determine the association rate of NGF with the high affinity binding site, a strategy was developed to distinguish binding to low affinity from high affinity sites. Even a t low NGF concentrations (2.5 x M) approximately 20% of the specific binding represented a low affinity interaction with p75. This was apparent by examining the proportion of NGF which rapidly dissociated from sites pre-equilibrated with lo-" M NGF (Fig. 40). Therefore, to minimize the binding of NGF to these low affinity p75 sites, cells expressing high affinity sites were pre-equilibrated with 2.5 x M 1251-NGF, and then incubated with an excess of unlabeled NGF for 30 s to allow NGF to dissociate from low affinity p75 sites. This protocol resulted in greater than 90% dissociation of NGF from p75 sites (Fig. 6A), but less than 5% dissociation of the bound NGF from p140trk sites (data not shown). Thus, the association rate for high affinity sites on trk-PC12 cells (Fig.   6B) was calculated as k+l = 2.0 x lo7 M-, s-l, a rate which is approximately 25 times as rapid as the association of NGF to the p140trk site in the PC12 subline expressing p140trk alone. This rapid association rate suggests that the interaction may b'e diffusion-limited. Using an independent neural crest cell line coexpressing p75 and p140trk (trk-A875), the association rate for high affinity sites yielded a comparable value of k+l = 2.3 x lo7 M -~ s-l (Fig. 6C). The experimental strategy employed in these studies, however, cannot distinguish the contribution of "low affinity" trkA binding to the calculation of high affinity site binding. Although this contribution is low (less than 20%), its inclusion minimizes differences between the association of p140trk and of high affinity binding sites.
A derived Kd can be calculated from the independently derived association and dissociation rates as the ratio of the k.J k+l. As shown in Table I, the derived Kd is on the order of 10-l' M for cells expressing either p75 or p140trk, but is 25-fold lower for the high affinity site (2 x M) detected on co-expressing cells. These calculations compare favorably with the Kd derived from the independent equilibrium binding experiments described above. These Kd values are slightly higher in absolute values, due to the slower dissociation rate measured by the dilution, rather than the displacement method. Thus, in cells coexpressing p75 and p140frk, the distinguishing kinetic parameter for the high affinity site is an accelerated on-rate. These measurements establish that high affinity site formation is dependent upon expression of both receptor subunits. Many mammalian receptors display more than one binding constant, including receptors for growth factors such as EGF, interleukin (1L)-2, IL3, IL-6, and receptors for smaller ligands such as the protein-coupled p-adrenergic receptor. The complex binding patterns that have been observed may result from a single receptor whose afflnity is affected by accessory membrane proteins, multiple receptor subtypes, or multiple binding confo~ations within a single receptor complex. The characterization of two functional receptor subunits for cytokines (Nicola and M e h l f , 1992) has indicated that high affinity ligand binding requires coexpression of two subunits, each of which independently binds ligand with a lower affinity. A common feature of this class of receptors is that the high and low affinity classes of binding sites differ nearly 100-fold in the equilibrium binding constants. Additionally, for many two-site receptors analyzed to date, the low affinity site is 10-fold more abundant than the high affinity species in vivo.
The p75 neurotrophin receptor is distinguished by fast association and dissociation rates (Radeke et al., 1987;Rodriguez et al., 1990), whereas ~1 4 0 " displays a very slow dissociation behavior . The results in this study demonstrate that the major difference in the binding between the two NGF receptors is derived not only from a disparity in the d~ssociation rates, but also in the rate of association. NGF displayed a far more rapid on-rate to p75 than p140trk at 0.4 "6. The half-time for association of NGF with p75 was only 30 s, whereas the association half-time (40 min) was considerably slower for p140trk. When both receptors are coexpressed, the rate of NGF association is accelerated, resulting in a higher equilibrium binding constant. This same behavior was observed, irrespective of the cell type analyzed, indicating that the kinetic values are a property of the two NGF receptors, and are not due to inherent or undetected differences in the cells themselves.
Previous studies described a high affinity NGF-binding site (& 1~: 10 PM), and a low affinity site with a X, of 1 n M (Sutter et at., 1979;Landreth and Shooter, 1980;Schechter and Bothwell, 1981). The measurements were pedormed using either chick sensory neurons or PC12 cells, which display both high and low affinity sites, and are very similar to the analysis described here for heterologous cells that coexpress p75 and p140trk. Sutter et al. (1979) calculated off-rates for low affinity (~7 5 ) binding, and for the "high affinity site" that are strikingly similar to those determined here. The slow dissociation of NGF from p140trk, and its similarity to the slow dissociation rate of the high affinity receptor, has led to the conclusion that ~140"'~ behaves alone as a high affinity receptor. As this study clearly demonstrates, the accelerated association rate of the high affinity site, that can be readily distinguished %om the slow association rate of p140Crk, is the defining kinetic parameter.
This difference in association rates has been difficult to measure, primarily because the number of high affinity sites is much lower than the number of low affinity sites. Indeed, our initial kinetic experiments using PC12 cells, which display only 3% high affinity sites at low temperatures, were inconclusive, One previous study has attempted to define the association of NGF with ~1 4 0 " '~ using affinity cross-linking (Harkman et al., 1992). Although kinetic constants cannot be calculated using this technique, the results are compatib~e with the association rate constants reported here. Several lines of investigation have suggested that the low affinity p75 participates in high affinity binding with the trk tyrosine kinase receptor. First, deletion mutations in p75 have been found to eliminate high affinity binding (Hempstead et al., 1990;Battleman et al., 19931, suggesting that coexpression of an intact p75 protein with rat p140trk is required. In addition, reconstitution studies of human p75 and rat p140t* by membrane fusion and expression of receptor cDNAs in 3T3, COS, and melanoma cells (Hempstead et al., 1991) also suggested that high aEnity binding required coexpression and binding to both NGF receptors. Other equilibrium binding studies, however, have provided evidence that human p140trk in the absence of p75, is capable of giving high affinity binding in fibroblast whereas the majority of the binding was measured with a nanomolar equilibrium binding constant. In addition, dose response studies (Glass et al., 1991;Cordon-Cardo et al., 1991;Ip et al., 1993) have reported that similar concentrations of NGF are required to elicit responses in cells expressing both p75 and p140tTk, as compared with cells expressing only p140trk. These findings have lead to the conclusion that high affinity NGF binding does not require expression of the p75 neurotrophin receptor subunit. Several explanations may account for these discrepant results. First, many of the fibroblast studies were undertaken with cells that markedly overexpress trk, which could result in increased receptor dimerization of p140trk, and thus a small number of high affinity sites . In vivo, the levels of p75 and trk normally exist at a ratio of approximately 10:l and are tightly regulated ~Holtzman, et al., Verge et al., 1992). It should be noted that 10-15% of the binding of sensory neurons constitutes high affinity binding sites (Sutter et al., 1979). These percentage values were not observed in binding studies of trk in fibroblast cells (Kaplan et af., 1991;Hempstead et al., 1991;Battleman et al., 1993). The inability to observe an effect of p75 upon high affinity binding may be due to difficulties of coexpressing high levels of p75 with trkA (see Table I from Jing el al., 1992). Second, significant numbers of p140trk receptors are internalized at 37 "C, giving rise to noncompetable binding sites, which could alter measurements of equilibrium binding constants. Third, functional responses have been frequently equated with high affinity binding, but high affinity site formation and signal transduction may represent two separate receptor functions. For example, high affinity binding in heterologous cells does not necessarily lead to differentiating responses such as neurite induction (Hempstead et al., 1989;Pleasure et al., 1990 (Ibanez et al., 1992). Survival and neurite outgrowth of NGF-treated MAH cells required only expression of p14Wrk, without concomitant expression of p75 (Birren et al., 1992). While these results were interpreted as distinguishing between models of signaling involving trk and p75, the interaction of NGF with a high affinity binding site was not examined (Ibanez et al., 1991;Birren et al., 19921, and many of the biological responses were obtained using relatively high concentrations (2 x M to 2 x M) of NGF. In more recent studies, NGF mutants which bind less readily to p75 display distinctive responses, as assessed by neurite outgrowth and cell survival (Ibanez et al. 1993). Other studies utilizing antibodies directed against the p75 receptor which block NGF-p75 binding have suggested that p75 is not necessary for NGF action (Weskamp and Reichardt, 1991); however, the number of high affinity sites was considerably reduced under these conditions. All of these studies suggest that biological signaling may be separable from binding to high a f f i t y sites, when both high and low affinity binding sites are present.
Each of these previous studies can be reconciled with the binding analysis presented here if the high afEnity NGF-binding site is considered to be either (a) a multimeric complex of p75-p140trk proteins, or ( b ) an altered conformation of p140td, formed in the presence of p75. Delineation of these models will require further and more extensive studies utilizing either mutant NGF receptors or mutant NGF molecules which display altered interactions with either p75 or ~140". It is important to emphasize that the results in this study were obtained in neural crest cells expressing p140frk and/or p75 receptors, USing NGF as the ligand. The kinetic analysis of the interactions of BDNF, NT-3, and NT-4/5 with trk receptors or with p75 (Rodriguez-%bar et al., 1990(Rodriguez-%bar et al., , 1992) may be distinctive from those described here. Strikingly similar conclusions have been reached with the IL-2 receptor (Grant et al., 19921, which is composed of at least two different binding proteins with different dissociation constants. The a (Taq) binding subunit, binds 1251-IL-2 with low affinity. Like the low affinity p75 NGF receptor, the a subunit is found in relative abundance and is a "fast" receptor, with short on-rate and off-rates of dissociation. The p subunit confers intermediate IL-2 binding affinity in T cells. A third subunit, y, is required for signal transduction and internalization of the receptor complex (Takeshita et al., 1992). Coexpression of a and p subunits produces a higher affinity IL-2-binding site, by generating a site with an accelerated association rate, analogous to the NGF receptor situation. Furthermore, the cell expression of the IL-2 a subunit is capable of altering the binding characteristics of the IL-2 p with its ligand, by a mechanism which does not require IL-2 to bind to the a subunit (Grant et al., 1992). Lastly, coexpression of the y subunit with either the a or p subunits can modulate the interactions of 11;-2 with both the a and j3 subunits (Voss et al., 1993). Taken together, these studies support a model for the IL-2 receptor in which both coexpression of mult~ple receptor subunits and ligand binding to these subunits induces conformational changes in the receptor complex leading to a high affinity, functionally competent receptor species (Taniguchi and Minami, 1993).
Why have polypeptide growth factor receptors evolved a multisubunit system to regulate receptor afhities? Growth factor receptors may require differing affinities to mediate distinctive biological functions. For NGF, the expression and functional activities of ~1 4 0 ' '~ alone are capable of giving rise to initial neurotrophin signal transduction by increased autophosphorylation (Kaplan et al., 1991; and increased cellular protein phosphorylation. The coexpression of p75 in many NGF-responsive neurons may serve a number of crucial functions in discrete neuronal populations, including increasing the affinity of neurotrophin binding during competition fog limited concentrations of trophic factors (Barde 1989), participation in ligand internaIization after binding (Bernd and Greene, 19841, transport of NGF intracellularly during retrograde transport (Johnson et al., 1987;DiStefano et al. 1992), and d i s c~m i n a~o n between different, but closely related, neurotrophin factors (Rodri~ez-bar et al., 1992;Benedetti et at., 1993), or as a mediator of apoptosis (Rabizadeh et at. 1993). For multipolar neuronal cells undergoing changes in electrical excitability, rapid ion fluxes, and high rates of protein and RNA biosynthesis, the necessity of a high affinity, multisubunit signaling receptor complex to respond to extracellular survival factors may be an essential mechanism for signal transduction.