Relationship among types of nerve growth factor receptors on PC12 cells.

We analyzed the kinetics and thermodynamics of 125I-nerve growth factor (125I-NGF) binding to NGF-receptor on PC12 cells. We used conditions of pseudo-first order kinetics and techniques to quantitate internalized complexes, "slow" or high affinity binding complexes, and cell surface "fast" or low affinity complexes. Two possible models were examined: binding to two independent receptors at the cell surface (i.e. high and low affinity forms of NGF-receptor) and a model for consecutive formation of fast, low affinity binding followed by slow, high affinity binding or internalization. Our data are consistent with the consecutive model only. The rates of association and dissociation of NGF with slow, high affinity sites and internalized, acid wash-resistant sites are indistinguishable from each other. We also analyzed, in detail, the two assays primarily used to distinguish slow binding complexes from internalized complexes. Scatchard analysis of total binding and dissociation of pre-equilibrated 125I-NGF in the presence of unlabeled NGF at high concentration (cold wash). Neither of these assays shows any evidence that the slow, high affinity binding step is different from internalization of the 125I-NGF-receptor complex. Based on this analysis, there are only two detectable forms of NGF-receptor on PC12 cells: complexes on the surface of the cells with a binding affinity of 0.5 nM at 37 degrees C and complexes internalized by the cells. Furthermore, the data are consistent with a model in which NGF-receptor is internalized constitutively and independently of occupancy by NGF. We also examined the fate of internalized 125I-NGF. In the first 60 min after contact with PC12 cells, no degradation of 125I-NGF was observed. Moreover, a significant amount of 125I-NGF recirculates to the cell surface and is released as intact, Mr = 13,000 NGF. The cells were also stimulated by NGF in a primary neurite outgrowth assay with an ED50 of 2-16 pM under conditions of low initial cell numbers in a large extracellular volume of NGF-containing medium. Thus, low level occupancy of the cell surface receptors, Kd = 0.5 nM, for several days is sufficient to stimulate neurite outgrowth. This indicates the presence of spare NGF-receptors on the surface PC12 cells.

We analyzed the kinetics and thermodynamics of "'I-nerve growth factor ("'1-NGF) binding to NGFreceptor on PC12 cells. We used conditions of pseudofirst order kinetics and techniques to quantitate internalized complexes, "slow" or high affinity binding complexes, and cell surface "fast" or low affinity complexes. Two possible models were examined: binding to two independent receptors at the cell surface (i.e. high and low affinity forms of NGF-receptor) and a model for consecutive formation of fast, low affinity binding followed by slow, high affinity binding or internalization.
Our data are consistent with the consecutive model only. The rates of association and dissociation of NGF with slow, high affinity sites and internalized, acid wash-resistant sites are indistinguishable from each other. We also analyzed, in detail, the two assays primarily used to distinguish slow binding complexes from internalized complexes. Scatchard analysis of total binding and dissociation of pre-equilibrated "  in the presence of unlabeled NGF at high concentration (cold wash). Neither of these assays shows any evidence that the slow, high affinity binding step is different from internalization of the "'1-NGFreceptor complex.
Based on this analysis, there are only two detectable forms of NGF-receptor on PC12 cells: complexes on the surface of the cells with a binding affinity of 0.5 nM at 37 "C and complexes internalized by the cells. Furthermore, the data are consistent with a model in which NGF-receptor is internalized constitutively and independently of occupancy by NGF. We also examined the fate of internalized lzBI-NGF.
In the first 60 min after contact with PC12 cells, no degradation of "'1-NGF was observed. Moreover, a significant amount of 12'I-NGF recirculates to the cell surface and is released as intact, M, = 13,000 NGF. The cells were also stimulated by NGF in a primary neurite outgrowth assay with an EDso of 2-16 pM under conditions of low initial cell numbers in a large extracellular volume of NGF-containing medium. Thus, low level occupancy of the cell surface receptors, Kd = 0.5 nM, for several days is sufficient to stimulate neurite outgrowth. This indicates the presence of spare NGF-receptors on the surface PC 12 cells.
The binding of NGF' to its specific receptor on PC12 cells is inconsistent with a simple, single step interaction of NGF with receptor (l-lo). The most striking demonstration of the complexity of binding kinetics is observed when lZ51-NGF, bound to PC12 cells after incubation at 37 "C, is dissociated from those cells; the dissociation is clearly biphasic (1,2,4,6). Observation of deviation from linear Scatchard plots has also been used to support the argument that more than one form of NGF-receptor is present (1,(3)(4)(5)(8)(9)(10). Detailed kinetics of aggregate NGF complex formation and dissociation from NGF-receptor in PC12 cells were recently presented by Woodruff and Neet (1). Both association and dissociation kinetics showed the presence of at least two distinct complexes. It appears that the weaker complex forms and decomposes rapidly, whereas the high affinity complex forms and decomposes slowly. In the literature, as well as in the following discussion, it is important to note that several terms are used to describe each of the two forms of NGF-receptors. Low affinity, "fast" and type II receptor refer to one form, and high affinity, "slow" and type I receptor refer to another form of receptor. A major unresolved issue is the relationship between high and low affinity binding. A priori, the presence of two distinct sites can be rationalized either by the presence of two distinct and independent forms of receptor on the surface of the cells or by a pathway where a free receptor on the surface of the cell forms first a low affinity complex which then transforms to the high affinity form (1). To date, the experimental evidence for neither model is compelling. The two pathways should, in principle, be distinguishable both by thermodynamic and kinetic methods. In fact, parallel pathways should yield two distinct equilibrium binding curves, whereas the consecutive pathway should yield only one binding curve once the slower equilibrium is established. However, in practice, the notorious sensitivity of Scatchard plots to slight experimental errors and the experimental difficulties inherent in measuring slow binding in cultured cells made the results to date ambiguous. The kinetic methods, on the other hand, would only distinguish between the two pathways if the kinetics of formation and decomposition of the two complexes could be measured independently instead of measuring only the sum of the two concurrently.
Interpretation of data is further complicated by partial internalization of the NGF-receptor complex at physiological temperatures. That internalized complexes do occur is not in dispute. A scheme describing internalization of ""I-NGF mediated only by high affinity receptors has been proposed (3). In order to distinguish between high affinity and internalized receptors, they must be quantitated separately and unambiguously. Equilibration with lZ51-NGF at 37 "C followed by incubation at 0 "C with a high concentration of unlabeled NGF does not necessarily distinguish high affinity binding from internalization, since a high external concentration of unlabeled NGF cannot compete for binding at internalized 12701 1251-NGF. NGF complexes. This approach is not sufficient to distinguish the forms. Similarly, Scatchard plots are affected by internalization as well as by the presence of more than one form of receptor and do not allow quantitative distinctions to be drawn. Even observing the difference between 1251-NGF retained by cells after acid washing compared to the amount released following incubation with a high concentration of unlabeled NGF is not sufficient, if nonspecific binding under each washing condition is not rigorously accounted for.
At the onset of this work, we felt that measuring the binding equilibria and the kinetics of formation and decomposition of the two complexes individually and separately should establish whether NGF binds by the independent or the consecutive mechanism. The rapid decomposition of the cell surface NGFreceptor complexes in acidic media or after dilution in the presence of a high concentration of unlabeled NGF at 0 "C provided two tools to assess the concentration of internalized and slow or high affinity binding, respectively. However, as pointed out above, internalized complexes are not always distinguishable from high affinity binding as assessed using these techniques. This leaves open the second major question regarding the binding kinetics of NGF. Is there a relationship between high affinity binding and internalized complexes? Nevertheless, these approaches do allow a comparison to be drawn between the rate of accumulation of internal complexes and slow, high affinity binding.
Regardless of which dissociation conditions are chosen, the extrapolation of the binding kinetics to infinite time should yield reliable values for the concentration of the high and low affinity complexes. With these tools, it was possible to analyze the kinetics and thermodynamics of the formation of complexes of NGF with receptor.
PC12 cells represent an accepted model, not only for the study of binding of NGF to its specific receptor (l-3, 5, 7), but are also used as a model for the biology of neural differentiation. PC12 cells also provide a rich source of homologous cells without the limitations of material inherent in systems dependent on harvesting cells as primary cultures of sympathetic neurons from animals.
In the following report, we propose to show that the kinetics and thermodynamics of binding of "'I-NGF to PC12 cells are only consistent with the initial formation of a single type of cell surface NGF-receptor complex known variously as fast, low affinity or type II binding. Consecutively, a portion of this fast complex is then moved to an internal compartment. Furthermore, we present data which indicate that the slow binding complexes are not distinct from internalized NGFreceptor complexes. Measurement of slow complexes is complicated by nonspecific binding such that there is no compelling evidence to support the conclusion that slow binding is anything other than internalization of the NGF-receptor complex. In this regard, our work is consistent with the work of Eveleth and Bradshaw (34). They observed that type I (slow or high affinity) and type II (fast or low affinity) NGFreceptor species can be interconverted by changing the position of the receptor from the surface to an internal compartment. In our work, we used a different experimental approach and report the kinetics of the initial cell surface binding and relate that to the conversion to internalized (type I, high affinity, or slow) binding complexes. The results from both of these experimental approaches support the same general conclusions regarding the properties of NGF-receptor. Thus, the heterogeneity of NGF binding arises not from the presence of two different species of receptor in the cells, but is a consequence of the pathway by which NGF:receptor complex is processed in the cell. was added, and the incubations were continued at the temperature indicated for an additional 60 min. Total cpm associated with the cells were determined after rapid centrifugation followed by freezing and cutting the tips of the tubes containing the cell pellets into fresh tubes for y counting. The curues shown are theoretical fits calculated using the parameters for & and B msx indicated in Table 1A fitted to the equation: The data points represent means of triplicates & S. D.

RESULTS
To begin examination of the interaction between NGF and its receptor(s) on PC12 cells, we observed the binding of NGF over a 105-fold range of NGF concentrations. For these experiments we used a single, low concentration of '251-NGF for each experiment (0.01 nM, 0.032 nM, or 0.1 nM) and 13-19 concentrations of unlabeled NGF ranging from 0.004 nM to 1200 nM. Results from 0.01 nM and 0.1 nM lz51-NGF are shown in Fig. 1. The use of this broad range of concentrations of unlabeled NGF allows detection of binding sites with affinities across the relevant range of NGF concentrations. Furthermore, the results of this approach can be plotted and examined directly without resorting to Scatchard plots. The existence of two independent sites should result in a double sigmoidal plot. The highest affinity sites were evaluated using a weighted nonlinear analysis of the Langmuir binding isotherm determined from data pooled from all three concentrations of lZ51-NGF that are given in Table 1A. Only one species of sites with a single affinity was detectable in the relevant range of NGF concentrations. Fits of the experimental data to a model of two binding sites did not improve the congruency of the experimental points with the theoretical curve. These results alone demonstrate that the presence of more than one form of NGF-receptor cannot be due to independent species of binding sites.
We also observed the effect of temperature on the binding of lz51-NGF. As seen in Table lA, the affinity of lz51-NGF for PC12 cells is lower at 0 "C than at 37 "C. This observation agrees with those of Woodruff and Neet (1). It is consistent with a model where the interaction of NGF with its specific receptor is mediated predominantly by hydrophobic interactions.
We followed up the equilibrium binding studies by observ-' Portions of this paper (including "Materials and Methods," part of "Results," Tables 2 and 3, Figs. 9 and 10, and Appendix) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. a Fitted by linear regression to the equation kexp = k-, + [NGF] lzl, these results translate to a k, = 0.025 & 0.004 nM-' s-l and a k-, = 0.0079 + 0.0009 s-l. These were used to calculate an equilibrium dissociation constant of Kd = 0.32 + 0.1 nM. * Kd = 0.60 f 0.14 nM (see text).
ing the kinetics of association of lz51-NGF with intact PC12 cells. We used three different concentrations of lz51-NGF under conditions sufficient to satisfy requirements for pseudofirst order kinetics, i.e. less than 10% of the Y-NGF was consumed in the binding reaction.
The pattern for total binding (open circles) of lz51-NGF shown in Figs. 2 and 7 was consistently observed. The early phase of the data for total binding (circles) was analyzed using an integrated equation for first order kinetics. The results are consistent with pseudofirst order kinetics as shown by the agreement of the experimental points and the calculated theoretical curves (Figs. 2 and 7). We observed a linear dependency of ,&, on the concentration of iz51-NGF. If the early phase of the binding is a one-step equilibrium, then the experimental rate constant (k,,,,) of the forward reaction should be given by k., = kl.
From these kinetic measurements, we also calculated the equilibrium binding constant, & = k-Jkl = 0.32 f 0.1 nM. This value is independently derived but eminently consistent with the value derived from equilibrium binding experiments, Kd = 0.46 f 0.11 nM, Table 1. Additionally, the k1 = 2.5 X lo7 M-' s-l is consistent with a diffusion-controlled association (15) between NGF and its receptor on the surface of PC12 cells.
We also measured, over a limited range of radioactive ligand, the equilibrium binding of Y-NGF to the receptor. The data were again analyzed according to the scheme of a one-step equilibrium, i.e. Each sample was harvested after the elapsed time indicated. Total associated counts were determined by the addition of 4 ml of ice cold KRH/A followed by immediate filtration (circles). Internalized lz51-NGF (squares) was determined by rapidly mixing 4 ml of ice cold solution containing 150 mM choline Cl + 50 mM acetic acid (pH 2.5) + 0.1% FBS to each sample, incubating for 10 min in an ice water mixture (0 "C), followed by rapid filtration. The individual values represent means of triplicates. The lines shown are theoretical curves generated from the kinetic parameters calculated for the data shown. Sections of the curves shown as solid lines were used to calculate kinetic rate constants. The experiment shown is representative of a total of five experiments performed differing only with respect to the concentration of lZ51-NGF used (0.1 or 0.2 nM).
We found that the data were consistent with this equation, and we calculated Kd = 0.60 1 0.14 nM. This value is consistent with the Kd derived from the kinetic parameters (Table  1B) and with the Kd derived from isotope dilution experiments (Fig. 1). In conclusion, both the kinetic and equilibrium binding experiments characterizing the fast binding are consistent with a single, diffusion-limited reaction between '*'I-NGF and a single form of receptor on the surface of PC12 cells.
Although our data are consistent with rapid complex formation involving a single species of receptor during the first 60 s after mixing lz51-NGF with NGF-receptor on PC12 cells, incubation over a 30-45-min interval results in a gradual additional accumulation of cell-associated iz51-NGF. This may be explained by association with a second species of receptor over this prolonged period of time or by conversion of some of the rapidly associated receptor to a different state. This could include internalization of the lZ51-NGF. NGF-receptor complex.
First, we observed the amount and proportion of lZ51-NGF internalized after incubation at 37 "C for 1 h. As the concentration of NGF increased, the amount of NGF internalized also increased (Fig. 3A). However, the proportion of bound NGF internalized was constant. Although a slight negative slope is evident in Fig. 3B (A) and on the proportion of internalized NGF (B). "'1-NGF at the concentrations indicated was incubated with samples containing 1.25 x lo6 cells in 100 ~1 of KRH/ A for 90 min at 37 "C. Total associated iZ51-NGF was determined by rapidly mixing the 100-~1 sample with 5 ml of ice cold KRH/A and filtering immediately. Internalized lZ51-NGF was determined by rapidly mixing the cells with 6 ml of ice cold KRH/A supplemented with choline Cl and acetic acid sufficient to bring the final concentrations to 150 mM and 50 mM (pH 2.5), respectively. These samples were held on ice for 10 min before filtration. The nercentaee of internal cpm was calculated as: % internal = cpm internal total cell associated cpm x 100 Nonspecific binding was determined using separate samples containing 10 pg/ml unlabeled NGF during the initial incubation in the lOO-~1 samples. Nonspecific binding was subtracted from the total cellassociated cpm as well as from the cpm internal. The values shown are the means (-C S. E.) from two sets of triplicates performed in two separate experiments.
of only one of the two forms of NGF-receptor. This would result in a change in the proportion of internalized lZ51-NGF. NGF-receptor when observed over a range of concentrations of ligand.
Next, we investigated the kinetic characteristics of internalization.
The technique previously used to measure internalization (3,13) includes washing with 0.5 M NaCl and 0.2 M acetate to remove lz51-NGF bound to receptors on the surface of the cells. Preliminary experiments indicated that these high salt and acid concentrations introduce a significant degree of sample-to-sample variability within the assay. This makes analysis of the early time course of internalization very difficult. Therefore, in our assay to determine the kinetics of internalization, we maintained physiological ionic strength using 0.15 M choline Cl instead of NaCl and dropped the concentration of acid to 50 XnM. We validated the assay with cells incubated at 0 "C, where surface binding occurs but internalization is minimal, and at 37 "C, where both surface binding and internalization occur efficiently. As shown in Fig.  4, within 1 min of dilution into a solution containing 0.15 M choline Cl, 50 mM acetate (pH 3.5), and 0.1% albumin, the surface binding of 1251-NGF on PC12 cells was reduced to undetectable levels, but internalized lz51-NGF was maintained (Fig. 4).
Using this modified assay, we measured the kinetics of internalization. Fig. 2 shows the time course of internalization for '251-NGF-receptor complex. Association with receptor in the internal compartment appears more complex than the simple, first order kinetics of association for total specific binding. According to Model I, the initial rate of internalization should be maximal, whereas according to Model II, there should be a latency period corresponding to the fast binding. Statistical analysis (analysis of variance) showed that during the first 40-60 s after addition of lz51-NGF to the cells, the amount of '251-NGF present in the internal compartment was Individual samples contained 5 x lo5 PC12 cells in 200 ~1 of KRH/A with 0.1 nM "'1-NGF and were incubated for 60 min at 37 "C (filled circks) or 0 "C (open circles). At the conclusion of this incubation period, 4 ml of ice cold solution containing 150 mM choline Cl, 50 mM acetic acid, and 0.1% FBS was added to each tube. These samples were incubated for the times indicated before rapid filtration. The points shown represent means -t S. D. of three replicates. not significantly different from nonspecific binding controls. This 40-60-s delay between addition of 12*1-NGF and significant internalization was observed consistently in each of five independent experiments. These results are consistent with Model II but not with Model I.
The experimental rate constant for the five experiments used to observe the internalization rate was used to calculate the rate constants for internalization and externalization of receptor, ki and h. From Model 2, the experimental rate constant for internalization/externalization is kexp = ki + k,,.
From these two equations, the calculated values for ki and /Q are 0.080 min-' and 0.18 min-', respectively.
We examined the fate of internalized lz51-NGF. For this determination, PC12 cells were preincubated with lZ61-NGF at 3'7 "C (or 0 "C as control) to allow equilibration with the internal compartment of the cells. The cells were washed three times to remove extracellular lz51-NGF and most of the surface-bound lZ51-NGF. The wash procedure took approximately 10 min, and, with the dissociation constant calculated for the fast component from the data described in Fig. 8, tH = 50 s, this should result in dissociation of greater than 99% of the surface-bound (fast) ligand. Incubation of the cells was then continued for 60 min at either 37 "C (or 0 "C as control) to allow processing and/or intracellular circulation. As shown in Fig. 5A, a significant proportion of the previously internalized '251-NGF rapidly appeared extracellularly.
The amount spontaneously released did not differ significantly from the amount released following a wash with acetic acid and salt. As determined by SDS-PAGE (Fig. 5B), the lz51-NGF released during this time course was intact NGF with M, = 13,000.
From these results it appears that a significant proportion of the lZ51-NGF can bypass a lysosomalfdegradative pathway in the cell and can recirculate to the cell surface after internalization where it can be released into the extracellular space. We conclude from the studies described above that Model II is the appropriate model to describe binding to NGFreceptor. We also conclude that the second step is considerably slower than the first step, making it possible to observe the second step largely independent of the first step. We observed that the second step of the reaction does not go to completion: i.e. approximately 30% of the receptor is internalized. This indicates that NGF-receptor is not only going The solution was split into 2 aliquots: the incubation for 1 aliquot was continued at 0 "C; the 2nd aliquot was transferred to a 37 "C water bath. Incubation at these temperatures was continued for a total of 60 additional min. loo-p1 samples were removed at the times indicated. Triplicate samples from each incubation temperature were microfuged (10,000 X g) to pellet the cells. The supernatants were removed, and SDS-PAGE sample buffer was added to the supernatants to produce final concentrations of 1% SDS, 12 mM Tris buffer (pH 6.8), and 50 mM dithiothreitol. A third set of triplicate samples was removed from the 37 "C incubation, brought to 150 mM choline Cl and 50 mM acetic acid (pH 2.51, and incubated on ice for 10 min before centrifugation and processing as described immediately above. The values shown in A are the mean cpm of lL'I-NGF in the supernatants after incubation at 0 "C (circles), 37 "C (triangles), or incubation at 37 "C followed by the acid wash procedure (squares). B, 50-4 samples from the 37 "C incubation (after washing extracellular ""I-NGF with KRH/A and corresponding to the triangles in part A, above) were run on a 15% polyacrylamide gel. The time after initiation of the incubation period at 37 "C is given above the appropriate lane on the gel. The dye front is at the bottom of the photograph.
We used the constants for association and dissociation from cell surface receptors and internalization and externalization of receptors derived from the kinetic experiments in these equations. Since the integrated form of this system of differential equations is too complex for use in this analysis, fifth order Runge-Kutta numerical integration was used to determine integrated values for these equations, and the resulting curves were superimposed on data generated in an experimental data set of binding and internalization.
As shown in Fig.  6, the results from the numerical integration of these differential equations are in substantial agreement with the experimental data. This provides experimental support for the model.
A distinction had been made between the high affinity (slow) NGF-receptor and internalized receptor (2,3,5,10,13). We set out to test whether we could distinguish between the early time course of internalization and the early time course of binding to the slow binding component. As shown in Fig. 7, the binding of ""I-NGF to slow NGF-receptor on PC12 cells is very similar to the accumulation of internalized ""I-NGF. NGF-receptor complex. In five independent experinto the cell but also returning to the cell surface. This is also supported by the data shown in Fig. 5. When the total number of binding sites is compared after incubation at 37 "C with the number of binding sites detected after incubation at 0 "C, we (Table l), as well as Woodruff and Neet (l), observed that approximately one-third more receptor is detected at 37 "C. These observations are consistent with 30% of the receptor being internal when incubation is carried out at 0 "C; i.e. even before occupancy of the receptor by NGF. It appears that the slow step "uncovers" new sites. After equilibration of binding at 37 "C, independently assessing the proportion of internalized receptor also indicates that approximately 30% of the receptor is internal (Fig. 3). These results indicate that both before and after occupancy, 30% of the receptor is inside the cells. Thus, these results are consistent with internalization and return of receptor to the cell surface being independent of receptor occupancy with NGF. With this information, the model for NGF binding and internalization can be described as follows: NGF + Rc where C = NGF-receptor complex on the cell surface, I' = internal NGF-NGF-receptor complex, and Ri = internal NGF-receptor (not occupied by NGF). In this model based on the analysis above, the values of k. and Izi for occupied and unoccupied receptor are equal. Hence, the values of ki and k~ for occupied receptor (C) are equal to the ki and k, for unoccupied receptor (RJ. The following series of differential equations results from this model: The individual points shown were determined as described in Fig. 2 and are the mean values for triplicate determinations. The lines shown are theoretical curves generated using fifth order Runge-Kutta integration of the system of differential equations with substituted kinetic constants, as described in the text. Each sample was harvested after the elapsed time indicated. Total associated counts were determined by the addition of 4 ml of ice cold KRH/A followed by immediate filtration (circles). Slow binding (squares) was determined by rapidly mixing 4 ml of ice cold solution containing 2 rg/ml unlabeled NGF (75 nM) to each sample, incubating for 10 min in an ice water mixture (0 "C), followed by rapid filtration. The individual values represent means of triplicates. The lines shown are theoretical curves generated from the kinetic parameters calculated for the data shown, Sections of the curves shown as solid lines were used to calculate kinetic rate constants. The experiment shown is representative of a total of five experiments performed differing only with respect to the concentration of iz51-NGF used (0.1 nM or 0.2 nM).
iments, a statistically significant 0, < 0.05) 40-60-s delay was observed after addition of lz51-NGF before significant slow receptor binding was detected. Thus, both internal and slow binding were detected after a statistically significant 40-60-s delay following addition of 'Y-NGF to the cells. Additionally, the k,, values calculated from internal and slow binding experiments were not significantly different from each other. Thus, there is no obvious kinetic difference between the association to the slow binding sites and the rate of internalization of NGF. NGF-receptor complex. We also tried to distinguish between the rate of dissociation of slow binding and the rate at which internalized receptor reappears on the surface of PC12 cells and the '251-NGFA NGF-receptor complex subsequently dissociates. This was done by binding lZ51-NGF to equilibrium at 37 "C for 30 min to equilibrate both the internal and external compartments. Next, the cells were diluted 20-fold into KRH/A containing 70 nM unlabeled NGF and incubated for time intervals, as indicated, at 37 "C. One set of samples was rapidly filtered at each of the times indicated. This treatment should result in dissociation from all of the fast binding sites and should allow measurement of the dissociation rate from the slow sites at 37 "C. The second set of samples was also diluted into KRH/ A containing 70 nM NGF, but additionally was brought to a final concentration of 0.15 M choline Cl + 50 mM acetate (pH 3.5) for 2 min before filtration in order to remove any slow binding which was not also internalized.
In other words, this procedure measured the rate of return of lz51-NGF to the cell surface. As shown in Fig. 8, the dissociation from slow receptor is not significantly different from the return of internalized receptor to the cell surface. The data from both experiments were fit to the following equation: cpm bound = C!i.exp (-t.kel) + C?.exp (-t.keP) where C, and C2 represent the capacities of each site, kel and ke2 represent dissociation rate constants, and t is time after dilution.
The rapid dissociation phase indicates dissociation incubated in an ice water bath for the times shown. At the conclusion of the timed interval, each sample was either rapidly filtered for the slow binding determination, or the samples were brought to a final concentration of 150 mM choline Cl + 50 mM acetic acid (pH 2.5), and the incubation on ice was continued for an additional 2 min before filtration. The values shown are the means of triplicate samples from a single experiment. The curues shown are theoretical curves generated from the parameters for the sum of two exponentials calculated from the data shown using the equation given in the text. A duplicate experiment produced nearly identical results, and those results were combined with the results from the experiment shown to generate the values presented in the text. rates of 0.014 (f0.003) s-l and 0.020 (f0.003) s-l for the slow and internal experiments, respectively. These values are not significantly different from the values calculated for kT1 presented above. Assuming a two-step dissociation, the rate constants for the second phase of the dissociation are 0.023 (+0.002) min-' and 0.021 (f0.002) min-' for slow and internal dissociations, respectively. Within a factor of 2, these values agree with values presented previously by Woodruff and Neet (1) for the slow phase of dissociation.
More importantly, these values do not significantly differ from each other, in support of the idea that dissociation from slow and internal receptor do not differ.
The values for the slower phase dissociation rate constants calculated from the data shown in Fig. 8,0.02 min-', do differ significantly from those calculated from the initial rates of internalization, 0.18 min-', as described for data shown in Fig. 2. Therefore, the dissociation data were also fit to an equation of the sum of three exponential terms imposing the rate constants for k-, and b derived from the kinetic data. The dissociation data are fully consistent with this model containing a third component which is substantially slower than those corresponding to k-, and k+ The origin of this third dissociation component is not understood. However, we did note that during the course of the incubation before beginning dissociation a significant degree of cell clumping occurs. Partitioning of some "'1-NGF into the space between cells in clumps could explain the appearance of this third dissociation component. Further experiments will be required in order to determine precisely the origin of this component and, for the present, we cannot totally eliminate the possibility that it has some biological relevance. However, the presence of this third component with a very slow dissociation rate and observed only in this experimental protocol does not alter the major conclusion from these experiments. Dissociation from the slow component of binding defined by resistance to dissociation in the presence of a high concentration of unlabeled NGF is indistinguishable from dissociation from internalized sites defined by resistance to dissociation in the presence of an acid wash.

DISCUSSION
Analysis of either the kinetics or the thermodynamics of fast binding indicates the presence of a single type of binding site with an equilibrium dissociation binding constant of approximately 0.5 nM. The initial binding of lz51-NGF to receptor on the PC12 cell surface is completely explained by a simple, first order mechanism. A second type of site was detected beginning approximately 60 s after contact with lz51-NGF. This second type of site can be explained by internalization of '?-NGF-receptor complexes. Interestingly, the rate of receptor internalization is not affected by NGF occupancy. This conclusion is supported by two results: (a) the proportion of NGF-receptor internalized is not affected by the concentration of NGF present, and (b) in a model of binding and internalization assuming equal rates of internalization for occupied and unoccupied receptor, the experimental data fit the curves determined from numerical integration of the rate equations very well. Thus, it appears that the NGF-receptor is constitutively internalized. This is similar to the behavior of transferrin receptor (38, 39). The rates of internalization and of reappearance at the cell surface for NGF described above do not differ substantially from the rates observed for transferrin receptor (38). Similar phenomena have also been described for insulin (40). These results provide a relatively simple model to explain binding of NGF to its receptor and subsequent internalization of the complex.
Previously, Woodruff and Neet (1) performed an extensive analysis of the binding characteristics of NGF-receptor. From this analysis, they concluded that at least two forms of the NGF-receptor were detectable after incubation of lz51-NGF at 37 "C. Our results agree with that interpretation. We observed two kinetically distinguishable forms of NGF-receptor after incubation at 37 "C. However, Woodruff and Neet (1) only measured aggregate complex formation and, thus, were not able to distinguish between two possible models for the origin of the slow, high affinity NGF-receptor: (a) two independent sites or (b) sequential formation of the second type of site from the first type. By using a concentration of lZ51-NGF below the Kd of binding combined with a low concentration of receptor such that the free NGF concentration was not substantially reduced during the binding reaction, our experiments were carried out under pseudo-first order conditions. Thus, the data generated were analyzed without resorting to fitting second order rate equations. We also designed experiments to quantitate fast binding separate from slow or internalized receptors. This allowed us to distinguish between the model for two independent receptors and the model for sequential formation of the second kinetically distinguishable type of binding. Only the sequential model is consistent with our data.
Association of NGF-receptor with cytoskeleton in the absence of NGF (2) is readily explained, if the receptor is constitutively internalized. Thus, at any time, approximately 30% of NGF-receptor would be internal, while the rest is present on the cell surface, even in the absence of NGF. Triton extraction (2,(25)(26)(27) should routinely find a proportion of NGF-receptor associated with the cytoskeletal extract of cells, and the degree of association with the cytoskeleton could be modified by pretreatment with wheat germ agglutinin (25). It is even possible that the affinity of receptor when associated with cytoskeletal proteins is different from the affinity of receptor on the cell surface. However, measurement of binding under these conditions is probably complicated by the need to diffuse the ligand into the dense cytoskeletal extract, especially after treatment with wheat germ agglutinin which agglutinates PC12 cells. 3 Recent work (34) indicates that the difference between type II (fast) NGF-receptor and type I (slow) receptor is the location of the receptor in the cell. In this case, slow receptor was shown to be associated with sequestration and internalization while fast receptor was shown to be representative of receptor free on the cell surface. Interconversion of receptors was accomplished by agents which modulated the position of receptor within the cell. The conclusions drawn here are in agreement with the experiments described by Eveleth and Bradshaw (34). Bothwell et al. (29) reported on mutant PC12 cell sublines which lacked the slow form of receptor and also failed to differentiate in the presence of NGF. Although it is not clear how many or which genes were mutated in these cells, it is interesting to speculate that these cell sublines include a mutation at the NGF-receptor gene which prevents signal transduction and/or eliminates the signal to internalize the NGF . receptor complex. NR18 cells, a specific subline of these nonresponsive cells which fail to express any detectable receptor, were transfected with the gene for the human NGF receptor (30). After transfection, the NR18 cells regained the ability to induce c-fos transcription after treatment with NGF. Since both high and low affinity forms of receptor were reported, these analyses still leave open the question of the role of high affinity NGF-receptors. It is not clear that the induction of c-fos could only be mediated by the high affinity form of receptor.
It is possible that our failure to detect the presence of slow receptor distinguishable from internalized receptor is due to a difference in the strain of PC12 cells used by us compared to the strains used by other investigators. However, even if this is true, the slow form of NGF-receptor is not required for neurite outgrowth in our strain of PC12 cells, since we can readily observe neurite outgrowth at low concentrations of NGF (Fig. 10). Thus, for our PC12 cells, the absence of slow receptors does not preclude biological activity, although our data do not address the issue of whether internalization is required for biological activity. However, with the present information in hand it is now more practical to address the issue of the role of NGF internalization by NGF-receptor in the biological activity of this growth factor. These data also suggest another potential role for NGF-receptor, the ability to move intact NGF across cells. The role of NGF-receptor in the transcytosis of NGF is worthy of further investigation.
Since the rate of formation of slow binding is indistinguishable from the rate of internalization of lz51-NGF, since the slow component of dissociation from PC12 cells is indistinguishable from the rate of externalization of NGF under conditions described above (Fig. 8), and since some of the properties of binding to slow receptor are consistent with binding to a low affinity component rather than a high affinity component, it is reasonable to hypothesize that the slow binding has been confused with two properties of NGF binding: 1) internalization of the NGF. NGF-receptor complex, and 2) the presence of very low affinity, very high capacity (nonspecific) binding sites. This leaves a straightforward interpretation of the binding properties of NGF. After ac-3 S. Buxser and D. Decker, unpublished observations. counting for a potentially confusing amount of nonspecific binding and with internalization of the NGF. NGF-receptor 16. complex, all of the data are consistent with the presence of a single class of binding sites on the surface of PC12 cells. The

17
' biologically relevant sites have an affinity of approximately :i: 0.5 nM, can be internalized by the cells, and require only a low level of occupancy, albeit for a long period of time, to 20. initiate activation of the neurite outgrowth response.