Binding of oxytocin to uterine cells in vitro. Occurrence of several binding site populations and reidentification of oxytocin receptors.

Myometrial and endometrial cells of sheep, rat, and calf in monolayer cell culture display at least three populations of binding sites for oxytocin, with dissociation constants (Kd) of approximately 5 X 10(-9), 4 X 10(-7), and greater than 10(-5) mol/liter, respectively. Binding of the tritium-labeled oxytocin (concentration range, 10(-11) to 5 X 10(-4) M) to the first two sites is displaceable by cold oxytocin. The ratio of binding capacities of the high to medium affinity site appears to average 1:18. Dissociation rate constants for these sites (22 degrees C) are roughly 10(-4) and 2 X 10(-3) s-1, respectively. The capacity of the low affinity site varies in individual cell preparations and is between 5 and 66 times that of the medium affinity site. The low affinity binding sites may not be fully saturable and may follow a nonasymptotic binding isotherm. Logarithms of Kd and binding capacity for individual binding sites are linearly correlated. The coexistence of the three sites was also proven by cluster analysis based on similarities between Kd, binding capacity, and Hill coefficient. Only minor systematic species and cell type differences occur in these properties. The value of Kd for the oxytocin receptor in rat myometrium, derived recently from a stepwise irreversible inhibition of uterotonic response to oxytocin, is close to 2.5 X 10(-7) mol/liter. Additional pharmacological data (pA2 values of structural analogues of oxytocin acting as competitive inhibitors) also reveal a Kd value of 3 X 10(-7). It is, therefore, concluded that the receptors for oxytocin in rat myometrium are identical with the medium affinity site.

ity of the low affinity site varies in individual cell preparations and is between 5 and 66 times that of the medium affinity site. The low affinity binding sites may not be fully saturable and may follow a nonasymptotic binding isotherm. Logarithms of & and binding capacity for individual binding sites are linearly correlated. The coexistence of the three sites was also proven by cluster analysis based on similarities between Kd, binding capacity, and Hill coefficient.
Only minor systematic species and cell type differences occur in these properties. The value of Kd for the oxytocin receptor in rat myometrium, derived recently from a stepwise irreversible inhibition of uterotonic response to oxytocin, is close to 2.5 X molfliter. Additional pharmacological data (PA2 values of structural analogues of oxytocin acting as competitive inhibitors) also reveal a & value of 3 X It is, therefore, concluded that the receptors for oxytocin in rat myometrium are identical with the medium affinity site.
The interaction of oxytocin with its receptors in the rat, sheep, and human uterus has recently been the subject of several studies. In all of them, the dissociation constant of the intermediary hormone-receptor complex (&), in the presence of high Mg2+ and Mn2+ concentrations (up to 10 mM), has been reported to be about 2 X lo-' mol/liter (1)(2)(3)(4)(5). This value has been considered to be very likely, since the concentration of oxytocin which causes a half-maximal response on the isolated rat uterus strip (ECM) is numerically similar. Additional pharmacological evidence, however, indicates a considerably weaker binding. So, for instance, K d resulting from dose-response analysis after partial irreversible inhibition of the receptors on isolated rat uterus is approximately 2.5 X mol/liter (6, 7). Similar  These discrepancies caused us to re-examine the binding isotherm of oxytocin on myometrial and endometrial cells. In order to avoid binding on sites that are inaccessible to the hormone in intact tissue and to better mimic the physiological conditions, we have carried out our experiments on dissociated and shortly cultured cells, at magnesium concentrations similar to extracellular values.

MATERIALS AND METHODS AND RESULTS'
Model of the Binding Equilibrium and Computatwns-Relation between bound and free concentrations of a ligand at a constant concentration of the binding macromolecule (and at a constant temperature) is referred to as "binding isotherm." For current models of binding equilibria on one single binding site population (i.e. a population in which binding energies of individual sites, and consequently also Kd values, are distributed around a single mean value according to a distribution law of statistical thermodynamics), such a binding isotherm is frequently approximated by the equation where B, is the binding capacity of the site and h the Hill coefficient. Mutual interaction of the sites within this population is reflected in the value of h ( h # 1 for sites displaying cooperative behavior). Parameters in Equation 1 can be assessed by optimalization of the power coefficient h within a preselected interval. The procedure has been described earlier (15).
An isotherm which describes binding of a ligand in a system with several binding site populations (a common situation in any biological system) is formulated as a sum of terms corresponding to individual populations. Thus, when Equation 1 is applicable to all populations of such a multisite system,  under the conditions that all binding sites with K d j < K d , k are already saturated and that the binding on sites with K d j > K d , k is minimal within the cf range in question. The linearized form (4) enables optimalization of the parameter h k and computation of parameters Bo, Bm,k, and K d , k in a similar way as for yields straight lines only within cf ranges around individual K d values (16); in a broader cf range, the plot is curvilinear (Fig. 1). Vice versa, linear segments in Equation 5 can be employed for computation of the binding parameters. To find these segments, the data were ordered in a series with ascending cf values. The shortest sequence investigated contained five points and was successively extended by one point each time. Thus, the fit was carried out with points 1 through 5 , l through 6, etc. to 1 through n, then with points 2 through 6, 2 through 7, etc., until all subseries ((n -3) (n -4)/2 in number) were computed. The computation itself consisted of two steps: (i) the estimate of constants K d , B,, Bo, and h as mentioned above; (ii) computation of the correlation coefficient r for linearized Scatchard plot (Equation 5). Optionally, the points within the groups can undergo smoothing by a polynomial of second to fourth degree. Distinct populations can be identified when the ratio of their K d values (larger-tosmaller) is at least 50-100. An example is shown in Fig. 2.
The results indicate that K d values obtained from those sequences of points which give an approximately linear plot according to Equation 5 appear only within certain "peaks" (right-hand panel). These peaks obviously represent individual populations of binding sites; their mean binding parameters (geometric means) were computed from all groups of data points which yield a significantly linear plot according to Equation 5 (at least on the 5% probability level). The plot of related K d and E , values, called the "affinity spectrum" (17) of the system (Fig. 2, lower right), shows the binding capacity of these sites.
Analogous binding profiles were obtained by the recently published "affinity spectrum method" which detects, in certain instances, individual binding site populations even more directly (17). (These computations were performed by Dr. H. J, Tobler, Sandoz AG, Basel, Switzerland.) The method also enables an estimate of the so-called "nonspecific binding" (operationally, the binding which is not displaceable by an excess of "cold" ligand), in this case less than 0.04% of the total. It neglects, however, possible inequality of their h values. The upper left-handpanel in Fig. 2 demonstrates that the fit yields differing h values for these sites.

DISCUSSION
Sheep, rat, and calf myometrial cells carry several types of binding sites for oxytocin. Those sites differ in their thermodynamic and kinetic properties and most probably in biochemical properties (e.g. metal activation (3)) as well. The binding capacities of the three populations identified by analysis of the binding isotherm were found to correlate, without any known cause, with the equilibrium dissociation constants (Fig. 4). Displacement of oxytocin from the two sites with highest affinity follows the predicted model (Equation 6), whereas the linearization by Equation 10 failed for the third site, for which K d = mol/liter (cf. Table I). The low affinity site displays apparent departures from the regular displacement process; the population may consist of "nonspecific," although perhaps saturable, binding elements, which bind oxytocin in a noncomplementary way.
The phenomenon of distinct classes of binding sites on target cells is not uncommon for neurohypophyseal and also other peptide hormones. Two distinct classes of binding sites for oxytocin have already been found on rat epididymal adipocytes (21); they seem to initialize antagonistic effects on glucose uptake and metabolism in these cells (22). Their Kd values, 1.1 X IO-' and 7.5 X mol/liter, are very similar to those on myometrial cells (vide supra). Arginine vasopressin was also reported to bind to two binding sites in rat brain membranes (23), with Kd values of 4.2 X 10"O and 1.3 X lo-* mol/liter, the lower one only being detected at pH 8 after repeated freezing and thawing of the membranes, or by adding 5 mM NiCl,. Despite the complex pH and metal activation that may cause difficulties in interpretation of binding profiles, the coexistence of several binding sites can be proven also in this instance.   These sites are "specific" in the conventional sense of this word; the tracer is displaceable by the nonlabeled peptide. It seems very likely that multiple populations of specific binding sites for neurohypophyseal hormones exist in many, if not in all, cellular systems responding to these agents. However, the majority of reports does not explicitly mention more than one single receptor population. Are our experiments inconsistent compared to the others, or are there discrepancies in the evaluation of these experiments?
As shown in the literature (17, 24) and under "Results," recording of the whole spectrum of binding sites in the given biological system requires a study of binding and/or displacement in a broad range of ligand (tracer) concentrations. Experiments carried out in a narrow concentration interval may reveal only a part of the spectrum. Binding sites derived under these circumstances constitute only an accidental selection from the population of all sites; their assignment to a receptor may be erroneous. However, no generally valid recommendation can be made in this respect. Recent reports (17) suggest a ligand concentration span of 4-5 orders of magnitude. This requirement can be met only when using tracers of varying specific radioactivity. With a few exceptions, such conditions have not been respected in the earlier experiments. It wouldn't be amiss to say that also the methodology presented in many reports was not optimal, one meets frequent misinterpretations in evaluation of the binding studies, in the use of linearization procedures, in derivation of binding constants, and the like. Still, the high affinity binding site seems identical in all instances; the answer to the first part of the question posed above is, therefore, definitely no, to the second part basically no.
Several circumstances may indeed account for a heterogeneity which might be only apparent, i.e. associated with larger departures from the assumed models. Receptor internalization, for instance, may represent such a deviation.
Although not yet identified in the uterine cells, it is rather likely to exist there, and if so, it may exercise an effect on steady-state measurements of K d which is not fully predictable. On the other hand, several observations, including the similarity of intact myometrial cells and their membrane preparations with regard to the binding site profiles: speak circumstantially against a major influence of this potential mechanism.
Which binding sites may now be delineated as receptors in the pharmacological sense, i.e. units which initiate a cellular response? It became almost common practice to rest the proof D. Crankshaw   In qualitative terms, however, a simple and long known experiment demonstrates it quite lucidly; when the total number of receptors in a responding biological system is stepwise decreased by irreversible inhibition, the EC, value increases while the K d value should remain unchanged (27). This phenomenon of "receptor reserve" is also known for the response of uterus to oxytocin (28, 29), its EC50 value corresponding to a subtotal inhibition of oxytocin receptors is around

M,
compared to lo-' M in noninhibited muscle. Having used an improved method (7) by Furchgott and Bursztyn (6), we have computed an average K d = 2.5 X mollliter from these partially inhibited responses. This allows the conclusion that the in vitro uterotonic response to oxytocin in rat, at low magnesium concentrations, is triggered on these sites. There are not great differences in binding for the three species investigated, and it is likely that a similar conclusion also holds for calf and sheep uterus which have not yet been thoroughly investigated pharmacologically. Rather low affinity for oxytocin is indeed not astonishing when one considers the effect of receptor reserve resulting from amplification within the stimulus-response coupling. In fact, under these circumstances, high affinity binding would appear uneconomic from both thermodynamic and regulatory viewpoints. The system discussed here is certainly no exception in this respect, and low affinity hormone-receptor binding may be a very common feature. Oxytocln tritiated on p r l t l o n r 3 and 5 of 2-tymrlne was prepared In collaborat l o n r l t h h r r h a n Internatlonal. Cardlff

29.
Kd-valuer occurs smund the Id% 3 x ml/l, and p l y n m i l l s (3rd to 5th degree) f l t t r d t o the nldpointr shw a flattenlng wlth a . d e l r with p I 2 (eq. 12) fit the exponential models (eq. 11 and 12) rho" only negligeable numeric differences.
hiqhcr eancentrations. Oespite I s i g n i f i c a n t %-tern.