Interactions of oxytocin and vasopressin with bovine neurophysins I and II. Effects of hormone binding on the protein quaternary structure: a simple model.

The effects of hormone binding on the reversible monomer in equilibrium dimer equilibrium of bovine neurophysins I or II in solution have been studied by sedimentation equilibrium measurements performed in conjunction with equilibrium dialysis experiments. Under normal solution conditions saturating amounts of oxytocin displace the neurophysin dimerization equilibrium toward the associated form of the protein to give a dimeric complex with two oxytocin molecules bound per dimer. Vasopressin exerts different influences on this oligomerization process. At low fractional saturation this ligand exhibits a behavior similar to oxytocin with a higher affinity for the neurophysin dimer than the monomer. But in contrast, at higher fractional saturation, vasopressin strongly displaces the aggregation equilibrium toward a monomeric complex bearing two vasopressin molecules. However, in the presence of a high concentration of LiCl two oxytocin molecules are bound per neurophysin protomer (10,000 daltons). These observations, together with earlier data for vasopressin binding, suggest that each neurophysin molecule possesses two structurally distinct hormone binding sites. These observations can be rationalized in a simple schematic model of hormone binding to neurophysin in which oxytocin favors a dimeric form with one hormone binding site available per 10,000 daltons while vasopressin favors the monomeric form with two hormone binding sites available per 10,000 daltons.

The effects of hormone binding on the reversible monomer + dimer equilibrium of bovine neurophysins I or II in solution have been studied by sedimentation equilibrium measurements performed in conjunction with equilibrium dialysis experiments. Under normal solution conditions saturating amounts of oxytocin displace the neurophysin dimerization equilibrium toward the associated form of the protein to give a dimeric complex with two oxytocin molecules bound per dimer. Vasopressin exerts different influences on this oligomerization process. At low fractional saturation this ligand exhibits a behavior similar to oxytocin with a higher affinity for the neurophysin dimer than the monomer. But in contrast, at higher fractional saturation, vasopressin strongly displaces the aggregation equilibrium toward a monomeric complex bearing two vasopressin molecules. However, in the presence of a high concentration of LiCl two oxytocin molecules are bound per neurophysin protomer (10,000 daltons). These observations, together with earlier data for vasopressin binding, suggest that each neurophysin molecule possesses two structurally distinct hormone binding sites. These observations can be rationalized in a simple schematic model of hormone binding to neurophysin in which oxytocin favors a dimeric form with one hormone binding site available per 10,000 daltons while vasopressin favors the monomeric form with two hormone binding sites available per 10,000 daltons.
Analysis of the binding of the nonapeptide hormones oxytocin and vasopressin to their carrier proteins, the neurophysins, has been the subject of several conflicting reports in recent years (l-6). Studies conducted by various authors were done using different techniques with different relative sensitivities and accuracies, such as equilibrium dialysis (l-6), NMR (7-ll), and CD (12, 13) spectroscopy. The situation was also complicated by the fact that the protein appeared to possess properties of a self-associating system in solution (14,15 contribution of cooperativity to the binding process; and (d) the existence of differences between oxytocin and vasopressin in the mode of complex formation (2, 12). Although most authors seemed to agree that, at equilibrium, only one oxytocin molecule could be bound per 10,000 daltons (1, 3), work in this laboratory had first shown that two vasopressin molecules could be bound per 10,000 daltons with apparent similar affinities'(1, 2). This unusual difference between the behavior of oxytocin and vasopressin binding has not yet been clearly explained.
In order to clarify this problem we have further investigated several aspects of the binding process using equilibrium dialysis under various conditions in connection with corresponding rigorous measurements of the apparent weight average molecular weight exhibited by the protein. hormone complexes.
The observations we report here strongly suggest that each neurophysin molecule of 10,000 daltons possesses two distinct hormone binding sites which can be occupied by oxytocin or vasopressin under certain conditions. We show that the relative capacity of oxytocin, or vasopressin, to bind at the second hormone binding site is correlated with specific differential effects of hormone binding on the neuro- Neurophysins-Highly purified bovine neurophysins I and II were prepared as previously described (1) by isoelectric focusing from an acetone powder of freshly collected bovine pituitaries.
The samples routinely tested for lipid content were found to contain no more than 0.5% by weight of glyceride derivatives as judged by gas chromatographic analysis of the fatty acid methyl esters produced after alkaline hydrolysis of the samples (16). Neurophysin samples were routinely tested for homogeneity using gel electrophoresis, gel isoelectric focusing, and amino acid composition.
Hormone-Oxytocin and vasopressin were generous gifts of Sandoz (Basle). The tritiated hormones (10 to 30 Ci/mmol) were also prepared as previously reported (17,18) and routinely tested for their pharmacological activities and their radiochemical purity by the usual electrophoretic and chromatographic tests (17,18). Equilibrium Dialysis-Hormone binding studies were run at 24", pH 5.60 (0.1 M sodium acetate buffer), as described in detail previously (1). Refined analysis of the Scatchard plots required that a minimum of 9 or 10 different hormone concentrations be tested (from 5 x IO-' M to 10e3 M) at each protein concentration.
Each point was run in quadruplicate.
Counting of the radioactive samples was done by means of a liquid scintillation spectrometer (Intertechnique SL 30). Concentrations of neurophysins were evaluated on a Cary 118 C spectrophotometer assuming an tH = 3400 cm-' M-' at 260 nm. Fractional saturation ratios V were expressed as bound ligand (C,) concentrations per protein molar concentration considering the molecular weights of neurophysins I an d II equal, respectively, to 9,560 and 10,041 (19,20). C, is the free ligand concentration.
Sedimentation Equilibrium Experiments-The measurements were made using a Beckman Spinco model E analytical ultracentrifuge and an equilibrium cell with two compartments.
Protein, or protein plus hormone, samples were put in one compartment while the buffer alone, or the hormone in the buffer, was in the other compartment. Neurophysin concentration was 0.4 or 0.8 mg/ml in the absence of ligand, and 0.5 mg/ml in the presence of peptide. Ultracentrifugation was performed at 20" at 30,000 rpm for 5 days. After 96 hours, the solute distribution was checked to see whether equilibrium was reached. The protein distribution was obtained using Rayleigh interference optics. The interferometric optics were aligned using the method of Richard and Schachman (21). Photographic plates were read using a Nikon microcomparator model 6C by measurement of five fringe coordinates taken at each 0.1 mm. Data from these studies were analyzed by the meniscus depletion method of Yphantis (22), by plots of ln concentration against RZ with a WANG 720 calculator using programs developed by P. Dessen and G. Batelier. These experiments were run on samples dialyzed 18 hours against 0.1 M sodium acetate buffer, pH 5.60, at 5". When the experiments were run in the presence of ligand, the protein, in the presence of the hormone, was dialyzed against the same hormone solution for 18 hours at 5".
Partial Specific Volume-For all experiments in 0.1 M sodium acetate buffer partial specific volumes of 0.709 for neurophysin II and 0.706 for neurophysin I were used for calculations (23). When the ultracentrifugation was run in 1.4 M LiCl, the partial specific volume U of the protein was determined in the following way: a solution of 6.5 mg/ml of protein was thoroughly dialyzed against 0.1 M sodium acetate buffer plus 1.4 M LiCl, pH 5.60. The difference of density between the protein solution and the dialysis buffer was measured with a digital Anton Paar model DMA 02 C densimeter carefully thermostated at 20.20 * 0.01". Circular Dichroism Spectra-Circular dichroism spectra were recorded in a Jouan dichrograph III, with a sensitivity of lo-' A units/mm, using 0.1.cm optical pathway cuvettes and a protein concentration of 0.5 mg/ml.

Sedimentation
Equilibrium FIG. 1. High speed sedimentation equilibrium of native bovine neurophysin II in 0.1 M sodium acetate buffer, pH 5.60, at 20": 0, in the absence of oxytocin (initial concentration of protein 0.8 mg/ml); 0, in the presence of a saturating amounts (lOms M) of oxytocin (initial concentration of protein 0.5 mg/ml). As the latter experiment was done at 28,060 rpm instead of 30,000 rpm for the former, the points were replotted for 30,000 rpm with arbitrary position along the AR2 axis. Inset, replots of the curve (0) in terms of weight average molecular weight (Mw) versus concentration.
The solid line represents the theoretical variation of Mw uersus concentration calculated assuming an apparent monomer + dimer equilibrium with an association constant X, = 5.8 x 10s Mu'. Concentrations-When the binding of oxytocin (from 5 x lo-' M to 10m3 M) to neurophysin II at various concentrations ranging from 5 pM to 0.3 mM was measured and the data plotted according to Scatchard (24), typical curves were obtained (Fig. 2). The type of curvilinearity observed is characteristic of ligand binding-coupled effects on a selfpolymerizing protein (25)(26)(27)(28). In all cases the maximal observed values of the fractional saturation were close to 1.0. Since the previously measured effects of oxytocin on the i?Iw of the protein indicates that, at saturation, the neurophysin .oxytocin complex is dimeric, this proves that the dimeric form of neurophysin II under these conditions possesses two hormone binding sites occupied by oxytocin.
In addition, the observation, from the sigmoidal plots of Fig. 2, that the average slope of the curves increases with increasing protein concentrations strongly confirms the previous evidence that the affinity of oxytocin for the neurophysin dimer is higher than for the monomeric form of the protein. Binding of (Lys*) Vusopressin to Neurophysin II-Refined analysis of the data for vasopressin binding to neurophysin II indicated the existence of curvilinearity in the lower and upper parts of the Scatchard plots (Fig. 3). As in the case of oxytocin some positive cooperativity could be observed at low 5 values (i, < l), and the upper part of these plots is essentially similar to those obtained for oxytocin. In the second part of the binding isotherms, there is observed for high V values (V > l), a reverse effect indicative that neurophysin II could be saturated by the binding of a second hormone molecule. This result was repeatedly observed with different biologically active preparations of tritiated vasopressin. However, in our hands, it was found that the capacity of binding a second vasopressin molecule was lessened or eventually was no longer detected when samples of tritiated vasopressin with a decreased biological activity were tested. In that case, a behavior very much like that of oxytocin was observed, indicative that structural changes occurring in these hormone samples led to a loss of the vasopressin character of the molecules.
Clearly these observations confirm previous results that using fully biologically active vasopressin under carefully controlled conditions, neurophysin II can bind two vasopressin molecules per protein monomer of 10,000 daltons (1). The affinity of the second peptide ligand molecule appears to be lower than that of the first molecule.
Taken together those features for vasopressin binding to neurophysin II and the previous observations made with oxytocin strongly suggest that ligand binding may have different effects on the proportion of either the associated, or dissociated, neurophysin species in solution at different V ratios.
Equilibrium Sedimentation Studies of Neurophysin ZZ in Presence of (Lys') Vusopressin-To test for this possibility, the weight average molecular weight (Rw) of neurophysin II was measured in the presence of vasopressin concentrations corresponding respectively to b = 0.9 and I, = 1.5 on the Scatchard plots (neurophysin II concentration was 50 FM) (Fig. 3). For such an experiment, vasopressin (lo-" M) was added to the neurophysin II solution (initial concentration 0.5 mg/ml) at pH 5.60. Fig. 4  In contrast, when the same initial concentration of neurophysin II was exposed to a lo-fold excess of (Lys") vasopressin ( 10m3 M), the curve obtained indicated clearly that the proportion of dimeric complexes was significantly lower (Fig. 4). Up to a protein concentration of 0.5 mg/ml the complex was essentially monomeric. Thus, at a protein concentration of 0.5 mg/ml, the conditions of the equilibrium dialysis where the protein alone is expected to have a l?Iw = 12,500, the ??Iw of the complex was 14,300 + 715. But in the lower part of the liquid column, in the centrifugation cell, the proportion of the higher molecular weight species increased, since the vasopressin:neurophysin ratio decreased. In conclusion, vasopressin behaves very much like oxytocin for low vasopressin:neurophysin ratios, i.e. exhibits a higher affinity for the dimer. But the opposite effect was detected as the proportion of vasopressin increased, and the complex was found to be predominantly monomeric in solution.

Sedimentation Equilibrium Measurements of Molecular
Weight of Neurophysin Z-At pH 5.60, the plot of In concentration of neurophysin I versus AR2 gives also a curve typical of a self-associating system and the inset of Fig. 3 represents the observed values of the weight average molecular weight (nw) at each concentration.
The solid line represents the theoretical variation of iGlw with concentration for a monomer * dimer equilibrium with an association constant of 7.7 x 10' Mm'. The behavior of both neurophysins I and II was essentially similar except that the Rw increased slightly faster with increasing protein concentration in the case of neurophysin I compared with neurophysin II, indicative of a higher monomer-monomer affinity in the first case.
Binding of (Ly.9) Vasopressin to Neurophysin Z-In this case, the Scatchard plots were slightly different from those of neurophysin II (Fig. 3). The curvature was less marked and only a small difference in the slope of the binding isotherms was observable at high versus low i ratio. This suggests that the ligand has probably very similar effects on the oligomerization of neurophysin II and neurophysin I, but that for the latter, vasopressin may have less pronounced effects on the capacity of the protein to bind a second hormone molecule.  (29) recently reported on the dimeric nature both of the hormone .neurophysin II complex and of the protein alone from simple measurements of sedimentation velocities. These authors claimed that ligand binding induced only conformational rearrangements of the pre-existing dimeric protein. This is in contrast with the previous report of Breslow et al. (15) of a monomer * dimer equilibrium and with the above reported evidence that ligand drastically affects this equilibrium. One possible explanation for these discrepancies may reside in the fact that these authors worked with an initial protein concentration sufficiently high to have a large proportion of dimer in solution.
It is clear that even in binding.measurements conducted at very low protein concentration (0.5 or 0.05 mg/ml) where the proportion of dimer in the absence of ligand is relatively small (Fig. I), some sign of positive cooperativity could be detected in the, binding isotherms. These phenomena in multisubunit protein systems are known to be due to modulation of the strength of subunit interactions coupled to ligand binding (27)(28)(29)(30). In some cases, ligand binding promotes association of the subunits, but in others, the opposite effects can be detected. The above reported observations demonstrate that oxytocin strongly displaces the monomer + dimer equilibrium of the protein in favor of the associated form. Thermodynamically this effect is brought about by the fact that the hormonal ligand has a higher affinity for the dimer than for the monomer.
This conclusion is substantiated by the observation that the apparent binding constant is higher, the higher the neurophysin concentration, consequently the proportion of dimer. This gives a satisfactory explanation to the reported discrepancies between the Ka values previously measured in our laboratory (1) at low neurophysin concentrations (0.5 mg/ml) compared with those of other authors (3, 4) carried out at protein concentrations close to 2 or 6 mg/ml. Since a maximum of one molecule of oxytocin could be bound per protein fraction of M, = 10,000 at saturation, and considering the fact that, for this value of I, the protein is predominantly in the associated, dimeric form it is clear that the neurophysin dimer possesses two occupied hormone sites at saturation.
The situation in the case of (Lys') vasopressin binding is rather more complicated since opposite effects on the protein quaternary structure could be observed over the saturation process. At low values of i, clearly the behavior of (Lys8) vasopressin is very much like oxytocin and the hormone exhibits a preference for binding to the dimeric neurophysin. Consequently a large proportion of dimeric complex (Mw = 22,000) was found in the presence of vasopressin and when values of b were smaller than 1.0. In contrast, at a ligand concentration closer to the 2:l hormone to protein stoichiometry, a reverse process was induced by ligand binding.
In that case the vasopressin 'neurophysin complex was obviously monomeric.
It is now possible to describe the noncovalent association of oxytocin and (Lys8) vasopressin to bovine neurophysin II at pH 5.60 as a system where ligand binding is coupled to the association of two monomers with four available binding sites into a dimer possessing two available sites. The scheme of Fig. 6 represents this phenomenon tentatively.
K,, K',, and K, are, respectively, the association constants for the successive equilibria, and X, is the equilibrium constant for the dimerization reaction. The essential features of this schematic representation are the following: (a) the existence of individual monomeric protein molecules possessing two structurally distinct and, possibly, thermodynamically nonequivalent sites; (b) the possibility that these monomers associate noncovalently as a dimer made of two equivalent protomers and possessing only two sites, these two sites exhibiting a higher affinity for oxytocin than the original sites of the nonassociated protomers; (c) it is presumed that two out of the four sites of the dimer, in some fashion and possibly consecutive to conformational rearrangements associated with dimerization, are made unavailable to ligand binding. This is a very simple representation of hormone association to the oligomerizing protein. This model was tested and values of the association constants K, and K, were evaluated in the case where the ligand is oxytocin (see "Appendix"). They were, respectively, found to be equal to K, = 5 x 10" Mm', K, = 2.5 x lo5 M-'. However, for very low V values attempts to get the best curve fit with the model lead to rather poor results, suggesting that the cooperativity associated with dimerization cannot account for the curvature observed at P < 0. 25 6. A possible schematic model of oxytocin and vasopressin binding to bovine neurophysin II. M is the monomer, D is the dimer, L is oxytocin, and L' is vasopressin. In this model it was assumed, to simplify the equations, that the two binding sites on the dimer are structurally and thermodynamically equivalent, and that K, or K, are identical for the 2 mol of bound ligand. X, is the dimerization constant for the unliganded protein and K,, K',, and K, are the association constants ef the described reactions. Only vasopressin is presumed to be able to give the ML', form.
oxytocin. Given this approximation, a value of K', was obtained and found to be equal to 0.8 x 10' MM'.
Although we have no experimental evidence that neurophysin I quaternary structure is similarly affected by ligand binding, it seems reasonable to assume that most of the steps involved in the present model are also applicable to neurophysin I. In this scheme, it can be seen that only the DL, form can be attained in solution in the case of oxytocin. It was not possible to assess the existence of a monomeric neurophysin bearing two oxytocin molecules. However, results obtained with LiCl suggest that the binding of two more oxytocin molecules may occur on a rearranged dimer. In the case of (Lys8) vasopressin the monomeric form ML', was shown to exist in solution at high ligand binding ratio. This can be rationalized, referring to the scheme proposed in Fig. 6 where fM and fD are the fractions of protomers in the monomer and dimer, respectively. In the proposed model for hormone binding to neurophysin, assumptions are made on two classes of independent sites per protomer: (a) one class of sites exhibiting a higher affinity for the ligand in the dimer than in the monomer (K, and K, are the respective affinity constants and K, > K,; (b) another class of sites in which occupation by the ligand is in competition with the association areas of the monomer (the association constant is K', in this case 10.