TERRELL L. HILL 145 ASPECTS OF POLYMERIZATION IN PROTEINS OF THE MUSCLE FIBRIL BY T.-C. TSAO AND K. BAILEY Biochemical Laboratory, Cambridge Received 12th May, 1952 Molecular studies are reported for the three main proteins of the myofibril, tropo- myosin, myosin and actin. The structure of the tropomyosin particle is discussed and aspects of its polymerization compared with that of actin. The much larger myosin particle has been studied largely from the standpoint of the subunits it contains, and the conditions under which they are liberated by fragmentation of the whole molecule. The proteins of the myofibril are being studied in many laboratories from differ- ent aspects, sometimes from the enzymic standpoint, sometimes from the physico- chemical, and all studies are largely directed to an understanding of the nature of the interaction of the contractile proteins with each other and with adenosine triphosphate.It has become quite clear, from the very different approach of Weber’s group in Germany,l and from Marsh’s study,2 carried out with the authors’ group in Cambridge, that ATP serves two distinct functions: its presence is a necessary condition of relaxation phenomena, and its splitting, that of contraction. In Cambridge it has been shown that muscle contains an inhibitor of the latter process, and that the stimulus is somehow connected with the release of this in- hi bition. So far, these are factual, descriptive statements for which no meaning exists on a molecular level, and it seems all-important at present to understand the size, shape and make-up of the components of the myofibril before we can begin to propose theories of the mechanism of contraction.We shall deal with the three proteins which as far as we know make up the whole, or at least the bulk, of the myofibril : tropomyosin, myosin and actin ; and inasmuch as molecular studies in all three cases cannot be divorced from the interesting features of polymerization phenomena, we shall discuss mainly this aspect. TROPOMYOSIN.~-T~~S protein, which exists in skeletal, cardiac and smooth muscle, has several unusual properties. Though soluble in aqueous solvents after isolation, it is not readily shed into the muscle press juice, nor into salt ex- tractants which are capable of withdrawing the myosin component. It resembles actin in that it remains largely with the stroma protein.It is readily obtained if muscle fibre is first treated with ethanol and ether, and then extracted, not with water, but strong salt solution. The work of Hamoir 4 with fish muscle suggests that in sifu it may exist as a ribonucleic acid complex, which would explain the necessity for salt as inducing a process of metathesis. But the treatment with organic solvents suggests also complex-formation with lipid. In many properties, tropomyosin is like myosin ; many of the amino-acids are present in comparable amount, many of the physicochemical properties are similar, and like myosin the intramolecular pattern is of the a-keratin type. Analytically, the absence of tryptophan, and the presence of larger amounts of lysine and146 POLY MERIZATlON I N PROTEINS glutamic acid, certainly distinguish it from myosin, as does the lower particle weight (53,000 against 850,000).The total charged groups represent 45 % of the total residues, of which 27 % of the total residues are anionic. No protein is known with a higher mixed valence than this. Although tropomyosin is more soluble in water than myosin, it nevertheless has globulin properties, which suggest that the charge distribution is not symmetrical. This fact, coupled with the asymmetry of the molecule, probably explains why tropomyosin solutions freed of salt become so very viscous, due simply to the end-to-end aggregation of particles. That the ag- gregation is purely electrostatic is shown by the fact that the addition of salt to salt-free sols causes an immediate drop in viscosity, and this can be very large, from yrel of 50 to one of about 3 at I = 0.1 ( c = 0.8 %).It would be interesting to know the dipole moment of such a molecule. The particle weight of tropomyosin under various conditions 5 is given in fig. 1, i.e. (i> in neutral salt solutions of varying ionic strength, (ii) in neutral urea and (iii) in acid at pH 2. The conditions in (ii) and (iii) are such as to induce FIG. 1 .-Particle weight of tropomyosin under various conditions. depolymerization and were expected to reveal whether the particle which undergoes polymerization in neutral salt solutions is capable of fragmenting into subunits. It can be seen that the average particle weight by osmotic pressure measurements in salt solutions tends towards that in the depolymerizing solvents.That the primary particle, which we shall call the monomer, does not contain subunits, received confirmation in other ways. From the intrinsic viscosity, obtained under varying conditions, it was deduced that the monomer of 53,000 possessed an asymmetry (unhydrated) of about 30. From the X-ray data and chemical analysis, and accepting the Astbury model for the cc-fold, it is possible to calculate the asymmetry of a single, double or triple chain model, which works out at about 78, 26 and 13 respectively. Only the double chain model thus seemed acceptable, and this was all the more favoured when end group assay of the N-terminal amino acids showed none to be present.6 A double chain model involving spiralization as in the Pauling-Corey hypothesis7 gives less good agreement with viscosity data (about 19), but we would not suggest that our data can distinguish between them. The simplest interpretation from all the evidence was a cyclic chain of a configuration, disposed like a deflated bicycle tube hanging from a point on its circumference. The length is 385A approximately and the width 14.5A.T .- C . TSAO AND K . BAILEY 147 The reasons for the polymerizability of tropomyosin are not far to seek. The trend of the viscosity measurements as salt is removed shows (table 1) that up to the trimer stage the process is end-to-end. But if salt-free sols are dried down on a glass plate to facilitate aggregation, and the aggregates washed off on to a supporting film, electromicrographs show very large fibrils, 3000-6000 8, long and 200-300 8, wide.This shows that side-to-side aggregation is also possible, and will, of course. be encouraged by the preliminary end-to-end process. The first stages must involve the aggregation of particles which are essentially bipolar at neutral pH, and such linear aggregates are then capable of aligning side by side. TABLE 1 . -SHAPE OF RABBIT-TROPOMYOSTN PARTICLES IN SOLUTION axial ratio (Simha equation) medium intrinsic viscosity viscosity increment hydrated solvent and ionic strength unhydrated (25 % water) 6.5 6.5 6.5 6.5 6-5 2.1 6.5 12.0 NaCl (0.1) NaCl (0.2) NaCl (0-3) NaCI (0-6) NaCl (1.1) HC1 (0.3) urea (0-3) NaOH (0.3) I -40 1.00 0.70 0-59 0.57 0.523 0.523 0.39 197 53 45 141 44 37 99 35 30 83 32 27 80 31 26 74 30 25 74 30 25 55 25 21 Still unexplained is the fact that after dissolving in concentrated urea and dia- lyzing, solutions of tropomyosin are much more viscous than before, and after aggregation by drying on a glass plate, form tremendously long aggregates (2-6,~) which have a fairly uniform width.As we have seen, urea does not fragment the monomer, but it does induce some change, because although tropomyosin treated with acid at pH 2 or alkali at pH 12 will crystallize isoelectrically at ionic strength 0.4, it will not do so after contact with urea. What significance the aggregation of particles by electrostatic forces holds for questions which relate to the biogenesis of natural fibre molecules is not known.It could be the initial orienting mechanism which is later implemented by stronger secondary valence forces along the length of the aggregates, or even by covalent bonds. MYOSIN.-Turning now to myosin, the best data on the particle weight are those of Weber’s group: 8 s;;O = 7.1 x 10-13, D = 0.87 x 10-7, whence M r= 860,000 i 30,000, a value checked by osmotic pressure measurements.9 Conventional interpretation of the frictional ratio in terms of shape suggests a molecule about 2300 A long and only 23 8, thick. It was observed that solutions of myosin tend to aggregate in a series of steps, that of s = 15 x 10-13 being particularly stable. This aggregation, which Weber calls denaturation must, from the large increase in sedimentation constant, involve an increase in the dia- meter of the myosin particles ; but whether it is a physical process, or due perhaps to the formation of intermolecular disulphide bonds between the many cysteine residues which myosin contains has not been explored. What significance, if any, this type of aggregation has in muscle is not known.For the present, we are more concerned with the make-up of the particle of 850,000 molecular weight. These undoubtedly are built up of smaller units, as was first indicated by Weber and Stover 10 and later by Snellman and Erdos,ll and it is only by following the time course and conditions of fragmentation that we can hope to obtain in- formation on the mode of biogenesis. If myosin is dissolved in concentrated urea, fragmentation (at room temperature) is quite a slow process.When kept for a month and ethanol added in small aliquots, the main fraction (I) precipitates between 20 and 35 % (v/v) ethanol, but thereafter a small fraction (In continues148 POLYMERlZATION I N PROTEINS to precipitate up to 70 % ethanol. After removal of urea, fraction I is a syneres- ing gel, insoluble in water and salt, fraction I1 a water-soluble protein, which when separated from some contamination with fraction I, constitutes 4-5 % of the original myosin. Fraction I1 is not split off from myosin immediately after dis- persing in urea, though the exact time course of its liberation has not been followed in the very early stages. After storage for 6 months, the behaviour towards ethanol has changed : no fraction appears until the ethanol concentration is 65 %, so that the now-modified fraction J and fraction I1 come out together, and the whole of the protein is water- soluble.No further change can be detected between 6 months and 2 years, the threshold concentration of ethanol and the intrinsic viscosity of the system remaining constant. At this steady state, fractions I and 11 can be separated by ammonium sulphate fractionation, and it is then found that fraction I1 con- stitutes 8 % of the total protein. Though by the action of urea fraction I1 is broken off the myosin particle, there is no evidence that fraction I, whether in its water-insoluble or -soluble form, is depolymerized. The particle weight of the soluble form in water at pH 7 determined by the polarization of fluorescence technique of Weber,R 16 is very large indeed, and its sedimentation constant in urea approximately 8.7 x 10-13, though the material is rather polydisperse.However, depolymerization of fraction I readily occurs when the pH is changed to a value between 10 and 1 1 , a range in which the lysine €-amino groups are presumed to have lost their charge. The average particle weight both by osmotic pressure and polarization techniques is then about 170,000 (table 2). On this evidence, the change of solubility of fraction I cannot be due to a depolymerization process. Whatever the initial action of urea, there would seem to follow a slow intramolecular rearrangement of groups, such that the polar side chains are freer to interact with water, or have lost their asymmetry of distribution, or both. The assay of N-terminal groups in native myosin by Sanger’s method 6 does not allow more than a total of about 1 group in 500,000, and this in reality com- prises three different amino acids.It was thought that these were assignable to an impurity, and that myosin, like tropomyosin, might be built up of cyclopeptides. After long treatment in urea, or in alkali at pH 10.7 (table 2) the proportion of N-terminal groups does not increase, showing that fragmentation is not due to the rupture of covalent bonds. The bulk of the N-terminal amino acids are in fact found in fraction 11, and amount to a total of 1 group in 16,000; this value agrees with that obtained by physical methods (table 2). From the physical evidence given above it seems likely that fraction I1 is an intrinsic component of the myosin particle.First of all, electrophoresis and sedimentation show the myosh used to be monodisperse, and the slow liberation of fraction I1 by urea does not sug- gest that it is a loosely held impurity; nor is it due to the presence of actin, which interacts specifically with myosin, because the average particle weight of actin treated with urea under comparable conditions is quite large (about 74,000). The overall picture that we have of the myosin particle is one consisting of 4-5 units, probably cyclic, of average particle weight 170,OOO, associated with 4-5 open polypeptides of fairly low average particle weight (about 15,000). These latter are split off only by the action of concentrated urea leaving the larger units, which we may regard as the framework, still associated.In alkali, however, the framework itself breaks up into its component parts, and if the open chain components have not already been split off by urea, they remain combined with the larger units. It is premature to speculate upon the types of linkage which exist between the various subunits in the myosin molecule, but the depolymer- ization which accompanies the loss of charge of the +amino groups suggests that these latter play a part in holding the framework together. A similar mechanism has been put forward for serum albumin.12fraction and depoly- merizing conditions myosin: in 6.7 M urea, 0.06 M phosphate, pH 6.5, kept for 6-24 months at room temp. myosin FRACTION I1 ethanol-urea frac- tionation after 1-2 months in conc.urea ammonium sulphate fractionation after 6-24 months in conc. urea FRACTION I not separated ethanol-urea frac- tionation after 1-2 months in conc. urea ammonium sul- phate fractionation after 6-24 months in conc. urea T . - C . TSAO AND K . BAILEY TABLE 2.-DIMENSIONS OF MYOSIN FRAGMENTS average particle wt. + axial solvent ratio by no O.P. 1 lP viscosity 6.7 M urea, 0.6 M phosphate, pH 6-5 0.1 M borate, pH 10.7 0.06 M phosphate, 0.2 M KCI, pH 6.5 water, pH 7 6.7 M urea, 0.06 M phosphate, pH 6.5 6-7 M urea, 0.1 M phosphate, 0.1 M KCI, 0.05 M thio- glycollic acid, pH 7.0 water, pH 7 0.1 M borate, 0.1 M KCI, pH 10.7 170,000 - 14,000 - - 17,000 122,000 - - > 300,000 165,000 - - 175,000 149 -30 1 terminal amino- group/900,000 g protein -40 1 terminal amino- group/500,000 g protein -10 1 terminal amino- group/ 16,000 g protein - * ~ ~ ~ = 8 .7 x 10-13 (C = 0.4 %) -40 still including 4 % of fraction I1 -30 - f O.P. by osmotic pressure measurements. * kindly measured by Dr. A. G. Ogston. 1 / p by fluorescence polarization measurements. AcTm-The mechanism of the polymerization of G-actin to F-actin has been the subject of numerous speculations 13 both from the chemical and physico- chemical standpoints. The characteristic properties of F-actin are the high, anomalous viscosity, flow birefringence and high sedimentation rate, which have all been interpreted as due to the linear aggregation of globular units into long filaments. An apparent confirmation of this mechanism has been afforded by the fibrils seen under the electron microscope.14~ 15 A more detailed study of the polymerization by Weber’s fluorescence-polarization method 16 seems to suggest that such large units do not exist in solution, and alternative explanations may need to be sought for the properties given above.Whatever the final interpretation, we are mainly concerned at present with the dimensions of the primary unit, and of its participation in a monomer-dimer transformation. The actin used was prepared by a new method, and unlike the Straub pre- paration, was electrophoretically homogeneous. Various preparations were subjected to osmotic pressure measurements in 0.6 M KI, which is sometimes used as a depolymerizing agent in protein chemistry.17 The results were found150 POLYMERIZATION I N PROTEINS to be somewhat erratic, indicating in some experiments the presence of a 74,000 unit, at other times one of 140,000, and in very dilute solution the transition of the latter to the former with progressive dilution.An analysis of the condi- tions employed in these experiments has not entirely explained the results, but did suggest that the presence of ATP encouraged the formation of the monomeric state. This was confirmed by the application of Weber's polarization-fluorescence technique. If typical F-actin preparations are depolymerized in simple buffers at pH 2-2 or 10 or in neutral 0.6 M KI, or if G-actin 18 with or without the addition of traces of Mg ion are examined, the polarization of fluorescence indicates a rota- tional relaxation time p corresponding to the dimer.G-actin in neutral urea or in buffer (without urea) at pH 11, or at pH 8 in presence of ATP, possesses relaxation times intermediate between that of dimer and monomer. The lowest relaxation time is obtained in presence of a chelating agent (versenate) at temper- atures above 20" C ; preliminary measurements of the intrinsic viscosity suggest an axial ratio greater than 10, and this order of asymmetry combined with the value of p = 13.8 x 10-8 (at 25" C) gives a particle weight of 70,000, corresponding to the monomeric form indicated by the osmotic pressure measurements. Below 20" C in presence of versenate, the monomeric form passes over to the dimeric, and a similar change of state with temperature occurs in presence of ADP or hexametaphosphate.The above results suggest rather strongly that the units of the dimeric form of actin are held together by a divalent metal, linked possibly through the nucleo- tide prosthetic group. ATP, ADP, pyrophosphate and polyphosphate all com- pete for the metal and cause dissociation, versenate, an excellent chelating agent, being most efficient. The dissociation which is effected at high pH is probably due to electrostatic repulsion. DISCUSSION The polymerization-depolymerization phenomena brought about under such diverse conditions in these three proteins of the myofibril are interesting to compare. The features which are most important in such comparison are the reversibility of the process, the types of primary unit involved, and their structural features.In the case of tropomyosin, the polymerization is freely reversible and electrostatic in nature, and the units participating are possibly cyclic in structure and possess no subunits. Myosin, on the other hand, appears to con- tain two types of subunit, one possibly cyclic and the other an open chain, each requiring different conditions for their liberation ; in contrast with tropomyosin, this process is slow and irreversible. It is not easy to prove that the open chain components of the myosin particle are an intrinsic part of the molecule, and in reality it is difficult to decide the meaning of purity with large molecules which possess a specific physiological function. It must be remembered that myosin possesses an enzyme character, and as an enzyme also possesses a uniquely high molecular weight.Whether the enzyme property resides in the " framework " or in the small complement of open chains is a very relevant question, and might influence our views on what constitutes the myosin particle as a physiological unit. Quite independently of such considerations, the forces which hold the framework together, and the open chains to the framework, seem to be of different character. Whereas salt effects the disaggregation of tropomyosin, with actin it elicits all the properties which ar rather typical of polymeric molecules-high, anomalous viscosity and flow birefringence. Though we have cause to question the accepted structure of F-actin, it seems fairly certain that it is formed in two separate stages ; 19 first, the primary unit or monomer, itself asymmetric, passes to a dimeric state in which divalent metal ions are important.These ions may act as a bridge throughT . - C . TSAO AND K . BAILEY 151 the ADP or ATP prosthetic group which have been shown always to occur in actin preparations. It does not seem probable that they act by modifyingthe charge on the protein through adsorption : mercury in organic combination with one valency free at once diminishes the viscosity of F-actin and prevents the formation of the latter from G-actin. (We do not consider that mercury acts by combination with SH groups, or that SH groups are important in the polymer- ization of actin.20) The formation of F-actin, whatever its nature, involves only the interaction of the dimeric form.The formation of the dimer, o r of the monomer from the dimer, is a relatively slow process under the conditions so far studied, and in time scale is comparable to the formation and depolymerization of the mercury-linked dimer of serum albumin ; 21 it differs very notably from the rapid transformations observed in tropomyosin. We are very much indebted to Mr. G. S. Adair, F.R.S., and Dr. G. Weber for help with methods and for valuable discussions. The research has been financed by grants from Prof. W. T. Astbury, F.R.S., and from Imperial Chemical Industries, to whom we are greatly indebted. 1 Weber and Portzehl, Adv. Protein. Chem., 1952, 7, 161. 2 Marsh, Biochim. Biophys. Acta (in press). 3 Bailey, Biochem. J., 1948, 43, 271. 4Hamoir, Biochem. J., 1951, 50, 140. 5 Tsao, Bailey and Adair, Biochem. J., 1951, 49, 27. 6 Bailey, Biochem. J., 1951, 49, 23. 7 Pauling and Corey, Proc. Nut. Acad. Sci., 1951, 37, 235. 8 Portzehl, Schramm and Weber, 2. Nuturforsch., 1950, 5b, 61. 9 Portzehl, 2. Nuturforsch., 1950, 5b, 75. 10 Weber and Stover, Biochem. Z., 1933, 259,269. 11 Snellman and Erdos, Biochim. Biophys. Acta, 1948, 2, 650. 12 Weber, Biochem. J., 1952, 51, 155. 13 summarized in ref. (1). 14 Jakus and Hall, J. Biol. Chem., 1947, 167, 705. 15 Rozsa, Szent-Gyorgyi and Wyckoff, Biochim. Biophys. Acta, 1949, 3, 561. 16 Weber, Biochem. J., 1952, 51, 145. 17 Szent-Gyorgyi, J. Biol. Chem., 1951, 192, 361. 18 This does not imply an identity with the actin which is obtained by direct extraction of acetone-treated muscle fibre; it is obtained by dialysis of polymerized actin against 0.002 M NaHC03. 19 Feuer, MolnAr, Pettko and Straub, Hung. Physiol. Actu, 1948, 1, 150. 20 Turba and Kuschinsky, Biochim. Biophys. Actu, 1952, 8, 76. 21 Hughes, J. Amer. Chem. Soc., 1947, 69, 1836.