Radius of gyration of multis.

ubunit
macromolecules: apphcatmn to
myosin heads, myosin rod and
whole myosin
Juan A. Soivez, Angel Iniesta and Jose Garcifi de la Torre*
Departamento de Quimica Fisica, Facultad de Ciencias, Quimicas y Matematicas, Universidad de Murcia,
30001 Murcia, Spain
(Received 30 October 1986; revised 3 April 1987)
The radius of gyration, R, of macromolecules composed by subunits of arbitrary shape can be easily obtained
flora the radii of gyration of the subunits and the position of their centres. The procedure can be applied also to
obtain R for particles having cavities, as illustrated for a model of myosin head. B'hen the macromolecule
presents segmental flexibility, R must be calculated as a conformational average. As an example of the
procedure, we analyse the experimental values of R for whole myosin and myosin rod. Our analysis supports
strongly the existence of a small but appreciable flexibility in the myosin rod.
Keywords: Radius of gyration; myosin; conformation; flexibility

Introduction
In the low-angle region, the techniques of electromagnetic
scattering (light, X-rays or neutrons) by macromolecules
in solution are particularly easy to interpret. The
dependence of scattered intensity on scattering angle is in
part determined by the radius of gyration, R, which is a
measure of the global size of the macromolecule. When
combined with other physical data like molecular weight
andpartial specific volume, it can give information about
the macromolecular shape.
Well-known formulas for R are available for simple,
symmetrical shapes such as ellipsoids or cylinders.
However, in the field of biological macromolecules, one is
often facing rather complex, irregular structures that
cannot be described adequately by such simple models.
This situation has motivated much theoretical work
aimed at the interpretation of hydrodynamic properties
(sedimentation, diffusion, viscosity) of complex biological
macromolecules 1. R is a practical alternative to
hydrodynamic measurements; indeed, its sensitivity to
size and shape should be comparable to that of the
sedimentation or translational diffusion coefficients.
For macromolecules of arbitrary shape, R can be
calculated from a model in which point-like or in which
small, spherical elements are distributed throughout.
Such is the kind of model needed to compute the angular
variation of the scattering intensity using Debye's
formula. This approach can be nearly exact if the model
contains a very high number of scattering elements, so
that the shape of the macromolecule is accurately
described, but at the cost of much computational effort.
Models composed of spherical elements have the
advantage of allowing the joint interpretation of
hydrodynamic and X-ray results; interesting illustrations
of joint modelling are available in recent literature 2,3.
* To whom correspondence should be sent.
0141-8130/88/010039-05503.00
~) 1988 Butterworth & Co. (Publishers) Ltd

In this paper, we explore the possibility of modelling

the macromolecule for the calculation of R, using a

moderate number of building blocks of known radius of
gyration and various shapes, like spheres, cylinders or
ellipsoids. In some instances, these blocks may actually
correspond to functionally significant subunits. Then,
from the experimental R of the whole macromolecule,
and the R (or dimensions) of the subunits, one can infer
plausible structures for the spatial arrangements of the
subunits. Concretely, we consider two alternative
formulae for R of multisubunit structures. The formulae
can be also applied to particles having cavities, as
illustrated here for a model of myosin head. Another
particular situation is that of segmentally flexible
macromolecules, for which R is a conformational
average. A typical example, considered in this paper, is
whole myosin.

Theory
The particle under study is assumed to be composed of n
subunits whose radii of gyration are Ri, i = 1. . . . . n. If fi is
the mass fraction of subunit i and ei is the distance from
the centre of mass of the subunit to the centre of mass of
the particle, R can be calculated as
R 2= ~ fi(c~+R 2)

(1)

i=l

There is an alternative formula which does not require

the previous calculation of the centre of mass of the
particle, but only the distances r o between the centres of
all the i,i-pairs of subunits:

R 2=

f, R2+E 2 f , flr2j
i=l

(2)

i<j

Equations (1) and (2) are not reported in a recent

monograph on X-ray scattering 4. After deriving these

Int. J. Biol. Macromol., 1988, Vol 10, F e b r u a r y

39

Radius of gyration of multisubunit macromolecules: J. A. Solvez et al.

Z

Applications to myosin proteins

Myosin is a typical example of a macromolecule in which
rigid subunits can be clearly differentiated. Furthermore,
the joints at which the myosin subunits are connected are
usually supposed to be flexible. The segmental flexibility
of myosin plays an essential role in muscle contraction 9.
A schematic model for the myosin molecule is depicted
in Figure 1. Two myosin heads (subfragment S1) are
connected to the myosin rod. The rod may not be
straight; instead, it is assumed to contain a small, 'soft'
region which can act as a flexible joint (joint J1 in Figure
1), and divides the rod into two subfragments, $2 and
LMM. The heads are also flexibly joined to the $2
subfragment of the rod by means of the joint denoted J2 in

s1

~y
x r"

$2

Figure 1.

Figure 1 Model for whole myosin in an instantaneous

conformation. Subfragments S1, $2 and light meromyosin
(LMM) are indicated. The orientation of the heads are specified
by polar angles 01, ~'l (shown) and 02, ~2 (not shown), referred
to a system of Cartesian coordinates such that the rod ($2 plus
LMM) lies on the y, z plane. The conformation of the rod is
described by angle ~t
equations we were kindly informed by a referee that
they were presented in a review article on ribosome
structure by Damaschun et al. 5, quoting previous, lessknown works of their group 6-s. In the Appendix of this
paper we give the derivation of equations (1) and (2).
The procedure based on equation (1) would be as
follows. From the geometry of the particle, the position
vectors of the centres of the subunits with respect to an
arbitrary origin are evaluated first. Next, the centre of
the particle, C, is calculated using the subunit volumes
and densities. Then, the position vectors of the subunits
are referred to C. Finally, these positions and the
individual radii are substituted into equation (1) to find R.
The alternative procedure based on equation (2) is
more appealing because the intercentre distances can be
evaluated directly from the geometry, and the centre of
the whole particle is not needed. However, the calculation
from equation (2) requires a number of operations
proportional to n 2. In special cases in which n is very
large, computing time may be appreciably longer than in
the other procedure, for which the number of operations
is proportional to n.
For a particle composed by only two subunits,
equation (2) takes a simple form:

R 2 = f l R2 q-f2R 2 +flf2r22

Int. J. Biol. Macromol., 1988, Vol. 10, February

Myosin heads
Garrigos et al. 11 have proposed that chymotryptic,
light-chain-2(LC2)-free S1 has the shape of a prolate
ellipsoid with a hole near one of its extremities. This hole
would be the place where LC2 is located. For simplicity, it
was assumed that the hole is also ellipsoidal, and has the
same axial ratio as the main ellipsoid. The model is
depicted in Figure 2, where its geometrical parameters
are defined. With p = 2 . 3 , a * = 6 . 0 n m , d = 1.5 nm and
~ = 4 . 0 n m , the theoretical results for a number of
hydrodynamic properties and the angular dependence of
X-ray scattering were in excellent agreement with
experimental data 11.
Equation (3) can be used to obtain the radius of
gyration of a particle having a hole. If RF, R and RH are
the radii of gyration of the filled particle, hollow particle
and hole, respectively, and VF, V and VH are their
volumes, with V= VF--VH, it follows from equation (3)
that
R2

2
2
= RF/f-R~LfH/f-rl2 zfH

(4)

a ~

,=

- -i

--

(3)

which is known as the Parallel Axes Theorem of solution

scattering.
Equations (1), (2) and (3) tell us that R depends on the
dimensions of the subunits and the position of their
centres but, if the latter are fixed, R does not depend on
their orientation. Let us suppose, for instance, a dimer
composed of two rods of length L and diameter d joined
side-to-side with their axes parallel. For this model,
r12=d and RE=R2=L2/12 +d2/8 so that RE=L2/12+
3d2/8. Now, if one of the rods is arbitrarily rotated around
its centre, so that the dimer takes a cross-like shape, R is
unchanged because r12 remains constant.

40

Hydrodynamic properties of myosin fragments and

whole myosin have been extensively used to learn about
myosin structure and flexibility (see for instance, Ref. 10).
In this paper, we shown how data on the radii of gyration
can be useful for the same purpose.

,8

Figure 2 Model for LC2-free Sl. a* and b* are the semimajor

and semiminor axes of the ellipsoid, respectively. The axial ratio
is p = a*/b*. The ellipsoidal hole is centred at a distance d from
the centre of the ellipsoid. Its major and minor semiaxes are ct
and fl, respectively, with the same axial ratio, p=ct/fl, as the
main ellipsoid

Radius of gyration of multisubunit macromolecules: J. A. Solvez et al.

50

40

ely-

30

20
0

I
30

I
60

I
90

I
120

I
150

180

0 (A), a (B, C, D) (degrees)

Figure 3

Radius of gyration, R, of whole myosin. Curve A: R

versus 0t=02 for Ls2=LLMM=72nm, Ct=0 and $t=~b2=0.

Curve B: R versus ot for LS2=LLMM=72nm, 01=02=30 ,
~b~=~k2=0. Curve C: variation of the value of R,
conformationally averaged over the orientation of the heads,
with the fixed angle ~t. Ls2 = LLMM= 72 rim. Curve D: The same
as for curve C with Ls2 =43 nm, LLMi = 113 nm

where f = V/Vv and fu = Vu/VF are the volume fractions

(equal to mass fractions since we are assuming uniform
density), and the distance between the centres of the hole
and hollow particle is r12 = d/f. Using expressions for the
volume and radius of gyration of ellipsoids, we obtain, in
the notation of Figure 2,
R 2 = [(p2 + 2)/5](b.S _ fls)/(b.a _flu)
_ d2b ,afla/(b,a _ fl3)2

(5)

Due to an arithmetic error, equation (8) in Ref. 11 differs

slightly from the correct expression, equation (5).
Repeating the calculations we have seen that the effect
from this error on the previous analysis of the shape of
LC2-free S1 is very small. For the parameters given above,
equation (5) leads to R = 3.3 nm, which coincides with the
experimental value ~2.
It must be pointed out that the size and shape of myosin
heads are objects of much recent controversy t-16. Our
purpose here was only to illustrate the applicability of
calculations of R in the case of a specific model.

We note that the actual shape of S 1 is not relevant in the

calculations for the whole molecule. The length of rodlike
subfragments $2 and L M M are not known accurately.
Two rather extreme choices are Ls2 =/-~MM = 72 nm and
Ls2 = 43 nm, and LLMM= 113 nm. The diameter of $2 and
L M M is taken as 1.7 nm, which is compatible with
experimental data for the molecular weight of the rod,
and slightly smaller than the hydrodynamic radius 7.
An instantaneous conformation of myosin is specified
by the polar angle of the heads, 01, ~k1, 02, ~k2, and the
angle in the rod, ct. These angles are shown in Figure 1.
The experimental radius of gyration of the myosin
molecule, R~xp= 45-47 nm (Refs 17 and 18), corresponds
to a conformational average over these angles, Thus it
seems possible to obtain information on myosin flexibility
by comparing R~xpwith calculated values. However, the
dependence of R on the conformation of the rod is much
more pronounced, as shown by curve B in Figure 3.
It is instructive to calculate R firstly for instantaneous
conformations of the model, i.e. for fixed angle. Some
results are presented in Figure 3. Curve A is for a straight
rod (ct=0) with Ls2=LLMM=72nm and varying
orientation of the heads. It shows that R depends very
weakly on the conformation of the heads. In fact, the
change in R between a fully extended conformation (0 t =
02'-'30 ) and a retracted one ( 0 1 = 0 2 2 1 5 0 ) is about

10%.
According to current views of myosin flexibility, the
mobility of the heads is rather extensive, so that J1 is
nearly a universal joint. Therefore, the average over 0~,
~bl, 02, ~k2 can be a simple, unweighted one. We note that
if the flexibility of joint J1 were somewhat restricted, the
error introduced by the small influence of the heads'
conformation on R would be small. Then, we decided to
carry out the average by numerical simulation to avoid
cumbersome algebra. For fixed angle, ~, in the rod, about
400 allowed conformations of the heads were generated
picking at random 01 and 02 with a sine distribution, and
ff~ and ~k2 with a uniform distribution. A conformation is
allowed if there is no overlap between the four subunits.
This procedure was repeated for varying values of ct, and
the results are presented in curves C and D of Figure 3.
In a simple approach 1 the flexibility of the rod is
characterized by a certain equivalent angle, Ct~q,such that
the value of a given radius calculated for ~q, and averaged
over the random orientation of the heads, coincides with
the experimental datum. Thus, Ct~qcan be regarded as
some mean bending angle. The result would be Ct,q- 0 if
the rod was rigid and straight. However, values of ~,q of
up to 60 have been found in the analysis of
hydrodynamic properties 1. In the present study, ~q is
estimated interpolating R~xpin curves C and D. We find
atoq---20-45 with curve C for LS2= L L M M = 72 nm, and
~ q - 600-70 with curve D for Ls2 = 43 nm and LLMM=
113nm. These values, as well as those from
hydrodynamic properties, are significantly different from
0 and therefore confirm the existence of a moderate
flexibility of the myosin rod.

Whole myosin
In the calculations for whole myosin, subfragment S1
is considered as a particle with a volume of 142nm a, and
a radius of gyration of 3.3 nm, as discussed in the previous
section. Its centre of mass is located at a distance of 8 nm
from joint J1. This distance is obtained adding an excess
of 2 nm to the long semiaxis of S1, which is about 6 nm.

Myosin rod
The flexibility of the myosin rod can be analysed with
more detail in the rod itself (headless myosin) since it has
only two subunits and a single joint. The proper model is
a broken, or hinged rod having two rodlike subunits of
length L1 ( = Ls2) and L 2 (-----L L M M ) . The radius of gyration

Int. J. Biol. Macromol., 1988, Vol 10, February

41

Radius of oyration of multisubunit macromolecules: J. A. Solvez et al.

of the broken rod can be obtained from equations (3) with

fl=LJL=7, where L = L t + L 2 is the total length,

f 2 = 1 - 7 , R2=L~/12, (d2/8,~L2/12), R2=L2/12, and
r22 = L2/4 + L2/4 + L 1L2 c o s a/2. The result is
(R2)=(L2/2)[1/6+72(1-7)2((cos~-l)]
(6)
where ( - - - - ) denote conformational average. We note
that only conformation-dependent quantity is cos ~. A
formula equivalent to equation (6) was derived by Garcia
Molina and Garcia de la Torre 19 though in a more
indirect way.
Hvidt et al. 2 have measured the radius of gyration of
the myosin rod, obtaining values in the range 37-41 nm
with a negligible temperature dependence. From
equation (6) we find (cos a) = 0.45-0.93 if Lsz = LLMM =
72nm and (cos a) < 0.29 if L s 2 = 4 3 n m and LLMM=
113 nm. These values deviate to a lesser or greater extent
from the rigid limit (cos a) = 1, and is again a proof of the
flexibility of myosin rod. For the first choice, the rod is
quite stiff (Ct~q=20~50) while for the second one its
behaviour is almost completely flexible (~q> 75). We
recall that these two choices are extremes. As R increases
with increasing L and decreasing 7, the first choice
corresponds to lowest R, while for the second one has the
highest value. Thus, the parameters obtained for the two
choices of length should be regarded as lower and upper
bounds for the true values of myosin.
Flexibility of the joint in the rod can be quantified by
means of an elastic potential
V(a) = k~ 2

(7)

where k is the elastic constant. Then, the conformational

average needed for R is (Ref. 19)

Acknowledgement
This work was supported by grant 561/84 from the
Comision Asesora de Investigacion Cienfifica Tbcnica.

References
1
2
3
4
5
6
7
8
9
I0
11
12
13
14
15
16
17
18
19
20

<cos a> =

fo o

d ~ cos a sin ~ e - q "

/fo o

dct sin ct e-~2

(8)

Appendix

with

Q=k./2kbT

(9)

where kbTis Boltzmann factor. In Ref. 19, the upper limits

of the integrals in equation (8) where rt instead of ~ .
However, the influence of this difference in the resulting
values of (cos ~) is negligible except for very low Q. The
plot of (cos ~) versus Q is practically the same as that in
Fioure 1 of Ref. 19. Interpolating the (cos ct) values for
myosin rod obtained from equation (6), we find Q = 0.5-8
if Ls2 = LLMM= 72 nm and Q < 0.3 if Ls2 = 43 nm and
LLMM= 113 nm.
The agreement between the values of the flexibility
parameter, ~q, of myosin and myosin rod is better for
L S 2 = LLM M ~---72 nm than for Ls2 = 43 nm, LLMM= 113 rim.
For the first set of lengths the myosin rod is found to be
rather stiff, but not straight, while for the other set the
flexibility is perhaps too high. We should note that our
analysis is based on a single property, namely, the radius
of gyration, which depends on several geometrical and
conformational parameters of the molecule, and therefore
we cannot ignore the possibility of over-interpretation of
our model. Nonetheless, different choices of those
parameters lead to coincident qualitative conclusions:
our analysis of the radius of gyration of myosin and
myosin rod confirms the moderate but appreciable
flexibility of the rod.

42

Garci/l de la Torre, J. and Bloomfield, V. A. Q. Rev. Biophys.

1981, 14, 81
Perkins, S. J. Biochem. J. 1985, 228, 13
Perkins, S. J. and Sire, R. B. Eur. J. Biochem. 1986, 157, 155
Glatter, O. and Kratky, O. (Eds) 'Small Angle X-ray Scattering',
Academic, New York, 1982
Damaschun, G., Muller, J. J. and Bielka, H. Meth. Enzymol.
1979, 59, 706
Damaschun, G., Muller, J. J. and Purschel, H. V. Acta Biol.
Med. Ger. 1968, 20, 379
Damaschun, G., Fichtner, P., Purschel, H. V. and Reich, J. G.
Acta Biol. Med. Get. 1968, 21,308
Damaschun, G. and Purschel, H. V. Acta Biol. Med. Ger. 1970,
24, 59
Harvey, S. C. and Cheung, H. in 'Muscle and Nonmuscle
Motility', (Eds R. M. Dowben and H. Cheung), Plenum, New
York, 1982, p. 279
Garcia de la Torre, J. and Bloomfield, V. A. Biochemistry 1980,
19, 5118
Garrigos, M., Morel, J. E. and Garcia de la Torre, J.
Biochemistry 1983, 22, 4961
Mendelson, R. Nature 1982, 298, 665
Mendelson, R. Nature 1985, 318, 20
Highsmith, S. and Eden, D. Biochemistry 1986, 25, 2237
Bachouchi, N., Gulik, A., Garrigos, M. and Morel, J. E.
Biochemistry 1985, 24, 6305
Craig, R., Trinick, J. and Knight, P. Nature 1986, 320, 688
Holtzer, A. and Lowey, S. J. Am. Chem. Soc. 1959, 81, 1370
Herbert, T. J. and Carlson, F. D. Biopolymers 1971, 10, 2231
Garcia Molina, J. J. and Garci/i de la Torre, J. Int. J. Biol.
Macromol. 1984, 6, 170
Hvidt, S., Chang, T. and Yu, H. Biopolymers 1984, 23, 1283

Int. J. Biol. Macromol., 1988, Vol. 10, February

The radius of gyration, R, of a particle which occupies a

volume V is given by

R2= fv r2p(r) d~ / fv p(r) d~

(A1)

where p(r) is some density at a point P whose position

vector with respect to the particle's centre, C is r. This
centre has the property

vrp(r) dz = 0

(A2)

Ifr' and e' are the position vectors of the general point P
and the centre C, respectively, with respect to an arbitrary
origin 0, so that r' = e ' + r, we have from equation (A2)

c'=fvr'p(r')d~/fj(r')d~

(A3)

Now we assume that the particle is composed of n

subunits. The volume, centre and radius of gyration of
subunit i are denoted as V, C; and R~, respectively. For
simplicity, we restrict ourselves to subunits of uniform
density. If pi is the density of subunit i, equations (A1) and
(A2) lead to
R 2=

(tlfv )/(tl gi)

Pi

r 2 dz

(A4)

Radius of gyration of muhisubunit macromolecules:J. A. Solvez et al.

and

and
p

C-----

r d'r

fvridr = 0

i
i

The centre of subunit i, C~, is placed at

e [ = f r' dx/Vi

(A6)

,d V~

so that
e ' = ~. f,c~

(A7)

i=1

(A8)

For the particular case in which all the subunits have

the same density, f~isjust the volume fraction of subunit i.
Particularizing equations (A1) and (A2) for subunit i
we have

g~=~,fv,r~d,

where ri is the vector from Ci to P. Now, if ei is the

position vector of C i with respect to C, we have r = c~+ r~.
Substituting this into equation (A4) and using equations
(A9) and (A10) we finally find equation (1).
To derive equation (2) we note that it follows from
equations (A2) and (A10) that

Z f,ei=O

(All)

and therefore

where

f i= piVi/ (i~=l piVi)

(A10)

(A9)

~. f.ff~cj = 0
i j
Now, if

(A12)

r0=c i-ci

(A13)

the law of cosines gives

e~cj-_ (ci2 + cj2 _

r~)/2

(A14)

and substituting equation (A14) into equation (A12) and

equation (1) we obtain equation (2).

Int. J. Biol. Macromol., 1988, Vol 10, February

43