CHAPTER XL

THE STRUCTURE OF VIRUSES AS DETERMINED BY X-RAY
DIFFRACTION*

Rosalind E. Franklin, D. L. D. Caspar, and A. Klug

HE PIONEER X-RAY INVESTIGATIONS by Ber-
T nal and Fankuchen (5, 9) were among the
earliest and most informative physical studies
on tobacco mosaic virus (TMV) . This work was
undertaken on material provided by Bawden and
Pirie (5) within a year of the first isolation and
purification of TMV by Stanley (89). Likewise
when the first spherical virus was crystallized
(6), the preliminary X-ray investigations (11)
followed soon thereafter. The first X-ray photo-
graphs of TMV taken over twenty years ago
demonstrated that the individual virus parti-
cles are built up of subunits arranged in a regu-
lar way (IO), but this observation made little
impression on the general direction of research
into virus structure until fairly recently. Vi-
ruses were looked at for some time rather as
giant molecules. Physico-chemical studies were
concerned largely with characterizing a virus
particle in terms of its size, shape, molecular
weight, and hydration, while the biochemical in-
vestigations emphasized over-all composition and
chemical reactivity.

The first approach to the study of substructure
in viruses by physico-chemical methods was
made by Schramm, who, in work published in
detail in 1947 (83), described the alkaline deg-
radation of TMV. The breakdown proceeds in a
stepwise fashion until the small, relatively ho-

* The X-ray diffraction studies described here
have been facilitated by the cooperation of work-
ers in virus laboratories in the United States, Great
Britain, and Germany. The work at Birkbeck Col-
lege has been supported by the Agricultural Re-
search Council (R.E.F.), the Nuffield Foundation
(A.K.), and the United States Public Health Serv-
ice (Research Grant E. 1772; R.E.F. and A.K.).
The work at Yale has been supported by a grant
from the John A. Hartford Foundation to the
Biophysics Department. The preparation of this
manuscript has been assisted by a travel grant to
me of us (D. L. D. C) from the Rockefeller
Foundation.

mogeneous A-protein is produced which can be
separated from the ribonucleic acid (RNA) of
the virus. It was, however, not until 1952 that
the end-group analyses of Harris and Knight
(51) provided a clear indication that the pro-
tein part of TMV consists of a large number of
small, chemically identical protein molecules. It
is now possible to identify this chemical sub-
unit of TMV with the structural subunit "seen"
by X-rays. Substructure has also been revealed,
by both biochemical and X-ray analysis, in some
other small viruses which contain RNA. This
review will be concerned mainly with the struc-
tural studies on small RNA-containing plant
viruses.

The X-ray diffraction method
and virus substructure

We cannot here go into the principles of X-ray
diffraction techniques, and can hardly do more
than mention the method of analysis. The reader
wishing to know more about the X-ray analysis
of biological structures is referred to the excel-
lent account for the noncrystallographer by
Crick and Kendrew (22). X-ray diffraction is
particularly useful for studying any structure
which repeats regularly in space. The scale of
structure which can be investigated ranges from
a few angstroms, as in simple atomic and ionic
crystals, to hundreds of angstroms, as in virus
crystals. X-rays pick out the ordered part of a
structure in such a way that the diffraction dia-
gram of a substance is directly related to the
structural regularity within it. The order in a
structure arises from the regular arrangement
of smaller parts (asymmetric units), and it is
this arrangement that determines the position
of the reflections ("spots") on the X-ray dia-
gram. The important point is that the asym-
metric units are identical and all have the same
environment.

447

4-18 The Structure of Viruses

Crick and Watson (23) have suggested that
all small viruses are built up on a framework of
protein subunits packed together in a regular
manner. This hypothesis provides a simple ex-
planation of the regular shape of small virus
particles, and was based largely on the esample
of ThIV. In another paper (24)  reasons were
presented why one might expect small viruses
to be made up of subunits. The infectivity of a
rirus is carried by its RNA (27, 46) and the
function of the protein part is presumably to
provide some form of packaging for the specific
RNA. It appears that` the RNA "codes" the
virus protein, and it is reasonable to expect that
the RNA may only be able to "code" the pro-
tein in the form of small identical molecules.
Such protein subunits can then aggregate in a
regular may to form the protein shell or coat of
the virus.

Regular packing of a number of identical units
so that the same kind of contacts between units
are used over and over again will necessarily
result in a symmetrical structure. This means
that a virus particle can be constructed out of
subunits in only a Iited number of ways since
there are only a rather small number of kinds
of symmetry possible for a particle. A discus-
sion of symmetry cannot be undertaken here,
and for details the reader is referred to the pa-
pers of Crick and Watson.
The well-defined symmetry to be expected for
virus particles is very favorable for X-ray
structure analysis since the symmetry of the
particle is indicated in a relatively direct way
by the X-ray diagrams. Determination of the
symmetry does not indicate what the asym-
metric unit looks like. It does, however, tell us
how many there are and how they are related
and thus gives some indication of the size of the
structural subunit and the overall appearance
of the virus particle.
A physical limitation of the X-ray method is
that a very large number of particles must be
examined at one time in order to have enough
diffracting material to obtain a photograph. For
this photograph to indicate the symmetry of a
single virus particle, the particles must be regu-
larly arranged in space, either in a crystal for
the spherical viruses, or in an oriented gel
( "paracrystal") for the rod-shaped particles
like TMV.

The basic problem in X-ray structure analysis
is that the diffraction pattern itself does not con-

tain all the information necessary to completely
determine the structure which gave rise to it.
The ultimate goal of a structure analysis is to
calculate (by Fourier synthesis) the three di-
mensional electron density" distribution in the
asymmetric unit to a resolution compatible with
the quality of the experimental data. There is,
however, no standard or direct method by which
this can be done. Each X-ray reflection or spot
on the photograph corresponds to a diffracted
wave which is characterized by three quantities :
direction, magnitude, and phase. Of these, the
first two can be determined directly, but in tak-
ing the photograph all information as to the
phase angle is lost. In order to reconstruct the
electron density of the diffracting structure this
phase angle must somehow be determined. For
complicated structures there is only one gen-
erally certain way of doing this. This is the
method of isomorphous replacement (481, in
which the diffraction pattern is modified by
binding a heavy atom, such as mercury, in a
regular way to each asymmetric unit. The
changes of intensity in the pattern due to the
presence of the heavy atom may be used first to
determine the position of the heavy atom and
then to calculate the phase angles. It should be
noted that for the purpose of phase detennina-
tion it is not necessary to know the chemical na-
ture of the binding of the heavy atom to the
virus; all that is required is that it be specific,
and that the replacement is isomorphous, i.e.,
it does not significantly change the structural
arrangem~nt of the virus.

Although it now appears that the method of
isomorphous replacement affords the only hope
for determining the detailed internal structure
of a virus, it should be stressed that a great
deal of structural information can nevertheless
be obtained by fairly direct interpretation of the
X-ray diagrams. As discussed above, determina-
tion of the symmetry gives considerable insight
into the substructure of a virus; by comparing
certain parts of the diagram with predictions
based on simple models some of the significant
morphological features can be deduced, and
comparison of the patterns given by different
strains can also reveal general structural regu-
larities.

* For most biological structures, which are con-
stituted for the most part of atoms of low atomic
number, the electron density is approximatelY
proportional to the mass density.

w

MAXIMUM RADIUS

MEAN RADIUS

PLATE XL.-~.-A model of to-
bacco mosaic virus, based
mainly on the X-ray diffraction
studies of Franklin. About one-
tenth of the total length of the
virus particle is shown.
The virus protein is in the
form of a Iarge number of small,
identical subunits set in helical
array about the particle axis.
The structure repeats after 69 A
in the axial direction, and the
repeat contains 49 subunits dis-
tributed over three turns of the
flat, major helix of pltch 23 A.
The shape of the subunits as
represented is rather schematic,
but is such that the helical ar-
ray has a hollow core of diam-
eter 3540 A and each subunit
appears as a protuberance at
both the inner and outer sur-
faces, giving the particle a max-
imum diameter of 180 A.
The ribonucleic acid is deeply
embedded in the protein ar-
ray and is in the form of a sin-
gle long-chain molecule which
follows the line of the flat helix,
at a radius of 40 A. Some of the
protein subunits have been re-
moved from the model to show
the RNA, which is represented
schematically by a smooth coil.

42oxP RADIUS of HOLE

PLATE s~-2.--S-ray diffraction diagram (by Franklin) of an orientated gel of mercury-substitul
TJIV, common btrain, taken with a high resolution focusing camera and crystal-monochromatin
CuIi~l, radiation. The diagram is very similar to that of TMV itself already published elsewhere
43) but is of even higher aualitv.

d
d
0

-."

The virus particles have their long axes in the vertical direction and the pattern corresponds to
the cylindrically averaged intensity diffracted by a single particle.
The horizontal stratification (``layer-lines") corresponds to the axial repeat period of 69 d in the vi-
rus particle. The intensity distribution along a layer line does not consist of discrete reflections ag
shown by a true crystal, but is continuous with maxima showing up as spots.
The orientation of the specimen is particularly good as shown by the small degree of arcing of the
spots on the diagram.

B

PLATE x~i-3.--(.4) Schematic model illustrating one way in which 60 subunits may be packed according
to icosahedral symmetry to form the protein shell of a spherical virus. The subunits are arranged at
the vertices of a snub dodecahedron (B), which has the same rotational symmetry as the icosahedron
which it is shown inscribed.

Ping-pong balls have been used for convenience in building the model, but do not represent the
actual shape of protein subunits. Whatever shape is used in a model of this kind, the geometrical real-
tion between each subunit and its neighbors remains the same for all subunits.

L

5

2

A B

PLATE xL-4.--X-ray diffraction diagrams (precession photographs) of single crystals of (A) tomato
bushy stunt virus (IS), and (B) turnip yellow mosaic virus (63), taken ill each case with the X-ray beam
parallel to the edge of the cubic unit cell. Only the central region of each pattern is shown.
Many such photographs must be taken along different crystal directions in order to record the com-
plete three-dimensional diffraction pattern of a crystal.
The directions of the twofold, threefold, and fivefold rotation axes corresponding to icosahedral
symmetry, which lie in the plane of the photograph, are marked by arrows. The twofold axes arc parallel
to the cube edges of the crystal, but the other two sets do not lie along crystallographic axes. There is a
concentration of strong reflections ("spikes") in these directions. There are twice as many such spikes
in the TPMV pattern as in the BSV, because the virus particles in the crystal of TYXIV arc arranged
in two different orientations (rotated 90" to each other about the cube edge of the crystal). In tile BSV
crystal, all the virus particles have the same orientation.

ROSALIND E. FRANKLIN, D. L. D. CASPAR ~~ft A. KLUG

449

Tobacco mosaic vim

TMV is at present the only virus for which
&e X-ray structure analysis has been carried
far enough for us to be able to build a model
(plate XL-1) which gives a reasonably realistic
representation of the structure. Most of the in-
forniation which has gone into the construction
of this model has been obtained from the work
of Rosalind Franklin. Detailed structure analy-
sis requires X-ray photographs of very high
technical quality (Plate XG2), and these can
only be obtained from very well-orientated gels.
To illustrate the prospects and difficulties of the
X-ray diffraction method as applied to viruses,
we shall discuss the studies on TMV in some
detail.
The helicat structure of TMV.-The early
X-ray studies of Bernal and Fankuchen indi-
cated that TMV has a regular substructure, but
at that time the arrangement of subunits was not
dear. In 1954 Watson (91) indicated that the
essential features of the X-ray pattern could be
accounted for if the virus particle were built up
of a helical array of subunits set about the long
axis of the particle. In the axial repeat period
of 69 A there would have to be 3n -t 1 (where
n is an integer) subunits equally spaced along
three turns of the helix. Further X-ray studies
by Franklin (30) and Franklin and Klug (41)
completely confirmed this helical structure, in
particular by showing that certain unusual fea-
tures of the X-ray diagram (layer-line splitting)
could only be accounted for by this type of
structure.
The value of n is now known to be 16, so
that there are 49 subunits in three turns of the
helix or 16% per turn. Direct determination of
the value of n proved very difficult and the
earlier estimates were too small. This number
was determined by Frah and Holmes (39,
40), using a mercury-substituted TMV pre-
pared by Fraenkel-Conrat, in which a methyl-
mercury group is bound to the sulphur of the
single cysteine residue (2, 64) of each protein
subunit. The mercury atoms serve as markers
for the subunits and are necessarily distributed
on a helix with the same axial repeat as TMV.
From a detailed quantitative comparison of the
X-ray diffraction diagrams of the mercury-sub-
stituted ThlV and of normal TMV, it was possi-
ble to deduce that there are 49 mercury atoms
equally spaced along three turns of a helix of

radius 56 A. Therefore there must be the same
number of protein subunits in three turns.
On returning to the X-ray dagram of normal
TAW itself, and using the recently determined
maximum radius of the particle (17, 42, 43), it
was found that a self-consistent interpr~tation
of the parts of the diagram due to the outside
of the particle is obtained if there are 49 struc-
tural subunits in three turns of the helix. Since
these independent measurements refer to the
structural unit of the virus, and the measure-
ments on the mercury positions refer to the
chemical subunit, it appears that the structural
and chemical subunit are identical.
Further confirmation of the identity of the
TAN subunit demonstrated by chemical and
X-ray methods is provided by a comparison of
the molecular weight of the structural unit as
determined from the number per virus particle
with that of the chemical unit given by end
group analysis (14, 28, 52, 74) and amino acid
composition (64, 73, 80). A recent detailed
study of monodisperse TMV preparations by
Boedtker and Simmons (13) by a number of
physico-chemical methods gives a molecular
weight of 39 (21) x lo", and the particle length
of these same preparations as measured uith
the electron microscope by Hall (49) is 3000 i:
50 I%. These values agree very well with many
of the earlier measurements of the molecular
weight (68, 77, 84) and particle length (76, 97).
Since there are 49 subunits in a 69 A length of
the virus, there are 49 x 3000/69 = 2130 & 40
structural subunits in the particle of length
3000 2 50 A. Five to 6 per cent by weight of
TMV consists of RNA (64, 79), thus the mo-
lecular weight of the 2130 protein subunits is
37 (ztl) x lo6. Hence the molecular weight of
the protein structural subunits is 17,300 e 800.
This is in complete agreement with the molecu-
lar weight of 17-18,000 obtained by the chemical
methods. It appears likely from these results
that all the protein of TMV consists of small
identical protein subunits organized in the heli-
cal structure.

There is of course no structural reason why
a helix should contain an exact integral ratio
of subunits to turns, and 49/3 is merely the
nearest rational approximation. In fact it has
been shown (41) that the number of units in
three turns is not exactly integral, and that this
departure from an integer varies slightly from
strain to strain (32), corresponding presumably

450 The Structure of Viruses

to slight differences in the packing of subunits.
The effects are, however, very small and corre-
spond to a difference of only one to three sub-
units in the more than two thousand contained
in the virus particle of length 3000 -4.
dIorphology and ~nteTna1 ~~ru~ture of TMV.-
The diameter of TMV has for some time been
quoted to be 152 A on the basis of the X-ray
measurements (10) of the interparticle distance
in dry orientated TMV preparations. Inter-
particle distances measured by the electron mi-
croscope agree well with this figure, although
observers had occasionally been puzzled by the
fact that isolated particles seemed to be thicker
(95). It is now clear that the maximum diameter
of the particle is a good deal greater than 152 A
and that this figure represents the packing dis-
tance between particles intermeshed to form a
close-packed hexagonal array.
This conclusion was reached by Franklin and
I(l11g (42), who showed that the outer surface
of the virus could be represented by a helically
grooved model with a mean radius of 76 A, but
with a maximum radius of about 85-90 A. The
helical grooves are only a way of describing the
spaces between protruding subunits near the
outside of the virus.

The fact that the particle has a maximum ra-
dius greater than 76 A has been definitely
shown by the radial density distribution (17)
determined by the method of isomorphous re-
placement as described below. The best esti-
mate of the maximum radial extension of the
particle comes from some recent work of Frank-
lin (36) on the binding of osmium to TMV. It
is found that osmium atoms attach to TMV at
a radius of 90 A, indicating that there is still
some protein at this radius.
To obtain more information on the structure
it is necessary, as already mentioned, to resort
to isomorphous replacement. This was first suc-
cessfully accomplished for TMY by Caspar (16,
17) who was able to prepare a lead-substituted
TAN. From the changes produced in the equa-
torial X-ray scattering, it was found that equal
amounts of lead are bound at radial distances
of 25 and 84 A from the particle axis, though
only an average of about half a lead atom per
protein subunit could be bound to each of the
two distinct sites in the virus. The location of
the bound Iead made possible a definite deter-

mination of the signs* of those equatorial diffrac-
tion maxims which correspond to the cylindri-
cally averaged electron density of the virus.
Using these signs and the observed magnitudes,
the Fourier transform was calculated to give the
cylindrically averaged radial density distribu-
tion in the particle. The result is shown by the
full curve in Fig. XL-l(a).
Caspar's allocation of signs was confirmed by
a second successful application of the method of
isomorphous replacement to TMV by Franklin
(33, 43), using the mercury-substituted TMV
prepared by Fraenkel-Conrat. One can therefore
have complete confidence in the correctness of
the radial density distribution shown in Fig.
XL-la. It should, however, be mentioned that
the smali fluctuations which occur near the axis
are of no significance since in this region the ef-
fect of  experiment^ error is large.
The most significant features of the radial
density distribution are as follows: (a) The vi-
rus particle has a hole of diameter 35-40 A
extending along the axis; (b) The maximum ra-
dius of the particle is close to 90 A; (c) There
are regions of high density at radii of 25, 40,
66, and 78 A.
The axial hole of TMV is occupied by water
when the virus is in solution, and the inside sur-
face of the virus is as accessible to small mole-
cules as the outside. When TMV is dried from
solutions containing small solute molecules, these
molecules fill in or line the hole, and in this way
Huxley (57) has recently shown the hole di-
rectly in electron micrographs.
Electron density carries no chemical label, so
the location of the RNA cannot be determined
from the density distribution of TMV alone,
but studies on nucleic acid-free particles, de-
scribed below, demonstrate that the very high
density region about a radius of 40 A is due to
the RNA. The other well-defined maxima must
be due to the configuration of the polypeptide
chain which makes up the protein subunit. Al-
though some plausible suggestions regarding the
folding of the polypeptide backbone (30, 34, 91)
have been made, no definite conclusions can be
drawn from the present, limited evidence.

To obtain further information about the struc-

* For this part of the X-ray diagram the phase
angles are only 0 or 180". In this case each dif-
fraction maxima is characterized by a sign, either
plus or minus, and a single heavy atom substitu-
tion is sufficient to determine the signs.

ROSALIND E. FRANKLIN, D. L. D. CASPAR AND A. KLUG

451

RADIUS h

20 40 (0 eo IC

RADIUS I,

-FIG. x~1 .-(a) The cylindrically averaged radial density distribution in (i) tobacco
mosaic virus, common strain (full curve) (17, 43) ; (ii) repolymerized, nucleic-acid
free, tobacco mosaic virus protein (dotted curve) (33). The curves show the difference
between the electron density of the particles and that o$ water, plotted as a function
of radial distance from the particle axis. The high peak at 40 %I in (i) is absent in (ii)
and must therefore be due to the RNA in the virus. The actual difference in the 40 A
region is about 10 to 12 per cent, i.e., twice the percentage of RNA in the virus, because
the density of RNA above that of water is tmice as great as for protein. (b)-(e)
show the radial density distributions (55) in some strains of TMV: (b) Strain U2
(87); (c) Cucumber virus 4; (d) A strain of TMV taken from Nigerian cowpea and
grown in Burley tobacco plants (4); (e) The same as (d), grown in the Prince variety
of French bean.

ture of the protein subunits it is necessary to  rather than signs have to be determined, and for
investigate other parts of the X-ray diagram be-  this purpose at least two isomorphous replace-
sides the equator. Here, however, the diffraction  ments are necessary. High-quality photographs
amplitudes are complex so that phase angles (Plate XG2) of the mercury-derivative of

452 The Structure of Viruses

TMV show significant changes in the nonequa-
torial refections, but comparable photographs
of the lead-derivative have not yet been ob-
tained. It is thus not yet possible to determine
the phase angles unambi~uousl~r. In any case it
would probably be necessary to obtain more
than two isomorphous replacements.
Some preliminary work of Franklin on the
co~nparison of third layer-line of the mercury
derivative with that of TMV has led to the con-
struction of a map (37) giving a view of the
electron density in TMV projected down the
basic helix. Because of the uncertainties in the
phase determination, the results cannot be re-
garded as established, but it does seem that cer-
tain features (34) shown by the map are prob-
ably correct, since they are compatible with the
direct interpretation of other parts of the X-ray
diagram. It is found that the shape of the pro-
tein subunit is such that it protrudes at the
inner as well as the outer surface. The internal
and external protuberances do not occur oppo-
site one another, thus suggesting that the gen-
eral lie of the protein subunits is not strictly
perpendicular to the particle axis, but some-
what skew to it (see Plate XGI). Furthermore,
the shape of the protein subunits is such that
their helical packing results in a set of holes in
the virus particle at a radius of 55-60 A. Inner
regions of the virus may be accessibie to small
molecules through these holes and the external
grooves. It is interesting to note that the cys-
teine, as located by the position of the bound
methyl-mercur~, lies at about a radius of 56 A,
and thus close to this hole. The fact that the
free--SH group of the native ~6rus will react
with methyl mercury nitrate (26) and iodine
(2, 26), while there is no reaction with par-
achloromercuribenzoate (PCMB) and other
equally large sulfhydryl reagents (2) unless the
virus is denatured, may be explained if the
spaces between the subunits are acting as "mo-
lecular sieves" which accommodate the ,mailer,
but not the larger molecules.
Strains of TAW.-It was shown by Bernal
and Fankuchen (IO), and later by Franklin
(32), that different strains of ThfV give X-ray
diffraction patterns which bear a strong resem-
blance to one another, though small, measurable
differences can be detected. Following the deter-
mination of signs of the equatorial reffections
of the common strain (17, 331, it became possi-
ble to deduce the signs in the different strains

as well. The radial density distribution of CV4
was calculated in this way by both Kilkson (59)
and by Holmes and Franklin (55), who also in-
vestigated three other strains of TMV. The re-
sults of Holmes and Franklin are given in Fig.
XL-1 (b-e) and they show that the structural
arrangement of protein and nucleic acid is fun-
damentally similar in all the strains of TMV and
in CV4. The similarities and differences are dis-
cussed in full in the original paper, but perhaps
the most interesting conclusion concerns CY4,
which has been considered (65) as not being a
strain of TMV.
The earlier X-ray investigations (lo) showed
that the packing diameter of CV4 was smaller
than that of TMV by about 3 per cent. Fig.
XGl(c) shows, however, that the maximum
diameter is the same in each case, but that the
weight of protein in the outer part of the par-
ticle is considerably less in CV4 than in TMV.
The closer packing in dried CV4 thus probably
arises simply from a deeper interlocking of
knobs and grooves. The structure of CV4 ~ts in-
dicated by the X-ray diagrams is as closely hi-
lar to that of the common strain of TMV as any
of the other strains investigated.

The only other rod-shaped virus besides TMV
and related strains that has been investigated
by X-ray diffraction is potato virus X. The
photographs that have been obtained (10) are
not as good as those of TMV, presumably be-
cause the sinuous PVX particles are mort! diffi-
cult to orientate. The photographs do suggest,
however, that the structure is helical with a
pitch close to that of TMV, but that the sub-
unit packing repeats after two instead of three
turns.
The structure of p~l~rne~zed RNA-free pro-
tein.-Rod-shaped particles of TMV protein
freed from RNA may be prepared by breaking
down the virus in weak alkali into the low mo-
lecular weight A-protein component, which can
then be separated from the degraded RNA and
subsequently repolymerized by readjusting the
pH (83). From repolymerized A-protein pre-
pared by Schramm, Franklin (31) was able to
obtain a de ell-orientated gel which gives an X-
ray diagram similar to that of native TMV, and
which shows that the subunits in the repolymer-
ized protein have a helical arrangement very
similar, if not identica1, to that of TMV. How-
ever, the diagram is less perfect, indicating that
there is a small degree of disorder in the struc-

ROSALIND E. FRANKLIN, D. L. D. CASPAR AND A. &uo

453

ture. This is probably due to the possibility of
slight positional rearrangements of the protein
subunits when the RNA is absent.
The existence of different packing arrange-
ments when protein subunits are polymerized to
form TMV-like particles in the absence of RNA
has been demonstrated by a number of X-ray
studies. Franklin (38) examined the polymerized
rods formed from an abnormal protein present
in infected plants (Commoner's B8 component
[21]) and found the X-ray pattern to be sig-
nificantly different from that of TMV; on the
other hand, polymerized rods of abnormal pro-
tein (Takahashi's protein X [go]) prepared by
Newmark were reported by Rich, et al. (81) to
give a pattern very similar to that of TMV.
Furthermore, an A-protein preparation made
and examined by Caspar (19) gave a pattern
which corresponds to a mixture of the B8 and
TMV types. In view of this latter observation
it is probable that the variant pattern obtained
from B8 does not necessarily reflect a funda-
mental difference between this abnormal pro-
tein and the A-protein obtained from TMV it-
self. When gels of these various polymerized
protein rods are dried, the axial repeat distance
shortens somewhat (31, 38, 81). This shows that
the bonding between subunits is rather weak;
the intact virus shows no such change, presum-
ably because in this case the protein structure
is stabilized by the RNA (31).
The location and configuration of the RNA
in TMV.-The over-all similarity between the
X-ray pattern of the reaggregated A-protein
preparations studied by Franklin and that of
intact TMV indicates that the main differences
between the two patterns are due to the RNA.
From the equatorial scattering of the RNA-free
protein, she (33) was able to obtain the radial
density distribution in the protein particle. This
is shown in Fig. XLl(a) where it is compared
with the density distribution in the intact virus.
The only major difference between the two
curves is that the very prominent density maxi-
mum at a radius of 40 ,& in the virus is replaced
by a density minimum in the RNA-free protein,
and moreover it corresponds in amount to the
quantity of RNA present in the virus (about
5-6 per cent). This shows clearly that the RNA
must be located at a radius of 40 A. Ample con-
finnation that the structural difference between
the virus and the RNA-free protein lies mainly

at this radius comes from the nonequatorial re-
gions of the X-ray diagrams.
Franklin's results on the location of the RNA
required a revision of the obvious interpretation
of electron micrographs (53, 85) of partially de-
graded virus particles in which the exposed RNA
appears to form a central core. The X-ray results
refer to the intact virus in wet gel specimens
which have not undergone any special treatment,
and what was observed in the electron micro-
graphs is undoubtedly a collapsed form of the
RNA. Indeed, from what is now known of the
configuration of the RNA in the virus, one would
expect a collapse to take place when the protein
in removed, since it is the protein which main-
tains the RNA configuration in the intact virus.
The location of the RNA at a radius of 40 A
does not by itself indicate how the RNA is ar-
ranged in three dimensions in the virus, nor how
many chains of RNA there are (43). To de-
termine the configuration of the RNA, it is nec-
essary to compare the nonequatorial reflections
of TMV and of repolymerized A-protein, and
this has recently been done by Franklin (35).
From her work she was able to determine with
a high degree of certainty the general configura-
tion of the RNA in the virus.

Fig. XGl(a) shows that the RNA is em-
bedded tightly between protein subunits, so that
the chain direction of the nucleic acid molecule
must be directly related to the helical arrange-
ment of the protein subunits. Moreover, the per-
fection of the X-ray pattern of ThlV strongly
indicates that there is no, or very little, material
in the virus that does not follow a helical ar-
rangement. It is, of course, the phosphate-sugar
backbone of the RNA which will have the sym-
metry and not the sequence of bases (23).

Because the symmetry of the RNA conforms
to that of the protein there must be an integral
number of nuhleotides associated in a regular
way with each subunit. A critical survey (35) of
the phosphorus analyses of TMV (64, 79) gives
three phosphorus atoms per subunit as the best
integral value. Taking the molecular weight of
the subunit as 17,000, this corresponds to a
phosphorus content of 0.52 per cent, which is in
agreement with reported analyses.

From the number of nucleotides per protein
subunit, the known radial location of the RNA,
and the helical parameters of the protein sub-
units, it is possible to predict all possible ways
of forming a regular structure for the RNA, be-

454 The Structure of Viruses

cause of the restriction that the distance between
successive phospho~s atoms along an RNA
chain cannot be greater than 7.5 A. (The point
that this kind of prediction was possible was
made in 1955 by Caspar and Watson in an un-
published manuscript.) It turns out that with
three phosphorus atoms per subunit there are
only two regular ways of doing this. One arrange-
ment is that in which a single RNA molecule
follows the line of the main protein helix of pitch
23 A, and the other involves 16 chains of RNA
running at a small angle to the particle axis.
A comparison of the X-ray diagrams of TMV
and of the RNA-free repolymerized A-protein
shows that the over-all pattern of the strong
changes can be accounted for by the single chain
structure. Furthermore, the 16-chain structure
is definitely inconsistent with some of the
changes. The X-ray results are incorporated into
the model of the virus in Plate XG1, where the
RNA is represented 8s a smooth helix of diam-
eter 80 -4 and pitch 23 A. No attempt has been
made to show the detailed molecular configura-
tion of the RNA as this is still unknown. Two
points are, however, clear. There is only a single
strand of RNA, since there is not enough mate-
rial for a twin-strand structure lie DNA (92),
and moreover this strand is not fully extended
since the I>-P distance is only about 5 fL. The
kind of chain folding involved remains a problem
to be solved. Comparison of the birefringence of
TMV with that of the repolymerized protein
(31, 38) indicates that the base groups are more
nearly parallel than perpendicular to the particle
axis, and this is in accord with measurements of
the ultravio~et dichroism (78, S6). This indica-
tion is compatible with a single chain of RNA
following a flat helix, since the bases might be
expected to lie approximately perpendicular to
the chain direction. It should be noted that since
the pitch of the helix is 23 A, there is no pos-
sibility of hydrogen bonding between bases on
successive turns of the helix.
It should be stated that the X-ray diffraction
method could not detect discontinuities or breaks
in the covalent linkages of the RNA chain, if any
were present. However, with this reservation,
the X-ray results are in agreement with the sug-
gestions from various physico-chemical studies
(12, 45, 47, 56, 75) that the RNA in TMV is in
the form of a single large unit of molecular
weight equal to about 2 x 10`. It is also possible
to ~iiiderstand now why the virus has a well-

defined length in contrast to the variable lengths
(29,83) found for rods of polymerized A-protein
and of abnormal protein, since the length of the
RNA chain presumably determines the length of
the virus. Perhaps the best confirmation of the
conclusion from the X-ray analysis that the
RNA is in the form of a single chain comes from
the recent electron micrographs of Hart (54) on
degraded TMV examined by a method slightly
different from that used in his earlier work. A
strand of fibrous material is exposed in each
partially degraded virus particle, and the extra-
polated length is longer than any possible on the
multiple-chain models of the RNA.
These results on the configuration of RNA in
TMV provide the first detailed picture of the
structure of any ribonucleoprotein. In contrast
to deosyribonucleoproteins, where the in vivo
nucleic acid configuration (93) appears to be
the same double heIix as for extracted DNA
(92), the configuration of RNA in ribonucleo-
proteins appears to be imposed on it by its pack-
ing with the protein. Besides the detailed evi-
dence for TMV, X-ray studies of spherica1
viruses and microsome particles (44) indicate
that there is no intrinsic in vivo structure for
RNA comparable to the DNA double heIix. Ex-
tracted RNA appears to have a random coil con-
figuration and may be single stranded (12, 45),
yet it can take up a very regular structure in the
nucleoprotein complex.

~p~e~ica~ viruses

"Spherical" is not the most apt description of
the small viruses that are not long rods, since
high-resolution electron microscopy has shown
that many are polyhedral in shape (58, 94) and
some are significantly longer in one direction
than another (3, 67). The existence of sub-
structure in the spherical viruses, as shown by
the X-ray patterns obtained from virus crystals,
suggests that the virus surface will consist of a
regular arrangement of protuberances, which
could give the virus n polyhedral appearance.
The term "polyhedral" has, however, already
been used to describe insect virus inclusions
which are structures of an entirely different
scale. In the absence of a more acceptabfe term
for the small nonrodshaped virus particles we
will continue to refer to them as spherical with
the reservation that many may not look much
Iike spheres.

Spherically averaged structure.-Before dis-

ROSALIND E. FRANKLIN, D. L. D. CASPAR AM) A. KLUG

455

c&ng the diffraction patterns from Virus Single
crystals, which indicate their highly regular in-
ternal structure, we shall consider the results
obtained by Beeman and his colleagues (69, 82)
from low angle X-ray scattering studies on virus
solutions, which refer only to the spherically
averaged structure. Calculation of the spheri-
cauy averaged radial density (1) from measure-
ment of the low angle scattering is analogous to
the calculation of the cylindrically averaged ra-
  density of TMV using the equatorial scat-
tering from oriented gels, with the significant
difference that much less detail in the structure
can be resolved in the spherical viruses than in
TMV by this method.

The low angle scattering from a number of
small plant viruses (69, 82) is very similar to
that of a uniform density sphere, although there
some indication of a central hole, particularly
for brome grass mosaic virus (l), analogous to
the axial hole of TMV. Accurate calculations of
the mean particle diameters have been made
which are consistent with values obtained by
electron microscopy and hydrodynamic methods,
if account is taken of the effect of internal hy-
dration on the different types of measurements.
For those viruses which have associated with
them closely related RNA-free protein parti-
cles, for example turnip yellow mosaic virus
(TYMV) (72) and wild cucumber mosaic virus
(WCMV) (88) , a comparison of the low angle
X-ray scattering from the virus and protein
particles provides significant information about
the distribution of protein and RN.4 in the virus.
Markham (71) has shown by phy~co-chemical
methods that the serologically related virus and
protein particles of TYMV have the same diam-
eter and surface properties, and he suggested
that in both of them the protein is in the form
of a spherical shell. Low angle scattering studies
(63, 82) have confirmed that the protein par-
ticle, of mean external diameter about 280 A, is
essentially a spherical shell of mean thickness
35 if. More detailed results have been obtained
by Anderegg (1) for WCMV by comparison of
the density distributions for the top and bottom
components of this spherical virus (Fig. XL-2).
It is clear that most of the RNA is centrally lo-
cated and most of the protein goes to make up
the protective shell of these viruses.

Symmetry and substructure.-In order to
study substructure within the protein and RNA
Parts of the virus it is necessary to work with

FIG. x~-2.-The spherically averaged radial
density distributions of the top and bot-
tom components of wild cucumber mosaic
virus (1). The former contains no RNA,
thus it can be concluded from the differ-
ence between the two curves that the RNA
in the virus is centrally located and that
most of the protein is in an outer shell.
The virus itself (solid line) appears to
have a central hole, but this and other
details seen in the density distributions
must be regarded with caution, because of
the limited resolution, which is a good deal
lower (50 A) than that of the curves in
Fig. XL-1 (10 A). The overall features are,
however, significant.
(We are grateful to Dr. J. W. Anderegg
for providing this diagram.)

single crystals. X-ray patterns have been ob-
tained from six different crystalline viruses (18,
20, 25, 62, 70) which suggest that, like TMV,
they are buiJt up of regularly arranged subunits.
Only bushy stunt virus (BSV) and TYMV have
yet been studied in any detail, and even for these
two viruses our knowledge is less advanced than
for TRIV, mainly because up to now no iso-
morphous replacement has been achieved. We
shall therefore not discuss the spherical viruses
in the same detail as we did TMV.
The X-ray techniques and methods of analy-
sis differ a good deal in detail from those used
for TMV. In place of the difficulty of preparing
well-orientated gels there is now the problem of
growing and mounting crystals large enough for
diffraction studies. As in the case of TRIV, special
theoretical methods (60) have to be developed

456 The ~~~ct~re of Viwses

Tetruhedron
3 two-fold axes
4 three-fold axes

Octu~edron
3 four-fold axes
4 three-fold axes
6 two-fold axes

Icosahedron
6 five-fold axes
10 three-fold axes
15 two-fold axes

FIG. x~-3.-Regular polyhedra in which the re-
spective arrangements of rotation axes represent
the three types of cubic symmetry possible for a
spherical virus. In each case the three-fold axes
are in directions perpendicular to the faces. Note
that a virus particle having one of these types of
symmetry does not necessarily look like the regu-
lar polyhedron chosen to represent the symmetry
(cf. Plate XL-3).

to deal with the special problems of symmetry
not encountered in the X-ray analysis of crystals
of simpler organic substances and proteins.
We have already discussed how regular pack-
ing of subunits leads to a symmetrical structure.
In a helix, each unit is related to the next by a
rotation plus a translation, and the length of a
helix, such as TMV, is not determined by its
symmetry since the structure can repeat in-
definitely along a line. For a particle of finite
extent, built up of subunits regularly packed
about a central point rather than along a line,
only rotational symmetry is permissible if the
subunits consist of optically active molecules.
Only three general types of point group, or finite
collection of symmetry axes, are possible (22,
23) : first, those containing a single n-fold rota-
tion axis*; second, those possessing, in addition,
twofold axes perpendicular to the main axis;
and third, the cubic point groups. It is the latter
which are of particular interest in connection
with spherical virus structure since subunits ar-
ranged with cubic symmetry will all be equi-
distant from the center of the point group, and
will thus form an approximately spherical or

* An n-fold rotation axis can be represented by
a wheel with n equally spaced identical spokes; ?z
is necessarily an integer.

polyhedral shell. Three kinds of cubic symmetry
are possible for a virus particle and the arrange-
ments of rotation axes can be illustrated by the
tetrahedron, octahedron, and icosahedron (Fig.
XL-3); the number of as~metric units re-
quired by each of these three point groups is,
respectively, 12, 24, and 60. Plate XL-3(A) il-
lustrates one way in which 60 units (ping-pong
balls) can be put together to form an approxi-
mately spherical particle using icosahedral sym-
metry. The ping-pong balls are arranged in po-
sitions corresponding to the vertices of the snub
dodecahedron illustrated in Plate XL-3 (B) . By
changing the shape of the subunits,  a more
polyhedral-shaped particle could be produced.

In the twenty years following the first crys-
tallization of a spherical virus a number of X-
ray measurements (7, 8, 10, 11, 15, 25) of inter-
particle distances and pachng amangements of
viruses in crystals were made. The first photo-
graphs of a single crystal were obtained by
Crowfoot and Schmidt (25) from a derivative
of tobacco necrosis virus, and shortly afterwards
Carlisle and Dornberger (15) obtained photo-
graphs from single crystals of BSV. However, it
was not until the systematic studies of Caspar
(18) on crystals of BSV, and of mug, Finch, and
Franklin (62,63) on crystals of TYMV that evi-
dence regarding particle symmetry was obtained.
Cubic symmetry has been definitely established
for these two plant viruses, but the results, ai-
though already very informative, represent only
the begi~ng of the inv~stigation of internai
structure.
The choice of crystalline BSV for the first of
these studies proved very fortunate since all the
particles have the same orientation in the crys-
tal, and thus the X-ray diagram relates directly
to the diffraction from a single virus particle.
The X-ray patterns (Plate XM[A]) show that
the crystals have tetrahedral symmetry which
implies that the individual virus particles have
at least this symmetry and must therefore be
made up of 12 identical subunits. The virus may
have higher symmetry (but not lower) than that
of the crystal. Examination of the distribution of
strong reflections in the X-ray patterns (18)
indicates that the virus particles have fivefold
as well as the threefold and twofold rotation axes
required by the crystal symmetry; this suggests
that the virus has icosahedral symmetry which
would require that it be made up of 60 identicd
structural subunits. It is not possible to con-

ROSALIND E. FRANKLIN, D. L. D. CASPAR AND A. KLUG

457

clude, on the basis of the present evidence,
whether this symmetry applies to both the pro-
tein and the RNA backbone, or whether all the
60-protein subunits indicated are chemically
identical.
TYLlV crystallizes with a more complicated
cubic arrangement of the virus particles (8, 63)
than does BSV; the particles occur in two differ-
ent orientations, so that the unit cell (the repeat-
ing unit of the crystal) contains eight times as
many virus particles as the simpler unit cell of
BSV. The additional complexity of the crystal
has proved an advantage in analyzing the sym-
metry of the virus particles, although without
the evidence already obtained from BSV this
symmetry might not have been recognized. The
fivefold symmetry in BSV was recognized by the
concentration of strong reflections in certain di-
rections; in the X-ray patterns (Plate XL-4
[B]) from TYMV crystals there are twice as
many "spikes" of strong reflections, due to the
two different orientations of the particles in the
crystal. The analysis of these patterns (63)
shows that TYMV, like BSV, has icosahedral
symmetry. This, however, is not the complete
story. The larger unit cell of TYMV gives rise
to more reflections than does a BSV crystal, and
these additional reflections show that the sym-
metry of the particles is not completely icosa-
hedral. The most likely explanation of this result
is that though the protein may be made up of 60
identical subunits, there are a smaller number of
structural subunits for the RNA (perhaps 12).
As for TICIV, symmetry will only apply to the
RNA backbone, and, even if it is a continuous
chemical chain, it can be folded to give struc-
tural subunits.

The RNA-free particles associated with TYMV
crystallize in the same form as the intact virus
(7, 72). This provides an exceptional opportu-
nity to study both the substructure of the pro-
tein and RNA in a spherical virus, as has already
been done for TMV and repolymerised A-pro-
tein. This comparison is only in its earliest stages
(61), but it does seem to confirm the earlier sug-
gestion that the protein shell of the virus may
be made up of 60 identical subunits, while the
RNA may have a lower symmetry.
End-group analyses on TYMV (50), as well
as for BSV (66), indicate that there are at least
120 protein molecules in the virus and that
therefore the structural subunit (asymmetric
unit) may contain two chemical subunits. It is

interesting to note that there is a very close
structural similarit,y between BSV and TYMV,
in spite of the considerable difference in composi-
tion and mass of the particles. On the basis of
the present X-ray analyses the molecular weights
of the protein structural subunits are, respec-
tively, 125,000 and 50,000 for BSV and TYRIV.
The structural correspondence between these two
significantly different virus particles is probably
not fortuitous, but is most likely a reflection of
the fact that this type of cubic symmetry is a
very efficient way for nature to build a compact
particle out of smaller protein subunits.
On the basis of Crick and Watson's (23, 24)
suggestions discussed earlier, we might expect
that many small spherical viruses arc built up by
using cubic symmetry, and will thus cont'ain
some multiple of twelve identical protein sub-
units arranged to form a spherical shell. Evidence
for cubic symmetry in tobacco ring spot virus
(20) has recently been obtained; and there is at
least an indication of some kind of symmetry,
which may be cubic, in Coxsackie (20), a small
animal virus similar in many respects to the
small plant viruses. All small virus crystals need
not have cubic symmetry, as the recent X-ray
studies of Magdoff (70) on southern bean nio-
saic virus (SBMV) indicate. Although the crys-
tals are not cubic, there is evidence for subunits
in the individual virus particle. If more asym-
metrically shaped viruses, such as alfalfa mosaic
(3), are built up of subunits, these units could
not possibly be arranged according to cubic
symmetry. The virus particle could still have
one of the lower types of rotational symmetry
mentioned earlier.
It should be noted that cubic symmetry need
not be confined to the small RNA-containing vi-
ruses only. Particles of tipula iridescent virus
(TIV) , a DNA-containing insect virus of over
one hundred times the weight of small RNA-
containing viruses, have been shown by Williams
(96) to be regular icosahedra ; thus, morphologi-
cally at least, this virus has the same symmetry
as small plant viruses such as BSV and TYMV.

Conclusions

X-ray diffraction is at present the only
method available for studying the internal three-
dimensional configuration of viruses and other
highly organized, biological structures in the hy-
drated native state. Detailed X-ray diagrams
obtained from viruses tell us that the virus par-

458 The Structure of Viruses

ticles have a highly regular internal structure,
and moreover give a good deal of fairly direct in-
formation about their size, shape, and symmetry.
However, the only way we can yet be reasonably
certain of extracting the wealth of detailed struc-
tural information contained in the diffraction
patterns obtained from virus crystals or para-
crystals is by studying virus derivatives struc-
turally related to the intact virus, in particular,
heavy-atom substituted virus particles. For the
preparation of these modified viruses, we depend
on the assistance of virologists and biochemists.
Even with a number of suitable isomorphous
heavy-atom derivatives of a given virus, there is
an ultimate limitation to the detail that can be
"seen," which is determined by the perfection of
the virus structure. The X-ray diagrams tell us
that although one virus particle is very much
like all other particles of the same strain, there
are small differences in the positions of the
atoms, and perhaps even in the exact chemical
composition, between one particle and another.
However, in viruses such as TMV, BSV, and
TYMV, these differences are very small and, to
a resolution of about 3 A, one particle is identical
with a11 its sister particles. Hence, although we
may never expect to determine the positions of
all the atoms in a virus particle, we can hope
eventually to determine the configuration of the
polypeptide chain(s) in the virus protein sub-
unit, the molecular arrangement of the ribonu-
cleic acid, and its structural relation to the pro-
tein in the intact virus. Some of this information
has already been obtained by Franklin for TPIIV,
but for the spherical viruses the analysis of de-
tailed structure has only just begun.
The investigations of substructure in both
TAN and the crystalline spherical viruses have
already made possible some generalizations about
the way in which small viruses are put together.
In particular, it seems likely that the parts are
made by a type of subassembly process before
being assembled to build the virus. The forces
holding together the protein subunits in the vi-
rus particle are like those between globular pro-
tein molecules in a crystal. The configuration
of the RNA is determined by its regular packing
with the protein. These generalizations are prob-
ably applicable to all small RNA-containing vi-
ruses and may also apply to other particulate
nucleoproteins.

A tribute to Dr. Franklin

By W. M. Stanley

Dr. Rosalind Franklin of London, who was
scheduled to present the above paper at the
symposium, died on April 16, four months before
the meeting took place. In a way this paper can
be regarded as a memorial tribute to Dr. Frank-
lin from the co-authors.
Dr. Franklin was born July 25, 1920, in Lon-
don and received her Ph.D. degree from Cam-
bridge University in 1945. She has the distinction
during her rather short lifetime of having made
great contributions to two quite different areas
of research: first, in the study of coke and coal
and then later on, as you well know, in studies
on the structure of viral nucleoproteins. She
moved to Birkbeck College at the beginning of
1953 and started her work on tobacco mosaic vi-
rus, using the techniques she had developed
earlier, and made notable advances in the study
of this virus. She first verified and refined Wat-
son's helical hypothesis for the structure of this
virus and then made very important contribu-
tions having to do with the location of its nu-
cleic acid. By using tobacco mosaic virus con-
taining substituted mercury atoms as markers
she mas able to locate key points in the viral
structure. She developed a scale model of the
tobacco mosaic virus molecule and this was one
of the central features of the virus exhibit at the
Brussels World Exposition.
Dr. Franklin's life, I think, is an example of
complete devotion to scientific research. She was
a woman of great intelligence and wide culture
and her main interest was devoted to discovering
the ever more complex and significant patterns
underlying the processes of nature. In addition
to this, she was essentially an international cour-
ier of good mill and scientific information. She
visited several active virus laboratories, includ-
ing, of course, those at Cambridge and Roth-
amsted in England, and the laboratory at Tubin-
gen. She was quite well at home on the continent,
having worked earlier for some years in Paris.
She made several trips to the Virus Laboratory
in Berkeley. She discussed her work with many
investigators and took materials of interest to
her from all of these areas and succeeded in a
way in blending certain aspects of the scientific
research of these centers together into a coherent
picture.

ROSALIND E. FRANKLIN, D. L. D. CASPAR A~I A. KLUG

459

I think one of her most outstanding char-
acteristics was her courage. It is now known that
she was quite aware of the fatal nature of her
last illness. Those of us who were fairly close to
her knew little about it; she never spoke about
it; but she continued right on to the last to work
and plan as though her life were to continue.
She died within a few minutes of the time that
her last scientific paper was due to be read at
a conference of the Faraday Society. Her death
has certainly cut off a life of great promise, but
she had already done work which will insure
for her a notable place in the history of bio-
logical science.

Literature cited

1. Anderegg, J. W. 1957. Proceedings of the
American Biophysical Society Meeting, Co-

..

lumbus, Ohio.
2.  Anson, M. L., and W. M. Stanley. 1941. Some
effects of iodine and other reagents on the

structure and activity of tobacco mosaic vi-
rus. J. Gen. Physiol. 24: 679-90.

3. Bancroft, J. B., and P. Kaesberg. 1958. The
size and shape of alfalfa mosaic virus. Nature
  181: 720-21.

4. Bawden, F. C. 1958. Reversible changes in
strains of tobacco mosaic virus from legumi-
nous plants. J. Gen. Microbiol. 18: 751-66.
5.  Bawden, F. C., N. W. Pirie, J. D. Bernal, and
I. Fankuchen. 1936. Liquid crystalline sub-
stances from virus infected plants. Nature
  138: 1051-52.

6.  Bawden, F. C., and N. W. Pirie. 1938. A plant
virus preparation in a fully crystalline state.
  Nature 141 : 513-14.

7.  Bernal, J. D., and C. H. Carlisle. 1948. Unit
cell measurements of wet and dry crystalline
turnip yellow mosaic virus. Nature 162: 139-
  40.

8.  Bernal, J. D., and C. H. Carlisle. 1951. The
place of X-ray diffraction methods in the
estimation of the size and mutual arrange-
ments of colloidal particles. Discussions Fara-
day SOC. 11 : 227-29.

9.  Bernal, J. D., and I. Fankuchen. 1937. Struc-
ture types of protein "crystals" from virus-
infected plants. Nature 139: 923-24.

10.  Bernal, J. D., and I. Fankuchen. 1941. X-ray
  and crystallographic studies of plant virus
  preparations. J. Gen. Physiol. 25: 111-65.

11.  Bernal, J. D., I. Fankuchen, and D. P. Riley.
  1938. Structure of the crystals of tomato
  bushy stunt virus preparations. Nature 142 :
  1075.

12.  Boedtker, H. 1958. Personal communication.
13. Boedtker, H., and N. S. Simmons. 1958. The
  preparation and characterization of essentially
  uniform tobacco mosaic virus particles. J.

Am. Chem. SOC. 80: 2550-56.

14. Braunitzer, G. 1954. Uber der Nachweis der
  Carboxyl-Endgruppe im TMV durch Spaltung
  mit Hydrazin. Z. Naturforsch. 9b: 675.
  Carlisle, C. H., and K. Dornberger. 1948. Some
  X-ray measurements on single crystals of to-
  mato bushy-stunt virus. Acta Cryst. 1 : 194-96.

16.  Caspar, D. L. D. 1955. The radial structure of
  tobacco mosaic virus. Thesis, Yale University,
  New Haven, Conn.

17. - . 1956. Radial density distribution in
  the tobacco mosaic virus particle. Nature 177:
   928.

18.  - . 1956. Structure of bushy stunt virus.

  Nature 177: 475-76.
19. - . Unpublished.
20. Casper, D. L. D., and R. Langridge. Un-

published.

21.  Commoner, B., M. Yamada, S. D. Rodenberg,
  T-Y. Wang, and E. Basler. 1953. The proteins
  synthesized in tissue infected with tobacco
  mosaic virus. Science 118: 529-34.

22. Crick, F. H. C., and J. C. Kendrew. 1957.
  X-ray analysis and protein structure. Ad-
  vances in Protein Chem. 12: 133-214.

23. Crick, F. H. C., and J. D. Watson. 1956.
  Structure of small viruses. Nature 177 : 473-75.
24. Crick, F. H. C., and J. D. Watson. 1957.
  Virus structure: general principles. Ciba
  Foundation Symposium "The Nature of

Viruses" Churchill, London. p. 5-18.

25. Crowfoot, D., and G. M. J. Schmidt. 1945.
  X-ray crystallographic measurements on a
  single crystal of a tobacco necrosis virus de-
  rivative. Nature 155: 504605.
26. Fraenkel-Conrat, H. 1955. The reaction of
  tobacco mosaic virus with iodine. J. Biol.
  Chem. 217: 373-81.

27.  - . 1956. The role of the nucleic acid in
  the reconstitution of active tobacco mosaic
  virus. J. Am. Chem. SOC. 78: 882.

28.  Fraenkel-Conrat, H., and B. Singer. 1954. The
  peptide chains of tobacco mosaic virus. J. Am.
   Chem. SOC. 76: 180-83.

29. Fraenkel-Conrat, H., and R. C. Williams.
  1955. Reconstitution of active tobacco mosaic
  virus from its inactive protein and nucleic
  acid components. Proc. Nat. Acad. Sci. US.
  41 : 690-98.
30.  Franklin, R. E. 1955. Structure of tobacco mo-
   saic virus. Nature 175 : 379-81.

31. - . 1955. Structural resemblance between
  Schramm's repolymerised A-protein and to-
  bacco mosaic virus. Biochim. et Biophys. Acta
   18: 313-14.

cumber virus 4 and three strains of tobacco
mosaic virus. Biochim. et Biophys. Acta 19:
203-1 1.

33.  - . 1956. Location of the ribonucleic acid
  in the tobacco mosaic virus particle. Nature
  177: 92UO.

34.  - . 1959. X-ray diffraction studies of the
  structure of the proteins of tobacco mosaic
  virus. In A. Xeuberger, ed. IUPAC Sympo-

15.

32.

- ' . 1956. X-ray diffraction studies of cu-

460 The Stmcture of Viruses

35.

36.

37.
38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.
51.

52.

53.

sium on Protein Structxrc, Mcthuen, London.
(In press.)

   . 1959. The structure of the ribonucieic
acid in tobacco mosaic virus. (To be pub-
lished.)
-. The binding of osmium to tobacco
mosaic virus. (To be published.)

Franklin, R. E., and B. Commoner. 1955. S-
ray diffraction by an abnormal protein (B8)
associated with the tobacco mosaic virus.

. Unpublished.

Nature 175 : 107677.
Franklin, R. E., and K. C. Holmes. 1956. The
helical arrangement of the protein sub-units
in tobacco mosaic virus. BioFhim. et Biophys.
Actn 21 : 40W06.
Franklin, R. E., and K. C. Holmes. 1958.
Tohncco mosaic virus: application of the
method of isomorphous replacement to the
determination of the helical parameters and
radial density distribution. .4cta Cryst. 11 :
213-20.
Franklin, R. E., and A. Klug. 1955. The split-
ting of layer-lines in X-ray fibre diagrams of
helical structures. Application to tobacco mo-
saic virus. Acta Cryst. 8: 777-80.
Franklin, R. E., and A. Klug. 1956. The na-
ture of the helical groove on the tobacco
mosaic virus particle. Biochim. et Biophys.
Acta 19: 403-16.
Franklin, R. E., il. Klug, and K. C. Holmes.
1956. X-ray diffraction studies of the structure
and morphology of tobacco mosaic virus.
Ciba Foundation Symposium on "The Nature
of Viruses," Churchill, London. p. 39-55.
Franklin. R. E., A. Klug, J. T. Finch, and
K. C. Holmes. 1959. On the structure of some
ribo~~~cleoprotein particles. Discussions Fara-
day SOC. (In press.)
Giercr, A. 1957. Structure and biological func-
tion of ribonucleic acid from tobacco mosaic
virus. Xature 179 : 1297-99.
Gierer, A., and G. Schramm. 1956. Infectivity
of ribonucleic acid from tobacco mosaic virus.
Nature 177: 702-3.
Ginoza, W. 1958. Kinetics of heat inactiva-
tion of ribonucleic acid of tobacco mosaic vi-
rus. h'ature 181 : 958-61.
Green, D. W., V. M. Ingram, and M. F.
Pcrutz. 1954. Structure of haemoglobin. IV
Sign determination by isomorphous replace-
ment method. Proc. Roy. SOC. London (A)
225: 287-307.
Hall, C. E. 1958. Lengths of tobacco mosaic
virus from electron microscopy. J. Am. Chem.
SOC. SO: 255647.
Harris, J. I. 195s. Personal communication.
Harris, J. I., and C. -4. Knight. 1952. Action
of carboxypeptidase on tobacco mosaic virus.
Nature I70 : 613-14.
Harris, J. I., and C. A. Knight. 1955. Studies
on the action of carboxypeptidase on tobacco
mosaic virus. J. Biol. Chem. 214: 21530.
Hart, R. G. 1955. Electron-microscopic evi-

54.

55.

56.

57.

5s.

59.

60.

61.
62.

63.

64.

65.

66.

67.

68.

69.

70.
71.

72.

73.

dence for the localization of ribon~ic~eic acid
in the particles of tobacco mosaic virus. Proc.
Xat. Acad. Sci. U.S. 41: 261-64.

   . 1958. The nucleic acid fiber of the to-
bacco mosaic virus particle. Biochim. et Bio-
phys. Acta 28: 457-64.
Holmes, K. C., and R. E. Franklin. 1958. The
radial density distribution in Some strains of
tobacco mosaic virus. Virology 6: 328-36.
Hopkins, G. R., and R. L. Sinsheimer. 1955.
Visible and ultraviolet light scattering by to-
bacco mosaic virus nucleic acid. Biochim. et
Biophys. Acta 17: 476-84.
Huxley, H. E. 1957. Some observations on the
structure of tobacco mosaic virus. Proc. Stock-
holm Conference on Electron Microscopy.
Almguist and Wiksell, Stockholm. p. 260.
Kaesberg, P. 1956. Structure of small "spheri-
cal" viruses. Science 124: 626-28.
Kilkson, R. 1957. The cylindrically averaged
electron density distribution in cucumber vi-
rus 4. Arch. Biochem. Biophys. 67: 53-58.
Klug, A. The Fourier transforms of the cubic
point groups 23, 432 and 532. (To be pub-
lished.)
Klug, 8., and J. T. Finch. Unpublished.
Klug, A., J. T. Finch, and R. E. Franklin.
1957. Structure of turnip yellow mosaic virus.
Nature 179: fj83-84.
Klug, A., J. T. Finch, and R. E. Franklin.
1957. Structure of turnip yellow mosaic virus:
X-ray diffraction studies. Biochim. et Bio-
phys. Acta 25 i 242-52.
Knight, C. A. 1954. The chemical constitution
of viruses. Advances in Virus Research 2:

   . 1955. Are cucumber viruses 3 ahd 4
strains of tobacco mosaic virus? Virology 1:
261-67.
Xu, C.-I., V. Shore, and C. A. Knight. 1958.
The peptide chains of some plant viruses.
Virology 6 : 226-33.
Labaw, L. W., and R. W. G. Wyckoff. 1957.
The structure of southern bean mosaic virus
protein crystals. Arch. Biochem. Biophys. 67 :
225-36.
Lauffer, M. A. 1944. The size and shape of to-
bacco mosaic virus particles. J. Am. Chem.
SOC. 66: 118s-94.
Leonard, B. R., J. W. Anderegg, S. Shulman,
P. Icaesberg, and W. W. Beeman. 1953. An
X-ray investigation of the sizes and hydra-
tions of three spherical virus macromolecules
in solution. Biochim. et Biophys. Acta 12:
499507.
Magdoff, B. E. Personal communica~ion.
Markham, R. 1951. Physicochemical studies
of the turnip yellow mosaic virus. Discussions
Faraday SOC. 11 : 221-27.
Markham, R., and K. M. Smith. 1949. Studies
on the virus of turnip yellow mosaic. Para-
sitology 39 : 330-42.
Newmark, P., and D. Fraser. 1956. Composi-
tion of an abnormal protein present in to-

153-82.

ROSALIND E. FRANKLIN, D. L. D. CASPAR AND A. KLUG

461

74.

75 *

76.

77.

78.

79.

80.

231.

82.

83.

84.

bacco plants infected with tobacco mosaic
virus. J. Am. Chem. SOC. 78: 158S90.
Niu, GI., and H. Fraenkel-Conrat. 1955.
C-terminal amino-acid sequence of tobacco
mosaic virus protein. Biochim. et Biophys.
Acta 16: 597-98.
Northrop, T. G., and R. L. Sinsheimer. 1954.
Light scattering by tobacco mosaic virus nu-
cleic acid. J. Chem. Phys. 22: 703-707.
Oster, G. 1950. Two-phase formation in solu-
tions of tobacco mosaic virus and the prob-
lem of long-range forces. J. Gen. Physiol. 33:
445-73.
Oster, G., P. M. Doty, and B. H. Zimm. 1947.
Light scattering studies of tobacco mosaic
virus. J. Am. Chem. SOC. 69: 1195-97.
Perutz, M. F., M. Jope, and R. Barer. 1950.
Observations of proteins in polarized ultra-
violet light. Discussions Faraday SOC. 9:
423-27.
Pirie, N. W. 1956. Some components of to-
bacco mosaic virus preparations made in dif-
ferent ways. Biochem. J. 63: 316-25.
Ramachandran, L. K. 1958. The amino acid
composition of tobacco mosaic virus protein.
Virology 5 : 244-55.
Rich, A., J. D. Dunitz, and P. Newmark. 1955.
Structure of polymerized tobacco plant pro-
tein and tobacco mosaic virus. Nature 175:
1074-75.
Schmidt, P., P. Kaesberg, and W. W. Beeman.
1954. Small-angle X-ray scattering from turnip
yellow mosaic virus. Biochim. et Biophys.
Acta 14: 1-11.
Schramm, G. 1947. ober die spaltung des
Tabakmosaikvirus und die wiedervereinigung
der Spaltstiicke zu hohermolekularen pro-
teinen. I and 11. Z. Naturforsch. 2b: 112-21;
249-57.
Schramm, G., and S. Bergold. 1947. Uber das

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

1955. An infectious nucleoprotein from to-
bacco mosaic virus. Nature 175: 549-51.
Seeds, W. E., and M. H. F. Wilkins. 1950. Ul-
tra-violet micro-spectrographic studies of nu-
cleoproteins and crystals of biological inter-
est. Discussions Faraday SOC. 9: 419-23.
Siegel, A,, and S. G. Wildman. 1954. Some
natural relationships among strains of tobacco
mosaic virus. Phytopathology 44 : 277-82.
Sinclair, J. B., P. H. Geil, and P. Iiaesberg.
1957. Biophysical studies of wild cucumber
mosaic virus. Phytopathology 47 : 372-77.
Stanley, W. M. 1935. Isolation of a crystalline
protein possessing the properties of tobacco-
mosaic virus. Science 81 : 644-45.
Takahashi, W. N., and M. Ishii. 1952. An ab-
normal protein associated with tobacco mo-
saic virus infection. Nature 169: 419-20.
Watson, J. D. 1954. The structure of tobacco
mosaic virus. I. X-ray evidence of a helical
arrangement of sub-units around the longi-
tudinal axis. Biochim. et Biophys. Acta  13:
10-19.
Watson, J. D., and F. H. C. Crick. 1953. Mo-
lecular structure of nucleic acids. A structure
for deoxyribose nucleic acids. Nature 171 :

Wilkins, M. H. F. 1956. Physical studies of the
molecular structure of deoxyribose nucleic
acid and nucleoprotein. Cold Spring Harbor
Symposia Quant. Biol. 21 : 7H7.
Williams, R. C. 1954. Electron microscopy of
viruses. Advances in Virus Research 2: 183-
239.
   . 1957. Discussion in Ciba Foundation
Symposium. "The Nature of Viruses." Church-
ill, London. p. 52-53.
Williams, R. C., and I<. M. Smith. 1958. The
polyhedral form of tipula iridescent virus. Bio-
chim. et Biophys. Acta 28: 464-69.
Williams, R. C., and R. L. Steere. 1951. Elec-

737-38.

molekulargewicht des tabakmosaikvirus. 2. tron microscopic observations on the unit of
Naturforsch. 2b: 108-12.               length of the particles of tobacco mosaic vi-
Schramm, G., G. Schumacher, and W. Zillig.     rus. J. Sm. Chem. Soc. 73: 2057-61.

85.