DENSITY DIFFERENCES BET WEEN GENETIC MARKERS IN
PNEUMOCOCCAL TRANSFORMING PRINCIPLE

By Ronaxip Rotre* anp HARRIETT Epurussi-Tayuor

GATES AND CRELLIN LABORATORIES OF CHEMISTRY, f NORMAN W. CHURCH LABORATORY OF CHEMICAL
BIOLOGY, CALIFORNIA INSTITUTE OF TECHNOLOGY, PASADENA, AND LABORATOIRE DE GENETIQUE
PHYSIOLOGIQUE DU C.N.R.S., GIF-SUR-YVETTE, FRANCE

Communicated by Linus Pauling, July 3, 1961

It is known that DNA carries genetic information, and it is commonly assumed
that the identity and linear order of the genes are determined by the linear sequence
of nucleotides in DNA. However, little is known about the detailed relationship
(or “code”’) between the DNA and proteins of a given species. A linkage group

of genes may correspond to a unique nucleotide Sequence, or it is possible that a
large number of different nucleotide sequences could in principle carry the same
genetic information.

The DNA of a given bacterial species is restricted by some mechanism within
narrow limits of nucleotide composition, relative to the observed range of composi-
tions for the various species.!:? To explain the wide differences in nucleotide
composition between DNA’s from species that presumably have considerable

equivalent to C, and G is equivalent to T, so that the AT and GC pairs are inter-
changeable.*}4 Given such a code, mistakes in replication could lead to changes
in nucleotide composition with no change in genetic information, and DNA mole-
cules containing a given genetic marker would have a range of possible nucleotide
compositions equal to that for the total DNA. Differences in nucleotide composi-
tion between molecules containing different. genetic markers, and the nucleotide

around some equilibrium value.®
In the experiments to be described, the nucleotide compositions of DNA mole-

cules containing specific genetic markers have been investigated, taking advantage
of the fact that the buoyant density of a DNA molecule depends upon its nucleotide
composition, 2 and that the biological activity of transforming principle may be
determined after isolation from a CsCl density gradient.

Materials and M ethods.— Pneumococcal DNA containing the genetic markers streptomycin re-
sistance (.S"), optochin resistance (Q’), bryamycin resistance (B‘), and aminopterin resistance
{A’(136)] in various combinations was studied. A bacterial lysate containing about 20 ug of
DNA was centrifuged in CsCl solution of density 1.70 g cm~*, buffered at pH 8.5 with 0.01 M
tris-(hydroxymethyl)-aminoethanol, and containing 0.001 M sodium versenate. After 48 hr at
35,000 rpm in the Sw39 head of the Spinco preparative ultracentrifuge, the CsCl solution was
fractionated by making a small pinhole in the bottom of the lusteroid centrifuge tube and collect-
Vou. 47, 1961 GENETICS: ROLFE AND EPHRUSSI-TAYLOR 1451

ing the effluent drops individually. The relationship between CsCl solution density and drop
number was established by refractive index measurements® of individual drops (Fig. 1). The
drops were assayed for biological activity as follows: One milliliter of a culture of receptor strain
R36A, grown to maximal competence and chilled, was added to a cold test tube containing 0.01
ml of the CsCl drop to be assayed. The mixture was incubated 20 minutes at 37°C, chilled, and

 

POT TT ee
1.410 - 4

1.405>F- A ae

o
e
an
&
Pad
oa
me]
£
> 400F- 4
2
S
o
-
+
a
x
1.395 _

 

 

 

 

0 20 40 60 80 100 (20
Drop Number,N

Fig. 1.—The relationship between CsCl drop number and refrac-
tive index, for the various experiments. Refractive indices for so-
dium D light at 25°C. 42%°c for effluent CsCl drops are plotted
against drop number. The values are correct to within an additive
amount which was constant for a given experiment. Data from
experiments 5, 7, 8, and 10 are presented.

plated on blood agar after appropriate dilution. The plates were incubated 2 hr at 37°C, to allow
expression of phenotype, overlayered with an equal volume of antbiotic-containing agar, to select
the transformants, and colonies were counted 24 to 36 hr later. Antibiotics used were dihydro-
streptomycin (200 ug/ml), bryamycin’ (1.5 ug/ml), optochin (5 ug/ml), or aminopterin (0.75 X
10-5 M). Reaction of the cells with the DNA is restricted to 10 min, which is the duration of
competence of the receptor strain under the culture conditions used.

The number of transformants for a given marker in the assay system was proportional to the
DNA concentration, in the range employed in these experiments. Ratios of numbers of trans-
formants for the different markers found in replicate cultures were constant for a given medium
and DNA preparation, and were independent of DNA concentration. However, these ratios
were not completely independent of CsCl and other ion concentrations in the medium and it was
1452 GENETICS: ROLFE AND EPHRUSSI-TAYLOR Proc. N. A. 8,

found necessary to employ a standard amount of the CsCl solution in the assay culture. Differ-

ences in CsCl concentration of the drops from opposite ends of the density gradient were insuf-
ficient to modify the transformation ratios.

Density-gradient distributions of transforming activity for the different genetic markers have
been plotted according to the following expression:

FN) = T(N)/2 TN) (1)

where F(N) is the frequency of transformation for the Nth drop, and T(N) is the absolute number
of transformants for the given marker obtainable from the Nth drop. Mean drop number N
and variance o? were determined for each density-gradient distribution using the formulae

N = NF) (2)
o = DN — NyR) (3)

Variances expressed in drops? were converted to variances in gm? cm~* using the relationship

2 wy 2 2 ~6

(é vy} ° gm? cm (4)
where dp/dN is the density gradient. In units of gm?cem~, variances are characteristic for a given
DNA sample, independent of the density gradient.

Results DNA molecules containing different genetic markers occupy different
portions of the density-gradient distribution of the total DNA of pneumococcus.
This can be seen in Figure 2, which shows the density-gradient analysis of a sample
of genetically marked DNA, simultaneously labeled with P*?. Comparing the
distributions of P*? and of biological activity for streptomycin resistance, we observe
that the marker S’ is associated with DNA molecules which have an average buoyant
density greater than that of the bulk of the DNA. In addition, the density-gradient
distributions of the S’ and B’ marked molecules differ from each other, and are con-
siderably narrower than the distribution for the total DNA. Density-gradient
distributions for pairs of unlinked genetic markers were compared in several
experiments (Figs. 2-5), and significant differences in mean drop number NV were
detected for all pairs except (S’, A’). Density differences (Table 1) between pairs

TABLE 1

TABULATED PARAMETERS* FoR Densiry-GRADIENT DISTRIBUTIONS
OF GENETICALLY Markep PngumococcaL DNA

-—Ap(S", m) gm em73 X 104—

 

dp/dN Nm oh Fm?
gm emet drop — number gm?em=6 x Glog Ts/Tm (Ny — Nor) — din Ts/Tm,
Expt. x< 103 Marker +£0.01 106 dp X dp/dN dp
5 1.37 Sr 76.706 3.1 tee Lee a
Q 77.086 2.9 80 5.2 5.5
10 2.10 St 42.818 3.1 Les Lee Lee
Qr 42.994 3.3 57 3.8 4.2
Ar 42.809 3.1 57 4.0 4.2
8 1.54 Sv 70.866 3.7 a a a
B 71.512 4.7 123 10 10.6
Pp 71.906 5.5 Lee Lee an
7 1.63 S 58.153 7.2 tee a Lan
B 58 .905 6.3 141 12.2 22

*

* The following are tabulated: dp/dN gm :m~3 drop-!: the gradient of CsCl-solution density p with
drop number N; Nw: mean drop number for the distribution of biological activity for the genetic marker
m; &m? gm? cm’: variance of the distribution of biological activity for the marker m; Tsr/Tm (N):
ratio of the number of transformants for the marker S7 to the number of transformants for the marker
m obtainable from the Nth drop; Ap(St, m) gm cm—?: the difference in mean buoyant density between
DNA molecules containing the marker m and those containing the marker Sr.
VoL. 47, 1961 GENETICS: ROLFE AND EPHRUSSI-TAYLOR 1453

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T T T T | T l J T | T
0.4-- =
325 WITT
st P2LZZ2 777.
_~ r G 7)
S B' o000 LY A
w GQ 4
O03 Yeoooe TV, _
e 43 >
= Y POOOO 1
s 4 ° 7
co P ° 2
2 P7774 ° 4
rs L $ a
LY °
S 4 ° ° 4
= vy) ° °
o G ° ° 1
E 0.2 -- v ° b 4 4
Ss y ° b 4
w g ° D
S Y ° P
rae ly ° >
te VY ° 4
Y ° D
Ye0000 7
2 boOo00
Yo o77
0.1. rc “ ° 4
° °
YY oO °
iv ° °
Z ° oco0/
° oA
e000 °
ZZ So777
zrgseet oo00
| i | l I L I 1 ; t |

 

-66 67 68 69 70 v7 72 73 74 75 76
Drop Number,N

Fie. 2.—Equilibrium density gradient distributions for genetically marked, *P containing
pneumococcal DNA. D. pneumoniae carrying the genetic markers for streptomycin resistance
S’, and bryamycin resistant B’, was grown in a medium containing #P. Genetically marked
DNA labeled to the extent of one #P atom per 10 molecules was isolated from these organisms
and studied in the CsCl density gradient. Distributions of radioactivity and of biological activity
for each marker were determined. For the markers 8S’ and B’ transformation frequency F(N)
is plotted against effluent drop number N. For ®P activity, the fraction of total counts multi-
plied by the factor 1.55 is plotted against N.

of markers Ap(A,B) were determined from differences in mean drop number
AN(A,B) and the density gradient dp/dN using the relationship

Ap(A,B) = AN(A,B) (<2) (5)

For the markers S’, Q" and A’, the density gradient distributions of biological
activity are unimodal and not grossly asymmetrical.2 Due to differences in drop
size, the values of dp/dN and the width of the marker distributions measured in
drops differed in different experiments. However, as can be seen in Table 1,
the variances of the distributions for the different markers in a single experiment,
and for the same marker in different experiments, were closely similar when ex-
pressed in gm? cm~6.9

If the DNA molecules containing a given genetic marker are homogeneous in
density and molecular weight, their density-gradient distribution will be gaussian.
Gaussian character cannot be judged directly from the few observed points for
1454 GENETICS: ROLFE AND EPHRUSSI-TAYLOR Proc. N. A. 8,

 

 

0.3;-—

0.2

 

 

Transformation Frequency, F(N)

0.1

 

 

 

 

 

 

 

O-6-6-4-0-4-4,

 

0 I L I ! ! ; teogee
74 75 #76 77 +78 + +79 + #«280

Drop Number, N
Fig. 3.

each marker distribution. However, a study of ratios of numbers of transformants
for pairs of genetic markers (7'4/7's), as a function of drop number V , suggests that
the density-gradient distributions may be gaussian.

Assume that the density-gradient distributions for the genetic markers A,B are
gaussian with equal variances. If p is solution density, C;(p) is the concentration
at p of molecules carrying the marker 7, and p, is the buoyant density of these mole-
cules, we can write

(po — p,)?
C0) = Os(0) exp ( — C= 22 (6)
G;
af = 6; 74= A,B
Then the logarithm of the ratio C4/Cz should be a linear function of solution density p
Cc

Ino (2) = ap + B;
a= A043) (7)

a
Vou. 47, 1961 GENETICS: ROLFE AND EPHRUSSI-TAYLOR 1455

 

 

 

 

 

 

 

 

 

 

 

 

 

l l T ] T
eeccece
e e
,_$
r
¢---% s
q > Ql eeee
4 9
04- 4 4 A = 4
q >
? 4
9 9
4 >
ern} q
> >
= e b
aL e @
_ e §
9g 03 e 9 4
e 4
$s e ¢
a e >
» q@eece ,
uw 4 >
c q q
2 9 ¢
= q ¢
:
«= 0.2 9 4q ~“
2 q peecee
2 q ©
q e
2 q | 0
Ke q e
q >
q >
q >
> >
0.1 p- > > 4
> >
> 9
> >
> 4
, beece
q e
oss Lm oe
eevee
! ! | l |
4) 42 43 44 45
Drop Number, N
Fig. 4.
Px” 7 pa? C'4(p4)
B=" tina)
o C's(Pp)

Plots of log (4/73) versus drop number N for the various pairs of genetic markers
are fairly linear (Figs. 6-8). Since we may write

In (C,/C,) = 2.3 log (T/T) + constant (8)

the density difference Ap(A,B) may be calculated from the slope [d log(7'4/Ts) |/dN,
the density gradient dp/dN, and the variance ¢* according to the expression

d log (T/T »)
2.3 aN

Ap(A,B) = apldN a (9)

Density differences between markers calculated on the assumption that the distri-
1456 GENETICS: ROLFE AND EPHRUSSI-TAYLOR Proc. N. A.S.

 

 

 

 

 

 

 

 

 

 

 

 

 

T T T ] T q T q T T ]
sf
¢0000 B’ c000
0.3 f- q ° . 4
> z °
ra Gooos °
al q °
> e °
2 ©
o q b
7
® q ®
re q D
c 02F- q q 4
° q
s q >
q ¢
E q p 200000
o q q ° °
G q b ° °
< q ° °
° D ° °
a q PpOOad °
0.1 -—- D ° 4
°
° °
° °
° °
° 2
©0000008 9990 °
§ 0000d
| comvs | ! I ! | | | ! 20090 |
53 54 55 56 57 58 $9 60 61 62 63 64

Drop Number, N

Fie. 5.

Fies. 3-5.—Equilibrium density gradient distributions of transforming activity for the genetic
markers 8’, Q’, Br. For the different genetic markers in experiments 5, 10, and 7, transformation
frequency F(N) is plotted against drop number V. In experiment 10, a mixture of two kinds of
DNA was studied: (1) A purified DNA marked with A’ and (2) a DNA marked with S’ and Qr
isolated from a bacterial lysate.

butions are gaussian (Table 1) agree well with differences calculated according to
equation (5).

It has been shown that the DNA of pneumococcus contains classes of DNA
molecules associated with different genetic markers. These classes differ in mean:
buoyant density, but are nearly identical with respect to the variance of their
density gradient distributions 2 (marker). If we assume that the DNA of pnheumo-
coccus consists entirely of such classes of molecules, we may use the variance a?
(marker) to calculate lower bounds for the number average molecular weight’ WM,
and for the density heterogeneity!! «, of the DNA. The true values of these
parameters will exceed the calculated lower bounds only to the extent that mole-

cules containing a specific genetic marker are themselves heterogeneous in density.
Thus we may write

35
M,, (total DNA) = M, (marker molecules) > ——“—— (10)

a?(marker)

a, > Gy 2 3%, — 6? (marker) gm? cm-6 (11)
Vou. 47, 1961 GENETICS: ROLFE AND EPHRUSSI-TAYLOR 1457

 

1.6 TTT T_T TT

 

 

 

74 76 78 80
Drop Number ,N
Fic. 6 (Expr. 5).

where &?, is the contribution of density heterogeneity to the total variance of the
DNA @?,. Taking the observed value a? (marker) = 3 X 10-* gm? cm~— we obtain

M, > 9X 108

corrected to the sodium salt of DNA, but neglecting solvation effects.12 Using
values of a7, (P**) and 6? (S’) from experiment 8 (Table 1) and substituting in
formula 11, we obtain

2.35 X 10-3 > @, 2 1.34 X 10-3 gm em-?

Taking the density differences between DNA molecules containing different
genetic markers as entirely due to differences in nucleotide composition, we use
the relationship?

pose = 1.658 + 0.100 GC gm em-3 (12)

where po5ec is the buoyant density in CsCl at 25°C and GC is the mole fraction of
guanine-cytosine, to calculate the compositional differences between molecules
containing the markers S’, B’, A’, and Q’. Similarly, we estimate the nucleotide
compositional heterogeneity of the DNA from its heterogeneity in density.
Assuming a molecular weight of 9 X 10, molecules containing the marker S’
have 12,000 nucleotide pairs; they have 50 more GC pairs than molecules marked
with Q’, 120 more than those marked with B’, and differ in nucleotide composition
from molecules containing A’ by less than 10 GC pairs. For randomly constructed
1458 GENETICS: ROLFE AND EPHRUSSI-TAYLOR Proc. N. A. 8,

 

TTT 17
L s/o! ON 4
Lak \ 4
\
d
LOK 4
\
FIM fF 7
°o
o
S&S 08 >
A‘ /Q' a
06 4
si/a! \,
0.4F =
a
1 L l ! L

 

 

 

41 42 43 44 45
Drop Number, N

Fie. 7 (Expr. 10).

molecules of this size (similar results also hold if M , is greater than 9 X 108) let
us write!3

nm = 12,000 nucleotide pairs
probability of GC = 1 — (probability of AT) = 04 =p
standard deviation = [np (1 — p)]'’* = 53 GC pairs/molecule

The standard deviation of 53 GC pairs for the random model is much less than a
lower bound for the standard deviation calculated from the observed lower bound
for the density heterogeneity of the DNA

2.34 per cent GC > S.D. (GC) = 10 6,
2 1.34 per cent GC or 160 GC pairs per molecule.

With respect to their own mean, the nucleotide compositions found for the
markers studied have a standard deviation of only 64 guanine-cytosine pairs;
however it is interesting that the observed compositions for the markers all lie
on the guanine-cytosine-rich side of the total DNA distribution, with a mean com-
position differing from that of the total DNA by 150 GC pairs, or 1.3 per cent
Vou. 47, 1961 GENETICS: ROLFE AND EPHRUSSI-TAYLOR 1459

 

 

 

 

 

54 60 62 EXP 7
1 oT TT TT
0.4 4
0.2 |- \ 4
Oo = EXP 7
° @=:EXP8
\
e
0.0 - _
hy -
tu? oO
° ‘
oa
°
— -0.2 4
-0.4 ~ 4
-O0.6 4
a | I
68 76 EXP 8

 

Drop Number, N

Fig. 8 (Expts. 7 AND 8).

Figs. 6-8.—Logarithmic plots of ratios of numbers of transformants for pairs of genetic markers
versus drop number for experiments 5, 10, 7, and 8.

guanine-cytosine. Let us assume that (1) the DNA molecules of pneumococcus
that carry genetic information have the same mean composition as the bulk of
the DNA and (2) the four markers studied represent a random sample of the total
marker population. Using these assumptions, let us calculate P, the probability
of observing the compositions found for the various genetic markers. There are
two alternative ways of making the calculation, corresponding to two alternative
ways of expressing the deviations of observed compositions for the markers from
the mean composition for the DNA. The deviations (Table 2) may be expressed
in terms of (1) the standard deviation observed for the distribution of nucleotide
compositions among the DNA molecules (160 guanine-cytosine pairs per molecule),
or in terms of (2) the calculated standard deviation for randomly constructed
molecules (53 guanine-cytosine pairs per molecule). In each case P is the proba-
1460 GENETICS: ROLFE AND EPHRUSSI-TAYLOR Proc. N. A. 8.

TABLE 2

Nuc.eotipe Compositions oF MoLEcuLEs ContTaIninc Sprciric GENETIC MARKERS IN PNEvuMo-
coccaL DNA; Deviations From MEAN COMPOSITION OF Tota, Pyeumococcan DNA
——Deviation from Mean DNA Composition in Units of.

Standard Deviation for the Distribution of |
Nucleotide Compositions

Genetic Mole fraction, Guanine-cytosine, Observed in Calculated from
marker guanine-cytosine pairs per mole:ule the total DNA a random model
Sv +0.016 +190 1.2 3.6
Ar +0.016 +190 1.2 3.6
Br +0.006 + 70 0.45 1.3
Q +0.012 +140 0.9 2.6
Mean +0.013 +150 Lee

bility that all of the marker compositions exceed the population mean by at least
X standard deviations; thus in case 1, X = 0.45 and P is less than 0.013 and in
case 2, X = 1.3 and P is less than 0.0001.

Discussion.—These experiments demonstrate that the DNA molecules carrying a
specific genetic marker have the following properties.

(1) They are more homogeneous in buoyant density than the bulk of the DNA.

(2) They have a mean density characteristic for the marker;' it differs from
that for other unlinked markers and from that for the bulk of the DNA.

(3) The variances of their density gradient distribution are essentially inde-
pendent of the marker in question. The molecular weights obtained from
the variances are in the range of values reported for the best preparations of
pneumococcal DNA, as determined by sedimentation velocity studies,

The experimental observations are inconsistent with a model for DNA com-
position in which (1) the genetic markers of Ppneumococcus are evenly distributed
among its DNA molecules and (2) the nucleotide composition of a particular
molecule is a random variable with a fixed, species-determined probability of oc-
currence for the guanine-cytosine pair.

It appears likely that the nucleotide composition of a DNA molecule contain-
ing a specific genetic marker is uniquely determined by the compositions of the
genes of a definite linkage group comprising this molecule. This hypothesis will
require modification if the existence of density heterogeneity in molecules containing
the bryamycin marker is confirmed, and if in addition it can be shown that unlinked
genes for the B’ phenotype are not present in the DNA. Finally it would be
interesting to examine the nucleotide compositions of molecules containing other
genetic markers of pneumococcus, to investigate the possibility that genetic infor-
mation is confined to a special, guanine-cytosine rich fraction of the total DNA.

The authors are indebted to Professor Jean Weigle for helpful criticism of the manuscript.

* Postdoctoral Fellow of the U. 8. Publie Health Service.

} Contribution No. 2719.

*Sueoka, N., Marmur, J., and Doty, P., Nature, 183, 1429 (1959).

? Rolfe, R., and Meselson, M., these PRocEEDINGS, 45, 1039 (1959).

5 Sinsheimer, R. L., J. Molec. Biol., 1, 218 (1959).

4 Crick, F. H. C., in Structure and Function of Genetic Elements, Brookhaven Symposia in
Biology, No. 12 (1959).

5 Sueoka, N., J. Molec. Biol., 3, 31 (1961).

° Weigle, J., Meselson, M., and Paigen, K., J. Molec. Biol., 1, 379 (1959).

7 Kindly supplied by the Bristol Drug Company.

§ Distributions for the marker B’ differed importantly from those for the other markers, The
variance of the B’ distribution was significantly larger than that for S” in experiment 8, and in
Vou. 47, 1961 GENETICS: V. C. TWITTY 1461

experiment 7, the distribution for B’ was bimodal. These observations may result from gross dif-
ferences in drop size, but if they are not experimental artifacts, they suggest that molecules con-
taining the B’ marker are not homogeneous in buoyant density.

9 A partially purified pneumococcal DNA preparation was used instead of a bacterial lysate in
experiment 7. The large values of the variances for the marker distributions may be due to a
small quantity of low molecular weight DNA in the preparation. The agreement between the

r r
values of dog te /"s in experiments 7 and 8 indicates that the bulk of the DNA was of a size
p
‘ comparabie to that of the DNA studied in the other experiments. The slightly increased value
for «2(S) in experiment 8 may be due to molecular scissions associated with P*® decay.

0 Meselson, M., Stahl, F., and Vinograd, J., these PROCEEDINGS, 43, 581 (1957).

1 Sueoka, N., these ProceEpines, 45, 1480 (1959).

12 Hearst, J., Thesis, California Institute of Technology, 1961.

18 The nucleotide composition of the DNA from the strain of pneumococcus used is 0.40 mole

fraction guanine-cytosine, both by chromatographic analysis (R. Litman personal communication)
and buoyant density (ps°c = 1.698 gm cm~).
_ 314A given genetic marker may be introduced into different strains of pneumococcus by trans-
formation. Experiments in progress in these laboratories suggest that (1) the characteristic den-
sity of a marker is the same in different strains of pneumococcus, and (2) independent occurrences
of a given marker in different strains of pneumococcus have the same characteristic density.