GENETIC RECOMBINATION IN DNA-INDUCED
TRANSFORMATION OF PNEUMOCOCCUS.
I. THE PROBLEM OF RELATIVE EFFICIENCY OF
TRANSFORMING FACTORS?

HARRIETT EPHRUSSI-TAYLOR, A. MICHEL SICARD? anp ROBERT KAMEN?*
Developmental Biology Center, Western Reserve University, Cleveland, Ohio

Received November 6, 1964

FO OWING absorption of DNA by competent cells, genetic transformation
ensues with a probability that is characteristic for each particular genetic
marker (Horcuxiss and Marmur 1954; Lerman and Totmacn 1957). Thus,
specific incorporation, defined as the amount of DNA which must be incorpo-
rated in order to obtain one transformant, varies according to the genetic marker
employed.

With the development of more elaborate genetic marker systems, notably,
systems of linked genes, various causes were postulated as operating in determin-
ing the probability that, given DNA incorporation, a certain type of transformant
would appear. Thus, arguing from the fact that pairs of linked genes appear
together in transformants less often than the same genes appear singly (Hotcx-
Kiss and Evans 1958), it has been suggested that mutant markers which have
different probabilities of integration into a transformed cell have different linear
dimensions (Lacks and Hotcuxiss 1960; Epurussi-Taytor 1961).

In ah interesting series of experiments, SCHAEFFER (1958) showed that where
transformations are performed between different species of Hemophilus, related-
ness, which is presumably an expression of homology of base-pair sequence in
DNA, is crucial in determining the probability of integration of a genetic marker
into a transformant. Further, in intraspecific transformation in pneumococcus,
Green (1959) found that certain combinations of donor and recipient led to low
transformation frequencies for a particular streptomycin resistance gene. The
depressed probability of integration was further shown to be due to the presence
of a “depressor” region closely associated with the marker gene, but separable
from it by recombination (GREEN 1959; RorHHEim 1962). Both of these lines of
work indicate the importance of structural homology in determining the proba-
bility of integration.

Still another cause of low probability of genetic integration has been proposed,
though never demonstrated experimentally. Were a particular genetic marker
to be located always near the end of a DNA molecule, one might expect it to be

1 Supported by Public Health Service grant GM-09917-3.
2 Present address: Laboratoire de Génétique Physiologique du C.N.R.S., Gif-sur-Yvette, S. et O., France.
3 Fellow in the National Science Foundation undergraduate summer training program.

Genetics 51: 455-475 March 1965.
456 H. EPHRUSSI-TAYLOR et al.

recombined into a transformant with a low probability (Eprrusst-TayLor
1961). This suggestion is based on a particular model of the recombination
process, and it should be noted that other models could lead to the opposite
prediction. For example, if recombination requires strand separation of the donor
DNA, markers near ends might participate more actively in the recombination
process owing to easier strand separation.

Finally, with the discovery that physical uptake of DNA by competent cells is
molecular-weight dependent (RosENBERG, SIROTNAK and Cava.ienrt, 1959; Litr
1958), in that larger molecules are more readily absorbed, one could suppose that
different markers, upon extraction, are associated with DNA fragments of differ-
ent size. Hence, different markers could exhibit different probabilities of integra-
tion, owing to selective absorption of the larger fragments (Marmur, ANDERSON,
Maruews, Berns, Gasewska, Lane and Dory 1961).

Distinguishing between these various possible factors requires primarily that
the experimenter be in possession of a system of genetic analysis with which
fine-structure studies can be performed. Given a single densely marked region of
a DNA molecule, the majority of these possible causes of high or low efficiency
can be tested. Having developed such a system (Sicarp 1964), we have under-
taken to examine what determines the transforming efficiency of the mutant
markers in the densely marked region.

The experiments to be described here concern 73 mutants, 39 of which have
been mapped in 30 sites, within what appears to be a single functional unit of
the pneumococcal genome (Sicarp 1964). These sites are all genetically linked
(Sicarp and Epwrussi-Tay tor, in preparation). All mutants in this region are
resistant to aminopterin, and sensitive to an imbalance in the molar concentra-
tions of isoleucine, leucine and valine. Thus, transformations can be selectively
scored in crosses of wild-type cells by mutant DNA, or mutant cells by wild-type
DNA. Owing to the special advantages of this genetic system, it will be shown
below that the probability of genetic integration of these particular markers is
determined by highly local factors; i.e., either by the nature of the mutation, or
by the base composition of the region in which a mutation has occurred.

MATERIALS AND METHODS

Media, transformation techniques and strains employed in our laboratory have been recently
described (Sicarp 1964). Pertinent to what follows is the fact that all mutants have been isolated
within a single wild-type line, to render genetic homology as great as possible. Competence is
obtained by transfer of a small aliquot of frozen stock culture (preculture, or PC) into the ap-
propriate medium, where it develops in an acute fashion at a predictable time after transfer.
Since competence appears and disappears in an interval of 10 to 15 minutes. it is not necessary
to arrest the reaction of cells with DNA by the addition of DNase. Variability in transformation
could be expected to occur because of slight physiological differences in batches of PC, or in
batches of competent cells made from the same PC. The incidence of such variability on estima-
tions of transforming efficiency will be examined below. By “complexes” we mean cells which
have fixed DNA irreversibly, but which are not yet transformed.

Mutants induced with ethyl methane sulfate (EMS) were obtained by growing a strepto-
mycin-resistant transformant of the wild-type line in 0.5% EMS for about six generations, and
plating in 1 X 10-5 aminopterin.
RELATIVE TRANSFORMATION EFFICIENCY 457

All mutants having the dual property of resistance to 1 x 10-5 m aminopterin, and sensitivity
to an imbalance of branched amino acids, map in a single region of a DNA molecule and are
presumed to be in a single cistron referred to as the amiA cistron. Mutants bearing the numbers
r{ and r2, r21 through r40, and above 133 are of spontaneous origin. Mutants numbered from
13 through r20 are HNO,-induced (technique of Lirman and Epxrussi-Ta¥Lor 1959), while
r41 through r53 are EMS-induced.

Ultraviolet irradiations were carried out on dilute solutions of DNA (about 10 #g/ml), using
a lamp emitting light at 257 my, at a dose rate of 17.1 ergs per sec per mm? (Black Light Eastern
Corp., model R-57). Solutions were stirred during irradiation.

Statistical analyses were carried out according to KimBaLL (1961), who has treated specifically
the problem of evaluating the errors in measuring ratios. Dr. Norman RusurorrH has been of
invaluable assistance in adapting KimBa.1’s method to the present data.

EXPERIMENTAL RESULTS

The efficiency of the amiA mutants. This is determined as a ratio; ie., the
number of transformants of a particular type, relative to the number of trans-
formants for a standard reference gene, str-r41. The reference gene serves to
measure, essentially, the amount of DNA fixed by a particular lot of cells. This
practice is widely used in place of measuring P*? incorporation since it is more
convenient, and avoids the problem of radiation damage to the DNA (GREEN
1959; Lacks and Horcuxiss 1960; Epprati-Evizur, SRINivasaAN and ZAMEN-
HoF 1961; ANAGNosTOPOULOSs and CrAwForD 1961). The distribution of the effi-
ciencies of the 73 amiA mutants is shown in Figure 1. In our experience, these
values are reproducible, and variability lies within the error of the plating tech-

 

L. H E MUTANTS

  
  
  
 
 
  

Y
“ee
EA A A
OA,
COAL
06 086 o- 120«14 Lé

SPONTANEOUS

(1 uno,
EMS

   

   
    

LE mutants

 
 

| E
|

0.06 0.08 GIO Al2 Ol O16 O18 0.20 0.22 0.24

 

 

Ficure 1.—Frequency distributions of transforming efficiencies of mutants selected in the
amiA locus.
458 H, EPHRUSSI-TAYLOR et al.

nique. This is in contrast with the experience of Iver and Ravin (1962) working
with several erythromycin resistance factors, who found considerable variation
depending on the batch of recipient cells employed. The efficiency measurements
performed on eight amylomaltase marker genes by Lacks and Horcuxiss
(1960) also show in several instances greater standard deviations than we have
found. On the other hand, relative constancy of efficiency determinations has
been noted by Srrornax, Lunt and Hurcutnson (1964) fora series of amethop-
terin resistant mutants in pneumococcus. However, the methods of statistical
evaluation employed in the above-mentioned studies have not been indicated, so
that it is impossible to compare published data in a significant way.

The first 30 mutants, and many of the remaining 43, have been used not only
as DNA donors in one-point crosses of wild-type recipient by amiA-r str-r41
DNA, but also as recipients in one-point crosses of amiA-r cells by amiA-s
str-r41 DNA. With the possible exception of mutant 730, efficiency for any given
mutant is the same in both directions of the one-point crosses. Mutant r30
usually gives a value of about 0.6 in the cross of mutant cells by wild-type DNA,
and about 0.8 in the reverse cross. In paired experiments done the same day,
this difference in the two directions of one-point crossing of site 30 has been found
to be significant at the 5% level. Since each cross to be described in the report to
follow the present one (on mapping of the genes) includes a control cross of
mutant recipient by wild-type (for amiA) donor DNA, more information is
available on the reproducibility of the efficiency measurement in the one-point
transformation cross of mutant to wild type. Table 1 presents data from inde-
pendent experiments of this type, demonstrating the constancy of efficiency.

The striking feature of the data of Figure 1 and Table 1 is that mutants within
the amiA region fall into two nonoverlapping classes with respect to efficiency.
These classes will be called HE and LE classes, for high and low efficiency.

Relationship between efficiency of a mutant and its origin: The distribution of
the 73 amiA-r mutants into two efficiency classes relative to origin of the mutant
is shown in Table 2. Spontaneous mutation gives rise to the two classes with
roughly equal frequency, whereas EMS has given rise only to LE mutants. Of
the 18 mutants isolated following transformation of wild-type cells with HNO.-
treated wild-type DNA (see Sicarp 1964), only two are HE (r3 and ri).
Since mutagenesis by transformation of cells with HNO,-treated DNA increases
the incidence of amiA mutants by a factor of 10 relative to background muta-
tion, on the average one out of every ten presumably induced mutants could in
fact be of spontaneous origin. Both of the HE mutants isolated in the HNO,-
induced group fail to show recombination with spontaneous mutants rf and 72
which themselves do not recombine, thus marking a “hot-spot” of spontaneous
mutation. This suggests that mutants r3 and r19, of the HNO, group, are in
reality of spontaneous origin. A further reason for believing that HNO, induces
only LE mutants is the fact that the very process by which these mutants are
obtained would favor the detection of HE mutants. The HNO.-altered DNA is
used as a transforming factor, and HE induced mutations would have a ten times
greater chance of appearing in transformants than would LE mutations. There
: al
KELATIVE TRANSFORMATION EPVICIENCY 459

TABLE 1

Statistical variability of efficiency measurements

 

Complex

 

Cross PC No. No. Ratio Limits
amiA-r1 X wild type 1 1 1.05 0.93 —1.19
2 1 1.21 1.05 — 1.38
amiA-r? * wild type 1 1 0.082 0.072 — 0.101
1 2 0.086 0.075 — 0.095
amiA-r15 X wild type 1 1 0.0936 0.082 — 0.107
amiA-r16 X wild type { 1 0.116 0.105 — 0.128
1 0.122 0.112 — 0.134
2 1 0.111 0.102 — 0.120
ami-17 X wild type 1 1 0.089 0.072 — 0.098
1 2 0.093 0.083 — 0.101
2 1 0.104 0.098 — 0.111
2 2, 0.133 0.117 — 0.152
ami-22 * wild type 2 2 1.13 1.05 ~— 1.22
2 3 1.01 0.92 —1.10
2 4 0.915 0.84 — 1.00
ami-29 * wild type 1 1 1.03 0.96 —1.11
1 (bis) 0.88 0.80 —0.96
2 1.03 0.94 — 1.13
ami-30 X wild type 1 1 0.634 0.589 — 0.683
1 1 (bis) 0.696 0.646 — 0.749
1 2 0.727 0.659 — 0.805
1 3 0.664 0.597 — 0.738
2 1 0.667 0.653 — 0.702

 

Values are included which were obtained (a) in different platings of same batch of DNA-bacterial complexes; (b) in
different preparations of complexes from a single preculture; (c) in completely independent experiments from different
lots of frozen preculture. PC refers to batch of preculture. Variability is expressed as upper and lower limits of the
95% confidence interval.

TABLE 2

Incidence of HE and LE mutants according to their origins

 

Type of mutant
HE LE

 

Spontaneous, lot 1 7 5
Spontaneous, lot 2 5 4
Spontaneous, lot 3 8 9
Spontaneous, lot 4 g 2
Total 29 20
percent 53 47
HNO,-induced 2* 16
percent 0(11) 100(89)
EMS-induced 0 13
percent 0 100

 

* Probably of spontaneous origin.
460 H. EPHRUSSI-TAYLOR et all.

is, therefore, strong reason to believe that treatment of DNA with HNO, yields
only LE mutants.

Mapping the amiA mutants: The transforming efficiency of a mutant plays a
very curious and important role in two-factor mapping experiments (Lacks and
Horcuxiss 1960), in which wild-type recombinants are scored in crosses of
mutant by mutant. If a particular cross involves a pair of mutants which have
different efficiencies, as defined above, the frequency of wild-type recombinants
observed will very much depend upon whether the LE site is in the donor DNA,
or the recipient cell. When the LE site is in the recipient cell, recombinants will
be very much rarer. To counter this difficulty, Lacks and Horcnxiss proposed
that recombination values should be corrected by dividing the observed recombi-
nation value by the efficiency of the site present in the recipient cell. In spite of
the fact that such presumed corrections can leave differences as great as threefold
or more in recombination frequencies of crosses in reverse polarity, several at-
tempts at mapping by this method have been published (Epurati-Exizur,
Srinivasin and ZAMENHOF 1961; ANAGNOSTOPOULOS and Crawrorp 1961;
Srrotnak, Lunt and HutcHINson 1964). The persistence of these differences
demonstrates that there is no simple correlation between efficiency, in one-pomt
transformations, and recombination frequency, in two-point transformations.
This question will be treated in extenso in the next paper of this series, now in
preparation. A further difficulty with the mapping method proposed by Lacks
and Horcrxiss is that suitable statistical methods to compare numbers which,
each, are a ratio of two ratios have apparently not been developed.

The mutants which have been mapped at the amiA locus have been located
with respect to each other on the basis of two sorts of evidence: (1) Some mutants
do not recombine and are presumed to lie at the same site; (2) In a series of
crosses in which a given mutant strain serves as recipient, the approximate
alignment of all other genes is made on the basis of the relative frequencies of
wild-type recombinants observed. The independently obtamed alignments are
then compared. It is found that a high degree of concordance is observed for sites
that are relatively close together. Therefore, by varying the site in the recipient,
it is possible to align sites at any point of the map. This method avoids assigning
a specific meaning to the actual recombination values observed. Mutants can be
ordered with a fair degree of confidence in this way, but distances between
them remain unknown, except in a gross way. Mapping data will be presented
in the next paper of this series, but the map itself must be presented here in order
to discuss the efficiency problem (see Figure 2). Most of the mutants numbered
from 31 through 73 have not been completely mapped as yet. They fall at a
large number of new sites, and appear to extend map length by a factor of the
order of 2.

Site r30 has been indicated to be a multisite mutant, covering sites 777 and 15.
The reasons for doing this, instead of placing 730 between 5 and 17, are the
following. Crosses of strain r17 by DNA of r5 yielded 9 amiA-s recombinants for
57,450 str-r transformants scored. Strain 130 crossed by DNA r/7 has shown no
amiA-s recombinants for 258,150 str-r transformants scored; and crossed by
RELATIVE TRANSFORMATION EFFICIENCY 461

 

Ig
10 3 1s
3a 21 13 18 30 --
39 40 49 41 42 22 29 9 20 24 45 ¢ mze7 i) aves It 23 2 28 438 as 53 57
i L i L n L n L 1 i uf 1 1 1 4 1 nl 1 L Mv 1 L ‘ul

 

Figure 2.—Map of the genes whose positions have thus far been determined. Mutants which
recombine with very low frequency are placed close together. Others are equally spaced, indicat-
ing that we cannot evaluate their relative distances with any degree of certainty. The order of
the closely linked pairs of genes (5 and 17; 26 and 31; 4 and 16) may be the inverse of that

shown.

DNA 75, no amiA-s recombinants for 207,000 str-r transformants scored. It is
thus unlikely that 730 lies between r5 and r/7. Further, mapping data show that
30 is much closer to r53 than is either r/7 or r5, indicating that r30 extends
appreciably to the left of the position occupied by r7/7. Figure 3 shows the
recombination relationships of the mutants at the extreme right end of the map,
in crosses with r30 as recipient, with r/7 as recipient, and with 753 as recipient.

Relationship between efficiency and position on the map: Inspection of the
map shows two things. First, there is no over-all correlation between position on
the map and efficiency. Second, with the exception of mutant 730, independent
mutations at the same site are either all HE or all LE. The first observation

amiA—r30 AS RECIPIENT:

ss

 

'
1
t r
4 6 35 5315 17

1
RH 0.0005
-———i 0,009
Ht SO-.006
-————————SC0.0005

amiA -r!i7 aS RECIPIENT:

 

mH a.ooo6e

e——A 900I7
+ 0.0048
-———— + 0.0056
Si. 4

 

amiA-r53 AS RECIPIENT: 30 __
i
\
'
q T T Tt T T
4 46 35 53,5 7
'
0.0077 rite! 0.00003
0.012 et 0.0023
0.013 eH 0.0032

Figure 3.—Mapping the 730 region, using strains 130, ri7 and r53 as recipients. Recipients
r17 and r53 would be expected to yield low frequencies of recombinants since r{7 and r53 are
LE genes. In spite of this, strain r30 x DNA r53 gives fewer wild-type recombinants than
strain r17 X DNA 753, indicating that mutation r30 extends leftwards. ending close to site 753.
This is consistent with the cross of strain 153 by DNAs 730, r5 and r17, where 730 also maps
very much closer to site r53 than do 75 and r17. Gene order in crosses of recipient r30 is not the
same as the order observed in crosses of recipients ri7 and r53. The latter order is confirmed in
crosses (not shown) of recipient r28, by all sites to its right. Site 728 lies to the left of r4.
462 HH, EPHRUSSI-TAYLOR et al.

suggests that low efficiency has nothing to do with proximity of a site to the end
of a molecule; were this the case, there should be a gradient with respect to
efficiency. The second observation suggests that efficiency is site specific.

Mutant 730 is a most interesting exception in that it is a mutant whose effi-
ciency lies at the lower limit of the HE class, yet it covers two LE sites. We are
searching for more mutants in this region of the map in an attempt to determine
the length of mutation 730, and whether it covers HE sites as well as LE sites.

As the map stands at present, there is a strong indication that efficiency is site-
specific and independent of the direction of the cross, and that it is not correlated
with any major discontinuity; ie., “end,” either of the DNA molecule, or of the
pairing process.

Differences in sizes of mutant markers as a cause of different efficiencies: A
general correlation between efficiency, on the one hand, and linear dimensions of
a mutation, on the other, was suggested by the study of Lacks and HotcHxiss
(1960). Of a group of eight mutants affecting a single enzyme, two mapped as
small deletions (or multisite mutants), and one as a large deletion. Strains
bearing these mutants could be transformed back to wild type, and the efficiency
of such transformation appeared to be inversely correlated with the extent of the
deletion. Furthermore, heating of wild-type DNA at temperatures causing de-
purinization but not strand separation resulted in an inactivation of the wild-type
sites corresponding to the presumed size of the various mutations. In general, the
larger the region required for transformation, as indicated by genetic data, the
more rapid the inactivation proved to be.

We have considered whether our mutants differ in size, and whether such
differences, if any, are correlated with efficiency. At present, appreciable size-
differences seem excluded for the following reasons: (1) The amiA cistron is
fairly intensely marked by the mutants thus far mapped, and only one has
proven definitely to be multisite (r30). It is unlikely that many of the others
represent gross mutational alterations. (2) In the course of the present study,
spontaneous reversion has thus far been observed for the following mutants:
HE—r1, r22 and r29; LE—r6, r8, r9 and r10. All revertants which have shown
up have been examined for the presence of suppressor mutations by transforming
wild-type cells with DNA of the revertants, and screening for the appearance of
ami-r transformants. All revertants appear to be true reversions. We cannot,
however, exclude the possibility of there being suppressors extremely closely
linked to the original mutant site. As the reversion data stand they suggest that
neither efficiency class has the reverse mutation pattern of a deletion or multisite
mutation.

While there is no genetic evidence of gross size differences between HE as
opposed to LE genes, one can nonetheless ask whether such differences do exist.
Following Lacxs and Horcuxiss (1960), and Rocer and Horcuxiss (1961),
heat inactivation at subdenaturation temperatures was performed as a test for
size differences.

The consequences of heating DNA at temperatures below those at which DNA
melts have been shown to depend on pH (Rocer and Horcuxiss 1961) and upon
RELATIVE TRANSFORMATION EFFICIENCY 463

the way the DNA is prepared (Gurnoza and GuILD 1961). Simple inactivation
pehavior is obtained on DNA deproteinized by chloroform and octy] alcohol only
if the pH lies below neutrality. Accordingly, the experiments reported here were
done at pH 5.4 (0.02 m phosphate buffer). To protect against strand separation,
the molarity of NaCl was raised to 1 m. Wild-type DNA bearing the str-r47 gene
was heated at 83° C, and samples removed at various times. Remaining biological
activity was titered on a variety of amiA mutant recipient strains. In other ex-
periments, DNA bearing gene str-r41 and an amiA-r mutant gene was heated,
and surviving activities titered on the wild-type strain.

Various HE and LE sites were examined to see whether they show any
differences in subcritical heat inactivation rate. No significant differences were —
observed for the reference gene str-r41 and HE sites s22 and s/, on the one hand,
and LE sites s6, s4, and sf7, on the other. HE site s29 is, possibly, slightly more
resistant to subcritical heating than the reference gene str-r47. Subcritical heat
damage does not, however, affect differentially HE and LE mutants.

The mapped sites 30, 5 and 17 provide an occasion to test the validity of sub-
critical heat inactivation as a measure of marker size. Inactivation of s30 or
730, as well as of sf7 and r5, proceed at the same rate. Since the multisite mutant
730 and its s allele have the same sensitivity to subcritical heating as the sites
with which it shows no recombination, it is doubtful that subcritical heat inacti-
vation is really a sensitive measure of the dimensions of mutant sites. The inacti-
vation curves for these sites are shown in Figure #.

Further doubt is cast on the general usefulness of subcritical heating to mea-
sure dimensions of mutant marker genes by the fact that the str-r47 gene is itself
a multisite mutant of very appreciable dimensions (Ravin and Dr Sa 1964).
Yet, it proves to have the same resistance as virtually all of the amiA genes
examined, which, with one exception, map as nonoverlapping mutants in the
most intensely marked segment of DNA thus far known in a transformation
system.

Finally, it should be noted that the multisite mutation r30 (or its allele s30) is
more, not less, readily integrated than the sites r5 and r/7, which 30 covers. This
indicates that there is no simple relationship between efficiency and linear dimen-
sions of genetic markers, just as there is no simple relationship between sub-
critical heat inactivation and efficiency.

Proximity to an end, and molecular weight heterogeneity of DNA as possible
causes of efficiency differences: The possibility that our LE genetic markers are
situated on DNA fragments of low molecular weight and therefore are less well
absorbed by competent cells, is excluded by the results of crosses of the individual
mutants by a single DNA preparation, wild-type for the amiA cistron. Since all
of the amiA markers are linked, the same kind of DNA particle clearly is re-
sponsible for transformation of the various sites back to wild type, be they HE
or LE.

While the mapping data do suggest that there is no obvious relationship be-
tween position of a mutant site within the cistron and its efficiency, a possible
way of retaining the “proximity to an end” hypothesis as an explanation of low
464 H. EPHRUSSI-TAYLOR et al.

 

Oo 180
100 32 = $9 2 .
. © ami-$30 4
t & ami-$29 4
r 6 @ ami- si7 1
50 F 4
L : 4 J
e@

 

 
    

O amia-r 30 4
© amiA-r5 4

Percent of initial activity

g

 

 

1 L
30 60 30 120 Igo
Minutes heated at 83°C

 

Ficure 4—~Subcritical heat inactivation of mutant 730, its s allele, and r or s alleles of the
sites r30 covers. Upper curve, 30 and sf7. Lower curve 730 and r5. In the upper curve, data on
HE site s20 are included for comparison.

efficiency has occurred to us. Were the pneumococcal chromosome circular, and
were a mutational event leading to breaking the circle not lethal, one could
suppose that LE mutations_are always associated with a chromosomal break,
while HE mutations are not. Thus, each LE mutant would be adjacent to an end
of the chromosome, and would be near the end of a DNA molecule upon break-
down of the chromosome during the DNA extraction procedure. Further, when
a LE mutant is serving as recipient cell in transformation, recombination could
be strongly influenced by proximity of the mutant site in the recipient cell to
the chromosomal break. One could, in this way, have LE mutants at any point
of the cistron, which would show equally low efficiency in both directions of the
crossing. The presence of a break near each LE mutant would not disturb the
relative positions of the sites on the map, although it might disturb mapping with
two-point crosses if sites of a cross were far enough apart to have been separated
by a break. Since mapping has proven possible only when the sites in two-point
crosses are quite close, recombination data alone could possibly not reveal the
breaks, Another test of the “proximity to an end” hypothesis has thus been
sought.

Ultraviolet light (UV) inactivates genetic markers when DNA is exposed in
vitro (LaATARJET, REBEYROTTE and DemERsEMAN 1957; LERMAN and TotmacH

 
RELATIVE TRANSFORMATION EFFICIENCY 465

1959; Lirman and Epurussi-Tayior 1959). It has been previously observed that
various pneumococcal markers used in our laboratory (str-r41, amiA-rf and
opt-r2) have different UV sensitivities (Lirman and Epnrussi-Taytor 1959).
The first two mutants, both highly efficient as transforming factors, are UV
resistant, while the last mutant, showing a low efficiency, is UV sensitive, as
shown in the curves of Lirman and Eprrusst-Taytor. Marked differences in
UV inactivation rates have been reported by others, working with transforming
DNA in several bacterial systems. Possible explanations of differing sensitivities
to UV have been enumerated and discussed (Marnmur et al. 1961). In personal
discussions, LERMAN has suggested that genes near the end of a DNA molecule
might be expected to be more resistant to the inactivating effects of UV irradia-
tion, owing to the loss of excitation energy at the end of the molecule. This hy-
pothesis was invoked to explain differences in rates of UV inactivation of differ-
ent genetic markers in transforming DNA. In view of recent work on the mode of
action of UV on transforming DNA (SreTriow and Seritow 1962) and in par-
ticular the demonstration that dimerization of excited thymine residues is an
important cause of inactivation, we can assume a somewhat different form of
LERMAN’s hypothesis insofar as this type of lesion is concerned, namely, that
thymine residues in terminal sequences give rise more readily to dimers, owing
to the possibility of easy strand separation and rotation of single strands at the
ends of the molecule. Other suggested causes of different inactivation rates are:
difference in target size of individual markers; differences in sizes of the mole-
cules on which the markers are located; different localized sequences of bases in
the DNA molecules which undergo photochemical changes with greater ease
(Marmour, et al. 1961). Clearly, by studying the UV sensitivity of a significant
number of closely linked mutant sites, several of these possible explanations can
be eliminated. In the only reported study which approaches this end, the pair of
linked markers ery-r2 ery-r3 was examined (Marmour, ef al. 1961). It was
concluded that the greater UV sensitivity of the ery-r3 marker relative to the
ery-r2 is not due.to differences in molecular weight of the DNA molecules in-
volved, since both were on the same molecule. On the other hand, thermal sta-
bility at subdenaturation temperatures indicated a greater dimension of marker
ery-r3, There remained, thus, in the study of Marmur, et al. (1961), three
plausible explanations of greater UV sensitivity of the ery-r3 marker: greater
target size, special local base composition favoring photochemical damage, and
proximity to an end.

The series of mutants shown in Figure 2 offers a unique occasion for testing the
correlation between UV sensitivity and the efficiency of a transforming factor,
under conditions in which we shall be able to narrow down the causes of UV
sensitivity to a single cause: namely, highly localized factors which can, in last
analysis, only be structural features in the neighborhood of the genetic markers
in question, or of the markers themselves.

By irradiating with UV dilute solutions of the DNAs from various strains and
titering residual activity of both amiA-r markers and the reference marker
str-r41 in these solutions, it has been possible to show a perfect correlation be-
466 H. EPHRUSSI-TAYLOR et al.

 

 

 

100 T T 1 ¥ T T v T T T T T ¥
eo 4
4
Q
e
x
~ - 4
2
°°
Qo " hi h
— I
Siok A\o ” " +
= C efficiency 4
c
& g 4
~ 4
3 L.
< i 7
@ . =
oO
3
a - " coon 7
AY low: efficiency
1 1 L i L 1 1 L 1 1 1 4. 1 1

 

(0) l 5 10
Minutes of irradiation at dose rate I71 ergs/ sec /mm*

Ficure 5.—Relative rates of inactivation of HE and LE mutants by UV irradiation.

tween UV sensitivity and efficiency of amiA marker genes in transformation.
Five HE markers examined (amiA-r1, r2,r19, r22, and r36) showed a sensitivity
to UV similar to that of the relatively resistant reference gene str-r41. Seven LE
genes (amiA-r21, r13, r17, r6, r26, r46, and r53) showed in every instance a
much greater sensitivity to UV inactivation. Figure 5 shows several of the inacti-
vation curves observed.

We have excluded above the possibility that efficiency differences are caused
by the amiA genes being on DNA fragments of different size, and shown that it
is unlikely that they have widely different linear dimensions, both of which
structrual features have been invoked to explain not only efficiency differences,
but also increased UV sensitivity of transforming factors. In view of the high
degree of correlation just demonstrated between UV sensitivity and efficiency
we can conclude that the UV sensitivity and efficiency of the amiA mutants are
determined by the same structural feature of the DNA, and that this feature is
neither molecular weight heterogeneity, nor differences in linear dimesions.

If we retain the idea that low efficiency and UV sensitivity are correlated with
proximity of the marker site to an end of a DNA molecule, we can suppose that
greater UV sensitivity is due either (a) to higher yield of damage per photon
absorbed near the end of DNA molecules, or (b) toa greater effect of a damaging
hit on recombination, for marker genes near an end. As mentioned already,
transformation of mutant cells to wild type can be performed and scored quanti-
RELATIVE TRANSFORMATION EFFICIENCY 467

tatively, just as well as transformation of wild type to mutant. LE mutants are
spread throughout the amiA cistron, and in order to invoke proximity to an end
as a cause of low efficiency, it was necessary to suppose that in each instance the
event leading to mutation caused a break in the DNA, near each mutated site.
However, the same sites in wild-type DNA should not all be proximal to breaks.
(If the DNA is broken randomly upon extraction all marker genes have equal
probability of being near an end. If it is broken nonrandomly, a restricted
number of sites would be near ends.) We can, thus, test whether the wild-type
alleles of LE mutations show high or low sensitivity to UV irradiation. To do
this, it is enough to irradiate a single DNA preparation, wild-type for amiA but
bearing str-r41, and follow the inactivation of normal amiA-s alleles by titering
on the corresponding amiA-r strains. If proximity of a particular mutant site to
an end is the cause of low efficiency and of UV sensitivity, then we would expect
its wild allele to be UV resistant, although showing low efficiency when tested
on a recipient strain bearing the mutant allele (because of proximity to a break
in the chromosome of the recipient cell).

UV inactivation curves have been determined for the wild-type sites cor-
responding to a variety of mutant sites within the amiA cistron by the UV tech-

 

 

 

 

 

 

100 100
50 r 4a
40F © amid-si? r
© eo" i?
3ob r
Oo a
!
ot s t
Ly omiA-s46
a" 146 Oo
a
ory vos OE
2 C
2 @ str-r al ~ -
8 + amia-s22 2
>
z BY]
€ _ L
3 2
2 z f
s
= 3
3 —
= v $
é ¥ . $
° 3
= a *
Sore c
5 L
a 5
[ Qj str 4
O amia-130 mutant DNA
@& str-r4 a
@ amia-s3o fs "Ye ONA
1 1 1 or 1 Fa
2 3 4 5 | 2 3 4 5 6 a a 3 10 ul (2 i3 14
MINUTES OF UV. IRRADIATION Minutes of irradiation ot dose rote I7.1 ergs/sec/mm?
Ficure 6.—Relative rates of inactivation of Ficune 7.—Relative rates of inactivation of

s alleles of HE and LE classes, by UV irradia- rand s alleles of multisite mutant 30, compared
tion. The curves for both s and r alleles are with the reference gene str-r41,
shown for the LE genes.
468 H. EPHRUSSI-TAYLOR et al.

niques used above, using synthetic medium supplemented with isoleucine for
scoring amiA-s transformants in the amiA-r recipient populations. Some of the
results of such a study are shown in Figure 6; the curves for each HE marker
tested, str-r41, amiA-s1, amiA-s19, amiA-s22 and amiA-s36, are essentially the
same and all are equivalent to the inactivation curves for the corresponding
mutant sites. Of the LE marker genes tested, s12, s/7, s46 and s53, site r17 seems
to be slightly more resistant than its s allele. Otherwise r and s alleles show essen-
tially the same sensitivity to UV. From the fact that both mutant and wild-type
alleles of the LE mutants tested are UV sensitive, it may be concluded that the
“proximity to an end” hypothesis is not readily tenable for explaining low effi-
ciency and UV sensitivity of the LE mutants in the amiA cistron.

UV sensitivity of gene amiA-r30 and its wild-type allele: The characteristics of
gene 730 are that it maps as a multisite mutant, and that it shows relatively high
efficiency as a transforming factor, in spite of covering two LE sites. With respect
to UV irradiation, the mutant allele shows a sensitivity identical to that of the
str-r41 gene, while its wild-type allele shows the sensitivity of a typical LE site
(Figure 7). This is the first instance in which a great difference has been found
between a mutant site and its wild-type allele. The sensitivity difference is the
mverse of that postulated if mutation were to create a break in the DNA
molecule adjacent to the mutant site. These results are compatible, however, with
mutation 730 being a deletion: the mutant would have virtually no linear
dimension, and have lost a segment of DNA particularly sensitive to UV (sites
5 and 17). Its wild-type allele would have appreciable linear dimension, and
contain a UV-sensitive segment. Inactivation of 730 would be due to narrowing,
by UV damage, of the zone within which switching must occur during recombi-
nation, according to the model of Srani (1959), whereas inactivation of 530
would occur, in addition, through hits in the UV-sensitive segment corresponding
to the 730 deletion.

DISCUSSION

The experiments described show that the efficiency with which a given genetic
marker is transferred to, or transcribed into a recombinant chromosome in trans-
formation is locally determined. These results do not exclude the possibility that
in some specific instances, gross structural features such as molecular weight
heterogeneity, or position relative to the end of a DNA molecule, are operating
to determine efficiency. However, the present investigation is the first designed
specifically to explore factors determining efficiency, and has employed a genetic
system possessing unusually advantageous features. It has failed to support the
notion that either of the gross structural features just mentioned plays any role
in determining differential efficiencies of various genetic sites in a given DNA
preparation and has shown that efficiency is in all likelihood site-specific for each
marker examined, and independent of the polarity of transformation cross. In the
absence of any experimental proof that efficiency differences between individual
genetic markers in the same DNA preparation can be influenced by gross struc-
RELATIVE TRANSFORMATION EFFICIENCY 469

tural features, the weight of existing evidence is in favor of efficiency being
determined by local structural features.

The four most striking findings of the present study, which will now be dis-
cussed, are the following: (1) Mutants which do not show the characteristics of
deletions or multisite mutants are distributed into two nonoverlapping classes
with respect to efficiency: high (HE) and low (LE = 0.1 x HE). (2) A multisite
mutation which covers two LE sites shows a relatively high efficiency. (3) Spon-
taneous mutation gives rise to mutants of both efficiency classes, while the muta-
gens thus far examined give rise only to LE mutants. (4) There is a strict cor-
relation between UV sensitivity of a site and the efficiency with which the
unirradiated site will be included in a recombinant chromosome through trans-
formation.

The existence of two nonoverlapping efficiency classes into which our first 73
mutants fall can only mean that the recombination process by which donor sites
are integrated into a recipient-cell genome is subject to discontinuous variation,
under the control of localized structural conditions. amiA-r30, the one proven
multisite mutant, shows the lowest efficiency of the HE class. It covers two re-
combining LE sites, which are also UV-sensitive sites. From the fact that wild-
type site 30 is UV sensitive, while mutant site 30 is UV resistant, one can infer
that mutant 730 is a deletion which eliminated a segment of UV-sensitive se-
quence. One arrives also at the conclusion that a deletion mutation makes a rela-
tively efficient transforming factor. If one examines the efficiencies of the few
mapped multisite mutants known for pneumococcus, notably the three reported
by Lacks and Horcuxiss (1960) in the amylomaltase locus, and the several
mapped in the str-r locus by Rornerm and Ravin (1961 ), Rormerm (1962) and
Ravin and Dz Sa (1964), no sharp discontinuity in the distribution of efficiencies
is discerned: they are, relative to str-r41, 0.05, 0.13 (2), 0.39, 0.41, 0.61, 0.87, 1.0.
(It should be noted that str-r41 is itself a multisite mutation). We are, therefore
examining the hypothesis that mutants which do not clearly fall into either the
HE or LE class are multisite mutants. Mutant markers amiA-r32 and r40 show
efficiencies of over 1.6 in crosses with wild type, irrespective of the polarity of the
cross. Whether they are multisite, or point mutations of a third efficiency class,
or simply mutants falling in the extreme upper end of the HE distribution,
remains to be discovered.

At present, however, we have only two clear efficiency classes into which our
mutants fall. There are, in DNA, two types of base pairs. Is it possible that
whenever a mutation is a transition of GC to AT it falls in one efficiency class,
and whenever the mutation is an AT to GC transition it falls in the other? Both
kinds of transitions would result in no hydrogen bond at the level of the mutated
base pair in a paired structure composed of a single strand of DNA of the re-
cipient cell, and the complementary single strand of the donor. One could raise
the question whether, for instance, when an unpaired couple has G in the re-
cipient strand and a mutant T opposite it in the donor strand, the T is rejected
from the final recombinant structure more often than accepted, thus leading to
470 H. EPHRUSSI-TAYLOR ef al.

low efficiency for mutants which are GC AT transitions. This model in its
simplest form is excluded by the results of the reciprocal transformation experi-
ments, which show that efficiency is independent of the direction of the cross.
However, this type of model could be retained if we suppose instead that every
time there is an unpaired G or C in the recipient strand, the recombination
process excludes the donor base, whereas an unpaired A or T leads either to no
bias, or preferential inclusion of the donor base. (The choice of base pairs here
is arbitrary.)

The next most obvious “highly localized structural feature” which could
determine efficiency is the composition of the base sequence adjacent to the mu-
tation, Let us consider for a moment whether our present information furnishes
a clue as to what sequence(s) might give rise to LE mutants.

If one tries to construct an argument on the basis of the fact that the two
mutagens successfully employed, HNO, and EMS, give rise only to LE mutants,
several difficulties are encountered. The first is that the chemical action for
neither agent is so unique and specific that their mutagenic effects can be at-
tributed to known chemical modifications of DNA. The one feature common to
the two mutagens is that under the conditions in which they are active, guanines
are altered and eventually hydrolyzed off of the polymer chain (Lirman 1961;
Lawiey 1961). If the removal of guanine (leading either to a transition or a
point deletion, for instance) were the cause of induction of mutations by these
two agents, one could advance the proposition that such mutations are frequent
in GC tracts, and that GC tracts of the donor are selectively rejected by the
recombination process, causing low probability of inclusion of the mutation into
the recombinant chromosome. Here a second difficulty arises. While the action of
the mutagens on guanine may be the sole mutagenic event, there is no known
reason why a GC pair adjacent to AT base pairs should not be affected some of
the time by the mutagen. Knowing what kind of chemical change is mutagenic
is not enough. What we need also to know is the nature of the base pairs adjacent
to the altered base.

There is a strong correlation between great UV sensitivity and low efficiency.
This means that the special structural features leading to low efficiency also lead
to great UV sensitivity. UV irradiation of DNA causes thymine dimer formation,
and this has been shown to account for about 70% of the inactivation of two
different genetic markers in DNA of Hemophilus, at high UV doses (SerLow and
SeTLow 1962; 1963). The two genes examined by the SerLows showed somewhat
different UV inactivation rates. Therefore, there is an indication from their
studies that sensitivity differences of genetic markers are due to photochemical
damages other than thymine dimer formation. However, nothing is known of
the genetic complexity of the two Hemophilus markers examined, or whether
gross structural factors are influencing their UV sensitivities. A detailed study of
the mapped amiA genetic markers in pneumococcus, using the techniques of
SerLow and Seriow for evaluating the fraction of damage due to thymine dimer
formation, would provide more crucial evidence concerning the chemical basis
of different UV sensitivity of transforming factors, We have attempted such a
RELATIVE TRANSFORMATION EFFICIENCY 471

study using yeast photoreactivating enzyme to destroy thymine dimers (WULFF
and Ruperr 1962) but have encountered complications, Photoreactivating en-
zyme prepared by us from Baker’s yeast contains a factor which inactivates
unirradiated DNA, and which is partially inhibited by visible light. No clear
experimental results on UV damaged DNA can be expected under the circum-
stances. It will be necessary either to purify the enzyme preparation or resort to
monochromatic UV, in order to evaluate the role of thymine dimer formation in
inactivation of our LE and HE mutations.

One can well ask, however, whether the different UV sensitivities of high and
low efficiency mutants cannot be explained after all on a target-theory basis,
rather than by supposing special base sequences to be particularly UV sensitive.
Were the genetic integration of LE marker genes to proceed obligatorily through
the insertion of relatively long segments of DNA, while that of HE genes pri-
marily through the insertion of short sequences, the observed UV inactivation
results would be explained. One would have to assume as a correlary that de-
purination does not affect the recombination process, since subcritical heating
yields no differential inactivation of HE and LE sites. This hypothesis is the
equivalent of saying that there are two qualitatively different recombination
processes: the one, the rarer event, for the insertion of long segments of donor
DNA; the other, a frequent event, for the insertion (or transcription) of short
sequences of donor DNA into recombinants. The decision as to which process will
be successful in inserting a particular mutant could depend on the base composi-
tion around the mutant, or the structure of the mutant itself (transition, trans-
version, point deletion). Since HE site r30 and LE site r53 are exceedingly close
together on the map, both processes would have to be able to occur in this region,
but in a mutually exclusive fashion. The frequent process can be visualized as a
rapid “switching” process which selectively eliminates LE sites from the re-
combinant structure. Only when the rarer, qualitatively different process occurs
would LE sites be inserted into a recombinant. This is a hypothesis that can be
tested by genetic experiments, which are now under way.

If efficiency and UV sensitivity are determined primarily by the base composi-
tion adjacent to the mutant site, then, clearly, another correlation can be pro-
posed; namely, that EMS and HNO; are mutagenic only when they alter a base
adjacent to the particular sequence responsible for both UV sensitivity and low
efficiency. It is quite clear from the work of Cuampe and Benzer (1962) that
base-analogue and other chemical mutagenesis in double stranded DNA is a far
more complicated process than is often supposed. They attempted to identify the
base transitions involved in a large number of mutations, induced in a variety of
ways, by measuring the ability of mutants to revert under the influence of base
analogues and hydroxylamine. Their rather low degree of success in obtaining
clear identification revealed the complexity of mutagenesis, and led CHAMPE
and Benzer to suggest that composition of base sequence adjacent to an altered
or substituted base might be playing a role in the mutation process, Thus, our
suggestion that EMS and HNO, cause mutations only when they alter a base
adjacent to a particular base sequence finds ample echo in the experiments cited.
472 H. EPHRUSSI-TAYLOR et al.

If, for example, the alteration of a base adjacent to A-T repeats (or G-C repeats)
were to lead to a transition, then it need not matter which particular base be
altered by the mutagen. We could then accept the prevailing evidence that
HNO.-induced mutation is caused primarily by alteration of cytosine and
adanine (ScHusTER and VIELMUTTER 1961; Lirman 1961; Wirrmann 1960;
WitTMANN and WirtmMann-Liepouip 1963), while that of EMS is due to altera-
tion of guanine (Law ey 1961).

An alternate explanation of efficiency differences can be advanced, but it seems
less likely than that just discussed. This is that the mutational event itself creates,
or not, an incompatibility between donor and recipient DNAs. An inversion, or a
very small deletion might, for example, cause pairing difficulties. The reason for
considering this explanation less likely is that all of the mutations induced by
HNO, and EMS, and half of the spontaneous mutations would have to be of this
class. This seems a very high proportion of the total mutants examined. In other
organisms, an appreciable number of HNO.-induced mutations have, indeed,
proven to be missense mutations resulting from single base transitions (Tsucrra
and FraENKEL-Conrat 1960; Tsucira 1962; Wirrmann 1960; Wirrmann and
WITTMANN-LiEBOLD 1963), causing single amino acid substitution in proteins,
not deletions or gross alterations of base sequence.

Differences in the ease with which highly localized points of a chromosome can
participate im recombination are not generally recognized in genetic systems
other than transformation and transduction. That such differences probably do
exist, however, has already been pointed out by Epurussi-TayLor (1961). This
question will be re-examined in connection with the presentation of the results of
two-factor crosses, in the report to follow.

The possibility that we may be able to recognize deletions on the basis of
(a) their transforming efficiency when in donor DNA and (b) differential UV
sensitivity of the deletion and its wild-type allele, is one promising perspective
opened up by the present study. The possibility that one may be able to map the
position of a particular type of base-sequence simply by mapping LE and HE
mutants is another. We are, therefore, making every effort to determine the
precise causes of high and low efficiency of transforming factors.

ACKNOWLEDGMENTS

We thank Mr. Josep Veness for his able technical assistance, Miss Jupy Oswa to for isolat-
ing and characterizing mutants 57 through 73, and Dr. Drew Scuwartz for numerous discus-
sions in the course of the work,

SUMMARY

A group of 73 mutants in a single functional unit which determines resistance
to aminopterin and sensitivity to an imbalance of branched amino acids dis-
tributes into two nonoverlapping classes with respect to the efficiency with which
they are integrated into recombinant cells. High and low efficiency differ by a
factor of about 10. Spontaneous mutants distribute equally between the two
classes. Ethyl methane sulfate and HNO, induce only low efficiency mutants.
RELATIVE TRANSFORMATION EFFICIENCY 473

Thirty-nine of the mutants have been mapped at 30 sites, and while the re-
mainder have not yet been completely mapped, it is clear that they lie at a
number of additional sites. There is no simple relation between position on the
map and efficiency. Independent mutations at the same site, yielding non-
recombining mutants, are thus far all of the same efficiency class. Efficiency is
the same in both directions of transformation: mutant cells by wild-type DNA
or wild-type cells by mutant DNA. Existing reversion data suggest that neither
class can be described as deletions. Genetic evidence indicates therefore, that
efficiency is site-specific, and determined by highly localized factors. |

The consequences for biological activity of damaging transforming DNA in
vitro have been examined, studying effects on mutant and wild-type alleles at the
same site, and at different sites representing the two efficiency classes. Sub-
critical heat inactivation (depurinization) affects identically all marker genes
tested, be they at high or low efficiency sites, mutant or wild-type. The one
multisite mutant thus far found, r30, is inactivated by depurinization at the same
rate as two recombining alleles it covers. Doubt is thus cast on subcritical heat
inactivation as a sensitive measure of the dimensions of a mutant gene. UV
irradiation has a strongly differential effect on mutants of the two efficiency
classes: low efficiency genes are far more inactivated by low UV doses than are
high efficiency genes. With one exception, UV sensitivity is shown to be very
similar or identical for mutant and wild-type alleles at a given site. The one
exception is mutant 730 which is UV resistant, while its wild-type allele is UV
sensitive. This is compatible with 730 being a deletion of a base sequence re-
sponsible for both low efficiency and UV sensitivity. This sequence contains two
low efficiency sites, 5 and 17. UV sensitivity as well as low efficiency appear, both,
to be site specific and caused by the same structural features of the DNA mole-
cule, or of the recombination process.

On the basis of these observations the following hypotheses are advanced, and
are currently under investigation: (1) That multisite mutations in general may
be characterized by different UV sensitivities of the mutant and its wild-type
allele. (2) That recombination in transformation proceeds through two distinct
processes. In the one, responsible for genetic integration of high efficiency mu-
tants, recombinants are constructed by a process in which rapid switching is
occurring between donor and recipient DNA sequences. Low efficiency mutants
are screened out of recombinants by this process, either because of special base
composition in the region of the mutant, or because of the nature of the mutant
(transversion, transition, point deletion). The second, rarer and qualitatively
distinct process is presumed to involve the transfer to the recombinant of rela-
tively long sequences of donor DNA, including both high and low efficiency sites.

LITERATURE CITED

ANacnostopou.os, C., and I. P. Crawrorp, 1961 Transformation studies on the linkage of
markers in the tryptophan pathway in Bacillus subtilis. Proc. Natl, Acad. Sci. U.S. 37:
378-390.
474 H. EPHRUSSI-TAYLOR et al.

Cuampg, S., and S. Benzen, 1962 Reversal of mutant phenotypes by 5-fluorouracil: an approach
to nucleotide sequences in messenger-RNA. Proc. Natl. Acad. Sci. U.S, 48: 532-546.

ErurattExizor, E., P. R. Srmvivasan, and S. Zamenuor, 1961 Genetic analysis, by means of
transformation, of histidine linkage groups in Bacillus subtilis. Proc. Natl. Acad. Sci. U.S. 47:
56-63.

Epurussi-Tayior, H., 1961 Recombination analysis in microbial systems. pp. 39-75. Growth
in living systems. Edited by M. K. Zarrow. Basic Books, New York.

Green, D. M., 1959 A host specific variation affecting relative frequency of transformation of
two markers in pneumococcus. Exptl. Cell Res. 18: 466-480.

Gurnoza, W., and W. BR. Guiry, 1961 On the imactivation of transforming DNA by tempera-
tures below the melting point. Proc. Natl. Acad. Sci. U.S. 47: 633-638.

Horcuuiss, R. D., and A. E. Evans, 1958 Analysis of the complex sulfonamide locus of pneu-
mococcus. Cold Spring Harbor Sym. Quant. Biol. 23: 85-98.

Horcrxtss, R. D., and J. Marmur, 1954 Double marker transformations as evidence of linked
factors in desoxyribonucleate transforming agents. Proc. Natl. Acad. Sci. U.S, 40: 55-60.

Iver, V. N., and A. W. Ravin, 1962 Integration and expression of different lengths of DNA
during the transformation of penumococcus to erythromycin resistance. Genetics 47: 1355-
1368. :

Kimpatt, A. W., 1961 Confidence intervals for recombination experiments with microorgan-
isms. Biometrics 17: 150-153. ,

Lacks, S., and R. D. Horcuxiss, 1960 A study of the genetic material determining an enzyme
activity in pneumococcus. Biochim. Biophys. Acta 39: 508-517.

Laragver, R., N. REBEYROTTE, and P, DemersemMan, 1957 Action de peroxydes organiques sur
un facteur transformant du pneumocoque. pp. 61-69. Les perorydes organiques en radio-
biologie. Masson, Paris.

Law ey, P.D., 1961 The action of alkylating agents on deoxyribonucleic acid. J. Chimie Phy-
sique 58: 10411-1020.

Lerman, L. S., and L. J. Tormacu, 1957 Genetic transformation. I. Cellular incorporation of
DNA accompanying transformation in pneumococcus. Biochim. Biophys. Acta 26: 68-82.

1959 Genetic transformation. II. The significance of damage to the DNA molecule.
Biochim. Biophys. Acta 33: 371-387.

Lirman, R. M., 1961 Genetic and chemical alterations in the transforming DNA of pneumo-
coccus caused by ultraviolet light and by nitrous acid. J. Chimie Physique 58: 997-1004.

 

Litman, R. M., and H. Epurussi-Tayior, 1959 Inactivation et mutation des facteurs génétiques
de V’acide desoxyribonucleique du pneumocoque par ultraviolet et par l’acide nitreur. Compt.

Rend, 249: 838-840.

Lirr, M., 1958 The critical size of transforming deoxyribonucleic acid. Ph.D. Thesis, Depart-
ment of Chemistry, Harvard University, Cambridge, Mass.
Marmur, H., W. F. Anperson, L. Marnews, K. Berns, E. Gasewsxa, D. Lang, and P. Dory,

1961 The effects of ultraviolet light on the biological and physical chemical properties of
deoxyribonucleic acids. J. Cell, Comp. Physiol. 58: (Suppl.) 33-55.

Ravin, A. W., and J. D. H. De Sa, 1964+ Genetic linkage of mutational sites affecting similar
characters in pneumococcus and streptococcus. J. Bacteriol. 87: 86-96.

Roser, M., and R. D. Horcuxiss, 1961 Selective heat inactivation of pneumococcal transforming
desoxyribonucleate, Proc. Natl. Acad. Sci. U.S. 47: 653-669.

Rosenserc, B. H., F. M. Srrornax, and L. F. Cavariert, 1959 On the size of genetic determi-
nants in pneumococcus and the nature of variables involved in transformation. Proc. Natl.

Acad. Sci. U.S. 45: 144-156.
RELATIVE TRANSFORMATION EFFICIENCY 475

Rorueiar. M. B.. 1962 A genetic factor on the DNA molecule capable of depressing the fre-
quency of transformation at a linked site. Genetics 47: 599-609. :

Rornem, M., and A. W. Ravin, 1961 The mapping of genetic loci affecting streptomycin
resistance in pneumococcus. Genetics 46: 1619-1634.

ScuaErrer, P., 1958 Interspecific reactions in bacterial transformation. pp. 60-74. The Biological
Replication of Macromolecules. Cambridge University Press, London.

Scuusrer, H., and W. Vie_murrer, 1961 Studies on the inactivating and mutagenic effect of
nitrous acid and hydroxylamine on viruses. J. Chimie Physique 58: 1005-1010.

Sertow, J. K., and R. B. Sertow, 1963 Nature of the photoreactivable ultraviolet lesion in
deoxyribonucleic acid. Nature 197: 560-562.

Sertow, R. B., and J. K. Serrow, 1962 Evidence that ultraviolet induced thymine dimers in
DNA cause biological damage. Proc. Natl. Acad. Sci. U.S. 48: 1250-1257.

Sicarp, A. M., 1964 A new synthetic medium for Diplococcus pneumoniae and its use for the
study of reciprocal transformations at the amiA locus. Genetics 50: 31-44.
Sirotnak, F. M., RB. B. Lunt, and D. J. Hurcutson, 1964 The distribuuon of mutational sites
affecting resistance to amethopterin in Diplococcus pneumoniae. Genetics 49: 439-454.
Svan, F. W., 1959 Radiobiology of bacteriophage. The Viruses. Vol. 2, pp. 353-385. Edited by
F. M. Burnet and W. M. Sranuey. Academic Press, New York.

Tsucrra, A., 1962 The proteins of mutants of TMV. J. Mol. Biol. 5: 284-300.

Tsucrra, A., and FraeNKeL-Conrat, H., 1960 The amino acid composition and C-terminal
sequence of a chemically evoked mutant of TMV. Proc. Natl. Acad. Sci. U.S. 46: 636-642.

Wirrmann, H. G., 1960 Comparison of the tryptic peptides of chemically induced and spon-
taneous mutants of TMV. Virology 12: 609-612.

Wirrmann, H. G., and B. Wirrmann-Lizsoip, 1963 Tobacco mosaic virus mutants and the
genetic coding problem. Cold Spring Harbor Symp. Quant. Biol. 28: 580-595.

Wutrr, D. L., and C. S. Rupert, 1962 Disappearance of thymine photodimer in ultraviolet
irradiated DNA upon treatment with a photoreactivating enzyme from Baker's yeast. Bio-
chem. Biophys. Res. Comm. 7: 237-239,