A MINUTE CIRCULAR DNA FROM ESCHERICHIA COLI 15*

By Nicuotas R. CozzarE.ui,t Reeis B. Ketiy,t anp ARTHUR KORNBERG

DEPARTMENT OF BIOCHEMISTRY, STANFORD UNIVERSITY SCHOOL OF MEDICINE,
PALO ALTO, CALIFORNIA

Communicated May 10, 1968

In the course of preparing the intracellular replicative form of phage ¢X174
DNA (RFI), 2 we discovered an even smaller duplex circular DNA. This new
DNA, synthesized by uninfected as well as infected cells of Escherichia coli, may
be either an episome or a plasmid.’ For lack of evidence indicating its integra-
tion into the chromosome, we regard it provisionally asa plasmid. In this report
we describe some of the physical properties of this DNA and its distribution
among strains of &. coli. The function of this plasmid remains unknown.

Experimental Procedure.—Materials: Sources were as follows: CsCl (optical grade)
was obtained from Harshaw; Brij 58, a nonionic detergent, from Atlas Chemical; Sepha-
rose 4B from Pharmacia; egg-white lysozyme from Calbiochem; pancreatic DNase from
Worthington; H*-deoxythymidine from New England Nuclear Corp.; and ethidium
bromide, a gift from Boots Pure Drug Co., Nottingham, England.

Methods: Cells were labeled with H*-thymidine (1-2 * 10’ cpm/yzmole) by growth in
Fraser and Jerrel’s glycerol-casamino acids medium;' cells labeled with P#? (5-9 x 108
cpm/pmole) were grown in a medium described by Hershey,® supplemented with a mix-
ture of the 20 natural amino acids. Cultures were harvested in the logarithmic phase at
an Aso of 1.5-2.0.

Cytoplasmic DNA was released from cells by the following modification® of the Godson
and Sinsheimer lysis procedure.’ To about 5 X 10” cells suspended in 1 ml of 25% sucrose,
0.05 M Tris HCl, pH 8.0, were added 0.2 ml of lysozyme (5 mg/ml in 0.25 M Tris HCl,
pH 8.0) and 0.4 ml of 0.25 M Na,EDTA. Conversion to spheroplasts was essentially
complete after a 10-min incubation at 25°. The spheroplasts were ruptured by a further
10-min incubation at 25° with Brij added to a final concentration of 0.5%. Cellular
debris and 90-98% of the total DNA were removed by sedimentation at 60,000 X g
for15 min. At this stage, extracts may be analyzed for plasmid DNA by sucrose gradient
sedimentation.

For the further purification of plasmid DNA, the supernatant fluid was extracted three
times with redistilled phenol and the phenol removed by extraction with ethyl ether.
The solution was concentrated to an Ass of 500 by rotary evaporation and then digested
with pancreatic ribonuclease (0.3 ug/ml for 10 min at 37°) to the point where 10% of
the RNA became acid-soluble. The RNA fragments were removed by filtration through
a column of Sepharose 4B with a bed volume ten times that of the sample. The duplex
covalently closed plasmid DNA was separated from circular duplexes bearing single-
strand breaks and linear forms by banding in a solution of CsCl (p = 1.57 gm cm~*) and
ethidium bromide (100 ug/ml) for 36 hr or more at 40,000 rpm and 5° in the 50 rotor of
the Spinco model L ultracentrifuge.? The last step was repeated if the product was less
than 90% pure as judged by sedimentation in alkali. Ethidium bromide was removed
by extraction with an equal volume of isopropanol or by passage through a Dowex-50
column.’ The final material was dialyzed against 0.05 M potassium phosphate buffer,
pH 7.2,and1 mM EDTA. The yield of plasmid DNA from a 20-liter culture harvested
at an Ass of 1.5 was about 0.4 umole (expressed as nucleotide residues).

Sucrose gradient sedimentation of DNA samples was carried out as previously de-
scribed.* The buffer contained: for neutral low-salt sedimentation, 10 mM Tris HCl, pH
8.0, and 1 mM EDTA titrated to pH 8.0 with NaOH; for neutral high-salt sedimentation,
1 M NaCl in addition; and for alkaline high-salt-sedimentation, 0.8 M NaCl, 1 mM
EDTA, and 0.3 44 NaOH. Analytical determinations of buoyant density were made in

992
VoL. 60, 1968 BIOCHEMISTRY: COZZARELLI ET AL. 993

CsCl solution (p = 1.70 gm cm) in the Spinco model E ultracentrifuge at 44,000 rpm.
Kel-F centerpieces were employed in standard cells. Ultraviolet absorption photographs
were taken at regular intervals and subsequently scanned with a Joyce-Loebl recording
microdensitometer.

DNA was viewed with a Siemens Elmiskop 1A electron microscope using the protein
film technique of Kleinschmidt e? al. as modified by Inman e? al.!1_ The hypophase was
water. Magnification was determined from a diffraction grating of known line spacing.

Colicin 15 was assayed by the procedure of Mukai?? with Z. coli strain TAU™ as the
indicator and the colicin-resistant strain THU" as a control. The dose of UV light
used to induce colicin production was 900 ergs mm~?.

Results —Duplez circular plasmid DNA: In cells of strain THU, infected
with ¢X174, a form of DNA in addition to RFI of ¢X174 was discovered. This
new DNA had a density in an ethidium bromide-CsCl gradient characteristic of
closed circular duplex DNA and sedimented less rapidly than ¢X174 RFI. The
ratios of sedimentation velocities in alkaline high-salt, neutral high-salt, and
neutral low-salt sucrose gradients were 0.65, 0.76, and 0.71, respectively. DNA
with these same sedimentation properties was also found in uninfected cells.
Thus, this new DNA appears to be a plasmid form of host cell DNA. The ratio
of the alkaline to neutral sedimentation rates for the plasmid DNA is several
times that anticipated for a linear duplex" and is consistent with a closed circular
duplex structure.'* This supposition was tested by the action of pancreatic
DNase. Purified plasmid DNA, sedimented to equilibrium in a CsCl gradient in
the presence of ethidium bromide, yielded over 90 per cent of the DNA at a
density value characteristic of 6X174 RFI (Fig. 1A). When single-strand scis-
sions were introduced with pancreatic DNase, the DNA shifted into the lighter
peak occupied by an RFII marker (Figs. 1B, C). This effect of pancreatic DNase
on the banding pattern strongly supports a closed circular duplex structure of the
plasmid DNA.’ ,

Contour length: The circular form of the plasmid DNA was also revealed by
electron microscopy with the Kleinschmidt protein film technique (Fig. 2). The
contour length was found to be 0.96 » + 0.13 (sp) for the plasmid DNA and
2.26 » + 0.13 (sp) for 6X 174 RFTI included on the same grids. The two samples
of DNA were mixed and spread together to provide an internal control for the
variations in absolute length measured by electron microscopy.’” '* Based on a
molecular weight of the ¢X174 RF of 3.4 X 106 daltons,'* that of the plasmid
DNA is calculated to be 1.45 K 108. The molecular length distributions shown
in Figure 3, as well as the standard deviations, show that the length variation is
similar for both types of DNA and thus that the cytoplasmic DNA has a rather
homogeneous size distribution.

In a separate experiment in which no ¢X174 RFII standard was added, only
small circles and no polymeric forms® ®. #1. % 22 were found among 93 molecules
examined. Furthermore, when the cytoplasmic DNA was analyzed on an alka-
line sucrose gradient, no peak of radioactivity (less than 0.3%) was detectable at
the position anticipated for dimers.

Buoyant density: The buoyant densities of plasmid and T7 DNA were com-
pared with that of poly dAT (p = 1.678)?5 as a standard. The plasmid DNA
(p = 1.710) was 0.001 gm cm~ lighter than the T7 DNA. Since T7 and E. colt
994 BIOCHEMISTRY: COZZARELLI ET AL. Proc. N. A. 8.

 

1000

 

 

r BDNose ¥ 7
| somyug/ml 4

    
     

 

on —2— 19 sf 20 Fig. 1.—The effect of pancreatic DNase
. treatment on banding of plasmid DNA in a
CsCl-ethidium bromide gradient. Samples
were centrifuged in a 1.5-ml solution of CsCl
(0.75 gm/ml), 5 mM EDTA, and ethidium
bromide (100 ug/ml) at 45,000 rpm, 10°C,
| for 48 hr in the SB405 rotor of the Inter-
+ national B60 ultracentrifuge. The arrows
+ indicate the position of a marker of ¢X174
4  RFIIDNA. The plasmid DNA was treated
. with DNase at the indicated concentrations
for 60 min at 37°.

 

 

Hl 1
v 5 10 15 20
Fraction number

 

DNA have equal densities,?* the calculated GC content?* of the plasmid DNA is
comparable to that of Z. coli chromosomal DNA. The half-width of the plasmid
DNA band was 0.005 gm cm~—, which is in the expected range for a DNA of its
size,” indicating a lack of extensive density heterogeneity.

Molecular weight: Since the relationship between sedimentation velocity and
molecular weight for closed duplexes is complicated by the number of tertiary
turns,'* the sedimentation velocities of the open or “nicked” form of the plasmid
DNA and of ¢X174 RFII were compared. To ensure that the pancreatic DNase
was introducing only single-chain and not double-chain scissions into the P?*-
labeled plasmid DNA, conditions were chosen which supported the conversion of
at least 90 per cent of H?-lableled ¢X174 RFI (in the same reaction tube) to the
nicked circular RFII. The neutral and alkaline sedimentation velocities gave
molecular weight values of 1.53-1.63 X 10° daltons (Table 1), in good agreement
with the value of 1.45 X 10° obtained from the contour lengths.

Tertiary structure: ‘To determine whether the closed molecules contain tertiary
turns, the ratio of the neutral sedimentation rates of the closed and open forms of
the plasmid DNA were compared with that for ¢X174 RF and with the published
value for phage ADNA.™ The presence of tertiary turns in a DNA increases its
sedimentation coefficient.'° For plasmid DNA, at both high and low ionic
strengths, this ratio was greater than one, which indicates that the DNA contains
tertiary turns under these conditions (Table 2). The increase in the ratio of
sedimentation velocities at low ionic strength was marked for \DNA, insignifi-
Vou. 60, 1968 BIOCHEMISTRY: COZZARELLI ET AL. 995

 

Fic. 2.—Electron micrograph of plasmid and ¢X174 RFIT DNA. The smaller circles are the
plasmid INA, and the larger the viral RF.

cant for the plasmid, and intermediate for ¢X174 RI, suggesting that this ionic
strength dependence is a function of molecular weight.

Number of plasmids per chromosome: This value was estimated from the frac-
tion of the total DNA recovered as plasmid DNA in the supernatant after deter-
gent lysis. Spheroplasts of strain THU, uniformly labeled with H*-thymidine,
were lysed with Brij, and the amount of plasmid DNA in the supernatant fluid
was measured by sedimentation in an alkaline sucrose gradient with a marker of
purified P3-labeled plasmid DNA. Three such determinations gave percentage
values for the plasmid DNA of 0.64, 0.77, and 0.80 of the total cellular DNA.
Assuming a molecular weight of 2.8 X 10° for the resting &. coli chromosome,”

and 1.5 X 10° for the plasmid DNA, there are, on the average, 12 to 15 copies of
‘ plasmid DNA per chromosome. Our estimate of the number of plasmids might
be low because of breakage of DNA during handling, the possible natural occur-
rence of open forms (analogous to ¢X174 RII), or incomplete extraction. The
first two sources of error are probably not significant because single-stranded DNA
of molecular weight 7 10° was not detectable by alkaline sucrose sedimentation.
of DNA from Brij-lysed cells. To determine whether plasmid DNA was sedi-
mented with the host DNA after Brij lysis, the yield with this technique was
compared to that after Sarcosy! lysis** and phenol extraction of the total cellular
DNA. Both extraction procedures gave yields which differed by less than 50 per
cent.

Occurrence of plasmids in various iv. colt strains: The studies deseribed up to
this point were conducted with strain THU, a derivative of strain 15. Nine
other strains of 1, colt were also tested for the presence of the plasmid DNA
996 BIOCHEMISTRY: COZZARELLI ET AL. Proc, N. ALS.

   

 

oO:
8504
3
e
3%
w
230 re : : :
2 Fig. 3.—Distribution of contour
5 20- lengths of plasmid DNA and
- @X174 RFI. Linear or aggre-
10k gated molecules were not included
in these measurements; the ma-
jority of the forms were circular.

Oo 04 08 12 16 20 24 28
Contour length, microns

(Table 3). H®-thymidine-labeled cells were lvsed with Brij, and the supernatant
fraction was analyzed on an alkaline sucrose gradient (and in several cases also
on a neutral suerose gradient) with a marker of purified P-labeled cytoplasmic
DNA from strain THU. All derivatives of strain 15 showed a similar peak of
radioactivity coincident with the P?? marker. Figure 4A shows one such experi-
ment. Also, the DNA from strain 15T—D3 was purified and its contour length
determined. The length, 1.03 1 + 0.09 (sp), was the same within experimental
error as that determined for the plasmid from strain THU. In contrast to these
results with derivatives of strain 15, no cytoplasmic DNA (<0.02% of the total
DNA or <0.4 plasmid per chromosome) was detected in the strain K12 and B
derivatives. One such experiment is shown in Figure 4B. It is possible that
plasmids are synthesized by these strains but that extrachromosomal DNA is
degraded or is not released by our lysis procedure. This possibility appears un-
likely for the IX12 strains, at least, since high yields of ¢X174 RII can be ob-
tained from these strains by this procedure. It therefore appears that, among
the strains examined, the plasmid DNA is specific to strain 15.

Discussion.—All the derivatives of strain 15 studied possess the plasmid DNA

Taste 1. Molecular weight of plasmid DNA.

 

Buffers
Neutral Neutral Alkaline
high salt low salt high salt
Relative sedimentation velocity of ¢X174 RFII to
strain 15 plasmid DNA (open form) 1.32 1.29 1.35
Mol wt of plasmid DNA ( x 10~*) 1.43 1.68 1.61

The sedimentation buffers are described in Methods. The ratio of sedimentation velocities is the
average of 2-4 determinations and the average error ix 1%. The molecular weights were calculated
from the equation of Studier."

TABLE 2. Effect of single-strand scissions on sedimentation velocity of circular DNA.

Relative sedimentation

velocities of closed vs. EB. coli Circular
open forms plasmid oX174 RF DNA
High salt buffer 1.24 1.25 1.32*
Low salt. buffer 1.26 1.37 1.67*

The sedimentation buffers are described in Methods.
* From Bode and Kaiser.
Vou. 60, 1968 BIOCHEMISTRY: COZZARELLI ET AL. 997

TaBLe 3. Occurrence of plasmid DNA among ten strains of F. colt.

E. colt Colicin 15 UV-light Plasmid
Strain subspecies Reference production sensitivity DNA
THU 15 14 - Ss +
TAU 15 13 + Ss +
TAU-e 5 29 - 8 +
TAU-bar 15 30 — R +
15T~- 15 31 + S +
15T-D3 15 32 + S +
JGI51 15 33 ~ T +
B3 B 34 _ R -
HF4704 Kl12 3) - R -
HF4733 Kl2 36 ~ R -

All strains are thymidine auxotrophs; additional requirements are cited in the references. The
lytic response to UV light was determined by irradiation of cells with 900 ergs mm ~ of UV light and
subsequent turbidemetric monitoring of growth. For sensitive (S) strains, there was about a fivefold
decrease in optical density several hours after irradiation; for the tolerant (7) strain there was only
a slight drop in turbidity; resistant (#) strains continued to grow. Colicin assays were performed
on chloroform-treated cultures 6-8 hr after UV irradiation, the time of maximal clearing for sensi-
tive strains. Strains listed as not producing colicin failed to do so even after a tenfold concentration.
Strains THU, HF4704, and HF4733 were obtained from R. L. Sinsheimer;® TAU and TAU-bar from
P.C. Hanawalt;* TAU-e and 15T ~D3 from K. Matsubara;® 15T ~and JG 151 from A. Yudelevich;%4
und B3 from N. E. Melechen.**

even though some of these mutants were isolated as long as 14 vears ago.?!
Moreover, the only common parent of these strains is strain 15 itself which was
first described in 1944.%7 °° This genetic stability suggests a mechanism for
maintaining the plasmid in the population, perhaps an orderly segregation of
plasmids to daughter cells. It is further possible that occasional cured strains
are at a selective disadvantage or can be reinfected with the plasmid.

The plasmid DNA level of 15 copies per chromosome is high, especially since
it was found in exponentially growing cells and thus represents a steady-state
value.** By contrast there is only about one Flac or FColVColBtrycys*!
particle per chromosome. Multiple copies of a drug-resistance factor were found
only after transfer from an £. coli donor to a Proteus mirabilis recipient.** Tf we
assume that the many copies of plasmid DNA are genetically identical, a reason-
able assumption in light of their uniform molecular weight and density homoge-
neity, then it is puzzling why the cell maintains such extensive genetic redundancy.
Still a further question is the mechanism whereby the cell maintains the number
of copies of the chromosome and the plasmid at such markedly different levels.
The ease of identification of the plasmid DNA may make it a favorable system
for study of this mechanism of replicon” control.

The failure to detect the plasmid in IX12 or B led us to look at the properties of
strain 15 that are not shared by B and X12. Some derivatives of strain 14 pro-
duce a particulate colicin (colicin 15) which is apparently a defective phage. *?
It is not likely that the small plasmid DNA codes for the defective phage because
the virus appears to be large and thus would be expected to have a much larger
DNA. Also, four of the strains analyzed, THU, TAU-bar, TAU-e, and JG151, do
not produce colicin 15 but nevertheless contain the plasmid DNA (Table 3).
Strain TAU-c was obtained from TAU under curing conditions”? and specifically
lost the ability to produce colicin while retaining its plasmid DNA. Strain JG151
has also lost two other unusual properties associated with strain 15: high sus-
998 BIOCHEMISTRY: COZZARELLI ET AL. Proc. N.A.S.

 

 

 

 

 

 

600r—y T 1 T T T T oT
A. Strain JG151 B. Strain Bs
500 -1F- 4
e 4007 VW 180 2
‘ g
.

E 500 ‘e i de {603
v $ qi ?
z l I 4
200+ tf: I +40

” $ i
e t {
100F- \ st: | \ -120
p We vececeat! tetas f 4 *~
d4 ‘4 i ! | Meepebosssheogtenet tose 3
30

{
5 10 158 «620-2530 5 10 15 20 25
Fraction number

Fic. 4.—Neutral high salt-sucrose sedimentation of H?-labeled DNA
from strain JG151 (A) and strain B3 (B). P#?-labeled plasmid DNA
from strain THU was included for reference.

ceptibility to thymineless death,?? and the induction of a new methylase by UV
irradiation.‘4 The induction of lysis by UV irradiation is still another property of
some derivatives of strain 15 that also is not necessarily correlated with the pres-
ence of the plasmid DNA (Table 3). Thus far we have been unable to ascribe
any biological function to this DNA.

The strain 15 plasmid DNA is the smallest monodisperse DNA species thus far
described,’ and the smallest. microbial DNA of any kind.® One of its strands
could code for a polypeptide of only 75,000 in molecular weight and thus could
contain very few cistrons. Its small size, relative abundance, ease of separation
from chromosomal DNA, and its location in E. colt make it a convenient source of
material for the investigation of DNA structure, replication, and function.

Summary.—Strains of Escherichia coli 15 contain a supertwisted, circular,
plasmid DNA with a molecular weight of 1.5 million, the smallest known micro-
bial DNA. There are about. 15 plasmids per chromosome in the strain 15 cells
but none was detected in strains of FE. coli X12 or B.

* This work was supported in part by research grants from the National Institutes of Health
(USPHS) and the National Aeronautics and Space Administration,

+ National Science Foundation postdoctoral fellow.

t Helen Hay Whitney postdoctoral fellow.

1 Abbreviations: RFI and RFII denote the covalently closed and open duplex replicative
forms of @X174 DNA, respectively; EDTA, ethylenediaminetetraacetate; Tris HCI, tris
(hydroxymethyl)aminomethane HCI; poly dAT, polydeoxyadenylate-deoxythymidylate.

2 Burton, A., and R. L. Sinsheimer, J. Mol. Biol., 14, 327 (1965).

3 Jacob, F., and E. L. Wollman, Seruality and the Genetics of Bacteria (New York: Academic
Press, 1961).

4Fraser, D., and E. A. Jerrel, J. Biol. Chem., 205, 291 (1953).

5 Hershey, A. D., Virology, 1, 108 (1954).

®§ Komano, T., and R. L. Sinsheimer, Biochim. Biophys. Acta, 155, 295 (1968).

7 Godson, G. N., and R. L. Sinsheimer, Biochim. Biophys. Acta, 149, 476 (1967).

% Radloff, R., W. Bauer, and J. Vinograd, these Procegpings, 57, 1514 (1967).

’ 9 Goulian, M., and A. Kornberg, these Procegpines, 58, 1723 (1967).
VoL. 60, 1968 BIOCHEMISTRY: COZZARELLI ET AL. 999

10 Kleinschmidt, A. K., D. Lang, D. Jacherts, and R. K. Zahn, Biochim. Biophys. Acta, 61
857 (1962).

11 Inman, R. B., C. L. Schildkraut, and A. Kornberg, J. Mol. Biol., 11, 285 (1965).

12 Mukai, F. H., J. Gen. Microbiol., 23, 539 (1960).

13 Barner, H. D., and 8S. 8S. Cohen, Biochim. Biophys. Acta, 30, 12 (1958).

4 Stern, J. L., M. Sekiguchi, H. D. Barner, and 8. 8. Cohen, J. Mol. Biol., 8, 629 (1964).

% Studier, F. W., J. Mol. Biol., 11, 373 (1965).

16 Vinograd, J., and J. Lebowitz, J. Gen. Physiol., 49, 103 (1966).

17 Inman, R. B., J. Mol. Biol., 25, 209 (1967).

18 Lang, D., H. Bujard, B. Wolff, and D. Russel, J. Mol. Biol., 23, 163 (1967).

19 Sinsheimer, R. L., J. Mol. Biol., 1, 43 (1959).

2 Clayton, D. A., and J. Vinograd, Nature, 216, 652 (1967).

21 Roth, T. F., and D. R. Helinski, these Proceepines, 58, 650 (1967).

22 Rush, M. G., A. K. Kleinschmidt, W. Hellman, and R. C. Warner, these PRocEEDINGs,
58, 1676 (1967).

23 Schildkraut, C. L., J. Marmur, and P. Doty, J. Mol. Brol., 4, 430 (1962).

24 The half-width of the plasmid DNA band is 1.56 times that of 6X 174 RF, measured under
the same conditions. The calewlated ratio of molecular weights is therefore 2.43 (Meselson,
M., F, W. Stahl, and J. Vinograd, these ProceEepinos, 43, 581 (1957)), in good agreement with
that determined by electron microscopy.

% Bode, V. C., and A. D. Kaiser, J. Mol. Biol., 14, 399 (1965).

2¢ J. Wang has found that the tertiary structures of ¢X174 RFI and plasmid DNA are
relaxed by the same number of molecules of ethidium bromide per unit length of DNA (personal
communication). .

7 Cairns, J., in Cold Spring Harbor Symposia on Quantitative Biology, vol. 28 (1963), p. 43.

% Young, E. T., Il, and R. L. Sinsheimer, J. Mol. Biol., 30, 147 (1967).

29 Isolated from strain TAU after two growth cycles in the presence of 35 ug/ml acridine
orange, followed by thymidine deprivation.

* Hanawalt, P. C., Nature, 198, 286 (1963).

31 Barner, H. D., and S. 8. Cohen, J. Bacteriol., 68, 80 (1954).

32 Matsubara, K., A. Taketo, and Y. Takagi, J. Biochem. (Japan), 54, 225 (1962).

33 Tshibashi, M., and Y. Hirota, J. Bacteriol., 90, 1496 (1965).

34 Melechen, N. E., and P. D. Skaar, Virology, 16, 21 (1962).

% Lindqvist, B. H., and R. L. Sinsheimer, J. Mol. Biol., 28, 87 (1967).

36 A strain lacking endonuclease I isolated by P. H. Flanders.

7 Roepke, R. R., R. L. Libby, and M. H. Small, J. Bacteriol., 48, 401 (1944).

% Roepke, R. R., J. Bacteriol., 93, 1188 (1967).

39 The exposure of THU to 900 ergs mm~? of UV light, sufficient to initiate cell lysis within
1 hr, did not cause any change in the amount of plasmid DNA.

Jacob, F., S. Brenner, and F. Cuzin, in Cold Spring Harbor Symposia on Quantilative
Biology, vol. 28 (1963), p. 329.

“| Hickson, F. T., T. R. Roth, and D. R. Helinski, these Proceevines, 58, 1731 (1967).

42 Rownd, R., R. Nakaya, and A. Nakamura, J. Mol. Biol., 17, 376 (1966).

48 Nomura, M., Ann. Rev. Microbiol., 21, 257 (1967).

44 Yudelevich, A., and M. Gold, Federation Proc., 26, 450 (1967).

* Thomas, C. A., and L. A. MacHattie, Ann. Rev. Biochem., 36, 485 (1967).