Proc. Nat. Acad. Sci. USA
Vol. 72, No. 8, pp. 2999-
, August 1975
Distinct penicillin binding proteins involved in the division,
elongation, and shape of Escherichia coli K12
(P-lactamantibiotics/slab gel electrophoresis/binding protein mutants)
BRIAN G. SPRATT
Department of Biochemical Sciences, Moffett Laboratories, Princeton University, Princeton, New Jersey 08540
Communicated by Arthur B. Pardee, May 20,1975
cell division, cell elongation, and cell shape in E. coil are
shown to be due to the presence of three essential penicillin
binding proteins with distinct roles in these three processes.
(A) Cell shape: ,-Lactams that specifically result in the pro-
duction of ovoid cells bind to penicillin binding protein 2
(molecular weight 66,000). A mutant has been isolated that
fails to bindft-lactamsto protein 2, and that grows as round
cells. (B) Cell division:f-Lactamsthat specifically inhibit cell
division bind preferentially to penicillin binding protein 3
(molecular weight 60,000). A temperature-sensitive cell divi-
sion mutant has been shown to have a thermolabile protein
3. (C) Cell elongation: Oneft-lactamthat preferentially inhib-
its cell elongation and causes cell Iysis binds preferentially to
binding protein 1 (molecular weight 91,000). Evidence is pre-
sented that penicillin bulge formation is due to the inhibition
of proteins 2 and 3 in the absence of inhibition of protein 1.
The varied effects of #-lactam antibiotics on
The,B-lactam group of antibiotics (penicillins and cephalos-
porins) exert a variety of effects on Escherichia coli. Most
typical fl-lactams inhibit cell division at low concentrations
and produce cell lysis at high concentrations (1, 2). In addi-
tion, some of these compounds produce bulges in cells at in-
termediate concentrations (2). Another penicillin derivative
called FL1060) produces ovoid-shaped cells at low concen-
trations without specifically inhibiting cell division or caus-
ing typical penicillin lysis (3-5).
Although it has been suggested (2, 6-8) that the effects of
f,-lactam antibiotics on cell division, elongation, and shape
are due to the presence of distinct penicillin-sensitive en-
zymes specifically involved in peptidoglycan metabolism for
these three processes, little evidence has been produced to
substantiate this view. Multiple penicillin-sensitive enzymes
(7, 9, 10) and penicillin binding proteins (8, 9, 11) with dif-
ferent sensitivities to fl-lactam antibiotics are present in E.
coli, and recently we have shown that one of the six penicil-
lin binding proteins (protein 2) is specifically involved in the
maintenance of the rod shape of E. coli (8).
Three of the E. coil penicillin binding proteins (proteins 4,
5, and 6) are not thought to be involved in the effects of typ-
ical ,B-lactams on cell elongation and division since several
antibiotics fail to bind to these proteins at concentrations far
above those that affect the growth of the cells (Spratt, in
The remaining two binding proteins (proteins 1 and 3) are
the likely sites at which typical ,B-lactams act to inhibit cell
division and cell elongation.
In this paper, I report that one of these proteins is specifi-
cally required for cell division and that the other is required
for cell elongation; I also suggest a model to explain the var-
ied effects off3-lactamantibiotics by their relative affinities
for three proteins involved in cell division, elongation, and
the maintenance of cell shape.
The organism used in these studies was E. coil K12 (strain
KN126). It was grown in Difco Pennassay broth at 370 with
vigorous aeration and harvested at late logarithmic phase.
Mutants B6 and 6-30 were grown in the same medium at
Assay of Penicillin Binding Proteins. ['4C]Benzylpenicil-
lin or [14C]mecillinam (FL1060) were bound to purified cell
envelopes [prepared by sonication and differential centrifu-
gation (8, 12)], the inner membranes selectively solubilized
with Sarkosyl NL-97 (13), and the binding proteins sepa-
rated on sodium dodecyl sulfate (NaDodSO4)/polyacrylam-
ide slab gels and detected by fluorography (14) as described
In experiments in which the residual penicillin binding
proteins are measured after growth of cells in the presence
of unlabeledf3-lactamantibiotics, the membranes were pre-
pared without the use of 2-mercaptoethanol to avoid the loss
of bound penicillin.
Penicillin binding proteins were quantitated by densitom-
etry of the x-ray films and measurement of the areas under
each of the peaks.
Chemicals. [14C]Mecillinam (53 mCi/mmol) and unla-
beled mecillinam were generous gifts from Dr. F. Lund of
Leo Laboratories, Ballerup, Denmark. [14C]Benzylpenicillin
(54 mCi/mmol) was obtained from Amersham/Searle. Am-
picillin was a gift of Bristol Laboratories. Cephalexin and ce-
phaloridine were gifts of Eli Lilly and Co. Benzylpenicillin
was a gift of E. R. Squibb and Sons.
Identification of jB-lactam antibiotics with preferential
effects on cell division, elongation, and cell shape
To identify penicillin binding proteins with distinct roles in
cell division, elongation, and shape, we screened 13 antibiot-
ics* for those which showed preferential effects on one of
these three processes. Each antibiotic was added at a range
of concentrations to cells of E. coil KN126 growing exponen-
tially in Penassay broth at 370, and the cells were examined
at 15-min intervals under the phase contrast microscope.
*6-Aminopenicillanic acid, ampicillin, benzylpenicillin, cephalex-
in, cephaloglycin, cephaloridine, cephalothin, cephalosporin G,
cephamycin C, cefoxitin, dihydroampicillin, mecillinam, and
Abbreviation: NaDodSO4, sodium dodecyl sulfate.
Proc. Nat. Acad. Sci. USA 72 (1975)
to envelopes prepared from strains KN126 and B6. [14C]Benzyl-
penicillin (31 tig/ml)or [14C]mecillinam (2 pg/ml) were bound to
purified cell envelopes prepared from strains KN126 or B6 grown
in Penassay broth at 300 as described (8). The reaction was ter-
minated, and the inner membrane selectively solubilized with Sar-
kosyl NL-97 (13), and separated on NaDodSO4/polyacrylamide
slab gels as described (8). The binding proteins were detected by
fluorography (14) on Kodak RPRoyal x-ray film for 48 days at
-700. [14C]Benzylpenicillinwas bound for 10 min at 300 (A) or 420
(B) to envelopes prepared from strain KN126. [14C]Benzylpenicil-
lin was bound for 10 min at 300 (C) or 420 (D) to envelopes pre-
pared from strain B6.["ClMecillinamwas bound for 10 min at 300
(E) or 420 (F) to envelopes prepared from strain KN126. [14C]Me-
cillinam was bound for 10 min at 30° (G) or 420 (H) to envelopes
prepared from strain B6. Proteins 5 and 6 are not fully resolved on
this gel. In addition to the six binding proteins we consistently de-
tect, two further minor binding proteins are also detected on this
gel. Electrophoresis was from top to bottom.
Binding of [14Cjbenzylpenicillin and [14C]mecillinam
Cell Shape. A binding protein involved in the mainte-
nance of cell shape can be readily identified using the ami-
dinopenicillanic acid, mecillinam (3, 8). Mecillinam is an
ideal probe for cell shape, since it is specific for shape and
does not initially inhibit cell division or cause typical penicil-
Cell Division. Most otherfl-lactamantibiotics inhibited
division (without inhibiting cell elongation) at low concen-
trations and thereby caused filamentation; cell lysis occurred
at higher concentrations. To probe for a penicillin binding
protein involved in cell division we selected cephalexin, am-
picillin, and benzylpenicillin, as these antibiotics showed the
largest concentration range over which cell division was spe-
cifically inhibited without cell lysis occurring.
Cell Elongation. A binding protein involved in cell elon-
gation was probed for with cephaloridine, since thisfl-lac-
tam caused cell lysis (presumably by inhibiting cell elonga-
tion) at the lowest effective concentrations.
Bulge Formation. This was observed with only two of the
(3-lactams examined. One of these, ampicillin, was used as a
probe of bulge formation. At low concentrations ampicillin
inhibited cell division, resulting in the production of fila-
ments. At slightly higher concentrations bulges were pro-
duced in the middle of the filaments, while lysisoccurred at
still higher concentrations.
Involvement of penicillin binding protein 2 in cell
Mecillinam competes for the binding of ['4C]benzylpenicil-
lin to only one of the six penicillin binding proteins (protein
2), and we have suggested that protein 2 is involved in the
ence of cephalexin and benzylpenicillin. Cells of E. coli KN126
were grown at 370 in Penassay broth for 15 min (A) with no addi-
tions, (B) with 10 pg/ml of cephalexin, and (C) with 12 pg/ml of
benzylpenicillin. Cells were then cooled on ice and purifiedcell en-
velopes were prepared without the use of mercaptoethanol (see
Methods). [14C]Benzylpenicillin (31 pg/ml) was added for 10 min
at 300, and the binding proteins were fractionated and detected as
in Fig. 1. Densitometer tracings were made from the x-ray film.
Binding proteins 5 and 6 are not shown. Aliquots of cells from (B)
and (C) were also grown further at 370 to ensure that cell division
was inhibited under these conditions.
Penicillin binding proteins in cells grown in the pres-
maintenance of cell shape (8). Since ["4C]mecilhinam is now
available to us, we have studied the binding of this com-
pound to E. coil directly. [14C]Mecillinam, at physiologically
effective concentrations (0.01-1.0 gg/ml), binds exclusively
to a protein in the E. coli inner membrane that has the same
mobility on NaDodS04/polyacrylamide slab gels as penicil-
lin binding protein 2. No binding of mecillinam to the outer
membrane was detected. Binding of ["4C]mecillinam was
50% saturated at 0.02Ag/mlfor 10 min at 300, which com-
pares excellently with the value obtained previously (8). At
higher concentrations of [14C]mecillinamvery slight binding
occurred to some of the other penicillin binding proteins.
Fig. lE and F shows the binding of [14C]mecillinam (2jig/
ml) to the E. coil inner membrane proteins.
Characterization of a mutant with an altered penicillin
binding protein 2
Mutants resistant to mecillinam (10ttg/ml)were isolated at
300 on Penassay broth plates after nitrosoguanidine mutage-
nesis, and their penicillin binding proteins were examined.
One of the five mutants examined (strain B6) failed to bind
either ["4C]mecillinam or [14C]benzylpenicillin to binding
protein 2 (Fig. 1). Strain B6 grows slowly as round cells at
300, in the absence of mecillinam, and fails to form colonies
at 420. It seems likely that strain B6 possesses a binding pro-
tein 2 that not only fails to bind mecillinam and benzylpeni-
cillin, but also functions poorly and results in the growth of
the strain as round cells. it is also possible that protein 2 is to-
tally absent in this strain.
Involvement of penicillin binding protein 3 in cell
When cells of strain KN126 are grown for 15 min at 370 in
Penassay broth in the presence of a (3-lactam that is a good
specific inhibitor of cell division, thefl-lactambinds mainly
to protein 3. Fig. 2 shows the binding proteins still accessible
to ["4C]benzylpenicillin after growth in the presence of unla-
beled cephalexin or benzylpenicillin (at concentrations that
just inhibit cell division). In both
[14C]benzylpenicillin to protein 3 is grealy reduced comn-
pared to untreated control cells, while the binding to pro-
, the binding of
Proc. Nat. Acad. Sci. USA 72(1975)
z > 60
V,) t 20
PG/ML xMIN AMPICILLIN
proteins. Washed cell envelopes of strain KN126 were preincubat-
ed with a range of ampicillin concentrations and the binding pro-
teins remaining accessible to benzylpenicillin detected by the addi-
tion of a saturating concentration of [14C]benzylpenicillin (31
gg/ml for 10 min at 300). The binding proteins were fractionated
and detected as described in Fig. 1, and the amount of each bind-
ing protein remaining accessible to [14C]benzylpenicillin was mea-
sured by densitometry of the x-ray film. 0, Protein 1; X, protein 2;
0, protein 3.
Competition of ampicillin for the penicillin binding
teins 1, 2, and 4 is little altered. Proteins 5 and 6 are also un-
affected (data not shown).
We have also implicated protein 3 in cell division by
studying the affinities of the penicillin binding proteins for
fl-lactamsthat are good inhibitors of cell division. Fig. 3
shows the binding of a saturating concentration of [14C]ben-
zylpenicillin to cell envelopes after prebinding a range of
ampicillin concentrations. In common with all fl-lactams
that are good specific inhibitors of division, ampicillin com-
petes more effectively for penicillin binding protein 3 than
for proteins 1 or 2.
Characterization of a mutant with an altered penicillin
binding protein 3
Since penicillin resistance can be acquired by a variety of
means unrelated to changes in penicillin binding proteins
(10), we adopted a selective technique (15) to isolate a mu-
tant with a thermolabile binding protein 3.
Cells were mutagenized with nitrosoguanine, and colonies
resistant to a low level of cephalexin (5 jug/ml) were isolated
at 300 on Penassay broth plates. From these colonies temper-
ature-sensitive cell division mutants were isolated. By using
this selection procedure it was hoped to isolate mutants that
had a protein 3 that was more resistant to cephalexin at 300
and was thermolabile at 420. Inactivation of protein 3 at 420
would result in the inhibition of cell division. Membranes
envelopes were prepared from cells of strain 6-30 grown in Penas-
say broth at 300. (A) [14C]Benzylpenicillin (31
for 10 min at 300. (B) [14C]Benzylpenicillin (31jg/ml)was added
for 2 min at 420. (C) Cell envelopes were incubated at 42° for 11
min and then [14C]benzylpenicillin (31 Ag/ml) was added for a fur-
ther 2 min at 420. Penicillin binding proteins were fractionated
and detected as described in Fig. 1. Densitometer tracings were
made from the x-ray film. Binding proteins 5 and 6 are not shown.
Penicillin binding proteins in strain 6-30. Washed cell
sg/ml) was added
pG/MLx MIN CEPHALORIDINE
Competition of cephaloridine for the penicillin binding
proteins. Washed cell envelopes of strain KN126 were preincubat-
ed with a range of cephaloridine concentrations, and the binding
proteins remaining accessible to benzylpenicillin were detected by
the addition of a saturating concentration of [14C]benzylpenicillin
(31 pg/ml for 10 min at 30°). The binding proteins were fractionat-
ed, detected, and quantitated as described in Fig. 3. 0, Protein 1;
X, protein 2; 0, protein 3.
were prepared from mutants grown in Penassay broth at
300, and the binding of [14C]benzylpenicillin was assayed at
300 and 420.
Fig. 4 shows the binding to strain 6-30 at 300 and 420. At
300 this strain binds less [14C]benzylpenicillin to protein 3
than does the parent strain [compare the ratio of peak 2 to
peak 3 in the mutant (Fig. 4A) with that of the parent (Figs.
1A and B and 2A)]. At 420 the ability to bind penicillin is
lost within 11 min in the mutant but not in the parent strain.
Since no other binding proteins are thermolabile in strain
6-30, this temperature-sensitive cell division mutant has a
thermolabile protein 3. Since our other experiments have
implicated protein 3 in cell division, the temperature-sensi-
tive cell division phenotype appears to be caused by a ther-
molabile penicillin binding protein 3.
Involvement of penicillin binding protein1in cell
The cell elongation probe, cephaloridine, is unusual in that it
is the onlyfl-lactamwe found that causes cell lysis at the
lowest effective concentrations.
Cephaloridine is also the onlyfl-lactamthat shows a high-
er affinity for binding protein 1 than for protein 2 or 3. Fig.
5 shows the binding of
[14C]benzylpenicillin to proteins 1, 2, and 3 after prebinding
a range of cephaloridine concentrations. We believe that
protein 1 is involved in cell elongation and that inhibition of
this protein by ,-lactamsresults in cell lysis.
Involvement of penicillin binding proteins 2 and 3 in
the formation of penicillin bulges
Since bulge formation is an alteration of cell shape, we con-
sidered the possibility that it was due to the inhibition of
protein 2. Mostfl-lactamsbind with low affinity to protein 2
such that mecillinam-like cell shape changes are not ob-
served. If af3-lactamshowed higher affinities for proteins 2
and 3 than for protein 1, cell shape and division should be
altered before lysis occurred.
Since inhibition of protein 2 (by mecillinam) results in
continuing cell division and the maintenance of normal rod
shape for about 50 min at 370, the effect of additionally in-
hibiting cell division (blocking protein 3) should be to pro-
duce a normal rod-shaped filament during the first 50 min.
After 50 min the blocking of protein 2 should result in the
production of the cell shape changes (bulges) within these
filaments. In support of this model, the appearance of bulges
by ampicillin can be mimicked by the simultaneous addition
a saturating concentration of
Proc. Nat. Acad. Sci. USA 72 (1975)
Effects of j3-lactam antibiotics on growth of E. coli predicted by their relative affinities
for penicillin binding proteins
Morphological effects produced by three arbitrary concentrations of
Relative affinities of
binding proteins 1,
2, and 3 for a ,-
j3-Lactam showing this
1 > 2or3
2 > 1 > 3
2 > 3 > 1
2 > 1 or 3
3 > 1 > 2
3 > 2 > 1
*1=cell elongation; 2=cell shape; 3=cell division.
Filaments with bulges
Filaments with bulges
of afl-lactamthat binds to protein 2 (mecillinam) and one
that binds preferentially to protein 3 (cephalexin).
As expected from this model, ampicillin had a high affini-
ty for protein 3, slightly less affinity for protein 2, and even
less affinity for protein 1. One other antibiotic (dihydroamp-
icillin) also produces bulges and shows a similar order of af-
finities for binding proteins 1, 2, and 3.
Table 1 summarizes a model by which the particular effects
of a fl-lactam antibiotic on the division, elongation, and
shape of E. coli are explained by the relative affinity of that
antibiotic for three penicillin binding proteins with specific
roles in these three processes.
The evidence for a penicillin binding protein involved in
the maintenance of the rod shape of E. coli is very strong.
Mecillinam is a powerful probe for this protein (protein 2)
since it has a specific action on cell shape over a very wide
concentration range (3). Studies with both [14C]mecillinam
and [14C]benzylpenicillin show that the mecillinam binds
exclusively to one inner membrane protein (molecular
weight 66,000). A mecillinam-resistant mutant (strain B6)
has been isolated that fails to bind [14C]mecillinam or
[14C]benzylpenicillin to this protein. The enzymatic function
of this protein seems to be seriously impaired since the mu-
tant grows slowly as round cells at 300 and fails to form colo-
nies at 420. A study of revertants of strain B6 will be re-
quired to understand the relationship between temperature-
sensitivity, mecillinam-resistance, cell shape, and the altered
The evidence for the involvement of distinct penicillin
binding proteins in cell division and cell elongation is less
strong than the evidence for a protein involved in cell shape,
since no fl-lactamwith absolute specificity for the former
proteins is known. A good correlation does exist between a
(3-lactam being a good inhibitor of cell division and having a
high affinity for protein 3 (molecular weight 60,000). Fur-
thermore, when cells were grown for 15 min in the presence
of a concentration of benzylpenicillin or cephalexin that just
inhibits division, the (-lactams were bound preferentially to
protein 3. Strong additional evidence for the involvement of
protein 3 in cell division comes from the isolation of a tem-
perature-sensitive, cell division mutant (strain 6-30) with a
thermolabile binding protein 3.
The mutants B6 and 6-30 are examples of mutants resis-
tant tof3-lactamsin which the resistance has been correlated
with an altered penicillin binding protein. Strain 6-30 is also
an example of a cell division mutant in which the biochemi-
cal basis of the lesion is in some way understood.
Penicillin binding protein 1 (molecular weight 91,000) is
considered to be involved in cell elongation. /3-Lactams that
cause inhibition of cell elongation and cell lysis at the lowest
effective concentrations show a higher affinity for protein 1
than for protein 2 or 3.
A possible explanation for bulge formation by some peni-
cillins is that it is due to the inhibition of protein 2 in addi-
tion to protein 3 (see Table 1). Most (3-lactams do not pro-
duce bulges in our strain, since protein 1 has a higher affini-
ty for mostf3-lactamsthan protein 2, such that lysis occurs
before any effects on cell shape are observed. The only two
fl-lactams that give typical penicillin bulges in our strain
(ampicillin and dihydroampicillin) are also the only twofl-
lactams that show higher affinities for proteins 2 and 3 than
for protein 1.
The precise role of the three essential penicillin binding
proteins is not clear. Inhibition of protein 1 stops cell elonga-
tion and results in cell lysis, presumably by the continued ac-
tion of peptidoglycan hydrolases in the absence of peptido-
glycan synthesis. Protein 1 is proposed to be required for all
peptidoglycan synthesis. It possibly acts as a transpeptidase
which introduces new precursors into the growing peptido-
glycan chains (16).
Peptidoglycan synthesis continues when protein 2 or 3 (or
both) are inhibited, and the peptidoglycan produced under
these conditions is not seriously defective, since ovoid cells
(produced by inhibiting protein 2), filaments (produced by
inhibiting protein 3), and cells with bulges (produced by in-
hibiting proteins 2 and 3) are osmotically stable.
Protein 3 may be required for the synthesis of peptidogly-
can for cross walls, but not for that for side walls. It is also
possible, however, that this protein acts transiently at the
end of a round of DNA replication to alter the direction of
peptidoglycan growth at the potential division site. Synthesis
of both cross walls and side walls would then be due to the
action of protein 1.
Inhibition of protein 2 does not appear to affect pre-exist-
ing cell wall growth sites, since the resulting shape change
t Note Added in Proof. A revertant of strain B6 has been obtained
which has a normal rod morphology, grows at 42°, and is sensitive
to mecillinam. This revertant has regained the ability to bind
["4C]mecillinam and [14C]benzylpenicillin to penicillin binding
Proc. Nat. Acad. Sci. USA 72(1975)
does not occur for about 50 minutes in an exponential cul-
ture (5), and never occurs if DNA synthesis is blocked (17).
It is interesting to note that bulge formation also required
continuing DNA synthesis (2, 18), and only occurs after a
similar time lag (2). One possibility is that protein 2 acts
transiently in the cell cycle to ensure that elongation at
newly introduced growth sites occurs in the correct rod con-
figuration (James, manuscript in preparation).
I would especially like to thank Dr. Frantz Lund for the generous
gift of [14C]mecillinam and Dr. Arthur B. Pardee for support and
encouragement. I would also like to thank Drs. Jack L. Strominger
and Austin Newton for reading the manuscript and Ms. Mary
Cupples for providing clean glassware. Supported by U.S. Public
Health Service Grant CA11595 to Dr. A. B. Pardee.
Lederberg, J. (1957) J. Bacteriol. 73, 144.
Schwarz, U., Asmus, A. & Frank, H. (1960) J. Mol. Biol. 41,
Lund, F. & Tybring, L. (1972) Nature New Biol. 238, 135-
Melchior, N. H., Blom, J., Tybring, L. & Birch-Anderson, A.
(1973) Acta Pathol. Microblol. Scand. Sec. B 81, 393-407.
James, R., Haga, J. Y. & Pardee, A. B. (1975) J. Bacteriol. 122,
Strominger, J. L., Blumberg, P. M., Suginaka, H., Umbreit, J.
& Wickus, G. G. (1971) Proc. Roy. Soc. Ser. B. 179,369-383.
Nguyen-Disteche, M., Pollock, J. J., Ghuysen, J-M., Puig, J.,
Reynolds, P., Perkins, H. R., Coyette, J. & Salton, M. R. J.
(1974) Eur. J. Biochem. 41, 457-463.
Spratt, B. G. & Pardee, A. B. (1975) Nature 254,516-517.
Strominger, J. L., Willoughby, E., Kamiryo, T., Blumberg, P.
M. & Yocum, R. R. (1974) Ann. N.Y. Acad. Sd. 235,210-244.
Blumberg, P. M. & Strominger, J. L. (1974) Bacteriol. Rev. 38,
Suginaka, H., Blumberg, P. M. & Strominger, J. L. (1972) J.
Biol. Chem. 247,5279-5288.
Inouye, M. & Guthrie, J. P. (1969) Proc. Nat. Acad. Sci. USA
Filip, C., Fletcher, G., Wulff, J. L. & Earhart, C. F. (1973) J.
Bonner, W. M. & Laskey, R. A. (1974) Eur. J. Biochem. 46,
Kamiryo, T. & Strominger, J. L. (1974) J. Bacteriol. 117,
Mirelman, D., Bracha, R. & Sharon, N. (1974) Ann. N.Y.
Acad. Sci. 235,326-4344.
Matsuhashi, S., Kamiryo, T., Blumberg, P. M., Linnett, P.,
Willoughby, E. & Strominger, J. L. (1974) J. Bacteriol. 117,
Burdett, I. D. J. & Murray, R. G. E. (1974) J. Bacteriol. 119,