Eur. J . Biochem. 72, 341-352 (1977)
Properties of the Penicillin-Binding Proteins of Escherichia coli K 12 Brian G. SPRATT Department of Genetics, University of Leicester (Received September 14 / October 30, 1976)
B e n ~ y l [ '~ Cpenicillin ] binds to six proteins with molecular weights of between 40000 and 91 000 in the inner membrane of Escherichia coli. Two additional binding proteins with molecular weights of 29000 and 32000 were sometimes detected. All proteins were accessible to ben~yl['~C]penicillin in whole cells. Proteins 5 and 6 released bound ben~yI['~C]penicillin with half times of 5 and 19 min at 30 'C but the other binding proteins showed less than 50% release during a 60-min period at 30 "C. The rate of release of bound penicillin from some of the proteins was greatly stimulated by 2-mei-captoethanol and neutral hydroxylamine. Release of ben~yI['~C]penicillin did not occur ~ tthe ' binding proteins were denatured in anionic detergent and so was probably enzymic. No additional binding proteins were detected with two ['4C]cephalosporins. These p-lactams bound to either all or some of those proteins to which ben~yl['~C]penicillinbound. No binding proteins have been detected in the outer membrane of E. coli with any p-['4C]lactam. The binding of a range of unlabelled penicillins and cephalosporins were studied by measuring their competition for the binding of berl~yl['~C]penicillinto the six penicillin-binding proteins. These results, together with those obtained by direct binding experiments with P-[14C]lactams,showed that penicillins bind to all six proteins but that at least some cephalosporins fail to bind, or bind very slowly, to proteins 2, 5 and 6, although they bind to the other proteins. Since these cephalosporins inhibited cell division and caused cell lysis at concentrations where we could detect no binding to proteins 2, 5 and 6, we believe that these latter proteins are not the targets at which S-lactams bind to elicit the above physiological responses. The binding properties of proteins 1, 3 and 4 correlate reasonably well with those expected for the above killing targets.
The p-lactam group of antibiotics (penicillins and cephalosporins) kill bacteria by inhibiting the terminal stages of peptidoglycan metabolism [l - 31. The introduction of nascent precursor chains into the growing peptidoglycan appears to be the primary target of their action [4,5]. In recent years two complementary approaches have been taken to study the interaction of these antibiotics with bacterial cells. The first of these involves the identification and purification of enzymes which are inhibited by P-lactams. In Escherichia coli three enzymic reactions have been shown to be penicillin-sensitive : D-alanine carboxypeptidase 1, peptidoglycan trdnspeptidase and peptidoglycan endopeptidase [6- lo]. In all bacteria that have been studied in any detail multiple penicillin-sensitive enzymes have been identified and in several cases purified enzymes have been shown to catalyse more than one of the above reactions [6,9, lo].
The second of these approaches depends on the old observation that penicillin irreversibly binds to bacterial membranes [I I ] and the recognition that this reflects the irreversible binding of ,8-lactams to the enzymes which they inhibit [2,6]. This implies that penicillin-sensitive enzymes and penicillin-binding proteins are synonymous, and that penicillin-sensitive enzymes can be detected as those proteins which bind penicillin. Multiple penicillin-binding proteins have been detected in the membranes of all bacteria that have been examined [6,12,13], and it is now clear that the interaction of p-lactams with bacterial cells is considerably more complex than once was thought. The physiological and morphological response of E.coli to P-lactam antibiotics is also complex (e.g p-lactams can inhibit cell division, cause cell lysis, bulge formation or the formation of osmotically-stable ovoid cells) and we have recently been identifying the physjological targets to which P-lactams bind to
342
elicit the above responses [13,14]. For this purpose we have developed a method for detecting penicillinbinding proteins on dodecylsulphate-polyacrylamide slab gels and in this paper describe the properties of the E. coli penicillin-binding proteins.
MATERIALS AND METHODS Organisms E. coli strain KN126 was used in all experiments. It has the genotype F- trpE9829 tyr- ilv- sup-126 and was obtained from Dr T. Nagata.
Growth of Cells and Preparation of Membranes Cells were routinely grown in 3-1 batch cultures in pennassay broth (Difco Antibiotic Medium No. 3) at 37 "C with vigorous aeration and were harvested in mid to late exponential growth phase. After cooling on ice, the cells were centrifuged at 6000 x g for 6 rnin and resuspended in 80 ml of ice-cold 10 mM sodium phosphate buffer (pH 7.0). 2-Mercaptoethanol was added (final concentration of 0.14 M) and the cells were broken open by three 30-s pulses of sonication with intervening 30-s periods of cooling. Unbroken cells were removed by centrifugation at 8000 x g for 20 min at 4 "C in the SS-34 rotor of a Sorvall RC2B centrifuge, and the cell membranes were pelleted out of the supernatant by centrifugation at 100000 x g for 40 rnin at 4 "C in the 50Ti rotor of a Beckman ultracentrifuge. The membranes were resuspended in 20 ml of ice-cold phosphate buffer (as above), washed twice by centrifugation at 100000 x g, and finally stored in the above buffer at a concentration of approximately 40 mg protein/ml in a Revco -80 "C freezer. These washed cell membranes consist of inner (cytoplasmic) membrane, outer membrane and peptidoglycan but will be referred to as membranes for brevity.
Binding of Benzyl[ ''C]penicillin to E. coli Membranes Membranes were thawed slowly and adjusted to 5 - 10 mg protein/ml and to 50 mM sodium phosphate (buffer pH 7.0). MgC12, which had been included in the assay mixture in earlier experiments [13], was omitted as it was found to be inessential for binding. For the standard assay of penicillin-binding proteins 20 pl of ben~yl['~C]penicillinwas bound to 200 pl of membranes for 10 rnin at 30 "C and the reaction was terminated by the addition of 5 pl of non-radioactive benzylpenicillin (120 mg/ml) and 10 pl of 20% (w/v) sodium lauroyl sarcosinate (Sarkosyl NL-97). The addition of the anionic detergent sarkosyl has a dual
Penicillin-Binding Proteins of Escherichia coli
purpose. Firstly it denatures the penicillin-binding proteins and prevents the enzymic loss of bound penicillin (see Results), and secondly, it selectively and completely solubilises the proteins of the inner membrane whilst leaving the outer membrane insoluble 1151. After solubilisation of the inner membrane for 20 min at room temperature, the sarkosyl-insoluble outer membrane and peptidoglycan were removed by centrifugation at 100000 x g for 40 min at 10 "C in the 50Ti rotor of a Beckman ultracentrifuge using 10 ml Oak-Ridge-type polycarbonate screw-cap tubes. In recent experiments the removal of the sarkosylinsoluble fraction by centrifugation at 40000 x g in the SM24 rotor of a Sorvall RC2B centrifuge has been used with identical results. 100 p1 of the sarkosylsoluble supernatant were added to 50 p1 of gel sample buffer (0.2 M Tris-HC1, pH 6.8 ; 3% w/v sodium dodecylsulphate; 30'x vjv glycerol; 0.002 bromophenol blue). To study binding to the outer membrane, the sarkosyl-insoluble fraction from above was re-extracted with 2 ml of 1% sarkosyl for 5 min at room temperature, pelleted by centrifugation at 100000 x g for 40 rnin at 10 "C, and resuspended in 200 pl of distilled water. 100 pl was added to 50 p1 of gel sample buffer. All samples were then fractionated by dodecylsulphate-polyacrylamide slab gel electrophoresis as described below. Experiments with other P-['4C]lactams were performed in a similar manner.
Slab Gel Electrophoresis and Detection of Penicillin-Binding Proteins 20 pl of 2-mercaptoethanol were added to each of the above samples immediately before they were heated for 3 rnin in a boiling water bath. 50-65 p1 of each sample (containing about 200 pg protein) were loaded into the gel slots of a 10% sodium dodecylsulphate polyacrylamide slab gel. The apparatus and the discontinuous buffer system have been described in detail by Laemmli and Favre [16]. Electrophoresis was performed at a constant current of 20 mA (initial voltage of about 60 V increasing to about 200 V at the end of electrophoresis) until the bromophenol blue tracking dye had migrated 11 cm into the separating gel. The gel was removed from the apparatus, stained for 1 h with 0.1% Coomassie brilliant blue R (Sigma) in 50% methanol 10% acetic acid, diffusiondestained with several changes of 5% methanol 10% acetic acid, and prepared for fluorography (scintillation autoradiography) by the incorporation of the scintillant PPO into the gel matrix as described by Bonner and Laskey [17]. Staining with Coomassie brilliant blue was used only to monitor the quality of
B. G. Spratt
the protein fractionation on the gel since the stained bands were not visible clearly after fluorography. The PPO-impregnated gel was dried under vacuum on to Whatman number 17 filter paper and placed on Kodak RP Royal X-Omat X-ray film at -80 "C. X-ray films were developed in Kodak liquid X-ray developer for 5 rnin at 19 "C without agitation, washed in 3% acetic acid and fixed for 5 min at 19 "C in Kodak X-ray fixer. Conventional autoradiography (without incorporation of PPO) was used in one experiment and in this case Ilford industrial G X-ray film was used. The use of fluorography is essential to obtain the maximal sensitivity of 14C detection on the X-ray film and even with this technique the exposure times required are still inconveniently long. The binding proteins can be detected by eye after about 10 days fluorography but a considerably longer exposure is required to obtain image densities sufficient for accurate densitometry of the minor proteins. Fluorography probably introduces some errors into the quantitisation of the results since the image density on the X-ray films is not linear with the amount of radioactivity applied [18]. This problem can be eliminated by pre-fogging of the X-ray films [18]. A further substantial increase in sensitivity is obtained by the removal of the outer membrane proteins before applying samples to the gels. This is possible as we have failed to detect any penicillinbinding proteins in the E.coli outer membrane. Although the outer membrane constitutes only about 50% of the total protein of the membranes, the majority of the outer membrane consists of three proteins with almost identical mobilities on dodecylsulphate-polyacrylamide gels [ 191 and this results in overloading when relatively small amounts of outer membrane protein are applied to the gels. In contrast, the inner membrane consists of many proteins in approximately similar amounts, with the result that much larger amounts of protein can be applied before distortion of the gels due to overloading occurs. Extraction of the inner membrane is complete under the conditions that we use, however, an increase in the concentration of sarkosyl to at least 2% can be used without interfering with the running of the gels or significantly extracting outer membrane proteins. Apparent molecular weights of the penicillinbinding proteins were measured by comparison of their mobilities on dodecylsulphate-polyacrylamide slab gels with that of eight proteins of known molecular weights. The proteins and their molecular weights were : P-galactosidase (130000), phosphorylase A (lOOOOO), bovine serum albumin (68 000), pyruvate kinase (57 OW), ovalbumin (43 000), glyceraldehyde-3phosphate dehydrogenase (36000), carbonic anhydrase (29000), and chymotrypsinogen (25 700).
343
Measurement of the Binding of P-Lactams by Competition with Benzyl[14C]penicillin
The binding of non-radioactive P-lactams was studied by measuring their competition for the binding of ben~yl['~C]penicillin to each of the proteins. 200 pl of membranes (5 - 10 mg protein/ml in 50 mM sodium phosphate buffer pH 7.0) were pre-incubated with 10 p1 of either distilled water or dilutions of the non-radioactive p-lactam for 10 rnin at 30 "C and then 20 pl of 50 pCi/ml ben~yl[~~C]penicillin were added for a further 10 min at 30 "C and the binding terminated by the addition of 5 p1 of non-radioactive benzylpenicillin (120 mg/ml) and 10 p1 of 20% (w/v) sarkosyl as described above. The final concentration of ben~yl['~C]penicillin (30 pg/ml) was approximately three times that required to obtain saturation of all binding proteins under the conditions used. Samples were fractionated by electrophoresis and the level of ben~yl['~C]penicillin bound to each protein measured as described above.
Release of Bound Benzyl[ ''C]penicilEin
To measure the spontaneous rate of release of the product of bound ben~yl['~C]penicillin(which is not benzylpenicillin itself but probably either benzylpenicilloic acid [9] or phenacetylglycine and thiazolidine ring fragmentation products [20, 21]), membranes (5 - 10 mg protein/ml in 50 mM sodium phosphate buffer pH 7.0) were labelled with benzyl[14C] penicillin (16.5 pg/ml final concentration) for 10 rnin at 30 "C; a 1000-fold excess of non-radioactive benzylpenicillin was added, and samples of 200 p1 removed immediately and at intervals during continued incubation at 30 "C into tubes containing 10 p1 of 20% (w/v) sarkosyl. Denaturation of the penicillinbinding proteins in sarkosyl prevented further release of ben~yl['~C]penicillin since release is enzymic (see Results). The samples were fractionated by electrophoresis and fluorographs prepared as described above. The spontaneous release of other fi-['"C] lactams was measured in a similar fashion. Various compounds (e.g. 2-mercaptoethanol, neutral hydroxylamine) were tested for their effect on the spontaneous rate of ben~yl['~C]penicillin release by adding them immediately after the excess non-radioactive benzylpenicillin and sampling as above into sarkosyl. In some experiments where the excess benzylpenicillin was known, or suspected, to react with the test compound, the further binding of ben~yl['~C]penicillinwas prevented by the addition of 0.1 vol. of penicillinase (100000 units/ml) for 5 min at 30 "C. Penicillinase does not remove bound penicillin.
Penicillin-Binding Proteins of Escherichia coli
344
Quantit isation of Penicillin-Binding Proteins by Densitometry of X-Ray Films X-ray films of sufficient image density for accurate quantitisation (but not so dense as to approach saturation of the film emulsion) were scanned with a Joyce-Loebl M k l 1lc microdensitometer. The peaks obtained were carefully cut out and weighed. Corrections were made, when necessary, for differences in widths of the gel slots.
Protein 1
2
3
Measurement of Minimal inhibitory Concentrations of p- Lac tams
5
6
'
Minimdl inhibitory concentrations weredetermined using E. coli KN126 growing exponentially at 37 "C in penassay broth with initial cell densities of approximately 2 x lo7 organisms/ml. The minimal inhibitory concentration was defined as the lowest concentration of the p-lactam which produced a clear alteration of the normal rod morphology of the cells (depending on the 8-lactam this could be the appearance of cell lysis, filamentation or ovoid cells).
Chemicals Ben~yI['~C]penicillinwas purchased from the Radiochemical Centre, Amersham, England (specific activity 53 - 54 Ci/mol). ['4C]Cefoxitin (2.8 Ci/mol) and ['4C]cephacetrile (4.6 Ci/nioI) were generously provided by Dr Edward Stapley (Merck Sharp and Dohme, Rahway, New Jersey, U.S.A.) and D r W. Zimmermann (Ciba-Geigy, Basel, Switzerland) respectively. All three radioactive p-lactams were of high purity and were labelled in the R, sidechain (position 6 of penicillins ;position 7 of cephalosporins). Reference samples of 8-lactams were kindly provided by the following companies: ampicillin by Bristol Laboratories (Syracuse, New York, U.S.A.), cephacetrile and cephradine by Ciba-Geigy Ltd (Basel, Switzerland), mecillinam by Leo Laboratories, (Ballerup, Denmark), cephalexin, cephaloridine and cephalothin by Eli Lilly andCo. (Indianapolis, Indiana, U.S.A.), cefoxitin by Merck, Sharp and Dohme (Rahway, New Jersey, U.S.A.), benzylpenicillin, cephalosporin G and penicillin V by Squibb and Sons (Princeton, New Jersey, U.S.A.). /j-['"C]Lactams were stored as solids at -2O';C and were dissolved in 10 mM sodium phosphate buffer pH 7.0 immediately before use. Aqueous solutions were stored at - 80 "C and could be used for at least a week without the appearance of non-specific binding problems due to degradation products of the fl-lactams. Solutions of nonradioactive p-lactams were always prepared fresh immediately before each experiment. The solid compounds were stored desiccated at - 20 "C.
Fig. 1 . Detection uf rhe E. coli penicillin-binding proteins utzd their location in the inner memhrune. 40 p1 of ben~yl['~C]penicillin (31 pg/ml final concentration) were added to 400 pl of membranes for 10 min at 30 "C. To measure penicillin-binding proteins in the total membrane 100 pI of the assay mixture was added to 50 pl of gel sample buffer. A further 200 p1 were added to 10 pl of 20"/> (w/v) sarkosyl and the sarkosyl-soluble (Inner membrane) and sarkosyl-insoluble (outer membrane) fractions were obtained as described in Methods. All three samples were fractionated on a 10% polyacrylamide slab gel and a fluorograph prepared after exposure of the gel against Kodak RP Royal X-Omat X-ray film for 32 days at -80 'C. (A) Total membrane; (B) outer membrane; (C) inner membrane
Penicillinase was obtained from Wellcome Reagents Ltd (Beckenham, England). Sarkosyl NL-97 was purchased from Geigy Industria1 Chemicals (New York, U.S.A.).
RESULTS All the penicillin-binding proteins are located in the E.coli inner membrane, defined as that fraction of the cell envelope which is solubilised by lo( sarkosyl at room temperature. Fig.1 shows the binding of ben~yl['~C]peniciIlin to total cell envelopes and to the sarkosyl-soluble and sarkosyl-insoluble fractions. No penicillin-binding proteins have been detected in the sarkosyl-insoluble (outer membrane) fraction with any of the P-['4C]lactams that we have studied. In all further experiments only the sarkosyl-soluble (inner membrane) fraction was analysed for penicillinbinding proteins. All the proteins are accessible to ben~yI[~~C]penicillin in whole cells since they could be detected in membranes prepared from whole cells which had been incubated with ben~yl['~C]penicillin.
345
B. G. Spratt
In this experiment mercaptoethanol was omitted during sonication as this compound releases bound penicillin (see below). Since several of the chemicals which are used in dodecylsulphate-polyacrylamide gel electrophoresis (e.g . ammonium persulphate, 2-mercaptoethanol) have been reported to release bound penicillin [12], we have re-investigated the effects of these compounds on the recovery of the penicillin-binding proteins. 2-Mercaptoethanol released bound penicillin rapidly from proteins 5 and 6 if it was added to the binding assay mixture before the addition of sarkosyl (see below). However this release was clearly enzymic since it was completely prevented by denaturation of the proteins in 1% sarkosyl. Once denatured, we could detect no differences in the levels of the penicillinbinding proteins in samples which had been boiled in gel sample buffer in the presence or absence of 10% (v/v) 2-mercaptoethanol (data not shown). Similarly we could find no effect of ammonium persulphate (0.1%) on the release of bound penicillin once the proteins had been denatured in sarkosyl. In earlier experiments [13] we included 10 mM MgC1, in the penicillin-binding protein assay mixture but since we find no stimulation of binding by Mg2+, or any inhibition of binding by 1 mM EDTA, we now omit MgCl, from the assay. Pre-incubation of samples of membrane with N-ethylmaleimide (5 mM) or p hydroxymercuribenzoate (1.5 mM) for 10 min at 30 "C did not prevent binding to any of the proteins. The pattern of six proteins is extremely reproducible. In some gels additional minor penicillinbinding proteins were observed. These bands were in general non-reproducible, although two bands, proteins 7 and 8, with molecular weights 29000 and 32000 were apparent in several gels. An identical pattern of penicillin-binding proteins was found in all E.coli K12 strains that we have examined. E.coli B/r strains also showed an identical pattern although protein 7 was more prominent. The apparent molecular weights of the proteins were determined by comparing their mobility on a dodecylsulphate-polyacrylamide slab gel with that of eight proteins of known molecular weights. Table 1 shows the apparent molecular weights, and the relative abundance of each penicillin-binding protein, as well as an estimate of the number of molecules of each protein per bacterium (see below). The penicillin-binding proteins are minor components of the E.coli inner membrane. Fig.2 shows the location of these proteins on a Coomassie-bluestained gel of the inner membrane (note the absence of the characteristic major outer membrane proteins [19] with molecular weights 36000 on this gel of the sarkosyl-soluble membrane). The major proteins (5 and 6) can be tentatively identified as minor stained bands but the others are below the resolution of detection as stained bands.
Table 1. MoJecufar weights and relative abundance of the penicillinbinding proteins Protein
Apparent molecular weight
Binding of penicillin
Molecules/cell"
%total
91 OOO 66 000 60000 49000 42000 40000
1 2 3 4
5 6 a
8.1 0.7 1.9 4.0 64.7 20.6
230 20 50 110 1800 570
See Results for details
Kinetics of Binding of p-[ 14C]Lactams
Benzyl[14C]penicillin was added in 2-fold increasing concentrations (final concentrations of 0.015 -31.3 pg/ml) to aliquots of membranes for 10 min at 30 "C and the extent of binding to each of the proteins measured as described in Methods. The binding of penicillin to proteins 1-4 was treated as a simple bimolecular irreversible reaction. For such a reaction the degree of acylation by penicillin should be proportional to the product of the penicillin concentration and the time of incubation [22]. Fig.3 shows that proteins 1- 3 bind penicillin in accord with the above kinetics, but protein 4 shows more complex kinetics. Proteins 5 and 6 were not plotted in this way as they released bound penicillin with half lives which are of the same order as the incubation time of the assay (see below). Similar kinetics of ben~yl['~C]penicillinbinding were obtained when a fixed concentration of penicillin was bound for increasing incubation times. Saturation of all penicillin-binding proteins was essentially complete at 80 pg m l - ' min. The concentration of ben~yl['~C]penicillinin our standard binding assay was over 3.5 times this level (313 pg ml-' min) to ensure saturation of all penicillinbinding proteins. Although experiments with ['4C]cefoxitin and ['4C]cephacetrile were undertaken (see below) the specific activity of these materials was too low to obtain images on the X-ray films of sufficient density for accurate quantitisation. It was, however, observed that [14C]cephacetrilebound to the same six proteins as ben~yl['~C]penicillin and that ['4C]cefoxitin bound to all except protein 2. No additional penicillinbinding proteins were observed with any p-['"C] lactams. Reversihility of the Binding of p-[ ''C']Lnctan~.s The penicillin-binding proteins were saturated with benzyl['"C]penicillin, excess non-radioactive
Penicillin-Binding Proteins of Escherichia coli
346
M,
1
130-
Protein 100-
-1
-2 57.--c
--3 Benzylpenicillin Ipg rn-lrnin)
L3
36
29
26
-
4-5 4-6
--
Fig. 2. Location of the penicillin-binding proteins on a Coomassieblue-stained gel of the inner membrane. Ben~yl['~C]penicillin (31 pg/ml final concentration) was bound t o membranes for 10 min at 30 'C and the sarkosyl-soluble inner membrane proteins were fractionated on a 10% polyacrylamide slab gel. Eight proteins of known molecular weights were run in a neighbouring gel slot. The Coomassie-blue-stained gel was autoradiographed on Ilford industrial G X-ray film for 90 days and the positions of the penicillinbinding proteins and the molecular weight standards marked o n a photograph of the stained gel
Fig. 3. The kinetics of binding of benzyl[ ''C]penicillin. Ben~yl['~C] penicillin was bound to aliquots of membranes for 10 min at 30 "C at ?-fold increasing concentrations from 0.015- 31.3 pg in1 (w. from 0.15 - 31 3 pg m1-l min). The binding was terminated by the addition of excess non-radioactive benzylpenicillin, the sarkosylsoluble proteins fractionated on a lox, polyacrylamide slab gel, and a fluorograph prepared. The extent of binding to each of the penicillin-binding proteins 1, 2, 3 and 4 was measured by microdensitometry of the X-ray film. P is the saturation level of binding and P, is the level of binding at penicillin concentration .Y (pg ml-' min)
benzylpenicillin added, and the release of bound ben~yl[~~C]penicillin from each of the proteins measured during continued incubation at 30 "C. Fig.4 shows the rapid release of penicillin from proteins 5 and 6 but only very slight release from the other proteins. Half times of release for proteins 5 and 6 were 5 and 19 min respectively at 30 "C. Accurate half times of release for proteins 1-4 could not be obtained because of the scatter in the data points, but at least 50% of the benzyl['4C]penicillin remained bound to each of these proteins during 1 h at 30 "C. [14C]Cefoxitin was not significantly released from proteins 1,3,5 or 6 during a 45-min period at 30 "C, but it was released from protein 4 with a half time of approximately 15 min (cefoxitin does not bind to protein 2). The effect of various compounds on the rate of release of bound benzylpenicillin were studied. NO difference in the rate of release from any of the proteins was detected whether the ben~yl['~C]penicillin was diluted with excess non-radioactive benzylpenicillin or removed with penicillinase, indicating that
341
(final concentration 0.05 M) also stimulated release from proteins 5 and 6 to a slight extent. Neutral hydroxjlamine (final concentration 0.2 M) stimulated release from all proteins with the exception of protein 4. It is possible that higher concentrations of hydroxylamine would effect release from protein 4 but interference of the electrophoresis by higher concentrations of the compound prevented this from being studied. Table 2 summarises the results obtained on the release of ben~yl[~~C]penicillin.
benzylpenicillin itself (15 mg/ml) did not stimulate release. The rate of benzylpenicillin release from proteins 5 and 6 was markedly enhanced by 2mercaptoethanol (final concentration 0.14 M), whereas proteins 1-4, which do not spontaneously release benzypenicillin at an appreciable rate, are not stimulated to d o so by 2-mercaptoethanol. Dithiothreitol
Protein
Competition of’/3-Lactams,forBenzyl[ ‘‘C]penicillin Binding
0
------a&.
50
Since most B-lactams were not available radioactively labelled, we have studied their binding by the ability to compete with ben~yl[’~C]penicillin. Such experiments are not entirely satisfactory as competition by a 8-lactam depends both on its affinity for the binding proteins and on its rate of release. Clearly a p-lactam will show neglible competition if it has the same affinity for a penicillinbinding protein as ben~yl[‘~C]penicillin but is released 100 times as fast. Fig.5 shows the competition of three concentrations of cephalothin, cephalosporin G and penicillin V for the penicillin-binding proteins. Fig. 6 shows the result of a more complete competition experiment with cephradine and Table 3 gives the minimal inhibitory concentrations of a selection of 8-lactams and the concentrations required to reduce ben~yl[’~C]penicilIin binding to each protein by 50% under the conditions of the assay. Although penicillins showed competition with all proteins, several cephalosporins show competition for proteins 1,3 and 4 but little or none for proteins 2,5 and 6. This failure to obtain competition could either be due to their failure to bind to the latter proteins or due to their rapid rate of release. We have attempted to distinguish between these possibilities
$ 100 1
--0
0
0
0 -
0
V
x
0
\
Time(rnin1 Fig. 4. Releusr of hound henzyl[ ‘“C]penici/lin.0.2 ml of 50 pCi/ml benzyl[14C]penicill~nwas added t o 4 ml of membranes for 10 min at 30 “C (final penicillin concentration 16.5 pg/ml). 60 mg of nonradioactive benzylpenicillin were added and 0.2-ml samples were removed immediately and at intervals during continued incubation at 30 “C. 10 pl of 20% (w/v) sarkosyl were added to each sample to prevent further release of bound benzyl[14C]penicillin and the sarkosyl-soluble proteinswerefractionated on a 10% polyacrylamide slab gel and a fluorograph prepared. The level of b e n ~ y l [ ’ ~ C ] penicillin remaining bound to each penicillin-binding protein was measured by microdensitometry of the X-ray film
Table 2. Relrme ofhenzyl[ ’‘C]penicillin ,from the pmic~illiii-hiiitliii,~ prorc’iii.r The rate of release of bound ben~yl[’~C]penicillin from each penicillin-binding protein was measured at 30 ’ C in 0.05 M sodium phosphate buffer p H 7.0 as described in Methods. It was measured either after addition of excess unlabelled benzylpenicillin (A) o r after addition of penicillinase (B) Additions t o basic assay mixture
Measured
Time at 30 ‘C for 50% release of bound ben~yl[’~C]penicillin ~~
No addition No addition 2-Mercaptoethanol(O.14 M) Dithiothreitol(O.05 M ) Dithiothreitol(O.05 M) Neutral hydroxylamine (0.2 M )
A B A A
B B
6
1
2
3
4
> 60
> 60 > 60 > 30 > 30 > 30 18
> 60 > 60 > 30 > 30 > 30 8
> 60
5
> 60 > 30 > 30 > 30 > 30
5 30 > 30 > 30 22
5
19 20 2 12 7
Properties of the Penicillin-Binding Proteins of Escherichia coli K 12 Brian G. SPRATT Department of Genetics, University of Leicester (Received September 14 / October 30, 1976)
B e n ~ y l [ '~ Cpenicillin ] binds to six proteins with molecular weights of between 40000 and 91 000 in the inner membrane of Escherichia coli. Two additional binding proteins with molecular weights of 29000 and 32000 were sometimes detected. All proteins were accessible to ben~yl['~C]penicillin in whole cells. Proteins 5 and 6 released bound ben~yI['~C]penicillin with half times of 5 and 19 min at 30 'C but the other binding proteins showed less than 50% release during a 60-min period at 30 "C. The rate of release of bound penicillin from some of the proteins was greatly stimulated by 2-mei-captoethanol and neutral hydroxylamine. Release of ben~yI['~C]penicillin did not occur ~ tthe ' binding proteins were denatured in anionic detergent and so was probably enzymic. No additional binding proteins were detected with two ['4C]cephalosporins. These p-lactams bound to either all or some of those proteins to which ben~yl['~C]penicillinbound. No binding proteins have been detected in the outer membrane of E. coli with any p-['4C]lactam. The binding of a range of unlabelled penicillins and cephalosporins were studied by measuring their competition for the binding of berl~yl['~C]penicillinto the six penicillin-binding proteins. These results, together with those obtained by direct binding experiments with P-[14C]lactams,showed that penicillins bind to all six proteins but that at least some cephalosporins fail to bind, or bind very slowly, to proteins 2, 5 and 6, although they bind to the other proteins. Since these cephalosporins inhibited cell division and caused cell lysis at concentrations where we could detect no binding to proteins 2, 5 and 6, we believe that these latter proteins are not the targets at which S-lactams bind to elicit the above physiological responses. The binding properties of proteins 1, 3 and 4 correlate reasonably well with those expected for the above killing targets.
The p-lactam group of antibiotics (penicillins and cephalosporins) kill bacteria by inhibiting the terminal stages of peptidoglycan metabolism [l - 31. The introduction of nascent precursor chains into the growing peptidoglycan appears to be the primary target of their action [4,5]. In recent years two complementary approaches have been taken to study the interaction of these antibiotics with bacterial cells. The first of these involves the identification and purification of enzymes which are inhibited by P-lactams. In Escherichia coli three enzymic reactions have been shown to be penicillin-sensitive : D-alanine carboxypeptidase 1, peptidoglycan trdnspeptidase and peptidoglycan endopeptidase [6- lo]. In all bacteria that have been studied in any detail multiple penicillin-sensitive enzymes have been identified and in several cases purified enzymes have been shown to catalyse more than one of the above reactions [6,9, lo].
The second of these approaches depends on the old observation that penicillin irreversibly binds to bacterial membranes [I I ] and the recognition that this reflects the irreversible binding of ,8-lactams to the enzymes which they inhibit [2,6]. This implies that penicillin-sensitive enzymes and penicillin-binding proteins are synonymous, and that penicillin-sensitive enzymes can be detected as those proteins which bind penicillin. Multiple penicillin-binding proteins have been detected in the membranes of all bacteria that have been examined [6,12,13], and it is now clear that the interaction of p-lactams with bacterial cells is considerably more complex than once was thought. The physiological and morphological response of E.coli to P-lactam antibiotics is also complex (e.g p-lactams can inhibit cell division, cause cell lysis, bulge formation or the formation of osmotically-stable ovoid cells) and we have recently been identifying the physjological targets to which P-lactams bind to
342
elicit the above responses [13,14]. For this purpose we have developed a method for detecting penicillinbinding proteins on dodecylsulphate-polyacrylamide slab gels and in this paper describe the properties of the E. coli penicillin-binding proteins.
MATERIALS AND METHODS Organisms E. coli strain KN126 was used in all experiments. It has the genotype F- trpE9829 tyr- ilv- sup-126 and was obtained from Dr T. Nagata.
Growth of Cells and Preparation of Membranes Cells were routinely grown in 3-1 batch cultures in pennassay broth (Difco Antibiotic Medium No. 3) at 37 "C with vigorous aeration and were harvested in mid to late exponential growth phase. After cooling on ice, the cells were centrifuged at 6000 x g for 6 rnin and resuspended in 80 ml of ice-cold 10 mM sodium phosphate buffer (pH 7.0). 2-Mercaptoethanol was added (final concentration of 0.14 M) and the cells were broken open by three 30-s pulses of sonication with intervening 30-s periods of cooling. Unbroken cells were removed by centrifugation at 8000 x g for 20 min at 4 "C in the SS-34 rotor of a Sorvall RC2B centrifuge, and the cell membranes were pelleted out of the supernatant by centrifugation at 100000 x g for 40 rnin at 4 "C in the 50Ti rotor of a Beckman ultracentrifuge. The membranes were resuspended in 20 ml of ice-cold phosphate buffer (as above), washed twice by centrifugation at 100000 x g, and finally stored in the above buffer at a concentration of approximately 40 mg protein/ml in a Revco -80 "C freezer. These washed cell membranes consist of inner (cytoplasmic) membrane, outer membrane and peptidoglycan but will be referred to as membranes for brevity.
Binding of Benzyl[ ''C]penicillin to E. coli Membranes Membranes were thawed slowly and adjusted to 5 - 10 mg protein/ml and to 50 mM sodium phosphate (buffer pH 7.0). MgC12, which had been included in the assay mixture in earlier experiments [13], was omitted as it was found to be inessential for binding. For the standard assay of penicillin-binding proteins 20 pl of ben~yl['~C]penicillinwas bound to 200 pl of membranes for 10 rnin at 30 "C and the reaction was terminated by the addition of 5 pl of non-radioactive benzylpenicillin (120 mg/ml) and 10 pl of 20% (w/v) sodium lauroyl sarcosinate (Sarkosyl NL-97). The addition of the anionic detergent sarkosyl has a dual
Penicillin-Binding Proteins of Escherichia coli
purpose. Firstly it denatures the penicillin-binding proteins and prevents the enzymic loss of bound penicillin (see Results), and secondly, it selectively and completely solubilises the proteins of the inner membrane whilst leaving the outer membrane insoluble 1151. After solubilisation of the inner membrane for 20 min at room temperature, the sarkosyl-insoluble outer membrane and peptidoglycan were removed by centrifugation at 100000 x g for 40 min at 10 "C in the 50Ti rotor of a Beckman ultracentrifuge using 10 ml Oak-Ridge-type polycarbonate screw-cap tubes. In recent experiments the removal of the sarkosylinsoluble fraction by centrifugation at 40000 x g in the SM24 rotor of a Sorvall RC2B centrifuge has been used with identical results. 100 p1 of the sarkosylsoluble supernatant were added to 50 p1 of gel sample buffer (0.2 M Tris-HC1, pH 6.8 ; 3% w/v sodium dodecylsulphate; 30'x vjv glycerol; 0.002 bromophenol blue). To study binding to the outer membrane, the sarkosyl-insoluble fraction from above was re-extracted with 2 ml of 1% sarkosyl for 5 min at room temperature, pelleted by centrifugation at 100000 x g for 40 rnin at 10 "C, and resuspended in 200 pl of distilled water. 100 pl was added to 50 p1 of gel sample buffer. All samples were then fractionated by dodecylsulphate-polyacrylamide slab gel electrophoresis as described below. Experiments with other P-['4C]lactams were performed in a similar manner.
Slab Gel Electrophoresis and Detection of Penicillin-Binding Proteins 20 pl of 2-mercaptoethanol were added to each of the above samples immediately before they were heated for 3 rnin in a boiling water bath. 50-65 p1 of each sample (containing about 200 pg protein) were loaded into the gel slots of a 10% sodium dodecylsulphate polyacrylamide slab gel. The apparatus and the discontinuous buffer system have been described in detail by Laemmli and Favre [16]. Electrophoresis was performed at a constant current of 20 mA (initial voltage of about 60 V increasing to about 200 V at the end of electrophoresis) until the bromophenol blue tracking dye had migrated 11 cm into the separating gel. The gel was removed from the apparatus, stained for 1 h with 0.1% Coomassie brilliant blue R (Sigma) in 50% methanol 10% acetic acid, diffusiondestained with several changes of 5% methanol 10% acetic acid, and prepared for fluorography (scintillation autoradiography) by the incorporation of the scintillant PPO into the gel matrix as described by Bonner and Laskey [17]. Staining with Coomassie brilliant blue was used only to monitor the quality of
B. G. Spratt
the protein fractionation on the gel since the stained bands were not visible clearly after fluorography. The PPO-impregnated gel was dried under vacuum on to Whatman number 17 filter paper and placed on Kodak RP Royal X-Omat X-ray film at -80 "C. X-ray films were developed in Kodak liquid X-ray developer for 5 rnin at 19 "C without agitation, washed in 3% acetic acid and fixed for 5 min at 19 "C in Kodak X-ray fixer. Conventional autoradiography (without incorporation of PPO) was used in one experiment and in this case Ilford industrial G X-ray film was used. The use of fluorography is essential to obtain the maximal sensitivity of 14C detection on the X-ray film and even with this technique the exposure times required are still inconveniently long. The binding proteins can be detected by eye after about 10 days fluorography but a considerably longer exposure is required to obtain image densities sufficient for accurate densitometry of the minor proteins. Fluorography probably introduces some errors into the quantitisation of the results since the image density on the X-ray films is not linear with the amount of radioactivity applied [18]. This problem can be eliminated by pre-fogging of the X-ray films [18]. A further substantial increase in sensitivity is obtained by the removal of the outer membrane proteins before applying samples to the gels. This is possible as we have failed to detect any penicillinbinding proteins in the E.coli outer membrane. Although the outer membrane constitutes only about 50% of the total protein of the membranes, the majority of the outer membrane consists of three proteins with almost identical mobilities on dodecylsulphate-polyacrylamide gels [ 191 and this results in overloading when relatively small amounts of outer membrane protein are applied to the gels. In contrast, the inner membrane consists of many proteins in approximately similar amounts, with the result that much larger amounts of protein can be applied before distortion of the gels due to overloading occurs. Extraction of the inner membrane is complete under the conditions that we use, however, an increase in the concentration of sarkosyl to at least 2% can be used without interfering with the running of the gels or significantly extracting outer membrane proteins. Apparent molecular weights of the penicillinbinding proteins were measured by comparison of their mobilities on dodecylsulphate-polyacrylamide slab gels with that of eight proteins of known molecular weights. The proteins and their molecular weights were : P-galactosidase (130000), phosphorylase A (lOOOOO), bovine serum albumin (68 000), pyruvate kinase (57 OW), ovalbumin (43 000), glyceraldehyde-3phosphate dehydrogenase (36000), carbonic anhydrase (29000), and chymotrypsinogen (25 700).
343
Measurement of the Binding of P-Lactams by Competition with Benzyl[14C]penicillin
The binding of non-radioactive P-lactams was studied by measuring their competition for the binding of ben~yl['~C]penicillin to each of the proteins. 200 pl of membranes (5 - 10 mg protein/ml in 50 mM sodium phosphate buffer pH 7.0) were pre-incubated with 10 p1 of either distilled water or dilutions of the non-radioactive p-lactam for 10 rnin at 30 "C and then 20 pl of 50 pCi/ml ben~yl[~~C]penicillin were added for a further 10 min at 30 "C and the binding terminated by the addition of 5 p1 of non-radioactive benzylpenicillin (120 mg/ml) and 10 p1 of 20% (w/v) sarkosyl as described above. The final concentration of ben~yl['~C]penicillin (30 pg/ml) was approximately three times that required to obtain saturation of all binding proteins under the conditions used. Samples were fractionated by electrophoresis and the level of ben~yl['~C]penicillin bound to each protein measured as described above.
Release of Bound Benzyl[ ''C]penicilEin
To measure the spontaneous rate of release of the product of bound ben~yl['~C]penicillin(which is not benzylpenicillin itself but probably either benzylpenicilloic acid [9] or phenacetylglycine and thiazolidine ring fragmentation products [20, 21]), membranes (5 - 10 mg protein/ml in 50 mM sodium phosphate buffer pH 7.0) were labelled with benzyl[14C] penicillin (16.5 pg/ml final concentration) for 10 rnin at 30 "C; a 1000-fold excess of non-radioactive benzylpenicillin was added, and samples of 200 p1 removed immediately and at intervals during continued incubation at 30 "C into tubes containing 10 p1 of 20% (w/v) sarkosyl. Denaturation of the penicillinbinding proteins in sarkosyl prevented further release of ben~yl['~C]penicillin since release is enzymic (see Results). The samples were fractionated by electrophoresis and fluorographs prepared as described above. The spontaneous release of other fi-['"C] lactams was measured in a similar fashion. Various compounds (e.g. 2-mercaptoethanol, neutral hydroxylamine) were tested for their effect on the spontaneous rate of ben~yl['~C]penicillin release by adding them immediately after the excess non-radioactive benzylpenicillin and sampling as above into sarkosyl. In some experiments where the excess benzylpenicillin was known, or suspected, to react with the test compound, the further binding of ben~yl['~C]penicillinwas prevented by the addition of 0.1 vol. of penicillinase (100000 units/ml) for 5 min at 30 "C. Penicillinase does not remove bound penicillin.
Penicillin-Binding Proteins of Escherichia coli
344
Quantit isation of Penicillin-Binding Proteins by Densitometry of X-Ray Films X-ray films of sufficient image density for accurate quantitisation (but not so dense as to approach saturation of the film emulsion) were scanned with a Joyce-Loebl M k l 1lc microdensitometer. The peaks obtained were carefully cut out and weighed. Corrections were made, when necessary, for differences in widths of the gel slots.
Protein 1
2
3
Measurement of Minimal inhibitory Concentrations of p- Lac tams
5
6
'
Minimdl inhibitory concentrations weredetermined using E. coli KN126 growing exponentially at 37 "C in penassay broth with initial cell densities of approximately 2 x lo7 organisms/ml. The minimal inhibitory concentration was defined as the lowest concentration of the p-lactam which produced a clear alteration of the normal rod morphology of the cells (depending on the 8-lactam this could be the appearance of cell lysis, filamentation or ovoid cells).
Chemicals Ben~yI['~C]penicillinwas purchased from the Radiochemical Centre, Amersham, England (specific activity 53 - 54 Ci/mol). ['4C]Cefoxitin (2.8 Ci/mol) and ['4C]cephacetrile (4.6 Ci/nioI) were generously provided by Dr Edward Stapley (Merck Sharp and Dohme, Rahway, New Jersey, U.S.A.) and D r W. Zimmermann (Ciba-Geigy, Basel, Switzerland) respectively. All three radioactive p-lactams were of high purity and were labelled in the R, sidechain (position 6 of penicillins ;position 7 of cephalosporins). Reference samples of 8-lactams were kindly provided by the following companies: ampicillin by Bristol Laboratories (Syracuse, New York, U.S.A.), cephacetrile and cephradine by Ciba-Geigy Ltd (Basel, Switzerland), mecillinam by Leo Laboratories, (Ballerup, Denmark), cephalexin, cephaloridine and cephalothin by Eli Lilly andCo. (Indianapolis, Indiana, U.S.A.), cefoxitin by Merck, Sharp and Dohme (Rahway, New Jersey, U.S.A.), benzylpenicillin, cephalosporin G and penicillin V by Squibb and Sons (Princeton, New Jersey, U.S.A.). /j-['"C]Lactams were stored as solids at -2O';C and were dissolved in 10 mM sodium phosphate buffer pH 7.0 immediately before use. Aqueous solutions were stored at - 80 "C and could be used for at least a week without the appearance of non-specific binding problems due to degradation products of the fl-lactams. Solutions of nonradioactive p-lactams were always prepared fresh immediately before each experiment. The solid compounds were stored desiccated at - 20 "C.
Fig. 1 . Detection uf rhe E. coli penicillin-binding proteins utzd their location in the inner memhrune. 40 p1 of ben~yl['~C]penicillin (31 pg/ml final concentration) were added to 400 pl of membranes for 10 min at 30 "C. To measure penicillin-binding proteins in the total membrane 100 pI of the assay mixture was added to 50 pl of gel sample buffer. A further 200 p1 were added to 10 pl of 20"/> (w/v) sarkosyl and the sarkosyl-soluble (Inner membrane) and sarkosyl-insoluble (outer membrane) fractions were obtained as described in Methods. All three samples were fractionated on a 10% polyacrylamide slab gel and a fluorograph prepared after exposure of the gel against Kodak RP Royal X-Omat X-ray film for 32 days at -80 'C. (A) Total membrane; (B) outer membrane; (C) inner membrane
Penicillinase was obtained from Wellcome Reagents Ltd (Beckenham, England). Sarkosyl NL-97 was purchased from Geigy Industria1 Chemicals (New York, U.S.A.).
RESULTS All the penicillin-binding proteins are located in the E.coli inner membrane, defined as that fraction of the cell envelope which is solubilised by lo( sarkosyl at room temperature. Fig.1 shows the binding of ben~yl['~C]peniciIlin to total cell envelopes and to the sarkosyl-soluble and sarkosyl-insoluble fractions. No penicillin-binding proteins have been detected in the sarkosyl-insoluble (outer membrane) fraction with any of the P-['4C]lactams that we have studied. In all further experiments only the sarkosyl-soluble (inner membrane) fraction was analysed for penicillinbinding proteins. All the proteins are accessible to ben~yI[~~C]penicillin in whole cells since they could be detected in membranes prepared from whole cells which had been incubated with ben~yl['~C]penicillin.
345
B. G. Spratt
In this experiment mercaptoethanol was omitted during sonication as this compound releases bound penicillin (see below). Since several of the chemicals which are used in dodecylsulphate-polyacrylamide gel electrophoresis (e.g . ammonium persulphate, 2-mercaptoethanol) have been reported to release bound penicillin [12], we have re-investigated the effects of these compounds on the recovery of the penicillin-binding proteins. 2-Mercaptoethanol released bound penicillin rapidly from proteins 5 and 6 if it was added to the binding assay mixture before the addition of sarkosyl (see below). However this release was clearly enzymic since it was completely prevented by denaturation of the proteins in 1% sarkosyl. Once denatured, we could detect no differences in the levels of the penicillinbinding proteins in samples which had been boiled in gel sample buffer in the presence or absence of 10% (v/v) 2-mercaptoethanol (data not shown). Similarly we could find no effect of ammonium persulphate (0.1%) on the release of bound penicillin once the proteins had been denatured in sarkosyl. In earlier experiments [13] we included 10 mM MgC1, in the penicillin-binding protein assay mixture but since we find no stimulation of binding by Mg2+, or any inhibition of binding by 1 mM EDTA, we now omit MgCl, from the assay. Pre-incubation of samples of membrane with N-ethylmaleimide (5 mM) or p hydroxymercuribenzoate (1.5 mM) for 10 min at 30 "C did not prevent binding to any of the proteins. The pattern of six proteins is extremely reproducible. In some gels additional minor penicillinbinding proteins were observed. These bands were in general non-reproducible, although two bands, proteins 7 and 8, with molecular weights 29000 and 32000 were apparent in several gels. An identical pattern of penicillin-binding proteins was found in all E.coli K12 strains that we have examined. E.coli B/r strains also showed an identical pattern although protein 7 was more prominent. The apparent molecular weights of the proteins were determined by comparing their mobility on a dodecylsulphate-polyacrylamide slab gel with that of eight proteins of known molecular weights. Table 1 shows the apparent molecular weights, and the relative abundance of each penicillin-binding protein, as well as an estimate of the number of molecules of each protein per bacterium (see below). The penicillin-binding proteins are minor components of the E.coli inner membrane. Fig.2 shows the location of these proteins on a Coomassie-bluestained gel of the inner membrane (note the absence of the characteristic major outer membrane proteins [19] with molecular weights 36000 on this gel of the sarkosyl-soluble membrane). The major proteins (5 and 6) can be tentatively identified as minor stained bands but the others are below the resolution of detection as stained bands.
Table 1. MoJecufar weights and relative abundance of the penicillinbinding proteins Protein
Apparent molecular weight
Binding of penicillin
Molecules/cell"
%total
91 OOO 66 000 60000 49000 42000 40000
1 2 3 4
5 6 a
8.1 0.7 1.9 4.0 64.7 20.6
230 20 50 110 1800 570
See Results for details
Kinetics of Binding of p-[ 14C]Lactams
Benzyl[14C]penicillin was added in 2-fold increasing concentrations (final concentrations of 0.015 -31.3 pg/ml) to aliquots of membranes for 10 min at 30 "C and the extent of binding to each of the proteins measured as described in Methods. The binding of penicillin to proteins 1-4 was treated as a simple bimolecular irreversible reaction. For such a reaction the degree of acylation by penicillin should be proportional to the product of the penicillin concentration and the time of incubation [22]. Fig.3 shows that proteins 1- 3 bind penicillin in accord with the above kinetics, but protein 4 shows more complex kinetics. Proteins 5 and 6 were not plotted in this way as they released bound penicillin with half lives which are of the same order as the incubation time of the assay (see below). Similar kinetics of ben~yl['~C]penicillinbinding were obtained when a fixed concentration of penicillin was bound for increasing incubation times. Saturation of all penicillin-binding proteins was essentially complete at 80 pg m l - ' min. The concentration of ben~yl['~C]penicillinin our standard binding assay was over 3.5 times this level (313 pg ml-' min) to ensure saturation of all penicillinbinding proteins. Although experiments with ['4C]cefoxitin and ['4C]cephacetrile were undertaken (see below) the specific activity of these materials was too low to obtain images on the X-ray films of sufficient density for accurate quantitisation. It was, however, observed that [14C]cephacetrilebound to the same six proteins as ben~yl['~C]penicillin and that ['4C]cefoxitin bound to all except protein 2. No additional penicillinbinding proteins were observed with any p-['"C] lactams. Reversihility of the Binding of p-[ ''C']Lnctan~.s The penicillin-binding proteins were saturated with benzyl['"C]penicillin, excess non-radioactive
Penicillin-Binding Proteins of Escherichia coli
346
M,
1
130-
Protein 100-
-1
-2 57.--c
--3 Benzylpenicillin Ipg rn-lrnin)
L3
36
29
26
-
4-5 4-6
--
Fig. 2. Location of the penicillin-binding proteins on a Coomassieblue-stained gel of the inner membrane. Ben~yl['~C]penicillin (31 pg/ml final concentration) was bound t o membranes for 10 min at 30 'C and the sarkosyl-soluble inner membrane proteins were fractionated on a 10% polyacrylamide slab gel. Eight proteins of known molecular weights were run in a neighbouring gel slot. The Coomassie-blue-stained gel was autoradiographed on Ilford industrial G X-ray film for 90 days and the positions of the penicillinbinding proteins and the molecular weight standards marked o n a photograph of the stained gel
Fig. 3. The kinetics of binding of benzyl[ ''C]penicillin. Ben~yl['~C] penicillin was bound to aliquots of membranes for 10 min at 30 "C at ?-fold increasing concentrations from 0.015- 31.3 pg in1 (w. from 0.15 - 31 3 pg m1-l min). The binding was terminated by the addition of excess non-radioactive benzylpenicillin, the sarkosylsoluble proteins fractionated on a lox, polyacrylamide slab gel, and a fluorograph prepared. The extent of binding to each of the penicillin-binding proteins 1, 2, 3 and 4 was measured by microdensitometry of the X-ray film. P is the saturation level of binding and P, is the level of binding at penicillin concentration .Y (pg ml-' min)
benzylpenicillin added, and the release of bound ben~yl[~~C]penicillin from each of the proteins measured during continued incubation at 30 "C. Fig.4 shows the rapid release of penicillin from proteins 5 and 6 but only very slight release from the other proteins. Half times of release for proteins 5 and 6 were 5 and 19 min respectively at 30 "C. Accurate half times of release for proteins 1-4 could not be obtained because of the scatter in the data points, but at least 50% of the benzyl['4C]penicillin remained bound to each of these proteins during 1 h at 30 "C. [14C]Cefoxitin was not significantly released from proteins 1,3,5 or 6 during a 45-min period at 30 "C, but it was released from protein 4 with a half time of approximately 15 min (cefoxitin does not bind to protein 2). The effect of various compounds on the rate of release of bound benzylpenicillin were studied. NO difference in the rate of release from any of the proteins was detected whether the ben~yl['~C]penicillin was diluted with excess non-radioactive benzylpenicillin or removed with penicillinase, indicating that
341
(final concentration 0.05 M) also stimulated release from proteins 5 and 6 to a slight extent. Neutral hydroxjlamine (final concentration 0.2 M) stimulated release from all proteins with the exception of protein 4. It is possible that higher concentrations of hydroxylamine would effect release from protein 4 but interference of the electrophoresis by higher concentrations of the compound prevented this from being studied. Table 2 summarises the results obtained on the release of ben~yl[~~C]penicillin.
benzylpenicillin itself (15 mg/ml) did not stimulate release. The rate of benzylpenicillin release from proteins 5 and 6 was markedly enhanced by 2mercaptoethanol (final concentration 0.14 M), whereas proteins 1-4, which do not spontaneously release benzypenicillin at an appreciable rate, are not stimulated to d o so by 2-mercaptoethanol. Dithiothreitol
Protein
Competition of’/3-Lactams,forBenzyl[ ‘‘C]penicillin Binding
0
------a&.
50
Since most B-lactams were not available radioactively labelled, we have studied their binding by the ability to compete with ben~yl[’~C]penicillin. Such experiments are not entirely satisfactory as competition by a 8-lactam depends both on its affinity for the binding proteins and on its rate of release. Clearly a p-lactam will show neglible competition if it has the same affinity for a penicillinbinding protein as ben~yl[‘~C]penicillin but is released 100 times as fast. Fig.5 shows the competition of three concentrations of cephalothin, cephalosporin G and penicillin V for the penicillin-binding proteins. Fig. 6 shows the result of a more complete competition experiment with cephradine and Table 3 gives the minimal inhibitory concentrations of a selection of 8-lactams and the concentrations required to reduce ben~yl[’~C]penicilIin binding to each protein by 50% under the conditions of the assay. Although penicillins showed competition with all proteins, several cephalosporins show competition for proteins 1,3 and 4 but little or none for proteins 2,5 and 6. This failure to obtain competition could either be due to their failure to bind to the latter proteins or due to their rapid rate of release. We have attempted to distinguish between these possibilities
$ 100 1
--0
0
0
0 -
0
V
x
0
\
Time(rnin1 Fig. 4. Releusr of hound henzyl[ ‘“C]penici/lin.0.2 ml of 50 pCi/ml benzyl[14C]penicill~nwas added t o 4 ml of membranes for 10 min at 30 “C (final penicillin concentration 16.5 pg/ml). 60 mg of nonradioactive benzylpenicillin were added and 0.2-ml samples were removed immediately and at intervals during continued incubation at 30 “C. 10 pl of 20% (w/v) sarkosyl were added to each sample to prevent further release of bound benzyl[14C]penicillin and the sarkosyl-soluble proteinswerefractionated on a 10% polyacrylamide slab gel and a fluorograph prepared. The level of b e n ~ y l [ ’ ~ C ] penicillin remaining bound to each penicillin-binding protein was measured by microdensitometry of the X-ray film
Table 2. Relrme ofhenzyl[ ’‘C]penicillin ,from the pmic~illiii-hiiitliii,~ prorc’iii.r The rate of release of bound ben~yl[’~C]penicillin from each penicillin-binding protein was measured at 30 ’ C in 0.05 M sodium phosphate buffer p H 7.0 as described in Methods. It was measured either after addition of excess unlabelled benzylpenicillin (A) o r after addition of penicillinase (B) Additions t o basic assay mixture
Measured
Time at 30 ‘C for 50% release of bound ben~yl[’~C]penicillin ~~
No addition No addition 2-Mercaptoethanol(O.14 M) Dithiothreitol(O.05 M ) Dithiothreitol(O.05 M) Neutral hydroxylamine (0.2 M )
A B A A
B B
6
1
2
3
4
> 60
> 60 > 60 > 30 > 30 > 30 18
> 60 > 60 > 30 > 30 > 30 8
> 60
5
> 60 > 30 > 30 > 30 > 30
5 30 > 30 > 30 22
5
19 20 2 12 7
