MICROBIOLOGY LETTERS

FEMS Microbiology Letters 129 (1995) 215-220

Depletion of anionic phospholipids has no observable effect on the anchoring of penicillin binding protein 5 to the inner

membrane of Escherichia coli

Frederick Harris, Lee K. Chatfield, David A. Phoenix *

Unicersity of Central Lancashire, Department of Applied Biology, Preston PRl 2HE, UK

Received 14 March 1995; revised 14 April 1995; accepted 18 April 1995

Abstract

Escherichiu coli penicillin-binding protein 5 (PBPS) is anchored to the periplasmic face of the inner membrane via a C-terminal amphiphilic a-helix. The results of washing experiments have suggested an electrostatic contribution to the anchoring mechanism which may involve the cationic region of the C-terminal a-helix. Similarities between this anchor

domain and some surface active agents, such as melittin, suggest that the cationic region of the PBP5 anchor may require the presence of anionic phospholipids for membrane interaction. Washing experiments performed on membranes of HDLll, an E. cofi mutant in which the expression of the major anionic phospholipids is under lac control, found no such requirement. The results are discussed in relation to the hypothesis that the cationic region may interact with other sources of negative

charge, possibly arising from a PBP complex.

Keywords: Escherickia coli; Penicillin-binding protein; Membrane anchor; Anionic phospholipid

1. Introduction

Penicillin-binding proteins (PBPs) are a group of penicillin-sensitive enzymes which possess active site serine DD-peptidase activity. PBPs are involved in

prokaryotic cell wall assembly where they catalyse the final stages of peptidoglycan biosynthesis [l]. In

Escherichiu coli there are seven major PBPs which have been well characterised. These are all mem-

brane-bound proteins and are anchored to the periplasmic face of the inner membrane via one of

* Corresponding author. Tel.: +44 (1772) 893 519; Fax: +44 (1772) 892 903; e-mail: d.a.phoenix@Uclan.ac.uk

two general mechanisms. The high molecular mass

PBPs (PBPla, lb, 2 and 3) possess N-terminal hy- drophobic amino acid sequences which span the

bilayer whereas the low molecular mass PBPs (PBP4, 5 and 6) are thought to interact with the membrane via C-terminal amphiphilic a-helices [2]. These (Y-

helical anchors appear to have structural similarities to regions of some surface active molecules and therefore may have similar requirements for mem-

brane interaction. In this paper we investigate such a possibility for the anchor region of PBPS.

PBPS possesses DD-carboxypeptidase activity and is believed to have a regulatory role in E. coli cell wall assembly [2j. The protein is 374 amino acid residues in length and deletion experiments have

0378-1097/95/$09.50 0 1995 Federation of European Microbiological Societies. All rights reserved

SSDI 0378-1097(95)00160-3

216 E. Harris rr 01. /FkM‘MS Microhrology Lenen I20 (19Y.T) 215-220

Fig. I. The PBPS anchor region projected onto a Schiffcr-

Edmundson wheel. The helical sequence is formed by the C-

terminal 18 amino acid residues (~1357-374) of PBPS. The

sequence is assumed to be a-helical with each successive residue

100” from the previous one. Hydrophobic residues are shaded.

The positively charged amino acids, histidine and lysinc are

shown in the upper hydrophilic portion of the wheel [IS].

shown that the C-terminal 18 residues are essential

for efficient membrane interaction [3,4]. When pro- jected onto a Schiffer-Edmundson wheel [5], the

C-terminal residues are seen to have the potential to form an amphiphilic a-helix (Fig. 1). This a-helix has been shown to possess a large hydrophobic moment which is characteristic of surface-seeking

proteins and polypeptides [6]. Surface active molecules tend to have highly am-

phiphilic o-helical regions which adopt a preferred orientation at the membrane interface. In this orienta- tion the hydrophilic face of the a-helix projects into

the aqueous environment whilst the hydrophobic face interacts with some component of the lipid bilayer

core [7]. This model may hold for the PBPS anchor region,

since it has been shown that the interaction of the

PBPS anchor with the inner membrane is predomi- nantly hydrophobic and is particularly susceptible to chaotropic agents such as thiocyanate. However, it was also found that this susceptibility was dependent on pH. At low pH, PBPS extraction was resistant to perturbants but as pH was increased above neutrality, progressively greater amounts of PBP.5 were dis- placed from the inner membrane [8]. This observa- tion suggests the involvement of an electrostatic contribution in the anchoring mechanism.

Electrostatic interactions are also a feature of the membrane interactions of some surface active pro-

teins and polypeptides. For example, melittin from bee venom [9] and the bacterial toxin colicin A [ 101

require the presence of anionic phospholipids for

membrane association. Interaction between the nega- tively charged phospholipid headgroups and cationic regions within the structure of the protein or

polypeptide acts to stabilise membrane association. The cationic regions involved in these interactions

are frequently located on the hydrophilic face of the

amphiphilic cy-helices found in such molecules. For

example, melittin possesses positively charged N- terminal and C-terminal regions, both of which are

situated on the hydrophilic face of the polypeptide’s single a-helix [9].

PBPS has four basic amino acid residues (two

histidines and two lysines) on the hydrophilic face of its anchor region. The presence of the histidine residues provides a cationic region within the PBPS anchor which could vary in magnitude in a pH-de-

pendent manner which mimics that of the anchor region’s susceptibility to perturbants. This suggests

that this positively charged region may contribute to an electrostatic interaction in PBPS membrane bind-

ing. Since the PBPS anchor possesses molecular architecture of a similar nature to other surface ac- tive agents, it seems reasonable to postulate that this electrostatic interaction may involve anionic phos-

pholipids, by analogy with cationic regions of these surface active molecules.

The main anionic phospholipids found in

[1 11. If an electrostatic interaction

involving anionic phospholipids does contribute to the PBPS anchoring process, then removing PG and DPG from the membrane should destabilise mem-

brane binding. Under these conditions, increased amounts of PBPS may be expected to be displaced from the inner membrane by perturbants.

To investigate this possibility, the E. coli strain HDLll/pLG364 was used in washing experiments. In this strain the pgsA gene, which encodes phos- phatidylglycerolphosphate synthetase, an enzyme in- volved in PG and DPG synthesis, has been placed under the control of the luc promoter [12], and pgsA

F. Harris et al. / FEMS Microbiology Letters 129 (1995) 215-220 217

expression can be induced by addition of the lac inducer isopropylthiogalactoside (IPTG). In the ab- sence of IPTG, phosphatidylglycerolphosphate syn- thetase is still produced at a low, basal level and this is sufficient to permit the viability of this strain

given the additional presence of a lpp2 deletion [13]. The plasmid pLG364 carries the dacA 11191 PBPS

allele, and overproduces this protein.

2. Materials and methods

2.1. Bacterial strains and growth conditions

The E. coli strains used were: SP1048 (his, tsx, supF, srE::TnlO, AdacCl, AdacA::Km”; [8]) and HDLll (pgsA:: kan, +(lacOP-pgsA+)l, LacZ’, lacy ::Tn9, lpp2, zdg::TnIO; 1121). SP1048 and

HDLll were each transformed with the multicopy plasmid pLG364 [14] which carries the dacA 11191 mutation of PBPS on a BamHI-EcoRI fragment in

pBR322 [ 151. All strains were grown in nutrient broth (Lab M,

No. 2) supplemented with ampicillin (25 pg ml-’ 1

at 37” C with aeration. When required, HDLll strains were grown in the presence of 50 pM IPTG to induce high levels of pgs4 gene expression.

2.2. Determination of phospholipid content of bacte- rial strains

Cultures (1 1) were grown in the presence of 10 pCi of [r4C]sodium acetate (specific activity; 50.2

mCi mmoll’, Sigma) to an OD,,,, of 0.6. Lipids

were extracted from cells [16] and separated by thin

layer chromatography (Silica gel, type G, Sigma) using chloroform/methanol/acetic acid/water

(170:30:20:7) as solvent. The lipids were visualised by iodine vapour and identified using appropriate standards. Incorporation of [r4C]acetate into lipids was quantified using a Beckman LS5801 scintillation counter.

2.3. Preparation of envelope fractions

Cultures (1 1) were grown to an OD,,,, of 0.6. Envelope fractions were prepared by osmotic lysis [17]. DNA released from cells was sheared mechani-

tally using 23 G and 25 G syringes and unlysed cells

were removed by low-speed centrifugation (1100 X

g, 15 min, 4” Cl. Samples (8 ml) of the supernatant were centrifuged (100000 X g, 60 min, 4” C) and the

resulting pellets were stored at - 20” C.

2.4. The effect of perturbants on the anchoring of PBP.5 in envelope fractions

Envelope pellets were resuspended in 1 ml of 4 M

urea, 2 M sodium thiocyanate or 2 M sodium chlo- ride at pH 6, 7 or 8 and left on ice for 1 h at 0” C.

Solutions were prepared in 10 mM phosphate buffer.

Envelope pellets were also resuspended in 10 mM

phosphate buffer alone at pH 6, 7 or 8 [S]. The volume of the resuspended pellets was made up to 8

ml with the appropriate perturbant at the appropriate pH and centrifuged (100 000 X g, 60 min, 4” Cl. The pellet was resuspended in a minimum of resuspen-

sion buffer (13 parts of wash buffer (KH,PO,, 3 g 1-l; Na,HPO, .2H,O, 8.9 g 1-l; NaCl, 4.0 g I-‘;

MgSO, .7H,O, 0.1 g I-‘); 5 parts of saturated Tris

base; 2 parts of phenylmethylsulfonylfluoride in dimethyl sulfoxide (6 mg ml-’ 1). Trichloracetic acid

was added to the supernatant to give a final concen- tration of 10% (w/v), the mixture left on ice for 1 h,

centrifuged (3000 X g, 15 min, 4” C) and the pellet resuspended in a minimum of resuspension buffer (lo-25 ~11. Protein in the resuspended pellets was resolved by sodium dodecylsulfate polyacrylamide

gel electrophoresis (12% resolving gel) [18]. PBPS was visualised by Western blotting [19] with the exception that 3,3’-diaminobenzidene (Sigma) was

used as substrate and quantified by densitometry using a Shimadzu CS-9000 dual wavelength flying

spot scanner at 500 nm.

3. Results and discussion

When HDLll/pLG364 was grown in the pres- ence of IPTG, the overall level of anionic phospho- lipids was comparable to the control strain

SP1048/pLG364 (Fig. 2). However, when the or- ganism was grown in the absence of IPTG, this level was decreased by approximately 60%. The predomi- nant anionic phospholipid in the membrane is PG, and the level in uninduced cells was reduced by

218 p. Harris et al. / FEMS Microbiology Letters 129 ~199.5) 215-220

100 1 T

,L \ (a) (b)

4 I CC)

Fig. 2. Phospholipid content of bacterial strains examined. The

relative percentages of the major phospholipids in the membranes

of the E. cok strains (a) SP1048/pLG364, (b) HDLl 1 /pLG364

grown in the presence of 50 FM IPTG and (c) HDLI l/pLG364

grown in the absence of IPTG were determined by radio-labelling

with [ “Clacetate and quantification via scintillation counting, all

as previously described. Phosphatidylethanolamine ( 0 ) was the

major component with contributions from the anionic phospho-

lipids phosphatidylglycerol (0 ) and diphosphatidylglyccrol and

phosphatidic acid which co-chromatographed together (W ). The

total anionic phospholipid content is also shown (0 ). Error bars

denote the standard deviation for n = 3.

85%. This agrees well with data by Van der Goot et al. [lo], who showed that the surface active bacterial

toxin, colicin A, has an in vivo requirement for the presence of anionic phospholipids if it is to interact

with the membrane. Using HDLll, they demon- strated that the toxin was fully active against this

strain in the presence of induced levels of acidic phospholipids, but toxicity was reduced by a factor of three when in the presence of non-induced levels of these phospholipids [lo].

Despite structural similarities between the mem-

brane interactive regions of colicin A and the anchor domain of PBPS, we failed to detect a similar re- quirement for the presence of anionic phospholipids in PBPS membrane binding using HDLll. Even a reduction of approximately 60% in acidic phospho- lipids led to no significant difference in the amounts of PBP5 displaced from membranes of HDLll by perturbant action. The levels of PBPS displaced from membrane fragments of SP1048/pLG364 and

HDLll/pLG364 grown in the presence of IPTG are comparable to those displaced from HDLl 1 /pLG364

grown in the absence of IPTG (Fig. 3). This would

appear to preclude anionic phospholipids as a PBPS anchoring requirement, although the possibility ex-

ists that the magnitude of their interaction is too

small to be detected by this experimental system. However, the results do not exclude phospholipids,

charge interaction or the cationic region of the PBPS

anchor from contributing to the anchoring process.

The trends shown by the results of these washing

experiments are in agreement with the findings of previous studies [g] and indicate that the PBPS an- chor region becomes progressively more susceptible

to chaotropic agents with increasing pH, but main- tains its resistance to extraction by ionic perturbants. The overall pH dependence of perturbant action reit-

erates the possibility of an electrostatic contribution to membrane binding and, if anionic phospholipids

are not involved in the binding process, this could

mean that the anchor’s positively charged region may interact with negative charge from other sources.

One possibility is that the cationic region may interact with negatively charged regions in the ec- tomembranous domain of PBPS. Such an interaction could serve to stabilise membrane binding and may be associated with changes in the conformation of

the protein. It has been shown that the overall con- formation of PBP5 can affect the strength of binding

and, in vivo, anchoring may be related to the enzy- matic activity of the protein [20]. PBPS may undergo

conformational changes upon membrane interaction since it has been found that purified PBPS shows

resistance to proteolytic action when reconstituted into vesicles but that it is readily degraded by pro- teases in the unreconstituted form (Phoenix and Pratt,

unpublished data). Another possibility is that the positively charged

region of the PBPS anchor may interact with other proteins, either a receptor protein or as part of a protein complex. Phoenix and Pratt (unpublished data) found that PBPS would not reconstitute into vesicles where the cytoplasmic face was accessible

but would reconstitute into vesicles where the periplasmic face was accessible. These findings are consistent with a model for PBPS anchoring which does not require anionic phospholipids but which requires instead some other component of the

F. Harris et al. / FEMS Microbiology Lerlers I.?9 (1995) 215-220 219

pif 7

(aI

PH’ PHI

(h)

2!i 1 2s ,

PB 6 P” ’ PBS pH6 pH7 PHR

(C) VJ)

Fig. 3. The relative amounts of PBP5 displaced from membrane fragments of strains of E. coli by the action of chaotropic and ionic

pertubants. Membrane fragments prepared from SP1048/pLG364 (a), HDLl l/pLG364 grown in the presence of 50 PM IPTG ( q ) and

HDLll/pLG364 grown in the absence of IPTG Cm) were treated with (a) 4 M urea, (b) 2 M sodium thiocyanatc, or Cc) 2 M sodium

chloride all in 10 mM phosphate buffer at pH 6, 7 or 8. Fragments were also treated with (d) 10 mM phosphate buffer at pH 6, 7 or 8 as a

control. PBPS was visualised by Western blotting and quantified by densitometry, all as previously described. Error bars denote the standard

deviation for n = 3.

periplasmic face of the inner membrane - possibly a

proteinaceous component. In conclusion, we tentatively suggest that PBPS in

the native state has a minor electrostatic component which contributes to membrane binding, and which may involve the cationic region of the PBPS anchor domain. The anchor region appears not to interact with anionic phospholipids, but may interact with other negatively charged regions, either in PBP5 itself or those of other membrane-bound proteins.

Interestingly, evidence from cross-linking studies has

suggested that PBPS, PBP3 and either PBPla or lb have the capacity to form a protein complex [21], and at present we are investigating this possibility

further.

References

[I] Ghuysen, J.M. (1991) Serine p-lactamases and penicillin

binding proteins. Annu. Rev. Microbial. 45, 37-67.

[2] Gittins. R.G.. Phoenix, D.A. and Pratt. J.M. tlYY4) Multiple

[31

[41

[-‘I

161

[71

[Xl

[‘)I

[lOI

[I II

[I21

mechanisms of membrane anchoring of Escher-ichirr UI[I

penicillin binding proteins. FEMS Microhiol. Rev. 13. I-I?.

Pratt, J.M.. Jackson. M.E. and Holland. I.B. C IYX6) The

C-terminus of the penicillin hinding protein 5 is essential for

localisation to the Evchcrichrcr co/i inner mcmhranc. EMBO

J. 5. 23YY-2405.

Jackson, M.E. and Pratt. J.M. (lYX7) An IX amino acid

amphiphilic helix forms the mcmhranc anchoring domain of

the Eschericl~iu co/i penicillin hinding protcin 5. Mol. Mi-

crobiol. I, 23-2X.

Schiffer, M. and Edmundson, A.B. tIYh7) USC of helical

wheels to reprcscnt the structures of proteins and to identify

segments with helical potential. Biophys. J. 7. I2 I - 135.

Phoenix, D.A. (1900) Investigation into structural matures of

the Escherichia coli penicillin-binding protein 5 C-terminal

anchor. Biochem. Sot. Trans. IX. 948-94Y.

Eisenhcrg, D. ( 19X4) Three dimensional structure of mcnt-

hranc and surface proteins. Annu. Rev. Biochcm. 5.1. SY-

623.

Phoenix. D.A. and Pratt. J.M. ( IYYOJ pH-induced insertion ot

the amphiphilic (r-helical anchor of E,dwrrchicr w/t peni-

cillin binding protein 5. Eur. J. Biochem. IYO. 365-360.

Dempsey. C.E. (1990) The action of mclitin on memhrancs.

Biochim. Biophyx. Act. 1031. 143-161.

Van dcr Goot. F.G.. Didat, N.. Pattus. F.. Dowhan. W. and

Lctcllier. L. tlYY3) Role of acidic phospholipids in the

translocation and channel activity of colicin\ A and N in

Escherichiu co/i cells. Eur. J. Biochcm. 213, 217-271.

Gcnnis, R.B. (IYXYJ Biomcmhranes, Molecular Structure and

Function. p. 22. Springer-Verlag. London.

Kusters, R., Dowhan, W. and dc Kruijff, B. (IYYIJ Ncga

tivcly charged phospholipids r&ore prePhoE translocation

[I31

II41

[ISI

[Ihl

[I71

[ISI

[IYI

[Nl

[2ll

across phosphatidylglyccrol-dcplctcd Eschrrichiu co/i inner

membranes. J. Biol. Chem. 266. X654-8662.

Asai. Y.. Katayosc. Y.. Hikita, C., Ohta. A. and Shihuya. I.

(19XYJ Suppression of the lethal effect of acidic phospholipid

deficiency hy dcfectivc formation of the major outer-mcm-

branc lipoprotein in Flrchericlria co/i. J. Bactcriol. 171.

6X67-686’).

Hanahan, D. (1986) In: DNA Cloning (Glovcr. D.M., Ed.),

Vol. I, pp. 109%135. IRL Press, Oxford.

Broome-Smith, J.K. and Spratt, B.G. (IY84) An amino acid

substitution that blocks the deacylation step in the enzyme

mechanism of penicillin-hinding protein 5 of Eschrrichiu

coli. FEBS Lett. 165. 1X5-189.

Bolton, P. and Hatwood. J.L. (1977) Fatty acid biosynthesis

by a particulate preparation from germinating pea. Biochcm

J. 16X, 261-269.

Osborne. M.J. Gander, J.E. and Parisi, E. tlY72) Mechanism

of assembly of the outer membrane of Salmoncll~ h’-

phimrtrium. J. Biol. Chem. 247, 3973-39X6.

Lacmmli, U.K. (1970) Clcavagc of structural protcina during

the asscmhly of the head of hactcriophage T4. Nature 277.

6X0-685.

Towhin. H.. Staehelin. I‘. and Gordon, J. tlY7Y) Elcc

trophorctic transfer of proteins from polyacrylamide gels to

nitroccllulosc sheets: proccdurc and some applications. Proc.

Natl. Acad. Sci. USA 76. 4350-43.54.

Phoenix. D.A. and Pratt, J.M. tlYY3) Mcmhranc interaction

nf Eschcnchirr coli penicillin binding protein 5 is modulated

hy the cctomcmhranous domain. FEBS Lctt. 322, 215-21X.

Said. I.M. and Holtjc, J. (19x3) In: The Target of Penicillin

(Hakenback, R., Holtje. R. and Labischinski, H., Eds.). pp.

13%444. Walter de Gruyter. New York, NY.