Penicillin and Beyond: Evolution, Protein Fold, Multimodular Polypeptides, and Multiprotein Complexes

Mary

Ann I,¡chert.Inc.

Penicillin and

_Beyond:

_Evolution,

Protein

_Fold,

Multimodular

Polypeptides,

and

_{Multiprotein Complexes}

JEAN-MARIE

GHUYSEN,

PAULETTE

CHARLIER,

JACQUES

COYETTE,

COLETTE

DUEZ,

EVELINE

FONZÉ,

CLAUDINE

_FRAIPONT,

COLETTE

_GOFFIN,

BERNARD

JORIS,

and

MARTINE

NGUYEN-DISTÈCHE

ABSTRACT

As the

_protein

_sequence

and structure

databases

_expand,

the

_{relationships}

between

_proteins,

the notion of

protein superfamily,

and

the

_driving

forces

of

evolution

are better understood.

Key

steps

of

the

synthesis

of

the bacterial cell

wall

_{peptidoglycan}

are

revisited in

light

of

these

advances. The reactions

through

which

the

D-alanyl-D-alanine depeptide

is

formed, utilized,

and

_hydrolyzed

and

the sites

of action of

the

_glycopeptide

and

_/3-lactam

antibiotics illustrate

the

_concept

_according

towhich newenzyme

functions

evolve asaresult

of

tinkering

of

_{existing proteins.}

This

occurs

by

the

acquisition

of

local

structural

changes,

the fusion into

mul-timodular

_{polypeptides,}

and the

association into

_multiprotein

complexes.

INTRODUCTION

THE

TRANSLATION OF GENETIC INFORMATION into

_biological

activity

is achieved

_by

the conversion ofa

newly

synthe-sized

_polypeptide

chain into acompact,

correctly

folded

pro-tein.8

The

folding

code is still far from

being

understood. Nascent

_proteins

fold

_rapidly,

in secondsorless,in

spite

of the

fact that the timethatwould be neededtosearch all

_potentially

accessible conformations is astronomical. The solution tothis

paradox

is that anessential step in

protein folding

is the

for-mation ofa"molten

globule."

This

species

lacks the

persistent

tertiary

interactions characteristic of the native state,but it

al-ready

possesses extensive

secondary

structuresthat are

major

elements of the native

_topology.

Fromthis

intermediate,

the

re-maining

search forcorrect

folding

is

_only

over alimited

con-formationalarea.

Moreover,

thecells possess_manyfactors that

assist

folding

and minimize and/orcorrect

_misfolding

events.

The distribution of amino acid residue

_types

along

the

polypeptide

chain isa

major

determinant of

secondary

and

ter-tiary

structures.

Yet,

the number of distinct folds

adopted

by

the

_proteins

is limited. Proteins

_having

25%,

ormore,of their sequences incommon

_adopt

thesamefoldedstructures.

But,

at

the same

time,

an

increasing

number of

proteins

are

being

re-vealed that have similar folds and

statistically

insignificant

sim-ilarities.32

Hence,

_proteins

unrelatedin sequence and function

may

diverge

froma common

protein

ancestorwhile

_retaining

thesamebasic

polypeptide

fold. It has been

reported32

that2511

polypeptide

chains cluster into 212 amino acid sequence fam-ilies

(25%,

or_more,

identities)

and into

only

80

single

domain

polypeptide

foldfamilies. In consequence,aclassification has

been consideredthatextends the

_{sequence-based superfamilies}

toinclude

_proteins

with similar three-dimensionalstructuresbut

no_sequence

similarity.

One may alsonotethat nine

_superfolds

dominate the

_protein

database,

representing

morethan 30% of

all determined

structures.32

Often,

evolution obscures the func-tion.

The

_pathway

of the bacterial cell wall

_{peptidoglycan}

synthesis

shown in

_Figure

1is that of Escherichiacoli.It

_applies

toall bac-teria

_possessing

a wall

peptidoglycan.

From the MurA

UDP-N-acetylglucosamine enolpyruvate

transferase,4

which

catalyzes

the

first committedstepof the

_pathway

inthe

_cytoplasm,

tothe

peni-cillin-binding proteins

that assemble the

_polymer

from the

disac-charide-pentapeptide-lipid

IIintermediateontheouterface of the

membrane,

all the reactions are

bacteria-specific.

The

lipid

II

in-termediate is a

ß-\,

4-linked

iV-acetylmuramyl-iV-acetylglu-cosamine

disaccharide,

the

N-acetylmuramic

acid of which is sub-stituted

_by

a

D-alanyl-D-alanine-terminated pentapeptide

via a

D-lactyl-L-alanine

amide bondand theC-l atomis attachedtothe

intracellular end ofatransmembrane

undecaprenyl lipid

carrier via a

pyrophosphate. Lipid

IIisa

_key

intermediate. It

is,

atthesame

Centre

_{d'Ingénierie}

desProtéines,Université de

_Liège,

InstitutdeChimie, B6,B-4000 Sart Tilman

_(Liège

1),

Belgium.

Penicillin -* Receptor

vvc

/

rv

^f

"'' COOH Peptidoglycan , hydrolases luí Induction

/î-lactamase

_synthesis

/

Mia

. 1bl

jG

_{L-Ala-D-GlupL-Xaa-D-Ala-CO}

2-M M

/|

x I

_G

L-Ala-D-Glur—L-Xaa—O-Ala—CO-NH /

I—I

_\w

l-Ala-D-Glur-L-Xaa—D-AlaTCO-NH i KO

¡¡J^ij-assembty

(Tpase: Tglyase)

-S6-cell shape fc60 —cell septation »¿9 — recycling 42-cell _shape 32 -recycling UDP-MurNAclMI

L-Ala¿

D-Glu

—L-Xaa-'D-Ala-D-Alà'.

NH2 -"

ib-Ala-D-Ala/^

Ddl D-Ala+D-Ala MurF MurE L-Xaa

\u>

NH, MurD

L-Ala-LD-Glu-L-Xaa-'.p-Ala—D-Ala)

urA

k'Ren0lpyrUVate

NH2 '"--• UDP-GUNAc-enolpyruvate MurB ^NADPH MurC D-Glu L-Ala UDP-MurNAc

FIG. 1. Wall

_{peptidoglycan synthesis pathway.}

The PBP_patternshown is that ofE. coli in whichcasethe diaminoacidresidue

L-Xaais

_{meio-diaminopimelic}

acid.

_G,

_{N-acetylglucosamine;}

M,

N-acetylmuramic

acid;

Tpase,

transpeptidase;

Tglyase,

transg-lycosylase.

Thesites of

_cleavage

of

_{peptidoglycan hydrolases}

areshown:

1,

N-acetylmuramidase;

2,

AZ-acetylmuramoyl-L-alanine

amidase; 3,

endopeptidase.

The PBPsare inactivated

by penicillin.

Penicillin is

_hydrolyzed

_by

the

_{/3-lactamases.}

In some

bacte-ria,

/3-lactamase

synthesis inducibility

is mediated

_by

a _receptor

_(see

_Fig.

7).

The

PBPs,

the

_majority

of the

_{/3-lactamases,}

and the

_BlaR-type

_penicillin

_receptors

_belong

tothe

_superfamily

of

_penicilloyl

serine transferases.

time,

the

_product

of the

_"cytosolic"

stageand the substrate of the "wall"stageof the

_synthesis.

The Ddl D-Ala-D-Ala

_ligase,

the low-molecular-mass

peni-cill-inbinding

proteins

(PBPs),

and the

/3-lactamases

haveone

single catalytic

function. Some

_{peptidoglycan hydrolases,}

the

BlaR-type penicillin

receptors, and the

_{high-molecular-mass}

PBPsaremultimodular

_{polypeptides.}

Sets ofPBPs and

non-penicillin-binding proteins

associate into

_multiprotein

com-plexes

(not

shownin

_Fig.

₁₎

andform

_{morphogenetic}

networks. Thesesystemsof

_increasing

_complexity

are examined

succes-sively.

With few

_exceptions,

references are made

_only

to

pa-pers

published

from 1993 to 1995.

D-ALA-D-ALA

AND

d-ALA-d-LACTATE

(ATP:ADP

+

_P¡)

_LIGASES

In E.

coli,

the

_lipid

IIintermediateis formed

_by

_the sequen-tial addition of

L-Ala, d-G1u,

meso-A2pm,

and a

preformed

D-Ala-D-Ala to

_{UDP-/V-acetylmuramic}

acid

_by

the

MurC,

MurD, MurE,

and MurF

_adding

_enzymes,

_{respectively (Fig.}

_1).

TheD-Ala-D-Ala

_dipeptide

is

_{synthesized by}

theDdl

_adding

en-zyme. The MraY and MurG transferases

catalyze

the

attach-ment of the

_{A'-acetylmuramyl pentapeptide}

tothe

_lipid

carrier and the

_subsequent

addition of

_{A'-acetylglucosamine,}

respec-tively.

The

synthesis

of the D-Ala-D-Ala

_{dipeptide (Fig.}

₂₎

_begins

with the attachment ofafirstD-alanine residueonthe

y-phos-phate

ofadenosine

_triphosphate

(ATP)

to

yield

an

acyl

phos-phate,

followed

_by

attack

_by

the amino_{group of the second}

D-alanine residue to

produce

a tetrahedral

intermediate,

which

then eliminates the

_phosphate

group to

_give

the D-Ala-D-Ala

dipeptide.

The DdlB D-Ala-D-Ala

_ligase

of E. coli is made of

three

_domains,14

each folded arounda4- to6-stranded

/3-sheet

core,and the

ATP-binding

site issandwiched between the

/3-sheets

ofthe

_{carboxy-terminal}

andcentral domains. A helix

dipole

andthe

hydrogen-bonded

catalytic

triad

_{El5, S150,}

and

Y216 assist

_binding

and

_{deprotonation}

_steps.

The insertion of

_{lipid-transported,}

butas_yetnon-cross-linked

disaccharide

_pentapeptide

units,

inthe

_growing

wall

peptido-glycan

mustbe achieved

by transglycosylation

at the level of the

_glycan

chains and

_by

_{transpeptidation}

at the level of the

peptide

chains if the process is to

_yield

an insoluble network.

0>. ATP D-Ala AMP

Sl50'

--.

_H2N

_R255

-H3N

s

/-H'

,G276

D-alanylphosphate

"-"-'_R275 D-Ala

H^_r^:H-NH--.N

V216.

\ .

/^C°")hN-L2

s.H^C

y

,H0

S281

~°?n

H3N*

CH3

H2NX/CH3

0,P2~_{3 -0}

/f~-n

₀ -,__*

N^

*

HN:

)=o H-NH

I

hpo;

/

rC02 H3C

COJ

TI

H3C

D-Ala-D-Ala

FIG. 2.

Synthesis

of the D-Ala-D-Ala

_dipeptide

_by

the Ddl

(ATP:APP

+

_P¡)

_ligase.

isareaction in which the

_{carboxy-terminal D-alanyl-D-alanine}

moiety

ofa

pentapeptide

_precursorserves as

carbonyl

donor. The

_glycopeptide

antibiotics bind

_tightly

tothe

_dipeptide

moi-ety

ofthe

_{lipid-transported disaccharide-pentapeptide}

precur-sors,

preventing cross-linking.

Resistance to the

_glycopeptide

_vancomycin

in enterococci results from

changes

in the

_{peptidoglycan}

_biosynthetic

path-way.37

In VanA and VanB

_strains,

the

_dipeptide

D-Ala-D-Ala ofthe

_{peptidoglycan}

_precursoris

_replaced

_by

the

_depsipeptide

D-Ala-D-lactate. This

_change

doesnotlimit the

_activity

ofthe

transpeptidase

that

_catalyzes

_{cross-linking,}

but it results

_in,

at

least,

a1000-folddecreased

binding affinity

of

vancomycin

for

the

_{peptidoglycan}

_{precursor. The D-Ala-D-Ala and}

D-Ala-D-lac-tate

_ligases

_{have 30% of their sequences in}

_{common.14 Hence,}

conversion ofone

ligase

into theother

_requires

a

_great

extent

oflocal

_changes

but it doesnotalterthefold

_topology.

The

cru-tial E15 and S150 are

conserved,

butnotable differences also

occurin theactive

site,

themost

significant

one

_being

the sub-stitution Y216-H.K.

The DdlB D-Ala-D-Ala

_ligase

andthe

_y-glutamyl

cysteine-glycine ligase

(or

glutathione

synthetase)

couple

activation of

an

acyl

_{group and}

hydrolysis

of ATPintoADP and

_P¡

to

pro-vide the

_{thermodynamic}

_driving

force for

_peptide

_bond syn-thesis. The two

synthetases

_perform

different functions

(glu-tathione is the

_major

determinant of the oxidation-reduction state of the

cells).

They

lack amino acid sequence

similarity

(10% identities).

Yet

_they

showaremarkable fold

similarity.15

Theircommon

signature

fold and

catalytic

site may be

charac-teristics of a

particular superfamily

of

ADP-forming peptide

synthetases.

The

_{MurC, MurD, MurE,}

and MurF

_ligases

and the

_{y-glutamic acid-cysteine ligase}

(the

reaction

_product

of which is the substrate of the

_glutathione

_synthetase)

_also per-form

_peptide

bond formation with concomitant

_hydrolysis

of ATP into ADP and

_P¡.

_They

_might

be other members of the

same

superfamily.

MONOFUNCTIONAL PENICILLIN-BINDING

PROTEINS AND

_{j3-LACTAMASES}

Serine-assisted

_{transpeptidation}

between a

D-Ala-D-Ala-ter-minated

_pentapeptide

precursor

acting

as

carbonyl

donor and

the_{w-amino group of the}L-Xaaresidue of another

_peptide

act-ing

asamino

_acceptor

doesnot

_require

an

input

of energy

and,

therefore,

can resultin

_peptide

bondformation atexocellular siteswhereATPisnotavailable.

The

_{transpeptidation}

reaction

_requires

a

_precise

_proton

ab-straction-donation

_(Fig.

3).

Instep

1,

the C-terminal D-Ala-D-Ala

_dipeptide

_moiety

ofa

pentapeptide

_precursormustbindto

the _{active site of the enzyme in}a

position

that allows the

pro-ton ofthe

_yOH

of the active-site serine

(S*)

to be

abstracted,

theactivated

_OyS*

toattackthe

carbonyl

of the D-Ala-CONH-D-Ala scissile

_bond,

and the abstracted protontobe back-do-nated tothe

_{adjacent nitrogen}

atom. In step

2,

the serine

(S*)

ester-linked

_peptidyl

enzyme must

_adopt

aconformation that

allows the _proton of_{the &j-amino group of the L-Xaa residue} ofanother

_peptide

tobe

abstracted,

the activated

ÑH

toattack the

_carbonyl

of theester

bond,

and the abstracted_protontobe

back-donatedtothe

OyS*

atom._{Backbone amino groups of the}

enzyme

cavity

(denoted

E-NHin

_{Fig. 3)}

_polarize

the

_carbonyl

of the D-Ala-D-Ala

_peptide

bond in

_step

1 and the

_carbonyl

of the

_peptidyl

_enzymeesterbond in

_step

2.

Because the

_dipeptide

D-Ala-D-Ala

(in

its extended

confor-mation)

and

_penicillin

are

nearly

isosteric,

the

_{transpeptidase}

also reacts with

_penicillin.

But because the scissile

/3-lactam

amidebond is

_endocyclic,

the serine

(S*)

ester-linked

penicil-loyl

enzyme is very

long

lived. The

transpeptidase

is inacti-vated and behaves as aPBP

(Fig. 4).

An

_evolutionary

scenariohasbeen

_proposed18

_through

which

acquisition

of new functions from a

putative

DD-transpepti-dase/PBP

ancestorisachieved

by

local

_{changes (Figs.}

3 and

4).

Catalyzed hydrolysis

of theesterbond of the

_peptidyl

_enzyme with conservation of the inertness of the

_penicilloyl

_enzyme

gave rise to the monofunctional

DD-carboxypeptidases/PBPs.

They

may control the extent of

_{peptidoglycan cross-linking.}

Conversely,

catalyzed hydrolysis

of the

_penicilloyl

_{enzyme with} lossof

_{peptidase activity}

_{gave rise}tothe

defensive,

penicillin-hydrolyzing ß-lactamases.

The monofunctional PBPs and the

_majority

ofthe

/3-lacta-masesknown

today

are

acyl

serinetransferases.

They

fall into

several amino acid sequence classes

Similarity

among

mem-bers ofa

given

class

(i.e.,

intraclass

_similarity)

forms a

con-tinuum,

the cut-off

_points

_being

> 20% identities. Interclass

similarity

is almost nonexistent.

_Similarity

isnot

_always

related

tofunction. There ismore

similarity

between the

Streptomyces

Carbonyl donor TI Peptidyt enzyme R~-D-Ala-C-NH-D-Ata-COO" n" 0

R~-D-Ala-C-vOr-5*-E

Ammo x_nh acceptor -R~D-Ala-C=01'-S~E Water HOH

R—D-Ala-C-NH —D-Ala-COO" R~~D-Ala-C-0)-S-E

/ N *v. " .0. oí- X o H' ~H

I

I R~-D-Ata-C-OI'-S-E -0. NH-X N N D-Ala Product R^D-Ala—C-NH-X 0 DD-transpeptidase R^D-Ala-C-OH II 0 DD-carboxypeptidase

E-S^OI'H

FIG. 3.

_Acyl

transfer reactions on

D-alanine-D-alanyl-terminated peptides

via formation ofa serine-ester-linked

peptidyl

en-zyme. Attack of the

peptidyl

enzyme

by

anamino_acceptorleadsto

_{transpeptidation}

of the

carbonyl

donor. Attack

by

waterleads

to

hydrolysis.

TI,

terahedral intermediate.

between the

_Streptomyces

R61

_{DD-carboxypeptidase/PBP}

and the class C

_{/3-lactamases}

thanbetween theclass Aand the class C

_{/3-lactamases.}

The K15

_PBP,

the R61

_PBP,

andseveral class A and class C

_/3-actamases

are oftwo domains

_(Fig.

5).18

One domain is of a-helices and the other domain is a five-stranded

/3-sheet

corethat is covered

by

additional a-helices. The active site that

is sandwiched between thetwo

domains,

isadense

hydrogen-bonding

network

_{interconnecting}

watermoleculesand the side

Carbonyl donor Tl Pencilloyl enzyme

COO" 0=_{C tL} y _o—c^7 O_su 0?

\H

I H HOHA\ H I COO" I I _/SZc C00" HOH OH C-H_I C00" E-S-OiH

/}-

lactamase

FIG. 4.

_Acyl

transfer reactionson

penicillin

viaformationof

aserine-ester-linked

penicilloyl

_{enzyme. With the}

PBPs,

the

re-action

_stops,

atleast fora

long

_time,

at the level ofthe

peni-cilloyl

enzyme. With the

/3-lactamases,

the reaction

proceeds

to

_hydrolysis

of

_penicillin.

chains of amino acid residues at the

_boundary

of the

_cavity.

This fold

_topology

accommodates_many local structural varia-tions. Withx

denoting

_anyamino acid

residue,

the tetrad S*xxK

is located

_centrally

atthe amino end ofana-helix of the all-a domain. The triad

_{[K/H] [T/S]G}

isonthe innermost

/33

strand

ofthe

_/3-sheet

ononeside of the

_cavity,

and the triad SGC

(the

K15

PBP),

YxN

(the

R61 PBP and the class C

/3-lactamases),

orSDN

(the

classA

/3-lactamases)

ison a

loop

connecting

two

helices ofthe all-a domainonthe otherside of the

cavity.

The

classA

/3-lactamases

haveanadditional active-site

defining

mo-tif,

the

_pentapeptide

ExELN,

locatedattheentranceof the

cav-ity

nearthe bottom of the

_/3-3

strand.

_Compared

with the K15

PBPand the class A

/3-lactamases,

the R61 PBP and the class C

_{/3-lactamases}

have additional

_loops

and

_secondary

structures

away from the active site.

In

spite

ofdifferences in

function,

the monofunctional

dd-peptidases/PBPs

and

_{/3-lactamases}

have retained much of the

same fold and much of the same active-site

signature

in the form of the motifs

S*xxK,

[S/Y]xN

or

analogue,

and

[K/H/R][T/S]G.

They

alsohaveretained much of thesame

cat-alytic machinery.

They

each

_catalyze

_rupture

of the

/3-lactam

amide bond with transfer of the

_carbonyl

carbon tothe serine

(S*)

residue and formation ofa serine ester-linked

penicilloyl

enzyme. On the basis of thiscommon_property,

they

forma

su-perfamily

of

_peniciloyl

serine transferases.

_They

also

_catalyze

acyl

transferon

acyclic

carbonyl

donors

R1

-CONH-CH(R2)-COX-CH(R3)-COOH

where X is

NH,

O,

orS. The substituents

Rl, R2,

and

R3,

the natureof the scissile

_(peptide,

ester,

thiolester) bond,

and the reaction

_pathways

areclass- and

_{enzyme-specific.5-21}

de-pend

onthe accuracy of fit of the

ligands

(peptide,

_ester,

thio-lester,

/3-lactam

carbonyl

donors;

amino

_acceptors)

in the

en-zyme

cavity. Catalysis

also

depends

onthe

efficacy

with which

amino acidresidues oftheactivesitefulfill the

_required

func-tion of

_general

base

_{catalyst (abstracting}

the _proton of the

yOHS*

instep1 and that ofwateror anaminoacceptorinstep

2)

and

_provide

an

itinerary through

which the abstracted pro-toncanbe back-donatedtothe

_right

atomsin each _stepof the reaction.

Anextensive

_study

of the PBPs and

_{/3-lactamases}

of known three-dimensional structure

by

site-directed

mutagenesis

and molecular

_modeling

has failed to

_identify,

with

_certainty,

the

routethat the

_proton

uses

during

catalysis.17

At this level of the

investigation,

10~10

m,

_quantum

effects rule the nanoworld and

the_protonshuttlecanbedisclosed

_only

_by

the_{methods of}

quan-tum

_chemistry.

Such methodsare

being developed.

Studies

car-riedouton

chymotrypsin

have ledtotheconceptthat the

_charge

relay

ofan

acyl

serine transferase is

created,

de novo,

by

the

interacting

partners,the_enzyme,and the

_ligand.6'7

The creation of the

_{charge relay}

results from the combined effects of the

ac-tive-site

environment,

the deformation

_{undergone by}

the bound

ligand(s),

the relaxation

_{undergone by}

_{the enzyme}

_polypeptide

backbone,

and the freedom ofone orseveral watermolecules.

Hence,

enzymes ofasame

family

or even a sameclasscan use morethanone

proton-shuttle

route

_depending

onstructural fea-tures of the active site

and/or

the bound

_ligand.

The

_naturally

occurring /3-lactamases

of classes

_{A, C,}

and D and theeasewith

which

_|8-lactamase

mutants _{emerge among} clinical isolates

supportthis

_concept.

In recent_years,atleast 26

_{ß-lactamases}

of

_{varying specificities}

have been identified.

_They

each arose

by

alteration of amino acid residues in the class A TEM-1

ß-lactamase. Evolution is

_occurring

beforeour_eyes.

In

conclusion,

the monofunctional

_penicilloyl

serine

trans-ferases have evolved and arestill

evolving

with

preservation

ofa characteristic

signature

fold and much of the same

ser-ine-assisted

_acyl

transfer

_{machinery. They}

illustrate the

prin-ciple according

towhich evolution obscures the function. The

catalytic properties

ofamember of the

superfamily

cannotbe deduced from its amino acid sequence and even fold

topol-ogy. Direct biochemical evidence is

required.

Finally,

the monofunctional

_penicilloyl

serine transferases are

highly

adaptable

structures._{The essential serine residue may be}

ac-tivated

_by

different

_general

bases and several proton shuttle

routesmay be used.

KTGS

KTGS EPELN

FIG. 5.

_Peptide

fold of monofunctional

_penicilloyl

serine transferases. The

Streptomyces

K15

_{DD-peptidase/PBP}

functionsas a

transpeptidase

on

D-alanyl-D-alanine-terminated

peptides.

The

Streptomyces

R61

DD-peptidase/PBP

functions

mainly

as a

car-boxypeptidase.

The E. coliTEM-1

_/3-lactamase

andE.cloacae P99

_/3-lactamase

aremembers of the amino acid sequence class

Aand class

C,

respectively.

The active-site

defining

motifsare indicated. For

references,

see

Ghuysen.18

The atomic coordinates of the K15PBPwill be

_published

shortly.

MULTIMODULAR WALL PEPTIDOGLYCAN

HYDROLASES

MonofunctionalPBPsof E. coliare, atthesame

time,

DD-car-boxypeptidases

and

_{peptidoglycan}

_{hydrolases. They}

_hydrolyze

the

carboxy

terminal

_{D-alanyl-D-alanine peptide}

bond

_(made

_by

the D-Ala-D-Ala

_ligase)

of the

_pentapeptide

_precursors.

_They

also

hy-drolyze

the

_carboxy

terminal

_{D-alanyl-(D)-meio-diaminopimelic}

acidbond

(made

by

the

_{transpeptidase)}

that_{cross-links the} pep-tide unit in the

_completed

_{peptidoglycan.}

Streptomyces

albus Gsecretes a

non-penicillin-binding

dd-carboxypeptidase/peptidoglycan

hydrolase.

Thismétallo

(zinc)

enzyme is constructed oftwomodules

_(Fig.

6).19

Thearrow on

the

_{right points}

toward the

_{zinc-containing}

active siteborne

_by

the 132 amino acid residue

_{carboxy-terminal,}

_{catalytic (C)}

mod-ule.Thearrow onthe left

_points

toward the

_cavity

borne

_by

the

81 amino acid residue

_{amino-terminal,}

_noncatalytic

_(n-C)

mod-ule. The crevice

(18.6

Â/13.5

À)

of the n-C moduleis defined

by

two a-helical _repeats

_(a2

and

_a3)

each 16 amino acid

residues

_long,

connected

_by

a

heptapeptide loop.

The n-Cmodule of theZn

_DD-peptidase

is the

_prototype

of

an_{amino acid sequence}

_family

ofn-C modules also foundin the Bacillus subtilis CwlA and B.

_{licheniformis}

CwlL

A'-acetyl-muramoyl-L-alanine

amidases and the

_{Corynebacterium}

aceto-butylicum N-acetylmuramidase (lysozyme). Similarity

between the n-C modules is

_{high (30-50% identities),}

_{and, therefore,}

one canbe confident that

they

have the same fold. However

(with

IP

_denoting

an

intervening peptide)

themodular

_design

ofthe

_{peptidoglycan hydrolases}

is different:

NH2-[n-C]-[C]-COOH

for theZn

_DD-peptidase

NH2-[C]-[n-C]-COOH

forthe CwlA amidase

NH2-[C]-[n-C]-[n-C]-COOH

for the

_lysozyme

NH2-[C]-[n-C]-IP-[n-C]-COOH

for the CwlL amidase

Depending

onthe_enzymes,the n-C moduleoccursatthe amino

orthe

_carboxy

end of the C module inone ortwo

copies

and the

_copies

are either

_contiguous

orconnected

by

an

interven-ing peptide.

Acquisition

ofa

substrate-binding

module fused tothe

cat-alytic

module is an

evolutionary advantage

forexocellular

en-zymes that interact with and

_hydrolyze

bonds in a

polymeric

substrate.

_{Peptidoglycan}

_hydrolases

have achieved this feat

FIG. 6.

_Peptide

fold of the bimodularZn

DD-carboxypepti-dase/endopeptidase

of

_Streptomyces

albus G.

_Reproduced

from

Ghuysen

et

al.,19

with

_permission

ofElsevier.

through

the

_interchange

and local structural _{alterations of} spe-cialized modules. The

_likely

functionof these modulesas

sub-strate

_{recognition/binding}

sites would

_{depend mainly}

on the

conserved amino acid residues. Substrate

_specificity

and di-rected

_{topological activity}

would

_depend

ontheoccurrence of

nonconserved amino acid residues and the _{location and copy}

number of then-C modules.

BIMODULAR PENICILLIN

RECEPTORS

Thebacterial cellsare sensitive to

virtually

_everyaspect of their environment and theseenvironmental

_changes

are

moni-tored

_by

_specialized

_sensory transducers. The dominantforms of

_signal

transduction

_proceed

via

_phosphoryl

transfer

_pathways

(the

so-called

_{two-component}

_regulatory

_systems)

or are

asso-ciated with sitesof

_methylation

and

_{demethylation.}

Bacteriahave

_{developed unique}

transduction

_pathways

that detect

/3-lactam

molecules in the environment of the cell and switch on the

transcription

of the

_{/3-lactamase encoding}

_gene. Induction of

_{/3-lactamase synthesis}

in a _{number of}

gram-nega-tive bacteria istheresult of

_/3-lactam

antibiotic-induced and

pep-tidoglycan hydrolases-mediated deregulation

of the cell wall

re-cycling

process.26-27

Inthe

_{gram-positive}

Bacillus

_{licheniformis,}

the inductionof

_{/3-lactamase synthesis}

rests_{upon the presence in}

the

membrane,

ofa

penicillin

_receptorthat results fromafusion

event

through

whicha

penicilloyl

serinetransferase

_polypeptide

islinkedtothe

_carboxy

endofa

signal transducer.29

Regulation

of

_{transcription}

of the

_{/3-lactamase-encoding}

blaP gene inB.

_{licheniformis}

involves three-chromosome-borne

regulatory

genes, blal encodesa 15-kDa repressor; blaRl

en-codesa

penicillin

_receptor;the

product

of blaR2 is of unknown

function. Membrane

_{topology experiments, predictional}

stud-ies,

andconformational

_analyses29

_(unpublished

data from this

laboratory

andR.

Brasseur)

strongly

suggestthat the 601amino acid residue blaR1 -encoded

_penicillin

_receptor BlaR has the

multipartite

organization

shown in

_Figure

7. Central to this model is afour a-helix bundle defined

by

four transmembrane

segmentsTM1 toTM4. A 63 amino acid residue extracellular domainconnectsTM2 and TM3.A189amino acid residue in-tracellulardomain_{that possesses the}

_signature

ofa

metallo-pep-tidase

(Zn2+

_binding

site)

connects TM3 and TM4. A 261 amino acid residue extracellular

domain,

the

_penicillin

sensor,

is fusedtothe

_carboxy

end ofTM4.Thissensor_{possesses the}

active-site

_defining

motifs

(S*TYK,

YCN, KTGT)

of the

peni-cilloyl

serinetransferases

_superfamily.

The sensor canbe

pro-duced

_{independently}

from the restofBlaRas a water-soluble

polypeptide

in the

_periplasm

ofE. coli. Theisolated

polypep-tide binds

_penicillin

and behavesas a

high

_affinity

PBP.

On thebasis of this

model,

a

likely

andtestable mechanism of

_signal

transmission

_by

_{BlaR may beput}forward. Penicillin-inducedconformational

_changes

in the

_{penicillin-bound}

sensor

and the

_interacting

63 amino acid residue extracellular domain would be transmitted via the four a-helix bundleto the intra-cellular domain withconcomitant activation of the

_putative

met-allopeptidase. Degradation

of the Blal repressoror release of an

antirepressor

in the

_cytosol

_by

the "activated"

_peptidase

would result in

_derepression

of

_{/3-lactamase synthesis.}

BlaR is the

_prototype

ofanamino acid sequence

family

of

Sensor

K539TG Y476GN

S*¿02TYK

601-_I_I_L

Zn

_binding

site

Signal

emission

FIG. 7. Schematic

_{representation}

ofthe bimodular

penicillin-receptor

BlaR of B.

_{licheniformis.}

The

_penicillin

sensor

belongs

to_{the amino acid sequence class}D

_{/3-lactamases.}

and

_low-affinity

PBP2'

synthesis

inS.

aureus.29

The S342-R601

polypeptide

sensorof BlaR

is, itself,

amember of the amino acid

sequence class D

/3-lactamases (32%

identities withtheOxa-2

/3-lactamase),

suggesting

a common

signature

fold.

Hence,

a

peni-cilloyl

serine transferase ofa

given

class

(a

/3-lactamase)

_may

ac-quire

a new _property

(penicillin sensing)

through

local

changes

and fusiontoanother

_polypeptide,

andthe

_resulting

chimeric pro-tein may

acquire

a newfunction

(gene

regulation).

MULTIMODULAR

PENICILLIN-BINDING

PROTEINS

The

_{high-molecular-mass}

PBPsarealso of modular

design'2

(Table

1,

Fig.

8).

The amino acid sequence dataare from the

literature.2,12'34

A

_{penicillin-binding}

(PB)

module that

belongs

to _and

pos-sesses the

S*xxK,

SxN,

and

K[T/S]G

markers of the

penicill-loyl

serine transferases

_superfamily

is fusedtothe

_carboxy

end ofa

non-penicillin-binding

(n-PB)

module ina

single

polypep-tide chainthatfoldsontheouterface ofthe

_plasma

membrane. The

_polypeptide

itself is fused to an amino-terminal

trans-membrane-anchoring

module. Inserts_mayoccurthatare

_large

enough

toform additional modules.

By analogy

with the

mono-functional

PBPs,

the PB modulesareassumedto startabout 60 amino acid residues

_upstream

fromthe S*xxK motif andto

ter-minate about 60 amino acid residues downstream from the

K[T/S]G

motif._{The n-PB modules also possess}

_specific

amino acid

markers,

but the

_signature

of the n-PB modules of the class

Amultimodular PBPs is different from that of the n-PB mod-ules of theclass B multimodularPBPs.

Asa

polypeptide

chain increases in

length,

finding

the

right

fold becomes more difficult because the

possibilities

of

mis-folding

increase.

Study

of derivatives of multimodular PBPs

overproduced

from

_{appropriate expression}

(and secretion)

vec-tors shows that the

_acquisition

ofa

stable,

penicillin-binding

fold

_topology

_by

the PB module is membrane-anchor

indepen-dent but

_requires

concomitant

_biogenesis

of then-PBmodule. As shown with the E. coli

PBPlb31

and

PBP316

_(unpublished

data),

the S. aureus

PBP2',4041

and the E. hirae PBP5

(un-published

data),

replacement

of the membrane anchor

by

a

cleavable

_{signal peptide}

orsubstitution of the

genuine

anchor

by

another transmembrane

_anchoring

device has noeffect on

the

_{thermostability}

and

_{penicillin-binding capacity}

of the PBP

mutants.

_However,

elimination of both the membrane anchor and the n-PB module

_(or

partof

it)

orelimination of then-PB

module with conservation of the membrane anchor isnot

tol-erated.

Moreover,

expression

of E. coli

_ftsl

genes

encoding

PBP3 mutants in which El93 of motif 3* of the n-PB module is

_replaced

by

DorN

_gives

rise tomembrane-bound

_proteins

that are _very

unstable,

suggesting

that motif 3*

plays

arole in

the

_folding

process

(unpublished

datafromthis

laboratory

and

J.

_Ayala).

Intraclass

_similarity

between then-PB modules and the PB

modules,

respectively,

is acontinuum with acut-off

point

of

about 20% identities

(Table

2).

Interclass

_similarity

between the

n-PBmodules is nonexistent and interclass

similarity

between the PB modules is_{very low.} The PB modules of the classA PBPsandthoseof classBPBPs have

diverged

sofarthattraces

of

_similarity

other thantheactive site

_defining

motifshave

al-most

_{completely disappeared.}

Then-PB modules of the class

A PBPs and those of the class B PBPs formtwodistinct

fam-Table 1. Sizeofthe MultimodularPBPs in NumberofAmino AcidResidues andPosition of theActive Site Serine S* ClassA ClassB S* B. subtilis

la/b

H.

_influenzae

la E. coli la E. coli lb M.

leprae

1 S. aureus2x S.

pneumoniae

la B.subtilis la S. oralis la 10. B.subtilis 4 914 864 850 844 821 727 719 714 637 624 390 452 465 510 398 398 370 359 371 388 1. S.

_pneumoniae

2x 2. S.

pneumoniae

2b 3. E. hirae

5,3r

4. S.aureus 2' 5. E. coli 2 6. E. coli 3 7. N.

_gonorrhoeae

2 8. N.

_meningitidis

750 680 678 667 633 588 582 582 337 385 422 405 330 307 312 310

ClassA Class B

7) KTG

190215:E.coli1a; H inftu.la

SxN

(5)

S*xxK

225-238:

_H.influ.1a;

E.cotila O

_Rx3xL

3)

RKx2ExxxxL

p

G[A/G][S/T]Txx,2Q

l)

EDx2,Fx2HxG

233:_{E.coli 1b} PB module n-PB

(cooh)

36-68 173: S. pneumo 2x

9)

Px2[N/Q][P/G)

25-30

(6.

135-157 172: S.pneumo. 2b SxN 52-59 34-50

(5) TGtE/D/G/*lx6[T/S/H]Px2D

31-45

(4<9[D/N]x3lT/S]x[D/Slx3Q

43-46

(3*)

Gx2GxEx3(E/D/Nl

module _2i-42

(2»)

Rx2PxG

90-110 162: S. pneumo.2b

(]»)

RGx3DR[N/s][G/N]x3

A 151-175: S.aureus 2' E.hirae 5,3r Membrane 6o~76 NH,

FIG. 8.

_Design

_{and amino acid sequence}

_signatures

of the multimodularPBPs of classesAand B. Thedistribution of the

con-served

motifs,

the average

length

of the_{intermotif sequence, and the} occurrenceof inserts

_(expressed

innumber of amino acid residues

_aa)

areshown. The data derive from the amino acid sequences of the PBPs listed in Table 1.

S*,

active-site

serine;

x,

variableaminoacid

_residue;

x,

hydrophobic

amino acid residue.

Table 2. Identities

(%)

between theAligned Amino Acid

Sequences

ofHigh-Molecular-Mass PBPsa n-PB module PBP Class A 1. M.

_leprae

1 2. E. coli la 3. H.

influ-enzae la 4. E. coli lb Class B 5. E.coli2 6. E. coli 3 7. E. hirae 3r 8. E. hirae 5 9. S. aureus 2' Class A 100 26 27 24 100 57 100 22 24 100 Interclass PB module Class A 100 17 16 18 100 52 100 24 24 100 Interclass 10 11 12 11 10 10 11 12 12 11 11 14 13 13 11 10 12 10 10

10

100 Class B 24 28 17 21 20 100 85 100 100 23 17 39 40 100

ilies. The two families have a characteristic amino acid

se-quence

and,

presumably,

fold

signature,

suggesting

that

they

have evolved from different

_polypeptide

ancestors.Asann-PB

module of class A is linkedtoaPBmodule of class Aandan

n-PBmodule of class Bis linkedtoaPB module of class

_B,

a

likely

corollary

of this

_{class-specific}

modular

_design

is that the class A PBPs

_perform

a different function

(or

different

func-tions)

from that

(those)

of the classB PBPs. In _consequence,

the effects of the inactivation ofthe class A PBPs onthe cell

viability

should be different from those of the classB PBPs. In

vitro,

the

_purified

class A PBPla and PBPlb of E. coli

(and,

perhaps,

by

extension other classA

PBPs)

are wall

pep-tidoglycan synthetases.

They catalyze

glycan

chain

_elongation

and

_peptide

_{cross-linking}

from the

_lipid

II intermediate.

Inhibition of the

_{n-PB/transglycosylase}

module ofPBPlb

_by

moenomycine

prevents

peptide cross-linking

whileinactivation of the

_{PB/transpeptidase}

module

_by

_penicillin

enhances

_glycan

chain

_elongation,

_showing

that thetwomodules interactwith each other.

Moreover,

PBPlb containsamembrane association

site in addition to the transmembrane

anchor31

and dimeric forms of PBPlb are in close association withthe

peptidogly-can.42

Invivo,inactivation of thePBmodules ofPBPlaandPBPlb

of E. coli

_by

j3-lactam

antibioticscauses

peptidoglycan

hydro-lases-induced cell

_lysis.

Deletion of the genes

encoding

PBPla and PBPlb is

fatal,

but deletion of either of the PBPla-orthe

PBPlb-encoding

geneis

tolerated,

suggesting

thatonePBPcan

compensate

for another.

In

vitro,

the

_purified

classBPBP3 of E.

coli21

_(unpublished

data)

and PBP2x ofS.

_pneumoniae29

catalyze

serine-assisted

hydrolysis

and

_aminolysis

of thiolester

_carbonyl

donors. Bacterial strains

_having

a reduced

affinity

for

penicillin

and

possessing

one or several classB PBPs with reduced

_affinity

for the

_{drug synthesize}

a wall

peptidoglycan

with a different

peptide moiety

from that of the wild

_type.36

Hence,

thePB

mod-ules of the class B PBPs are

involved,

one_way or

another,

in

peptidoglycan

cross-linking.

However,

the

_precise

natureof the

catalyzed

reactions remains tobe elucidated.

Indeed,

the iso-lated PBP3 ofE.coli is inertonthe

lipid

II

intermediate1

(un-published

results from this

_laboratory

and J. Van

_Heijenoort).

Moreover,

evolution may obscure the function ofa

protein

and

fusion between

_polypeptides

_{may result in the}

_acquisition

ofa new function

(see

preceding

sections).

Given that then-PBmodules oftheclassB PBPshavea

dif-ferent amino acid

signature

fromthatof the n-PB modules of the classA

PBPs,

they

maynothavea

transglycosylase

activ-ity.

The

_acyl

transferase

_activity

of their PBmodules

might

be

coupled

with the

_{transglycosylase activity}

of the n-PB modules ofsomeclass A PBPs. Such a situation

implies

that the class

A and class B multimodularPBPsinteract with eachother. One

may also note that the E. coli PBP3 forms dimers

_(personal

communication from N.

_Nanninga

andJ.

_Ayala).

In

vivo,

the

_primary

effects of the inactivation of the PB mod-ules of the classBPBP2 and PBP3 of E. coli

_by

_/3-lactam

an-tibioticsare

morphological

_{abnormalities,}

followed

by

cell

ly-sis. Inactivation of PBP2 results in

_growth

as

spherical

cells

(the

rod-shaped

maintenance

_machinery

is no

longer

func-tional).

Inactivation ofPBP3 results in

_growth

as filamentous

cells

(the

septation machinery

is no

longer functional).

Inactivation of either of thePBP2-or

PBP3-encoding

_{genes of}

E. coliand of the PBP2b-or

PBP2x-encoding

_{genes of S.}

pneu-moniae isnottolerated.

Like the monofunctional PBPs and

_{/3-lactamases,}

the multi-modular PBPs are

highly adaptable

structures. Multimodular PBP-mediated resistanceto

/3-lactam

_{antibiotics among}

gram-positive pathogens

has becomea serious health

problem.

In S.

_{pneumoniae,i0,23,3S}

the first PBPs tobe affected

dur-ing

selection of

_laboratory

mutants

_having

a reduced

affinity

for cefotaxime and

_piperacillin

are the class B PBP2x and

PBP2b,

respectively.

The modified PBP2x and PBP2b confer resistance upon transformation. Low

affinity

forms of PBP2x and PBP2b and the class A PBPlaare

_present

in

highly

resis-tantclinical isolates. Reduced

affinity

is the result of structural

changes affecting

either a limited number of amino acid

residuesor

large

blocks of amino acid sequences of thePB

mod-ules.

_Interspecies

recombinationaleventsoccurthat

replace

part

ofa

PBP-encoding

_{gene with}the

corresponding

partsfrom

ho-mologous

PBP-encoding

genes of

closely

related,

naturally

re-sistant

_species.

The

_low-affinity

class BPBP2' of

S.

aureus3

and PBP5 and PBP3r ofE.

hiraeu'33

allow the strains that

produce

them to

grow and manufacture a wall

peptidoglycan

under conditions

in which all the monofunctional PBPs andthe PB modules of all the other multimodular PBPs areinactivated

by penicillin.

The

_{staphylococcal}

PBP2' is chromosomal

and,

insome

strains,

its

_{penicillin-induced synthesis}

is

BlaR-mediated.29

The

ente-rococcal PBP5 is also chromosomal but its

_homologue

PBP3r is

_plasmid

borne in E. hirae S185R

_(unpublished

data).

This

plasmid

carries,

at

least,

two

_copies

of the

_{PBP3r-encoding}

gene,one_{copy of the}

streptomycin-resistance

markerstr,one

copy of the

erythromycin-resistance

marker erm, and several

copies

oftheinsertion module

IS/276,

asituation

eminently

fa-vorablefor the

_spread

of multiresistance_amongenterococci and related bacterial

_species.

Asthe

_origin

of the

_low-affinity

classB

PBPz, PBPs,

and PBP3r is

unknown,

the structural features

_responsible

for the low

_affinity

are also unknown.

Compared

tothe other classB

multimodular

PBPs,

the

_low-affinity

PBPshave anextended

n-PB module because of the presence ofa«100 amino acid

residue insert

_immediately

downstream from the membrane anchor. These inserts are related in amino acid sequence.

Presumably, they

have a samefold.

Expression

of genes

en-coding low-affinity

PBP derivatives in which the insert is

missing

or truncated

gives

rise to

_proteins

that are inert in terms of

penicillin-binding41

(unpublished

data),

showing

that the insert

_plays

arole in

_biogenesis.

The active-site

topol-ogy of the PB module confers resistance to the usual

penicillins

and

_{cephalosporins}

butnot

_necessarily

toall

/3-lac-tam

_{compounds. Cephalosporin}

derivatives are

_being

devel-oped

that are active

against

methicillin-resistant S. aureus

strains.22

A

_{plausible picture}

that arises from the above

_analysis

is that the enzyme activities

required

tobuild the

_{peptidoglycan}

poly-merwould be

_{provided by}

the

n-PB/transglycosylase

modules of the class A PBPs and the

PB/acyl

transferase modules of both class A and class B PBPs.

_Regulation

ofthese activities in a

cell-cycle-dependent

fashion would be mediated

by

the

n-PB modules oftheclassB PBPs. Recent

genetic

studies have

brought

to

_light

the existence of

_{morphogenetic}

networks. These networksare

multiprotein

complexes,

theconstitutive elements

ofwhicharesetsofmultimodular

_PBPs,

monofunctional

PBPs,

and

_{non-penicillin-binding}

_proteins.

MORPHOGENETIC NETWORKS

Figure

9 shows _{partof the gene}

_organization

in the

chro-mosomeof E. coli. The 4 and 75 min

_regions

contain the genes

encoding

the

_bienzyme

_{(transglycosylase/transpeptidase)}

class A PBPlb and

PBPla,

respectively.

The 2 min

_region,

_{which contains the gene}

_encoding

the mul-timodular classB

PBP3,

also contains genes that encode other

cell-septation

proteins.1-9

Figure

10 shows the cellular

local-izationofsome_componentsof the cell

septation

network

(also

called divisómeor

septator). FtsQ

is of unknown functionand

FtsL containsa leucine

zipper

motif. Each isontheouterface

of the

_plasma

membrane. FtsW is an

integral

membrane

pro-tein.

FtsA,

an

ATP-binding

protein

of the

eukaryotic

actin

fam-ily,35

is

_exposed

on the inner face of the

_plasma

membrane. MraW is a

protein bearing

an

S-adenosyl methionine-binding

motif and FtsZ isa

GTP-binding/GTPase protein.

Each is

cy-toplasmic.

FtsZ contains a short

glycine-rich

_segment

that is

strikingly

similarto the

_GTP-binding

motif ofthe

_eukaryotic

cytoskeletal

tubulin.13

About

10,000

molecule of FtsZoccur

percell.

_They

forma

ring

at the future division

site39

in the form ofa

*=AO-pm

filamentthat is

long enough

tosurround the

IN

^©0/

FIG. 10. Cellularlocalization of

_proteins

of the "cell

septa-tion"

_{morphogenetic}

network inE. coli.

cell 20 times. These genes form a cluster the

expression

of

whichis controlled

_by

a

gearbox

and metabolic_promoters.

Genes located outside the 2min

_region

of the chromosome also encode "cell

_{septation" proteins.}

SulA isacell-division

in-hibitor.

_ZipA

is another

_integral

membrane

_protein.

Slt-70isan

/V-acetylmuramidase

that

catalyzes

the

_hydrolysis

of the

glyco-sidic bond with transfer of the

_carbonyl

to the C-6

_hydroxyl

group,

yielding

a

(

1

_{-6)anhydrornuramic}

acid.25

FtsK possesses

probably

an N-terminal domain with several

membrane-span-ning

helices anda

large cytoplasmic

domain withan

ATP/GTP-r-Lipid

II intermediate

_synthesis

-i

2min

region

MurE MurF MurD

MraY

Î

MurC MurG Odl mraZ mraW

IT

mr4

T

F1st murE murF mraY murD ftsW tnurG MraZ MraW FtsL

PBP3

~1T~

FtsW

-Cell

_septation

murC ddl flsO ftsA UftsZ

~i

i

r

FtsQ FtsA FtsZ

Cell

J

_division

imin 1 ¿, min

region

21-22mm

region

52min 75 min 99.7 min

pond decA mrdd mrdA

1

RodA ftsK

3-

sulA FtsK SulA zipA ZipA

PBPlb

PBP5

PBP2

I—Cell-11—Cell—1

Cell

shape

'division septation

Peptidoglycan

assembly

ponA sit Y

r

SIt 70

PBPla

Lytic

Tglyase

Peptidoglycan

assembly

FIG. 9.

_Organization

_{of the genes in} thechromosome ofE. Coli involved in

_lipid

IIintermediate

_{synthesis, peptidoglycan}

as-sembly,

cell

_septation,

cell

_shape,

and cell division.

_ponA

and

_ponB

arealso called mrcA and

mrcB,

respectively. Tglyase,

binding

motif

_(L.

_Begg,

S.

Dewar,

and W.

_{Donachie; report}

presented

atthe

_workshop

"Structure,

Function andControlsin

Microbial Division." InstituteJuan

March, Madrid,

May

1995).

A

_single

base

_{change inftsK}

causes aconditionalblock in cell division that is

_{suppressed by}

deletion of the

_{PBP5-encoding}

dacA.

A

battery

of

_techniques

is

_being

usedto

study

protein-pro-teininteractions within the "cell

_septation

network."

_According

to _reports

_presented

at the

_workshop

mentioned

above,

PBP3

interacts,

presumably

through

its intracellular amino

_end,

with

FtsA,

SulA,

and FtsZ in the

cytoplam;

FtsZ itselfinteracts with the

_integral

membrane

_{protein ZipA;}

and PBP3

_interacts,

pre-sumably through

its

periplasmic

module,

with

PBPlb,

PBP7

(an

endopeptidase),

and Slt-70.

At thesame

time,

the structural

_requirements

of PBP3 forin vivo

_activity

are

being

studied

by complementation

experi-ments. InE. coli

RP41,

the

_{temperature-sensitive-/ttl}

2158-en-codedPBP3 has two mutations: G191-D in motif 3*ofthe

n-PB module and D266-Natthe

junction

between then-PB and PB

modules.1

At

42°C,

E. coli RP41 grows as filaments and

lyses

but

_rod-shaped,

cell division and

_viability

arerestored

by

transformation with a

plasmid carrying

the

wild-type ftsl.

In

contrast,

complementation

isnotachieved

_{by ftsl}

_genes

encod-ing

PBP3mutantsthat either lack the membrane

_anchor,

have

a membrane anchor different form the

genuine

one,orhave a

17 amino acid residue insert

immediately

upstreamfrom R60 of motif1*

(the

insert

_being

theR60-D76 sequence,exceptthat D76 is mutated into N;

unpublished

datafrom this

laboratory

and J.

_{Ayala). Complementation}

doesnotoccurin

spite

of the

fact that the

_produced

PBP3mutantshave thesame

thermosta-biltiy

and

_{penicillin-binding capacity}

asthe

_wild-type

PBP3. The

_acquisition

ofa

penicillin-binding

fold

topology by

the

PBmodule of PBP3

_depends

on anintact motif 3* ofthen-PB

module but it is

_{membrane-anchor-independent.}

Itnow_appears

that the in vivo

_activity

of PBP3

_requires

not

_only

a correct

penicillin-binding

fold

_topology,

but,

in

addition,

the presence of the

_genuine

membraneanchorandanintactenvironment of

motif 1*_{of the n-PB module.Ithas also been}

_reported

_{that the}

P565-G571 sequenceatthe end

(V577)

of thematurePBP3 is

not

_required

for

_{penicillin-binding}

but is essentialfor in vivo

activity.20

Likely,

the membrane

anchor,

features ofthe n-PB

module,

and the

_carboxy

end of the

_polypeptide

chainaresites

through

which PBP3_mayinteract with other_componentsof the "cell division"

_{morphogenetic}

network.

The 14 min

_region

of theE. coli

chromosome,

which

con-tains the gene

encoding

the multimodular class B

PBP2,

also contains genes that encode the

"cell-shape"

PBP5 and

RodA,

a

protein

very similartoFtsW. This

_complex

and ribosomal

ac-tivities appeartobe coordinated

_by

achain of

interacting

ele-ments,oneof which is

regulated by

the nucleotide

_guanosine

5'-diphosphate, 3'-diphosphate (ppGpp,

an RNA

_polymerase

effector),

itself

_synthesized

_by

the

_{SpoT/RelA, proteins.}

Remarkably,

the "cell

_shape"

and "cell

_septation"

morpho-genetic

networks areconnected. PBP2 isnot

_required

for

sep-tum

synthesis.

However,

loss of PBP2

_activity

results inablock

of cell division

and,

in the absence of

PBP2,

celldivision and

viability

are restored

by

increasing

the

pool

of

ppGpp

or the

level of

_FtsQ-A-Z.30

The "cell

_septation"

and "cell

_shape"

networksare

_probably

ubiquitous

in the bacterial

world,

with _manyindividual

varia-tions.

_Likely,

other networks remaintobe identified. The

step-wise increased resistanceto

_/3-lactam

compounds

of S.

pneu-moniae

_laboratory

mutantsdoesnot

always

correlate withPBP

changes,

but correlates with

_genetic

competence

deficiency.23

In a cefotaxime

_laboratory

mutant, increased resistance and

competence

deficiency

aremediated

by

a

point

mutation ina

histidine kinase

CiaH.23

CiaH and CiaRaremembers of

signal

transduction

_{pathways including}

the

_EnvZ/OmpR

osmoregula-tion inE. coli and the

VanS/VanR

vancomycin

resistance

in-ducibility

in E.

_faecium.

The

_{morphogenetic}

networksare still far from

being

under-stood.

_{They probably}

_{possess several}

_phosphoryl

transfer

path-ways.

They

maybe components of

_{multiple parallel,}

overlap-ping,

and

_interacting

signal

transduction

_systems.

There is cross-talk between the

_pathways.

These characteristicsare

typ-ical of

_{"(phospho?)neural"}

networks.24

CONCLUSION

The

synthesis

and

_assembly

of the bacterial cell wall

pepti-doglycan

require proteins

the

_primary

functions ofwhich are

the chemical transformation of metabolite

intermediates,

the

building

ofacellular structure, and the transfer and

processing

ofinformation

_{through integrated}

biochemical "circuits" that

are abletotransform an

input signal

intoan_output

signal

and

anoutput

signal

intoan

input signal.

The wall

_{peptidoglycan}

is a

bacterium-specific polymer.

Empirical

approaches

to the

_discovery

of "bacterial-cell-wall" inhibitors and the

_improvement

of

_existing

drugs

were

justified

when basicresearchwas still

struggling

to_{cope with the}

com-plexity

of the process and the molecularstructures. The situa-tion is

_changing.

With the_presentadvancedstateofour

knowl-edge

and the

_availability

of

_experimental

and theoretical tools ofever

increasing

incisiveness,

future antibacterial

chemother-apy

strategies

are

likely

to

_depend

onthe

understanding

of the

functioning

of

_existing

targetsatthe atomic level

_allowing

new

drugs

tobe

designed,

andonthe identification of the

constitu-tive elements and the

_"wiring"

of the

morphogenetic

networks

allowing

new_targetstobe discovered.

ACKNOWLEDGMENTS

This workwas

supported

inpart

by

the

Belgian

programon

Interuniversity

Poles of Attraction initiated

_by

the

_Belgian

State,

Prime Minister's

Office,

Servicesfédéraux des affaires

scien-tifiques, techniques

et culturelles

(PAI 19),

the Fonds de la Recherche

_Scientifique

Médicale

_{(contract 3.4531.92),}

and the Fonds de la Recherche Fondamentale Collective

_(contract

2.4534.95).

REFERENCES

1.

_Ayala,

J.A.,T._Garrido,M.A. dePedro,and M. Vicente. 1994. Molecular

_biology

of bacterial

_septation,

_{pp. 73-101. In R.} Hakenbeck and J.M.

_Ghuysen

(eds.), New

_{comprehensive}

bio-chemistry,

Vol.27,Bacterial cell wall. Elsevier SciencePublishers,