Ornithine lipid is required for optimal steady-state amounts of c-type cytochromes in Rhodobacter capsulatus

Ornithine lipid is required for optimal steady-stateamounts of c-type cytochromes in Rhodobactercapsulatus

Semra Aygun-Sunar,1,2 Sevnur Mandaci,2

Hans-George Koch,1,3 Ian V. J. Murray,4

Howard Goldfine5 and Fevzi Daldal1*1Department of Biology, Plant Science Institute,University of Pennsylvania, Philadelphia, PA 19104,USA.2TUBITAK, Research Institute for Genetic Engineeringand Biotechnology, PO Box 21, Gebze-Kocaeli, 41470Turkey.3Institute for Biochemistry and Molecular Biology,University of Freiburg Medical School, Freiburg, 79104Germany.Departments of 4Pharmacology and 5Microbiology,School of Medicine, University of Pennsylvania,Philadelphia, PA 19104, USA.

Summary

The c-type cytochromes are haemoproteins that aresubunits or physiological partners of electron trans-port chain components, like the cytochrome bc1

complex or the cbb3-type cytochrome c oxidase. Theirhaem moieties are covalently attached to the corre-sponding apocytochromes via a complex post-translational maturation process. During our studiesof cytochrome biogenesis, we uncovered a novelclass of mutants that are unable to produce ornithinelipid and that lack several c-type cytochromes.Molecular analyses of these mutants led us to theornithine lipid biosynthesis genes of Rhodobactercapsulatus. Herein, we have characterized thesemutants, and established the chemical structure ofthis non-phosphorus membrane lipid from R. capsu-latus. Ornithine lipids are known to induce potent hostimmune responses, including B-lymphocyte mitoge-nicity, adjuvanticity and macrophage activation. Yet,despite their widespread occurrence in Eubacteria,and the diverse biological effects they elicit inmammals, their physiological role in bacterial cellsremained hitherto poorly defined. Our findings nowindicate that under certain bacterial growth conditions

ornithine lipids are crucial for optimal steady-stateamounts of some extracytoplasmic proteins, includ-ing several c-type cytochromes, and attribute them anovel and important biological function.

Introduction

The c-type cytochromes (cyt) are major membrane-integral or periplasmic proteins that contain one or morecovalently attached haem groups. They are crucial for avariety of growth modes of bacteria as they are involved invarious cellular processes, including electron transportbetween photosynthetic and respiratory energy transduc-tion complexes for ATP production (Moore and Pettigrew,1990). In many Gram-negative bacteria, the c-type cytsare produced via elaborate post-translational processesthat involve specific maturation components (Turkarslanet al., 2006) as well as the extracytoplasmic thio-reductionand -oxidation pathways (Porat et al., 2004). In the purplenon-sulphur facultative phototroph Rhodobacter capsula-tus, anaerobic photosynthetic growth requires cyt c1,which is a c-type cyt subunit of the cyt bc1 complex, andthe periplasmic cyt c2 or the membrane-anchored cyt cy aselectron carriers (Jenney and Daldal, 1993). Similarly, themitochondria-like respiratory growth of this species relieson the cyt c1 as well as the cyt co and cyt cp, which are thec-type cyt subunits of its cbb3-type cyt c oxidase (cbb3-Cox), a close relative of mitochondrial cyt c oxidase(Zannoni, 1995) (Fig. 1). Mutants lacking these c-typecyts are photosynthesis deficient (Ps–) and exhibit nocbb3-Cox activity that could be detected by using the Nadi(a-naphthol + dimethylphenylene diamine → indophenolblue + H2O) reaction (i.e. Nadi– phenotype) (Keilin, 1966).However, they can still grow by respiration via an alternatepathway devoid of c-type cyts (Fig. 1).

Previously, we have described several groups ofR. capsulatus Nadi– mutants lacking the cbb3-Cox activ-ity (Koch et al., 1998). These strains contained muta-tions in either the structural genes (ccoNOQP) of thecbb3-Cox or another gene cluster (ccoGHIS) locatedimmediately downstream of it, required for the matura-tion of this enzyme (Koch et al., 2000). More recently,we have proposed a multistep subunit-assemblypathway for the cbb3-Cox enzyme (Kulajta et al., 2006).

Accepted 18 May, 2006. *For correspondence. E-mail [email protected]; Tel. (+1) 215 898 4394; Fax (+1) 215 898 8780.

Molecular Microbiology (2006) 61(2), 418–435 doi:10.1111/j.1365-2958.2006.05253.xFirst published online 15 June 2006

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd

During our studies on the biogenesis of the c-type cytsand their complexes, we have encountered anothergroup of R. capsulatus mutants that exhibited no cbb3-Cox activity and not complemented by the ccoNOQP-ccoGHIS clusters. In this work, we characterize thesemutants that contain suboptimal amounts of c-type cytsas a result of mutations located in either of two adjacentgenes (RRC00138 and RRC00139). We show that thesegene products are required for the biosynthesis of orni-thine lipid [OL, a-N-(3-acyloxyacyl)-ornithine]. Despitethe widespread occurrence of OL in bacteria and potentbiological effects that it elicits in mammals (Kawai et al.,1996), the specific function of this amino-lipid in cellsproducing it remained hitherto poorly defined (Lopez-Lara et al., 2005; Rojas-Jimenez et al., 2005). Here wedemonstrate that, in the absence of OL, R. capsulatuscells do not harbour a full complement of c-type cytsunder specific physiological conditions. These findingsreveal a previously unknown link between OL biosynthe-sis and optimal steady-state amounts of some extracy-toplasmic proteins, including various c-type cyts, andallow us to interlink two vital and seemingly distinct cel-lular processes, which are the lipid biosynthesis and thecyt c biogenesis in bacteria.

Results

Rhodobacter capsulatus Nadi- mutants that are notcomplemented with ccoNOQP-ccoGHIS

Screening of Nadi– (i.e. cyt c oxidase deficient) mutantsthat are not complemented with the ccoNOQP-ccoGHIS

gene clusters, known to affect the cbb3-Cox activity,defined a new group of mutants, including MR2, IJ1 (Kochet al., 1998) and AYG4 (Table 1). These mutants exhibitedcomplex phenotypic characteristics (Fig. 2). At 35°C, theywere Ps– and Nadi– on enriched medium while they werePs+ and Nadi– on minimal medium (Fig. 2, left panel).Analysis of the c-type cyts profiles of these mutantsgrown on enriched MPYE medium at 35°C, using3,3�,5,5�-tetramethylbenzidine (TMBZ)/SDS-PAGE, wasconducted. The data showed that, of the membrane-bound c-type cyts (cp, c1, cy and co with Mr of 32, 31, 29and 28 kDa respectively) (Gray et al., 1994), the mutantscontained predominantly cyt c1 of the cyt bc1 complexalbeit at lower amounts (Fig. 3A). The cyt cp, cyt cy and cytc2 (a periplasmic soluble protein, data not shown) wereabsent in all mutants, with the exception that smallamounts of cyt co could still be seen when the cbb3-Coxwas overproduced (Fig. 3A, lane 4). Estimation of theamounts of the cyt bc1 complex by using its enzymaticactivity (Experimental procedures) indicated that themutants contained about 25% of the wild-type amount ofthis enzyme (approximately 260 versus 1020 nmoles ofcyt c reduced per mg of protein per minute respectively),in agreement with the lowered amounts of cyt c1 seenby TMBZ/SDS-PAGE analyses (Fig. 3A). Clearly, themutants contained decreased steady-state amounts ofvarious c-type cyts to differing degrees.

Next, ccoN::lacZ (pXG2, Koch et al., 1998) andcycY::lacZ (pHM12, Myllykallio et al., 1997) or cycY::phoA(pHM11, Myllykallio et al., 1997) fusions were used toprobe whether or not the absence of the c-type cyts was

Respiratory Dehydrogenases

Qpool

Cytochrome bc complex

1

(cyt c )1

cyt c 2

cyt c y

or

Cytochrome cbb oxidase3

(cyt c and cyt c )o p

Reaction centre

Quinoloxidase

Nadi stain

Photosynthesis

Respiration

(Ps growth)

Cyt c oxidase branch

RespirationQuinol oxidase branch

O2

H2O

O2

H2O

Fig. 1. Respiratory and photosynthetic growth pathways of R. capsulatus and their c-type cyts. During respiration, reduced quinones (Qpool)are provided by the respiratory dehydrogenases, and subsequently oxidized by the cyt bc1 complex (cyt c1) reducing via the electron carrierscyt c2 or cyt cy the cyt cbb3 oxidase (cyt co and cyt cp), thus converting O2 to H2O. During photosynthesis, the electron carriers cyt c2 and cyt cy

convey electrons to the photochemical reaction centre, which reduces quinones (Qpool). The alternate, quinol oxidase-dependent, respiratorypathway not containing any c-type cyts is also shown. Mutants lacking at least the cyt bc1 complex and the cyt c2 and cyt cy are deficient inphotosynthesis (Ps–), and mutants missing at least the cyt co or cyt cp have no cyt c oxidase activity detectable by the Nadi stain.

Ornithine lipids and c-type cytochromes 419

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 418–435

Tab

le1.

Bac

teria

lstr

ains

and

plas

mid

sus

edin

this

stud

y.

Str

ain

orpl

asm

idD

escr

iptio

nR

elev

ant

char

acte

ristic

saS

ourc

eor

refe

renc

e

Str

ains

E.c

oli

HB

101

F–

D(gp

t-pr

oA)6

2le

uB6

supE

44ar

a-14

galK

2la

cY1

D(m

crC

-mrr

)rp

sL20

xyl-5

mtl-

1re

cA13

Str

R;

r B–

mB

–S

ambr

ook

etal

.(1

989)

R.c

apsu

latu

sM

T11

31b

crtD

121

RifR

Wild

-typ

e,N

adi+ ,

Ps+

Koc

het

al.

(199

8)M

T11

31/p

OX

15c

crtD

121

RifR

ccoN

OQ

Pdi

ploi

dTe

tR,

Nad

i+ ,P

s+K

och

etal

.(1

998)

IJ1

olsA

1(R

RC

0013

8-1)

Nad

i– ,P

s– /M

PY

E,

Ps+ /

Med

AK

och

etal

.(1

998)

MR

2ol

sB2

(RR

C00

139-

2)N

adi– ,

Ps– /

MP

YE

,P

s+ /M

edA

Koc

het

al.

(199

8)Y

262

GTA

over

prod

ucer

Yen

etal

.(1

979)

GK

32D(

ccoN

O::k

an)

Nad

i– ,P

s+K

och

etal

.(1

998)

CW

1D(

ccoG

HIS

::spe

)N

adi– ,

Ps+

Koc

het

al.

(200

0)A

YG

4ol

sB4

(RR

C00

139-

4)in

MT

1131

/pO

X15

TetR

,N

adi– ,

Ps– /

MP

YE

,P

s+ /M

edA

Thi

sst

udy

AY

G4-

25A

YG

4cu

red

ofpO

X15

TetS

,N

adi– ,

Ps– /

MP

YE

,P

s+ /M

edA

Thi

sst

udy

SA

1ol

sA::s

pe=

(RR

C00

138:

:spe

)N

adi– ,

Ps– /

MP

YE

,P

s+ /M

edA

Thi

sst

udy

SA

2R

RC

0013

6::s

peN

adi+ ,

Ps+

Thi

sst

udy

SA

3D[

(ilvD

-Exo

D):

:spe

]N

adi+ ,

Ps+

Thi

sst

udy

SA

4D(

olsA

::spe

)=

D(R

RC

0013

8::s

pe)

Nad

i– ,P

s– /M

PY

E,

Ps+ /

Med

AT

his

stud

yS

A5

D(R

RC

0014

0::s

pe)

Nad

i+ ,P

s+T

his

stud

yS

A6

D(ol

sB::s

pe)

=D(

RR

C00

139:

:spe

)N

adi– ,

Ps– /

MP

YE

,P

s+ /M

edA

Thi

sst

udy

SA

8D[

(ols

BA

)::s

pe]=

D[(R

RC

0013

9-13

8)::s

pe]

Nad

i– ,P

s– /M

PY

E,

Ps+ /

Med

AT

his

stud

yS

.mel

iloti

Sm

1021

Wild

type

Wei

ssen

may

eret

al.

(200

2)O

RLD

1ol

sA::k

anW

eiss

enm

ayer

etal

.(2

002)

AA

K1

olsB

::kan

Gao

etal

.(2

004)

Pla

smid

spR

K20

13tr

a+(R

K2)

Kan

R,

conj

ugat

ive

help

erD

itta

etal

.(1

985)

pRK

404

or41

5Te

tR,

broa

d-ho

st-r

ange

vect

orD

itta

etal

.(1

985)

pHP

45W

Wsp

ein

pHP

45ve

ctor

Spe

RA

mpR

Pre

ntki

and

Kris

ch(1

984)

pXG

2cc

oN::l

acZ

TetR

Koc

het

al.

(199

8)pH

M7

cycY

-FLA

GTe

tRM

ylly

kalli

oet

al.

(199

7)pH

M11

cycY

188:

:pho

ATe

tRK

anR

Myl

lyka

llio

etal

.(1

997)

pHM

12cy

cY::l

acZ

TetR

Myl

lyka

llio

etal

.(1

997)

pOX

15cc

oNO

QP

inpR

K40

4Te

tRK

och

etal

.(1

998)

pMR

C5.

9kb

chro

mos

omal

Eco

RI

frag

men

tin

pLA

FR

1Te

tRK

och

etal

.(1

998)

420 S. Aygun-Sunar et al.


pMR

C2

4.5

kbch

rom

osom

alE

coR

Ifr

agm

ent

inpR

K41

5Te

tRK

och

etal

.(1

998)

pMR

B/p

AY

17

kbch

rom

osom

alB

amH

Ifr

agm

ent

inpR

K40

4Te

tRK

och

etal

.(1

998)

,th

isst

udy

pMR

B4

1.5

kbch

rom

osom

alE

coR

Ifr

agm

ent

inpR

K41

5Te

tRT

his

stud

ypS

EM

1pM

RC

2cu

tw

ithB

amH

Ian

dlig

ated

(120

0bp

dele

tion)

TetR

Thi

sst

udy

pSE

M2

Sm

aI-W

spe

ofpH

P45

Win

sert

edin

toun

ique

Xho

Isi

teof

RR

C00

136

onpS

EM

1Te

tRS

peR

Thi

sst

udy

pSE

M3

1365

bpB

stE

IIbl

unt-

ende

dfr

agm

ent

cove

ring

ilvD

-Exo

Don

pSE

M1

repl

aced

with

Sm

aI-W

spe

ofpH

P45

WTe

tRS

peR

Thi

sst

udy

pSE

M4

Bam

HI-

Wsp

eof

pHP

45W

inse

rted

into

uniq

ueB

amH

Isi

teof

RR

C00

138

onpM

RC

TetR

Spe

RT

his

stud

ypS

EM

55.

3kb

Eco

RV

-Hin

dIII

frag

men

tof

pMR

Ccl

oned

into

the

sam

esi

tes

ofpB

R32

2(S

ambr

ook

etal

.,19

89)

Am

pRT

his

stud

y

pSE

M6

19bp

Hpa

Ifr

agm

ent

ofR

RC

0014

0on

pSE

M5

repl

aced

with

Sm

aI-W

spe

ofpH

P45

WS

peR

Am

pRT

his

stud

ypS

EM

72.

6kb

Pst

Ifr

agm

ent

ofpS

EM

6lig

ated

toP

stI

site

ofpR

K41

5Te

tRS

peR

Thi

sst

udy

pSE

M8

313

bpR

srII

frag

men

tof

RR

C00

139

onpS

EM

5re

plac

edw

ithS

maI

-Wsp

eof

pHP

45W

Spe

RA

mpR

Thi

sst

udy

pSE

M9

2.7

kbX

hoI-

Bsu

36I

frag

men

tof

pSE

M8

ligat

edin

toth

eB

amH

Isi

teof

pRK

415

TetR

Spe

RT

his

stud

ypS

EM

100.

5kb

Bst

XI-

Not

Ifr

agm

ent

ofR

RC

0013

8on

pSE

M5

repl

aced

with

Sm

aI-W

spe

ofpH

P45

WTe

tRS

peR

Thi

sst

udy

pSE

M11

5.3

kbE

coR

V-E

coR

Ifr

agm

ent

ofpS

EM

10lig

ated

into

the

Bam

HI

site

ofpR

K41

5Te

tRS

peR

Thi

sst

udy

pSE

M12

2.0

kbM

luI

frag

men

tof

pSE

M5

carr

ying

RR

C00

138-

RR

C00

139

repl

aced

with

Sm

aI-W

spe

ofpH

P45

WS

peR

Am

pRT

his

stud

y

pSE

M13

5.0

kbE

coR

V-B

su36

Ifr

agm

ent

ofpS

EM

12lig

ated

into

the

Bam

HI

site

ofpR

K41

5Te

tRS

peR

Thi

sst

udy

pBA

D/M

yc-H

isA

Am

pR,

arab

inos

e-in

duci

ble

vect

orIn

vitr

ogen

pSE

M19

777

bpP

CR

prod

uct

cont

aini

ngR

RC

0013

9cl

oned

into

Nco

I-S

caI

site

sof

pBA

D/M

yc-H

isA

Am

pRT

his

stud

ypS

EM

20N

siI

cut

pSE

M19

ligat

edin

toP

stI

cut

pRK

415

TetR

Am

pRT

his

stud

ypJ

G21

S.m

elilo

tiol

sBin

pRK

404

TetR

Kan

RG

aoet

al.

(200

4)

a.P

san

dN

adir

efer

toph

otos

ynth

etic

grow

than

dC

ox-d

epen

dent

cata

lysi

sof

a-na

phth

olto

indo

phen

olbl

ue,r

espe

ctiv

ely,

at35

°C.F

orco

rres

pond

ing

phen

otyp

esat

25°C

,see

the

text

and

Fig

.2.

Abb

revi

atio

nsof

antib

iotic

resi

stan

ces

are

asfo

llow

s:A

mp,

ampi

cilli

n;K

an,

kana

myc

ine;

Rif,

rifam

pici

n;S

pe,

spec

tinom

ycin

;Te

t,te

trac

yclin

e.b

.R

.cap

sula

tus

MT

1131

isre

ferr

edto

as‘w

ild-t

ype’

with

resp

ect

toits

c-ty

pecy

tspr

ofile

and

grow

thpr

oper

ties.

All

R.c

apsu

latu

sst

rain

s,ex

cept

Y26

2,ar

ede

rived

from

MT

1131

.c.

MT

1131

/pO

X15

over

prod

uces

the

wild

-typ

ecb

b 3-C

oxby

abou

tfiv

efol

d(K

och

etal

.,19

98).



due to general transcription-translation or secretiondefects. The b-galactosidase and alkaline phosphataseactivities exhibited by appropriate wild-type or mutant

transconjugants carrying these fusions were similar whencells were grown in MPYE under respiratory conditions(Fig. 3B). Thus, the decreased amounts of c-type cyts

PsEnriched MPYE medium Minimal Med A medium

Res

35 C 25 Coo

Enriched MPYE medium Minimal Med A medium

N+ N-(wt) (ΔolsA) N+ N-(wt) (ΔolsA) N+ N(wt) (ΔolsA)+ N+ N-(wt) (ΔolsA)

N-(ΔolsBA)N-(ΔolsB) N-(ΔolsBA)N-(ΔolsB) N (ΔolsBA)N (ΔolsB) N-(ΔolsBA)N-(ΔolsB)+ +

wt ΔolsA

ΔolsBAΔolsB

wt ΔolsA

ΔolsBAΔolsB

wt ΔolsA

ΔolsBAΔolsB

wt ΔolsA

ΔolsBAΔolsB

Fig. 2. Medium- and temperature-dependent complex growth phenotypes of R. capsulatus RRC00138 (olsA) and RRC00139 (olsB) mutants.Enriched MPYE or minimal MedA media containing plates were incubated at either 35°C (left two columns) or 25°C (right two columns), underrespiratory (top two rows) or photosynthetic (bottom two rows) growth conditions. Respiratory grown cells were subsequently stained for Nadi(N) reaction (Experimental procedures). Colonies that contain an active cyt c oxidase (N+) turn blue while those that lack it retain their parental‘green’ colour. R. capsulatus strains shown are wt (wild type, MT1131), DolsA (SA4), DolsB (SA6) and DolsBA (SA8) mutants. The initialisolates IJ1 and MR2/AYG4-25 (not shown) are allelic to SA4 and SA6 respectively, and exhibit the same phenotypes. OlsA– and OlsB–

mutants are Ps–/N– at 35°C and Ps+/N+ at 25°C on enriched media, but are Ps+/N– at both 35°C and 25°C on minimal media.

ccp1

yo

cc

A

MT1131 IJ1 MR2 AYG4 GK32 CW1

olsA1 olsB2 olsB4Δcco Δcco

32

25

NOQP GHIS

MPYE CcoN::lacZ CycY::lacZ CycY::phoA

β-ga

lact

osid

ase

activ

ity

400

600

800

1000

1200

200

B

400

600

800

1000

1200

200

Alk

alin

e ph

osph

atas

e ac

tivit

y

wtwt

olsA1 olsB2

Fig. 3. Membrane-associated c-type cyts profiles of R. capsulatus mutants grown at 35°C in enriched MPYE medium. Membrane fractionsisolated from R. capsulatus strains grown at 35°C in enriched MPYE medium were analysed using 16.5% Tricine SDS-PAGE, and the c-typecyts revealed using TMBZ, as described in Experimental procedures. The c-type cyt subunits of the cbb3-Cox (cp and co) and the cyt bc1

complex (c1) along with the electron carrier cyt cy (cy) are indicated on the left of panel A together with the 32.5 and 25 kDa molecular weightmarkers. The R. capsulatus wild type (MT1131), OlsA– (IJ1), OlsB– (MR2 and AYG4), DccoNO (GK32) and DccoGHIS (CW1) strains (Table 1)were grown at 35°C in enriched MPYE medium as indicated. Panel B shows the b-galactosidase and alkaline phosphatase activities exhibitedby the wild-type MT1131 (wt), olsA (IJ1) and olsB (MR2) mutants carrying ccoN::lacZ, cycY::lacZ or cycY::phoA reporter gene fusions (Table 1)grown in enriched MPYE medium at 35°C, as described in Experimental procedures.



found in the mutants did not appear to originate fromgeneral transcription-translation or secretion defectsunder these growth conditions.

Identification of the mutated genes in the MR2/AYG4-25and IJ1 strains

Genetic complementation of MR2, IJ1 andAYG4-25 (a TetS

derivative of AYG4 cured of its plasmid pOX15) mutants(Table 1) with a transferable R. capsulatus chromosomallibrary (Koch et al., 1998), yielded the plasmid pMRCcarrying two contiguous EcoRI fragments of 5.9 kb total(Fig. 4, top). Similar experiments using additional librariesyielded the plasmids pMRB/pAY1 carrying the 2.8 kbBamHI-EcoRI portion of pMRC, which complemented only

MR2 andAYG4-25, while another plasmid pMRB4 carryingonly the 1.5 kb EcoRI fragment of pMRC complementedneither of the mutants (Fig. 4). Determination of the 5� and3� ends nucleotide sequences of thus isolated chromo-somal inserts, and comparison with the R. capsulatusgenome sequence (http://ergo.integratedgenomics.com)revealed that pMRC contained two partial [RRC00132(ilvD) and RRC00143 (proA)] and several complete openreading frames (ORFs), annotated as RRC00133 (exoD),135, 136, 138 (plsC), 139, 140 and 142. Furthermore,these data also indicated that the BamHI site at the 5� endof pMRB/pAY1 corresponded to that located in RRC00138,and suggested that RRC00139 and RRC00138 might bedefective in MR2 (and AYG4-25) and IJ1 respectively(Fig. 4).

1 kb

pMRC

pMRB/pAY1

pMRB4

pSEM4(SA1)

~

RRC00138 RRC00139 proARR

C00136

RR

C00135

exoDilvD

EE

spe

Xh Rs Rs BsupSEM9(SA6, ΔolsB)

Ml Ml

pSEM13(SA8, ΔolsBA)

Complementation IJ1 MR2/AYG4-25

+ +

+

_+

+

_

_

Ba

RR

C00140

RR

C00142

pSEM11(SA4, ΔolsA)

_ _

_ _

_ _

olsA olsB

E

BaBa E

~

EE

EE Ba E

EV Bs EN

spe

Ba

spe

EV Bsu

spe

(olsA1) (olsB2/B4)

(olsA::spe)

Fig. 4. R. capsulatus RRC00138 (olsA) and RRC00139 (olsB) loci. ilvD (dihydroxy-acid dehydratase), exoD (exopolysaccharide synthesisprotein) and proA (gamma-glutamyl phosphate reductase) encompassing RRC00138-139 (olsAB) are shown along with RRC00135, 136, 140and 142 of unknown functions. Arrows depict the transcription directions for these loci. RRC00138 (annotated previously as1-acyl-sn-glycerol-3-phosphate acyltransferase, plsC) and RRC00139 (hypothetical cytosolic protein) correspond to olsA and olsB respectively,involved in OL biosynthesis. The plasmids shown underneath pMRC correspond to its derivatives carrying insertion or insertion-deletion allelesof olsA or olsB, constructed by using W-spe cassette. SA1, SA4, SA6 and SA8 refer to R. capsulatus strain MT1131 derivatives with thecorresponding chromosomal mutant alleles (Table 1), constructed as described in Experimental procedures. Complementation results betweenOlsB– (olsB2 or 4) and OlsA– (olsA1) mutants and various plasmids are shown on the right hand side, with + and – referring to the Nadiphenotypes. Ba, Bs, Bsu, E, EV, Ml, N, Rs and Xh indicate the restriction endonucleases BamHI, BstXI, Bsu36I, EcoRI, EcoRV, MluI, NotI,RsrII and XhoI cleavage sites respectively.



http://ergo.integratedgenomics.com

Inactivation of the chromosomal copy of eitherRRC00138 or RRC00139 leads to the absence of thecbb3-Cox activity

Interposon mutagenesis, used to knock out the chromo-somal copies of different intact ORFs located on pMRC(Experimental procedures), indicated that inactivation ofonly the RRC00138 [SA4, D(olsA::spe)] or RRC00139[SA6, D(olsB::spe)] or both [SA8, D(olsBA::spe)] yieldedNadi– mutants on both enriched MPYE and minimal MedA media at 35°C (Fig. 2, left panel) (Table 1). Conversely,inactivation of RRC00136 and ilvD-ExoD located down-stream of RRC00138, or RRC00140 located upstream ofRRC00139 yielded Nadi+ mutants (SA2, SA3 and SA5respectively, data not shown). Moreover, the c-type cytsprofiles of SA4, SA6 and SA8 were also identical to MR2(AYG4-25) and IJ1. They lacked various c-type cyts whengrown at 35°C on MPYE medium, in agreement with theirNadi/Ps phenotypes (Fig. 2, left panel) (compare Fig. 3Alanes 2–4 and Fig. 5). On the other hand, in minimalmedium, they contained c-type cyts at lower amounts(Fig. 5, lanes 6–8). As expected, they were fully comple-mented for both Nadi/Ps and c-type cyts phenotypes witha plasmid (pMRC) harbouring both RRC00138 andRRC00139 (Fig. 5, lane 9). In addition, the plasmids car-rying solely RRC00138 or RRC00139 complemented onlyIJ1 and SA4 or only MR2 (also AYG4-25) and SA6 respec-tively, and that the transconjugants SA6 [D(olsB::spe)]/pSEM4 (olsA::spe) and SA4 [D(olsA::spe)]/pSEM9[D(olsB::spe)] (Table 1) were Nadi+ (data not shown), indi-cating that the interposon insertion in RRC00139 did notabolish the expression of RRC00138. Finally, although themutants SA6, MR2 and AYG4 were all defective inRRC00139 (Table 1), only the latter strain harbouringpOX15 (ccoNOQP) exhibited a small amount of cyt co

(Fig. 3A, lane 4), due to its ccoNOQP merodiploid geno-type (Table 1).

On the R. capsulatus genome, RRC00138 was initiallyannotated as a plsC encoding 1-acyl-sn-glycerol-3-phosphate acyltransferase (AGPAT) due to its similarity to

this enzyme from various species (e.g. 18% identity +23% similarity to Escherichia coli, and 18% identity + 24%similarity to Saccharomyces cerevisiae proteins). InE. coli, AGPAT is a membrane-bound enzyme that cataly-ses the transfer of an acyl chain from an acyl-acyl carrierprotein (ACP) or acyl-CoA to 1-acyl-sn-glycerol-3-phosphate (lysophosphatidic acid, LPA) to produce 1,2diacyl-sn-glycerol-3-phosphate (phosphatidic acid, PA).PA is a precursor of prokaryotic membrane phospholipids,including phosphatidylethanolamine (PE), phosphatidylg-lycerol (PG) and cardiolipin (CL) (Raetz and Dowhan,1990; Cronan, 2003), and E. coli PlsC mutants are lethaldue to their inability to produce phospholipids (Coleman,1990). RRC00139 is annotated as a hypothetical cytosolicprotein of unknown function, but it exhibits homology toN-acyltransferases, also using acyl-ACP as a substrate(Stanley et al., 1998). These similarities raised the possi-bility that the absence of various c-type cyts in mutantswith defective RRC00138 or RRC00139 might be linkedto phospholipid biosynthesis. However, when coupledglycerol-3-phosphate acyltransferase (GPAT) + AGPATactivities of R. capsulatus were assayed using membraneparticles, acyl acceptor [14C]-G3P, and acyl donor cis-vaccenyl-ACP (Experimental procedures), no differencewas seen between an RRC00138 knock-out mutant andits wild-type parent (data not shown). Moreover, both wild-type and the mutant strains produced similar amounts ofPA, as evidenced by thin layer chromatography (TLC)of the radiolabelled reaction products (data not shown).The inactivation of RRC00138 not being lethal inR. capsulatus, and the mutants lacking it still producingnative amounts of PA, we reasoned that the absence ofvarious c-type cyts could not originate from a defect inphospholipid biosynthesis.

Total lipid compositions of R. capsulatus mutantsdefective in RRC00138 and RRC00139

More recently, a biosynthesis pathway for the non-phosphorus OL has been elucidated in Sinorhizobium

Fig. 5. Comparison of membrane-associatedc-type cyts profiles of R. capsulatus OL-lessmutants grown at 35°C in enriched MPYE orminimal MedA media. Membrane fractionsisolated from appropriate R. capsulatusstrains grown at 35°C in appropriate mediawere analysed as described in Fig. 3. Thestrains used wt (wild-type MT1131), DolsA(SA4), DolsB (SA6), D(olsBA) (SA8) andD(olsBA)/(olsBA)+ (SA8/pMRC) were grown at35°C in enriched MPYE or minimal MedAmedia as indicated, and the c-type cytsrevealed using TMBZ, as described inExperimental procedures.

cc

p1

yo

cc

32

25

MPYE Med A MPYE

wt ΔolsA ΔolsB ΔolsBA wt ΔolsA ΔolsB ΔolsBA ΔolsBA / (olsBA)+

MT1131 SA6 SA8SA4 SA6 SA8MT1131 SA4 SA8 /pMRC



meliloti. In this pathway, ornithine is N-acylated with a3-hydroxy-fatty acyl moiety from an acyl-ACP by anN-acyltransferase (OlsB) to form lyso-ornithine lipid (LOL)(Gao et al., 2004), which is then converted to OL by anO-acyltransferase (OlsA) using a second acyl-ACP (Weis-senmayer et al., 2002). Rhodobacter species contains OL(Asselineau, 1991), and RRC00138 and RRC00139showed similarity to S. meliloti olsA (29% identities + 25%similarities) and olsB (34% identities + 28% similarities)respectively. Upon separation of non-labelled, or [1-14C]acetate-labelled, total lipid extracts from R. capsulatuswild-type and mutant strains using two dimensional (2D)-TLC (Experimental procedures), the absence of CL andthe presence of a ninhydrin-positive spot distinct from PE,identified as OL in S. meliloti with the same 2D-TLCsolvent system (Weissenmayer et al., 2002), were noted(Fig. 6A and B). The R. capsulatus mutants lackingRRC00138 (olsA) (data not shown) or RRC00139 (olsB)(Fig. 6C), or both (data not shown), contained no spotcorresponding to OL in enriched MPYE or minimal MedAgrowth media, whereas spots corresponding to PE, phos-phatidylcholine (PC) and PG were readily identified bycomparison with appropriate standards. Significantly, theninhydrin-positive OL spot reappeared upon complemen-tation of these mutants with a plasmid carrying wild-typecopies of RRC00139-138 (olsBA) (Figs 6D and 7A).

Usually, OL constitutes a small fraction of total bacteriallipids, but in some species under phosphorus-limitinggrowth conditions, its amount increases to over 50% of

total lipids as a substitute for PE (Minnikin and Abdolra-himzadeh, 1974; Benning et al., 1995; Geiger et al.,1999). Whether this is also the case in R. capsulatus wastested using minimal MedA medium containing differentamounts of phosphate. When the phosphate concentra-tion in the growth medium was lowered from 20 mM (Fig.6A) to 0.1 mM (Fig. 6B), the R. capsulatus total lipid com-position changed, as seen by the increased abundance ofa lipid spot, tentatively assigned to be diacylglyceryltrim-ethylhomoserine (DGTS) (Benning et al., 1995: Geigeret al., 1999). However, no drastic changes could be seenin the amounts of the ninhydrin-positive spot, postulatedto be R. capsulatus OL (see below). Furthermore, similarlipid compositions were observed with cells grown inminimal MedA (Fig. 6A) or enriched MPYE media (datanot shown), suggesting that R. capsulatus OL synthesis isnot tightly regulated by phosphate availability.

Interspecies complementation between R. capsulatusand S. meliloti establishes that RRC00138 andRRC00139 are functional homologues of olsA and olsB

Proof that RRC00138 and RRC00139 are functionalhomologues of S. meliloti olsA and olsB, respectively,was obtained by interspecies complementation. Aplasmid carrying S. meliloti olsB was introduced into aR. capsulatus RRC00139 mutant (Rc-olsB) (Fig. 7B).Similarly, a plasmid carrying R. capsulatus RRC00139(olsB) or RRC00138-139 (olsBA) were introduced into

Fig. 6. Total lipids composition ofR. capsulatus strains grown at 35°C inminimal MedA media. [1-14C]-acetate labelledtotal lipids were extracted, and in all casessimilar amounts (60 000 cpm) were analysedby 2D-TLC, as described in Experimentalprocedures. Panels A and B correspond toR. capsulatus wild type (Rc-wild type)(MT1131) grown at 35°C in regular minimalMedA (containing 20 mM phosphate) or inminimal MedA containing only 0.1 mMphosphate (Pi) respectively. Similar total lipidprofiles were seen when wild-type strainMT1131 was grown in regular minimal MedA(panel A) or enriched MPYE media (notshown). An OlsB– mutant Rc-DolsB (SA6)without and with the plasmid pMRC carryingolsBA (Rc-DolsB/Rc-olsBA+) are shown inpanels C and D respectively. Similar total lipidprofiles were seen with OlsB– (SA6) or OlsA–

(SA4) and OlsBA– (SA8) mutants (not shown).Arrows indicate OL, and PE, PG, PC, DGTSand DMPE refer to phosphatidylethanolamine,phosphatidylglycerol, phosphatidylcholine,diacylglyceryltrimethylhomoserine andphosphatidyl-N,N dimethylethanolaminerespectively. The vertical and horizontalarrows at the origin O refer to the first andsecond dimension of solvent migrationsrespectively.

PG

PEOL

PC

O

A Rc-wild type

C

PG

PE

PC

O

B

PG

PEOL

Rc-wild type

O

PC

DMPE

DMPE

DMPEDGTS

DGTS

DGTS

Med A Med A/0.1 mM Pi

Med A

D

PG

PE

PC

OL

O

DMPE

DGTS

Med A

+Rc-ΔolsB Rc-ΔolsB / Rc-(olsBA)



OlsB– or OlsA–S. meliloti mutants respectively (Fig. 7Cand D). All heterologous transconjugants regained theability to produce OL as analysed by 2D-TLC. Further-more, in the case of R. capsulatus, they become Nadi+

and contained the full complement of c-type cyts in theirmembranes (data not shown). We therefore concludethat R. capsulatus RRC00138 and RRC00139 arefunctional homologues of S. meliloti olsA and olsBrespectively.

Chemical structure of R. capsulatus OL

The chemical identity of the postulated R. capsulatus OLwas defined by comparison with that known from S. melilotifollowing 2D-TLC separation of appropriate lipid spots, andmass spectrometric analyses. The spot corresponding toS. meliloti OL, which as expected was absent in the OlsA–

mutant (Weissenmayer et al., 2002), was used as a controland yielded an intense ion [M + H]+ at m/z 694 in the600–800 region of the mass spectrum (data not shown).Additional MS/MS analysis of the m/z 694 ion yieldedintense diffusion-induced collisional fragmentation prod-ucts at m/z 415, 397, 379, 361 and 133, 115, 70, whichcould be attributed to protonated ornithine ([Orn + H]+),ornithine B (Orn B) and ornithine immonium (Orn immo-nium) ions respectively. The full spectrum and the fragmen-tation pattern of m/z 694 ion were identical to previouslypublished data on S. meliloti OL (Geiger et al., 1999;Weissenmayer et al., 2002), and provided a robust experi-

mental control (data not shown). In the case of R. capsu-latus, the postulated OL spot yielded in the 600–800 regionof the mass spectrum an intense ion at m/z 680, which wasabsent in the olsBA double knock-out mutant (Fig. 8 inset).When this intense ion was subjected to MS/MS analysisfollowing diffusion-induced collisional fragmentation, ityielded in the low-mass (up to m/z 400) region intenseproduct ions at m/z of 397, 380, 361, 133, 115 and 70,which were identical to those seen for S. meliloti OL(Fig. 8). Thus, the m/z fragmentation pattern observed wasconsistent with the presence of ornithine in R. capsulatuslipid. The m/z-values of the intense ion missing in OL-lessmutants of R. capsulatus or S. meliloti (680 or 694 respec-tively) indicated that the OL of these species differed by anm/z difference of 14. The identical fragmentation patternsobserved for m/z-values up to 415 (corresponding to LOL[M + H]+) implied that this difference might not be on theLOL moiety (Geiger et al., 1999; Weissenmayer et al.,2002). The second acyl chain of OL contains a cyclopro-pane ring of m/z of 14 in the case of S. meliloti OL, althougha minor OL fraction containing a cis-vaccenyl moiety at thisposition is also present in this species (Geiger et al., 1999).However, R. capsulatus does not contain cyclopropanefatty acids (Grogan and Cronan, 1997) in agreement withthe absence of a cyclopropane fatty acid synthase in itsgenome. Thus, we tentatively conclude that the secondacyl chain of R. capsulatus OL is an 18:1-(D11) unsatur-ated fatty acid, the most abundant fatty acid in this species(Asselineau, 1991).

Fig. 7. R. capsulatus–S. meliloti interspeciescomplementation between appropriate OlsA–

or OlsB– mutants and OL overproduction.[1-14C]-acetate labelled total lipids wereextracted and in all cases similar amounts(60 000 cpm) were analysed by 2D-TLC, asdescribed in Experimental procedures. Notethat unlike R. capsulatus total lipids from awild-type S. meliloti also contain CL. TheS. meliloti OlsB– mutant Sm-DolsB (AAK1)harbouring plasmid pSEM20 carryingR. capsulatus olsB (Sm-DolsB/Rc-olsB+) andthe R. capsulatus OlsB– mutant Rc-DolsB(SA6) harbouring plasmid pJG21 carryingS. meliloti olsB (Rc-DolsB/Sm-olsB+) areshown in panels C and D respectively. Toillustrate OL overproduction, the R. capsulatusOlsA– [Rc-DolsA (SA4)] and S. meliloti OlsA–

[Sm-DolsA (ORLD1)] mutants harbouringplasmid pMRC carrying R. capsulatus olsBA+

are shown in panels A and B respectively. PC,PE, PG, OL, CL, DGTS and DMPE are as inFig. 6.

PG

PE

PCCL

DMPEOL

O

C Sm-ΔolsB / Rc-olsB

PE

PC

OL

B

PG

O

+

Rc-ΔolsB / Sm-olsB+

PEOL

D Sm-ΔolsA / Rc-(olsBA)

PGCL PC

O

DMPE

+

DMPE

DGTS

Med A

Med A

A

PG

PE

PC

OL

O

Rc-ΔolsA / Rc-(olsBA)+

DMPE

DGTS

Med A

Med A



The c-type cyts profiles of OL-less R. capsulatusmutants are partially restored upon growth at lowtemperature in enriched medium

The Ps and Nadi phenotypes of, and the amounts ofvarious c-type cyts in, R. capsulatus mutants lacking OLwere also examined after growth at 25°C in different

media (Fig. 2, right panel). When the mutants were grownat 25°C in enriched MPYE medium, instead of 35°C, theyregained the Nadi+ phenotype and contained membrane-associated c-type cyts, but at amounts lower than wildtype (Fig. 9A, lane 2). Conversely, at 25°C on minimalMedA medium, they remained Nadi– and contained evenlower amounts of various c-type cyts than at 35°C (Fig.

100 200 300 400 500 600 700 800

115.1

70

m/z, amu

379.4

133

397.5

415

O

O

NHNH

HOOCHOOC

O

NHNH2

670 680 690 700 710 720

2

4

6

8

10x 104

679.7

2

4

6

8

10x 10

4

Inte

nsity

, cps

m/z, amu

679.7

Intensity, cps

361.4

415

70/71 Orn immonium ion

397

133 [Orn+H]+

H_

115/116 Orn B ion

679.7

H O2_

Fig. 8. Chemical structure of R. capsulatus OL determined by mass spectrometry. Silica gel corresponding to the OL spot on 2D-TLC platesdeveloped as in Figs 6 and 7, was scraped off, extracted and submitted to MS and MS/MS analyses. Control samples were prepared fromboth wild type (Sm1021) and OlsA– (ORLD1) S. meliloti strains (not shown) and compared with wild type (MT1131) and OlsBA– (SA8)R. capsulatus samples. The insets at the right hand side show the major ion that is present in the wild type (bottom trace) but absent in thecorresponding mutant (top trace) in the m/z range of 670–720, identifying a major ion of 679.7 in R. capsulatus as compared with 693.9 inS. meliloti extracts. This ion was subjected to fragmentation and the product ions generated were analysed by MS/MS. The data obtained withS. meliloti were identical to those reported previously (Geiger et al., 1999; Weissenmayer et al., 2002) and used as controls. A similarfragmentation pattern up to m/z of 415 [attributed to LOL (Weissenmayer et al., 2002)] was observed with both species (not shown), leadingus to a tentative structure for R. capsulatus OL, based on the observed m/z difference of 14 compared with that of S. meliloti and the knownabsence of cyclopropane fatty acids in the former species (Grogan and Cronan, 1997).

ccp1

yo

cc

32

25

MPYE Med A

wt ΔolsA wt

25 Cο

PE

PG PC

OL

O

DGTSPE

PG PC

O

DGTS

MPYE, 25 Cο MPYE, 25 Cο

Rc-ΔolsARc-wild typeBΔolsA

MPYE, 35 Cο

Rc-ΔolsA

PE

PG

DGTS

PC

O

DMPE

A C D

Fig. 9. Membrane-associated c-type cyts profiles and total lipid composition of R. capsulatus OlsA– mutants grown at 25°C in enriched MPYEor minimal MedA media. Membrane fractions and total lipids extracts were prepared and analysed from R. capsulatus wild-type (Rc-wild type,MT1131) and OlsA– mutant (Rc-DolsA, SA4) strains as in Fig. 5, except that cells were grown at 25°C instead of 35°C. Note the increasedsteady-state amounts of various c-type cyts when cells are grown at 25°C in enriched MPYE media (panel A, lane 2), despite the absence ofOL (panels C and D). Various c-type cyts and lipids are as indicated in Figs 5 and 6 respectively.



9A, lane 4). Furthermore, these phenotypes were some-what reversible: if the R. capsulatus mutants lacking OLwere grown on enriched MPYE medium plates for2–3 days at 25°C (i.e. initially Nadi+ and contained c-typecyts) and then incubated at 35°C for another day, theyended up being Nadi– and lacked various c-type cyts.Conversely, if they were grown at 35°C on enriched MPYEmedium for 2 days (i.e. initially Nadi– and lacked variousc-type cyts) and then incubated at 25°C for another1–2 days, they became Nadi+ and contained c-type cyts.Thus, cells lacking OL contained different amounts ofc-type cyts when different temperature and media combi-nations were used for growth (Figs 6 and 9), indicatingthat the severity of the phenotypes manifested, and pos-sibly the function of OL, was tightly linked to the physi-ological growth conditions of R. capsulatus cells.

Absence of OL decreases the production of cyt cy

during growth in minimal medium at 35°C

Whether lower amounts of various c-type cyts found inOL-less mutants was due to the decreased production, orincreased degradation, of their mature forms wasaddressed using pulse-chase experiments, as describedin Experimental procedures. During growth in minimalmedium at 35°C, the rate of degradation in an olsBAdouble mutant (SA8) of a carboxyl-terminally FLAGepitope-tagged (and functional) variant of R. capsulatuscyt cy (Myllykallio et al., 1997) was compared with that inits wild-type parent MT1131 (Table 1). The data indicatedthat, upon pulse labelling, mutant cells lacking OL pro-duced much less of a full-length cyt cy-FLAG protein (asrecognized by anti-FLAG antibodies) in comparison with awild-type strain. On the other hand, the rate of cyt cy

degradation remained similar in both the mutant and wild-type strains up to 480 min post chase period (Fig. 10, onlythe first 60 min post chase period is shown). Thus, duringgrowth at 35°C in minimal MedA medium, absence of OLaffects more severely the production, rather than the deg-radation, of the cyt cy. This finding implies that OL appearsto be required for some aspect(s) of the biogenesisprocess of at least the cyt cy, which is an N-terminallymembrane anchored protein that acts as an electroncarrier (Turkarslan et al., 2006).

Optimal amounts of various membrane proteins that arenot c-type cyts are also affected in OL-less cells grownin enriched medium at 35°C

Total membrane proteins profiles of a mutant lacking OL,such as SA8 (DolsBA) and its wild-type parent MT1131grown in enriched MPYE medium at 35°C were also com-pared using SDS-PAGE analyses, and drastic changes inthe amounts of a variety of membrane-associated pro-

teins of unknown identities were found (Fig. 11). Remark-ably, in OL-less cells the amounts of several membraneproteins decreased, while those of others increased sig-nificantly, indicating that the defects induced by theabsence of OL affected many extracytoplasmic proteins,including the c-type cyts, when cells are grown in enrichedmedium at 35°C. These findings further illustrate theimportance of this non-phosphorus lipid in R. capsulatusand possibly other bacterial cells.

Discussion

This work demonstrated that mutations in either one ofR. capsulatus RRC00138 (olsA) and RRC00139 (olsB)genes abolished the synthesis of OL, and greatlydecreased the amounts of various c-type cyts and otherproteins in cell membranes, especially during growth at

B

C

0 1’ 2’ 5’ 10’ 30’ 60’

c

c

y

y

wt/cyt c -FLAG

ΔolsBA/cyt c -FLAG

chase time

0 1’ 2’ 5’ 10’ 30’ 60’chase time

y

y

A

cy

wt/cyt c -FLAGywt

U I I U

Fig. 10. Production and degradation of cyt cy in an OL-lessR. capsulatus mutant grown at 35°C in minimal MedA medium. TheR. capsulatus wild type (wt, MT1131) (panels A and B) and OL-less(DolsBA, SA8) (panel C) strains harbouring plasmid pHM7 thatcarry a carboxyl-terminally FLAG epitope-tagged version of cyt cy,along the wild-type strain MT1131 lacking pHM7 (cyt cy-FLAG) as acontrol (panel A), were pulse-labelled for 5 min using [14C] proteinhydrolysate, and then chased with 0.1% non-radioactive casaminoacids. The amount of cyt cy-FLAG recognized by a gel boundanti-FLAG antibody was monitored in function of time during thechase. The pellets corresponding to the gel boundimmunocomplexes (I) and their supernatants (U) were analysed bySDS-PAGE, and visualized using a phosphoimager, as described inthe Experimental procedures. For each time point, a total cell lysatecontaining approximately 1.25 ¥ 106 total cpm was used, and thecompleteness of cyt cy-FLAG immunoprecipitation was confirmedby re-incubating the supernatants with a fresh batch of anti-FLAGantibody.



35°C on enriched medium. Genetic complementation ofappropriate mutants with the corresponding genes con-comitantly restored both the protein and lipid defects.Thus, the two seemingly unrelated processes, namely OLbiosynthesis and the steady-state amounts of some extra-cytoplasmic proteins, including various c-type cyts, areintertwined under appropriate physiological conditions(Fig. 12). Interspecies complementation betweenS. meliloti (Weissenmayer et al., 2002; Gao et al., 2004)and R. capsulatus further established that RhodobacterRRC00139 and RRC00138 products are functional homo-logues of Rhizobial OlsB and OlsA, encoding theN-acyltransferase and O-acyltransferase catalysing OLbiosynthesis respectively. Mass spectrometry data sug-gested that although LOL produced by S. meliloti andR. capsulatus are chemically identical, the major lipid

products are slightly different (Fig. 8). Unlike the major OLspecies of S. meliloti, that of R. capsulatus lacks a cyclo-propane ring in its second acyl chain, in agreement withthe absence of a fatty acid cyclopropane synthase in thegenome of the latter species. Furthermore, we note that inthe absence of olsA neither R. capsulatus nor S. melilotimutants accumulate LOL, and that R. capsulatus olsBAoverproduce OL in S. meliloti. This finding suggests eitherthat OlsB activity is tightly regulated, or that LOL is quicklydegraded if not converted to OL. In contrast to OlsB,S. meliloti OlsA is unable to complement an appropriateR. capsulatus mutant, unlike its R. capsulatus counterpartthat could do so in S. meliloti (Weissenmayer et al., 2002),raising the possibility that OlsA enzymes from thesespecies may be different (Fig. 12).

Ornithine lipid is widespread among Gram-negativeand Gram-positive free-living bacteria, human pathogensand plant symbionts, including Pseudomonads, Mycobac-teria, Streptomyces and Rhizobia (Asselineau, 1991). Itelicits strong host immune responses in mammals,ranging from B-lymphocyte mitogenicity, adjuvanticity,haemagglutination and hypothermia (Kawai et al., 1996),reactive oxygen radicals production (Kawai et al., 2000a),macrophage activation via the Toll like receptor CD14-dependent pathway and interleukin-1 and prostaglandinE2 production (Kawai et al., 2000b). In bacteria, OL canact as a substitute for phospholipids under limiting phos-phate availability, but its specific function is poorly defined(Lopez-Lara et al., 2005; Rojas-Jimenez et al., 2005). Ourfindings now indicate for the first time that this aminoacid-derived non-phosphorus membrane lipid plays,either directly or indirectly, a crucial role for optimalsteady-state amounts of several extracytoplasmic pro-teins, including c-type cyts, under specific physiologicalconditions.

The complex and reversible phenotypes displayed byOL-less R. capsulatus mutants in response to the growthtemperature and the nature of growth medium are trulyremarkable. While the absence of OL apparently perturbsthe production of at least the cyt cy in OL-lessR. capsulatus cells grown in minimal medium at 35°C,whether this is also the case for other c-type cyts remainto be seen. Furthermore, when the same OL-less cells aregrown in enriched medium at 35°C, the steady-stateamounts of a number of extracytoplasmic proteins, includ-ing various c-type cyts, change more severely (Fig. 11).Thus, whether the absence of OL perturbs only the ratesof production, or also the rates of post-synthesis degra-dation, or both, of specific membrane proteins under dif-ferent physiological conditions, and if so, what are themolecular mechanism(s) underlying these changes areimportant questions, but are beyond the scope of thiswork. Clearly though, the drastic changes observed inR. capsulatus cells lacking OL underscore the importance

ΔolsBA

250148

60

30

42

22

WT

Fig. 11. Comparison of the membrane protein profiles of wild-typeand OL-less mutant strains of R. capsulatus. The R. capsulatuswild type (wt, MT1131) and OL-less (DolsBA, SA8) strains weregrown at 35°C in enriched MPYE medium, and intracytoplasmicmembrane vesicles were prepared as described in theExperimental procedures, and 50 mg of total proteins was subjectedto 15% SDS-PAGE analyses and stained with Coomassie brilliantblue. Increased (white arrowheads) and decreased (black arrowheads) steady-state amounts of various extracytoplasmic proteinsseen in the OL-less membranes are indicated on the left along themolecular weight markers used.



of this non-phosphorus membrane lipid in this, and pos-sibly in other, bacterial species.

What could be the role of a minor lipid like OL in bac-teria? In general, polar lipids of biological membranesprovide a permeability barrier and a fluid structure thatserves as a matrix for membrane-associated proteins(Raetz and Dowhan, 1990; Cronan, 2003). Certain lipids,due to their physicochemical properties, appear to playkey roles in the assembly of specific proteins into mem-branes to achieve critical functions, like cell division orDNA replication (Dowhan et al., 2004). Some lipids havealso been implicated in the formation of supramolecularassemblies of membrane proteins of related functions,like those of the respiratory or photosynthetic chains(Zhang et al., 2002), and defined interactions betweenphospholipids and proteins have been described (vanDalen and Kruijff, 2004; Dowhan et al., 2004). For

example, membrane proteins often cocrystallize with spe-cific lipids such as CL, sugar lipids or sulpholipids (Langeet al., 2001; Roszak et al., 2003), suggesting that protein–lipid interactions might be critical for their localization,folding, stability, assembly and enzymatic activity(Bogdanov and Dowhan, 1999; Sedlak and Robinson,1999). Thus, the absence of OL might induce drasticchanges in the properties of cell membranes, which inturn could manifest complex growth-dependent pheno-types, as observed in this work. Although the mechanismsunderlying these changes are not yet known, the pheno-types appear to be indirect manifestations of perturbedphysiologies of OL-less cells, and are likely to depend onadditional cellular component(s).

What could be the component(s) that causes the sub-optimal amounts of various c-type cyts, and possibly othermembrane proteins, in the absence of OL under various

cy

c2

cO cP

cbb3

bc1

bc1native

OL-less

Lyso-ornithine lipid (LOL)

Ornithine lipid (OL)

Ornithine

HOOC

NH2

NH2

NH2

NH

OHOOC

OH

NH2

NH

O

OO

HOOC

OlsB

3-hydroxy-acyl-ACP

ACP

OlsA

acyl-ACP ACP

cpc1

c1

cyc0

Fig. 12. Ornithine lipid biosynthesis in R. capsulatus and growth medium and temperature-dependent effects of the absence of OL on thepresence of various c-type cyts. In native membranes (top), OL and phospholipids are represented as yellow hexagons and green circlesrespectively. The c-type cyts containing membrane proteins are shown using the high-resolution structures, except for the cbb3-Cox (cbb3) ofR. capsulatus proteins. Only a small amount of cyt bc1 (bc1) is present in OL-less membranes (bottom). Membrane-associated c-type cytsprofiles in native and OL-less mutants, showing from the top to bottom the cyts cp, c1, cy and co bands stained using tetramethylbenzidine, andOL biosynthetic pathway are shown in bottom left and right respectively, to illustrate the previously unknown link between OL biosynthesis andc-type cyts biogenesis.



physiological conditions? A possibility is that in cellsgrowing in minimal medium at 35°C, OL is requireddirectly or indirectly for a yet to be identified step(s), orcomponent(s), of membrane protein production. In cellsgrown at 35°C on minimal MedA medium, lower steady-state amounts of the N-terminally membrane-anchoredcyt cy apparently stem from its decreased production,rather than increased turnover, as indicated by pulse-chase experiments (Fig. 10). Although to what extent thisis also the case for other c-type cyts remains to be seen,conceivably, perturbation of such component(s) in theabsence of OL could lead to decreased production ofvarious c-type cyts and other membrane proteins, andcomplex Ps and Nadi phenotypes, in manners reminis-cent of mutants defective in c-type cyts maturation (Langet al., 1996; Sanders et al., 2005). An additional possibilityis that, in the absence of OL, various c-type cyts and othermembrane proteins might also be rapidly degraded due toenvelope-related stresses (Raivio, 2005), caused bymodified membrane lipid compositions and differentgrowth conditions. The extremely low amounts of somec-type cyts in OL-less R. capsulatus cells grown inenriched MPYE medium at 35°C are consistent with thispossibility. It is known that growth temperatures influencebacterial membrane lipids composition and fluidity, andspecific lipids often affect optimal growth temperatures ofbacteria (Cronan, 2003; Los and Murata, 2004). In manyspecies, including R. capsulatus, higher growth tempera-tures induce multiple heat shock proteins of which severalare chaperones and proteases, such as DegP, involved inextracytoplasmic protein degradation, whose activitiesare temperature modulated (Onder et al., in preparation).Clearly, not only the phenotypes of the OL-lessR. capsulatus mutants are complex, but also, the mecha-nisms underlying these phenotypes might be multifac-eted, and even different, under various physiologicalconditions. In any event, this work has began to illustratethe importance of OL in R. capsulatus, and future studiesof mutants dissecting the links between membrane lipidsynthesis and the steady-state amounts of some extracy-toplasmic proteins, including c-type cyts, should prove tobe rewarding.

Experimental procedures

Bacterial strains and growth conditions

The bacterial strains and plasmids used are listed in Table 1.E. coli strains were grown aerobically in Luria–Bertani (LB)medium (Myllykallio et al., 1997), and R. capsulatus strainswere grown under respiratory or photosynthetic conditions,either at 25°C or 35°C as appropriate, in either minimal (MedA, containing succinate as carbon and NH4

+ as nitrogensources) (Sistrom, 1960) or enriched (MPYE, containingcomplex carbon and nitrogen sources provided by yeast

extracts and peptones) media supplemented with appropriateantibiotics, as described earlier (Myllykallio et al., 1997).

Molecular genetic techniques

Isolation of Nadi– (i.e. cbb3-Cox-minus) mutants has beendescribed previously (Koch et al., 1998), except that for newlyisolated mutants the ccoNOQP diploid strain MT1131/pOX15was used as a parent (Table 1). Standard molecular biologi-cal techniques were performed according to Sambrook et al.(1989), and all chromosomal insertion or insertion-deletionalleles were constructed by interposon mutagenesis usingthe spectinomycin-resistance (SpeR) cartridge from pHP45W(Prentki and Krisch, 1984) and the gene transfer agent (GTA),which is a phage-like particle capable of gene transduction(Yen et al., 1979). Plasmid pSEM1 was obtained by deletingthe 1.2 kb BamHI fragment of pMRC. An insertion mutation inORF2 (orf2::spe) was obtained by ligating the 2.0 kb SmaIcleaved SpeR cartridge from pHP45W into the blunt-endedunique XhoI site of ORF2 carried by pSEM1. pSEM3 carryinga deletion-insertion mutation covering both ilvD and exoD[D(ilvD-exoD)::spe] was obtained by replacing the 1.36 kbBstEII fragment of pSEM1 the with the 2.0 kb SmaI cleavedSpeR cartridge from pHP45W after blunting the ends asappropriate. An insertion allele of RRC00138 (olsA::spe) wasobtained by ligating the 2.0 kb BamHI cleaved SpeR cartridgefrom pHP45W into its unique BamHI site carried by pMRC,yielding pSEM4 (Fig. 4). To obtain a deletion-insertion alleleof RRC00138 [D(olsA::spe)], first the 5.3 kb EcoRV-HindIIIfragment of pMRC was subcloned into the same sites ofpBR322 to yield pSEM5. Then, the 0.52 kb BstXI-NotI blunt-ended fragment of pSEM5 was replaced with the 2.0 kb SmaIcleaved SpeR cartridge from pHP45W, yielding pSEM10. Adeletion-insertion allele of RRC00139 [D(olsB::spe)] wasobtained by substituting the 0.32 kb RsrII blunt-ended frag-ment internal to RRC00139 on pSEM5 with the 2.0 kb SmaIcleaved SpeR cartridge from pHP45W, yielding pSEM8. Aninsertion-deletion allele of ORF3 [D(orf3::spe)] was con-structed by replacing the 0.02 bp HpaI blunt-ended fragmentof ORF3 on pSEM5 by the 2.0 kb SmaI cleaved SpeR car-tridge from pHP45W, to yield pSEM6. Finally, a deletion-insertion allele covering both RRC00139 and RRC00138[D(olsBA::spe)] was constructed by replacing the 2.05 kb MluIblunt-ended fragment of pSEM5 with the 2.0 kb SmaI cleavedSpeR cartridge from pHP45W to yield pSEM12. Transferableplasmids carrying polar deletion-insertion alleles ofRRC00138 [D(olsA::spe)], RRC00139 [D(olsB::spe)], ORF3[D(orf3::spe)] and RRC00139-RRC00138 [D(olsBA)::spe)]were constructed by cloning the 5.32 kb EcoRV-EcoRI blunt-ended fragment of pSEM10 into the BamHI site, 2.75 kbXhoI-Bsu36I blunt-ended fragment of pSEM8 into the BamHIsite, the 2.63 kb PstI fragment of pSEM6 into the PstI site,and the 4.99 kb EcoRV-Bsu36I blunt-ended fragment ofpSEM12 into the BamHI site of the plasmid pRK404 (Dittaet al., 1985) or pRK415, to generate pSEM11, pSEM9,pSEM7 and pSEM13 respectively.

The plasmids pSEM2, pSEM3, pSEM4, pSEM7, pSEM9,pSEM11 and pSEM13 were conjugated into the GTA over-producer strain Y262. Following appropriate GTA crossesinto the wild-type strain MT1131 and selecting for SpeR

colonies, the single mutants SA2 (ORF2::spe), SA3



[D(ilvD-Exod)::spe], SA1 (olsA::spe), SA4 [D(olsA::spe)], SA5[D(ORF3::spe)], SA6 [D(olsB::spe)], and the double mutantSA8 [D(olsBA)::spe)] were obtained (Fig. 4 and Table 1). TheRRC00139 gene was PCR amplified using the plasmidpMRC (Fig. 4, Table 1) as a template, and the primer pairsRRC00139-NcoI (5�-CTTTGACCATGGCGGCAGCACCGC-3�) and RRC00139-SacI (5�-GCCAGCAGAGCTCCGCGTCCCTCCAC-3�) containing the NcoI and SacI restrictionenzymes sites, PCR products were digested with the appro-priate restriction enzymes, cloned into the same sites of theexpression vector pBAD/Myc-HisA (Invitrogen), yieldingpSEM19 (RRC00139), confirmed by DNA sequencing, andcloned into the broad host-range plasmid pRK415, yieldingpSEM20 (Table 1). Automated DNA sequencing with the Big-dye terminator cycle sequencing kit (Applied Biosystems)was used with the primers pBAD-Seq-F (5�-ATGCCATAGCATTTTTATCC-3�), pBAD-Seq-R (5�-GATTTAATCTGTATCAGG-3�). Homology searches and amino acid sequencealignments were done using MacVector (Accelerys) andappropriate software programs as described earlier (Myllykal-lio et al., 1997).

Biochemical techniques

Rhodobacter capsulatus intracytoplasmic membranevesicles (chromatophores) were prepared in 50 mM MOPS(pH 7.0), 1 mM KCl, 10 mM EDTA and 1 mM PMSF using aFrench pressure cell, and protein concentrations were deter-mined as described earlier (Gray et al., 1994). One hundredmicrograms of membrane proteins per lane was incubated at37°C for 10 min in the sample loading buffer, and after sepa-ration by 16.5% T (w/v) Tricine SDS-PAGE (Schägger andVon Jagow, 1987), cyts were visualized using TMBZ andH2O2 (Thomas et al., 1976). Combined GPAT and AGPATactivities were measured using [14C(U)]-G3P (NEN™ LifeScience) and cis-18:1-(D11) acyl carrier protein (cis-vaccenyl-ACP), as described by Goldfine (1969). cis-vaccenyl-ACPwas prepared by incubating 30 mg of vaccenic acid and 5 mgof E. coli ACP (Sigma-Aldrich) with excess of ACP acyltrans-ferase activity overproducing E. coli extracts (Clementz et al.,1996), purified through a DEAE-column, and its concentra-tion determined by titration with a sulphhydryl reagent(Ellman, 1959). LPA and PA produced during the combinedGPAT and AGPAT assay were analysed by TLC on Silica GelH plates developed with chloroform : methanol : acetic acid(39:9:3). TLC plates were scanned using a phosphoimager(Molecular Dynamics), and the radioactive spots were iden-tified by comparison with unlabelled LPA and PA (10 mg each)used as standards. Decylbenzohydroquinone: cyt c oxi-doreductase, b-galactosidase and alkaline phosphataseactivities were determined as described earlier (Gray et al.,1994; Myllykallio et al., 1997).

Total lipid analyses

Cellular lipids were extracted according to Benning andSomerville (1992) from 50 ml of cultures of appropriateR. capsulatus strains grown up to an A630 of 0.7–0.8. Briefly,cells were washed and resuspended in 0.5 ml of H2O, andlipids were extracted with 4 ml of chloroform : methanol (1 : 1

v/v). After adding 1.3 ml of a solution containing 1 M KCl and0.2 M H3PO4 and vigorous vortexing, samples were centri-fuged at 8000 g for 15 min, the lower chloroform phase con-taining the lipids was evaporated under a stream of argon,and the residue dissolved in chloroform at a final lipid con-centration of 10 mg ml-1. As needed, cellular lipids wereradiolabelled by growing R. capsulatus cells for 24 h in 1 mlof minimal medium (Med A) or 30 h in 1 ml of enrichedmedium (MPYE) supplemented with 2 mCi of [1-14C]-acetate(60 mCi mmol-1 specific activity). Labelled cells were resus-pended in 100 ml of H2O and lipids were extracted asdescribed in Bligh and Dyer (1959) using a mixture of 100 mlof chloroform and 200 ml of methanol, followed by an addi-tional extraction with 100 ml of chloroform. The chloroformphases were combined, evaporated under a stream of argon,and extracted lipids were dissolved in 100 ml of chloroform.2D-TLC was performed according to de Rudder et al. (1997)using either radiolabelled (60 000 total cpm) or non-labelledlipids (10 mg) deposited on heat-activated silica gel G60plates (Sigma-Aldrich). Plates were developed with chloro-form : methanol : water (14:6:1, v/v/vol) and chloroform :methanol : glacial acetic acid (13:5:2, v/v/vol) for the first andsecond dimensions respectively. Phospholipids and amine-containing lipids were detected using 0.3% (w/v) molybde-num blue and 0.3% ninhydrin in ethanol and heating at 120°Cfor 5–10 min respectively. PE, PG, PC and CL were identifiedby comparison with appropriate reference compounds(Sigma-Aldrich), and radiolabelled lipids visualized and quan-tified by using a phosphoimager equipped with theImageQuant software (Molecular Dynamics).

Mass spectrometry

The [1-14C]-acetate-labelled OL-containing spots werescraped off the TLC plates, extracted from silica gel with500 ml of chloroform : methanol (30:70, v/v), 0.1% formic acid(adjusted to pH 5.6 with NH4OH), centrifuged and the super-natants were analysed on an electrospray ionization massspectrometry with a turbo ion spray source on a QTrap(Applied Biosystems, Foster City, CA). Samples were directlyintroduced at a flow rate of 5 ml min-1, and a positive Q1 scanof 600–800 m/z was performed with a declustering potentialof 90 V, source voltage of 5500 V, curtain gas of 25, entrancepotential of 10 and an ion energy of 1. Fragmentation ofselected ions were performed using enhanced product ionscans, between the ranges of 50–800 m/z with a declusteringpotential of 50, entrance potential of 7, collision gas of 10,and collision energies of 60 and 45 for the ions 679.7 and693.7 m/z respectively.

Pulse-chase labelling and immunoprecipitation

The degradation rate of a carboxyl-terminally FLAG epitope-tagged and functional cyt cy protein (Myllykallio et al., 1997)was examined by pulse-chase labelling and immunoprecipi-tation using Anti-Flag affinity gel. The R. capsulatus wild-type(MT1131) and OlsBA– double mutant (SA8) strains carryingplasmid pHM7 (cyt cy-FLAG fusion construct, Table 1) weregrown in minimal Med A medium at 35°C to an optical densityat 630 nm of 0.75. Cells were pulse-labelled for 2, 5 or 15 min



with 1 mCi ml-1 final concentration of 14C-amino acid mixture,and chased with unlabelled casamino acids (0.1% final con-centration) either in the presence or absence of a proteinsynthesis inhibitor (200 mg ml-1 of chloramphenicol). Aliquots(1 ml) were removed at various times points (0, 0.5, 1, 2, 5,10, 30, 60, 120, 240 and 480 min) and placed on ice. Cellswere then pelleted, washed once with cold medium asneeded, resuspended in 100 ml of lysis buffer (CelLytic™ B2¥ Cell lysis reagent, Sigma B7310) supplemented withDNase (50 mg ml-1), RNase (50 mg ml-1), phenylmethylsul-phonic acid (1 mM) and e-amino-n-caproic acid (0.1 mM),lysed by vortexing for 2 min and incubation for 10 min at roomtemperature after addition of 0.2 mg ml-1 lysozyme. Then10 mM EDTA was added, and cell debris were pelleted at13.000 rpm for 15 min. Supernatants were transferred tofresh tubes containing 40 ml each of Anti-Flag M2 affinitygel (Sigma A2220) and incubated at 4°C on a rocker for3 h. Immunocomplexes were recovered by centrifugationand washed five times with 1¥ TBS (50 mM Tris-HClpH 7.4, 150 mM NaCl) buffer supplemented with 0.01%dodecylmaltoside. Washed pellets were resuspended in 1¥SDS-sample buffer [37.5 mM Tris, pH 7, 3% SDS, 7.5%glycerol (w/v), 0.0125% bromophenol blue and 2%b-mercaptoetanol], boiled and analysed by 15% SDS-PAGE.Gels were fixed in 25% methanol – 10% acetic acid for20 min, dried, and applied to a phosphoimager cassette.Radioactive protein bands were visualized, and quantified asneeded, by using a phosphoimager equipped with theImageQuant software (Molecular Dynamics).

Chemicals, reagents and enzymes

Restriction enzymes were purchased from New EnglandBiolabs. Radioactive chemicals, [1-14C]-acetate, L-[14C(U)]glycerol 3-phosphate disodium salt (G3P), and [14C(U)]protein hydrolysate (specific activity � 1.94GBq/miliatomcarbon) were purchased from Perkin Elmer, NEN™ LifeScience Products and Amersham Biosciences respectively.All other chemicals were from commercial sources and ofhighest available purity.

Acknowledgements

This work was supported by NIH Grants GM38237 (to F. D),AI45153 (to H. G) and AG20238 (to P. H. Axelsen), DOEGrant ER 9120053 (to F. D), TUBITAK Grant TBAG-2128 (toS. M), UNESCO-L’Oreal Co-Sponsored Fellowship ForYoung Women in Life Sciences (to S. A-S), DeutscheForschungsgemeinschaft/SFB388 (to H.-G. K), AmericanHealth Assist Foundation (to I.V.J. M) and Alzheimer’s Asso-ciation (to P.H. Axelson). We thank O. Geiger (UniversidadNacional Autónoma de México) for providing appropriateS. meliloti strains and plasmids, and N. Johnston (Universityof Pennsylvania) for advice concerning lipid isolation andchromatography.

References

Asselineau, J. (1991) Bacterial lipids containing amino acidsor peptides linked by amide bonds. Fortschr Chem OrgNaturst 56: 1–85.

Benning, C., and Somerville, C.R. (1992) Isolation andgenetic complementation of a sulfolipid-deficientmutant of Rhodobacter sphaeroides. J Bacteriol 174:2352–2360.

Benning, C., Huang, Z.H., and Gage, D.A. (1995) Accumu-lation of a novel glycolipid and a betaine lipid in cells ofRhodobacter sphaeroides grown under phosphatelimitation. Arch Biochem Biophys 317: 103–111.

Bligh, E.G., and Dyer, J.W. (1959) A rapid method of totallipid extraction and purification. Can J Biochem Physiol 37:911–917.

Bogdanov, M., and Dowhan, W. (1999) Phospholipid-assisted refolding of an integral membrane protein. J BiolChem 274: 12339–12345.

Clementz, T., Bednarski, J.J., and Raetz, C.R.H. (1996)Function of the htrB high temperature requirement gene ofEscherichia coli in the acylation of lipid A. J Biol Chem 271:12095–12102.

Coleman, J. (1990) Characterization of Escherichia coli cellsdeficient in 1-acyl-sn-glycerol-3 phosphate acyltransferaseactivity. J Biol Chem 265: 17215–17221.

Cronan, J.E., Jr (2003) Bacterial membrane lipids: where dowe stand? Annu Rev Microbiol 57: 203–224.

van Dalen, A., and de Kruijff, B. (2004) The role of lipids inmembrane insertion and translocation of bacterial proteins.Biochim Biophys Acta 1694: 97–109.

Ditta, G., Schmidhauser, T., Yacobson, E., Lu, P., Liang, X.,Finlay, D.R., et al. (1985) Plasmids related to the broadhost range vector, pRK290, useful for gene cloning and formonitoring expression. Plasmid 13: 149–153.

Dowhan, W., Mileykovskaya, E., and Bogdanov, M. (2004)Diversity and versatility of lipid–protein interactionsrevealed by molecular genetic approaches. BiochimBiophys Acta 1666: 19–39.

Ellman, G.L. (1959) Tissue sulfhydryl groups. Arch BiochemBiophys 82: 70–77.

Gao, J.-L., Weissenmayer, B., Taylor, A.M., Thomas-Oates,J., Lopez-Lara, I.M., and Geiger, O. (2004) Identification ofa gene required for the formation of lyso-ornithine lipid, anintermediate in the biosynthesis of ornithine-containinglipids. Mol Microbiol 53: 1757–1770.

Geiger, O., Rohrs, V., Weissenmayer, B., Finan, T.M., andThomas-Oates, J. (1999) The regulator gene phoB medi-ates phosphate stress-controlled synthesis of the mem-brane lipid diacylglyceryl-N,N,N-trimethylhomoserine inRhizobium (Sinorhizobium) meliloti. Mol Microbiol 32:63–73.

Goldfine, H. (1969) Filter paper disk assay for lipid synthesis.Methods Enzymol 14: 649–651.

Gray, K.A., Grooms, M., Myllykallio, H., Moomaw, C., Slaugh-ter, C., and Daldal, F. (1994) Rhodobacter capsulatus con-tains a novel cb-type cytochrome c oxidase without a CuA

center. Biochemistry 33: 3120–3127.Grogan, D.W., and Cronan, J.E., Jr (1997) Cyclopropane ring

formation in membrane lipids of bacteria. Microbiol Mol BiolRev 61: 429–441.

Jenney, F.E., and Daldal, F. (1993) A novel membrane-associated c-type cytochrome, cyt cy, can mediatethe photosynthetic growth of Rhodobacter capsulatusand Rhodobacter sphaeroides. EMBO J 12: 1283–1292.



Kawai, Y., Takasuka, N., Akagawa, K., and Naito, S. (1996)Hypothermic response of mice to ornithine-containinglipids and to endotoxin. Infect Immun 64: 2101–2105.

Kawai, Y., Okawarab, A.I., Okuyama, H., Kura, F., andSuzuki, K. (2000a) Modulation of chemotaxis, O(2)(-) pro-duction and myeloperoxidase release from human poly-morphonuclear leukocytes by the ornithine-containing lipidand the serine glycine-containing lipid of Flavobacterium.FEMS Immunol Med Microbiol 28: 205–209.

Kawai, Y., Takasuka, N., Inoue, K., Akagawa, K., and Nish-ijima, M. (2000b) Ornithine-containing lipids stimulateCD14-dependent TNF-alpha production from murinemacrophage-like J774.1 and RAW 264.7 cells. FEMSImmunol Med Microbiol 28: 197–203.

Keilin, D. (1966) The History of Cell Respiration andCytochromes. Cambridge: Cambridge University Press.

Koch, H.-G., Hwang, O., and Daldal, F. (1998) Isolation andcharacterization of Rhodobacter capsulatus mutantsaffected in cytochrome cbb3 oxidase activity. J Bacteriol180: 969–978.

Koch, H.-G., Winterstein, C., Saribas, A.S., Alben, J.O., andDaldal, F. (2000) Roles of the ccoGHIS gene products inthe biogenesis of the cbb3-type cytochrome c oxidase. JMol Biol 297: 49–65.

Kulajta, C., Thumfart, J.O., Haid, S., Daldal, F., and Koch, H.G.(2006) Multi-step assembly pathway of the cbb(3)-type cyto-chrome c oxidase complex. J Mol Biol 355: 989–1004.

Lang, S.E., Jenney, F.E., Jr and Daldal, F. (1996) Rhodo-bacter capsulatus CycH: a bipartite gene product withpleiotropic effects on the biogenesis of structurally differentc-type cytochromes. J Bacteriol 178: 5279–5290.

Lange, C., Nett, J.H., Trumpower, B.L., and Hunte, C. (2001)Specific roles of tightly bound phospholipids in the yeastcytochrome bc1 complex structure. EMBO J 20: 6591–6600.

Lopez-Lara, I.M., Gao, J.-L., Soto, M.J., Solares-Perez, A.,Weissenmayer, B., Sohlenkamp, C., et al. (2005)Phosphorus-free membrane lipids of Sinorhizobiummeliloti are not required for the symbiosis with Alfalfa butcontribute to increased cell yields under phosphorus-limiting conditions of growth. Mol Plant Microbe Interact 18:973–982.

Los, D.A., and Murata, N. (2004) Membrane fluidity and itsroles in the perception of environmental signals. BiochimBiophys Acta 1666: 142–157.

Minnikin, D.E., and Abdolrahimzadeh, H. (1974) The replace-ment of phosphatidylethanolamine and acidic phospholip-ids by an ornithine-amide lipid and a minor phosphorous-free lipid in Pseudomonas fluorescens NCMB 129. FEBSLett 43: 257–260.

Moore, G.R., and Pettigrew, G.W. (1990) Cytochromes c:Evolutionary, Structural and Physicochemical Aspects.Berlin: Springer-Verlag.

Myllykallio, H., Jenney, F.E., Jr, Moomaw, C.R., Slaughter,C.A., and Daldal, F. (1997) Cytochrome cy of Rhodobactercapsulatus is attached to the cytoplasmic membrane by anuncleaved signal sequence-like anchor. J Bacteriol 179:2623–2631.

Porat, A., Cho, S.H., and Beckwith, J. (2004) The unusualtransmembrane electron transporter DsbD and its homo-logues: a bacterial family of disulfide reductases. ResMicrobiol 155: 617–622.

Prentki, P., and Krisch, H.M. (1984) In vitro insertionalmutagenesis with a selectable DNA fragment. Gene 29:293–303.

Raetz, C.R.H., and Dowhan, W. (1990) Biosynthesis andfunction of phospholipids in Escherichia coli. J Biol Chem265: 1235–1238.

Raivio, T.L. (2005) Envelope stress responses and Gram-negative bacterial pathogenesis. Mol Microbiol 56: 1119–1128.

Rojas-Jimenez, K., Sohlenkamp, C., Geiger, O., Martinez-Romero, E., Werner, D., and Vinuesa, P. (2005) A CICchloride channel homolog and ornithine-containing mem-brane lipids of Rhizobium tropici CIAT899 are involved insymbiotic efficiency and acid tolerance. Mol Plant MicrobeInteract 18: 1175–1185.

Roszak, A.W., Howard, T.D., Southall, J., Gardiner, A.T.,Law, C.J., Isaacs, N.W., and Cogdell, R.J. (2003) Crystalstructure of the RC-LH1 core complex from Rhodop-seudomonas palustris. Science 302: 1969–1972.

de Rudder, K.E.E., Thomas-Oates, J.E., and Geiger, O.(1997) Rhizobium meliloti mutants deficient in phospholip-ids N-methyltransferase still contain phosphatidylcholine. JBacteriol 179: 6921–6928.

Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989) Molecu-lar Cloning: A Laboratory Manual. New York: Cold SpringHarbor Laboratory Press.

Sanders, C., Deshmukh, M., Astor, D., Kranz, R.G., andDaldal, F. (2005) Overproduction of CcmG and CcmFHRcfully suppresses the c-type cytochrome biogenesis defectof Rhodobacter capsulatus CcmI-null mutants. J Bacteriol187: 4245–4256.

Schägger, H., and Von Jagow, G. (1987) Tricine-sodiumdodecyl sulphate sulfate-polyacrylamide gel electrophore-sis for the separation of proteins in the range from 1 to 100kDa. Anal Biochem 166: 368–379.

Sedlak, E., and Robinson, N.C. (1999) Phospholipase A2

digestion of cardiolipin bound to bovine cytochrome coxidase alters both activity and quaternary structure. Bio-chemistry 38: 14966–14972.

Sistrom, W.R. (1960) A requirement for sodium in the growthmedium of Rhodopseudomonas sphaeroides. J GenMicrobiol 22: 77–85.

Stanley, P., Koronakis, V., and Hughes, C. (1998) Acylationof Escherichia coli hemolysin: a unique protein lipidationmechanism underlying toxin function. Microbiol Mol BiolRev 62: 309–333.

Thomas, P.E., Ryan, D., and Levin, W. (1976) An improvedstaining procedure for the detection of the peroxidaseactivity of cytochrome P-450 on sodium dodecyl sulfatepolyacrylamide gels. Anal Biochem 75: 168–176.

Turkarslan, S., Sanders, C., and Daldal, F. (2006) Extracyto-plasmic prosthetic group ligation to apoproteins: matura-tion of c-type cytochromes. Mol Microbiol 60: 537–541.

Weissenmayer, B., Gao, J.-L., Lopez-Lara, I.M., and Geiger,O. (2002) Identification of a gene required for the biosyn-thesis of ornithine-derived lipids. Mol Microbiol 45: 721–733.

Yen, H.C., Hu, N.T., and Marrs, B.L. (1979) Characterizationof the gene transfer agent made by an overproducermutant of Rhodopseudomonas capsulata. J Mol Biol 131:157–168.



Zannoni, D. (1995) Aerobic and anaerobic electron transportchains in anoxygenic phototrophic bacteria. In AnoxygenicPhotosynthetic Bacteria. Blankenship, R.E., Madigan,M.T., and Bauer, C.E. (eds). Dordrecht, The Netherlands:Kluwer Academic Publishers, pp. 949–971.

Zhang, M., Mileykovskaya, E., and Dowhan, W. (2002) Car-diolipin is required for supercomplex formation in theinner mitochondrial membrane. J Biol Chem 277: 43553–43556.



Ornithine lipid is required for optimal steady-state amounts of c-type cytochromes in Rhodobacter capsulatus

Documents