Báo cáo khoa học: Structural evidence of a-aminoacylated lipoproteins of Staphylococcus aureus pot

Figure 1A shows MALDI-TOF MS of the chloroform⁄ methanol organic phase representing a series of 14-Da interval peaks, explained by an increasing number of methylene CH2 groups in their f

Staphylococcus aureus

Miwako Asanuma1,*, Kenji Kurokawa2, Rie Ichikawa1, Kyoung-Hwa Ryu2, Jun-Ho Chae2,

Naoshi Dohmae1, Bok Luel Lee2and Hiroshi Nakayama1

1 Biomolecular Characterization Team, RIKEN Advanced Science Institute, Saitama, Japan

2 National Research Laboratory of Defense Proteins, Pusan National University, Busan, Korea

Introduction

Bacterial lipoproteins are lipidated proteins anchored

on the outside leaflet of bacterial cell membranes and

outer envelopes, and have diverse functions such as

nutrient uptake, cell-wall metabolism, adhesion and

transmembrane signaling The biosynthesis pathway of

bacterial lipoproteins has been established in

Escheri-chia coli and consists of three sequential enzymatic

reactions [1] Following apolipoprotein translocation

by Sec machinery, the first enzyme diacylglyceryl trans-ferase (Lgt) transfers a diacylglyceryl moiety from a membrane phospholipid to a sulfhydryl group of the +1 cysteine of a conserved ‘lipobox’ motif in the N-terminal signal peptide, making a thioether linkage The lipoprotein signal peptidase (Lsp) then cleaves the signal peptide at the N-terminus of the +1 S-diacylgly-ceryl cysteine Finally, the third enzyme apolipoprotein

Keywords

bacterial lipoprotein; Gram-positive bacteria;

mass spectrometry; N-acyltransferase; SitC

Correspondence

B L Lee, National Research Laboratory of

Defense Proteins, College of Pharmacy,

Pusan National University, Jangjeon Dong,

Geumjeong Gu, Busan 609-735, Korea

Fax: +82 51 513 2801

Tel: +82 51 510 2809

E-mail: brlee@pusan.ac.kr

H Nakayama, Biomolecular Characterization

Team, RIKEN Advanced Science Institute, 2-1

Hirosawa, Wako, Saitama 351-0198, Japan

Fax: +81 48 462 4704

Tel: +81 48 462 1419

E-mail: knife@riken.jp

*Present address

ERATO, Japan Science and Technology

Agency (JST), Tokyo, Japan

(Received 12 November 2010, revised

3 December 2010, accepted 9 December

2010)

doi:10.1111/j.1742-4658.2010.07990.x

Bacterial lipoproteins are known to be diacylated or triacylated and acti-vate mammalian immune cells via Toll-like receptor 2⁄ 6 or 2 ⁄ 1 heterodi-mer Because the genomes of low G+C content Gram-positive bacteria, such as Staphylococcus aureus, do not contain Escherichia coli-type apoli-poprotein N-acyltransferase, an enzyme converting diacylated liapoli-poproteins into triacylated forms, it has been widely believed that native lipoproteins

of S aureus are diacylated However, we recently demonstrated that one lipoprotein SitC purified from S aureus RN4220 strain was triacylated Almost simultaneously, another group reported that another lipoprotein SA2202 purified from S aureus SA113 strain was diacylated The determi-nation of exact lipidated structures of S aureus lipoproteins is thus crucial for elucidating the molecular basis of host–microorganism interactions Toward this purpose, we intensively used MS-based analyses Here, we demonstrate that SitC lipoprotein of S aureus RN4220 strain has two lipo-protein lipase-labile O-esterified fatty acids and one lipolipo-protein lipase-resis-tant fatty acid Further MS⁄ MS analysis of the lipoprotein lipase digest revealed that the lipoprotein lipase-resistant fatty acid was acylated to a-amino group of the N-terminal cysteine residue of SitC Triacylated forms of SitC with various length fatty acids were also confirmed in cell lysate of the RN4220 and Triton X-114 phase in three other S aureus strains, including SA113 strain and one Staphylococcus epidermidis strain Moreover, four other major lipoproteins including SA2202 in S aureus strains were identified as N-acylated These results strongly suggest that lipoproteins of S aureus are mainly in the N-acylated triacyl form

Abbreviations

BHI, Brain Heart Infusion; LB, Luria–Bertani; Lgt, diacylglyceryl transferase; Lnt, apolipoprotein N-acyltransferase; LPL, lipoprotein lipase; Pam3, N-palmitoyl-S- dipalmitoylglyceryl; TLR, Toll-like receptor; TX114, Triton X-114.

N-acyltransferase (Lnt) transfers an acyl group from

phospholipid to the newly generated a-amino group of

the S-diacylglyceryl cysteine (N-acylation reaction),

resulting in the generation of triacylated protein [1,2]

The N-acylation of lipoproteins is essential in

Gram-negative bacteria to transport lipoproteins from the

inner membrane to the outer membrane via the

lipopro-tein localization pathway [3,4]

Although Lgt and LSP are widely conserved in

eubacteria, Lnt has not been found in low G+C

content Gram-positive bacteria (Firmicutes) [5–8]

Recently, Tschumi et al identified an E coli Lnt

homolog in the high G+C content Gram-positive

bac-terium Mycobacbac-terium smegmatis and demonstrated its

Lnt activity [9], but its homolog in Firmicutes was not

identified Several reports have presented structural

and⁄ or indirect evidence of diacylated lipoproteins, for

example, dipalmitoyl macrophage-activating

lipo-peptide-2 kDa (MALP-2) from Mycoplasma

fermen-tans [10], SA2202 (SAOUHSC_02699) protein from

S aureus SA113 strain [11] and F0F1-type ATPase

subunit b from Mycobacterium pneumoniae [12]

There-fore, Firmicutes are widely regarded as having only

diacylated lipoproteins Despite the lack of an E coli

Lnt homolog in Firmicutes, however, chemical

analy-ses of lipoproteins in Bacillus subtilis and in S aureus

suggested N-acylated lipoproteins in these organisms

[13,14] We also recently used MS-based analysis to

demonstrate that the SitC lipoprotein from S aureus is

triacylated [15]; however, we could not show structural

evidence of N-acylation of the lipoprotein In addition,

some triacylated lipoproteins of Mollicutes, which are

closely related to Firmicutes, have been reported based

upon indirect evidence of nuclear factor-jB activation

through Toll-like receptor (TLR)1 and TLR2 [16,17]

Therefore, evidence of the N-acylation of lipoproteins

leading to triacylated forms in Firmicutes is ambiguous

and controversial

Microorganism invasion activates the innate immune

response in mammals Bacterial lipoproteins as a

path-ogen-associated molecular pattern [18] are sensed by

the hosts through TLR2 heterodimerized with TLR1

or TLR6: this signal induces the activation of innate

immunity and is necessary to control adaptive

immu-nity [19] In addition, TLR2 stimulation drives the

dif-ferentiation of hematopoietic progenitor cells [20]

Although TLR2 has been considered as a receptor for

various structurally unrelated pathogen-associated

molecular patterns such as lipoproteins, lipoteichoic

acid and peptidoglycan [18], recent studies suggest that

bacterial lipoproteins function as the major, if not sole,

ligand molecules for TLR2-activation [5,11,15,21,22]

To date, synthetic lipoprotein analogs, such as

N-pal-mitoyl-S-dipalmitoylglyceryl (Pam3)–Cys, Pam3CSK4 lipopeptide and MALP-2 [10], have been used to mimic the proinflammatory properties of bacterial lipoproteins, and have led to a model in which tria-cylated lipopeptides signal through TLR2⁄ TLR1 heterodimer, whereas diacylated lipopeptides signal through TLR2⁄ TLR6 heterodimer However, recent studies have demonstrated that some synthetic lipopep-tides [23–25] and at least one native lipoprotein [15] are inconsistent with this model Therefore, real struc-tural characterization of native lipoproteins from Gram-positive bacteria is crucial to elucidate the molecular basis of host–microorganisms interaction Here, we carefully analyzed the structure of the lipo-peptide moiety of S aureus lipoproteins Analyses using lipoprotein lipase (LPL) and MS⁄ MS revealed that S aureus SitC was N-acylated with various length fatty acids and thus was triacylated In addition, SitC

in three other strains of S aureus and one strain of

S epidermidis, and four other lipoproteins in S aureus were shown to be N-acylated These results strongly suggest that lipoproteins in S aureus are mainly in the N-acylated triacyl form

Results

N-Terminal structure of SitC from S aureus RN4220 strain

Staphylococcus aureus SitC is annotated as the sub-strate-binding component of the ATP-binding cassette transporter for iron [26], and is one of the predomi-nant lipoproteins functioning as a ligand of TLR2 [27] Our previous report provided clear MALDI-TOF MS data representing that the N-terminal lipopeptide of SitC is triacylated [15] Although the result is not enough to support the N-acylation of SitC protein, it

is still surprising because of the presumed absence of

E coli Lnt homologs in the S aureus genomes [28] Contrary to our findings, Tawaratsumida et al reported that another lipoprotein SA2202 of S aureus SA113 had a diacylated (dipalmitoylated) N-terminus, based on MS⁄ MS data [11] To clarify this discrep-ancy, we decided to determine the bona fide structure

of lipoproteins in S aureus Also, determination of the exact structure of the Gram-positive bacterial native lipoproteins is essential for the elucidation of the molecular mechanism of host–microorganism interac-tions

To characterize the acylated structure of S aureus lipoproteins, we used commercially available LPL which is known to degrade bacterial lipoprotein and reduce the TLR2-stimulating activity of lipoproteins

[21] At first, we characterized the specificity of the

enzyme using synthetic Pam3CSK4 triacyl lipopeptide

as a substrate, and found that the enzyme hydrolyzes

O-esterified fatty acids of the (di)acylglyceride moiety,

but not N-acylated fatty acids To examine the

cleav-age patterns of native SitC lipoproteins by LPL, SitC

protein prepared by Triton X-114 (TX114) phase

partitioning was separated by SDS⁄ PAGE and then

subjected to in-gel digestion with trypsin to make the

N-terminal lipopeptide of SitC in the presence of

n-decyl-b-D-glucopyranoside, followed by

chloro-form⁄ methanol extraction Figure 1A shows

MALDI-TOF MS of the chloroform⁄ methanol organic phase

representing a series of 14-Da interval peaks, explained

by an increasing number of methylene (CH2) groups in

their fatty acids between m⁄ z 1297 and 1409, which

corresponds with triacylated N-terminal lipopeptides

of SitC (Table 1) The result is consistent with the MS

data from our previous study [15], and suggests that

the N-terminal lipopeptides of SitC were highly

puri-fied, because other peptides generated by the tryptic

digest were rarely detected in the organic phase As a

positive control, we confirmed that S-dipalmitoylglyce-ryl–CSK4 and Pam3CSK4 lipopeptides were also recovered in the organic phase using this extraction method and represented specific signals on MALDI-TOF MS (data not shown) Recovery of TLR2-stimu-lating activity in the organic phase was also confirmed using TLR2-expressing Chinese hamster ovary cells (Fig S1), suggesting that the N-terminal lipopeptides

of SitC are responsible for the TLR2 stimulation After 5 h incubation with LPL and lipopeptides, a new series of 14-Da interval peaks was detected between

m⁄ z 1044 and 1115 (Fig 1B) The mass values of these ions corresponds to those of the diacyl-glyceryl CGTGGK (Table 1) generated by the release of one O-esterified fatty acid from the original triacylated lipopeptide After 17 h incubation, another series of peaks between m⁄ z 835 and 891 was detected (Fig 1C), corresponding to monoacyl-glyceryl CGTGGK generated by releasing of two O-esterified fatty acids from the triacylated lipopeptide (Table 1)

On further incubation, no additional series of peaks was detected Thus, the triacylated SitC lipoprotein

1086.75 2 1114.78

1381.02 1 1352.99 1409.05

A

B

1338.98

1072.73

b1 y5

y4 y3

[MH-thioglycerol] +

[M+H] +

[MH-2H2O] +

[MH-H2O] +

b°5 b°4

b5 y°5

b°1

0 50 100

D

[MH-C2H6O2] +

862.5 754.4

m/z

800 900 1000 1100 1200 1300 1400 1500

862.60 3

890.63

C

m/z

N CH O C (C18H35 O)

S

CH HO

CH2

CH2

CH2 HO

H

716.3 444.5

362.2 419.3

b b°

261.1 y

y°

E

754.4 844.3

826.4 800.5

641.2

Fig 1 Lipoprotein lipase analysis determines triacylated structures of SitC lipoprotein in Staphylococcus aureus RN4220 cells SitC lipopro-tein in the TX114 fraction prepared from exponential-growth phase RN4220 cells was separated by SDS ⁄ PAGE and digested in-gel with tryp-sin The resulting N-terminal lipopeptides of SitC were extracted to the organic phase, dried and resolved in water (A) or further incubated with LPL for 5 h (B) or 17 h (C) MALDI-TOF MS of each fraction is shown Group 1, 2 or 3 in the figures is consistent with a series of tria-cylated, diacylated or monoacylated S-glyceryl CGTGGK peptides of N-terminal SitC, respectively, as described in Table 1 The series of mass signals harboring 14-Da mass differences [an increasing number of methylene (CH2) groups] is due to various length saturated fatty acids (D) MALDI-IT MS ⁄ MS using 2,5-dihydroxybenzoic acid as a matrix was carried out for an LPL-resistant lipopeptide of m ⁄ z 862.5, as described in (C) Isolation width was ± 2 Da (E) The elucidated structure of N-octadecanoly-S-glycerylcysteinyl GTGGK from the observed fragment ions in (D) Peaks designated y or b correspond to y-type or b-type ions that have lost an H 2 O moiety, respectively.

purified from RN4220 cells is modified by two

O-ester-ified fatty acids and one LPL-resistant fatty acid

To determine the exact modification site of the

LPL-resistant fatty acid, one of the peaks after LPL

digestion corresponding to the octadecanoyl-glyceryl

CGTGGK with m⁄ z 862.60 shown in Fig 1C was

further analyzed by MALDI-ion trap (IT) MS⁄ MS

Figure 1D,E show the MS⁄ MS spectrum and the

elucidated structure of the lipopeptide, respectively

The C-terminus-containing y-series ions at m⁄ z 261.1

(y3), 362.2 (y4), 401.3 (y5; y5-H2O) and 419.3 (y5) in

the spectrum confirmed the amino acid sequence of

GTGGK, which is complemented by ions at m⁄ z

444.5, 641.2 and 698.3, which are assigned as

N-terminus-containing b-series ions (b1, b4 and b5,

respectively) This result strongly suggests that the

fatty acid modification site is at the N-terminal

cyste-ine residue Importantly, a characteristic fragment ion

for N-acyl-dehydroalanyl peptide generated by neutral

loss of thioglycerol was observed at m⁄ z 754.4,

indi-cating that the octadecanoyl group is linked at the

a-amino group of cysteine via an amide bond These

results are consistent with our previous observation

that the N-terminus of SitC was blocked when

Edman degradation sequencing was performed [15]

Therefore, the results of both LPL digestion and

MS⁄ MS of the N-terminal peptides demonstrated that

the N-terminal cysteine of SitC from S aureus

RN4220 cells is N-acylated with a saturated C16 to

C20 fatty acid in addition to the expected S-diacylgly-cerylation with two saturated fatty acids

Triacylated SitC is the major form in the cell lysate of S aureus RN4220 strain

We next examined whether the triacylated forms are the major molecular species of SitC in S aureus cells

To evaluate the overall molecular species of SitC and prevent the loss of specific molecular species during isolation of SitC through the TX114 phase-partitioning method, the cell lysate of S aureus RN4220 strain was directly subjected to SDS⁄ PAGE (Fig 2A) When an approximately 33-kDa band was in-gel digested and the resulting peptides were analyzed by LC-MS⁄ MS, the 33-kDa band was identified as SitC (data not shown) The digest was then extracted with chloro-form⁄ methanol and analyzed by MALDI-TOF MS

As shown in Fig 2B, the triacylated N-terminal lipopeptides of SitC were detected The arrows indicate peaks corresponding to triacylated lipopeptides with the sum of the carbon number for three fatty acids of 47–55 in Table 1 By contrast, any significant peaks with mass values corresponding to the diacylated N-terminal lipopeptides of SitC referred to in Table 1 were not detected in the MALDI mass spectrum (Fig 2C) These results indicate that the triacylated forms are the major forms of SitC in S aureus RN4220 strain

Table 1 Calculated and observed masses of the lipid-modified N-terminal peptides of SitC and those generated by lipoprotein lipase diges-tion shown in Fig 1.

Lipoprotein lipase digestion for a 5 h and b 17 h.

Characterization of the N-terminal structure of

SitC in other strains of S aureus and

S epidermidis

Because one lipoprotein in S aureus SA113 strain was

reported to be diacylated [11], we then asked whether

the SitC lipoproteins of three other strains of S

aur-eus, including SA113 strain, and of S epidermidis

ATCC12228 strain are diacylated or triacylated The

organic phase of the in-gel-digested SitC isolated from

the TX114 fraction of exponential-growth phase

S aureus SA113 cells grown in Luria–Bertani (LB)

medium was analyzed by MALDI-TOF MS As shown

in Fig 3A and Table 2, a series of 14-Da interval

peaks between m⁄ z 1283 and 1381, corresponding to

the triacylated N-terminal lipopeptides of SitC

modi-fied with saturated fatty acids (the sum of the carbon

C

Fig 2 SitC prepared from a crude cell lysate of S aureus RN4220 cell is also triacylated (A) SDS ⁄ PAGE profile visualized with Coomassie Brilliant Blue of a crude cell lysate or its TX114 fraction of S aureus RN4220 cells is shown The arrowhead indicates the migration position of SitC that was identified at 33 kDa in either the cell lysate or TX114 phase (B,C) MALDI-TOF mass spectrum of the in-gel tryptic digests

of a 33-kDa region of the cell lysate Arrows indicate the calculated mass positions of the triacylated (B) or diacylated (C) N-terminal lipopeptides of SitC Calculated mass values for the lipopeptides are shown in Table 1.

B

C

A

1367.05 1381.11 1353.05

1339.03 1325.02 1310.99 1296.98

1282.91

*

*

*

*

C46

C47

C51 C52 C53

1352.99 1381.02 1367.00 1338.98

1324.96

1395.04

C51

C54 C55

m/z

1367.02 1339.01

1353.01 1324.96

1381.02 1310.97

1280 1300 1320 1340 1360 1380 1400 1420

1409.04

C49

C51 C52

C53 C48

C50

C55

Fig 3 Triacylated structures of N-terminal peptides of SitC of other

three S aureus strains MALDI-TOF mass spectra of the organic

phase of in-gel tryptic digests of SitC protein, which was isolated

through TX114 phase extraction from exponentially growing S aureus

cells in LB medium of laboratory strain SA113 (A), clinically isolated

resistant strain MW2 (B) or clinically isolated

methicillin-sensitive strain MSSA476 (C) Asterisks in (A) are signals 2-Da smaller

than the triacylated peptides filled with saturated fatty acids.

number for three fatty acids is 46–53), was detected In

the case of clinically isolated S aureus strains MW2

and MSSA476, a series of 14-Da interval peaks

between m⁄ z 1311 and 1409, corresponding to the tria-cyl lipopeptides of SitC with the sum of the carbon number of the fatty acids equal to 48–55 (see Table 1), was also detected (Fig 3B,C) Because Tawaratsumida

et al used Brain Heart Infusion (BHI) medium when they determined the N-terminal lipopeptide structure

of SA2202 protein of SA113 strain [11], we used this medium for the SA113 strain As in case of LB med-ium, SitC lipopeptides isolated from SA113 cells grown

in BHI medium had a series of peaks corresponding to the triacylated forms (data not shown)

The total carbon number of the modified fatty acids

in the most abundant peak of the SitC lipopeptides derived from the SA113 strain was smaller than that from RN4220, MW2 or MSSA476 strain, indicating that shorter fatty acids were mainly attached to the tria-cylated lipopeptides of SitC of the SA113 strain (Figs 1A and 3) The usage of shorter fatty acids was also detected in the spectrum of the SitC lipopeptides isolated from SA113 cells grown in BHI medium (data not shown) Moreover, triacyl peptides 2 Da smaller than those loaded with saturated fatty acids were addi-tionally observed in SA113 cells grown in both LB and BHI medium, and are indicated by asterisks in Fig 3A These peaks would be due to the presence of an unsatu-rated fatty acid in the triacylated lipopeptides

Table 2 Calculated and observed masses of triacylated N-terminal

lipopeptides of SitC and SA2202 isolated from exponentially

grow-ing S aureus SA113 cells grown in Luria-Bertani medium.

Modified peptide Calculated [M + H] + Observed m ⁄ z D (ppm)

a

SitC,bSA2202.

200 400 600 800 1000 1200

A

B

H +

C

O

H +

O C

R 3

CH2 S

CH2 CH O

CH 2

O

R2

R1

H peptide

N CH N CH

O C

R 3

CH2 S

CH2 CH O

R2

O CH2

R1

H peptide

+ H2O2

Energy

C17

C17 C15

1339.0

*

[M+H] +

[M+H] +

y2 y3

y4

y5

y1 y°1

*

m/z

B

N C O C

R3

CH2 H peptide

H +

N-acyl-dehydroalanyl peptide ion

S

CH2 CH O

R2

O CH 2

R 1

HO 2,3-diacyloxypropane sulfenic acid Neutral loss

200 400 600 800 1000 1200

1355.0

y2y3

y4 y5

C15C18

C16 C19 C20

dAGTGGK-NH3

dAGTGGK

Fig 4 Staphylococcus aureus SA113 cells also have the N-acylated triacyl-forms of SitC lipoprotein (A) SitC lipoprotein of exponentially growing S aureus SA113 cells in BHI medium was prepared as in Fig 1A Among the observed N-terminal lipopeptides of SitC similar to those described in Table 2 (LB medium) a signal at m ⁄ z 1339.0 corresponding to C50-triacylated lipopeptides was analyzed by MALDI-TOF

MS ⁄ MS Asterisks indicate molecular ion peaks generated by the neutral loss of diacylthioglycerol (B) Lipopeptides used in (A) were oxi-dized by spotting hydrogen peroxide solution on to the sample–matrix co-crystal before MALDI-TOF MS MALDI-TOF MS ⁄ MS spectrum of the on-target oxidized C50-triacylated lipopeptide at m⁄ z 1355.0 is shown Signals marked with C15 to C20 indicate the N-acylated dehydro-alanyl peptide ions generated by neutral loss of 2,3-diacyloxypropane-1-sulfenic acid having different length fatty acids, as indicated by carbon numbers Peaks designated dAGTGGK or dAGTGGK-NH3 correspond to the dehydroalanyl peptide ion or its de-ammonium ion, respectively, which are secondarily generated from N-acyl-dehydroalanyl peptide ions by losing fatty acid (C) Reaction scheme of H2O2 oxidation and collision-induced dissociation Incorporated oxygen, shown in bold, adds 16 Da to lipopeptides, such as from m ⁄ z 1339.0 (A)

to m⁄ z 1355.0 (B) The rectangle indicates dehydroalanine R1, R2 and R3 indicate a hydrogen or acyl group.

To prove the N-acylated structure of SitC from other than the RN4220 strain, N-terminal lipopeptide from the SA113 strain grown in BHI medium was ana-lyzed by MS⁄ MS Figure 4A shows the MS ⁄ MS spectrum of the peak at m⁄ z 1339.0, corresponding to triacylated lipopeptide in which the sum of the carbon number for three fatty acids is 50 The y-series ions detected in the spectrum are essentially the same as those from the SitC lipopeptide from RN4220 (Fig 1D), confirming that the peptide moiety of N-ter-minus is identical between the two strains Weak peaks around m⁄ z 750 corresponding to N-acyl-dehydroala-nyl peptide ions generated by the neutral losses of dia-cylthioglycerol moieties, the hallmark of N-acylation, were also detected in the spectrum (Fig 4A) Because these characteristic peaks are usually weak, and some-times not detectable, we developed a method that can sensitively detect the neutral losses Figure 4B shows

MS⁄ MS spectrum of the lipopeptide of SitC on-target oxidized with H2O2, in which intense peaks between

m⁄ z 712 and 782 were obtained The increase in N-acylated dehydroalanyl peptide ions is explained by the fact that oxidized lipoproteins undergo facile neu-tral loss of 2,3-diacyloxypropane-1-sulfenic acid in

MS⁄ MS (or MALDI-MS) (Fig 4C) The reaction mechanism should be similar to the neutral loss of methane sulfenic acid from methionine sulfoxide by

MS⁄ MS [29] The result shown in Fig 4B clearly dem-onstrates that a saturated fatty acid (from C15 to C20)

is linked at the a-amino group of SitC

In addition to these S aureus strains, analysis of SitC purified from S epidermidis [26] gives a series of 14-Da interval peaks between m⁄ z 2512 and 2597, cor-responding to the N-terminal lipopeptides of the triacy-lated SitC (Fig 5A) Of these, the peak at m⁄ z 2568.5, corresponding to triacylated CGNHSNHEHHSHEGK (the sum of the carbon number for three fatty acids is 53), was analyzed by MALDI-TOF MS⁄ MS The

MS⁄ MS spectrum shows the characteristic N-acyl dehyroalanyl peptide ions with C17 to C20 saturated fatty acid (Fig 5B), and its elucidated structure strongly indicates that the S epidermidis SitC lipopro-tein is also in the N-acylated triacyl form (Fig 5C)

Characterization of other lipoproteins of

S aureus Because SA2202 lipoprotein in S aureus SA113 strain was reported to be diacylated with two palmitic acids rather than with different length fatty acids [11], we examined whether SA2202 in the SA113 strain was triacylated and also examined the composition of the fatty acids Using the methods described above for

2568.49

2596.52 2554.45

2582.51

2540.44

2526.41

2512.41

C53

C52 C51

C50

C49

C54 C55

A

m/z

m/z

y5

y6 y7 y8 y9

y10 y11 y12 y14

C18 C17 C19 C20

[M+H] +

2568.4

B

R2

y

O

N CH

CH2

CH

S

CH2

CH2

O

O

n = 20: 1968.2

19: 1954.5

17: 1926.3

R3

H

C

R1: ClH2l-1O

R2: CmH2m-1O

R3: CnH2n-1O

1606.5 1435.51299.71212.21097.9959.9 831.2 694.3 557.0

S

Fig 5 N-Acylated triacyl structure of S epidermidis SitC (A)

MALDI-TOF mass spectrum of an organic phase of in-gel tryptic

digest of SitC, isolated through TX114 phase extraction from

expo-nential-growth phase S epidermidis ATCC12228 cells in LB

med-ium The sum of carbon numbers for three fatty acids is labeled at

the peaks corresponding to the monoisotopic mass values of the

triacylated N-terminal lipopeptides of SitC (CGNHSNHEHHSHEGK).

(B) MALDI-TOF MS ⁄ MS spectrum of the precursor ion of

m ⁄ z 2568.5 (C53) shown in (A) Signals marked C17 to C20

indi-cate the N-acylated dehydroalanyl peptide ions generated by neutral

loss and with different length saturated fatty acids with the

indicated carbon numbers (C) The elucidated structure of

C53-tria-cylated N-terminal lipopeptides of S epidermidis SitC with

assign-ments of the observed fragment ions in panel B The l, m and n

are positive integers and l + m + n = 53.

SitC in RN4220 strain, lipopeptides of SA2202 from

the SA113 strain were prepared and analyzed by

MALDI-TOF MS As shown in Table 2, characteristic

peaks corresponding to triacyl N-terminal lipopeptides

of SA2202 with a total of 45 to 54 carbon numbers of

three saturated fatty acids were detected in the

MALDI-TOF mass spectrum of the in-gel tryptic

digest itself and its organic phase extract, but diacyl

lipopeptide signals were not Like SitC, SA2202 pro-tein in the SA113 strain was modified, mainly with shorter fatty acids than in RN4220 (Tables 2 and 3) and peaks containing unsaturated fatty acids were also observed (data not shown) Figure 6A,B shows the

MS⁄ MS spectrum and the elucidated structure of the most abundant peak, corresponding to the oxidized C50-triacyl lipopeptides The spectrum clearly showed the characteristic N-acyl-dehydroalanyl peptide ions with C15 to C20 saturated fatty acid, suggesting that the SA2202 lipoprotein in SA113 is the N-acylated triacyl form with different length fatty acids

We then asked whether other lipoproteins in the RN4220 strain were N-acylated To address this, we searched for other lipoproteins in the TX114 phase of

S aureus RN4220 LC-MS⁄ MS of the in-gel tryptic

Table 3 Calculated and observed masses of the triacylated

N-terminal lipopeptides of SA2202, SA0739, SA0771, SA2074, and

SA2158 proteins isolated from exponentially growing S aureus

RN4220 cells grown in LB medium.

Modified peptide Calculated [M+H] + Observed m ⁄ z D (ppm)

a SA2202, b SA0739, c SA0771, d SA2074, e SA2158.

R1

S

CH2 CH O

O CH2

R2

C17

1542.0 [M+H] +

y2 y3 y4 y5

C15

C18 C16

C19 C20 y6

m/z

R1: ClH2l-1O

R2: CmH2m-1O

R3: CnH2n-1O O

N

CH O C

CH2 S

H

R3

234.0 321.1

549.1 606.1

n = 20: 969.5

19: 955.5 18: 941.5 17: 927.5 16: 913.5 15: 899.5 O

A

B

Fig 6 Lipoprotein SA2202 of S aureus SA113 shows N-acylated triacyl-forms (A) MALDI-TOF MS ⁄ MS was carried out for N-termi-nal lipopeptides of SA2202 lipoprotein from exponentially growing SA113 cells as described in Fig 4B Precursor peaks used were molecular ions of the on-target oxidized C50-triacylated lipopeptide

at m ⁄ z 1542.0 Signals marked with C15 to C20 indicate the N-acyl-ated dehydroalanyl peptide ions that were generN-acyl-ated by neutral loss and had different length fatty acids with the indicated carbon num-bers (B) Elucidated structures of the oxidized N-terminal triacyl lipo-peptides R1, R2 and R3 indicate acyl group The l, m and n are positive integers and l + m + n = 50 The value of n was deter-mined to be from 15 to 20 (A) Calculated mass values of the N-ter-minal triacyl lipopeptides of SA2202 are found in Table 2.

digest of a minor band enabled us to identify five other

lipoproteins, SA0739, SA0771, SA2074 (ModA),

SA2158 and SA2202 To concentrate these lipoproteins,

the Zn-stained lipoproteins on the acrylamide gels were

collected and then eluted from gel pieces by simple

dif-fusion, concentrated and resubjected to SDS⁄ PAGE

The concentrated protein band was then in-gel digested

Using the methods described above for SitC, the

result-ing lipopeptides were extracted by chloroform⁄

metha-nol and analyzed by MALDI-TOF MS Table 3 shows the observed and calculated masses of these lipopeptide peaks in each lipoprotein A series of these peaks at an interval of m⁄ z 14 in each protein corresponded to the calculated molecular mass values of triacylated N-terminal lipopeptides of each In contrast, mass signals corresponding to diacylated lipopeptides were not observed (data not shown) Therefore, these five lipoproteins are suggested to be mainly triacylated N-Acylation of these triacylated lipoproteins from RN4220 was further demonstrated by MALDI-TOF

MS⁄ MS (Fig 7) Regarding SA2202 protein, Fig 7A shows the MS⁄ MS spectrum of the most abundant peak

at m⁄ z 1553.9, corresponding to the N-terminal tria-cylated CGNNSSK lipopeptide (the sum of the carbon number for three fatty acids was 52; see Table 3) The oxidized lipopeptides were also analyzed by MS⁄ MS (shown as an inset in Fig 7A) Both spectra represented the characteristic N-acyl-dehydroalanyl peptide ions due to neutral loss, whose signals were enhanced in the oxidized lipopeptides The fatty acid linked to the a-amino group was a saturated fatty acid with a length

of C16 to C20 Likewise, two lipoproteins (SA0739 and SA0771) were also successfully determined to be N-acyl-ated due to the detection of the N-acyl-dehydroalanyl peptides caused by the neutral losses using the oxidized lipopeptides (Fig 7B,C) Figure 7D shows the MS⁄ MS spectrum of the triacylated N-terminal lipopeptide of SA2074, which presents relatively week but significant peaks of the N-acyl-dehydroalanyl peptide ions (C15– C20) generated by the neutral loss of diacylthioglycerol, indicating N-acylation of the lipopeptide An MS⁄ MS spectrum of the triacylated lipopeptides of SA2158 pro-tein did not show significant signals because of the low intensity of the lipopeptide peaks

Discussion

This study presents, for the first time, the structure of N-terminal lipids of native S aureus lipoproteins Here,

we provide solid structural evidence for N-acylated triacyl forms of SitC and four other lipoproteins in

S aureus RN4220 using intensive MS-based analysis, combined with LPL or H2O2treatment The triacylated lipoproteins were confirmed in other three S aureus strains including SA113 and in one S epidermidis strain, strongly suggesting that lipoproteins of S aureus are mainly N-acylated triacyl forms Lipoproteins in Firmicutes were thought to be diacylated because of the absence of E coli-type Lnt in these bacteria [22]; the hypothesis was that immune cells could discrimi-nate between Gram-negative and Gram-positive bacte-ria by the ability of TLR2 to form heterodimers with

200 400 600 800 1000 1200 1400

m/z

C

B

A

HQ

HQD

y5 y4

y6

y7 y8

b5 b6

[M+H] +

1827.12

200 400 600 800 1000 1200 1400 1600 1800

C17 C18

1160 1200 1240 C15

C16 C19 C20

[M+H] +

y5 y2

y1

1553.9 C17

C18

900 940 980 C16

C19 C20

C

[M+H] +

y2

y4 y5

y6 y7 y8

y10 y9

2043.37

200 400 600 800 1000 1200 1400 1600 1800 2000

[M+H] +

y5

y4

y3

y2

y°1

1437.10

m/z

800 840 880 C16

C17 C18 C19 C20

1440

C17 C16 C18 C19

D

Fig 7 Other N-acylated triacyl-lipoproteins of S aureus RN4220

cells MALDI-TOF MS ⁄ MS spectrum of N-terminal lipopeptides of

SA2202 protein with C52 (A), SA0739 protein with C52 (B),

SA0771 protein with C52 (C) or SA2074 protein with C51 (D)

prepared from exponential-growth phase S aureus RN4220 cells in

LB as SitC of Fig 1A The mass of the precursor ion for each is

described in Table 3 The inset in each panel is the MALDI-TOF

MS ⁄ MS spectrum of on-target oxidized lipopeptides for each (A–C)

or the magnified view (D) N-Acyl-dehydroalanyl peptide ions

gener-ated by the neutral loss of 2,3-diacyloxypropane-1-sulfenic acid or

diacylthioglycerol were observed and are indicated by C15 to C20

in the insets Peaks designated with HQ and HQD in (B) are

internal fragment ions.

TLR1 or TLR6, in response to triacyl lipopeptides or

diacyl lipopeptides, respectively [30] However, our

study provides some clear evidence that these

predic-tions may need to be reconsidered

Contrary to our results, Hashimoto’s group reported

that the N-terminus of SA2202 lipoprotein in S aureus

SA113 was only an S-dipalmitoylglyceryl cysteine form

in S aureus SA113 [11] Discrepancies with our results

are likely due to differences in cell-culture, lipoprotein

preparation or analysis methods, because the SA113

strain we used was received from Hashimoto’s group

Differences in method resulted in several hundred-fold

differences in the recovery of lipoprotein for structural

analysis; their previous overall yield was only 1.6 lg of

each lipoprotein per l-culture [11], whereas we obtained

several hundred micrograms of SitC in our TX114

phase per l-culture (data not shown) In addition, we

developed the analytical method including in-gel

diges-tion and organic solvent extracdiges-tion in the presence of

0.1% n-decyl-b-D-glucopyranoside (see Experimental

procedures), which prevents the loss of very

hydropho-bic lipopeptides As a consequence, we detected a series

of peaks modified with different length saturated and

unsaturated fatty acids Because phospholipids in

bac-teria have various length fatty acids and Lgt transfers

a diacylglycerol moiety from phospholipids to the

lipo-protein precursor, it is reasonable to conclude that

lipoproteins are modified with different length fatty

acids Therefore, detection of only the

dipalmitoyl-glyceryl form in SA2202 [11] indicates insufficient

sensi-tivity to detect heterogeneity of the acylation in their

analysis Further, we showed the predominance of

tria-cylated SitC in cell lysate (Fig 2) Therefore, we believe

that we have obtained unbiased results, and that the

major form of lipoproteins in S aureus is the

triacylat-ed one Recently, a similar proctriacylat-edure with us allowtriacylat-ed

the detection of triacylated lipopeptides of LppX

lipo-protein from M smegmatis [9] However, our results

do not rule out the existence of diacylated lipoproteins

in the bacterium Because they are intermediate forms

during biosynthesis of the triacylated lipoproteins, they

are likely to be a minor component under our

condi-tions In fact, in addition to mass signals of the

tria-cylated form, relatively weak signals corresponding to

diacylated lipopeptides of SitC with various fatty acids

were also detected from high-temperature culture (data

not shown), suggesting that the degree of N-acylation

may depend on bacterial growth conditions

Our results also suggest that unsaturated fatty acid

was incorporated in lipoproteins of S aureus SA113

strain Although further analysis to determine the

modification site(s) and molecular species of the

unsat-urated fatty acid in bacterial lipoproteins is required,

its roles in ligand recognition and receptor activation for TLR2 are curious

In addition to our studies, several reports provide indirect evidence of triacylated lipoprotein(s) in cutes [13,14] and Mollicutes [16,17] Although Firmi-cutes do not have an E coli Lnt homolog [5–8], our results strongly suggest that S aureus and also S epide-rmidishave an unidentified enzyme which can catalyze the N-acylation of diacylated lipoproteins with a satu-rated fatty acid, whose structure is distinct from E coli and M smegmatis Lnt N-Acylation of lipoproteins in

E coliis characterized as being required for lipoprotein localization machinery LolCDE-dependent release of outer membrane-specific lipoproteins from the inner membrane [3], and deficiency of Lnt is known to cause mislocation of lipoproteins [31] Identification of the unidentified N-acylation enzyme in S aureus should facilitate understanding of the biological significance of the lipoprotein N-acylation

Experimental procedures

Bacterial strains, plasmids, and culture conditions

The S aureus SA113 strain was a gift from Dr M Hashim-oto (Kagoshima University, Japan) The methicillin-resistant

MSSA476 strain were obtained from the Network on Anti-microbial Resistance in Staphylococcus aureus (Chantilly, VA) The Staphylococcus epidermidis ATCC12228 strain was obtained from the Korean Culture Center of Microorgan-isms (Seoul, Korea) These S aureus and S epidermidis

exponential-growth phase, as described previously [15]

Purification of lipoproteins by TX114 phase partitioning

Lipoproteins were obtained using the TX114 phase-parti-tioning method, as described previously [15] Briefly, the harvested S aureus cells were disrupted with glass beads and centrifuged at 2000 g for 10 min to remove the glass beads and unbroken cells, and the resulting supernatant was further centrifuged at 20 000 g for 10 min to remove the peptidoglycan fraction The obtained supernatant was used as a cell lysate, or was supplemented with TX114 to a

removed and was replaced with the same volume of a

5 mm EDTA) This procedure was repeated twice The final