aac accepts, published online ahead of print on 26 may...
TRANSCRIPT
1
Title; Peptide-lipid huge toroidal pore: a new antimicrobial mechanism mediated by a 1
lactococcal bacteriocin, lacticin Q 2
3
Running title; Huge toroidal pore mediated by lacticin Q 4
5
Author; Fuminori Yoneyama,1 Yuichi Imura,
3 Kanako Ohno,
1 Takeshi Zendo,
1 Jiro 6
Nakayama,1 Katsumi Matsuzaki,
3 and Kenji Sonomoto
1,2 7
8
Address; Laboratory of Microbial Technology, Division of Microbial Science and 9
Technology, Department of Bioscience and Biotechnology, Faculty of Agriculture, Graduate 10
School,1 and Laboratory of Functional Food Design, Department of Functional Metabolic 11
Design, Bio-Architecture Center,2 Kyushu University, 6-10-1 Hakozaki, Higashi-ku, 12
Fukuoka 812-8581, Japan, and Graduate School of Pharmaceutical Sciences, Kyoto 13
University, 46-29 Yoshida-Shimo-adachi-cho, Sakyo-ku, Kyoto 606-8501, Japan3 14
15
*Corresponding author. Mailing address: Laboratory of Microbial Technology, Division of 16
Microbial Science and Technology, Department of Bioscience and Biotechnology, Faculty of 17
Agriculture, Graduate School, Kyushu University, 6-10-1 Hakozaki, Higashi-ku, Fukuoka 18
812-8581, Japan. Phone: (81) 92 642 3019. Fax: (81) 92 642 3019. E-mail: 19
Copyright © 2009, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved.Antimicrob. Agents Chemother. doi:10.1128/AAC.00209-09 AAC Accepts, published online ahead of print on 26 May 2009
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
2
1
Abstract 2
Lacticin Q is a pore-forming bacteriocin produced by Lactococcus lactis QU 5, and its 3
antimicrobial activity is in the nanomolar range. Lacticin Q induced calcein leakage from 4
negatively charged liposomes. However, no morphological changes in the liposomes were 5
observed by light scattering. Concomitantly with calcein leakage, lacticin Q was found to 6
translocate from the outer to the inner leaflet of the liposomes after initially binding to 7
membrane within two seconds. Also, lacticin Q induced lipid flip-flop. These results reveal 8
that the antimicrobial mechanism of lacticin Q is the toroidal pore model. This is the first 9
report on a bacteriocin of Gram-positive bacteria forming a toroidal pore. From liposomes, 10
lacticin Q leaked fluorescence-labeled dextran with a diameter of 4.6 nm. In addition, lacticin 11
Q caused leakage of small size proteins such as green fluorescence protein from live bacterial 12
cells. There are no other reports on antimicrobial peptides exhibiting protein leakage. The 13
proposed pore formation model of lacticin Q is as follows: (i) quick binding to outer 14
membrane leaflets; (ii) formation of at least 4.6 nm pores, causing protein leakage with lipid 15
flip-flop; (iii) migration of lacticin Q molecules from outer to inner membrane leaflets. 16
Consequently, we termed the novel pore model in the antimicrobial mechanism of lacticin Q, 17
a ‘‘huge toroidal pore’’. 18
19
Introduction 20
Bacteriocins are antimicrobial peptides or proteins produced by bacterial strains (7). 21
Small bacteriocins (<10 kDa) produced by lactic acid bacteria (LAB) are potential food 22
preservatives and alternatives to antibiotics (5, 7, 13). Antimicrobial mechanisms in the LAB 23
bacteriocins, nisin, pediocin PA-1, and lacticin 3147, are well characterized (3, 5, 9, 11, 13, 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
3
34). These LAB bacteriocins show strong antimicrobial activity against specific 1
Gram-positive bacteria in nanomolar concentration ranges. These bacteriocins cause ion 2
efflux via pore formation with the aid of initial receptors, the so-called docking molecules 3
(15). A peptidoglycan precursor, lipid II, is used as the docking molecule of nisin and lacticin 4
3147. These bacteriocins do not exhibit pore-forming activities against liposomes without the 5
docking molecule (34, 35). In addition, other killing mechanisms using lipid II are known to 6
involve inhibition of peptidoglycan biosynthesis (34, 35). Similarly, a membrane protein, 7
MptD, is the docking molecule for pediocin PA-1 and its homologs, and an inserted region in 8
MptD observed only in sensitive species such as Listeria and Enterococcus is the specific 9
recognition site (11, 14). This type of bacteriocins does not demonstrate pore-forming activity 10
against liposomes in the absence of MptD and against bacteria containing no specific region 11
in MptD (8, 26). In conclusion, docking molecules play essential roles in antimicrobial 12
activities of these LAB bacteriocins. 13
Multicellular eukaryotes also produce pore-forming antimicrobial peptides to prevent 14
microbial invasion (6, 38). Some antimicrobial peptides inhibit both Gram-positive and 15
-negative bacteria in the micromolar range through a membrane-permeabilizing mechanism 16
such as barrel stave, carpet, and toroidal pore. One of the most characterized peptides 17
involved in antimicrobial activities is magainin 2 produced by Xenopus laevis (18). Initially, 18
the cationic peptide magainin 2 binds to negatively charged bacterial membranes and forms 19
an amphiphilic α-helical structure. Magainin 2 then forms a pore, which is accompanied by 20
rapid lipid transbilayer movement, the so-called lipid flip-flop. The peptide-lipid 21
supramolecular complex pore is called the toroidal pore. The diameter of magainin 2 toroidal 22
pore, which can leak water-soluble substances, is estimated to be 2–3 nm. The pore causes 23
leakage of calcein (molecular weight, 622), but not of dextran with an average molecular 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
4
weight of around 4400 (19). Some magainin 2 molecules are translocated from the outer to 1
the inner membrane leaflets when the pore closes. While the peptide-lipid supramolecular 2
complex pores caused by peptides from some multicellular eukaryotes have been well 3
characterized, those created by LAB bacteriocins are poorly understood. 4
Previously, we discovered a novel LAB bacteriocin, lacticin Q, produced by 5
Lactococcus lactis QU 5 (10). Lacticin Q, composed of 53 amino acids without 6
intramolecular bridges, is a cationic membrane-permeabilizing peptide with wide bactericidal 7
activity against Gram-positive bacteria in the nanomolar range. The antimicrobial activity, pH 8
stability, and heat tolerance of lacticin Q are comparable with those of nisin, which shows 9
higher stability than other LAB bacteriocins. Unlike typical LAB bacteriocins, such as nisin, 10
the pore-forming activity of lacticin Q does not require a docking molecule, as confirmed by 11
its high activity against negatively charged liposomes (37). Lacticin Q forms an amphiphilic 12
α-helix, which is frequently observed in antimicrobial peptides (37, 38). 13
In the present study, we characterize the antimicrobial mechanism features of lacticin Q, 14
which are more similar to magainin 2 than typical LAB bacteriocins. On the other hand, 15
lacticin Q exerted strong antimicrobial activity in the nanomolar range, and the activity of 16
magainin 2 was in the micromolar range. We hypothesized that lacticin Q possesses unique 17
antimicrobial mechanism features. Here, we report a new lacticin Q-mediated antimicrobial 18
mechanism, termed the huge toroidal pore. 19
20
Materials and Methods 21
Antimicrobial peptides 22
According to a previous report (10), lacticin Q and nisin were purified from L. lactis QU 5 23
culture supernatant and a commercial preparation (Sigma, St. Louis, MO, USA), respectively. 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
5
Magainin 2 was chemically synthesized by standard solid-phase peptide synthesis (27). Purity 1
of peptides was checked by electrospray ionization time-of-flight mass spectrometry on a 2
JMS-T100LC (JEOL, Tokyo, Japan). 3
4
Liposome preparation 5
A zwitterionic phospholipid, egg yolk L-α-phosphatidylcoline (PC), and an anionic 6
phospholipid, L-α-phosphatidyl-DL-glycerol (PG), were purchased as the highest grade from 7
Sigma. Liposomes, large unilamellar vesicles (LUVs), were prepared as follows. A lipid film 8
(PC:PG, 1:1 molar ratio) was hydrated with buffer A (10 mM Tris-HCl, 75 mM NaCl, 1 mM 9
EDTA, pH 7.4). For leakage experiments, 70 mM calcein (Dojindo Laboratories, Kumamoto, 10
Japan) was added to buffer A when the lipid film was hydrated. The suspension underwent 10 11
freeze-thaw cycles and subsequently extruded through a 0.1 µm pore size polycarbonate filter. 12
For the calcein-entrapping LUVs, untrapped calcein was removed by gel filtration 13
chromatography using a 4% plain agarose beads column (Agarose Bead Technologies, 14
Madrid, Spain). The phosphorous-based lipid concentration (of lipid stocks and LUVs) was 15
determined by the Bartlett method (2). All other LUVs used in this study contained a 16
negatively charged lipid (PG) at a molar ratio of 50%. 17
18
Calcein leakage 19
Calcein leakage from LUVs was examined at an excitation wavelength (EX) of 490 nm and 20
emission wavelength (EM) of 520 nm on an F-7000 spectrofluorometer (Hitachi 21
High-Technologies, Tokyo, Japan). Peptides in buffer A were maintained at 30°C with 22
agitation prior to the addition of LUVs to the solution. Leakage of 100% was determined by 23
the addition of 0.1% (v/v) Triton X-100 (Wako Pure Chemical Industries, Osaka, Japan). 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
6
1
Light scattering 2
Peptide-induced morphological change in PC/PG (1:1) LUVs was monitored in buffer A 3
using 90° light scattering on the F-7000 spectrofluorometer at 30°C. Both EX and EM were 4
set as 400 nm. 5
6
Peptide translocation 7
A dansyl-labeled lipid, 8
N-(5-dimethylaminonaphthalene-1-sulfonyl)-1,2-dihexadecanoyl-sn-glycero-3-phosphoethan9
olamine (DNS-DHPE), was purchased from Invitrogen (Carlsbad, CA, USA). Dansyl-labeled 10
LUVs (PC/PG/DNS-DHPE, 4:5:1) and unlabeled LUVs (PC/PG, 1:1) were prepared by the 11
aforementioned procedures. Peptide migration from outer to inner membrane leaflets 12
(translocation) was detected by fluorescence resonance energy transfer (FRET) from the 13
tryptophan residues of lacticin Q to dansyl-labeled LUVs, as previously reported (21). 14
Dansyl-labeled LUVs were added to peptides in buffer A on the F-7000 spectrofluorometer at 15
EX of 280 nm and EM of 336 nm at 30°C. After incubation, excess of unlabeled LUVs were 16
added for desorption of untranslocated peptides from dansyl-labeled LUVs. Based on the 17
fluorescence intensities before and after the addition of unlabeled LUVs, the ratios of peptide 18
translocation were calculated, as previously reported (21). 19
Calcein leakage from the dansyl-labeled LUVs (PC/PG/DNS-DHPE, 4:5:1) was 20
monitored by the method described above. 21
22
Flip-flop 23
A pyrene-labeled lipid, 1-lauroyl-2-(1’pyrenebutyroyl)-sn-glycero-3-phosphocholine (pyPC), 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
7
was purchased from Invitrogen. The fluorescence of pyPC depends on the distance between 1
the molecules (17, 32). Symmetrically pyrene-labeled LUVs show monomer fluorescence 2
(Im) at 397 nm (exited at 341 nm). In the case of asymmetrically pyrene-labeled LUVs, 3
excimer fluorescence (Ie) at 479 nm is observed, whereas Im is lower than that of the 4
symmetrical LUV. The Ie/Im ratio is the indicator of transbilayer lipid movement. 5
Asymmetrically and symmetrically pyrene-labeled LUVs (PC/PG/pyPC, 47:50:3) were 6
prepared according to a previous report (25). Briefly, to prepare asymmetrically 7
pyrene-labeled LUVs, dried pyPC was dissolved in 50 µl of ethanol, and 1 ml of buffer A was 8
added to form pyPC micelle. The pyPC micelle (3% of final product) and unlabeled LUVs 9
(PC/PG, 47:50) were mixed at 37°C with agitation for a few minutes, and the mixture was 10
kept overnight at room temperature. Peptide-mediated lipid flip-flop in buffer A was detected 11
on the F-7000 spectrofluorometer at EX of 341 nm and EM of 350–500 nm. The fluorescence 12
spectra of symmetrically and asymmetrically pyrene-labeled LUVs served as positive and 13
negative controls, respectively. Flip-flop extent was calculated from spectrum Ie/Im ratio, as 14
previously reported (25). 15
Calcein leakage from the pyrene-labeled LUVs (PC/PG/pyPC, 47:50:3) was monitored 16
by the method described above. 17
18
Dextran leakage 19
Fluorescein isothiocyanate-labeled dextrans (FITC-dex) with average molecular weights of 20
around 10000, 20000, and 40000 (FD-10K, -20K, and -40K, respectively) were purchased 21
from Sigma. FITC-dex-entrapping dansyl-labeled LUVs (PC/PG/DNS-DHPE, 45:50:5) were 22
prepared by the method described above. FITC-dex (2.5 mM) was added into buffer A when 23
the lipid film was hydrated. Untrapped FITC-dex was removed by gel filtration 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
8
chromatography, as described above. 1
FITC-dex leakage was monitored according to a previous report (21). Briefly, 2
dansyl-labeled LUVs (50 µM lipids) harboring FITC-dex were incubated with peptides in 3
buffer A at 30°C. After 10 min, an excess of unlabeled LUVs (500 µM lipids; PC/PG, 1:1) 4
was added to stop the leakage. Leaked FITC-dex was eliminated by gel filtration 5
chromatography, and the LUVs fraction was collected. After addition of 0.1% Triton X-100, 6
the excitation spectrum of the fraction was measured on the F-7000 spectrofluorometer at EX 7
of 300–500 nm and EM of 520 nm. The spectrum was normalized at 490 nm. The extent of 8
FITC-dex leakage was determined by the fluorescence intensity at 340 nm (21). 9
10
Green fluorescence protein leakage 11
A green fluorescence protein (GFP)-encoding plasmid, pGreen (23), was a kind gift from the 12
National Institute of Genetics (Mishima, Japan). Gene gfp and two PstI sites located 13
immediately upstream and downstream of gfp were amplified by PCR using the DNA 14
polymerase, KOD plus (Toyobo, Osaka, Japan), with oligonucleotide primers, pGreen-F 15
(5’-CAGCTATGGTACCGGTAGA-3’) and pGreen-R (5’-CAGGTCGACTCTAGAGGAT-3’). 16
The PCR product and a lactococcal expression vector, pNZ8048 (22), were treated by PstI 17
(Nippon Gene, Toyama, Japan) and were ligated by Ligation high (Toyobo). The constructed 18
plasmid, pNZgfp, was introduced into L. lactis NZ9000 by electroporation, according to the 19
standard procedures (1). The transformant, L. lactis NZgfp, was grown in M17 medium 20
(Merck, Whitehouse Station, NJ, USA) supplemented with 0.5% glucose (w/v; GM17) and 21
50 µg/ml chloramphenicol at 30°C. 22
For the induction of GFP in L. lactis NZgfp, 10 ng/ml nisin was added 2 h after the 23
beginning of culture, and cells were grown for an additional 2.5 h. The cells were harvested 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
9
by centrifugation (6000 × g, 10 min, 4°C), and washed twice with cold buffer A. The cells 1
were added to the peptide solution in buffer A (1 ml) to adjust absorbance at 600 nm of 0.25 2
(ABS600 = 0.25), and incubated for 10 min at 30°C with agitation. The cells were then 3
harvested by centrifugation (6000 × g, 10 min, 4°C), and resuspended in 20 µl of buffer A. 4
Remaining GFP was monitored by fluorescence microscopy (Eclipse 80i/D-FL with GFP-B 5
filter; Nikon, Tokyo, Japan). The fluorescent image contrasted with the bright-field view. 6
Leaked GFP into the supernatant was measured on the F-7000 spectrofluorometer. GFP 7
expression in the L. lactis NZgfp cells and peptide treatment were performed according to the 8
above described procedures. The peptides-treated cells were eliminated by centrifugation. 9
The supernatant (800 µl), 1 µM (glycine-treated) Alexa Fluor 430 (Invitrogen; 10 µl), and 10
buffer A (1190 µl) were mixed. The excitation spectrum of the mixture was recorded at 30°C. 11
From the fluorescence intensities of 520 nm excited at 440 nm (FAF; Alexa Fluor 430) and 12
490 nm (FGFP; GFP), the ratio FGFP/FAF was calculated to determine the amount of GFP 13
leaked into the supernatant. As positive control, the FGFP/FAF of whole cells was recorded. 14
Fluorescence was seldom confirmed in GFP-unexpressed cells. 15
16
Protein leakage 17
L. lactis NZgfp and L. lactis IL1403, bacterial strains sensitive to lacticin Q, were cultured 18
overnight in GM17 supplemented with and without 50 µg/ml chloramphenicol, respectively. 19
Listeria innocua ATCC 33090T, also a strain sensitive to lacticin Q, was cultured in Tryptic 20
soy broth (Difco Laboratories, Detroit, MI, USA) supplemented with 0.6% (w/v) yeast 21
extract (Nacalai Tesque, Kyoto, Japan). Subsequently, the strains were inoculated to their 22
respective fresh media and grown to the logarithmic phase (ABS600 = 1.0) at 30°C. The cells 23
were washed two times by cold buffer A and resuspended. In 1.0 ml of buffer A, the cells 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
10
(ABS600 = 0.5) were treated by the peptide at the indicated concentration for 10 min at 30°C 1
with agitation. The cells were removed two times by centrifugation (6000 × g, 10 min, 4°C), 2
and the 900 µl of supernatant was desalted and concentrated using a PAGEprep Advance Kit 3
(Pierce Biotechnology, Rockford, IL, USA), and sodium dodecyl sulfate-polyacrylamide gel 4
electrophoresis (SDS-PAGE) and silver staining using 2D-silver stain II (Cosmo Bio, Tokyo, 5
Japan) were performed. 6
7
Result 8
Membrane permeabilization of lacticin Q 9
Lacticin Q-inducing calcein leakage from PC/PG LUVs was monitored (Fig. 1A). The 10
calcein leakage started at a very low concentration of 31.3 nM. 11
Next, we analyzed morphological change in LUVs by light scattering when lacticin Q 12
was exhibiting membrane permeabilization against LUVs. The absence of morphological 13
change in the LUVs was confirmed by the treatment with 250 nM lacticin Q (Fig. 1B). In 14
addition, a higher concentration of lacticin Q (2 µM) did not cause any morphological change 15
in LUVs (data not shown). These results revealed that lacticin Q does not affect the size of 16
the LUVs. 17
18
Characterization of the lacticin Q pore 19
We investigated peptide translocation by the detection of FRET. As shown in Fig. 2A, 20
tryptophan fluorescence energy of lacticin Q was transferred to DNS-DHPE when 21
dansyl-labeled LUVs were added (solid arrow), resulting in a rapid reduction in the 22
fluorescence intensity. After incubation, excess of unlabeled LUVs were added (dotted 23
arrows). Lacticin Q molecules locating on the outer leaflet of dansyl-labeled LUVs 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
11
redistributed to the unlabeled LUVs, and an increase of fluorescence intensity was observed. 1
However, depending on incubation time with dansyl-labeled LUVs, the recovered 2
fluorescence intensity was lower than that in the control (gray line) in which both 3
dansyl-labeled and excess of unlabeled LUVs were added simultaneously. These results 4
indicate that lacticin Q molecules translocated from the outer to the inner leaflets of LUVs. In 5
addition, fast FRET showed quick binding. The ratio of translocated lacticin Q was calculated 6
from the results of Fig. 2A, and the data are shown with the results of the calcein leakage 7
experiment (Fig. 2B). This indicates that changes in the ratios of translocated lacticin Q were 8
closely related to pore formation. 9
Next, lipid flip-flop was detected using symmetrically and asymmetrically 10
pyrene-labeled LUVs. In the case of symmetrically pyrene-labeled LUVs, high Im (397 nm) 11
was observed, and a peak corresponding to Ie (479 nm) was hardly detected (Fig. 3A, trace 1). 12
On the other hand, lower Im and higher Ie values were observed in the spectrum of the 13
asymmetrically pyrene-labeled LUVs (Fig. 3A, trace 2). Treatment with lacticin Q led to 14
enhancement of Im and reduction of Ie in the asymmetrically pyrene-labeled LUVs (Fig. 3A, 15
trace 3). The changes in pyrene fluorescence indicate that lipid flip-flop was induced by 16
lacticin Q. Time-dependent lipid flip-flop was analyzed from the ratio of Ie/Im, and these data 17
are plotted along with the results of the calcein leakage experiments (Fig. 3B). The ratio of 18
lipid flip-flop induced by lacticin Q was closely related to pore formation, as observed in the 19
case of peptide translocation. These features of the lacticin Q pore (calcein leakage, peptide 20
translocation, and lipid flip-flop) corresponded to those observed in toroidal pores created by 21
antimicrobial peptides from multicellular eukaryotes. 22
23
Sizing the lacticin Q pore on LUVs 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
12
We focused on the lacticin Q pore size, because lacticin Q exerts high antimicrobial activity 1
in the nanomolar range. Other antimicrobial peptides forming toroidal pores show activity in 2
the micromolar range. Dansyl-labeled LUVs containing various molecular weights of 3
FITC-dex were applied for sizing the lacticin Q pore. In the case of the FD-10K, the 4
concentration dependence of leakage was similar to that of calcein (Fig. 4). However, 5
leakages of FD-20K and FD-40K were clearly lower than that of FD-10K, indicating that the 6
average lacticin Q pore diameter was between the diameters of FD-10K and FD-20K. 7
8
Intracellular protein leakage by lacticin Q 9
Next, we investigated whether lacticin Q could leak a small protein, GFP (27 kDa), from L. 10
lactis NZgfp cells (Fig. 5). GFP-overexpressing L. lactis NZgfp cells were treated with 11
lacticin Q. Nisin and magainin 2 were also applied as controls. To visualize sufficient number 12
of cells, the cell density of this experiment (see materials and methods) was 5-fold higher 13
than that of in MIC determination described previously (36, 37). Only the lacticin Q 14
treatment (5 µM) led to the loss of GFP fluorescence in the cells, while almost all cells 15
retained GFP in the negative controls (no peptide added), nisin, and magainin 2 (Fig. 5A). 16
Conversely, GFP released by the peptides into the supernatant was monitored with a standard 17
fluorescence substance, Alexa Fluor 430 (Fig. 5B). Only lacticin Q caused an increase in the 18
fluorescence of GFP in agreement with microscopic observation shown in Fig. 5A. 19
We also attempted to detect intracellular proteins that leaked into the supernatant from 20
the peptide-treated cells, because no antimicrobial peptide causing protein leakage had been 21
previously reported. As shown in Fig. 6, lacticin Q resulted in leakage of intracellular 22
proteins from two L. lactis strains and from Li. innocua cells. In the case of L. lactis NZgfp, 1 23
µM lacticin Q caused protein leakage, while 5 µM lacticin Q caused further leakage. To 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
13
detect sufficient amount of protein that can be detected by SDS-PAGE, the cell 1
concentrations of this experiment were 10-fold higher than those in MIC determination 2
described previously (36, 37). It made the concentrations of lacticin Q for protein leakage 3
higher than those of MIC (0.07 µM for L. lactis NZ9000), and 0.1 µM lacticin Q did not 4
cause protein leakage. The protein bands were confirmed to be in the range 10–50 kDa. A 5
band around 30 kDa was observed when GFP was overexpressed, but was not detected 6
without GFP expression (data not shown). These findings support the GFP leakage following 7
the lacticin Q treatment shown in Fig. 5. Similar results were obtained in other strains, L. 8
lactis IL1403 and Li. innocua ATCC 33090T. In all cases, we detected many protein bands, 9
and the bands patterns were manifold depending on the indicator strains. The results reveal 10
that lacticin Q is an antimicrobial peptide causing random protein leakage. 11
12
Discussion 13
In this paper, we show that lacticin Q is the first LAB bacteriocin reported to exert membrane 14
permeation via the toroidal pore. In addition, the pore allows the leakage of macromolecules 15
from both liposomes and cells. We name this novel antimicrobial mechanism of lacticin Q a 16
‘‘huge toroidal pore’’ (HTP), and an HTP model is illustrated in Fig. 7. 17
First, lacticin Q rapidly binds to negatively charged membranes, forming an amphiphilic 18
α-helical structure (Fig. 7A). Previously, we reported that lacticin Q forms an amphiphilic 19
α-helical structure in aqueous conditions and in the presence of small unilamellar vesicles 20
(37). As shown in Fig. 2A, fluorescent energy from tryptophan residues in lacticin Q was 21
transferred to DNS-DHPE on the LUVs within two seconds, and an increase in fluorescence 22
also occurred rapidly upon redistribution to unlabeled LUVs. The observed FRET and its 23
cancellation by excess of unlabeled LUVs can be explained by quick binding between 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
14
lacticin Q and LUVs. In addition, rapid calcein leakage shown in Fig. 1A supports the 1
existence of rapid binding. 2
The membrane permeation by lacticin Q is due to pore formation, and not the result of 3
micellization or fusion of bilayers (Fig. 7B), because no morphological change in LUVs and 4
cells were observed by lacticin Q (Figs. 1B and 5A). Changes in light scattering derived from 5
peptide-induced micellization or fusion of liposomes have been reported (39); However, 6
pore-forming peptides do not affect scattering intensity (31). 7
As shown in Fig. 3, lacticin Q forms pores coupled with lipid flip-flop (Fig. 7B). Lipid 8
flip-flop, a main feature of the toroidal pore, has been characterized in several antimicrobial 9
peptides from multicellular eukaryotes, and the behavior of lacticin Q was similar to that 10
reported for magainin 2 (16). The lacticin Q pore corresponds to the toroidal pore model, but 11
to our knowledge, lacticin Q is the first bacteriocin causing lipid flip-flop in Gram-positive 12
bacteria. Colicin E1 produced by Escherichia coli is the sole bacteriocin reported to form a 13
toroidal pore (30). However, colicin E1 is a protein-type bacteriocin composed of 522 amino 14
acids. As described above, lacticin Q forming toroidal pore exerts its antimicrobial activity in 15
the nanomolar concentration range (MICs of 5–750 nM), but other toroidal pore type 16
antimicrobial peptides are in the micromolar range (MICs of 1–20 µM) (37, 38). Provably, 17
the strong activity of lacticin Q is due to the extraordinarily large size of its pores. 18
The diameter of the lacticin Q pore was determined by leakage of various sizes of 19
FITC-dex (Fig. 7B). Based on the supplier’s information and a previous report (29), the 20
averaged diameters of FD-10K and FD-20K were determined to be approximately 4.6 and 6.6 21
nm, respectively. Since the lacticin Q pore could similarly leak calcein and FD-10K (Fig. 4), 22
the average lacticin Q pore size should be more than 4.6 nm. Furthermore, since leakage of 23
FD-20K was lower than that of FD-10K, the average lacticin Q pore size should be less than 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
15
6.6 nm. This size is the largest among the reported antimicrobial peptides. For example, the 1
diameter of magainin 2 pores was determined to be 2–3 nm (18). In the lactococcal 2
bacteriocins, nisin and lacticin 3147 pore sizes were 2–2.5 and 0.6 nm, respectively (33, 34), 3
while their antimicrobial activities were in the nanomolar ranges, as in lacticin Q (4, 24). 4
Although magainin 2 analogues, magainin 2-PGLa mixture, and melittin allow leakage of a 5
FITC-dex of molecular weight of around 4400 or larger, the peptide concentrations needed 6
were higher than those for calcein leakage (12, 19, 21). 7
The huge pore of lacticin Q was also demonstrated on the bacterial membrane (Figs. 5, 8
6). We observed decreases of GFP fluorescence in the bacterial cells and increases in the 9
supernatant (Fig. 5). Since the size of GFP was determined as approximately 3 × 4 nm by 10
X-ray crystallography (28), the diameter of the lacticin Q pore on the biological membrane 11
would be larger than GFP size. In addition, lacticin Q caused random proteins from the 12
indicator strains to be leaked, which was rarely the case for nisin and magainin 2. Since the 13
cell densities in the protein leakage experiments were 10-fold higher (GFP leakage 14
experiment, 5-fold) than those in MIC determination, protein leakage by lacticin Q against 15
indicator strains in lower densities is expected even in the nanomolar range. Lacticin Q 16
leaked proteins with higher molecular weights than FITC-dexs (Figs. 4, 6), which we 17
speculate was due to the compact tertiary structures of some intracellular proteins. Regardless, 18
there are no other reports of antimicrobial peptides causing protein leakage. This type of 19
protein leakage needs a “huge” pore, therefore, the pore created by lacticin Q was termed a 20
“huge toroidal pore”. 21
Finally, some lacticin Q molecules migrate from the outer to the inner leaflet of the 22
membrane (peptide translocation; Fig. 7C). The ratio of translocation of lacticin Q was 23
related to calcein leakage (Fig. 2B), as shown in Fig. 1A, the rate of calcein leakage was 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
16
initially fast and subsequently slowed. These results indicate that a large amount of lacticin Q 1
molecules accumulated quickly on the outer leaflet of the membrane. After forming a 2
short-time toroidal pore, the pore was closed concomitantly with peptide translocation. As a 3
consequence, lacticin Q molecules which had accumulated on the outer membrane leaflet 4
were distributed to both leaflets, and the rate of pore formation was time-dependently reduced. 5
Short-duration pores have been observed in various antimicrobial peptides (21, 33). Whereas 6
peptide translocation has been previously reported in antimicrobial peptides from 7
multicellular eukaryote such as magainin 2, melittin, and tachyplesin I (17, 20, 21), this study 8
is the first on peptide translocation in a bacteriocin. 9
Nisin and lacticin 3147 pores are smaller than those created by lacticin Q, as described 10
above. However, the former bacteriocins show strong antimicrobial activities in the 11
nanomolar ranges. It has been reported that nisin and lacticin 3147 utilize lipid II as the 12
docking molecule for their antimicrobial mechanisms such as pore formation, inhibition of 13
peptidoglycan biosynthesis, and lipid II segregation (13, 34, 35). On the other hand, lacticin 14
Q forms a HTP without the presence of a docking molecule. The HTP of lacticin Q might be 15
an important factor in achieving high antimicrobial activity levels in the nanomolar range. 16
17
Acknowledgements 18
We thank S. Sugimoto of Kyusyu University, Fukuoka, Japan, for supporting this study. This 19
work was partially supported by a Grant-in-Aid for Scientific Research from the Japan 20
Society for the Promotion of Science (JSPS), Research project for utilizing advanced 21
technologies in agriculture, forestry and fisheries of the Ministry of Agriculture, Forestry and 22
Fisheries of Japan, and Takeda Science Foundation. 23
24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
17
References 1
1. Aso, Y., J. Nagao, H. Koga, K. Okuda, Y. Kanemasa, T. Sashihara, J. Nakayama, 2
and K. Sonomoto. 2004. Heterologous expression and functional analysis of the gene 3
cluster for the biosynthesis of and immunity to the lantibiotic, nukacin ISK-1. J. 4
Biosci. Bioeng. 98:429-436. 5
2. Bartlett, G. R. 1959. Phosphorus assay in column chromatography. J. Biol. Chem. 6
234:466-468. 7
3. Breukink, E. 2006. A lesson in efficient killing from two-component lantibiotics. 8
Mol. Microbiol. 61:271-273. 9
4. Breukink, E., H. E. van Heusden, P. J. Vollmerhaus, E. Swiezewska, L. Brunner, 10
S. Walker, A. J. Heck, and B. de Kruijff. 2003. Lipid II is an intrinsic component of 11
the pore induced by nisin in bacterial membranes. J. Biol. Chem. 278:19898-19903. 12
5. Breukink, E., I. Wiedemann, C. van Kraaij, O. P. Kuipers, H.-G. Sahl, and B. de 13
Kruijff. 1999. Use of the cell wall precursor lipid II by a pore-forming peptide 14
antibiotic. Science 286:2361-2364. 15
6. Brogden, K. A. 2005. Antimicrobial peptides: pore formers or metabolic inhibitors in 16
bacteria? Nat. Rev. Microbiol. 3:238-250. 17
7. Cotter, P. D., C. Hill, and R. P. Ross. 2005. Bacteriocins: developing innate 18
immunity for food. Nat. Rev. Microbiol. 3:777-788. 19
8. Dalet, K., Y. Cenatiempo, P. Cossart, and Y. Hechard. 2001. A σ(54)-dependent 20
PTS permease of the mannose family is responsible for sensitivity of Listeria 21
monocytogenes to mesentericin Y105. Microbiology 147:3263-3269. 22
9. Drider, D., G. Fimland, Y. Hechard, L. M. McMullen, and H. Prevost. 2006. The 23
continuing story of class IIa bacteriocins. Microbiol. Mol. Biol. Rev. 70:564-582. 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
18
10. Fujita, K., S. Ichimasa, T. Zendo, S. Koga, F. Yoneyama, J. Nakayama, and K. 1
Sonomoto. 2007. Structural analysis and characterization of lacticin Q, a novel 2
bacteriocin belonging to a new family of unmodified bacteriocins of gram-positive 3
bacteria. Appl. Environ. Microbiol. 73:2871-2877. 4
11. Gravesen, A., P. Warthoe, S. Knochel, and K. Thirstrup. 2000. Restriction 5
fragment differential display of pediocin-resistant Listeria monocytogenes 412 6
mutants shows consistent overexpression of a putative β-glucoside-specific PTS 7
system. Microbiology 146:1381-1389. 8
12. Hara, T., H. Kodama, M. Kondo, K. Wakamatsu, A. Takeda, T. Tachi, and K. 9
Matsuzaki. 2001. Effects of peptide dimerization on pore formation: Antiparallel 10
disulfide-dimerized magainin 2 analogue. Biopolymers 58:437-446. 11
13. Hasper, H. E., N. E. Kramer, J. L. Smith, J. D. Hillman, C. Zachariah, O. P. 12
Kuipers, B. de Kruijff, and E. Breukink. 2006. An alternative bactericidal 13
mechanism of action for lantibiotic peptides that target lipid II. Science 14
313:1636-1637. 15
14. Hechard, Y., C. Pelletier, Y. Cenatiempo, and J. Frere. 2001. Analysis of 16
σ(54)-dependent genes in Enterococcus faecalis: a mannose PTS permease 17
(EII(Man)) is involved in sensitivity to a bacteriocin, mesentericin Y105. 18
Microbiology 147:1575-1580. 19
15. Hechard, Y., and H.-G. Sahl. 2002. Mode of action of modified and unmodified 20
bacteriocins from Gram-positive bacteria. Biochimie 84:545-557. 21
16. Imura, Y., M. Nishida, and K. Matsuzaki. 2007. Action mechanism of PEGylated 22
magainin 2 analogue peptide. Biochim. Biophys. Acta 1768:2578-2585. 23
17. Imura, Y., M. Nishida, Y. Ogawa, Y. Takakura, and K. Matsuzaki. 2007. Action 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
19
mechanism of tachyplesin I and effects of PEGylation. Biochim. Biophys. Acta 1
1768:1160-1169. 2
18. Matsuzaki, K. 1999. Why and how are peptide-lipid interactions utilized for 3
self-defense? Magainins and tachyplesins as archetypes. Biochim. Biophys. Acta 4
1462:1-10. 5
19. Matsuzaki, K., Y. Mitani, K. Y. Akada, O. Murase, S. Yoneyama, M. Zasloff, 6
and K. Miyajima. 1998. Mechanism of synergism between antimicrobial peptides 7
magainin 2 and PGLa. Biochemistry 37:15144-15153. 8
20. Matsuzaki, K., O. Murase, N. Fujii, and K. Miyajima. 1995. Translocation of a 9
channel-forming antimicrobial peptide, magainin 2, across lipid bilayers by forming a 10
pore. Biochemistry 34:6521-6526. 11
21. Matsuzaki, K., S. Yoneyama, and K. Miyajima. 1997. Pore formation and 12
translocation of melittin. Biophys. J. 73:831-838. 13
22. Mierau, I., and M. Kleerebezem. 2005. 10 years of the nisin-controlled gene 14
expression system (NICE) in Lactococcus lactis. Appl. Microbiol. Biotechnol. 15
68:705-717. 16
23. Miller, W. G., and S. E. Lindow. 1997. An improved GFP cloning cassette designed 17
for prokaryotic transcriptional fusions. Gene 191:149-153. 18
24. Morgan, S. M., M. O'Connor P, P. D. Cotter, R. P. Ross, and C. Hill. 2005. 19
Sequential actions of the two component peptides of the lantibiotic lacticin 3147 20
explain its antimicrobial activity at nanomolar concentrations. Antimicrob. Agents. 21
Chemother. 49:2606-2611. 22
25. Muller, P., S. Schiller, T. Wieprecht, M. Dathe, and A. Herrmann. 2000. 23
Continuous measurement of rapid transbilayer movement of a pyrene-labeled 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
20
phospholipid analogue. Chem. Phys. Lipids. 106:89-99. 1
26. Naghmouchi, K., D. Drider, and I. Fliss. 2007. Action of divergicin M35, a class IIa 2
bacteriocin, on liposomes and Listeria. J. Appl. Microbiol. 102:1508-1517. 3
27. Nakayama, J., Y. Cao, T. Horii, S. Sakuda, A. D. Akkermans, W. M. de Vos, and 4
H. Nagasawa. 2001. Gelatinase biosynthesis-activating pheromone: a peptide lactone 5
that mediates a quorum sensing in Enterococcus faecalis. Mol. Microbiol. 6
41:145-154. 7
28. Ormo, M., A. B. Cubitt, K. Kallio, L. A. Gross, R. Y. Tsien, and S. J. Remington. 8
1996. Crystal structure of the Aequorea victoria green fluorescent protein. Science 9
273:1392-1395. 10
29. Scherrer, R., and P. Gerhardt. 1971. Molecular sieving by the Bacillus megaterium 11
cell wall and protoplast. J. Bacteriol. 107:718-735. 12
30. Sobko, A. A., E. A. Kotova, Y. N. Antonenko, S. D. Zakharov, and W. A. Cramer. 13
2006. Lipid dependence of the channel properties of a colicin E1-lipid toroidal pore. J. 14
Biol. Chem. 281:14408-14416. 15
31. Tachi, T., R. F. Epand, R. M. Epand, and K. Matsuzaki. 2002. Position-dependent 16
hydrophobicity of the antimicrobial magainin peptide affects the mode of 17
peptide-lipid interactions and selective toxicity. Biochemistry 41:10723-10731. 18
32. Vanderkooi, J. M., and J. B. Callis. 1974. Pyrene. A probe of lateral diffusion in the 19
hydrophobic region of membranes. Biochemistry 13:4000-4006. 20
33. Wiedemann, I., R. Benz, and H.-G. Sahl. 2004. Lipid II-mediated pore formation by 21
the peptide antibiotic nisin: a black lipid membrane study. J. Bacteriol. 22
186:3259-3261. 23
34. Wiedemann, I., T. Bottiger, R. R. Bonelli, A. Wiese, S. O. Hagge, T. Gutsmann, 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
21
U. Seydel, L. Deegan, C. Hill, P. Ross, and H.-G. Sahl. 2006. The mode of action of 1
the lantibiotic lacticin 3147--a complex mechanism involving specific interaction of 2
two peptides and the cell wall precursor lipid II. Mol. Microbiol. 61:285-296. 3
35. Wiedemann, I., E. Breukink, C. van Kraaij, O. P. Kuipers, G. Bierbaum, B. de 4
Kruijff, and H.-G. Sahl. 2001. Specific binding of nisin to the peptidoglycan 5
precursor lipid II combines pore formation and inhibition of cell wall biosynthesis for 6
potent antibiotic activity. J. Biol. Chem. 276:1772-1779. 7
36. Yoneyama, F., M. Fukao, T. Zendo, J. Nakayama, and K. Sonomoto. 2008. 8
Biosynthetic characterization and biochemical features of the third natural nisin 9
variant, nisin Q, produced by Lactococcus lactis 61-14. J. Appl. Microbiol. 10
105:1982-1990. 11
37. Yoneyama, F., Y. Imura, S. Ichimasa, K. Fujita, T. Zendo, J. Nakayama, K. 12
Matsuzaki, and K. Sonomoto. 2009. Lacticin Q, a lactococcal bacteriocin, causes 13
high-level membrane permeability in the absence of specific receptors. Appl. Environ. 14
Microbiol. 75:538-541. 15
38. Zasloff, M. 2002. Antimicrobial peptides of multicellular organisms. Nature 16
415:389-395. 17
39. Zhao, H., R. Sood, A. Jutila, S. Bose, G. Fimland, J. Nissen-Meyer, and P. K. 18
Kinnunen. 2006. Interaction of the antimicrobial peptide pheromone Plantaricin A 19
with model membranes: implications for a novel mechanism of action. Biochim. 20
Biophys. Acta. 1758:1461-1474. 21
22
23
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
22
1
Figure Legends 2
Figure 1. Effects of lacticin Q on LUVs, calcein leakage and morphological change. (A) 3
Lacticin Q-induced leakage of calcein entrapped in PC/PG (1:1) LUVs. From the top, traces 4
show the leakage with lacticin Q added at 250, 125, 62.5 and 31.3 nM concentrations. The 5
lipid concentration of PC/PG LUVs was 50 µM. As controls, 0% and 100% leakage were 6
obtained by the addition of buffer and 0.1% Triton X-100, respectively. (B) Light scattering 7
of PC/PG LUVs treated by lacticin Q. The arrow indicates the addition of PC/PG LUVs (50 8
µM lipids) to 250 nM lacticin Q (black line) and buffer (gray line). 9
10
Figure 2. Transbilayer movement (translocation) of lacticin Q. (A) Detection of lacticin Q 11
translocation using FRET. Tryptophan fluorescence of 125 nM lacticin Q was monitored at 12
EM of 336 nm and EX of 280 nm. Solid and dotted arrows show the addition of 13
dansyl-labeled LUVs (PC/PG/DNS-DHPE, 4:5:1) and excess unlabeled (PC/PG, 1:1) LUVs, 14
respectively. As a control (gray line), the dansyl-labeled and unlabeled LUVs were added 15
together. Lipid concentrations of dansyl-labeled and large excess unlabeled LUVs were 16
adjusted to 50 µM and 350 µM, respectively. (B) Co-occurrence of translocation of lacticin Q 17
and calcein leakage. Open circles show translocated lacticin Q molecules (%) calculated from 18
panel A data (see experimental procedures). The line indicates leakage of calcein entrapped in 19
the dansyl-labeled LUVs (PC/PG/DNS-DHPE, 4:5:1). The leakage experiment was 20
performed under the same conditions as in translocation detection (125 nM lacticin Q and 50 21
µM lipids). 22
23
Figure 3. Lipid flip-flop induced by lacticin Q. (A) Fluorescence spectra of pyrene-labeled 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
23
LUVs (PC/PG/pyPC, 47:50:3). Trace 1 indicates the fluorescence spectra of the 1
symmetrically labeled LUVs; traces 2 and 3 are asymmetrically labeled LUVs and 250 nM 2
lacticin Q-treated LUVs, respectively. The inset is an enlarged view of the same spectra at 3
450–500 nm. Lipid concentration of the pyrene-labeled LUVs was 50 µM. (B) Co-occurrence 4
of lipid flip-flop and calcein leakage by lacticin Q. The open circles show lipid flip-flop (%) 5
calculated from the ratio of Ie/Im (see experimental procedures). The line indicates leakage of 6
calcein entrapped in the pyrene-labeled LUVs (PC/PG/pyPC, 47:50:3). The leakage 7
experiment was performed under the same conditions as for lipid flip-flop (250 nM lacticin Q 8
and 50 µM lipids). 9
10
Figure 4. FITC-dex leakage by lacticin Q. Dansyl-labeled LUVs (PC/PG/DNS-DHPE, 11
45:50:5) entrapping FITC-dex were treated by various concentrations of lacticin Q for 10 min. 12
Open circles, triangles, and squares indicate leakage (%) of FD-10K, FD-20K, and FD-40K, 13
respectively. Gray circles show leakage (%) of calcein (molecular weight, 622) determined by 14
the method in Fig. 1. The concentration of lipids was 50 µM. 15
16
Figure 5. GFP leakage from L. lactis NZgfp. (A) Microscopic observation of 17
GFP-overexpressed and peptide-treated cells. The left and light pictures are bright-field views 18
and their fluorescence images, respectively. (B) Detection of GFP leaked into supernatant. 19
The ordinate axis shows FGFP/FAF: the ratio of fluorescence intensity of 520 nm excited by 20
490 nm (GFP) and 440 nm (Alexa Fluor 430). 21
22
Figure 6. SDS-PAGE of proteins leaked by peptide treatment. Indicator strains, L. lactis 23
NZgfp, L. lactis IL1403, and Li. innocua ATCC 33090T, were treated by lacticin Q, nisin, and 24
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
24
magainin 2 in buffer A, and the proteins leaked into the supernatant were detected by 1
SDS-PAGE with 12% polyacrylamide gel and silver staining. 2
3
Figure 7. A model of a huge toroidal pore (HTP) mediated by lacticin Q. (A) Lacticin Q 4
rapidly binds to negatively charged phospholipid bilayer. (B) Lacticin Q forms a HTP 5
accompanied by lipid flip-flop. The average diameter of the peptide-lipid supramolecular 6
complex pore is 4.6−6.6 nm resulting in protein leakage. (C) Some lacticin Q molecules show 7
transbilayer movement from the outer to the inner leaflet of the bilayer (translocation) when 8
the pore closes. 9
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
25
1
0 1 2 3 4 50
20
40
60
80
100
Le
akag
e (
%)
Time (min)
0 1 2 3 4 5
Time (min)
0
100
200
300
600
Scatt
eri
ng
inte
nsity
(a.
u.)
400
500
(A)
(B)
Fig. 1. Yoneyama et al.
250 nM (Lipid:peptide, 200:1)
125 nM
62.5 nM
31.3 nM
LUVs
250 nM lacticin Q
Buffer
2
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
26
1
0
15
30
45
60
Le
akag
e (
%)
5
0
10
15
20
25
Tra
nslo
catio
n (
%)
0 1 2 3 4 5
Time (min)
250
200
150
100
50
0
Flu
ore
sce
nce
inte
nsity (
a. u
.)
0 1 2 3 4 5
Time (min)
(A)
(B)
Fig. 2. Yoneyama et al.
2
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
27
1
100
30
40
50
60
70
450 500460 470 480 490
400
300
200
100
0350 400 450 500
Flu
ore
sce
nce
inte
nsity (
a. u
.)
Wavelength (nm)
80
60
40
20
0
60
45
30
15
00 1 2 3 4 5
Time (min)
Le
akag
e (
%)
Flip
-flo
p (
%)
(A)
(B)
1
3
2
1
3
2
Fig. 3. Yoneyama et al.
2
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
28
1
0
20
40
60
80
100
Le
akag
e (
%)
0 100 200 300 400 500
Lacticin Q concentration (nM)
FD-10K
FD-20K
FD-40K
calcein
Fig. 4. Yoneyama et al.
2
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
29
1
5 µ
M L
acticin
Q1
0 µ
M M
ag
ain
in2
10
µM
Nis
inN
o p
ep
tid
e
Bright-field Fluorescence(A)
(B)
0.90
0.65
0.40
0.15Flu
ore
sce
nce
(4
90
/44
0 n
m)
Fig. 5. Yoneyama et al.
2
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
30
1
Li. innocua ATCC 33090T
66
30
20
(kDa)
45
L. lactis IL1403L. lactis NZgfp
GFP
Fig. 6. Yoneyama et al.
2
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from
31
1
Binding
4.6–6.6 nm
Flip-flop
Translocation
(A)
(B)
(C)
Fig. 7. Yoneyama et al.
2
3
on June 22, 2018 by guesthttp://aac.asm
.org/D
ownloaded from