ORIGINAL PAPER

Rapid identification, by use of the LTQ Orbitrap hybrid FTmass spectrometer, of antifungal compounds producedby lactic acid bacteria

Brid Brosnan & Aidan Coffey & Elke K. Arendt &Ambrose Furey

Received: 16 December 2011 /Revised: 13 March 2012 /Accepted: 18 March 2012 /Published online: 19 April 2012# Springer-Verlag 2012

Abstract Fungal contamination of food causes health andeconomic concerns. Several species of lactic acid bacteria(LAB) have antifungal activity which may inhibit foodspoilage fungi. LAB have GRAS (generally recognised assafe) status, allowing them to be safely integrated into foodsystems as natural food preservatives. A method is de-scribed herein that enables rapid screening of LAB culturesfor 25 known antifungal compounds associated with LAB.This is the first chromatographic method developed whichenables the rapid identification of a wide range of antifungal

compounds by a single method with a short analysis time(23 min). Chromatographic separation was achieved on aPhenomenex Gemini C18 100A column (150 mm×2.0 mm;5 μm) by use of a mobile-phase gradient prepared from (A)water containing acetic acid (0.1%) and (B) acetonitrile con-taining acetic acid (0.1%), at a flow rate of 0.3 µL min−1. Thegradient involved a progressive ramp from 10–95% acetoni-trile over 13 min. The LC was coupled to a hybrid LTQOrbitrap XL fourier-transform mass spectrometer (FTMS)operated in negative ionisation mode. High mass accuracydata (<3 ppm) obtained by use of high resolution (30,000 K)enabled unequivocal identification of the target compounds.This method allows comprehensive profiling and comparisonof different LAB strains and is also capable of the identifica-tion of additional compounds produced by these bacteria.

Keywords LAB . Antifungal compounds . LTQOrbitrapXL . High mass accuracy data . LC–FTMS

Introduction

Lactic acid bacteria (LAB) are commonly associated withfermented foods as a result of their biopreservative capacity.Consumer demand for use of natural food preservatives hasbeen the impetus for much research devoted to the identifi-cation of bacterial strains with activity against agents offood spoilage without the need for the addition of chemicalpreservatives. LAB have been used as starter cultures in avariety of food products, for example fermented dairy prod-ucts, fermented meats and vegetables, sourdoughs, and si-lage [1–3]. The GRAS status of LAB promotes them as avery favourable option for integration as natural preserva-tives into food and feed. Starter cultures can be exempt fromEU regulations of additives and labelling if they can be

Published in the special paper collection Recent Advances in FoodAnalysis with guest editors J. Hajslova, R. Krska, and M. Nielen.

Electronic supplementary material The online version of this article(doi:10.1007/s00216-012-5955-1) contains supplementary material,which is available to authorized users.

B. Brosnan :A. FureyTeam Elucidate, Department of Chemistry,Cork Institute of Technology (CIT),Bishopstown,Cork, Ireland

A. CoffeyDepartment of Biological Sciences,Cork Institute of Technology,Bishopstown,Cork, Ireland

E. K. ArendtDepartment of Food and Nutritional Sciences,University College Cork,Cork, Ireland

A. Furey (*)Department of Chemistry, Proteobio,Mass Spectrometry Centre for Proteomics and Biotoxin Research,CIT, Bishopstown,Cork, Irelande-mail: ambrose.furey@cit.ie

Anal Bioanal Chem (2012) 403:2983–2995DOI 10.1007/s00216-012-5955-1

categorised as a processing aid [22]. The author Wessels[23] discusses the variants that occur within the regulationof LAB and their integration into the food chain. Some LABhave been shown to inhibit the growth of several species ofbacteria and moulds that have a negative effect on food fromthe perspective of spoilage and safety [4–6]. Recently, mem-bers of the LAB have been used to reduce fungal growth andprolong the shelf life of vegetables and bread as an alterna-tive to artificial preservatives [7–10].

Initially it was believed that the organic acids producedby LAB, namely lactic and acetic acids, were the mainagents of antifungal activity because of the lowering of thepH. However, studies have shown that the MIC (minimuminhibitory concentration) of these compounds alone is toolow to fully account for the antimicrobial activity observed[4, 11]. Other metabolites from these bacteria have beenidentified as the source of the antifungal activity. Theseinclude cyclic dipeptides, proteinaceous compounds, fattyacids and a variety of other low-molecular-weight com-pounds [3, 4, 11–16]. The individual MIC of these metab-olites are much higher than the amounts detected in culturebroths. It is, in fact, the synergy of all these compoundssecreted by the bacteria that seems to inhibit fungal growth[4, 11]. The method described herein allows comprehensiveinformation about these compounds to be gathered in orderto assess their synergistic effects and thus a full understand-ing of their antifungal activity.

An LC–MS method in negative ionisation mode wasdeveloped to identify low molecular weight compounds,fatty acids and nucleic acids. Chromatographic separationof the compounds was conducted to identify compoundswith the same chemical formula. A Thermo LTQ OrbitrapXL with high resolution capabilities was used to identify,unambiguously, the compounds present. Previously pub-lished methods have required complementary approachesto fully identify the compounds present in samples, includ-ing using both GC–MS and LC–MS to achieve full sampleprofiling [4, 11, 12, 16]. This method enables antifungalcompounds to be determined in a single chromatographicrun.

This is the first method enabling comprehensive,rapid separation of known antifungal compounds fol-lowed by identification with high mass accuracy. LABhave the potential to be integrated into food or feed forconsumption, without the need for labelling. Knowledgeof the active metabolites providing antifungal activitywill be of utmost importance. Identification and quanti-fication of compounds produced by LAB will providefurther data and aid explanation of the synergisticeffects that seem to provide the antimicrobial activity.This method can also be used to assess the presence ofantimicrobial compounds when they have been integrat-ed into food and feed products.

Materials and methods

Materials

Antifungal standard compounds (vanillic acid, (S)-(−)-2-hydroxyisocapric acid, ferulic acid, azelaic acid, 4-hydroxybenzoic acid, decanoic acid, caffeic acid, methylcin-namic acid, benzoic acid, hydrocinnamic acid, 3-(4-hydroxy-3-methoxyphenyl)propionic acid, 3,4-dihydroxyhydrocinnamicacid, 3-(4-hydroxyphenyl)propionic acid, ρ-coumaric acid, DL-ρ-hydroxyphenyllactic acid, D-glucuronic acid, cytidine, 2-deoxycytidine, 2-hydroxydodecanoic acid, DL-β-hydroxylauricacid, 1,2-dihydroxybenzene, DL-β-hydroxymyristic acid, 3-hydroxydecanoic acid, hydrocinnamic acid d9 OH, and salicylicacid),MRSbroth (used for the cultivationofmanyLAB), and themobile phase additive acetic acid were purchased from Sigma–Aldrich (Dublin, Ireland). Phenyllactic acid was obtained fromBachem (Weil am Rhein, Germany). Compound structures areshown in Fig. 1. LC–MS grade solvents were sourced fromThermo Fisher Scientific (Dublin, Ireland).

Preparation of standards

Preparation of standards involved using a stock solutioncontaining all the pertinent compounds. This stock solutionwas prepared by combining the individual standards(2 mg mL−1). The standards were prepared by dissolvingthe compounds in three working solutions:

1. the initial LCmobile phase (90%water, 10% acetonitrile);2. blank broth after liquid–liquid extraction (LLE ethyl

acetate×3); and3. blank broth (0.2 μm filtered)

in order to compare matrix effects on the signals obtainedfor the compounds. Deuterated hydrocinnamic acid (hydro-cinnamic acid-d9 OH) was used as an internal standard.

Sample preparation

Strains with strong antifungal activity were selected, namelyLactobacillus amylovorus FST2.1, Lactobacillus arizono-nas R13, Lactobacillus plantarum FST1.7, Lactobacillusreuteri R2, and Weisella cibaria PS2. An uninoculatedMRS broth (no bacteria added) was used as a negativecontrol. The antifungal strain was grown in MRS brothat 37 °C for 48 h. It was then centrifuged at 13,000 rpm(7,412g) and the supernatant was filter-sterilized(0.45 μm pore size). Samples were prepared in two waysfor analysis:

1. supernatant was filtered (0.22 μm); and2. supernatant was extracted with ethyl acetate (10 mL×1,

5 mL×2), the combined organic layers were dried using a

2984 B. Brosnan et al.

A B C

D E F

G H I

J K L

M N O

P Q R

S T U

V W X

Y Z

Fig. 1 Structures of the 25 known antifungal compounds (listed inorder of chromatographic elution): A, cytidine; B, 2-deoxycytidine; C,D-glucuronic acid; D, DL-ρ-hydroxyphenyllactic acid; E, 1,2-dihydrox-ybenzene; F, 3,4-dihydroxyhydrocinnamic acid; G, 4-hydroxybenzoicacid; H, caffeic acid; I, vanillic acid; J, (S)-(−)-2-hydroxyisocaproicacid; K, 3-(4-hydroxyphenyl)propionic acid; L, 3-(4-hydroxy-3-

methoxyphenyl)propanoic acid; M, p-coumaric acid; N, ferulic acid;O, azelaic acid; P, phenyllactic acid; Q, benzoic acid; R, hydrocinnamicacid; S, methylcinnamic acid; T, 3-hydroxydecanoic acid; U, DL-β-hydroxylauric acid; V, decanoic acid; W, DL-β-hydroxymyristic acid;X, 2-hydroxydodecanoic acid; Y, salicylic acid; Z, hydrocinnamic acidd9 OH (internal standard)

LC-FTMS of LAB antifungal compounds 2985

rotary evaporator and reconstituted in the initialmobile phase composition solution (90% water,10% acetonitrile).

Before injection for LC–MS analysis, samples were filtered(0.22 μm) and placed in an amber vial (2 mL) ready foranalysis. The blank MRS broth was treated in the same wayas the broth containing the antifungal strain.

Instrumentation and Analytical Conditions

LC instrumentation

Chromatography was performed with an Accela LC system(Thermo Fisher Scientific, Hemel Hempstead, UK).Separation of the compounds was achieved on a GeminiC18 column (150 mm×2 mm, 5 μm; Phenomenex,Macclesfield, UK) equipped with a SecurityGuard cartridge(C18, 4 mm×2 mm; Phenomenex). The column was main-tained at 30 °C and a flow rate of 300 μL min−1. Elution wasby use of a stepped gradient prepared from water containing0.1% acetic acid (component A) and acetonitrile contain-ing 0.1% acetic acid (component B). Initial conditionswere 10% B held for 3 min increasing to 95% B over10 min; this was held for 3 min before returning to theinitial starting conditions to equilibrate. To shorten theequilibration time to 10 min the LC flow rate wasincreased to 400 μL min−1 (9 min) and then revertedto the analytical flow-rate (1 min).

Mass spectrometry conditions

A hybrid LTQ Orbitrap XL FTMS (Thermo FisherScientific, Hemel Hempstead, UK) was connected to theLC system. It was operated in negative ionisation modewith a heated electrospray interface (HESI). A universalion source tune method was developed by running each ofthe compounds through the LC column and by varying thecapillary temperature, capillary voltage, tube lens voltage,sheath gas, and auxiliary gas independently. The maximumfor each was observed and the values chosen were: capillarytemperature 300 °C, capillary voltage −50 V, tube lens−110 V, sheath gas 45 arb units, auxiliary gas 15 arb units.The instrument was calibrated in accordance with the man-ufacturer’s instructions by using both the positive modecalibration solution (caffeine, MRFA, and ultramark 1621)and the negative mode calibration solution (sodium dodecylsulfate, sodium taurochlate, and ultramark 1621).Resolution of 30,000 FWHM was selected, owing to thelarge number of target compounds involved, because thisresolution resulted in the best reproducibility, giving suffi-cient data points for all compounds.

Validation

Validation was completed for all three standards (mobilephase, LLE blank MRS broth extract, and blank filteredMRS broth). A calibration curve containing six points andone control were run initially (n03) to assess the intradayspecificity, linearity, trueness, and precision of the method.This calibration was then repeated twice, consecutively, toobtain interday data (n09).

Results and discussion

A method has been developed that enables rapid detection ofthe main antifungal compounds (Fig. 1) produced by LAB. Incontrast with other methods, it enables identification of abroad range of compounds in a single analysis. Previousmethods required both GC–MS and LC–MS as complemen-tary methods to discern the compounds [4, 11, 12, 16]. Themost commonly employed methods are based on that reportedby Strom [12]. This method involves sample clean-up by SPEwith the eluent being fractioned through a C18 semi-prepcolumn. These fractions are then run through a Hypercarbcolumn. The fractions are collected and the compounds arefinally identified by use of three separate techniques: GC–MS,LC–MS, and NMR. The method presented here enables com-plete identification of the LAB compounds by use of a shortprocedure with assured identity as a result of the high massaccuracy data produced by the LTQ Orbitrap XL FTMS.

Chromatographic separation

Figure 2 illustrates the chromatographic separation and peakshape achieved for the twenty-five target compounds.Although there is some co-elution of the compounds thisis offset by the high mass accuracy capabilities of the LTQwhich enables unequivocal differentiation. Four sets ofcompounds in this mixture have the same chemical formulagiving them the same exact mass (Fig. 1): DL-ρ-hydroxy-phenyllactic acid and 3,4-dihydroxyhydrocinnamic acidwith m/z 181.05008 [M−H]− (compounds D and F, respec-tively); 4-hydroxybenzoic acid and salicylic acid with m/z137.02387 [M−H]− (compounds G and Y, respectively); 3-(4-hydroxy-3-methoxyphenyl)propionic acid and phenyllac-tic acid at m/z 165.05517 [M−H]− (compounds K and P,respectively), and DL-β-hydroxylauric acid and 2-hydroxydodecanoic acid at m/z 215.16472 [M−H]− (com-pounds U and X, respectively). Their chemical structuresdiffer only in the location of the hydroxyl groups. To iden-tify and quantify same mass compounds (D and F, G and Y,K and P, and U and X) chromatographic resolution wasnecessary (Fig. 2).

2986 B. Brosnan et al.

Matrix effects

Matrix effects were assessed by running standards preparedin different matrices to determine whether any suppression(or enhancement) of the analyte signal occurred as a resultof interfering compounds. Minimal matrix effects were ob-served. Previous methods for extraction of LABs used eitherprotracted sample-preparation techniques, for example SPEor LLE, or direct sample injection [3, 4, 11–16]. In thiswork, LLE and direct injection with standards prepared inthe mobile phase were compared. The reason for omissionof the SPE step was our initial observation of partitioning ofthe compounds between the load, wash, and elute steps (datanot included). It is of concern that most published methodsreport only analysis of the SPE eluent, which consideringthe vast differences in the polarity of the compounds willalmost certainly have led to substantial losses during theload and wash steps. This observation was previously de-scribed by Armaforte [15] who reported 10% recovery ofphenyllactic acid in the elute step and clear preference forthe compound in the wash step. Aramforte compared directinjection with SPE and showed that reproducibility and re-covery of phenyllactic acid were much better after directinjection. Thus SPE, although an excellent clean-up tech-nique, must be carefully optimised and validated for samplesthat contain target analytes with a wide range of polarity.

ControlMRS broths were incubated at 37 °C for 48 h, as perthe sample strain preparation. The broth was sterile-filtered

through a 0.45 μm mesh to prevent fungal and/or bacterialgrowth developing. For the direct injection method, the samplewas further filtered through a 0.22 μm filter and placed in anamber vial (2 mL) before LC injection. This was to ensure noparticulates were present in the samples injected (10 μL) whichcould lead to blockages of the LC injection port, tubing or pre-filter on the LC instrument, upstream of the SecurityGuardcartridge. The LLE method was optimised for 10 mL MRSbroth extracted with ethyl acetate (10 mL×1, 5 mL×2).Extraction was performed in triplicate and the organic layerswere combined and dried. The sample and standards were thenreconstituted and dissolved in the initial mobile phase (90:10water–acetonitrile).

For intra day studies several concentrations (10, 25, 50,100, 500, and 750 μg mL−1) of the standard antifungal com-pounds (n025) were added to the initial mobile phase solu-tion, extracted by LLE, and added to the crude broth extracts,as outlined in the section “Preparation of standards”. Table 2outlines the results for linear range, LOD, LOQ, trueness, andprecision for the antifungal standards which were subsequent-ly identified as being produced by LAB strain Lactobacillusamylovorus FST2.1 (Table 1). Electronic SupplementaryMaterial Table S2a and b list the results obtained for linearrange, LOD, LOQ, trueness, and precision for all 25 antifun-gal standards that were investigated as part of this study.

These samples were all analysed by LC–FTMS. Calibrationplots were generated for the standard mixture of compounds inthe three matrices and the RSD (relative standard deviation)

Fig. 2 Chromatogram obtained from the antifungal standards separat-ed on a Gemini C18 column as outlined in the section “ LC instru-mentation”: (i) C, D-glucuronic acid; D, DL-ρ-hydroxyphenyllacticacid; E, 1,2-dihydroxybenzene; F, 3,4-dihydroxyhydrocinnamic acid;H, caffeic acid; M, p-coumaric acid; O, azelaic acid; Q, benzoic acid;R, hydrocinnamic acid; S, methylcinnamic acid; T, 3-hydroxydecanoicacid; U, DL-β-hydroxylauric acid; X, 2-hydroxydodecanoic acid; (ii) I,

vanillic acid; J, (S)-(−)-2-hydroxyisocaproic acid; L, 3-(4-hydroxy-3-methoxyphenyl)propanoic acid; N, ferulic acid; W, DL-β-hydroxymyr-istic acid; (iii) G, 4-hydroxybenzoic acid; K, 3-(4-hydroxyphenyl)pro-pionic acid; P, phenyllactic acid; Y, salicylic acid; (iv) B, 2-deoxycytidine;(v) V, decanoic acid; (vi) A, cytidine. The deuterated internal standard(hydrocinnamic acid d9 OH, 200 µg mL–1) co-eluted with (i) R, hydro-cinnamic acid at 8.42 min and was therefore not included

LC-FTMS of LAB antifungal compounds 2987

values obtained were compared. Table 2 and ElectronicSupplementary Material Table S2a and b show linearityfor all the compounds was good. However, there were slightdifferences between the slopes of the lines observed for somecompounds from the LLE step (Table 2(ii), and ElectronicSupplementary Material Table S2a(ii) and b(ii)) and the crudebroth matrix (Table 2(iii), and Electronic SupplementaryMaterial Table S2a(iii) and b(iii)) compared with the mobilephase composition solution (Table 2(i), and ElectronicSupplementary Material Table S2a(i) and b(i)). RSD values<10% were obtained for both the mobile phase and the LLEspiked standards. RSD values for standards added to the crudefiltered broths varied substantially, ranging from 8–173%

for the 10 μg mL−1 spiked standard to 2.3–39% for the750 μg mL−1 spiked standard (Table 2a and b(iii)). This wasexpected because of the complex matrix of the crude broth, forwhich filtration (0.22 μm mesh) was the only clean-up stepused. It was frequently noted that blockage of the ion-transfertube of the hybrid Orbitrap FTMS occurred during analysis ofcrude broth filtered extracts. This resulted in an immediatedecrease of the ion signal during chromatographic runs. It wasdecided that the chromatographic mobile phase should flowconstantly into the H-ESI source between runs without use ofthe LC–MS software feature of diverting the flow to waste.This was done to prevent such blockages. Diverting portionsof the chromatographic flow to waste (during the first 1–2 min

Table 1 Name, chemical formula, theoretical mass [M−H]−, detectedmass [M−H]−, and calculated ppm error for 25 LAB standards deter-mined with the LTQ Orbitrap XL hybrid mass spectrometer. Also

included are the LAB compounds (n014) detected (masses [M−H]−

and ppm errors) and quantified (mg L−1) in the antifungal strainLactobacillus amylovorus FST2.1

Compound Chemicalformula

[M−H]−

theoretical m/zStandards Lactobacillus amylovorus FST2.1 broth

[M−H]−

found m/zppm error [M−H]−

found m/zppm error Concentration

(mg L−1)

A) Cytidine C9H13N3O5 242.07770 242.07726 1.82

B) 2-Deoxycytidine C9H13N3O4 226.08278 226.08263 0.66

C) D-Glucuronic acid C6H10O7 193.03483 193.03468 0.78

D) DL-ρ-Hydroxyphenyllactic acid C9H10O4 181.05008 181.04997 0.61 181.05000 0.44 228.12

E) 1,2-Dihydroxybenzene C6H6O2 109.02895 109.02908 −1.19

F) 3,4-Dihydroxyhydrocinnamic acid C9H10O4 181.05008 181.05000 0.44

G) 4-Hydroxybenzoic acid C7H6O3 137.02387 137.02380 0.51 137.02417 −1.68 1.51

H) Caffeic acid C9H8O4 179.03443 179.03436 0.39

I) Vanillic acid C8H8O4 167.03443 167.03435 0.48 167.03432 0.6 4.31

J) (S)-(−)-2-Hydroxyisocaproic acid C6H12O3 131.07082 131.07086 −0.31 131.07072 0.76 1057.23

K) 3-(4-Hydroxyphenyl)propionic acid C9H10O3 165.05517 165.05513 0.24

L) 3-(4-Hydroxy-3-methoxyphenyl)propanoic acid

C10H12O4 195.06573 195.06554 0.97

M) p-Coumaric acid C9H8O3 163.03952 163.03938 0.86 163.03944 0.49 14.01

N) Ferulic acid C10H10O4 193.05008 193.04994 0.73

O) Azelaic acid C9H16O4 187.09703 187.09673 1.6 187.09709 −0.32 1.21

P) Phenyllactic acid C10H12O3 165.05517 165.05495 1.33 165.05521 −0.24 1333.03

Q) Benzoic acid C7H6O2 121.02895 121.02901 −0.5 121.02904 −0.74 1.11

R) Hydrocinnamic acid C9H10O2 149.06025 149.06015 0.67 149.06017 0.54 451

S) Methylcinnamic acid C10H10O2 161.06025 161.06007 1.12

T) 3-Hydroxydecanoic acid C10H20O3 187.13342 187.13332 0.53 187.13336 0.32 170.52

U) DL-β-Hydroxylauric acid C12H24O3 215.16472 215.16447 1.16 215.16443 1.35 433.91

V) Decanoic acid C10H20O2 171.1385 171.13824 1.52 171.13865 −0.88 3.21

W) DL-β-Hydroxymyristic acid C14H28O3 243.19602 243.19565 1.52

X) 2-Hydroxydodecanoic acid C12H24O3 215.16472 215.16444 1.3 215.16440 1.49 427.52

Y) Salicylic acid C7H6O3 137.02387 137.02385 0.15 137.02386 0.07 9.91

Hydrocinnamic acid D9 C9HO2D9 158.11675 158.1165 1.58 158.11684 −0.57

1 10-fold concentrated2 1:10 dilution3 1:20 dilution

2988 B. Brosnan et al.

Tab

le2

Resultsforlin

earrang

e,LOD,L

OQ,trueness,andprecisionforkn

ownantifun

galstand

ards

which

weresubsequently

identifiedas

beingprod

uced

byLABstrain

Lactoba

cillu

sam

ylovorus

FST2.1(Table1).T

hese

13standardswereassessed

inthreematrixsolutio

ns:(i)standardsspiked

into

theinitial

mob

ileph

asecompo

sitio

n(90:10

%water–acetonitrile),(ii)standardsspiked

into

blankMRSbrothextractafterLLE,and(iii)

standardsspiked

into

blankcrud

eMRSbrothafterfiltration

Com

pound

Equationof

thelin

eLinearrange

(mgL−1)

R2

LOD

(mgL−1)

LOQ

(mgL−1)

Control

theoretical

concentration

(mgL−1)

Control

measured

concentration

(mgL−1)

RE(%

)RSD

(%,n=3)

Intraday

RSD

%(n=3)

10mgL−1

25mgL−1

50mgL−1

100mgL−1

500mgL−1

750mgL−1

D)DL-ρ-

Hydroxyphenyllactic

acid

i)Y=0.0011x−

0.0134

10–750

0.9985

0.4

0.1

250

236

5.21

5.65

4.23

5.04

5.77

5.58

4.03

7.57

ii)Y=0.0011x−

0.0049

10–750

0.9988

10.3

250

240

3.88

7.39

6.06

9.94

3.61

0.87

7.38

9.57

iii)Y=0.001x

−0.00115

25–750

0.999

10.3

250

233

6.64

1.94

n/d

3.10

7.41

11.38

7.51

6.61

G)4-Hydroxybenzoic

acid

i)Y=0.0013

x+0.0115

10–750

0.9957

0.13

0.4

250

279

−11.85

5.12

3.50

3.88

9.68

9.12

2.26

9.37

ii)Y=0.0011x+

0.0439

10–750

0.9959

0.2

0.07

250

239

4.07

5.27

9.13

4.30

1.74

3.67

5.31

9.19

iii)Y=0.0013

x−0.020

10–750

09947

0.3

0.1

250

282

−12.85

5.47

173.2

4.96

10.57

173.2

4.85

5.47

I)Vanillic

acid

i)Y=0.0015

x−0.0216

10–750

0.9984

0.5

0.15

250

228

8.73

5.12

2.56

3.42

6.76

6.51

3.47

5.40

ii)Y=0.0003x+

0.0156

10–750

0.9922

0.8

0.25

250

289

−15.93

6.40

3.34

4.64

4.37

1.43

4.10

9.12

iii)Y=0.0006

x+0.0046

25–750

0.9922

2.5

0.8

250

252

−0.94

5.81

n/d

8.05

9.66

12.50

6.40

15.29

J)(S)-(−)-2-

Hydroxyisocaproic

acid

i)Y=0.0031

x+0.0343

10–750

0.9987

0.07

0.035

250

280

−12.22

2.62

2.73

7.60

7.06

6.59

7.15

7.52

ii)Y=0.0021x+

0.0688

10–750

0.9988

0.08

0.035

250

221

11.45

5.38

2.74

3.92

0.69

7.89

8.00

6.74

iii)Y=0.0033

x+0.0461

10–750

0.9909

0.1

0.03

250

290

−16.09

6.60

173.2

12.57

4.35

8.38

8.36

10.50

L)3-(4-H

ydroxy-3-

methoxyphenyl)

propanoic

acid

i)Y=0.0013

x−0.0014

10–750

0.9996

0.05

0.015

250

266

−6.59

5.93

7.56

1.13

4.82

2.70

0.92

8.14

ii)Y=0.0007x+

0.0115

10–750

0.9945

0.6

0.2

250

263

−5.23

6.14

9.04

5.35

3.84

1.27

4.63

7.17

iii)Y=0.0014

x−0.0249

10–750

0.9895

0.5

0.15

250

204

18.18

7.15

65.16

3.33

12.58

25.07

2.76

39.06

M)p-Coumaric

acid

i)Y=0.001x

−0.013

10–750

0.9941

20.7

250

216

13.48

5.25

8.07

0.91

4.34

9.48

9.02

7.54

ii)Y=0.005x

−0.0035

10–750

0.9978

10.3

250

212

15.08

5.02

4.08

4.02

3.71

1.19

5.90

9.30

iii)Y=0.0008

x=0.0133

2–750

0.9942

10.3

250

211

15.25

9.36

n/f

8.30

9.12

12.66

6.00

22.68

O)Azelaic

acid

i)Y=0.0025

x+0.0211

10–750

0.9998

0.001

0.0003

250

265

−6.22

2.44

7.32

3.92

5.15

8.21

9.31

6.484

ii)Y=0.0024x+

0.0419

10–750

0.9985

0.002

0.0006

250

262

−5.02

6.72

4.11

2.97

1.63

1.21

4.34

.94

iii)Y=0.0024

x+0.0243

10–750

0.9995

0.05

0.015

250

245

1.89

3.90

67.57

17.19

5.81

6.30

1.03

29.33

P)Phenyllactic

acid

i)Y=0.0033

x+0.0256

10–750

0.9997

0.1

0.03

250

288

−15.26

1.45

9.83

1.82

7.39

8.58

5.59

4.05

ii)Y=0.0031x+

0.0742

10–750

0.9936

0.2

0.06

250

274

−9.80

4.75

7.78

3.21

5.11

4.40

6.48

4.29

iii)Y=0.0031

x+0.0303

25–750

0.9976

0.2

0.06

250

287

−15.16

1.45

n/d

76.39

12.27

4.95

3.08

22.63

Q)Benzoic

acid

i)Y=0.009x

+0.0216

10–750

0.998

0.09

0.03

250

263

−5.25

3.54

5.05

9.37

1.45

9.88

8.96

4.24

ii)Y=0.0008x+

0.0355

10–750

0.9943

0.4

0.1

250

272

−9.15

6.91

1.43

6.42

2.10

2.15

6.40

3.79

iii)Y=0.0008

x+0.0024

10–750

0.9992

0.6

0.2

250

275

−10.26

4.43

86.77

22.10

15.62

20.51

3.54

17.47

R)Hydrocinnam

icacid

i)Y=0.0007

x+0.0022

10–750

0.9999

0.4

0.1

250

249

0.10

2.77

4.09

1.98

3.99

3.20

7.00

4.96

ii)Y=0.0006x+

0.0032

10–750

0.9969

0.4

0.1

250

288

−15.56

1.80

3.31

4.22

9.33

2.66

5.88

7.90

iii)Y=0.0007

x−0.0014

10–750

0.9995

0.4

0.1

250

237

5.10

2.63

10.99

9.97

7.23

6.50

3.99

18.49

T)3-

Hydroxydecanoic

acid

i)Y=0.0034

x+0.0902

10–750

0.9988

0.05

0.015

250

265

−6.27

3.18

7.79

1.24

2.41

9.15

5.93

3.65

ii)Y=0.0027x+

0.1209

10–750

0.9941

0.05

0.015

250

280

−12.16

3.25

8.86

2.16

5.14

2.60

4.13

5.70

iii)Y=0.0027

x+0.1148

10–750

0.9934

0.8

0.25

250

300

−20.00

3.74

15.17

8.01

5.68

10.69

4.05

20.04

U)DL-β-H

ydroxylauric

acid

i)Y=0.0033

x−0.0009

10–750

0.9927

0.05

0.015

250

242

3.10

7.21

8.20

5.99

8.15

6.97

6.68

3.59

ii)Y=0.0022x+

0.0699

10–750

0.9959

0.2

0.06

250

273

9.43

3.40

4.38

5.66

2.43

4.55

7.29

8.42

iii)Y=0.001x

+0.1102

10–750

0.9435

0.5

0.015

250

579

−131.9

4.45

8.14

4.15

4.68

6.34

0.96

23.64

LC-FTMS of LAB antifungal compounds 2989

of a run, during the last 1–5 min of a run, and during columnequilibration periods) is a common solution to keep H-ESIsources clean of mobile phase salts and sample matrix con-taminants [17–21]. However, during such diversion periodsthe heated capillary (or ion transfer tube) has to be maintainedat 300 °C and without mobile phase flowmobile phase buffersand/or sample matrix contaminants could solidify and hardenwithin the ion transfer tube, thus blocking the tube. When thishappens chromatographic sequences have to be stopped andresumed only after the ion-transfer tube has been unblockedand/or cleaned. It was noted that if the divert feature was notactivated and the mobile phase was allowed to flow continu-ously into the ESI source throughout the chromatographicruns and during column equilibration periods between injec-tions, the ion-transfer tube would “self-clean” and neverblock, even during analysis of crude filtered broth samples.

Validation

The method was validated for standards prepared in mobilephase and MRS broth. Both intraday and interday data wereassessed as part of the validation. Calibration curves weregenerated and excellent linearity was observed. Coefficient ofvariation (R2) values of 0.9909–0.9999, 0.9922–0.9996, and0.9435–0.9995 were obtained for standards spiked into mobilephase solvent, into crude broth after LLE, and into crude brothbefore filtration, respectively. A control sample was used tofurther determine the accuracy of the method (Table 2,Electronic Supplementary Material Table S2a and b). Thecontrol was run three times (n03) and the RSD value for eachmatrix was calculated. For the mobile phase and LLE stand-ards RSD values were <15%, whereas significant varianceoccurred for the broth samples (Table 2(i), (ii), (iii),Electronic Supplementary Material Table S2a(i), (ii), (iii) andb(i), (ii), (iii)). To gauge the precision of the method, thecalibration set was repeated a further two times and the overallvalues were compared. The interday RSD value was deter-mined andwas ≤15% for themobile phase and broth LLE. The“self-clean” action used (as mentioned previously) to prevention transfer tube blockage by the broth samples enabled thesamples to be analysed continuously. However, large variancewas observed for the RSDs, leading to poor intraday results(Table 2, Electronic Supplementary Material Table S2a and b).A thoroughly validated method was developed for the antifun-gal standards spiked into mobile phase matrix and crude brothpost-LLE processing. However, because of the complexity ofthe crude broth filtered sample matrix, the resulting validationdid not conform to the necessary criteria required for completevalidation (Table 2(iii), Electronic Supplementary MaterialTable S2a(iii) and b(iii)). RSD values observed ranged from1.03–173.2% with approximately a quarter of the RSD values>15%.T

able

2(con

tinued)

Com

pound

Equationof

thelin

eLinearrange

(mgL−1)

R2

LOD

(mgL−1)

LOQ

(mgL−1)

Control

theoretical

concentration

(mgL−1)

Control

measured

concentration

(mgL−1)

RE(%

)RSD

(%,n=3)

Intraday

RSD

%(n=3)

10mgL−1

25mgL−1

50mgL−1

100mgL−1

500mgL−1

750mgL−1

V)Decanoicacid

i)Y=0.000008

x+0.0005

10–750

0.9942

1.5

0.5

250

246

1.33

9.52

3.01

9.05

9.21

6.15

4.41

4.93

ii)Y=0.0000006x

+0.0004

10–750

0.9947

10.3

250

243

2.58

2.74

3.29

1.66

2.35

6.20

3.77

6.19

iii)Y=0.000008

x−0.0002

100–750

0.9918

10.3

250

178

28.7

0.85

n/d

n/d

12.49

24.38

13.46

22.76

X)2-

Hydroxydodecanoic

acid

i)Y=0.0019

x+0.055

10–750

0.9947

0.02

0.006

250

213

13.42

2.50

7.69

5.17

2.33

9.12

10.94

2.40

ii)Y=0.002x

+0.1444

10–750

0.9951

0.1

0.03

250

268

4.51

6.22

1.26

9.52

6.40

8.24

6.87

10.26

iii)Y=0.013x

+0.1221

10–750

0.9645

10.3

250

476

−82.22

3.55

6.94

5.30

4.75

5.27

2.89

15.87

i)Stand

ards

spiked

into

mob

ileph

ase(90:10

water–acetonitrile)

ii)Stand

ards

spiked

into

blankMRSbrothextractafterLLE

iii)Stand

ards

spiked

into

blankcrud

eMRSbrothafterfiltration

n/d,

notdetected

2990 B. Brosnan et al.

Determination of the antifungal compounds presentin antifungal broth

Strains (Lactobacillus amylovorus FST2.1, Lactobacillusarizononas R13, Lactobacillus plantarum FST1.7,Lactobacillus reuteri R2, and Weisella cibaria PS2) withstrong antifungal activity were evaluated to determine thespecific antifungal compounds produced. The LLE methodwas used to prepare the strains. The antifungal cultures anda control MRS broth were extracted with ethyl acetate.Average recoveries for the twenty five target compoundswas 50%, further studies (unpublished data) are being car-ried out to improve this figure.

To confirm the presence of a compound, its retention timewas compared with its known standard retention time, within±0.3 min (Figs. 2 and 3) and the exact mass was determined tofive decimal places thus providing more information about theion present. Sample spectra of identified compounds areshown in Figs. 4 and 5a, b. From the masses obtained forthe identified compounds, a ppm error was calculated. Table 1shows the masses acquired for both the standards and theLactobacillus amylovorus FST2.1 sample. Low ppm errorvalues of less than 2 ppm were obtained, with most <1 ppm,definitively confirming the compound’s identity.

Antifungal compounds vanillic acid (compound I),(S)-(−)-2-hydroxyisocapric acid (compound J), azelaicacid (compound O), 4-hydroxybenzoic acid (compoundG), decanoic acid (compound V), benzoic acid (com-pound Q), hydrocinnamic acid (compound R), ρ-coumaric acid (compound M), DL-ρ-hydroxyphenyllactic

acid (compound D), phenyllactic acid (compound P), 2-hydroxydodecanoic acid (compound X), DL-β-hydroxylauricacid (compound U), 3-hydroxydecanoic acid (compound T),and salicylic acid (compound Y) were identified in the sample.The concentrations of these compounds in the samples variedimmensely. To quantify the target compounds over such largeconcentration ranges, three sample preparations were analysed:

A. A 10 mL sample was reconstituted in 1 mL of mobilephase giving a 10-fold concentration factor (this wasdone to provide a more intense signal for compoundsthat may be present in lower quantities);

B. A 1:10 dilution (now equivalent to direct injection ofthe crude broth), and

C. A 1:20 dilution of the concentrated sample (A) to obtainpeak areas within the calibration range for compoundspresent in higher quantities.

Compounds found in the “A” concentration range (with theconcentration detected in the broth) were: vanillic acid(4.3 mg L−1), azelaic acid (1.2 mg L−1), 4-hydroxybenzoicacid (1.5 mg L−1), decanoic acid (3.2 mg L−1), benzoic acid(1.1 mg L−1), hydrocinnamic acid (4.5 mg L−1), ρ-coumaricacid (14.0 mg L−1), DL-β-hydroxylauric acid (43.4 mg L−1),and salicylic acid (9.9 mg L−1). Compounds found in the“B” concentration range were: DL-ρ-hydroxyphenyllactic acid(228.1 mg L−1), 2-hydroxydodecanoic acid (433.9 mg L−1),and 3-hydroxydecanoic acid (170.5 mg L−1). The most abun-dant compounds in the sample, requiring 1:20 dilution (con-centration range “C”) for correct quantitative detection, were(S)-(−)-2-hydroxyisocapric acid and phenyllactic acid.

Fig. 3 Chromatogram obtained from the antifungal broth sample(Lactobacillus amylovorus FST2.1) separated on a Gemini C18 col-umn as outlined in the section “LC instrumentation”. Compoundsidentified included: (i) D, DL-ρ-hydroxyphenyllactic acid; M, p-cou-maric acid; O, azelaic acid; Q, benzoic acid; R, hydrocinnamic acid; T,3-hydroxydecanoic acid; U, DL-β-hydroxylauric acid; X, 2-

hydroxydodecanoic acid; (ii) I, vanillic acid; J, (S)-(−)-2-hydroxyiso-caproic acid; (iii) G, 4-hydroxybenzoic acid; P, phenyllactic acid; Y,salicylic acid; (iv) no compounds detected in this m/z region; (v) V,decanoic acid; vi) no compounds detected in this m/z region. The deuter-ated internal standard (hydrocinnamic acid d9 OH, 200 µg mL–1) co-elutedwith (i) R, hydrocinnamic acid at 8.42 min and was therefore not included

LC-FTMS of LAB antifungal compounds 2991

Concentrations of these compounds in the broth were 1,057.2and 1,333.0 mg L−1, respectively.

Electronic Supplementary Material Table S1 outlines theantifungal compounds, their theoretical masses [M−H]−, con-centrations (mg L−1), detected masses [M−H]−, and calculatedppm errors for four antifungal strains: Lactobacillus arizono-nas R13, Lactobacillus plantarum FST1.7, Lactobacillus reu-teri R2, and Weisella cibaria PS2. Compounds identified forall the strains tested were: vanillic acid (compound I), (S)-(−)-2-hydroxyisocapric acid (compound J), azelaic acid (com-pound O), 4-hydroxybenzoic acid (compound G), decanoicacid (compound V), benzoic acid (compound Q), hydrocin-namic acid (compound R), DL-ρ-hydroxyphenyllactic acid(compound D), phenyllactic acid (compound P), DL-β-hydroxylauric acid (compound U), 3-(4-hydroxy-3-methoxy-phenyl)propanoic acid (compound L), 3-hydroxydecanoic ac-id (compound T), and 1,2-dihydroxybenzene (compound E).Other compounds identified were DL-β-hydromyristic acid(compound W), found in Lactobacillus arizononas R13 andLactobacillus reuteri R2 broth, and salicylic acid (compoundY), only found in Lactobacillus arizononas R13 broth. Aswith the Lactobacillus amylovorus FST2.1 strain (Table 1), theconcentration ranges of the compounds detected were diverseand for their accurate quantitation (Electronic Supplementary

Material Table S1), the concentration dilution steps A, B, and Chad to be used.

Synergy studies of the LAB compounds have shown thatthe collective activity of the known compounds is muchhigher than their individual activity [4, 11]. Niku-Paavola[4] demonstrated this by combining different mixtures of thecompounds identified. Antagonistic effects and synergisticeffects were observed by use of mixtures. The methoddescribed herein should form a solid basis from which moreinformation on the collective and individual bioactivities ofthe LAB constituents can be investigated.

Application of method

The use of LAB as natural food and feed preservatives isbecoming more prominent. With use of appropriate sampleclean-up procedures this method can be used to test food and

Fig. 4 Spectra of the main compounds identified in Lactobacillusamylovorus FST2.1 broth. (i) D, DL-ρ-hydroxyphenyllactic acid; (ii)J, (S)-(−)-2-hydroxyisocaproic acid; (iii) P, phenyllactic acid; (iv) T, 3-

hydroxydecanoic acid; (v) U, DL-β-hydroxylauric acid; and (vi) X, 2-hydroxydodecanoic acid

Fig. 5 Magnified chromatograms and corresponding spectra of com-pounds present in low quantities (1.1–45 mg mL–1) in the sample broth(Lactobacillus amylovorus FST2.1). (a) (i and iv) M, p-coumaric acid;(ii and v) Q, benzoic acid; (iii and vi) I, vanillic acid. (b) (i and iv) G, 4-hydroxybenzoic acid; (i and v) Y, salicylic acid; (ii and vi) O, azelaicacid; (iii and vii) R, hydrocinnamic acid

b

2992 B. Brosnan et al.

166.5 167.0 167.5m/z

0

100

Rel

ativ

e Abu

ndan

ce

167.03432

121.0m/z

0

100

Rel

ativ

e Abu

ndan

ce121.02904

162.0 162.5 163.0 163.5m/z

0

100

Rel

ativ

e Abu

ndan

ce

163.03944

6 7 8 9Time (min)

0

100

Rel

ativ

e Abu

ndan

ce

7.5 8.0 8.5Time (min)

0

100

Rel

ativ

e Abu

ndan

ce

7.0 7.5Time (min)

0

100R

elat

ive A

bund

ance

(M) (Q) (I)

(M) (Q) (I)

(i) (ii) (iii)

(iv) (v) (vi)

a

b

6 7 8 9Time (min)

0

100

Rel

ativ

e A

bund

ance

(O)

5 10Time (min)

0

100

Rel

ativ

e A

bund

ance

(G)

(Y)

138.02765

100

Rel

ativ

e A

bund

ance

137.02410

136.04015

135.18024

135 136 137 138m/z

0

(G)

135 136 137 138 139m/z

100

Rel

ativ

e A

bund

ance

137.02386

138.02723136.03923

(Y)

7 8 9Time (min)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e A

bund

ance

(R)

186 187 188 189m/z

100

Rel

ativ

e A

bund

ance

187.09709

186.05550

188.10037189.05681

(O)

148.5 149.0 149.5m/z

0

100

Rel

ativ

e A

bund

ance

149.06017

149.67688148.73941

(R)

(i) (ii) (iii)

(iv)

(v)

(vi)

(vii)

LC-FTMS of LAB antifungal compounds 2993

feed samples and identify compounds produced by LAB strains.Indeed, this method should enable discovery of new com-pounds, variants, and metabolites. Ions with a mass of 500 Daor less requires accuracy of 0.0025 Da to enable unambiguousassignment of elemental composition [24]. For the target com-pounds in this study (e.g. 1,2-dihydroxybenzene [M−H]− ofm/z109.2895 (lowest mass) and DL-β-hydromyristic acid [M−H]−

of m/z 243.18602 (highest mass)) this corresponds to errors of23 and 10 ppm, respectively. As can be seen in Table 1 andElectronic SupplementaryMaterial Table S1, errors are less then3 ppm for all the target compounds, providing results withaccuracy of within 0.00032 Da (1,2-dihydroxybenzene) and0.00073 Da (DL-β-hydromyristic acid). These results definitive-ly confirm the presence of these analytes in the broth sample.

Conclusion

A rapid method has been developed that enables identificationof a wide range of antimicrobial compounds produced by LAB.High mass accuracy data were used to determine with confi-dence the exact chemical formulae of the compounds present.Obtaining ppm values below 3 ppm ensured definite chemicalformulae could be deduced. Use of Orbitrap MS technologycan provide unambiguous information enabling chemical char-acterisation. The method detailed herein is more comprehen-sive than those previously used, both in the number ofcompounds identified (n=25) and the use of one robust methodfor detection rather than the numerous methods formerly re-quired to detect such a vast number of compounds. It is also areliable method for identification of unknown and biotransfor-mation compounds that are produced by these bacteria.

Acknowledgements We gratefully acknowledge funding from theFood Institutional Research Measure (FIRM) Department of Agricul-ture, Fisheries and Food Ireland (project reference 08RDC607). TheCouncil of Directors, Technological Sector Research-Strand III 2006Grant Scheme, awarded to Dr A. Furey is also acknowledged forfunding the formation of the “Team Elucidate” research group. TheHigher Education Authority (Programme for Research in Third-LevelInstitutions, Cycle 4 (PRTLI IV) National Collaboration Programme onEnvironment and Climate Changes: Impacts and Responses is also ac-knowledged. We also thank Dr Mary Lehane and Dr Sharon Hutchinsonfor their critical reading of the manuscript.

References

1. Carr FJ, Chill D, Maida N (2002) The lactic acid bacteria: aliterature survey. Crit Rev Microbiol 28:281–370

2. Gobetti M, De Angelis M, Corsetti A, Di Cagno R (2005) Bio-chemistry and physiology of sourdough lactic acid bacteria. TrendsFood Sci Tech 16:57–69

3. Broberg A, Jacobsoson K, Strom K, Schnürer J (2007) Metaboliteprofiles of lactic acid bacteria in Grass Silage. Appl EnvironMicrobiol 73:5547–5552

4. Niku-Paavola ML, Laitila A, Mattila-Sandholm T, Haikara A(1998) New types of antimicrobial compounds produced byLactobacillus plantarum. J Appl Microbiol 86:29–35

5. Schnürer J, Magnusson J (2005) Antifungal lactic acid bacteria asbiopreservatives. Trends Food Sci Tech 16:70–78

6. Corsetti A, Gobbetti M, Rossi J, Damiani P (1998) Antimouldactivity of sourdough lactic acid bacteria identification of a mixtureof organic acids produced by Lactobacillus sanfrancisco CB1.Appl Microbiol Biotechnol 50:255–256

7. Valerio F, Lavermicocca P, Pascale M, Visconti A (2003) Produc-tion of phenyllactic acid by lactic acid bacteria an approach to theselection of strains contributing to food quality and preservation.FEMS Microbiol Lett 233:289–295

8. Ryan LAM, Dal Bello F, Arendt EK (2008) The use ofsourdough fermented by antifungal LAB to reduce theamount of calcium propionate in bread. I J Food Microbiol152:274–278

9. Rizzello CG, CassanaA CR, Gobbetti M (2011) Antifungal activityof sourdough fermented wheat germ used as an ingredient forbread making. Food Chem 127:952–959

10. Sathe SJ, Nawani NN, Dhakephalkar PK, Kapadnis BP(2007) Antifungal lactic acid bacteria with potential to pro-long shelf-life of fresh vegetables. J Appl Microbiol103:2622–2628

11. Ndagano D, Lamoureox T, Dortu C, Vandermoten S, Thonart P(2011) Antifungal activity of 2 lactic acid bacteria of the Weissellagenus isolated from food. J Food Sci. doi:10.1111/j.1750-3841.2011.02257.x

12. Ström K, Sjögren J, Broberg A, Schnürer J (2002) Lactoba-cillus plantarum MiLAB 393 produces the antifungal CyclicDipeptides Cyclo(L-Phe –L-Pro) and Cyclo(L-Phe_trans-4-OH-L-Pro) and 3-phenyllactic acid. Appl Environ Microbiol68:4322–4327

13. Sjögren J, Magnusson J, Broberg A, Schnürer J, Kenne L (2003)Antifungal 3-hydroxy fatty from Lactobacillus plantarum MiLAB14. Appl Environ Microbial 69:7554–7557

14. Lind H, Sj gren J, Gohil S, Kenne L, Schnürer J, Broberg A (2007)Antifungal compounds from cultures of dairy propionibacteriatype strains. FEMS Microbiol Lett 271:310–315

15. Armaforte E, Carri S, Caboni GF (2006) High-performance liquidchromatography determination of phenyllactic acid in MRS broth.J Chromatogra A 1131:281–284

16. Ryan LAM, Zannini E, Dal Bello F, Pawlowska A, Koehler P,Arendt EK (2011) Lactobacillus amylovorus DSM19280 as anovel food-grade antifungal agent for bakery products. I J FoodMicrobiol 146:276–283

17. Furey A, Lehane M, Gillman M, Puente-Fernandez P, James KJ(2001) n Determination of Domoic acid in shellfish using liquidchromatography electrospray ionisation ion-trap mass spectrome-try. J Chromatogr A 938:167–174

18. Furey A, Braña-Magdalena A, Lehane M, Moroney C, JamesKJ, Satake M, Yasumoto T (2002) Determination of Azaspir-acids in shellfish using liquid chromatography-tandem electro-spray mass spectrometry. Rapid Commun Mass Spectrom16:238–242

19. Furey A, Crowley J, Lehane M, James KJ (2002) Liquid chroma-tography with Electrospray Ion-trap mass spectrometry for thedetermination of Anatoxins in Cyanobacteria and drinking water.Rapid Commun Mass Spectrom 17:583–588

20. Allis O, Dauphard J, Hamilton B, Ni Shuilleabhain A, Lehane M,James KJ, Furey A (2007) Liquid chromatography-tandem massspectrometry application, for the determination of extracellularhepatotoxins in Irish lake and drinking waters. Anal Chem79:3436–3447

21. López-Rivera A, O’Callaghan K, Moriarty M, O’Driscoll D,Hamilton B, Lehane M, James KJ, Furey A (2010) First

2994 B. Brosnan et al.

Evidence of Azaspiracids (AZAs): a family of Lipophilicpolyether marine toxins in scallops (Argopecten purpuratus)and mussels (Mytilus chilensis) collected in two regions ofchile. Toxicon 55:692–701

22. European Parliament and Council (1995) Directive No. 95/2/EC of20 February 1995 on food additives other than colours andsweeteners

23. Wessels S, Axelsson L, Hansen EB, De Vuyst L, Laulund S,Lähteenmäki LS, Mollet B, Salminen S, Von Wright A (2004)The lactic acid bacteria, the food chain, and their regulation.Trends Food Sci Technol 15:498–505

24. Watson JT, Sparkman OD (2008) Introduction to mass spectrometry:Instrumentation, applications and strategies for data interpretation.4th Edn. Wiley

LC-FTMS of LAB antifungal compounds 2995