G

C

Vme

AEa

b

a

ARRAA

KLELH��

1

oaliCc

(aca

trG

0h

ARTICLE IN PRESS Model

HROMA-355085; No. of Pages 6

Journal of Chromatography A, xxx (2014) xxx– xxx

Contents lists available at ScienceDirect

Journal of Chromatography A

jou rn al hom epage: www.elsev ier .com/ locate /chroma

alidation of a high performance liquid chromatography-tandemass spectrometry method for �-hydroxy fatty acids as

nvironmental markers of lipopolysaccharide

lessandra Chiomintoa, Anna Maria Marcellonia, Giovanna Tranfob,∗,milia Pabaa, Enrico Pacib

INAIL Research, Department of Occupational Hygiene, via Fontana Candida, 1-00040 Monteporzio Catone (RM), ItalyINAIL Research, Department of Occupational Medicine, via Fontana Candida, 1-00040 Monteporzio Catone (RM), Italy

r t i c l e i n f o

rticle history:eceived 25 September 2013eceived in revised form 15 January 2014ccepted 17 January 2014vailable online xxx

eywords:

a b s t r a c t

The most commonly analytical method used for the assessment of the occupational exposure to airborneendotoxins is the LAL test, but other methods are reported in the literature.

The objective of this work was to develop and validate a method using HPLC coupled to triplequadrupole mass spectrometer for the quantitative determination of �-hydroxy-dodecanoic and �-hydroxy-tetradecanoic acids as chemical markers of lipopolysaccharide in Gram-negative strains. Matrixeffect was significant but compensated by the use of internal standard; accuracy was always more

ipopolysaccharidendotoxinAL testPLC-MS/MS-Hydroxy-dodecanoic acid-Hydroxy-tetradecanoic acid

than 85% and variability always less than 10%. LOD and LLOQ were respectively 2.5 �g/l and 10 �g/lfor �-hydroxy-dodecanoic acid, 1 �g/l and 5 �g/l for �-hydroxy-tetradecanoic acid.

The analysis of different bacterial strains showed that samples contain �-hydroxy-dodecanoic and�-hydroxy-tetradecanoic acids in variable ratio, and therefore the sum of their concentrations was con-sidered the chemical marker. The method showed a significant correlation of results with those of LAL-teston four bacterial strains.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

Endotoxins, also known as pyrogen or fever-causing toxin, areuter membrane components of Gram-negative bacteria and playn important role in occupational lung disease. They are made up ofipopolysaccharide (LPS) and composed by three regions: a glycol-pidic part, named Lipid A, a polysaccharide core, and an O-Specifichain, consisting of 20–40 repeating units, which generally mayontain up to eight different sugars (Fig. 1).

Lipid A consists of a phosphorylated N-acetyl glucosamineNAG) dimer with six � hydroxy fatty acids (FA) all saturated; fourre attached directly to the NAG dimer; while two additional acylhains are often attached to the beta hydroxy group. Beta hydroxycyl chains are usually between 10 and 16 carbons in length.

The structure of the lipid A portion is fairly well conserved, buthe nature (length and chemical composition) of the polysaccha-

Please cite this article in press as: A. Chiominto, et al., J. Chromatogr. A (20

ide side chain varies between genera, species, and even strains ofram-negative bacteria [1,2].

∗ Corresponding author. Tel.: +39 06 94181436; fax: +39 06 94181410.E-mail address: g.tranfo@inail.it (G. Tranfo).

021-9673/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.chroma.2014.01.052

The endotoxic activity of LPS resides in the lipid A domain,since polysaccharide-deprived free lipid A appears to exhibit simi-lar endotoxic activities as intact LPS [3,4]. It is a powerful biologicalresponse modifier that can stimulate the mammalian immune sys-tem; since lipid A is embedded in the outer membrane of bacterialcells, it probably exerts toxic effects only when released from mul-tiplying cells in a soluble form, or when the bacteria are lysed [5].

Several authors reported that chemical differences in the struc-ture of Lipid A are reflected by biological differences and differenttoxic effects [6,7].

The problem of assessing the occupational exposure to airborneendotoxins has been studied for years, but sampling and analysisprocedures are still not completely standardized and shared [8–10].The most widely used analytical method is the Limulus AmebocyteLysate (LAL) test, in its kinetic chromogenic version. It is a functionalassay based on enzymatic coagulation of the blood of a primitivemarine arthropod, the horseshoe crab (Limulus polyphemus) in thepresence of a minimal amount of endotoxin, due to the propertiesof the biologically active constituent LPS. Some critical aspects of

14), http://dx.doi.org/10.1016/j.chroma.2014.01.052

LAL test are reported in literature: the method is sensitive to thepresence of �-(1→3)-D-glucans, structural components of the wallof molds and yeasts, which seems to interfere in the enzymatic cas-cade of the LAL test, leading to amplification of the response, and

ARTICLE ING Model

CHROMA-355085; No. of Pages 6

2 A. Chiominto et al. / J. Chromato

tsswtiteacS

c�LtpLom

nmtaT

rba

wsr

2

2

A1ES

n

2.4. Sample preparation for HPLC-MS/MS

Fig. 1. Schematic architecture of LPS.

herefore to an overestimation of the results [11,12]; on the otheride, endotoxin exposure may be underestimated in environmentalamples by the LAL test because a substantial fraction can be non-ater soluble and not detected [13]. Rylander et al. [14] suggested

hat the Limulus method detects only about a third of the biolog-cally active endotoxin and that the remainder is present insidehe fragments of dust particles/bacterial cells, but still able to exertffects when deposited in the lung. Also the stability of the samplesfter repeated freezing/thawing may affect detectable endotoxinontent as reported by the guidelines of European Committee fortandardization [15].

A different approach to measure endotoxin is represented byhemical analysis, which is based on the release and detection of the-hydroxy fatty acids, used therefore as chemical markers of LPS.PS actually describes the protein-free form of endotoxin, but theseerms are frequently used synonymously [16]. Gas Chromatogra-hy (GC) coupled to Mass Spectrometry (MS) or High Performanceiquid chromatography (HPLC) coupled to a fluorescence detectorr UV/VIS spectrophotometry are the analytical methods com-only used for this purpose [17–21].The main objective of this work was to develop and validate a

ew analytical method using HPLC coupled to triple quadrupoleass spectrometer (HPLC-MS/MS) for the quantitative determina-

ion of �-hydroxy-dodecanoic and �-hydroxy-tetradecanoic acidss LPS chemical markers in Gram-negative strains from Americanype Culture Collection (ATCC).

In order to verify the applicability of this method to the envi-onmental monitoring of endotoxins, we evaluated the correlationetween the results of HPLC-MS/MS analysis and those of the LALssay on the same bacterial strains.

The method has been applied also on ATCC strains of fungi,hich are known to contain �-glucans, in order to estimate the pos-

ible interference; the stability of the samples analyzed followingepeated freezing and thawing cycles was also tested.

. Material and methods

.1. Bacterial and fungal strains

Four ATCC bacterial Gram-negative strains were used:eromonas hydrophila-35654 and Pseudomonas aeruginosa-5442 were purchased from D.I.D. S.p.A; Escherichia coli-8739 andnterobacter aerogenes-35029 were purchased from Biogenetics.r.l. and from Remel Europe Ltd. respectively.

Please cite this article in press as: A. Chiominto, et al., J. Chromatogr. A (20

Two fungal strains, Candida albicans-10231 and Aspergillusiger-9642, were provided by Oxoid LTD and Remel Europe Ltd.

PRESSgr. A xxx (2014) xxx– xxx

2.2. Analytical standards

Standard solutions of �-hydroxy-dodecanoic and �-hydroxy-tetradecanoic acids, along with �-hydroxy-tridecanoic acid used asinternal standard (Larodan Fine Chemicals, Malmo Sweden), wereprepared by dissolving in acetonitrile suitable amount of pure stan-dards, which were stored at 4 ◦C; from this solutions, by furtherdilutions, the calibration and quality control (QC) samples wereprepared.

2.3. HPLC-MS/MS analytical method

The HPLC device was a Perkin Elmer series 200, and a SupelcoHypersyl BDS C8 (150 × 4.6, 5 �m) chromatographic column, main-tained at 40 ◦C, was used, with mobile phase consisting of a gradientof acetonitrile (A) and acetic acid 1% (v/v) aqueous solution (B) asfollows: 1 min equilibration with 50% A and 50% B, then 9 min lin-ear gradient up to 100% of A; finally 1 min linear gradient up to theinitial conditions, for a total time of analysis of 10 min. The flowrate was maintained at 1000 �l/min.

The mass spectrometer was a AB/MDS Sciex API 4000, triplequadrupole mass spectrometer, equipped with a TurboIonSpray®Ion Source to create the ions in the droplets projected at a 90◦ angleto the orifice; the nebulizer gas is produced by a pair of high effi-ciency ceramic thermal injectors, placed at a 45◦ angle to the spray.Polypropylene glycol was used to calibrate the instrument workingin the negative ion, multiple reaction monitoring (MRM) mode. Thenegative-ion ESI MS/MS product-ion spectra of each analyte wereobtained by the automated “Infusion Quantitative Optimization”procedure, that identifies the mass of most intense peak detectedby the Q1 and then the mass of the most intense product ionin the Q3.:the optimum operating conditions of the electrosprayionization (ESI) were as follows: nebulizing gas (GS1) 50, turbogas (GS2) 15, curtain gas (CUR) 17, collision activated dissociationgas (CAD) 8, turbo-spray voltage −4500 V and turbo temperature600 ◦C. The mass spectrometer was operated in unit resolution forboth Q1 and Q3 in MRM mode, with a dwell time of 370 ms perMRM channel with a 5 ms pause between scans. Subsequently thescan parameters are refined by flow injection analysis (FIA), usingthe pure standards and the potentials optimized for each transitionare determined. For all analytes is declustering potential (DP) −55,entrance potential (EP) −10, collision energy (CE) −22, while colli-sion cell exit potential (CXP) is −7 for �-hydroxy-dodecanoic acidand −9 for �-hydroxy-tetradecanoic and �-hydroxy-tridecanoicacids. The m/z ion transitions (precursor→product), correspondingto the loss of a fragment of 46 Da typical of �-hydroxy fatty acidswere monitored for both the qualitative and the quantitative analy-sis. The m/z ion transitions (precursor→product), corresponding tothe loss of a fragment of 46 Da typical of �-hydroxy fatty acids weremonitored for the quantitative analysis, and, to further improve thespecificity, we also identified other transitions, corresponding tothe second highest peak in the product ion spectra of pure stan-dards. However the abundances were very low, being the loss of46 Da the main fragmentation step. Retention times and the differ-ence between the analytes and the internal standard ones (0.7 minfor both analytes) were reproducible during all the validation andanalysis experiments.

Transitions and retention times are reported in Table 1.The version of 1.4 Analyst† software was employed for instru-

ment control, using the section ‘Quant Method Editor’ to processquantitative data.

14), http://dx.doi.org/10.1016/j.chroma.2014.01.052

For HPLC-MS/MS analysis 2 ml of bacterial suspension inPyrogen-Free Water (PFW) (Lonza Sales Ltd, Switzerland) were

ARTICLE IN PRESSG Model

CHROMA-355085; No. of Pages 6

A. Chiominto et al. / J. Chromatogr. A xxx (2014) xxx– xxx 3

Table 1Transitions and retention times.

Analyte MW 1st ionic transition m/z 2nd ionic transition m/z Retention time

si2fbM

A

swGi1s

2

A(sb

0W(E

aotCpr

a

3-Hydroxy-dodecanoic acid 216 −215/169

3-Hydroxy-tetradecanoic acid 244 −243/1973-Hydroxy-tridecanoic acid 229 228/182

ubjected to hydrolysis by addition of 1 ml of 6 M sodium hydrox-de (Sigma-Aldrich, Milan) solution in water, in order to obtain a

M concentration in the sample, and heated at 90 ◦C under stirringor 40 min. After the hydrolysis, the suspension was neutralizedy addition of 6 M hydrochloric acid (Carlo Erba Reagenti SpA,ilan).After cooling, 40 �l of a 10 mg/l solution in acetonitrile (Sigma-

ldrich, Milan) of 3-hydroxy-tridecanoic acid were added.Hydrolyzed fatty acids were extracted with n-hexane, in three

ubsequent steps, each one with 1 ml; the 3 ml n-hexane extractsere finally dried in evaporator (MIVAC Quattro concentrator,enevac, UK) for 30 min at 60◦ C. The residue was redissolved

n 0.5 ml of acetonitrile, and after filtration with 0.2 �m Anotop0 IC membrane filters, was injected into the HPLC-MS/MSystem.

.5. Kinetic chromogenic LAL test

ATCC bacterial and fungal strains were isolated on Tryptone Soyagar (TSA, Oxoid-Thermo Fischer Scientific) and Malt Extract Agar

MEA, Oxoid-Thermo Fischer Scientific) respectively. Cell suspen-ions were prepared with PFW and subjected to mechanical lysis,y vortexing for 5 min with apyrogenic glass spheres.

A volume of 100 �l of serial 10-fold dilutions (1, 0.1, 0.01, 0.001,.0001, 0.00001) were analyzed by Kinetic QCL-LAL assay (Lonzaalkersville, MD USA) and were interpreted against a 5-point

concentration range from 0.005 to 50 EU/ml) standard curves ofscherichia coli (CSE).

All procedures were carried out using pyrogen-free reagents andll glassware was baked at 180 ◦C for 4 h to remove endotoxin. Tobtain information about possible enhancement or inhibition reac-ions of the LAL assay, a replicate of each sample was spiked with aSE standard (5 EU/ml final activity). The recovery of spiked sam-

Please cite this article in press as: A. Chiominto, et al., J. Chromatogr. A (20

les was in the range of 50–200%; otherwise the measurement wasepeated. Results are reported as EU/ml.

The same samples were subjected to the HPLC-MS/MS analysisccording to the procedure described above.

Fig. 2. LC-MS/MS chromatographic peaks obtained a

−215/121 5.05 min−243/174 6.45 min– 5.75 min

3. Results and discussion

3.1. Validation of the HPLC-MS/MS method

The validation procedure was carried out according to ICH [22].Three calibration curves for each analyte (�-hydroxy-dodecanoicand �-hydroxy-tetradecanoic acid) were generated in the range10–1000 �g/l using linear regression analysis according to theequation y = ax + b, where x is the ratio of the analyte peak area tothat of the internal standard, and y is the concentration for each cal-ibration standard. Mean a and b were calculated for each analyte.In order to determine the limits of detection and quantification(LOD and LOQ) we applied the approach based on the standarddeviation according to ICH. The analytical background was mea-sured by analyzing ten blank samples, considering the areas at theretention times of the two analytes of interest. The response is theratio between the background and internal standard areas; threeand ten times the standard deviation and the regression were usedto calculate the LOD and LOQ, which were respectively 2.5 �g/land 10 �g/l for �-hydroxy-dodecanoic acid, 1 �g/l and 5 �g/l for�-hydroxy-tetradecanoic acid. A LC-MS/MS chromatogram illus-trating the chromatographic peaks obtained at these concentrationlevels was added (Fig. 2).

. The concentration of the analytes in the unknown and qual-ity control samples were calculated from the regression equation.These results are summarized in Table 2. Table 3 reports the resultsobtained for accuracy, precision and recovery, determined on sam-ples of known concentrations tested as unknown samples (QC).The accuracy and precision at lowest concentration level closer tothe LLOQ (10 �g/l) were respectively 95% and 20% for �-hydroxy-dodecanoic acid and 115% and 12% for �-hydroxy-tetradecanoicacid.

In order to estimate the extraction recovery, three blank sam-

14), http://dx.doi.org/10.1016/j.chroma.2014.01.052

ples (2 ml of purified water) were spiked in triplicate with suitableamounts of both standards solutions to give 3 levels of QC con-centrations (150 �g/l, 400 �g/l and 750 �g/l) containing 400 �g/lof internal standard and submitted to the sample preparation

t concentration levels close to LODs (2.5 �g/l).

ARTICLE ING Model

CHROMA-355085; No. of Pages 6

4 A. Chiominto et al. / J. Chromato

Table 2Analytical method performances.

Analyte �-Hydroxy-dodecanoicacid

�-Hydroxy-tetradecanoicacid

Calibration range 10–1000 �g/L 10–1000 �g/LMean slope (a ± s.d.) 0.029 ± 0.009 0.005 ± 0.001Mean intercept (b ± s.d.) 0.007 ± 0.007 0.106 ± 0.127Linear regression mean(R2 ± ds)

0.999 ± 0.009 0.997 ± 0.004

pswsct

vftHtoi

bt

a

TV

LOD 2.5 �g/L 1 �g/LLLOQ 10 �g/L 5 �g/L

rocedure described above; three additional blank samples wereubmitted to the same procedure, after which they were spikedith suitable amounts of standard solutions in order to reach the

ame final concentrations. Extraction recovery was calculated byomparing the peak areas produced by HPLC-MS/MS analysis ofhe QC spiked before and after the extraction.

The extraction recovery ranges from 31 to 42%; even if thesealues are lower than 50% it was preferred to have cleaner samplesrom SPE, considering that these are very complex matrices, in ordero reduce matrix effects and interfering peaks in the subsequentPLC phase and mass spectrometry detection. On the other side,

his is not the case of trace analysis as LPS is the main constituentf gram-negative bacterial membranes and therefore it is presentn significant concentrations in the samples.

The matrix effect on the instrumental response was evaluated

Please cite this article in press as: A. Chiominto, et al., J. Chromatogr. A (20

y comparing the peak areas of the QC spiked after extraction tohose of the corresponding QC dissolved in acetonitrile.

We called “total recovery” the product of matrix effect recoverynd extraction recovery. Matrix effect is present, confirming that

able 3alidation results for �-hydroxy-dodecanoic acid and �-hydroxy-tetradecanoic acid.

QC conc. (�g/l) Intraday accuracy Intraday % CV Interday accuracy Interd

�-Hydroxy-dodecanoic acid150 104.3 6.7 100.7 15.8

400 96.8 4.2 110.6 8.5

750 102.5 5.0 106.1 6.2

�-Hydroxy-tetradecanoic acid150 97.1 4.3 85.4 4.5

400 95.8 4.2 102.6 12.0

750 95.0 5.2 93.5 7.8

Fig. 3. HPLC-MS/MS analysis of an Esc

PRESSgr. A xxx (2014) xxx– xxx

the MS/MS response is influenced by sample composition, but it iscompensated by the use of the �-hydroxy-tridecanoic acid as inter-nal standard, as it can be seen from the accuracy that was alwaysmore than 85% and the variability that is always less than 10%.

Intraday accuracy and precision were determined as follows: QCsamples were analyzed in the same day in triplicate, and the aver-age result compared to the nominal concentration was assumed asintraday accuracy; the standard deviation of the 3 replicates wasreported as the intraday precision (expressed as % CV).

Interday accuracy and precision were determined as follows: QCsamples were analyzed in three non consecutive days, and the aver-age result compared to the nominal concentration was assumed asinterday accuracy; the standard deviation of the 3 replicates wasreported as the interday precision (expressed as %CV).

3.2. Chemical analysis of bacterial strains

Bacterial suspensions were subjected to alkaline hydrolysis torelease the esterified fatty acids from the saccharide componentsof the membrane: as pointed out by Lyngby et al. [21], the hydrol-ysis in an acidic environment, even moderate, can result in furtherchemical reactions (dehydration and trans-esterification) of thefatty acids released, which can lower the yield of hydrolysis, result-ing in a underestimation of the content of endotoxin in the sample.

The analysis of different bacterial strains confirmed the vari-ability of the composition in �-hydroxy fatty acids between

14), http://dx.doi.org/10.1016/j.chroma.2014.01.052

different gram negative species (4, 13, and 18): only the �-hydroxy-tetradecanoic acid peak was detected in Aeromonas hydrophila,Enterobacter aerogenes and Escherichia coli (Fig. 3) while Pseu-domonas aeruginosa, shows besides the �-hydroxy-tetradecanoic

ay % CV Matrix effect recovery (%) SPE recovery (%) Total recovery (%)

160.5 33.5 53.8113.7 30.9 35.1107.5 28.9 31.1

73.9 36.5 27.0112.4 31.7 35.6

84.3 41.5 35.0

herichia coli hydrolyzed extract.

ARTICLE IN PRESSG Model

CHROMA-355085; No. of Pages 6

A. Chiominto et al. / J. Chromatogr. A xxx (2014) xxx– xxx 5

Fig. 4. HPLC-MS/MS analysis of a Pseudomonas aeruginosa hydrolyzed extract.

AL tes

aMtc

dss

3

vshtbM(gfiptt

Fig. 5. Correlation between HPLC-MS/MS and L

cid also the �-hydroxy-dodecanoic acid peak (Fig. 4). The HPLC-S/MS method allowed to quantify the �-hydroxy fatty acids in

he four bacterial strains, and the sum of their concentrations wasonsidered the chemical marker of LPS.

The stability of �-hydroxy fatty acids was verified analyzing 3ifferent samples of bacterial suspensions before and after beingubjected to 3 cycles of freezing and thawing: HPLC-MS/MS analysishowed responses with variability inside interday CV%.

.3. Correlation between the results of HPLC-MS/MS and LAL

The correlation between HPLC-MS/MS and LAL results waserified analyzing serial dilutions of the same bacterial suspen-ion with both methods, for the four bacterial strains (Aeromonasydrophila, Pseudomonas aeruginosa, Escherichia coli and Enterobac-er aerogenes). The data pooled together show a linear relationshipetween the logarithm of the analytical results found by HPLC-S/MS (expressed in �g/l) and those found with the LAL-test

expressed in EU/ml) for the four bacterial strains, with a veryood correlation coefficient, being R2 = 0.82 (Fig. 5). It is the

Please cite this article in press as: A. Chiominto, et al., J. Chromatogr. A (20

rst time that a similar correlation is shown and this result isromising for the possible use of this method to the environmen-al monitoring of endotoxins. Even if these results are referredo pure cultures of bacterial strains, while the concentration in

t on 19 samples of 4 different bacterial strains.

environmental samples could be very variable, a concentrationrange from 10 to 1000 �g/l of total fatty acids corresponds to arange of endotoxin concentration from 1 to 2000 EU/ml that arerealistic values for environmental samples; on the basis of this con-sideration also the method sensitivity can be considered adequatefor the purpose.

The HPLC-MS/MS analysis of two suspensions, one of yeast (Can-dida albicans) and the other of mold (Aspergillus niger), which areknown to contain �-glucans, did not detect traces of �-hydroxyfatty acids (results < LOD), confirming the absence of interferencefrom these compounds, whereas the LAL test showed a not negli-gible enhancement, giving a recovery of the spiked sample higherthan the maximum acceptable value of 200% (data not shown).

4. Conclusions

The analytical method here presented allowed to quantify theconcentration of �-hydroxy-dodecanoic and tetradecanoic acids inbacterial suspensions, and showed that a linear correlation existsbetween the results of HPLC-MS/MS and those of the LAL-test on

14), http://dx.doi.org/10.1016/j.chroma.2014.01.052

the four bacterial strains tested; its results are not affected by thepresence of �-glucans, and remain constant after repeated freez-ing and thawing cycles of the sample. Therefore it can be usedto evaluate the endotoxin content in environmental samples. The

ING Model

C

6 omato

dfao

mas

R

[

[

[[[[

[[

[[

[

ARTICLEHROMA-355085; No. of Pages 6

A. Chiominto et al. / J. Chr

etermination of LPS markers instead of LAL can be particularly use-ul when determining endotoxin in samples rich in molds, whichre known to activate the Limulus reaction leading to amplificationf the response, and therefore to an overestimation of the results.

These results encourage the application to the microbiologicalonitoring of both life and work environments, which should be

lso aimed to distinguish between the contribution of the wateroluble and insoluble fractions of LPS.

eferences

[1] S. Sforza, A. Silipo, A. Molinaro, R. Marchelli, M. Parrilli, R. Lanzetta, J. Mass.Spectrom. 39 (2004) 378.

[2] Z.P. Vasyurenko, L.S. Opanasenko, G.M. Koval’, A.I. Turyanitsa, N.M. Ruban,Mikrobiol. Z. 63 (3) (2001) 13.

[3] E.T. Rietschel, T. Kirikae, F.U. Schade, A.J. Ulmer, O. Holst, H. Brade, G. Schmidt,U. Mamat, H.D. Grimmecke, S. Kusumoto, U. Zähringer, Immunobiology 187(3-5) (1993) 169.

[4] D.S. Kabanov, I.R. Prokhorenko, Biochemistry (Mosc) 75 (4) (2010) 383.[5] K. Todar, Todar’s Online Textbook of Bacteriology, Available from:

Please cite this article in press as: A. Chiominto, et al., J. Chromatogr. A (20

www.textbookofbacteriology.net, 2011.[6] A.M.B. Kropinski, B. Jewell, J. Kuzio, F. Milazzo, D. Berry, Antibiot. Chemother.

36 (1985) 58.[7] V.A. Kulshin, U. Zähringer, B. Lindner, K.E. Jäger, B.A. Dmitriev, E.T. Rietschel,

Eur. J. Biochem. 198 (3) (1991) 697.

[

[

PRESSgr. A xxx (2014) xxx– xxx

[8] S. Spaan, D.J. Heederik, P.S. Thorne, I.M. Wouters, Appl. Environ. Microbiol. 73(19) (2007) 6134.

[9] S. Spaan, G. Doekes, D. Heederik, P.S. Thorne, I.M. Wouters, Appl. Environ.Microbiol. 74 (12) (2008) 3804.

10] E. Paba, G. Tranfo, F. Corsetti, A.M. Marcelloni, S. Iavicoli, Ind. Health. 51 (3)(2013) 237.

11] K. Brandenburg, J. Howe, T. Gutsman, P. Garidel, Curr. Med. Chem. 16 (21) (2009)2653.

12] M. Peters, P. Fritz, A. Bufe, Innate. Immun. 18 (5) (2012) 694.13] W. Eduard, M.H. Westby, L. Larsson, Am. J. Ind. Med. 46 (2004) 375.14] R. Rylander, Int. J. Occup. Environ. Health 3 (1997) S32.15] CEN/EN 14031, Determination of Airborne Endotoxins-Workplace Atmo-

spheres, European Committee for Standardization, Brussels, Belgium,2003.

16] M.S. Muhlebach, T.L. Noah, Am. J. Repir. Crit. Care Med. 165 (7) (2002) 911.17] S.J. Reynolds, D.K. Milton, D.J. Heederik, P.S. Thorne, K.J. Donham, E.A. Croteau,

K.M. Kelly, J. Douwes, D. Lewis, M. Whitmer, I. Connaughton, S. Koch, P. Malm-berg, B.M. Larsson, J. Deddens, A. Saraf, L. Larsson, J. Environ. Monit. 7 (12)(2005) 1371.

18] A. Sebastian, C. Pehrson, L. Larsson, J. Environ. Monit. 8 (5) (2006) 519.19] D. Pomorska, L. Larsson, C. Skorska, J. Sitskowska, J. Dutkiewicz, Ann. Agric.

Environ. Med. 14 (2) (2007) 291.20] P. Kim, J.H. Kim, Biotechnol. Bioprocess. Eng. 10 (6) (2005) 494.

14), http://dx.doi.org/10.1016/j.chroma.2014.01.052

21] J. Lyngby, L.H. Olsen, T. Eidem, E. Lundanes, E. Jantzen, Biologicals 30 (1) (2002)7.

22] International Conference on Harmonisation of Technical Requirements forRegistration of Pharmaceuticals for Human Use-ICH Harmonised TripartiteGuideline, Validation of Analytical Procedures: Text and Methodology, 2005.