Journal of Chromatography B, 871 (2008) 220­226
Contents lists available at ScienceDirect
Journal of Chromatography B
journal homepage: www.elsevier.com/locate/chromb
Phenotype differentiation of three E. coli strains by GC-FID
and GC­MS based metabolomics
Jing Tiana,b
, Chunyun Shia,b
, Peng Gaob
, Kailong Yuanb
, Dawei Yanga,b
, Xin Lub
, Guowang Xub,
a
Department of Modern Technology, Dalian Polytechnic University, Dalian 116034, China
b
Key Laboratory of Separation Science for Analytical Chemistry, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China
a r t i c l e i n f o
Article history:
Received 5 February 2008
Accepted 18 June 2008
Available online 27 June 2008
Keywords:
Metabolomics
E. coli
Phenotype
GC-FID
GC­MS
a b s t r a c t
Two mutants of E. coli with deletion of sdhAB and ackA-pta genes respectively and their wild-type strains
were subjected to gas chromatography-flame-ionization detection (GC-FID) and gas chromatographymass
spectrometry (GC­MS) metabolomics analysis. Intracellular metabolites of the three strains were
profiled by GC-FID firstly. Methodological evaluation of the employed platform indicated that the limit
of detection ranges were from 0.2 to 12.5 ng for some representative metabolites and the corresponding
recoveries were varied from 68.7 to 122.7%. Secondly, multivariable data analysis was applied to the
acquired data sets. As expected, the three phenotypes could be easily differentiated, and the perturbed
metabolite pools in the genetically modified strains were screened. Lastly, the metabolites playing key
roles in the differentiation were further identified by GC­MS. It was confirmed that succinic acid and
aspartic acid were similarly affected in the modified strains. But proline content was altered contrarily.
Additionally, deletion of sdhAB gene also affected the growth property of relevant mutant greatly. The
potential mechanism was postulated accordingly.
 2008 Elsevier B.V. All rights reserved.
1. Introduction
Metabolomics, a science dealing with small molecules,
addresses to a combination of data-rich analytical methods
together with chemometrics for metabolic profiling and interpreting
metabolic fingerprints in complex biological systems. It plays
a key role in genetic modified organism assays [1­3]. Owning to
the mostly inconspicuous phenotypic changes, it is difficult to correlate
genotypes to phenotypes. For instance, about 90% available
mutants of the Arabidopsis are the silent mutants [4]. Observed realities
suggest that a change in the activity of single enzyme--often
the result of genetic modification, usually produces a larger signal
in the metabolite concentrations than in the metabolic fluxes [5]. So
a metabolomics approach is a good option for detecting the modification
effects. Raamsdonk et al. [6] proved that pfk26 , pfk27
mutants and the wild-type Saccharomyces cerevisiae were of the
same growth rate, but they exhibited obvious differences in their
intracellular metabolite concentrations revealed by metabolomics
This paper is part of a special volume entitled "Hyphenated Techniques for Global
Metabolite Profiling", guest edited by Georgios Theodoridis and Ian D. Wilson.
 Corresponding author at: National Chromatographic R & A Center, Dalian Institute
of Chemical Physics, the Chinese Academy of Sciences, 457 Zhongshan Road,
Dalian 116023, China. Tel.: +86 411 84379530; fax: +86 411 84379559.
E-mail address: xugw@dicp.ac.cn (G. Xu).
analysis using 1H-nuclear magnetic resonance (NMR) spectroscopy.
According to their results, the content of fructose-6-phosphate was
higher in the mutants than in the wild-type, whereas, ATP/ADP
showed lower percentages in the mutants.
Metabolomic studies require reliable sensitivity, quantitation
data and robustness of analytical methodology. Currently,
NMR, gas chromatography-mass spectrometry (GC­MS), capillary
electrophoresis-mass spectrometry (CE­MS) and liquid
chromatography-mass spectrometry (LC­MS) are the typical techniques
prevailing in metabolomics world with varied features
[7­10]. GC­MSis a robust and unbiased approach to detecting unexpected
changes in transgenic lines [11]. Combined with the easily
accessible database of NIST (www.nist.gov), GC­MS gained more
application in different fields. Genetically or environmentally modified
plant systems have been extensively studied by using GC­MS
for many years [12­14]. Prokaryotic cells were also analyzed by
GC­MS metabolomics and different data processing strategies were
also developed accordingly to extract the implied information contained
in the generated data [15,16]. As a well-adopted strategy, gas
chromatography-flame-ionization detection (GC-FID) is thought to
be more reliable and sensitive for quantitative analysis than GC­MS,
whereas, GC­MS can provide more definite qualitative information
[17].
Previously, we had constructed two mutants of E. coli for the
purpose of high succinic acid production. Although the primary fermentation
results indicated that the production of succinic acid was
1570-0232/$ ­ see front matter  2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jchromb.2008.06.031
J. Tian et al. / J. Chromatogr. B 871 (2008) 220­226 221
not improved, the genome-wide changes affecting the metabolic
activity of the cells were expected to be in effect. Here, we employed
GC-FID and GC­MS metabolomics tactics to differentiate the phenotypic
changes of the mutant and the wild-type strains. GC-FID
was used to collect the metabolic profiling data for multivariate
analysis, the significantly changed metabolites were then subjected
to identification by GC­MS.
2. Experimental
2.1. Strains and medium
E. coli 1.1566 (wild-type), DC100 ( sdhAB) and DC101 ( ackApta)
were obtained from Group 1816 of Dalian Institute of Chemical
Physics, Chinese Academy of Sciences. DC100 was the mutant of
deletion of succinodehydrogenase (SDH) gene, and DC101 was the
acetate kinase gene knocked out mutant. LB broth was purchased
from Chinese Institute of Import & Export Commodity Inspection
Technology (Beijing, China). Cell concentrations were determined
by monitoring turbidity of the cultures at wavelength of 600 nm
(OD600). The turbidity of the culture was then converted to cell dry
weight (CDW)/ml. Residual glucose was quantified using Olympus
AU 2700 biochemical analyzer (Olympus, Tokyo, Japan) by employing
glucose oxidase method.
2.2. Chemicals and instruments
Dimethylformamide (DMF) and N,Obis(trimethylsilyl)trifluoroacetamide
(BSTFA) were the products
of Sigma­Aldrich (Steinheim, Germany). Methanol, chloroform
and pentaerythritol were all purchased from TEDIA (Fairfleld, OH).
The mixed standard stock solution of amino acids was prepared in
the distilled water (1 mg/ml), and organic acids were prepared in
methanol (Table 1).
Freeze drying system (FREEZONE4.5) was from LABCONCOC
(Kansas, MI). GC-FID detection was performed on an Agilent 6890N
GC system (Santa Clara, CA). GC­MS analysis was accomplished on
a Shimadzu QP 2010 GC­MS system (Kyoto, Japan). Water used
in the experiments was purified with Milli-Q system (Bedford,
MI).
2.3. Bacterial growing, quenching and extraction
Three E. coli strains were cultured in LB broth with constant agitation
(160 rpm) aerobically at 30 C for 12 h. Fresh culture of each
Table 1
Linearity and detection limit of standards
Compounds Linear range
( g/ml)
n r2
Limit of detection
(ng, on column)
l-Alanine 5­1000 8 0.9979 0.31
l-Leucine 5­1000 8 0.9944 0.56
l-Threonine 5­1000 8 0.9898 3.41
l-Proline 5­1000 8 0.9981 0.37
l-Glutamic acid 10­1000 6 0.9948 12.50
l-Phenylalanine 5­1000 7 0.994 2.50
l-Lysine 5­1000 7 0.9997 2.96
l-Aspartic acid 5­1000 7 0.9997 3.63
Phosphate 1­1000 8 0.9982 0.24
Glycerol 1­1000 8 0.995 0.48
Malate 1­1000 8 0.9998 0.53
Succinate 1­1000 8 0.9954 0.50
Myristic acid 1­1000 8 0.9997 0.84
Palmitic acid 1­1000 8 0.9967 0.37
Stearic aid 5­1000 7 0.9997 1.50
"n": number of concentration points across the range.
was diluted with aseptic broth to the final concentration of 7.5 × 108
colony forming unit/ml using a bioMerieux (France) DENSIMAT
transmissometer. 0.5 ml of bacterial suspension for each was added
to 50 ml LB broth in shake flask (150 ml) and then cultured with constant
agitation (160 rpm) until to exponential growth phase at 30 C
aerobically. Every strain was cultured in parallel for nine bottles,
which generated totally 27 samples.
20 ml of bacterial suspension of each strain was centrifuged
(3000 × g) at 4 C for 5 min. The pellet was washed with aseptic
physiological saline in triplicate at 4 C individually. 1 ml of
-45 C methanol was added to each and the mixtures were settled
at 4 C for 10 min. The turbidities of the suspensions were
adjusted to the lowest turbidity in the 27 samples using cold
methanol for the convenience of easy comparison. Subsequently,
1 ml of -45 C chloroform, 0.5 ml purified water and 9 l of
0.6 mg/ml pentaerythritol (internal standard) were added to 1 ml
bacterial suspension for each and vortexed for 1 min, respectively.
After centrifugation for 5 min at 6000 × g, the supernatant
was filtered through 0.22 m organic membrane. The filtrate was
lyophilized and stored in a -80 C refrigerator until further pro-
cessing.
2.4. Derivatization procedure
Because silylation regents were more effective to react with
hydroxyl, carboxyl and hydrosulfide groups, the BSTFA was selected
as the derivatization regent. The solvent used was DMF. The ratio of
BSTFA to DMF was 1:1 (v/v, totally 150 l). The derivatization was
accomplished at 60 C for 40 min.
2.5. GC­MS analysis
GC­MS analysis was accomplished on a Shimadzu QP 2010
GC­MS system using a 30 m × 0.25 mm (i.d.) fused-silica capillary
column chemically bonded with 0.25 m DB5-MS stationary
phase (J&W Scientific, Folsom, CA). The injectior temperature was
280 C and split ratio was 10:1. Helium was used as carrier gas
with linear velocity of 35.0 cm/s and an equilibration time of 3 min.
The column temperature was initially kept at 70 C for 5 min and
then increased to 280 C at 5 C/min, held for 5 min. The MS scan
parameters included a mass scan range of m/z 40­600, a scan
interval of 0.5 s, a scan speed of 1000 m/s, and a detector voltage
of 0.9 kV. The ion source temperature was set at 200 C, and
the interface temperature was 280 C. The solvent cut time was
6 min.
2.6. GC-FID analysis
GC-FID analysis was accomplished on an Agilent 6890 GC-FID
system. The detector temperature was 300 C. The other analytic
conditions including the column type and column temperature, the
injectior temperature, split ratio, carrier gas and the linear velocity
were the same as those of GC­MS analysis.
2.7. Data processing
Peaks with signal to noise ratios (S/N) higher than 10 in the GCFID
data set were selected, and their areas were quantitated against
the internal standard through dividing corresponding areas by the
internal standard area. The processed data were fed to SIMCA-P
software (Umea, Sweden) for multivariate data analysis using the
algorithm of principal component analysis (PCA). Variables playing
key roles in the pattern recognition model were extracted and then
identified via GC­MS assay utilizing commercial compounds.
222 J. Tian et al. / J. Chromatogr. B 871 (2008) 220­226
Fig. 1. Growth and the residual glucose curves of the three strains. Dashed line
indicated the sampling point at 6 h.
3. Results and discussion
3.1. Growth characteristics of the three strains
Under the aerobic conditions, the growth curves of three strains
were investigated. The wild-type and the DC101 shared with the
similar growth features (Fig. 1), but the strain of DC100 was characterized
of lower yield of biomass, nearly half of the wild-type
after 30 h cultivation. The residual glucose concentrations in the
fermentation liquid of the three strains changed similarly. This phenomenon
indicated that deletion of sdh gene severely compromised
the growth of E. coli. According to Fig. 1, these three strains were
all approached to exponential growth phase during the period of
5­8 h, so the sampling time was fixed to the time point of 6 h.
3.2. Analytical characteristics of the GC-FID method developed
In order to evaluate the applicability of the adopted strategy,
methodological appraisal of the GC-FID platform was conducted
briefly using some commonly found compounds in bacterial cells.
The linearity of response was determined by using serial standard
solutions ranging from 5 to 1000 g/ml for amino acids and
1­1000 g/ml for organic acids and other compounds. Calibration
curves for the tested compounds were adjusted using the internal
standard. For most of the investigated chemicals, the correlation
coefficients of determination (r2) were generally satisfied (Table 1).
Limit of detection was defined as the concentration of a compound
resulting in a peak with a signal-to-noise ratio (S/N ratio) higher
than 3, corresponding to the range of 0.2­12.5 ng for different individuals
(Table 1).
Table 2
Recoveries determined by spiking standards to E. coli cell suspension
Compounds Recoveries (%)
Average R.S.D. (n = 3)
l-Alanine 71.2 5.1
l-Threonine 72.3 5.4
l-Proline 68.7 1.3
l-Aspartic acid 77.4 12.5
Phosphate 122.7 1.7
Malate 79.8 1.2
Succinate 73.4 1.7
Myristic acid 83.6 1.8
Palmitic acid 85.0 1.6
At the same time, the serial standard solutions were used to
investigate the reproducibility. After corrected with internal standard,
the relative standard deviations (R.S.D.s) of peak areas for the
standards of different concentrations were acceptable, the R.S.D. of
peak areas varied from 0.7 to 6.2%. To study the extraction recoveries
of the metabolites from cellular samples, known amounts
of compounds were spiked to cell suspensions of E. coli prior to
extraction and lyophilization at the concentration of 15 ng/ l
for each. Determined by the quantitation results from the spiked
and unspiked samples, the recoveries varied from 68.7 to 122.7%
(Table 2). These parameters ensured that the data acquired from
GC-FID were acceptable for further analysis. Fig. 2 showed the typical
GC-FID chromatograms of intracellular metabolites of the three
strains.
3.3. Metabolic fingerprint PCA analysis
GC-FID data were used to compare the phenotypic differences
amid the three strains by using PCA. Processed data showed that
the three strains could be easily differentiated from each other
(Fig. 3). The PCA scores plot and loading plot of the wild-type
to ackA-pta strains and the wild-type to sdhAB strains were
shown in Figs. 4 and 5. In the loading plot, the farther the compounds
(variables) locate from the origin, the more important
they are for the differentiation pattern. These metabolites were
taken out for further identification by GC­MS. After t-test, it was
found that all compounds in Tables 3 and 4 were of p-value  0.05,
showing a significant difference between the wild-type and the
genetically modified strains. Obviously, the metabolite pools of
succinic acid, proline and aspartic acid were affected in both
mutants.
Owning to the advantages of GC­MS and GC-FID individually,
combining using these two platforms is more preferable in
Fig. 2. Intracellular metabolite GC-FID chromatograms of wild-type E. coli (A), DC101 (B) and DC100 (C). (a) Lactic acid; (b) phosphoric acid; (c) succinic acid; (d) pyrimidine;
(e) pentaerythritol; (f) proline; (g) aspartic acid; (h) phenylalanine; (i) glutamate; (j) uracil (peak identification was confirmed by GC­MS using commercial standards).
J. Tian et al. / J. Chromatogr. B 871 (2008) 220­226 223
Fig. 3. PCA scores plot (a) and loading plot (b) of the wild-type ( ), ackA-pta ( ) and sdhAB ( ) strains based on GC-FID data.
analytical biochemistry. Many types of samples and many endogenous
or exogenous compounds had been analyzed by this strategy
[18­21]. This could balance the quantitative and qualitative concerns
in most circumstance.
It can be anticipated that the concentration of succinic acid
would be elevated in the genetic modified strains compared to the
wild-type because of the blockage of branch pathways consuming
pyruvic acid and succinic acid. The enlargement of intracellular
metabolite pool of succinic acid was more significant in DC100 than
in DC101 (p < 0.05), because the deletion of sdhAB might be a substantially
direct signal to interrupt the consumption of succinic
acid.
Interestingly, the deletion of sdhAB gene severely compromised
the growth property of E. coli compared to the destruction of
ackA-pta gene. Under the aerobic condition, succinate is only an
intermediate of the tricarboxylic acid (TCA) cycle or the glyoxalate
shunt metabolism [22]. Destruction of sdhAB gene would result
in destruction of the integrity of TCA circle and cause the activaTable
3
Metabolites with statistical difference in the wild-type and DC101 ( ackA-pta)
Metabolites Concentrations (g/g CDW) p
Retention time (min) Compounds E. coli 1.1566 DC101 ( ackA-pta)
11.82 Alanine ­ 5.98 × 10-4
1.94 × 10-6
18.11 Glycerol 9.24 × 10-4
6.28 × 10-4
3.48 × 10-3
19.48 Succinic acid 6.16 × 10-4
1.87 × 10-3
9.18 × 10-10
20.15 Pyrimidine 1.30 × 10-4
5.08 × 10-4
3.64 × 10-19
25.89 Proline 5.77 × 10-4
3.82 × 10-4
3.05 × 10-10
26.03 Aspartic acid 6.50 × 10-4
3.04 × 10-4
2.15 × 10-5
26.37 Phenylalanine ­ 4.30 × 10-5
1.07 × 10-8
48.27 Uracil 1.98 × 10-4
3.97 × 10-4
5.24 × 10-14
"­": undetected.
224 J. Tian et al. / J. Chromatogr. B 871 (2008) 220­226
Fig. 4. PCA score plot (a) and loading plot (b) of the wild-type ( ) and ackA-pta ( ) strains based on GC-FID data.
tion of glyoxylate shunt which is usually not active in common
circumstances. For 1 mol of succinate generation, the net pyruvate
consumption of glyoxylate shunt is two times of TCA circle [23].
Intracellular succinate concentration was found elevated greatly
in DC100. The interrupted TCA circle brought about the deficiency
of energy generation; the higher content of succinate caused the
excessive consumption of pyruvate. Thus, the lower biomass formation
was observed in DC100.
Acetate is usually excreted by E. coli as an end product and E.
coli cannot endure higher concentration of acetate in most circumstances.
Acetate excretion is controlled by acetate switch. When
cultured in glucose medium, the switch is triggered following the
onset of stationary phase, at which point the primary carbon source
is fully consumed and acetate will be assimilated as carbon source
[24]. So destruction of acetate production might not severely affect
the growth of E. coli. Despite the accumulation of intracellular succinate,
we were not able to observe signs of its over-production in
the fermentation liquid. So it could be postulated that metabolism
of succinate and the relevant metabolites was under stringent regulation
in E. coli. Furthermore, in E. coli cell, succinate dehydrogenase
broadly interacted with other enzymes taking part in DNA replication
(e.g. DNA polymerase II), protein synthesis (e.g. 30S ribosomal
protein S2 and tyrosyl-tRNA synthetase), lipid metabolism (e.g.
long-chain-fatty-acid­CoA ligase) and transcriptional regulators
(e.g. transcriptional regulatory protein rtcR), etc. (www.ecocyc.org)
compared to ackA-pta gene product. So we could conclude that SDH
Table 4
Metabolites with statistical difference in the wild-type and DC100 ( sdhAB)
Metabolites Concentrations (g/g CDW) p
Retention time (min) Compounds E. coli 1.1566 DC100 ( sdhAB)
18.02 Phosphoric acid 1.03 × 10-3
8.42 × 10-4
5.00 × 10-3
19.48 Succinic acid 6.16 × 10-4
2.39 × 10-3
8.39 × 10-13
20.15 Pyrimidine 1.30 × 10-4
2.72 × 10-4
3.29 × 10-7
25.89 Proline 5.77 × 10-4
1.08 × 10-3
7.95 × 10-12
26.03 Aspartic acid 6.50 × 10-4
1.03 × 10-4
2.46 × 10-6
26.37 Phenylalanine ­ 5.25 × 10-5
6.44 × 10-10
28.97 Glutamic acid 4.45 × 10-4
1.61 × 10-3
3.76 × 10-14
35.06 Cadaverine 8.00 × 10-5
3.83 × 10-5
3.45 × 10-6
48.27 Uracil 1.98 × 10-4
1.46 × 10-4
2.59 × 10-6
"­": undetected.
J. Tian et al. / J. Chromatogr. B 871 (2008) 220­226 225
Fig. 5. PCA score plot (a) and loading plot (b) of the wild-type ( ) and sdhAB ( ) strains based on GC-FID data.
also played a crucial role in keeping the biological stabilization of
E. coli.
4. Conclusions
In conclusion, the recently introduced functional genomics technology,
especially metabolomics, is revolutionizing research in the
biological sciences. Although still in its infancy, metabolomics is a
convenient tool to understand bacterial functioning in microbiology.
Using GC-FID and GC­MS platform, the phenotypic difference
among three strains of E. coli could be easily discerned. Deletion of
sdhAB gene brought about a much more complex perturbation on
E. coli cell than deletion of ackA-pta gene. The most profound effect
stemming from deletion of sdhAB gene might be the perturbation
on broad interactions amid the cellular components and, potentially,
the involved metabolic pathways. This research work proved
again that metabolomics analysis is indeed a good approach for
phenotype differentiation.
Acknowledgments
The studies have been supported by National Basic Research Program
(2003CB716003, 2004CB719605 and 2007CB707800) of State
Ministry of Science and Technology of China, the National Natural
Science Foundation of China (no. 20776029) and the Knowledge
Innovation Program of the Chinese Academy of Sciences (K2006A8,
K2006A13).
References
[1] I. Wilson, Mol. Syst. Biol. 3 (2007) 113.
[2] D. Hwang, J.J. Smith, D.M. Leslie, A.D. Weston, A.G. Rust, S. Ramsey, P. de Atauri,
A.F. Siegel, H. Bolouri, J.D. Aitchison, L. Hood, Proc. Natl. Acad. Sci. U.S.A. 102
(2005) 17302.
[3] S. Draghici, P. Khatri, A.L. Tarca, K. Amin, A. Done, C. Voichita, C. Georgescu, R.
Romero, Genome Res. 17 (2007) 1537.
[4] R.D. King, K.E. Whelan, F.M. Jones, P.G. Reiser, C.H. Bryant, S.H. Muggleton, D.B.
Kell, S.G. Oliver, Nature 427 (2004) 247.
[5] D.A. Fell, J. Exp. Bot. 56 (2005) 267.
[6] L.M. Raamsdonk, B. Teusink, D. Broadhurst, N. Zhang, A. Hayes, M.C. Walsh, J.A.
Berden, K.M. Brindle, D.B. Kell, J.J. Rowland, H.V. Westerhoff, K. van Dam, S.G.
Oliver, Nat. Biotechnol. 19 (2001) 45.
[7] M.M. Koek, B. Muilwijk, M.J. van der Werf, T. Hankemeier, Anal. Chem. 78 (2006)
1272.
[8] W.B. Dunn, D.I. Ellis, Trends Anal. Chem. 24 (2005) 285.
[9] J. van der Greef, P. Stroobant, R. van der Heijden, Curr. Opin. Chem. Biol. 8 (2004)
559.
[10] O. Fiehn, W. Weckwerth, Eur. J. Biochem. 270 (2003) 579.
[11] U. Roessner, C. Wagner, J. Kopka, R.N. Trethewey, L. Willmitzer, Plant J. 23 (2000)
131.
[12] W. Weckwerth, M.E. Loureiro, K. Wenzel, O. Fiehn, Proc. Natl. Acad. Sci. U.S.A.
101 (2004) 7809.
[13] O. Fiehn, J. Kopka, P. Dormann, T. Altmann, R.N. Trethewey, L. Willmitzer, Nat.
Biotechnol. 18 (2000) 1157.
[14] U. Roessner, A. Luedemann, D. Brust, O. Fiehn, T. Linke, L. Willmitzer, A. Fernie,
Plant Cell 13 (2001) 11.
[15] N. Ishii, K. Nakahigashi, T. Baba, M. Robert, T. Soga, A. Kanai, T. Hirasawa, M.
Naba, K. Hirai, A. Hoque, P.Y. Ho, Y. Kakazu, K. Sugawara, S. Igarashi, S. Harada,
T. Masuda, N. Sugiyama, T. Togashi, M. Hasegawa, Y. Takai, K. Yugi, K. Arakawa, N.
Iwata, Y. Toya, Y. Nakayama, T. Nishioka, K. Shimizu, H. Mori, M. Tomita, Science
316 (2007) 593.
[16] H.H. Kanani, M.I. Klapa, Metab. Eng. 9 (2007) 39.
[17] M. Ligor, B. Buszewski, J. Sep. Sci. 31 (2008) 364.
[18] N.L. Koryagina, E.I. Savelieva, N.S. Khlebnikova, N.V. Goncharov, R.O. Jenkins,
A.S. Radilov, Anal. Bioanal. Chem. 386 (2006) 1395.
226 J. Tian et al. / J. Chromatogr. B 871 (2008) 220­226
[19] M.V. Russo, L. Di, V.P. Avino, Ann. Chim. 94 (2004) 741.
[20] M.A. LeBeau, M.A. Montgomery, M.L. Miller, S.G. Burmeister, J. Anal. Toxicol. 24
(2000) 421.
[21] M.A. Martinez, S. Ballesteros, T.C. Sanchez de la, A. Sanchiz, E. Almarza, A.
Garcia-Aguilera, J. Anal. Toxicol. 28 (2004) 609.
[22] M.J. Van der Werf, M.V. Guettler, M.K. Jain, J.G. Zeikus, Arch. Microbiol. 167
(1997) 332.
[23] G.N. Vemuri, M.A. Eiteman, E. Altman, Appl. Environ. Microbiol. 68 (2002) 1715.
[24] M. EI-Mansi, A.J. Cozzone, J. Schiloach, B.J. Eikmanns, Curr. Opin. Microbiol. 9
(2006) 173.