Journal of Pharmaceutical and Biomedical Analysis 41 (2006) 196­200
Development of an extraction method for mycobacterial
metabolome analysis
B.U. Jakia, S.G. Franzblaua, S.H. Choa, G.F. Paulia,b,
a Institute for Tuberculosis Research, College of Pharmacy, University of Illinois at Chicago, 833 S. Wood Street, Chicago, IL 60612-7231, USA
b Department of Medicinal Chemistry and Pharmacognosy, College of Pharmacy,
University of Illinois at Chicago, 833 S. Wood Street, Chicago, IL 60612-7231, USA
Received 24 August 2005; received in revised form 13 October 2005; accepted 13 October 2005
Available online 28 November 2005
Abstract
As a prerequisite for studying the intracellular metabolome of mycobacteria, several methods were evaluated for efficient breakage of the cell
using Mycobacterium bovis (BCG) as a model microorganism. Several pulping methods, treating with an Ultra-Turax
, deep-freezing in liquid
nitrogen followed by mechanical grinding, sonicating with probe head or cup horn and bead beating prior to solvent extraction were applied and
compared. Gravimetry, electron microscopy, and nuclear magnetic resonance spectrometry were used to analyze the extracts.
All analytical methods prove that sonicating is superior to mechanical grinding of deep-frozen cells. Two methods indicated that sonicating with
a probe head enhances the efficiency of cell disruption compared to sonicating with a cup horn. The highest extract yield and chemical diversity
were achieved by a combination of mechanical grinding and sonicating.
Within the scope of a metabolomic analysis, the method of choice to treat mycobacterial cells is a combination of deep-freezing in liquid nitrogen
and mechanical grinding followed by sonicating with a probe head.
 2005 Elsevier B.V. All rights reserved.
Keywords: Metabolome analysis; Extraction methods; Mycobacteria; NMR; qNMR; Breakage; Cell wall; Metabolites
1. Introduction
Metabolome analysis ­ in the generic sense of elucidating
the full complement of the low molecular weight compounds
of an organism, and as such including current definitions of
metabolomics and metabonomics [1,2] ­ is becoming recognized
as a complementary and perhaps most traceable approach
(including the insight necessary for rational drug discovery) to
a comprehensive understanding of any pathogen.
Since exhaustive organic extraction of the cell material is
the first step of any metabolome analytical method, finding a
way to open the cell wall is an essential prerequisite. Due to
differences in the architecture of bacterial cell surfaces the initial
choice of the cell breakage method is of great importance. The
mycobacteria, with their thick complex cell walls, are perhaps
the most challenging from this perspective.
 Corresponding author. Tel.: +1 312 355 1949; fax: +1 312 355 2693.
E-mail address: gfp@uic.edu (G.F. Pauli).
Without disputing the proven usefulness of available mechanical
breakage techniques, there are intrinsic limitations for the
extraction of mycobacterial cells and for the use in metabolome
analysis. Each method must carefully be assessed in light of the
physiochemical and physiological properties of the genus, in
this case Mycobacteria, and for the purpose of the cell fractionation.
If only certain cell fractions are to be isolated, a different
method might be useful as if a complete breakage of the cell wall
is desired. The French press cell, a method applying hydraulic
pressureandpressureshearing(hydrodynamicshear)andwidely
used for Gram-negative and some Gram-positive bacteria as well
as mycobacteria, normally breaks the cell wall without damaging
subcellular particles [3]. This may be an advantage, if certain
organelles are to be isolated, but is disadvantageous in a chemical
investigation aiming at small intracellular compartments and
their chemical components, which can be extracted only after a
complete degradation of the corresponding compartmental sub-
structure.
Cell breakage methods can be classified into two main
groups: mechanical and chemical. One mechanical method by
which degradation into smaller fragments can be achieved is
0731-7085/$ ­ see front matter  2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpba.2005.10.022
B.U. Jaki et al. / Journal of Pharmaceutical and Biomedical Analysis 41 (2006) 196­200 197
sonication. Rapid vibration of a resonating probe produces highintensity
sound waves, which generate microscopic air bubbles.
These transient cavities are thought to create high-shear gradients
by microstreaming [4]. Nevertheless, the reproducibility
of breakage is limited, since the result depends on several factors,
like treatment time and sample viscosity. Additionally, it
is very difficult to accommodate the French press cell and the
ultrasonic disintegration method with biosafety requirements.
Another approach is ballistic disintegration, which comes into
operation by the use of a bead beater [4]. Here, shear forces
develop when a suspension of cells together with small glass or
plastic beads is shaken or agitated, and will violently break the
bacterial cells [4,5]. A major disadvantage of this method is the
abrasion of chamber material (see results below), and its impracticality
when using organic solvents. The classical approach of
shearing or mechanical grinding is a simple method, where
frozen-lyophilized cells are broken by grinding cell paste or by
using an agate mortar and pestle [4,6,7]. The efficiency of this
process depends on the organism and the skills of the operator,
as well as time spent. This procedure has been efficiently used
for the breakage of archaebacteria [7]. Finally, some microorganisms
have been successfully lysed by microwave disruption
[8]. However, since this lysis method has been attributed primarily
to thermal effects, it appears unsuitable for a chemical
investigation, because the secondary metabolites, which are the
center of attention of a metabolomic investigation, might be heat
labile.
Chemical [9­13], osmotic [14] and alkali [3] lysis, are alternatives
to mechanical breakage methods. However, their major
disadvantage for metabolome investigation is the contamination
of the cell extract with chemicals and the unpredictable
occurrence of artifacts or reaction products during downstream
analysis. Several detergents, for example, contain aromatic
groups, which disturb UV measurements or show unfavorable
elution properties in chromatography due to their soapy
nature.
Within the scope of an isolation project to comprehend the
metabolome of mycobacteria falls the development of a reliable
method to break the mycobacterial cell walls. Mycobacteria are
known for their extremely resilient cell wall, a massive "core",
comprised of peptidoglycan which is covalently bond, via a
linker (l-Rha-d-GlcNAc-P), to a linear galactofuran, attached
to several strands of a highly branched arabinofuran, which
are in turn bound to mycolic acids [15­17]. Thus, a regime
of several pulping methods, treating with Ultra-Turax, deepfreezing
in liquid nitrogen followed by mechanical grinding,
sonicating with probe head or cup horn and bead beating was
compared. The first step, i.e. freezing the cells with liquid N2,
leads to actual "freezing" of all metabolic processes in the cells.
The mechanical cell disruption by means of sonication has the
advantage that the solvents used for extraction become part of
the disruption process themselves. Therefore, the first two steps
of a metabolome analysis, i.e. quenching and extraction, overlap
with the cell disruption process. An adequate choice of the
solvents is vital to separate the extracted material according to
polarity, which leads to the next steps, fractionation and analysis
of the cell material.
The vaccine strain of Mycobacterium bovis (BCG) was
chosen as a test microorganism because of reduced biosafety
requirements and high anatomical similarity to M. tuberculosis.
To evaluate the pulping effectiveness, the cells were subsequently
extracted by exhaustive maceration with CHCl3 and
MeOH. Electron micrographs of the cells were taken in all stages
of the procedure in order to visualize the cell wall and the degree
of destruction. In addition, 1H NMR spectra of the extracts were
acquired and the extract weights were recorded. Thus, the determination
of the efficiency of the extraction methods is based
on three parameters: optical inspection, chemical diversity, and
quantity/gravimetry.
2. Material and methods
2.1. Preparation of cell material
M. bovis, Romanian substrain I.C was obtained from the
National Institute of Research and Development for Microbiology
and Immunology Cantacuzino, a vaccine production facility
in Bucharest, Romania. The log-phase culture was grown in
Sauton's medium, washed in phosphate buffer and lyophilized.
It shall be noted that lyophilisation is not an essential part of
the presented extraction concept. The whole procedure is independent
of prior lyophilization of mycobacterial cells. The dried
cell material (200 g) was pre-extracted by using an Ultra-Turax
with CHCl3 followed by MeOH as solvents. From the residual
cell mass, six batches of 4 g dry weight each were deep-frozen
in liquid nitrogen and mechanically ground (g) with a pistil in
a mortar for 5 min. The resulting samples of each batch were
divided into three equal aliquots, which were weighed accurately.
One aliquot remained as ground (g) sample, the second
was further sonicated with a cup-holder resulting in sample
ground and sonicated with cup (gsc), the third aliquot was sonicated
with a probe head resulting in sample ground and sonicated
with probe (gsp). Six batches of all samples were used for further
analysis.
Twelve more batches of 2 g dry weight each, from the UltraTurax-treated
cell-mass were sonicated with both methods
resulting in samples sonicated with cup (sc) and sonicated with
probe (sp), six batches each, while six batches of 2 g samples
dry weight were processed with a bead beater (0.1 mm diameter
zirconia beads, 3 min) to yield six batches of sample bead
beaten (bb) (Table 1). All samples, except for sample b, which
contained massive chamber and/or rotor abrasion material, were
Table 1
Abbreviations and extract treatments
Abbreviation Treatment
g Ground
gsc Ground and sonicated with cup
gsp Ground and sonicated with probe
sc Sonicated with cup
sp Sonicated with probe
bb Bead beaten
198 B.U. Jaki et al. / Journal of Pharmaceutical and Biomedical Analysis 41 (2006) 196­200
exhaustively extracted by maceration with CHCl3, followed by
MeOH to give 60 extracts.
2.2. Electron microscopy
Electron micrographs of samples g, gsc, gsp, sc, sp as well as
from the untreated (u) cells were taken with a Hitachi S-3000N
variable pressure scanning electron microscope (SEM).
2.3. Sonicating
Sonicating was carried out with an ultrasonic liquid processor
(Misonix XL-2020 Sonicator, 600 W). Each sonicated sample
(10 mL CHCl3 as solvent, ice jacket) was treated three times for
60 s at a fixed frequency of 20 kHz using either a cup horn or
probe head. Cup horns are high-intensity ultrasonic water baths
that allow samples to be processed in completely closed containers.
The ultrasonic probe never comes in contact with the
sample so that sample loss, escape and/or cross-contamination
cannot occur. High-intensity probe heads are tipped horns, furnished
with replaceable tips, such as the microtip horns used in
this study.
2.4. Gravimetry
All weight measurements were carried out with an AND
analytical balance at a precision of 0.01 mg.
2.5. Nuclear magnetic resonance (NMR) spectroscopy
For NMR spectroscopy, 30.0 mg samples of the recombined
CHCl3 and MeOH extracts were partitioned in
CHCl3­MeOH­H2O two-phase systems, separated, evaporated
and subsequently dissolved in CHCl3 and CD3OD­D2O (2:1,
v/v), respectively, with an isotopic purity of 99.8% D, to give a
final volume of 1 mL corresponding to a filling height of 50 mm
in 5-mm tubes. The spectra were recorded on a Bruker Avance
500 spectrometer. Chemical shifts ( in ppm) were referenced
to the CD3OD multiplet at 3.300 ppm or to the CDCl3 singlet
at 7.240 ppm, respectively. For all NMR experiments, off-line
data analysis was performed using the NUTS software package,
Acorn NMR Inc., USA.
2.6. Bead beating
Zirkonia beads of 0.1 mm in diameter were mixed with 1 g
of cell material in 200 mL (total volume 350 mL) of methanol
in a Biospec Bead Beater with a stainless steel chamber. The
external jacket was filled with ice water and the mixture was
blended three times for 2 min.
3. Results
3.1. Gravimetric analysis
Gravimetric comparison of the extract yields shows that (i)
sonicating of the cell material (gsp, 12.43%, S.D. = 1.17; gsc,
Fig. 1. Gravimetric analysis of the extracts. g, ground cells; gsp, ground and sonicated
with probe head; gsc, ground and sonicated with cup horn; sc, sonicated
with cup-holder and sp, sonicated with probe head.
11.35%, S.D. = 0.65; sp, 8.24%, S.D. = 1.19 and sc, 7.07%,
S.D. = 1.09; n = 6) give higher yields than mechanical grinding
(g, 6.35%, S.D. = 0.52; n = 6) and that (ii) sonicating with probe
head (gsp, 12.43% and sp, 8.24%) in comparison with sonicating
with cup horn (gsc, 11.35% and sc, 7.07%) gives superior
results (Fig. 1).
3.2. Electron microscopic analysis
Comparison of the electron microscopic pictures shows that
all samples, which have been deep-frozen with liquid nitrogen
and mechanically ground, display disruptions of the mycobacterial
cell surface. Furthermore, they prove that sonicating
enhances cell destruction, whereas no remarkable differences
are detectable between sonicating using a probe head and sonicating
using a cup-holder (Fig. 2).
3.3. Nuclear magnetic resonance analysis
NMR as a method that allows comprehending the chemical
shifts and J patterns of the extracted compounds provides
structural information and therefore gives rise to the chemical
diversity and nature of the extracts. In contrast to the widespread
HPLC analysis, this method enables detection and assignment
of compound classes, without depending on UV chromophores
or elution properties. All the polar extracts of these samples, that
were sonicated, show prominent signals in the (hetero-)aromatic
region at 6.5­9.0 ppm. Sonicating with probe head improves
the signal integral intensity, a qualitative measure of the yield,
in those cases where the cells have been ground. These signal
groups can be assigned to intracellular material, namely nucleosides
(Figs. 3 and 4; adenosine: d = 8.205 ppm (H-2), 8.345 ppm
(H-8); guanosine: d = 7.830 ppm (H-8); uridine: d = 6.065 ppm
(H-4), 8.050 ppm (H-5)) [18], which in turn provides proof
that the cell walls are at least permeable, and more likely been
broken. Interestingly, these results show that chemical diversity
in the extracts can be improved by sonicating with a probe head
but not with a cup horn. Signals of sugar components are visible
between 3.0 and 4.5 ppm, but since severe overlapping does
not allow functional determination without chromatography, it
cannot be concluded whether they originate from intracellular
material or from the cell wall. Quantification of the nucleosides
by means of quantitative 1H NMR (qHNMR), performed
B.U. Jaki et al. / Journal of Pharmaceutical and Biomedical Analysis 41 (2006) 196­200 199
Fig. 2. Electron micrographs of the cell material in all stages of the treatment. Deep-freezing and grinding leads to broken cell surfaces. Sonicating enhances the
disruption of the cell wall.
Fig. 3. Chemical structures of the analyzed nucleosides. Sonication with probe head does not represent an advantage compared to sonication with cup horn in terms
of cell disruption (R, ribose).
Fig. 4. 1H NMR spectra of the polar fractions: diagnostic signals of the nucleosides adenosine and guanosine are contained in the sonicated fractions, while uridine
can only be detected in the fraction that was ground and sonicated with probe head. It shall be noted that, even at 500 MHz, a minimum of 128 scans had to be
accumulated in order to achieve sufficient signal/noise for identification of the highly coupled signals of the extract components and for reliable qNMR quantification.
200 B.U. Jaki et al. / Journal of Pharmaceutical and Biomedical Analysis 41 (2006) 196­200
Fig. 5. Relative content of nucleosides in equal amounts of the extracts prepared
by the different extraction methods, as determined by qHNMR. Signals specific
to the three nucleosides were quantitated (NMR integral) relative to the solvent
as an internal reference.
according to the general experimental procedure described by
Pauli et al. [19], provides a selective quantitative measure for
the deliberation of intracellular metabolites and confirms that
the amount of extracted intracellular compounds as a function
of the extraction method increases in the following order:
g < sc < sp < gsc < gsp (Fig. 5).
Non-polar fractions do not show significant differences in
signal diversity as detected by NMR. This is expected since
the majority of the extract is derived from cell wall and cell
wall-associated lipophilic material. The concentration of possible
cytosolic constituents is most likely too low to provide
significant NMR peak pattern in crude extracts and, thus, would
need further separation of the non-polar extracts.
3.4. Bead beating
Bead beating was found to be unsuitable for the purpose
of cracking mycobacterial cell walls for metabolomic analysis,
because massive abrasion of the chamber material and/or the
rotor was observed. The method was therefore abandoned.
4. Discussion
All three methods, gravimetry, scanning electron microscopy,
and nuclear magnetic resonance spectroscopy confirm that: (i)
sonicating optimizes the mycobacterial extraction process with
respect to extract mass, destructive modifications of cell surfaces,
and chemical diversity, whereas (ii) mechanical grinding
of the deep-frozen cells before extraction results in better extract
yields and improved cell cracking as evident from the optical
appearance of the extracted cells as well as increased NMR
signal diversity (chemical shift as an indicator of chemical diversity)
and integrals (qNMR quantitation). Furthermore, the NMR
results indicate that mycobacterial cells can be cracked by sonicating,
giving access to cytosolic metabolites. An increase in
chemical shift diversity is clearly achieved for polar extractibles
(intracellular/cytosolic material), whereas the non-polar extracts
(cell wall material) result in almost identical proton NMR spectra
of the crude extracts. Finally, it is noteworthy that the regimen
of (organic) extraction solvents can be adjusted to the polarity
of the target analytes of the metabolome. In order to extract
ionic/ionizable compounds, much more polar solvents can be
used, such as methanol­water mixtures, also under pH-defined
conditions. Since sonication only requires the presence of a liquid
phase for cell disruption, polar aqueous solvents are equally
suitable for combined extraction and mechanical disruption of
mycobacteria using the protocol described above.
Acknowledgements
The authors gratefully acknowledge the research support
from NIAID/NIH under grant R21-AI052847-01. Furthermore,
we thank the Research Resources Center at the University of
Illinois at Chicago for performing the SEM analysis.
References
[1] N.V. Reo, Drug Chem. Toxicol. 25 (2002) 375­382.
[2] J.C. Lindon, E. Holmes, J.K. Nicholson, Anal. Chem. 75 (2003)
385A­391A.
[3] G.D. Sprott, S.F. Koval, C.A. Schnaitman, in: P. Gerhardt, P.R.G.E. Murray,
W.A. Wood, N.R. Krieg (Eds.), Methods for General and Molecular
Bacteriology, ACM, Washington, 1994, pp. 72­103.
[4] D.E. Hughes, J.W.T. Wimpenny, D. Lloyd, in: D.W.R.J.R. Norris (Ed.),
Methods in Microbiology, Academic Press, New York, 1971, pp. 1­54.
[5] M.R.J. Salton, Methods Enzymol. 31 (1974) 653­667.
[6] K.F. Jarrel, D. Fafuy, A.M. Herbert, M.L. Kalmokoff, Can. J. Microbiol.
38 (1992) 65­68.
[7] C.F. Weil, D.S. Cram, B.A. Sherf, J.N. Reeve, J. Bacteriol. 170 (1988)
4718­4726.
[8] H. Fujikawa, H. Ushioda, Y. Kudo, Appl. Environ. Microbiol. 58 (1992)
920­924.
[9] A. Helenius, D.R. McCaslin, E. Fries, C. Tanford, Methods Enzymol.
56 (1979) 734­749.
[10] M.U. Zulauf, M. Fuerstenberger, P. Grabo, P. Jaeggi, M. Regenass, et
al., Methods Enzymol. 172 (1989) 528­538.
[11] C.A. Schnaitman, J. Bacteriol. 104 (1970) 890­901.
[12] C. Filip, G. Fletcher, J.L. Wulff, C.F. Earhart, J. Bacteriol. 115 (1973)
717­722.
[13] M. Kamekura, D. Oesterhelt, R. Wallace, P. Anderson, D.J. Kushner,
Appl. Environ. Microbiol. 54 (1988) 990­995.
[14] R.A. Laddaga, R.A. MacLeod, Can. J. Microbiol. 28 (1982) 318­324.
[15] G.S. Besra, D. Chatterjee, in: B.R. Bloom (Ed.), Tuberculosis: Pathogenesis,
Protection, and Control, ASM Press, Washington, 1994, pp.
285­302.
[16] P. Brennan, P. Draper, in: B.R. Bloom (Ed.), Tuberculosis: Pathogenesis,
Protection, and Control, ASM Press, Washington, 1994, pp. 285­302.
[17] P.J. Brennan, Tuberculosis 83 (2003) 91­97.
[18] G. Gosselin, M.-C. Bergogne, J.-L. Imbach, J. Heterocycl. Chem. 30
(1993) 1229­1233.
[19] G.F. Pauli, B.U. Jaki, D.C. Lankin, J. Nat. Prod. 68 (2005) 133­149.