Cholesteroid nature of free mycolic acids from M. tuberculosis

Article abstract of DOI:10.1016/j.chemphyslip.2008.01.004

Mycolic acids (MAs) are a major component of the cell walls of Mycobacterium tuberculosis and related organisms. These α-alkyl β-hydroxy long fatty acids have been the subject of numerous studies for their immunological properties. We previously reported that an interaction between cholesterol and mycolic acids could be responsible for the low accuracy in the serodiagnosis of TB when using free mycolic acid in an ELISA assay. The aim of this work was to investigate if this interaction could be due to a similarity in the structural properties between mycolic acids and cholesterol. The investigation revealed that patient sera cross-reacted with mycolic acids and cholesterol in an ELISA experiment suggesting that both molecules may present related functionality in a similar structural orientation. This relation was further supported by the interaction of mycolic acids with Amphotericin B (AmB), a known binding agent to ergosterol and cholesterol. Using a resonant mirror biosensor, we observed that AmB recognised both cholesterol and mycolic acids. In addition, a specific attraction was observed between mycolic acid and cholesterol by the accumulation of cholesterol from liposomes in suspension onto immobilized mycolic acids containing liposomes, detected with a biosensor technique. Combined, these results suggest that mycolic acids can assume a three-dimensional conformation similar to a sterol. This requires that mycolic acid exposes its hydroxyl group and assumes rigidity in its chain structure to generate a hydrophobic surface topology matching that of cholesterol. A particular folded conformation would be required for this, of which a few different types have already been proven to exist in monolayers of mycolic acids.

Full text of DOI:10.1016/j.chemphyslip.2008.01.004

Available online at www.sciencedirect.com
Chemistry and Physics of Lipids 152 (2008) 95–103
Cholesteroid nature of free mycolic acids from M. tuberculosis
Yolandy Benadiea,1, Madrey Deysela,1, D. Gilbert R. Sikoa, Vanessa V. Robertsa,
Sandra Van Wyngaardta, Simon T. Thanyania, Gianna Sekankaa,
Annemieke M.C. Ten Bokuma,2, Lynne A. Collettb, Johan Grootenc,
Mark S. Bairdd, Jan A. Verschoora,∗
a Department of Biochemistry, University of Pretoria, South Africa
b Department of Chemistry, University of Pretoria, South Africa
c Department of Molecular Biomedical Research, Molecular Immunology Unit, Ghent University, Belgium
d School of Chemistry, University of Wales, Bangor, United Kingdom
Received 13 August 2007; received in revised form 23 December 2007; accepted 28 January 2008
Available online 7 February 2008
Abstract
Mycolic acids (MAs) are a major component of the cell walls of Mycobacterium tuberculosis and related organisms. These ␣-alkyl ␤-hydroxy
long fatty acids have been the subject of numerous studies for their immunological properties. We previously reported that an interaction between
cholesterol and mycolic acids could be responsible for the low accuracy in the serodiagnosis of TB when using free mycolic acid in an ELISA
assay. The aim of this work was to investigate if this interaction could be due to a similarity in the structural properties between mycolic acids
and cholesterol. The investigation revealed that patient sera cross-reacted with mycolic acids and cholesterol in an ELISA experiment suggesting
that both molecules may present related functionality in a similar structural orientation. This relation was further supported by the interaction of
mycolic acids with Amphotericin B (AmB), a known binding agent to ergosterol and cholesterol. Using a resonant mirror biosensor, we observed
that AmB recognised both cholesterol and mycolic acids. In addition, a specific attraction was observed between mycolic acid and cholesterol by
the accumulation of cholesterol from liposomes in suspension onto immobilized mycolic acids containing liposomes, detected with a biosensor
technique. Combined, these results suggest that mycolic acids can assume a three-dimensional conformation similar to a sterol. This requires that
mycolic acid exposes its hydroxyl group and assumes rigidity in its chain structure to generate a hydrophobic surface topology matching that
of cholesterol. A particular folded conformation would be required for this, of which a few different types have already been proven to exist in
monolayers of mycolic acids.
© 2008 Elsevier Ireland Ltd. All rights reserved.
Keywords: Mycolic acids; Cholesterol; Lipid conformation; Amphotericin B; Resonant mirror biosensor
1. Introduction
demic and where the connection between TB and HIV/AIDS
infection is particularly worrying. In fact, TB is the major cause
of death in HIV/AIDS co-infected individuals (Uma Devi et al.,
2003).
In the search for new surrogate markers for the diagnosis
of tuberculosis, antibodies to cord factor (trehalose-6,6ꢀ-
dimycolate, TDM) have attracted much attention. Although
having a hydrophobic nature, TDM is a very immunogenic and
biologically active substance present in the mycobacteria and
a few related genera. TDM is a glycolipid consisting of tre-
halose, which is identical for all cord factors, and two of a set of
mycolic acids (MAs) (1, Fig. 1), which differ within and among
species and genera (Ryll et al., 2001). For example, Mycobac-
Tuberculosis is one the most life-threatening infectious dis-
eases. Its re-emergence has brought new interest in finding novel
methods for its treatment and diagnosis, especially in developing
countries where the population is most burdened by this pan-
∗
Corresponding author at: Department of Biochemistry, University of Preto-
ria, Pretoria 0002, South Africa.
E-mail address: Jan.Verschoor@up.ac.za (J.A. Verschoor).
These authors contributed equally towards the article.
Current address: Department of Infectious and Tropical Diseases, London
1
2
School of Hygiene and Tropical Medicine, London, United Kingdom.
0009-3084/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.chemphyslip.2008.01.004

96
Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
Fig. 1. Molecules under investigation for their similarity in function.
terium tuberculosis contains ␣-, methoxy- and keto-mycolic
acids subclasses, whereas Mycobacterium avium contains wax
ester—instead of methoxymycolic acids (Dobson et al., 1985).
Fujiwara et al. (1999) identified the antigenic epitope in cord
factor molecules to be the mycolic acids by proving that anti-
bodies against TDM from TB patients could distinguish between
mycolic acids subclasses.
We have reported the use of free mycolic acids as antigens
for the serodiagnosis of tuberculosis; they have consider-
able potential because HIV–TB co-infected patients maintain
high antibody levels to mycolic acids, despite the severity
of the immune deficiency measured by the fall in CD4 T
cell count (Schleicher et al., 2002). However, in an ELISA
assay, the use of free mycolic acids was not adequate for
serodiagnosis of tuberculosis (accuracy = 57%). An associa-
tion between mycolic acids and cholesterol was hypothesised
as a possible reason for this low accuracy (Schleicher et al.,
2002). Cholesterol may be non-specifically attracted to mycolic
acids by means of hydrophobic Van der Waals type of bind-
ing, or by a more specific interaction such as a hydrogen
bond, arising from conformational features present in the two
molecules.
acids could be imagined to assume a shape similar to that of
cholesterol.
The current study was designed to investigate the possibil-
ity that the low accuracy measured for a serodiagnostic ELISA
assay based on the use of mycolic acid as antigen could be
the result of the folding of free mycolic acids to resemble a
“cholesteroid” shape. In order to investigate this, we have used
an approach similar to that employed by Prendergast et al.
(1998) to suggest molecular mimicry between microbial and
self-structures in autoimmune diseases. The basic assumption
of this approach is that the specific molecular recognition of
two substances by an established binding agent indicates resem-
blance in the three-dimensional structure of the two compounds.
Therefore we further investigated the interaction of mycolic
acids and cholesterol with sera from human TB patients, using
the ELISA technique described by Schleicher et al. (2002). Sec-
ondly, using a resonant mirror biosensor (Cush et al., 1993;
Buckle et al., 1993; Athanassopoulou et al., 1999), we compared
the interaction of these two lipids with Amphotericin B (AmB),
a known cholesterol binding molecule (Baginski et al., 2002,
2005). Thirdly, the attraction between cholesterol and mycolic
acids was studied, also using the resonant mirror biosensor.
Recently, free mycolic acids have been demonstrated to
be able to adopt a folded conformation to give a hydrophobic
surface. In particular, the “W” conformation has the alkyl
chains folded to give four parallel arms, while in the “Z”
conformation three folded arms provide the three-dimensional,
curved hydrophobic surface (Villeneuve et al., 2007; Sekanka
et al., 2007). The existence of these conformations has been
suggested by analyses of Langmuir monolayers consisting of
free mycolic acids over a range of temperatures (Villeneuve et
al., 2005, 2007). In an extended form, the shape and structure
of a mycolic acid appears very different to that of cholesterol
and does not suggest particularly strong interactions between
these molecules. However, the folded conformations of mycolic
2. Materials and methods
2.1. General procedures
1H, 13C and COSY nuclear magnetic resonance (NMR) were
recorded on a Bruker advance DRX-500 spectrometer or Bruker
AC-300 spectrometer with chemical shifts in ppm. Analytical
thin layer chromatography (TLC) was performed on Alugram
SIL G/UV, layer: 0.2 mm silica gel 60 with fluorescent indi-
cator UV254 plates. The plates were viewed under ultraviolet
light (254 and 366 nm). Flash column chromatography was
performed using Merck silica gel. FT-IR was obtained with a

Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
97
PerkinElmer BX-1 instrument of 2.00 cm−1 resolution fitted
with a MIR source, an internal LiT aO3 detector and FT-IR Spec-
trum Software version 5.1. Optical rotations were measured on
a PerkinElmer model 341 polarimeter. For mycolic acids analy-
ses HPLC was performed using a Merck Hitachi Chromatograph
fitted with a Phenomenex Luna 5 ␮m C18 column and a Merck
Hitachi L-4500 Diode Array detector. For AmB derivative
purification a Waters high performance liquid chromatography
(HPLC) system was used with a Waters 610 fluid unit, Waters
600 controller and a Waters 996 Photodiode Array detector.
A Waters C18 reverse phase column (150 mm × 4.6 mm) was
used for optimization of methods and a Phenomenex Luna C-18
reverse phase column (250 mm × 10 mm, 10 ␮m) was used to
collect the product. Electrospray mass spectrometry performed
on a Micromass Micro Triple quadrupole mass spectrometer in
full scan mode, range 500–2000 amu, was used to characterise
the product. Sample infusion was directly into the electron spray
source by means of a Hamilton syringe pump at 10 ␮l/min.
The precipitate that formed was removed by filtration and the
mother liquor was concentrated to give a crude product (0.15 g)
that was purified by flash column chromatography on silica gel
in ethyl acetate:methanol:ethanol containing 1% triethylamine
(90:9:1, v/v) to give pure Nꢀ-(4-formylbenzylidene) isonicotino-
hydrazide (6, 0.12 g, 14% yield).
Molecular weight 253.2599 g mol−1
. Rf 0.57 (ethyl
acetate:methanol:ethanol with 1% Et3N (v/v) (90:9:1); mp
219–220 ◦C; IR νmax 3465, 3197 (NH), 1698 (CHO) cm−1. 1H
NMR (300 MHz, (CD3)2 SO) δ 12.25 (1H, s, NH), 10.06 (1H, s,
1a), 8.80 (2H, d, J = 5.4 Hz, 10a), 8.55 (1H, s, 6a), 8.01 (2H, d,
J = 9 Hz, 3a), 7.97 (2H, d, J = 9 Hz, 4a), 7.84 (2H, d, J = 5.4 Hz,
9a); 13C NMR (300 MHz, (CD3)2 SO) δ 193.1, 162.3, 150.7,
148, 140.7, 139.9, 137.4, 130.3, 128.1, 121.9 ppm; HR–MS (EI)
calculated for C14H10O2N3 [M+] 253.08513 g mol−1 found
253.08389 g mol−1, m/z (EI) 122 (74), 106 (1 0 0), 79 (10), 78
(47), 51 (22).
2.3.2. Preparation of N-(4-((2-
isonicotinoylhydrazono)methyl)benzyl)-AmB
(7, AmB derivative)
2.2. Preparation of methyl mycolic acids isolated from M.
tuberculosis (H37Rv)
In the next step, the aldehyde (6, 0.01 g, 0.04 mmol) and
Amphotericin B (0.043 g, 0.05 mmol) were stirred at RT in 2 ml
dimethylsulfoxide in the dark for an hour. The derivative that
formed was reduced in situ with NaBH4 (0.001 g, 0.04 mmol)
overnight and purified by RP-HPLC (methanol:H2O a gradient
from 20% to 100% methanol). The column eluent was mon-
itored at 300 and 407 nm and AmB derivative collected and
concentrated.
A mixture of mycolic acids (1) was isolated from the cell wall
of the virulent Erdman strain of M. tuberculosis as described
by Goodrum et al. (2001). HPLC and NMR spectra were in
agreementwiththedatareportedinliteratures(Stecketal., 1978;
Watanabe et al., 2001).
To form the methyl esters, mycolic acids (1, 100 mg,
∼0.1 mmol) were dissolved in a mixture of toluene:methanol
(5:1, 18 ml), trimethylsilyldiazomethane (TDM, 2 M solution,
0.2 ml, 0.4 mmol) added, followed by a further 4 additions of
TDM (0.1 ml, 0.2 mmol) every 45 min. The mixture was stirred
for 72 h, and then quenched by evaporation. The residue was dis-
solved in dichloromethane (15 ml) and water (10 ml) was added.
The two layers were separated and the water layer extracted
with dichloromethane (2× 10 ml). The combined organic layers
were dried and the solvent evaporated to give the desired com-
pound (98 mg, ∼97%). The NMR spectra of the compounds
obtained corresponded to those reported in the literatures for
methyl esters of mycolic acids (MAME, 2) (Watanabe et al.,
2001; Al Dulayymi et al., 2003, 2007; Koza and Baird, 2007).
The acetylated synthetic alpha-mycolic acid methyl ester
(␣-AMAME, 3) was kindly provided by Dr. J. Al Dulayymi,
University of Bangor, UK (Al Dulayymi et al., 2003, 2005).
Cholesterol (4), isoniazid, AmB, NaBH4, terephthalalde-
hyde and phosphatidylcholine were purchased from Fluka or
Sigma–Aldrich. All organic solvents were purchased from
Merck while the inorganic materials were from Sigma.
Molecular weight 1159.35 g mol−1 (calculated), melting
point not determined due to decomposition of compound at
20
about 40 ◦C, [␣]D (c 2.0, DMF), IR νmax 3391, (NH, OH’s),
1
1564 (double bonds) cm−1. H NMR (500 MHz, (CD3)2 SO)
δ 12.06 (1H, s, H–NH), 8.79 (2H, d, J = 3.3 Hz, H-10a), 8.46
(1H, s, H-6a), 7.83 (2H, d, J = 4.4 Hz, H-6a), 7.51 (2H, d,
J = 7.9 Hz, H-4a), 7.7 (2H, d, J = 7.9 Hz, H-3a), 6.45–6.07 (m,
12H, olefinic), 5.95 (1H, dd, J = 13, 5.3 Hz, H-20), 5.43 (1H, dd,
J = 11.3, 12 Hz, H-33), 5.21 (1H, m, H-37), 4.48 (1H, broad-s,
H-1ꢀ), 4.4 (1H, m, H-19), 4.24 (1H, m, H-11), 4.23 (1H, t, H-17),
4.06 (1H, m, H-3), 3.99 (1H, dt, H-15), 3.89 and 3.67 (2H, s,
H-1a), 3.73 (1H, d, H-2ꢀ), 3.54–3.46 (HOD plus 3H), 3.2 (1H, m,
H-5ꢀ), 3.09 (2H, m, H-4ꢀ, H-35), 2.82 (1H, m, H-3ꢀ), 2.28 (1H, s,
H-34), 2.16 (3H, m, H-2, H-18), 1.88–1.05 (14H, m, CH2, CH),
1.87 (1H, t, J = 6.7 Hz, H-16), 1.16 (3H, d, J = 5.5 Hz, CH3), 1.11
(3H, d, J = 5.6 Hz, CH3), 1.04 (3H, d, J = 5.6 Hz, CH3), 0.91
(3H, d, J = 6.5 Hz, CH3), 13C NMR (300 MHz, (CD3)2 SO) δ
174.4, 170.6, 161.54, 150.3, 149.1, 143.6, 140.51, 133.9, 136.8,
133.7–128.5, 131.9, 127.2, 121.5, 97.2, 77.1, 73.8, 74.5, 74.4,
69.8, 73.6, 69.4, 68.8, 66.2, 65.4, 65.2, 63, 57.1, 44.7, 49.5–44.3,
42.3, 42, 35.1, 29, 18.5, 12.1 ppm.
2.3. Synthesis of Amphotericin B derivative
2.4. ELISA
2.3.1. Preparation of
Nꢀ-(4-formylbenzylidene)isonicotinohydrazide (6)
Isoniazid (5, 0.5 g, 3.65 mmol) was added in portions over an
hour to a solution of terephthalaldehyde (0.49 g, 3.65 mmol) in
ethanol (50 ml) and stirred at room temperature (RT) overnight.
The human sera used in the ELISA experiments were from
two sources: the first was from a pulmonary TB positive col-
lection that was made in 1994 by the MRC Clinical and
Biomedical TB Research Unit at King George V Hospital, Dur-

98
Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
ban, KwaZulu-Natal and donated by Dr. P.B. Fourie. The second
source was from a collection made in the year 2000 from patients
for another study by Schleicher et al. (2002). From the latter, five
serumsampleswereselectedfrompatientswhoweredetermined
to be negative for both tuberculosis and HIV (TB−). These TB
negativepatientswerehospitalisedforvariousreasonsotherthan
TB or AIDS.
Mycolic acids (MA, 1), acetylated synthetic ␣-mycolic acid
methyl ester (␣-AMAME, 3) and cholesterol were used at final
concentrations of 60 ␮g/ml. To prepare the coating solutions,
the antigens were heated in PBS buffer for 20 min at 85 ◦C.
The hot solutions were sonicated and kept at 85 ◦C during load-
ing into the ELISA plates at 3 ␮g antigen per well. The assay
with sera was done according to Schleicher et al. (2002). The
results obtained were analysed for statistical differences using
the Student’s t-test.
2.5. Preparation of liposomes
Fig. 2. Biosensor activity sequence to obtain sensorgrams of the binding of
cholesterol from cholesterol-containing liposomes onto various immobilized
lipid-containing liposomes. 1: activation of surface with CPC, 2: PBS/AE wash
step, 3: addition of liposomes for coating, 4: PBS/AE wash step, 5: addition
of the test solution, 6: PBS/AE wash step. The cholesterol binding sensorgram
appears in the rectangle.
Empty liposomes were prepared consisting of only phos-
phatidyl choline (PC), whereas the other liposomes consisted
of PC in some ratio to the lipid under investigation. Mycolic
acids (MA, 1) liposomes and other chemically prepared mycolic
acid liposomes contained MA or MAME (2) or ␣-AMAME
(3) (7.8 mol%; mycolic acid: PC = 1 mg: 9 mg), whereas the
cholesterol (Chol, 4) liposomes contained Chol (50 mol%,
Chol:PC = 3 mg:6 mg). The appropriate lipids were initially dis-
solved in chloroform in an amber glass vial and vortexed to
ensure mixing. The samples were then dried at 80 ◦C under a
stream of N2 and sonicated in 2 ml saline for 5 min at room tem-
perature. Sonication was done using a Branson sonifier (model
B30, Branson Sonicpower Co. USA) with 30% duty cycles and
outputof5. Subsequentlytheliposomesweredividedinto200 ␮l
aliquots, freeze-dried and stored at −70 ◦C until required for
use. Before use, the liposomes were reconstituted with 2 ml of
phosphate buffered saline (PBS/AE) pH 7.4 containing EDTA
(1 mM), sodium azide (0.025%, m/v), heated at 80 ◦C for 15 min
and then sonified as above. The final liposome concentration
came to 500 ␮g/ml. The liposome suspensions were analysed
for their mycolic acid content by HPLC according to Goodrum
et al. (2001). Recovery of mycolic acids ranged between 70%
and 110% of the expected values.
pipetted into each cell of the cuvette to obtain a stable base-
line for 1 min. The PBS/AE was subsequently aspirated and the
surface activated with 50 ␮l of cetyl-pyridinium chloride (CPC,
0.02 mg/ml in PBS/AE) for 10 min. This was followed by wash-
ing 5 times with 60 ␮l PBS/AE and then substituting with 25 ␮l
PBS/AE for a new baseline before immobilization of mycolic
acids (or cholesterol, or a synthetic ␣-AMAME, or PC only)
containing liposomes to the surface for 20 min. The immobilized
liposomes were then finally washed 5 times with 60 ␮l PBS/AE.
The cuvette cell content was substituted with 25 ␮l of PBS/AE
and allowed to equilibrate for 5–10 min to achieve a stable base-
line. An addition of either 25 ␮l liposomes under investigation or
25 ␮l of a solution of AmB (1 × 10−4 M) or the AmB derivative
(7, Scheme 1, 1 × 10−4 M) in PBS/AE was made. Direct interac-
tion between the immobilized cholesterol/mycolic acids and the
dissolved Amphotericin B compounds or lipids contained in the
suspended liposomes was monitored for 5–10 min, after which
the cuvette was washed 3 times with 60 ␮l PBS/AE. Finally,
regeneration of the cuvette was effected by 5 times washing with
50 ␮l 95% ethanol for 1 min, followed by 7 times washing with
70 ␮l PBS/AE for 1 min. The surface was then finally treated
for 5 times with 50 ␮l potassium hydroxide (12.5 M) for 2 min,
followed by 7 times washing with 70 ␮l PBS/AE for 1 min. The
results obtained were analysed for statistical differences using
the Student’s t-test.
2.6. Measurements of interaction between mycolic acids,
cholesterol and Amphotericin B derivatives on an IAsys
affinity biosensor
ThebindinginteractionsamongAmphotericinB, cholesterol,
mycolic acids and acetylated synthetic ␣-mycolic acid methyl
ester were measured by means of an IAsys resonant mirror
biosensor (IAsys Affinity Sensors, Bar Hill, Cambridge, UK)
according to the activity sequence outlined in Fig. 2. IAsys soft-
ware was used to set the device at a data-sampling interval of
0.4 s, temperature of 25 ◦C and stirring rate of 75% for all exper-
iments on the biosensor. The cells of the cuvette were rinsed
with ethanol (95%) 3 times prior to use, followed by exten-
sive washing with PBS/AE. A 60-␮l volume of PBS/AE was
2.7. Measurements of direct interaction between mycolic
acids and cholesterol
After immobilization of either the mixture of natural mycolic
acids (1), their methyl esters (2), or protected acetylated syn-
thetic ␣-mycolic acid methyl ester (3) containing liposomes
using the same method as described above and illustrated in

Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
99
Scheme 1. Preparation of an AmB derivative: (i) terephthalaldehyde, (ii) AmB and (iii) NaBH4.
Fig. 2, 25 ␮l cholesterol-containing liposomes were added.
Direct interaction between the immobilized mycolic acids and
cholesterol was monitored for 10 min, after which the cuvette
was washed 3 times with 60 ␮l PBS/AE and regenerated as
before.
without alleviating the hydrophobic nature of the molecule that
can be expected to be the major force of non-specific bind-
ing from serum components. Moreover, the protection of the
polar groups in the mycolic motif also discourages any folded
arrangement with the merochain (Villeneuve et al., 2005, 2007).
Therefore, this change in the three-dimensional structure may
be hypothesised to preclude specific interactions with antibodies
against mycolic acids or cholesterol.
3. Results and discussion
Coating of the wells with the lipid antigens was confirmed
under the microscope as visible fatty deposits adsorbed on the
polystyrene. Eleven TB positive patient sera were randomly cho-
sen from the 1994 collection and five TB and HIV negative
patients were randomly selected from the 2000 collection. For
the TB negative controls, it was deemed important to exclude
HIV positive individuals, for whom false negative TB diagnosis
is known to occur at high frequency with the currently available
TB diagnostics (Uma Devi et al., 2003). The sera were diluted
1:20, which was found to be the highest dilution where signifi-
cant binding to the antigens on the plate could still be observed
with most sera.
As expected there was a tendency for TB positive patient
antibodies to bind more strongly to mycolic acid than those of
the TB negative patients (average absorbance value 0.76 com-
pared to 0.37 respectively, Fig. 3, P < 0.1). The selectivity of
binding with MAs was confirmed by the negligible antibody
activity of the same patients to the synthetic alpha-mycolic acid
in both groups of sera (Fig. 3, P < 0.0001). Both TB positive
and TB negative sera also recognised cholesterol as an antigen.
In either TB positive or TB negative patients, the difference in
ELISA signal between mycolic acid and cholesterol was not sig-
nificant (Fig. 3, P > 0.25). Although TB negative patients should
not present antibodies against M. tuberculosis mycolic acids, the
results obtained show that their sera recognise in a similar fash-
3.1. Interactions of patient sera to mycolic acids and
cholesterol
It is known that all humans have anti-cholesterol antibod-
ies (ACHA), which have been proven to be very specific for
their interaction with the sterol. ACHA recognise selectively
3␤-hydroxy-sterols in a stereospecific manner, but they can-
´ ´
not distinguish between enantiomers (Geva et al., 2001; Bıro et
al., 2007). Concurrently, tuberculosis patients have been shown
to produce antibodies against mycolic acids (Schleicher et al.,
2002). Pan et al. (1999) reported that even small changes in the
structure of the mycolic acids present in TDM are important
for their antigenicity. In particular, among the three subclasses
present in M. tuberculosis, methoxymycolic acids were shown to
be the most antigenic. For this reason, we expected TB negative
patient sera to react to cholesterol, but not to mycolic acids.
TB positive and TB negative sera were tested in ELISA for
antibody binding to natural mycolic acid and cholesterol. To ver-
ify the specificity of the interaction between sera and mycolic
acids, a synthetic ␣-mycolic acid (3), with the carboxylic acid
protected as methyl ester and the hydroxyl group as acetyl
ester, was used as negative control. This compound was cho-
sen because it lacks a hydrogen donor at either the ␣-carboxylic
acid, the ␤-hydroxyl group or in the meromycolate chain, which
diminishes the possibility of hydrogen bonds with antibodies

100
Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
Fig. 3. Normalised ELISA results of TB positive (n = 11) and TB negative sera
(n = 5) against natural mycolic acids (1) (MA, grey), cholesterol (4) (Chol,
hatched) and an acetylated, synthetic ␣-mycolic acid methyl ester (3) (␣-
AMAME, black). The error bars indicate the standard error of the mean.
ion either cholesterol or mycolic acids. Such congruence of the
titresofbothantibodyspecificitiescouldsuggestcross-reactivity
of recognition of cholesterol and mycolic acid by antibodies.
3.2. AmB cross-reactivity to mycolic acids and cholesterol
To further probe the possibility of resemblance in the nature
of mycolic acid and cholesterol to bind to a ligand, an alterna-
tive experiment was designed using Amphotericin B (AmB) as
a cholesterol binding agent. AmB is a macrolide antibiotic that
exerts its antifungal activity by binding to ergosterol and choles-
terol (Baginski et al., 2002). The major interaction between the
macrolide and the sterol is based on the formation of a hydrogen
bond between the amine or hydroxyl groups of the mycosamine
moiety of AmB and the hydroxyl group of ergosterol (Herve
et al., 1989; Matsumori et al., 2005). Other sterols, in particu-
lar cholesterol, are also recognised and bound by AmB, albeit
with one order of magnitude weaker affinity. This is due to
the more rigid hydrophobic sterol surface of ergosterol (which
has two extra double bonds and a methyl group compared to
that of cholesterol), making the Van der Waals interaction with
the hydrophobic part of AmB stronger (Readio and Bittman,
1982; Baginski et al., 2002). The difference in binding affinity
of AmB to ergosterol and cholesterol proves the specificity of
the interaction between AmB and its sterol ligands. According
to the approach of Prendergast et al. (1998), the specific molec-
ular recognition of two substances by an established binding
agent of one of them indicates a resemblance in the three-
dimensional structure of the two compounds. For this reason,
AmB was selected to assess its binding activity to cholesterol
and to mycolic acid using resonant mirror biosensor technol-
ogy. This technique was preferred to ELISA because it does not
need a fluorescent tag on the binding agent that may affect the
structure and interaction properties of AmB (Cush et al., 1993).
Moreover using this technique it is possible to use liposomes
which better mimic the biological environment within which
these lipids are presented (MacKenzie et al., 1997).
Fig. 4. Resonant mirror biosensor binding curves (A) and binding capacity (B)
of AmB or AmB derivative (7) on immobilized lipid antigens. Binding of AmB
on (a) mycolic acids (1) (thick line, +, hatched bar); (b) cholesterol (thin line,
o, white bar); (c) synthetic protected ␣-MA (3) (dotted line, *, grey bar) or (d)
PC liposomes (black bar). Binding of AmB derivative (7) on immobilized (e)
cholesterol liposomes (dashed line, -, horizontal stripes bar) or (f) mycolic acid
(1) containing liposomes (grid bar); n = 5 for each set. Immobilization of lipid
ligate containing liposomes was monitored to achieve binding of at least 2000
arc seconds before contacting with AmB or AmB derivative (7) solutions.
frequently employed cationic, amphipathic compound that gives
glazedandmetalsurfacesahydrophobiccharacter. Hydrophobic
lipidantigen-containingliposomescanthenadheretothesurface
(see Fig. 2). Following the coating of the cuvette cells with either
cholesterol-containing or empty liposomes, consisting only of
phosphatidylcholine, the biosensor registered a direct accumu-
lation of AmB on cholesterol liposomes, while it did not show
any binding to empty liposomes (Fig. 4). Remarkably, a binding
sensorgram comparable to that of immobilized cholesterol was
obtained when AmB was allowed to interact with immobilized
mycolic acids.
The binding between the AmB macrolide and cholesterol
could be abrogated by covalent modification of the single amino
group on the AmB macrolide by linking it to isoniazid to give
Amphotericin B-derivative (7). The mycolic acid binding was
similarly abrogated by the covalent modification of the AmB
macrolide amino group. This modification to the amino group
of the AmB molecule destroyed the latter’s ability to bind to both
cholesterol and the mycolic acid, showing that the attraction of
AmB for both these lipids is determined by the fine structure of
the ligand–receptor pair and possibly that this requires a specific
hydrogen bond between the binding partners. Such an analysis
would be in agreement with the literature, confirming the impor-
In order to immobilize the liposomes onto the surface of the
biosensor, a new procedure was used. This is based on the activa-
tion of the cuvette cell surface with cetyl-pyridinium chloride, a

Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
101
tance of the hydrogen bond for a stable interaction (Baginski et
al., 2002). Matsumori et al. (2005) demonstrated that AmB’s
binding to sterols could be hindered when its amino-sugar
moiety was forced in a conformation that prevented hydro-
gen bonding with its ligate. This suggests that the interaction
between AmB and mycolic acids could also be based on the
formation of a hydrogen bond with the amino-sugar moiety of
AmB. It is noteworthy that AmB did not bind to the synthetic
protected alpha-mycolic acid (3) immobilized on the sensor sur-
face (Fig. 4), thus supporting the specificity of this interaction
in terms of the requirement for both Van der Waal’s and hydro-
gen bonding in the recognition event. These results imply that
mycolic acid and cholesterol share structural features that are
similarly recognised by AmB and that may provide the basis for
the cross-reactivity observed with the TB patient antibodies.
Fig. 5. Resonant mirror biosensor binding curves for the attraction of cholesterol
to immobilized mycolic acids (1) (MA, thick line/+), their methyl esters (2)
(mMA, dashed line/*) or synthetic protected alpha-mycolic acid (3) (␣-MA,
thin line/-). Each line represents a typical curve of five repeats with the end
points of each indicated after 7.5 min of contact.
3.3. Interaction between mycolic acids and cholesterol
In order to determine whether the presumed structural relat-
edness between mycolic acid and cholesterol would also lead to
their interaction directly, the interaction between mycolic acids
and cholesterol liposomes was analysed using the biosensor
technique.
mycolic acid. The formation of a hydrogen bond between the
carboxylic and the ␤-hydroxyl of mycolic acid is particularly
favoured for the natural erythro configuration of 2R,3R 2-alkyl-
3-hydroxy acids and has been shown to have a stabilising effect
on the alignment of the alkyl chains, affecting the physical prop-
erties of these acids (Durand et al., 1979a,b). Therefore, the use
of methyl esters of mycolic acids as a negative control is partic-
ularly important, as this modification will destroy the specific
hydrogen bond in the mycolic motif that is expected to desta-
bilise the natural conformations of the alkyl chains to assume a
sterol fold, similar to that assumed by the steroid-precursor, the
linear 2,3-oxidosqualene, before its conversion into lanosterol
in cholesterol biosynthesis (Van Tamelen, 1982).
Fig. 5 shows that the mycolic acid liposomes-coated cuvette
surface accumulated cholesterol from the solution while the
methyl esters of mycolic acid (2) and the synthetic, acetylated
alpha-mycolic acid methyl ester (3) were unable to do so (no
significant difference between binding of cholesterol to (2) and
(3), P > 0.1, but highly significant difference of cholesterol bind-
Following coating of the cuvette surface with either mycolic
acid (1), mycolic acid methyl ester (2) or synthetic acetylated
␣-mycolic acid methyl ester (3), the cuvettes were exposed
to cholesterol-containing test liposomes and the sensorgrams
for the binding of cholesterol onto the different surfaces were
recorded (Fig. 5). A mixture of methyl esters of natural mycolic
acids (MAME), still containing hydrogen binding groups (␤-
hydroxylandtheoxygenatedgroupsinthemeromycolatechain),
was also used as a negative control for this test. The experiments
described earlier compared the interactions of mycolic acids and
cholesterol with binding agents known to be able to differentiate
between small conformational changes. The present experiment
measured the attraction between two lipids and therefore it was
important to verify that this is not just due to Van der Waals
interactions between two hydrophobic molecules but also to a
more specific conformational match between cholesterol and
Fig. 6. A pictorial representation of a hypothesised mechanism of interaction between a “W”-folded methoxy MA (Villeneuve et al., 2005, 2007) and cholesterol.
The absolute stereochemistry of the cyclopropane and the methoxy group are not completely clarified. Here for simplicity the S,R stereochemistry has been used for
the cyclopropyl group and the SS for the methoxy-group. RR is known to be the absolute configuration of the mycolic motif. The “W” fold is drawn here in a flat
plane, whereas in practice the fold shown in earlier models had a cylindrical three-dimensional structure (Villeneuve et al., 2005, 2007).

102
Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
ing to mycolic acid or either the methyl esters of mycolic acid
(2) and the synthetic, acetylated alpha-mycolic acid methyl ester
(3), P < 0.001). These results demonstrate a pronounced attrac-
tion between cholesterol and mycolic acid that is determined
by the presence of a free carboxylic acid group and the degree
of structural fit by a particular conformation that free mycolic
acids assume, which can be eliminated by formation of a sim-
ple methyl ester of the acid. In Fig. 6 a graphical representation
is attempted of a mycolic acid, folded to assume a cholesteroid
nature. A more accurate representation can be expected from vir-
tual models based on Langmuir–Blodgett analysis of synthetic
subclasses of mycolic acids interacting with cholesterol. The
current advances in the stereo-controlled chemical synthesis of
mycolic acids may make this possible in the near future.
administered intratracheally, mycolic acid could also prevent
experimentally induced asthma in mice (Korf et al., 2006). The
ability of mycolic acid to attract cholesterol to the macrophage
may be critical for these phenomena.
Third, cholesterol has been shown to play a role in the entry
and survival of M. tuberculosis in the host macrophage (Gatfield
and Pieters, 2000; de Chastellier and Thilo, 2006). The choles-
teroid nature of at least some mycolic acids may imply their
active participation in this manifestation of virulence.
The different subclasses of mycolic acids from M. tuberculo-
sis are currently under investigation to learn if the cholesteroid
nature of mycolic acid is due only to the mycolic motif or to
the other functional groups in the meromycolate chain. Finally,
an improved understanding of the nature of the mycolate coat
of M. tuberculosis could lead to the development of improved
diagnostics and treatments of tuberculosis.
4. Conclusion
The following observations support the hypothesis of a sim-
ilarity in nature between the mycolic acids of M. tuberculosis
and cholesterol that can affect the accuracy of serodiagnostic
assays based on antibodies to mycolic acids as surrogate markers
of active TB: (i) both mycolic acid and cholesterol are recog-
nised by AmB and human TB sera, (ii) both human TB sera and
AmB are able to distinguish between natural occurring mycolic
acid (1) and a closely related chemically synthetic structure, i.e.
acetylated alpha-mycolic acid methyl ester (3), (iii) TB nega-
Acknowledgements
We thank Mr. Eric Palmer of the Department of Chemistry,
University of Pretoria for performing the NMR analyses. We
also thank Dr. Elardus Erasmus from School of Biochemistry,
University of the Northwest, SA for performing the Electron-
spray mass spectrometry. This work was supported with grants
from NRF, MRC, UP and Claude Leon Harris Foundation.
´ ´
tive patient sera containing antibodies to cholesterol (Bıro et al.,
Appendix A. Supplementary data
2007) bind to mycolic acid comparatively well, but not to the
acetylated alpha-mycolic acid methyl ester (3) and (iv) the asso-
ciation of AmB to both cholesterol and mycolic acid could be
disrupted by covalent modification of the AmB that prevented
the formation of hydrogen bonds but not Van der Waals associa-
tion. Likewise, modification ofmycolicacid, aimingto disfavour
its arranging into the “W” or any other folded conformation and
preclude hydrogen bonding with ligate, abrogated its interaction
with patient sera, AmB and cholesterol. The specific attraction
between mycolic acid and cholesterol in liposomes probably
depends on a particular conformation that free MAs assume in
the phospholipid bilayer of liposomes, and this will probably
also apply to biological membranes. The existence of a con-
densed conformation for MAs has already been proposed by
several groups, based on Langmuir studies. Here we present the
first evidence that these folded conformations could possibly
exist in biological conditions to determine the antigenicity of
mycolic acids.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.chemphyslip.
2008.01.004.
References
Al Dulayymi, J.R., Baird, M.S., Roberts, E., 2005. The synthesis of a single
enantiomer of a major ␣-mycolic acid of M. tuberculosis. Tetrahedron 61
(50), 11939–11951.
Al Dulayymi, J.R., Baird, M.S., Roberts, E., Deysel, M., Verschoor, J., 2007.
The first syntheses of single enantiomers of the major methoxymycolic acid
of Mycobacterium tuberculosis. Tetrahedron 63 (12), 2571–2592.
Al Dulayymi, J.R., Baird, M.S., Roberts, E., 2003. The synthesis of single
enantiomers of a major alpha-mycolic acid of Mycobacterium tuberculosis.
Chem. Commun. 21 (2), 228–229.
Athanassopoulou, N., Davies, R.J., Edwards, P.R., Yeung, D., Maule, C.H., 1999.
CholeratoxinandGM1:amodelmembranestudywithIAsys. Biochem. Soc.
Trans. 27 (2), 340–343.
The implications of such a discovery and of the relatedness
between cholesterol and mycolic acids may be significant for
several reasons.
First it could confirm the explanation of the poor results
obtained with ELISA testing the usefulness of antibodies to
mycolic acids as surrogate markers for tuberculosis (Schleicher
et al., 2002).
Second, it impacts on the observation that mycolic acids,
intraperitoneally or intratracheally administered as liposomes in
mice, resulted in a pathogen-associated type of innate immune
response that mainly affects macrophages, converting them into
cholesterol-rich foam cells (Korf et al., 2005). Moreover, when
Baginski, M., Resat, H., Borowski, E., 2002. Comparative molecular dynamics
simulations of amphotericin B-cholesterol/ergosterol membrane channels.
Biochim. Biophys. Acta 1567, 63–78.
Baginski, M., Sternal, K., Czub, J., Borowski, E., 2005. Molecular modelling
of membrane activity of amphotericin B, a polyene macrolide antifungal
antibiotic. Acta Biochim. Pol. 52 (3), 655–658.
B´ıro´, A., Cervenak, L., Balogh, A., Lorincz, A., Uray, K., Horva´th, A., Romics,
L., Matko´, J., Fu¨st, G., Laszlo´, G., 2007. Novel anti-cholesterol mono-
clonal immunoglobulin G antibodies as probes and potential modulators
of membrane raft-dependent immune functions. J. Lipid Res. 48 (1), 19–29.
Buckle, P.E., Davies, R.J., Kinning, T., Yeung, D., Edwards, P.R., Pollard-
Knight, D., Lowe, C.R., 1993. The resonant mirror: a novel optical sensor for
direct sensing of biomolecular interactions. Part II. Applications. Biosens.
Bioelectron. 8 (7/8), 355–363.

Y. Benadie et al. / Chemistry and Physics of Lipids 152 (2008) 95–103
103
Cush, R., Cronin, J.M., Stewart, W.J., Maule, C.H., Molloy, J., Goddard, N.J.,
1993. The resonant mirror: a novel optical biosensor for direct sensing
of biomolecular interactions. Part I. Principle of operation and associated
instrumentation. Biosens. Bioelectron. 8 (7/8), 347–354.
de Chastellier, C., Thilo, L., 2006. Cholesterol depletion in Mycobacterium
avium-infected macrophages overcomes the block in phagosome matu-
ration and leads to the reversible sequestration of viable mycobacteria
in phagolysosome-derived autophagic vacuoles. Cell. Microbiol. 8 (2),
242–256.
Koza, G., Baird, M.S., 2007. The first synthesis of single enantiomers of keto-
mycolic acids. Tetrahedron Lett. 48 (12), 2165–2169.
MacKenzie, C.R., Hirama, T., Lee, K.K., Altman, E., Young, N.M., 1997.
Quantitative analysis of bacterial toxin affinity and specificity for glycolipid
receptors by surface plasmon resonance. J. Biol. Chem. 272 (9), 5533–5538.
Matsumori, N., Sawada, Y., Murata, M., 2005. Mycosamine orientation of
amphotericin B controlling interaction with ergosterol: sterol-dependent
activity of conformation-restricted derivatives with an amino-carbonyl
bridge. J. Am. Chem. Soc. (127), 10667–10675.
Dobson, G., Minnikin, D.E., Minnikin, S.M., Parlett, J.H., Goodfellow, M.,
Ridell, M., Magnusson, M., 1985. Systematic analysis of complex mycobac-
terial lipids. In: Goodfellow, M., Minnikin, D.E. (Eds.), Chemical Methods
of Bacterial Systematics. Academic Press, London, pp. 237–265.
Durand, E., Welby, M., Laneelle, G., Tocanne, J.F., 1979a. Phase behaviour of
cord factor and related bacterial glycolipid toxins A monolayer study. Eur.
J. Biochem. 93 (1), 103–112.
Pan, J., Fujiwara, N., Oka, S., Maekura, R., Ogura, T., Yano, I., 1999. Anti-
cord factor (trehalose 6,6’-dimycolate) IgG antibody in tuberculosis patients
recognizes mycolic acid subclasses. Microbiol. Immunol. 43 (9), 863–869.
Prendergast, M.M., Lastovica, A.J., Moran, A.P., 1998. Lipopolysaccharides
from Campylobacter jejuni O:41 strains associated with Guillain-Barre´
syndrome exhibit mimicry of GM1 ganglioside. Infect. Immun. 66 (8),
3649–3755.
Durand, E., Gillois, M., Tocanne, J.F., Laneelle, G., 1979b. Property and activity
of mycoloyl esters of methyl glucoside and trehalose. Effect on mitochon-
drial oxidative phosphorylation related to organization of suspensions and
to acyl-chain structures. Eur. J. Biochem. 94 (1), 109–118.
Fujiwara, N., Pan, J., Enomoto, K., Terano, Y., Honda, T., Yano, I., 1999.
Production and partial characterization of anti-cord factor (trehalose-6,6ꢀ-
dimycolate) IgG antibody in rabbits recognizing mycolic acid subclasses
of Mycobacterium tuberculosis or Mycobacterium avium. FEMS Immunol.
Med. Microbiol. 24 (2), 141–149.
Readio, J.D., Bittman, R., 1982. Equilibrium binding of amphotericin B and its
methyl ester and borate complex to sterols. Biochim. Biophys. Acta 685 (2),
219–224.
Ryll, R., Kumazawa, Y., Yano, I., 2001. Immunological properties of trehalose
dimycolate (cord factor) and other mycolic acid-containing glycolipids—a
review. Microbiol. Immunol. 45 (12), 801–811.
Schleicher, G.K., Feldman, C., Vermaak, Y., Verschoor, J.A., 2002. Prevalence
of anti-mycolic acid antibodies in patients with pulmonary tuberculosis co-
infected with HIV. Clin. Chem. Lab. Med. 40 (9), 882–887.
Gatfield, J., Pieters, J., 2000. Essential role for cholesterol in entry of mycobac-
teria into macrophages. Science 288 (5471), 1647–1650.
Sekanka, G., Baird, M., Minnikin, D., Grooten, J., 2007. Mycolic acids for the
control of tuberculosis. Expert Opin. Ther. Patents 17, 315–331.
Steck, P.A., Schwartz, B.A., Rosendahl, M.S., Gray, G.R., 1978. Mycolic acids.
A reinvestigation. J. Biol. Chem. 253 (16), 5625–5629.
Uma Devi, K.R., Ramalingam, B., Raja, A., 2003. Antibody response to
Mycobacterium tuberculosis 30 and 16kDa antigens in pulmonary tuber-
culosis with human immunodeficiency virus coinfection. Diagn. Microbiol.
Infect. Dis. 46 (3), 205–209.
Van Tamelen, E.E., 1982. Bioorganic characterization and mechanism of the 2,3-
oxidosqualene-lanosterol conversion. J. Am. Chem. Soc. 104, 6480–6482.
Villeneuve, M., Kawaia, M., Kanashima, H., Watanabe, M., Minnikin, D.E.,
Nakahara, H., 2005. Temperature dependence of the Langmuir monolayer
packing of mycolic acids from Mycobacterium tuberculosis. Biochim. Bio-
phys. Acta 1715 (2), 71–80.
Villeneuve, M., Kawai, M., Watanabe, M., Aoyagi, Y., Hitotsuyanagi, Y.,
Takeya, K., Gouda, H., Hirono, S., Minnikin, D.E., Nakahara, H., 2007.
Conformational behavior of oxygenated mycobacterial mycolic acids from
Mycobacterium bovis BCG. Biochim. Biophys. Acta 1768 (7), 1717–1726.
Watanabe, M., Aoyagi, Y., Ridell, M., Minnikin, D.E., 2001. Separation and
characterization of individual mycolic acids in representative mycobacteria.
Microbiology 147, 1825–1837.
Geva, M., Izhaky, D., Mickus, D.E., Rychnovsky, S.D., Addadi, L., 2001. Stere-
oselective recognition of monolayers of cholesterol, ent-cholesterol, and
epicholesterol by an antibody. Chem. Biochem. 2 (4), 265–271.
Goodrum, M.A., Siko, D.G., Niehues, T., Eichelbauer, D., Verschoor, J.A., 2001.
Mycolic acids from Mycobacterium tuberculosis: purification by countercur-
rent distribution and T-cell stimulation. Microbios 106 (413), 55–67.
Herve, M., Debouzy, J.C., Borowski, E., Cybulska, B., Gary-Bobo, C.M., 1989.
The role of the carboxyl and amino groups of polyene macrolides in their
interactions with sterols and their selective toxicity. A 31P-NMR study.
Biochim. Biophys. Acta 980 (3), 261–272.
Korf, J., Stolz, A., Verschoor, J.A., De Baetselier, P., Grooten, J., 2005.
The Mycobacterium tuberculosis cell wall component mycolic acid elic-
its pathogen-associated host innate immune responses. Eur. J. Immunol. 35
(3), 890–900.
Korf, J.E., Pynaert, G., Tournoy, K., Boonefaes, T., Van Oosterhout, A., Gin-
neberge, D., Haegeman, A., Verschoor, J.A., De Baetselier, P., Grooten, J.,
2006. Macrophage reprogramming by mycolic acid promotes a tolerogenic
response in experimental asthma. Am. J. Respir. Crit. Care Med. 174 (2),
152–160.