Carbohydrate-lipid interactions: Affinities of methylmannose polysaccharides for lipids in aqueous solution

Article abstract of DOI:10.1002/chem.201201222

The interactions between 3-O-methyl-mannose polysaccharides (MMPs), extracted from Mycobacterium smegmatis (consisting of a mixture of MMP-10, -11, -12 and -13) or obtained by chemical synthesis (MMP-5_s, -8 _s, -11_s and -14_s), and linear saturated and unsaturated fatty acids (FAs), and a commercial mixture of naphthenic acids (NAs) in aqueous solution at 25°C and pHa 8.5 were quantified by electrospray ionization mass spectrometry (ESI-MS). Association constants (K_a) for MMP binding to four FAs (myristic acid, palmitic acid, stearic acid and trans-parinaric acid) were measured by using an indirect ESI-MS assay, the proxy protein method. The K_a values are in the 10⁴-10⁵ M^-1 range and, based on results obtained for the binding of the synthetic MMPs with palmitic acid, increase with the size of the carbohydrate. Notably, the measured affinity of the extracted MMPs for trans-parinaric acid is two orders of magnitude smaller than the reported value, which was determined by using a fluorescence assay. Using a newly developed competitive binding assay, referred to as the proxy protein/proxy ligand ESI-MS method, it was shown that MMPs bind specifically to NAs in aqueous solution, with apparent affinities of approximately (5×10⁴) M^-1 for the mixture of NAs tested. This represents the first demonstration that MMPs can bind to hydrophobic species more complex than those containing linear alkyl/alkenyl chains. Moreover, the approach developed here represents a novel method for probing carbohydrate-lipid interactions. Copyright

Full text of DOI:10.1002/chem.201201222

FULL PAPER
DOI: 10.1002/chem.201201222
Carbohydrate–Lipid Interactions: Affinities of Methylmannose
Polysaccharides for Lipids in Aqueous Solution
Lan Liu, Yu Bai, Nian Sun, Li Xia, Todd L. Lowary,* and John S. Klassen*[a]
Abstract: The interactions between 3-
O-methyl-mannose polysaccharides
measured by using an indirect ESI-MS
assay, the “proxy protein” method. The
Ka values are in the 104–105 mꢀ1 range
and, based on results obtained for the
binding of the synthetic MMPs with
palmitic acid, increase with the size of
the carbohydrate. Notably, the meas-
ured affinity of the extracted MMPs
for trans-parinaric acid is two orders of
magnitude smaller than the reported
value, which was determined by using a
fluorescence assay. Using a newly de-
veloped competitive binding assay, re-
ferred to as the “proxy protein/proxy
ligand” ESI-MS method, it was shown
that MMPs bind specifically to NAs in
aqueous solution, with apparent affini-
ties of approximately (5ꢀ104)mꢀ1 for
the mixture of NAs tested. This repre-
sents the first demonstration that
MMPs can bind to hydrophobic species
more complex than those containing
linear alkyl/alkenyl chains. Moreover,
the approach developed here repre-
sents a novel method for probing car-
bohydrate–lipid interactions.
(MMPs), extracted from Mycobacteri-
um smegmatis (consisting of a mixture
of MMP-10, -11, -12 and -13) or ob-
tained by chemical synthesis (MMP-5s,
-8s, -11s and -14s), and linear saturated
and unsaturated fatty acids (FAs), and
a commercial mixture of naphthenic
acids (NAs) in aqueous solution at
258C and pH 8.5 were quantified by
electrospray ionization mass spectrom-
etry (ESI-MS). Association constants
(Ka) for MMP binding to four FAs
(myristic acid, palmitic acid, stearic
acid and trans-parinaric acid) were
Keywords: carbohydrates · electro-
spray ionization mass spectrometry ·
lipids
·
naphthenic acids
·
polysaccharides
Introduction
containing lipids, for example, long-chain linear fatty acids
(FAs), MMPs have been suggested to adopt a helical confor-
mation in which the methyl groups are on the inside of the
helix, thus providing a hydrophobic channel capable of lipid
binding, and a hydrophilic exterior.[8] Using fluorescence
measurements, very high association constants (Ka) of (1.0ꢀ
107)mꢀ1 and (2.5ꢀ106)mꢀ1, have been reported for MMP
binding to the important biochemical intermediate palmitoyl
CoA and the polyene parinaric acid, respectively.[9,10]
The remarkably high lipid affinities reported for MMPs
raise the interesting possibility of using these polysacchar-
ides as a sorbent to remove hydrophobic contaminants from
aqueous solutions. In this regard, of particular current rele-
vance are the tailings ponds accumulating as a byproduct of
the bitumen extraction process in the Athabasca oil sands in
Northern Alberta. Tailings pond waters (TPW) are an in-
creasing environmental concern and the primary toxic com-
ponents are believed to be naphthenic acids (NAs, Figure 1),
a complex mixture of alkyl-substituted cyclic and acyclic ali-
phatic carboxylic acids.[11] These compounds have the gener-
al molecular formula, CnH2n+ZO2, where Z is 0 or a negative
even integer. There is considerable interest in remediation
of TPW, and carbohydrate sorbents, such as dimethylami-
noethyl-cellulose[12] and b-cyclodextrin,[13] have been report-
ed for this purpose. However, a distinct advantage of
MMPs, compared to these other carbohydrates, is their in-
herently high affinity for hydrophobic molecules.
Polymethylated polysaccharides (PMPs) are cytoplasmic gly-
cans produced by mycobacteria that contain 10–20 carbohy-
drate residues.[1,2] These species have been postulated to
play a role in regulating lipid metabolism in the organisms
that produce them, through the formation of complexes
with long-chain fatty acids and their activated coenzyme-A
(CoA) derivatives.[3,4] For example, PMPs have been shown
to tolerize mycobacteria to the high concentration of the
long-chain acyl-CoA derivatives needed for the synthesis of
mycolic acids, archetypal mycobacterial lipids that are essen-
tial for viability.[5] In another study, these species have been
demonstrated to regulate fatty acid chain lengths of fatty
acids produced by Type I fatty acid synthetase in Mycobacte-
rium smegmatis.[6]
One class of PMPs are 3-O-methyl-mannose polysacchar-
ides (MMPs), a family of molecules that contain 10–13 man-
nopyranose (Manp) moieties.[7] In MMPs, every Manp resi-
due, except that at the non-reducing terminus, is methylated
on O-3 (Figure 1). When present in an aqueous solution
[a] Dr. L. Liu, Y. Bai, N. Sun, L. Xia, Prof. T. L. Lowary,
Prof. J. S. Klassen
Alberta Glycomics Centre and Department of Chemistry
University of Alberta, Edmonton, Alberta, T6G 2G2 (Canada)
E-mail: tlowary@ualberta.ca
john.klassen@ualberta.ca
We describe here the results of a quantitative study of the
interactions between MMPs, either extracted from mycobac-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201201222.
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

Figure 1. Structures of: a) extracted MMP-X; X=10, 11, 12, 13; b) synthetic MMP-Xs; X=5, 8, 11, 14; c) FAs, m=12 (MA), 14 (PA), 16 (SA); d) PnA;
e) representative structures of NAs; R=alkyl group.
teria (a mixture of MMP-10, -11, -12 and -13) or pure spe-
cies produced by chemical syntheses (MMP-5s, -8s, -11s and
-14s) and linear, saturated and unsaturated FAs, and a com-
mercial mixture of NAs in aqueous solution at 258C and
pH 8.5. Both direct and indirect (competitive) electrospray
ionization mass spectrometry (ESI-MS) assays were used to
quantify the interactions between the MMPs and the FAs
and NAs. However, due to the occurrence of in-source dis-
sociation, the direct ESI-MS assay[14–17] was found to signifi-
cantly underestimate the strength of the interactions. The af-
finities for the individual MMPs and the mixture of extract-
ed MMPs for four FAs, myristic acid, palmitic acid, stearic
acid and trans-parinaric acid (Figure 1), were measured by
using the “proxy protein” ESI-MS method.[18] Notably, the
affinity measured for trans-parinaric acid is two orders of
magnitude smaller than the reported value, which was deter-
mined by using a fluorescence assay.[9] The interactions be-
tween the MMPs and the commercial mixture of NAs were
investigated by using a newly developed competitive binding
assay referred to as the “proxy protein/proxy ligand” ESI-
MS method. The results of these measurements show, for
the first time, that MMPs do bind specifically to NAs, a
complex mixture of linear and cyclic carboxylic acids, in
aqueous solution, with apparent affinities of approximately
(5ꢀ104)mꢀ1.
slightly from the natural glycans, in which the residue at the
non-reducing terminus is unmethylated. Using an iterative
approach, a homologous series of MMP derivatives contain-
ing 5–20 monosaccharide residues was prepared.[19] In the in-
itial report on the synthesis of these compounds,[19a] the gly-
cosyl donor contained a non-participating protecting group,
which, given the required 1,2-trans stereochemistry in the
products, necessitated careful control of reaction conditions
to ensure good glycosylation stereoselectivities. Purification
of the compounds proved problematic in some cases and, in
a subsequent synthesis,[19b] donors employing participating
acyl groups were employed thus providing significantly im-
proved stereocontrol.
Mindful of this previous work, we endeavored to develop
a synthetic approach employing donors that contain an ester
moiety adjacent to the glycosylation site, and which would
lead to products that could be deprotected in a single step
under non-hydrogenolytic conditions. The final target com-
pounds were synthesized as 8-azidooctyl glycosides to
enable their possible conjugation to, for example, solid sup-
ports or protein carriers either through reduction to the
amine and subsequent amidation, or through click chemis-
try. It was, therefore, envisioned that MMP-5s, -8s, -11s and
-14s could be synthesized starting from three building
blocks, 1–3 (Scheme 1). The strategy was built around spe-
cies containing solely acyl-based protecting groups, and with
a levulinate ester serving as the requisite temporary protect-
ing group needed to facilitate chain extension.
Results and Discussion
Monosaccharide tricholoroacetimidate imidate 2 was pre-
pared as previously reported[24] and disaccharide thioglyco-
sides 1 and 3 were synthesized as described in the Support-
ing Information (Scheme S1 in the Supporting Information).
With these precursors in hand, they were assembled into the
oligosaccharide targets. The preparation of MMP-5s and -8s
is outlined in Scheme 2. The synthesis of MMP-11s and -14s
followed similar routes and details can be found in the Sup-
porting Information (Schemes S2 and S3). To synthesize
Synthesis of MMPs: Despite the intriguing biophysical prop-
erties of PMPs, the preparation of structurally defined ana-
logues of these glycans has received little attention. Indeed,
to date, only a handful of reports have addressed the chemi-
cal synthesis of PMPs or related analogues.[19–23] With regard
to the MMPs, previous synthetic work has reported the
preparation of analogues in which all of the mannopyranose
moieties are methylated.[19] These compounds thus differ
&
2
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Carbohydrate–Lipid Interactions
FULL PAPER
Binding measurements: MMPs isolated from mycobacteria
are a homologous mixture of species with 10–13 Manp resi-
dues. Both MALDI-MS and ESI-MS were used to analyze
the MMPs, which were extracted from M. smegmatis. Shown
in Figure 2a is a representative MALDI mass spectrum ob-
Figure 2. a) MALDI mass spectrum obtained for solution of extracted
MMP mixture in positive ion mode; b) ESI mass spectrum obtained for
an aqueous ammonium acetate (10 mm, pH 8.5) solution of extracted
MMPs (50 mm) in negative ion mode.
Scheme 1. Retrosynthetic analysis for MMP-Xs targets (where X=5, 8,
11 and 14).
tained in positive ion mode; an ESI mass spectrum obtained
in negative ion mode is shown in Figure 2b. Analysis of the
mass spectra reveals that MMP-10, -11, -12 and -13 repre-
sent the major components of the MMP mixture. The distri-
butions of MMPs in the MALDI and ESI mass spectra are
reasonably similar, with MMP-12 and -11 being the two
most abundant species. The fractional abundance (fMMP-X) of
each of the four MMP-X species (where X=10–13) was esti-
mated from the distribution of MMPs established from rep-
licate ESI mass spectra, Equation (1):
MMP-5s (Scheme 2a), disaccharide thioglycoside 1 was first
coupled with 8-azidooctanol by using activation with NIS
and silver triflate.[25] This reaction afforded the expected dis-
accharide 4 in 84% yield. The levulinoyl ester was then
cleaved by treatment with hydrazine acetate, giving an 86%
yield of the corresponding alcohol 5. Subsequent glycosyla-
tion of this compound, again by using thioglycoside 1 and
NIS/AgOTf, afforded 6, which was then deprotected with
hydrazine acetate. The product, tetrasaccharide alcohol 7,
was obtained in 71% overall yield in two steps from 5. The
final monosaccharide residue was incorporated into 7 by
using imidate 2, under the promotion of TMSOTf. Final
treatment of the pentasaccharide product 8 with sodium
methoxide removed all of the benzoate esters resulting in
MMP-5s (65% yield in two steps from 5).
The preparation of the octasaccharide target MMP-8s
(Scheme 2b) was achieved from tetrasaccharide 7, an inter-
mediate in the synthesis of MMP-5s. First, the tetrasacchar-
ide was elongated to a hexasaccharide (9) by glycosylation
with thioglycoside 1. Subsequent cleavage of the levulinate
group (yielding 10) was followed by glycosylation with thio-
glycoside 3 affording ocatasaccharide 11 in 85% yield over
the three steps from 7. Final treatment of 11 with sodium
methoxide afforded MMP-8s in 97% yield.
fMMP-X ¼ AbðMMP-XÞ=SAbðMMP-XÞ
ð1Þ
Assuming similar ESI response factors for the four
MMPs, the fMMP-X values are 0.04ꢁ0.01 (X=10), 0.37ꢁ0.01
(11), 0.53ꢁ0.02 (12) and 0.06ꢁ0.01 (13). From these values
and the molecular weights of individual MMP-X species
(MWMMP-X), the weighted average MW (MWMMP) of the MMP
mixture was calculated:
MW MMP ¼ SfMMP-X ðMW MMP-X
Þ
ð2Þ
The NA sample used in this work consisted of a mixture
of alkyl-substituted cyclic and acyclic aliphatic carboxylic
acids. Shown in Figure 3 is an ESI mass spectrum acquired
in negative ion mode for an aqueous solution of NAs at
pH 8.5 and 258C. Inspection of the mass spectrum reveals
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
3
&
ÞÞ
These are not the final page numbers!

T. L. Lowary, J. S. Klassen et al.
ion signals corresponding pre-
dominantly to species belong-
ing to the CnH2n+ZO2 series
with n=12 (Z=ꢀ2), 13 (Z=
ꢀ2), 14 (Z=ꢀ4), 15 (Z=ꢀ4),
and 16 (Z=ꢀ4). The average
molecular weight of the NA
sample was calculated follow-
ing the same procedure as used
for the MMPs.
The interactions between the
extracted MMPs and the com-
ponents of the NA mixture
were initially investigated by
using the direct ESI-MS assay.
Shown in Figure 4a is an illus-
trative ESI mass spectrum ac-
quired in negative ion mode
for an aqueous solution of
MMP (100 mm), NA (100 mm)
and imidazole (10 mm), which
was added to reduce the extent
of in-source dissociation.[14]
Signals
corresponding
to
(MMP+NA)ꢀ ions consisting
of the major components of
the NA mixture were detected.
Curiously, the complexes were
composed exclusively of MMP-
11 and no complexes contain-
ing the MMP-12, most abun-
dant MMP, were detected. This
result, on its own, suggests that
NA binding may be dependent
on the size of the MMPs, with
the smaller MMPs (i.e., MMP-
10) exhibiting higher affinities.
However, it is also possible
that, despite the presence of
the stabilizing additive imida-
Scheme 2. Synthesis of MMP-5s and MMP-8s targets.
zole in solution, the (MMP+
NA)ꢀ ions composed of the
longer MMPs undergo in-
source dissociation, such that the ESI mass spectrum does
not accurately reflect solution composition. The possibility
of in-source dissociation notwithstanding, the apparent Ka
(i.e., Ka,app) was determined to be (1ꢀ103)mꢀ1, based on the
relative abundance of all free MMP ions and the ligand
bound MMP-11 ions.
With the goal of testing the reliability of the direct ESI-
MS assay for quantifying the interactions between MMPs
and hydrophobic ligands, binding measurements were per-
formed on solutions containing the mixture of extracted
MMPs and PnA or PA. As noted above, MMPs are reported
to bind to long-chain FAs and their CoA derivatives with
high affinity in aqueous solution; for example, an affinity of
(2.5ꢀ106)mꢀ1 has been measured for PnA by using fluores-
Figure 3. ESI mass spectrum obtained for aqueous ammonium acetate
(10 mm, pH 8.5) solution of NAs (160 mm) in negative ion mode.
&
4
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Carbohydrate–Lipid Interactions
FULL PAPER
affinity and for which the gaseous ions of the corresponding
(Pproxy +L) complex are resistant to in-source dissociation.
For these measurements, the protein Lg, which possesses a
large hydrophobic cavity that can accommodate a wide vari-
ety of hydrophobic ligands,[26,27] was utilized. We recently
demonstrated that the interactions between Lg and long-
chain FAs, such as PA, can be quantified directly by ESI-MS
measurements.[14] For example, shown in Figure 5a is an ESI
Figure 4. ESI mass spectra obtained in negative ion mode for aqueous
ammonium acetate (10 mm, pH 8.5) solutions of: a) MMP (100 mm) with
NA (100 mm) imidazole (10 mm); b) MMP (75 mm) with PA (106 mm), imi-
dazole (10 mm); and c) MMP (75 mm) with PnA (96 mm), imidazole
(10 mm). The solution temperature for all measurements was 258C.
cence spectroscopy.[9] Shown in Figure 4b and c are illustra-
tive ESI mass spectra acquired in negative ion mode for
aqueous solutions of MMP (75 mm) with PA (106 mm) and
PnA (96 mm), respectively. ESI-MS analysis of the solution
containing PA reveals signal corresponding to (MMP+PA)ꢀ
ions consisting of each of the three major MMP species, that
is, X=11–13 (Figure 4b). This result suggests that hydropho-
bic ligand binding to the MMPs is, in fact, not strongly de-
pendent on the size of the MMP. However, the Ka,app calcu-
lated from the mass spectral data, (3ꢀ103)mꢀ1, is quite low
compared to previously reported values. In contrast, ESI-
MS analysis of the solution containing PnA reveals only
signal corresponding to complex composed of MMP-11, sim-
ilar to the results obtained for the NA sample (Figure 4c).
The Ka,app determined for the interaction with PnA, (4ꢀ
102)mꢀ1, is approximately four-orders of magnitude smaller
than the reported value.[9] Taken together, these results
strongly suggest that the gaseous deprotonated (MMP+L)ꢀ
ions composed of PA, PnA or the NAs, are prone to in-
source dissociation, and that the addition of imidazole is in-
sufficient to maintain the integrity of the complexes.
Figure 5. ESI mass spectra obtained in negative ion mode for aqueous
ammonium acetate (10 mm, pH 8.5) solutions of: a) Lg (12 mm), PA
(10 mm), imidazole (10 mm); b) Lg (12 mm), PA (10 mm), MMP (65 mm),
imidazole (10 mm); and c) Lg (12 mm), PA (10 mm), MMP (65 mm), NA
(74 mm), imidazole (10 mm). The solution temperature for all measure-
ments was 258C.
mass spectrum acquired in negative ion mode for solution
containing Lg (12 mm) and PA (10 mm) at pH 8.5 and 258C.
Ions corresponding to free Lg, that is, Lgqꢀ, at q=6 and 7,
were detected together with PA-bound Lg, (Lg+PA)7ꢀ.
From the relative abundance of ligand-bound and free Lg, a
Ka of (3.5ꢀ105)mꢀ1 was obtained (Table 1). This value is in
good agreement with values reported previously for this in-
teraction.[14]
Shown in Figure 5b is an ESI mass spectrum acquired in
negative ion mode for a solution of Lg (12 mm), PA (10 mm)
and MMP mixture (65 mm) at pH 8.5 and 258C. Although no
(MMP+PA)ꢀ ions were detected at these concentrations,
the addition of the MMPs to solution resulted in a dramatic
decrease in the relative abundance of the (Lg+PA)7ꢀ ion.
This observation confirms that the MMPs do bind PA under
these conditions. Following the procedure described in the
Experimental Section, a Ka value of (1.1ꢀ105)mꢀ1 was estab-
lished for the interaction between the MMPs and PA. This
To demonstrate conclusively the occurrence of in-source
dissociation an indirect assay, based on the recently devel-
oped “proxy protein” ESI-MS method,[18] was employed to
quantify the interactions between the MMPs and PA and
PnA. As noted above, this assay involves the use of a Pproxy
,
which binds specifically to the ligand of interest with known
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
5
&
ÞÞ
These are not the final page numbers!

T. L. Lowary, J. S. Klassen et al.
Table 1. Apparent association constants (Ka,app [mꢀ1]) for a mixture of
MMPs[a] binding to L (L=PA, SA, MA, PnA and NA), determined by
the “direct” ESI-MS assay, the “proxy protein” ESI-MS assay, and the
“proxy protein/proxy ligand” ESI-MS assay, at 258C and pH 8.5.[b]
Figure 5c shows an ESI mass spectrum acquired in nega-
tive ion mode for a solution of Lg (12 mm), PA (10 mm),
MMPs (65 mm) and NAs (74 mm) at pH 8.5 and 258C. Al-
though no ions corresponding to the (MMP+NA) were de-
tected at these concentrations, the addition of the NAs to
the solution resulted in a marked increase in the relative
abundance of the (Lg+PA)7ꢀ ion, from 0.30 (Figure 5b) to
0.55 (Figure 5c). This increase is consistent with the pres-
ence of specific interactions between the MMPs and NAs.
Analysis of the ESI-MS data by using the approach descri-
bed in the Experimental Section yields a Ka,app of (4.7ꢁ
0.9)ꢀ104 mꢀ1 for the interactions between the MMPs and
NAs. Analogous measurements were performed by using
MA, SA, or PnA as Lproxy. In all cases, there is good agree-
ment between the Ka values determined for the interactions
between the MMPs and NAs (Table 1). Based on all four
data sets, an average Ka value of (4.6ꢁ0.5)ꢀ104 mꢀ1 was es-
tablished.
L
Direct ESI-
MS
Proxy protein
ESI-MS
Proxy protein/proxy ligand
ESI-MS
MA
PA
SA
PnA
NA
–
N
–
–
–
–
E
–
E
A
–
N
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
[a] MMPs extracted from M. smegmatis. [b] Errors correspond to one
standard deviation.
value is approximately 100-times larger than the value es-
tablished from the direct ESI-MS measurements. Using this
competitive assay, MMP binding measurements were per-
formed on three other FAs, MA, SA and PnA. The results
of the binding measurements are listed in Table 1. Notably,
the Ka value measured for PnA, (4.3ꢀ104)mꢀ1, is also signifi-
cantly larger than that obtained from direct ESI-MS meas-
urements. Taken together, these results confirm that the de-
protonated (MMP+FA)ꢀ and, presumably, (MMP+NA)ꢀ
ions are prone to gas-phase dissociation during ESI-MS
analysis, and that an indirect measurement of these interac-
tions is required. It is also worth noting that the Ka value
measured for PnA is significantly smaller than the reported
value, which was measured by using a fluorescence assay.[9]
The reason for this discrepancy is not known but may be
due to an error in the concentration of the MMP mixture
used in the earlier study. Specifically, an overestimation of
the MMP concentration due to incomplete removal of
Tween (see the Experimental Section), which was used for
culturing the bacteria from which these glycans were isolat-
ed, would lead to an overestimation in the Ka value. In addi-
tion, this fluorescence assay has been noted[28] to be highly
sensitive to the presence of oxygen and trace amount of im-
purities, which are not expected to be limitations of the ESI-
MS assay we have developed.
In principle, the proxy protein ESI-MS method could also
be used to quantify the interactions between the MMPs and
the NAs. However, ESI-MS measurements performed on
solutions of Lg and the NA mixture revealed no evidence of
(Lg+NA) complexes (data not shown). Therefore, an alter-
native approach, the “proxy protein/proxy ligand” ESI-MS
assay, was used. As noted above, this assay employs a Pproxy
and an Lproxy. Both L and Lproxy bind competitively to the
MMPs, whereas only Lproxy binds to Pproxy. Furthermore, the
gaseous ions of the (Pproxy +Lproxy) complex are resistant to
in-source dissociation. Therefore, from the change in rela-
tive abundance of (Pproxy +Lproxy) complex, determined from
direct ESI-MS measurements, the affinity of the (MMP+L)
complex can be determined.
These studies represent the first demonstration that
MMPs can bind to cyclic or branched lipids. This finding is,
perhaps, not surprising given that the inner diameter of the
proposed helical structure of the MMPs is comparable to
the cavity present in b-cyclodextrin, which binds to an array
of branched and aromatic lipophilic molecules, albeit with
affinities lower than that reported here for the NA–MMP
interaction.[29,30] Moreover, in contrast to the cyclodextrins,
which have cavities of defined size constrained by their
cyclic nature, the acyclic structure of MMPs should allow
sufficient plasticity for the structure to “breathe”, thus al-
lowing binding to species other than simple fatty acids and
their CoA derivatives.
The data presented above suggest that MMPs extracted
from mycobacteria could, in principle, be used as a sorbent
for the removal of NAs from TPW. However, the use of
MMPs extracted from bacteria is impractical given the low
yields in which they can be isolated. Instead, implementa-
tion of this strategy requires obtaining MMPs by other
methods. One option is chemical synthesis and ideally purifi-
cation columns based upon these materials would be con-
structed by using the shortest MMP that effectively binds to
NAs. Another advantage of synthetic glycans over the natu-
ral materials, which are terminated at the reducing end with
a methyl group, is that they can be prepared with a function-
al group (e.g., an azide or its amine counterpart) to facilitate
their conjugation to a solid support. To establish the influ-
ence of size on the affinities of MMPs for NAs, the series of
synthetic MMP-Xs, where X=5, 8, 11, and 14, described
above (Figure 1) were evaluated for their ability to bind to
NAs. Using the proxy protein (where Pproxy =Lg) and proxy
protein/proxy ligand ESI-MS assays (where Pproxy =Lg and
L
proxy =PA) the affinities of the individual synthetic MMPs
for PA and the commercial mixture of NAs, respectively,
were measured at pH 8.5 and 258C (Table 2). It can be seen
that, for both PA and the NAs, the value of Ka increases
modestly with the length of MMP. The effect is more pro-
nounced for PA binding, with the affinity increasing from
(1.8ꢁ0.5)ꢀ104 (MMP-5s) to (8.8ꢁ0.3)ꢀ104 mꢀ1 (MMP-14s);
&
6
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Carbohydrate–Lipid Interactions
FULL PAPER
Table 2. Association constants (Ka, mꢀ1) measured for the binding of syn-
thetic MMP-Xs binding to PA (by using proxy protein ESI-MS method,
Experimental Section
P
proxy =Lg), and to NAs (by using the proxy protein/proxy ligand ESI-MS
MMPs, proteins and assay solutions: The mixture of MMPs (consisting of
species with 10–13 Manp residues, named MMP-10, -11, -12 and -13, re-
spectively, Figure 1) were extracted and purified from M. smegmatis as
described by Hindsgaul and Ballou.[31] Key steps in the purification were
affinity chromatography by using a silica-based absorbent functionalized
with palmitic acid followed by treatment of the eluate with decolorizing
charcoal. Although earlier work[31] had reported the MMPs could be ob-
tained pure following a single affinity chromatography-decolorization se-
quence, in our hands, multiple column passages were required to remove
the polyoxyethylenesorbitan monooleate (Tween) that was used in the
media for growing the bacteria. The presence of Tween was obvious in
the mass spectra obtained while carrying out our investigations. However,
given the similar chemical shifts of the hydrogen atoms in Tween and the
MMPs, this impurity cannot be readily seen by 1H NMR spectroscopy,
which had been used by Hindsgaul and Ballou to characterize their iso-
lated materials.[31]
assay, Pproxy =Lg, Lproxy =PA) at 25 8C and pH 8.5.[a]
MMP-Xs
Ka [mꢀ1
]
PA
(1.8ꢁ0.5)ꢀ104
(2.6ꢁ0.5)ꢀ104
(3.7ꢁ0.2)ꢀ104
(8.8ꢁ0.3)ꢀ104
NA
(3.6ꢁ0.5)ꢀ104
(3.9ꢁ0.6)ꢀ104
(4.5ꢁ0.5)ꢀ104
(7.6ꢁ0.7)ꢀ104
X=5
X=8
X=11
X=14
G
ACHTUNGTRENNUNG
E
ACHTUNGTRENNUNG
N
ACHTUNGTRENNUNG
N
ACHTUNGTRENNUNG
[a] Errors correspond to one standard deviation.
for the NAs, the values range from (3.6ꢁ0.5)ꢀ104 to (7.6ꢁ
0.7)ꢀ104 mꢀ1. Notably, the affinities measured for synthetic
MMPs for the NAs are very similar to apparent value meas-
ured for the extracted MMPs. Based on these results, it is
predicted that columns containing immobilized MMPs as
short as five monosaccharide residues (MMP-5s) would be
nearly as effective as those containing significantly longer
MMPs.
The synthesis of MMP-5s, -8s, -11s and -14s was carried out as described in
the Results and Discussion and in the Supporting Information. All exper-
imental data for the synthetic intermediates and final compounds can be
found in the Supporting Information.
Bovine b-lactoglobulin (Lg, monomer MW 18281 Da), palmitic acid (PA,
256.4 Da), myristic acid (MA, 228.4 Da), and stearic acid (SA, 284.8 Da)
were purchased from Sigma–Aldrich Canada (Oakville). trans-Parinaric
acid (PnA, MW 276.4 Da) was purchased from Cayman Chemical (Ann
Arbor, MI). A sample of commercially available Merichem naphthenic
acids (NA) was generously provided by Professor Greg G. Goss (Depart-
ment of Biological Sciences, University of Alberta). The structures of the
MMPs, FAs and NAs are shown in Figure 1.
Conclusion
In summary, the interactions between MMPs, extracted
from M. smegmatis or produced synthetically, and linear sa-
turated and unsaturated FAs, and a commercial mixture of
NAs, in aqueous solution (258C, pH 8.5) were quantified by
ESI-MS. Association constants for the binding of the MMPs
to four FAs, MA, PA, SA and PnA, were measured by using
the “proxy protein” ESI-MS assay. The measured Ka values
range from about 104 to approximately 105 mꢀ1, and increase
with the size of the MMP. Notably, the apparent affinity
measured for the mixture of extracted MMPs and PnA is
significantly smaller (by a factor of 100) than the reported
value. A newly developed competitive binding assay, refer-
red to as the “proxy protein/proxy ligand” ESI-MS method,
was used to quantify the interactions between the MMPs
and the commercial NA mixture. These results demonstrate,
for the first time, that MMPs bind specifically to cyclic,
branched-chain lipids. Moreover, we have demonstrated
that at least, some of the NAs in the mixture bind to the
MMPs with apparent affinities of approximately (5ꢀ
104)mꢀ1, which in turns suggests that these glycans may have
potential in the remediation of TPW. In preliminary investi-
gations, we have demonstrated that the polystyrene beads
functionalized with MMP-5s do indeed bind to NAs in TPW
and that the treated water has significantly reduced toxicity.
Finally, the ESI-MS method developed here represents an
attractive approach for probing carbohydrate–lipid interac-
tions. In particular, it does not suffer from the aforemen-
tioned limitations of the previously developed fluorescence
assay,[9,10] We also view it as superior to a more recently-re-
ported UV-based binding assay, which requires deconvolu-
tion of complex equilibria involving lipid aggregates.[28]
Stock solutions of MMPs were prepared by dissolving extracted and puri-
fied MMPs into Milli-Q water to yield a final concentration of 3m. The
stock solutions of FAs and NAs were prepared by dissolving a known
mass of each compound or mixture of compounds into methanol
(MeOH). Lg was dissolved and exchanged directly into Milli-Q water, by
using an Amicon microconcentrator with a MW cutoff of 10 kDa. The
concentration of the Lg solution was determined by lyophilizing a known
volume of the filtrate and measuring the mass of the protein. The protein
stock solution was stored at ꢀ208C. The ESI solutions were prepared
from the stock solutions. For the binding measurements, imidazole
(10 mm) was also added to reduce the occurrence of in-source dissocia-
tion.[14–16] Ammonium acetate buffer was added into the ESI solution to a
final concentration of 10 mm. Aqueous ammonium hydroxide was added
to adjust the pH of the solution to 8.5.
Mass spectrometry: Isolated MMPs were analyzed by matrix-assisted
laser desorption ionization (MALDI) MS in positive ion mode by using a
Voyager Elite MALDI time-of-flight (ToF) mass spectrometer (AB
Sciex, Framingham, MA).
A solution of 2,5-dihydroxybenzoic acid
(DHB, 20 mgmLꢀ1) in 3:1 H2O/MeOH was used as the matrix solution.
An aqueous solution of MMPs was mixed with the matrix solution at a
1:1 ratio. An aliquot (0.7 mL) of the mixture was loaded onto the
MALDI target plate by using a micropipette and allowed to dry. A small
volume of an NaCl (1 mm) solution was added to the spots to improve
ionization efficiency. Analysis of the MMPs by ESI-MS, as well as the
ESI-MS binding measurements, were performed in negative ion mode by
using
a 9.4 Tesla Apex II Fourier-transform ion cyclotron resonance
(FTICR) mass spectrometer (Bruker, Billerica, MA). Nanoflow ESI was
performed by using borosilicate tubes (1.0 mm o.d., 0.68 mm i.d.), pulled
to approximately 5 mm o.d. at one end by using a P-2000 micropipette
puller (Sutter Instruments, Novato, CA). Details of the instrumental pa-
rameters used for the binding measurements are given elsewhere.[14–16]
ESI-MS binding measurements: Both direct and competitive ESI-MS
assays were used to quantify the Ka values for the interaction between
MMPs and FA and NA ligands (L), Equation (3):
MMP þ L Ð ðMMP þ LÞ
ð3Þ
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
7
&
ÞÞ
These are not the final page numbers!

T. L. Lowary, J. S. Klassen et al.
A brief description of each assay is given below.
The procedure for determining Ka for the interaction between MMP and
L is similar to that described above. The concentration of free Lproxy can
be determined from the ratio (Rproxy) of Lproxy-bound and unbound Pproxy
1) Direct ESI-MS assay: The direct ES-MS assay is based on the quanti-
fication of the abundance (Ab) of ligand-bound and unbound ions in the
gas phase,[14, 16,32–36] for example, (MMP+L)ꢀ and MMPꢀ, respectively.
The measured abundance ratio (R) is assumed to be equivalent to the
equilibrium concentration ratio of bound and free MMP ions in solution,
Equation (4):
measured by ESI-MS and the known affinity (Ka,proxy) of Pproxy for Lproxy
,
Equations (7) and (8). From the values of [Lproxy and of [Pproxy
]
+
eq
Lproxy]eq, the concentration of MMP bound Lproxy can be found from the
equation of mass balance, Equation (15):
AbððMMP þ LÞꢀÞ
AbððMMPÞꢀÞ
½MMP þ Lproxyꢂeq ¼ ½Lproxyꢂo ꢀ ½Lproxyꢂeq ꢀ ½Pproxy þ Lproxyꢂeq
ð15Þ
½MMP þ Lꢂeq
½MMPꢂeq
¼
¼ R
ð4Þ
The concentration of free MMP in solution can be determined from
Equation (16):
From the measured R value and initial concentrations of MMP and L
(i.e., [MMP]o and [L]o, respectively), the association constant Ka,MMP can
be calculated by using Equation (5):
½MMP þ Lproxyꢂeq
½MMPꢂeq
¼
ð16Þ
Ka;ðMMPþL ½Lproxy
ꢂ
proxy Þ
eq
R
R
Ka;MMP
¼
ð5Þ
and the concentration of MMP bound to L can be found from Equa-
tion (17):
½Lꢂo ꢀ 1þR ½MMPꢂo
2) Proxy protein ESI-MS assay: A competitive binding assay,[18] employ-
½MMP þ Lꢂeq ¼ ½MMPꢂo ꢀ ½MMPꢂeq ꢀ ½MMP þ Lproxyꢂeq
Finally, Ka,(MMP+L) can be calculated by using Equation (12).
ð17Þ
ing a proxy protein (Pproxy) that binds specifically to L, Equation (6):
Pproxy þ L Ð ðPproxy þ LÞ
ð6Þ
was also employed to quantify the interactions of FAs with the MMPs.
Bovine Lg, which is known to bind reasonably strongly to long-chain FAs
in basic solutions and forms kinetically-stable gaseous (Lg+FA)qꢀ
ions,[14] served as Pproxy for these measurements. The concentration of free
L in solution is determined from the ratio (Rproxy) of ligand-bound and
unbound Pproxy measured by ESI-MS and the known affinity of Pproxy for
L (Ka,proxy), Equations (7) and (8):
Acknowledgements
The authors acknowledge the generous financial support provided by the
Centre for Oil Sands Innovation, the Alberta Glycomics Centre and the
Natural Sciences and Engineering Research Council of Canada.
P
AbððPproxy þ LÞqꢀÞ
½Pproxy þ Lꢂeq
q
P
¼
¼ Rproxy
ð7Þ
ð8Þ
[1] Y. C. Lee, C. E. Ballou, J. Biol. Chem. 1964, 239, PC3602–PC3603.
[2] V. Mendes, A. Maranha, S. Alarico, N. Empadinhas, Nat. Prod. Rep.
2012, 29, 834–844.
[3] M. Ilton, A. W. Jevans, E. D. McCarthy, D. Vance, H. B. White III,
K. Bloch, Proc. Natl. Acad. Sci. USA 1971, 68, 87–91.
[4] K. Bloch, Adv. Enzymol. Relat. Areas Mol. Biol. 1977, 45, 1–84.
[5] a) K. K. Yabusaki, C. E. Ballou, J. Biol. Chem. 1979, 254, 12314–
12317; b) L. S. Forsberg, A. Dell, D. J. Walton, C. E. Ballou, J. Biol.
Chem. 1982, 257, 3555–3563.
½Pproxyꢂeq
AbððPproxy
Þ
qꢀÞ
q
½Pproxy þ Lꢂeq Rproxy
Ka;proxy
¼
¼
½Pproxyꢂeq½Lꢂeq
½Lꢂeq
and the concentration of (Pproxy +L) can be calculated from Equation (9):
Rproxy
o 1 þ Rproxy
½Pproxy þ Lꢂeq ¼ ½Pproxy
ꢂ
ð9Þ
[6] N. Papaioannou, H. S. Cheon, Y. Lian, Y. Kishi, ChemBioChem
2007, 8, 1775–1780.
[7] M. Jackson, P. J. Brennan, J. Biol. Chem. 2008, 284, 1949–1953.
[8] J. E. Maggio, Proc. Natl. Acad. Sci. USA 1980, 77, 2582–2586.
[9] K. K. Yabusaki, C. E. Ballou, Proc. Natl. Acad. Sci. USA 1978, 75,
691–695.
where [Pproxy
] is the initial concentration of Pproxy. From mass balance
o
considerations, the concentration of (MMP+L) and MMP can be calcu-
lated by using Equation (10) and (11):
[10] C. E. Ballou, Pure Appl. Chem. 1981, 53, 107–112.
[11] J. S. Clemente, P. M. Fedorak, Chemosphere 2005, 60, 585–600.
[12] R. Frank, R. Kavanagh, B. Burnison, J. Headley, K. Peru, G. Derk-
raak, K. Solomon, Chemosphere 2006, 64, 1346–1352.
[13] M. H. Mohamed, L. D. Wilson, J. V. Headley, K. M. Peru, Process
Saf. Environ. Prot. 2008, 86, 237–243.
½MMP þ Lꢂeq ¼ ½Lꢂo ꢀ ½Lꢂeq ꢀ ½Pproxy þ Lꢂeq
½MMPꢂeq ¼ ½MMPꢂo ꢀ ½MMP þ Lꢂeq
ð10Þ
ð11Þ
The value of Ka,MMP can then be calculated by from Equation (12):
[14] L. Liu, E. N. Kitova, J. S. Klassen, J. Am. Soc. Mass Spectrom. 2011,
22, 310–318.
[15] D. Bagal, E. N. Kitova, L. Liu, A. El-Haweit, P. D. Schnier, J. S.
Klassen, Anal. Chem. 2009, 81, 7801–7806.
½MMP þ Lꢂeq
Ka;MMP
¼
ð12Þ
½Lꢂeq½MMPꢂeq
3) Proxy protein/proxy ligand ESI-MS assay: A newly developed ESI-
MS method was ultimately used to determine the affinities of the MMPs
for NAs. This assay incorporates both a Pproxy and a proxy ligand (Lproxy
to establish the Ka for the MMPs binding to L. Both L and Lproxy bind to
the MMPs, with association constants Ka,(MMP+L) and Ka,(MMP+Lproxy)
whereas only Lproxy binds to Pproxy, with the association constant Ka,proxy
[16] J. Sun, E. N. Kitova, J. S. Klassen, Anal. Chem. 2007, 79, 416–425.
[17] C. V. Robinson, E. W. Chung, B. B. Kragelund, J. Knudsen, R. T.
Aplin, F. M. Poulsen, C. M. Dobson, J. Am. Chem. Soc. 1996, 118,
8646–8653.
[18] A. El-Hawiet, E. N. Kitova, D. Arutyunov, D. J. Simpson, C. M. Szy-
manski, J. S. Klassen, Anal. Chem. 2012, 84, 3867–3870.
[19] a) M. C. Hsu, J. Lee, Y. Kishi, J. Org. Chem. 2007, 72, 1931–1940;
b) H.-S. Cheon, Y. Lian, Y. Kishi, Org. Lett. 2007, 9, 3323–3326.
[20] M. Meppen, Y. Wang, H.-S. Cheon, Y. Kishi, J. Org. Chem. 2007, 72,
1941–1950.
)
;
,
Equations (3), (13) and (14):
Pproxy þ Lproxy Ð ðPproxy þ Lproxy
Þ
ð13Þ
ð14Þ
MMP þ Lproxy Ð ðMMP þ Lproxy
Þ
[21] H.-S. Cheon, Y. Lian, Y. Kishi, Org. Lett. 2007, 9, 3327–3329.
&
8
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Carbohydrate–Lipid Interactions
FULL PAPER
[22] Y. Wang, H.-S. Cheon, Y. Kishi, Chem. Asian J. 2008, 3, 319–326.
[23] a) W. Liao, D. Lu, Carbohydr. Res. 1996, 296, 171–182; b) W. Liao,
D. Lu, Carbohydr. Res. 1997, 300, 347–349; c) M. Hirooka, M. Ter-
ayama, E. Mitani, S. Koto, A. Miura, K. Chiba, A. Takabatake, T.
Tashiro, Bull. Chem. Soc. Jpn. 2002, 75, 1301–1309; d) W. Liao, D.
Lu, A. Li, F. Kong, J. Carbohydr. Chem. 1997, 16, 877–890.
[24] B. Becker, R. H. Furneaux, F. Reck, O. A. Zubkov, Carbohydr. Res.
1999, 315, 148–158.
[30] J. Szejtli, Chem. Rev. 1998, 98, 1743–1754.
[31] O. Hindsgaul, C. E. Ballou, Biochemistry 1984, 23, 577–584.
[32] T. J. D. Jørgensen, P. Roepstorff, A. J. R. Heck, Anal. Chem. 1998,
70, 4427–4432.
[33] W. Wang, E. N. Kitova, J. S. Klassen, Anal. Chem. 2003, 75, 4945–
4955.
[34] A. Wortmann, M. C. Jecklin, D. Touboul, M. Badertscher, R.
Zenobi, J. Mass Spectrom. 2008, 43, 600–608.
[25] P. Konradsson, U. E. Udodong, B. Fraser-Reid, Tetrahedron Lett.
1990, 31, 4313–4316.
[35] Y. H. Yu, C. E. Kirkup, N. Pi, J. A. Leary, J. Am. Soc. Mass Spec-
trom. 2004, 15, 1400–1407.
[26] G. Kontopidis, C. Holt, L. Sawyer, J. Dairy Sci. 2004, 87, 785–796.
[27] B. Y. Qin, M. C. Bewley, L. K. Creamer, H. M. Baker, E. N. Baker,
G. B. Jameson, Biochemistry 1998, 37, 14014–14023.
[28] H.-S. Cheon, Y. Wang, J. Ma, Y. Kishi, ChemBioChem 2007, 8, 353–
359.
[36] B. Ganem, J. D. Henion, Bioorg. Med. Chem. 2003, 11, 311–314.
Received: April 10, 2012
Revised: June 6, 2012
Published online: && &&, 0000
[29] R. Bergeron, T. Machida, K. Bloch, J. Biol. Chem. 1975, 250, 1223–
1230.
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
9
&
ÞÞ
These are not the final page numbers!

T. L. Lowary, J. S. Klassen et al.
Carbohydrates
L. Liu, Y. Bai, N. Sun, L. Xia,
T. L. Lowary,*
J. S. Klassen* .................. &&&& — &&&&
Mixing sugar and oil: The interactions
between 3-O-methyl-mannose polysac-
charides, extracted from Mycobacte-
rium smegmatis or obtained by chemi-
cal synthesis, with fatty acids and a
commercial mixture of naphthenic
acids (see scheme) were quantified by
electrospray ionization mass spectrom-
etry.
Carbohydrate–Lipid Interactions:
Affinities of Methylmannose Polysac-
charides for Lipids in Aqueous Solu-
tion
&
10
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!