Differentiation between structurally homologous shiga 1 and shiga 2 toxins by using synthetic glycoconjugates

Article abstract of DOI:10.1002/anie.200703680

(Chemical Equation Presented) Choose your poison: Shiga toxins 1 and 2 are the major virulence factors of E. coli O157:H7. However, Shiga 2 is more potent than Shiga 1. Biotinylated glycoconjugates have been developed that differentiate between these stru

Full text of DOI:10.1002/anie.200703680

Angewandte
Chemie
DOI: 10.1002/anie.200703680
Glycoconjugates
Differentiation between Structurally Homologous Shiga 1 and Shiga 2
Toxins by Using Synthetic Glycoconjugates**
Ramesh R. Kale, Colleen M. McGannon, Cynthia Fuller-Schaefer, Duane M. Hatch,
Michael J. Flagler, Shantini D. Gamage, Alison A. Weiss,*and Suri S. Iyer*
Shiga toxins 1 and 2 (Stx1 and Stx2) are major virulence
factors of Escherichia coli O157:H7 and have been included
on the list of Select Biothreat Agents.[1] Of the estimated
70000 E. coli O157:H7 cases of disease per year in the United
States, 10–15% develop hemolytic uremic syndrome (HUS);
3–5% succumb during the acute phase of the disease; and an
equivalent number suffer brain damage and renal failure.[2]
Treatment of Shiga-toxin-mediated HUS remains primarily
supportive, as postdiarrheal antibiotic treatment is reported
to enhance toxin production and progression of HUS.[3]
Synthetic toxin neutralizers have proven to be ineffective in
affected patients, but could be useful as prophylactics if
diagnosis can be made early.[4–6] Furthermore, the presence of
Stx genes on the lysogenic bacteriophage[7] allows facile
transmission of Stx onto harmless serotypes of E. coli and
other enteric bacterial species, making it a serious emerging
threat to humans.[8]
Shiga toxin is a member of the AB5 family of toxins. The
Asubunit is enzymatically active and cleaves a single adenine
residue from the 28S ribosomal RNA, rendering the ribosome
incapable of protein synthesis. The B, or binding, subunit is a
homopentamer that binds to globotriaosylceramide (Gb3) in
lipid rafts on the cell surface and ultimately delivers the
Asubunit to its cytoplasmic target. The two major antigenic
variants of Shiga toxin, Stx1 and Stx2, share 55% amino acid
homology. Differences in affinity for Gb3 were observed by
using surface plasmon resonance spectroscopy; the affinities
Kd of Stx1 and Stx2 for Gb3 are 4.6 ” 10ꢀ8 m and 3.7 ” 10ꢀ7 m,
respectively.[9] However, Stx1 and Stx2 showed similar
affinities for Gb3 derivatives using electrospray ionization
mass spectrometry.[10] Spacer length, mode of presentation,
and assay conditions appear to influence binding pro-
foundly.[11–13] Strains of E. coli O157:H7 can produce Stx1,
Stx2, or both, but epidemiological studies suggest that Stx2-
(and not Stx1-) producing E. coli are associated with the
development of HUS.[14] These observations have been
confirmed by baboon[15] and murine[16] models. Indeed, the
lethal dose for Stx1 and Stx2 in mice has been reported to be
approximately 1200 and 2.4 ng, respectively.[16] Therefore, in
addition to the early and rapid detection of Shiga-toxin-
producing E. coli, identification of Shiga toxin variants
produced by the infecting strain to determine its pathogenic
potential is critical.[17]
In previous studies, the O-polysaccharides on several
lipopolysaccharide (LPS) serotypes expressed by harmless
E. coli strains have been shown to bind Stx2 specifically and
neutralize its toxicity to mammalian Vero cells in culture.[18]
The structure of one neutralizing polysaccharide, correspond-
ing to serogroup O117, had been previously determined.[19]
Interestingly, it resembles Gb3; however, there are significant
structural differences. Gb3 has a terminal a-1,4-digalactose
moiety (Figure 1, top), whereas the neutralizing polysaccha-
[*] C. M. McGannon, Dr. C. Fuller-Schaefer, M. J. Flagler,
Dr. S. D. Gamage, Prof. Dr. A. A. Weiss
Department of Molecular Genetics, Biochemistry and Microbiology
University of Cincinnati
Cincinnati, OH 45267-0524 (USA)
Fax: (+1)513-558-8474
E-mail: alison.weiss@uc.edu
Dr. R. R. Kale, D. M. Hatch, Prof. Dr. S. S. Iyer
Department of Chemistry
University of Cincinnati
Cincinnati, OH 45221-0172 (USA)
Fax: (+1)513-556-9239
E-mail: suri.iyer@uc.edu
Homepage: http://www.che.uc.edu/sensors/iyer.html
Figure 1. Structure of Gb3 (top) and the O117 LPS (bottom). The
hydroxy and N-acetyl groups have been shaded for emphasis.
[**] Financial support for this work was provided by the Center for
Chemical and Biosensors, Department of Chemistry, University of
Cincinnati (SSI), UC Research Council and Nanotechnology
Institute (A.A.W. and S.S.I.), NIAID (RO1 AI064893 A.A.W.) and
NIAID (U01-AI075498 A.A.W. and S.S.I.). S.S.I. thanks Dr. Stephan
Macha for mass spectral analysis.
ride has a modified terminal digalactose moiety (Figure 1,
bottom); the galactose residues possess a bulky N-acetyl
group at the 2-position. In contrast to Gb3, which binds to
both Stx1 and Stx2, the neutralizing polysaccharide was not
able to bind Stx1 or to neutralize its effects on Vero cells,
Supporting information (experimental details) for this article is
available on the WWW under http://www.angewandte.org or from
the author.
Angew. Chem. Int. Ed. 2008, 47, 1265 –1268
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1265

Communications
suggesting the N-acetyl group modifies binding specificity
towards Stx1.
On the basis of these observations, we hypothesized that
analogues of Gb3 could distinguish between Stx variants. We
developed a modular synthetic approach that allows variable
spacers and recognition elements to be incorporated without
considerable alteration of the overall synthetic strategy. An
illustration of the tailored glycoconjugate is depicted in
Figure 2 along with the structures of the synthetic glycocon-
shown in Scheme 1. Briefly, coupling of acceptor 4, obtained
by a two-step procedure from the known allyl-2-azido-4,6-
benzylidene-2-deoxy-b-d-galactopyranoside,[21] with trichlor-
oacetimidate donor 5[22] in the presence of catalytic amount of
TMSOTf yielded disaccharide 6 in reasonable yields (see the
Supporting Information). The azide functionalities were
reduced to the N-acetyl groups by using thiolacetic acid to
give 7. Extension of the allyl group by using a cross-
metathesis procedure was achieved with Grubbs catalyst to
yield the E,Z-isomers 8a,b as an insep-
arable mixture. Cleavage of the silyl
group and subsequent mesylation and
conversion to an azide yielded 9a,b in
68% yield over the three-step
sequence. The azide can easily be
coupled by 1,3-dipolar addition to
dimeric scaffold 10, bearing two
alkynes and a protected amine previ-
ously synthesized in our group.[23]
Thus, reaction of 2.1 equivalents of
the azide-bearing glycoconjugate with
10 yielded 11. Hydrogenation in the
presence of Pearlmanꢀs catalyst, and
reacylation yielded 12 in quantitative
yield. The final steps involved cleavage
of the Boc protecting group and cou-
pling to biotin to yield 13, which was
subjected to ZemplØn conditions to
give GC-1. The final product GC-1 was
purified by using a Biogel P-2 column
and lyophilized to yield a white foam.
Figure 2. Representation of the tailored biotinylated glycoconjuate and the structures of the five
molecules. The blue ellipse is the carbohydrate recognition element, the biotinylated scaffold and
the spacer are colored purple and red, respectively.
Similarly, GC-2a/b and GC-3a/b, in
which the terminal galactose of Gb3 is
replaced by an N-acetyl galactosamine
derivative, were synthesized (see the
Supporting Information).
jugates. The three components of the generic glycoconjugate
are 1) recognition element, 2) flexible spacer terminated in an
azide, and 3) dimeric scaffold bearing two alkynes. Using this
design, we synthesized biantennary ligands with varying N-
acetylation patterns. Each of the subunits in the B pentamer
of Stx possesses up to three Gb3 binding sites, and, therefore,
multiple binding sites are available. For initial studies we
chose a dimeric biotinylated scaffold to increase binding
affinity. As the length of the spacer has been shown to affect
the binding affinity,[11–13] we examined the influence of linker
length by synthesizing compounds with 6 (GC-1, GC-2a and
GC-3a) and 12 carbon atom (GC-2b and GC-3b) spacers. We
used biotin because it affords easy access to multivalency as
one streptavidin tetramer binds to four biotins and it can be
conjugated to commercial streptavidin matrices for toxin
capture.[20] The molecule was designed such that the biotin
and, hence, streptavidin is attached to the opposite end of the
rigid scaffold to minimize biotin interference in the binding
event.
Binding of Stx1, Stx2, and Stx2c to the synthetic glyco-
conjugates was assessed by ELISAanalysis as described
previously.[18] Toxin-containing culture supernatants, steri-
lized by filtration, were prepared from C600:H19B (Stx1),
C600:933W (Stx2), and O157 strain C394-03 (Stx2c).[24] The
biotinylated compounds were diluted in phosphate-buffered
saline (PBS) and added in excess to commercially available
precoated streptavidin microwell plates with a binding
capacity of around 125 pm per microwell. Binding was
allowed to proceed at room temperature for 2 hours. The
wells were washed three times with PBS and incubated with
PBS (negative control) or Stx-containing culture supernatant
at room temperature for 2 hours. The color was developed by
using commercially available polyclonal antibody to Shiga
toxin (Meridian Bioscience, Inc., Cincinnati, OH), alkaline
phosphatase conjugated goat antirabbit IgG, and p-nitro-
phenyl phosphate (Meridian Bioscience, Inc. or Sigma, St.
Louis, MO). The absorbance (l = 405 nm) was read by an
ELX800 microplate reader (Bio-Tek Instruments, Inc).
The results are shown in Figure 3. Similar to observations
for neutralizing LPS, Stx2 bound to the di- and mono-N-acetyl
substituted galactosamine GC-1 and GC-3a, respectively
(Figure 3A), whereas Stx1 failed to bind to either compound
We choose 2-deoxy-2-azidogalactosamine derivatives as
suitable starting materials as these derivatives yield a isomers
with high stereoselectivity. The synthesis of a representative
glycoconjugate, GalNAc(a-1,4)GalNAc, coupled to biotin is
1266
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1265 –1268

Angewandte
Chemie
Scheme 1. Synthesis of the biotinylated dimer of disaccharides. Reagents and conditions: a) TMSOTf, DCM, ꢀ208C!RT, 63%; b) AcSH, RT, 12 h,
57%; c) (1,1-dimethylethyl)dimethyl(4-pentenyloxy)silane, [((Cy)3P)2RuCl2(CHPh)], DCM, 16 h, 87%; d) TBAF, THF, RT, 3 h, 87%; e) MsCl,
(iPr)2EtN, DCM, ꢀ108C!RT, 5 h; f) NaN3, DMF, 658C, 5 h; 67% over two steps; g) CuSO4, sodium ascorbate, THF/H2O, RT, 2 days, 86%; h) H2,
Pd(OH)2, MeOH, RT, 1 atm, 12 h; i) Ac2O, pyridine, DMAP, 88% over two steps; j) TFA, TIPS, DCM, RT, 8 h; k) (+)-biotin, CDMT, NMM, THF/
DMF, 08C!RT, 32 h, 61% over two steps; l) NaOMe, MeOH, RT, 16 h, 86%. TMS=trimethylsilyl; Ac=acetyl; Bn=benzyl; Tf=trifluorometha-
nesulfonyl; DCM=dichloromethane; Cy=cyclohexyl; TBDMS=tert-butyldimethylsilyl; TBAF=tetrabutylammonium fluoride; Ms=methanesul-
fonyl; DMF=N,N-dimethylformamide; Boc=tert-butoxycarbonyl; DMAP=4-dimethylaminopyridine; TFA=trifluoroacetic acid; TIPS=triisopro-
pylsilyl; CDMT=2-chloro-4,6-dimethoxy-1,3,5-triazine; NMM=N-methylmorpholine.
(Figure 3B). The trisaccharide GC-3a appears to be a better
substrate for Stx2 than the disaccharide GC-1, as the former
exhibited enhanced toxin binding and was more sensitive
(toxin binding less than 100 ng per well). In contrast, GC-2a, a
trisaccharide analogue of Gb3, bound exclusively to Stx1 and
not to Stx2. The limit of detection was 10 ng of toxin per
microwell. We attribute the exclusive selectivity of GC-2a
towards Stx1 to the architecture of the biantennary analogue;
possibly, the shorter spacer permits binding to Stx1, but
constrains binding to Stx2.[10] Interestingly, increase in the
spacer length leads to loss of sensitivity; GC-2b, the Gb3
analogue with a 12-carbon-atom spacer bound to Stx1 with
lower affinity than GC-2a. The binding studies involving the
N-acetylated analogue GC-3b were very intriguing. Increase
in spacer length from 6 (GC-3a) to 12 carbon atoms (GC-3b)
led to loss of selectivity and sensitivity; GC-3b bound to both
Stx2 and Stx1 with equal affinity. Clearly, the role of the
spacer in the binding event remains to be determined.
Premier EHEC ELISA(Meridian Bioscience, Inc.), and the
assay was carried out as described before. As seen in
Figure 3C, Stx1 is efficiently captured without any interfer-
ence from a complex sample, and similar results were seen for
Stx2 (data not shown). We are particularly excited by these
results because the ligands could readily be applied to any
streptavidin-based biosensor platform to detect toxin sero-
types during the narrow window of opportunity of around
3 days between the onset of watery diarrhea and HUS.[1] We
also tested the ability of these glycoconjugates to bind to
Stx2c as this variant has been found in human clinical
samples.[24,25] Stx2c binds to GC-1 with higher affinity than
Stx2 (Figure 3D). Unlike Stx1 and Stx2, which are 55%
homologous, the emerging Stx2c variant differs from Stx2 by
only a few amino acids,[24,26] and the ability of these
glycoconjugate derivatives to recognize small changes is
remarkable.
In summary, our results indicate that it is feasible to
develop highly selective and sensitive synthetic glycoconju-
gate-based Shiga toxin detection reagents by introducing
simple manipulations in the structure of known saccharide
Next, we tested the ability of GC-2a to capture Stx1 in
clinical applications. Stx1 was spiked into a stool, diluted as
recommended for the commercially available diagnostic
Angew. Chem. Int. Ed. 2008, 47, 1265 –1268
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1267

Communications
[1] P. I. Tarr, C. A. Gordon, W. L. Chandler, Lancet 2005, 365, 1073.
[2] R. L. Siegler, J. Am. Med. Assoc. 2003, 290, 1379.
[3] C. S. Wong, S. Jelacic, R. L. Habeeb, S. L. Watkins, P. I. Tarr, N.
Engl. J. Med. 2000, 342, 1930.
[4] P. I. Kitov, J. M. Sadowska, G. Mulvey, G. D. Armstrong, H.
Ling, N. S. Pannu, R. J. Read, D. R. Bundle, Nature 2000, 403,
669.
[5] G. L. Mulvey, P. Marcato, P. I. Kitov, J. Sadowska, D. R. Bundle,
G. D. Armstrong, J. Infect. Dis. 2003, 187, 640.
[6] M. Watanabe, K. Matsuoka, E. Kita, K. Igai, N. Higashi, A.
Miyagawa, T. Watanabe, R. Yanoshita, Y. Samejima, D. Ter-
unuma, Y. Natori, K. Nishikawa, J. Infect. Dis. 2004, 189, 360.
[7] S. D. Gamage, A. K. Patton, J. F. Hanson, A. A. Weiss, Infect.
Immun. 2004, 72, 7131.
[8] E. Strauch, R. Lurz, L. Beutin, Infect. Immun. 2001, 69, 7588.
[9] Y. Miura, Y. Sasao, H. Dohi, Y. Nishida, K. Kobayashi, Anal.
Biochem. 2002, 310, 27.
[10] E. N. Kitova, P. I. Kitov, E. Paszkiewicz, J. Kim, G. L. Mulvey,
G. D. Armstrong, D. R. Bundle, J. S. Klassen, Glycobiology 2007,
17, 1127.
[11] H. Dohi, Y. Nishida, M. Mizuno, M. Shinkai, T. Kobayashi, T.
Takeda, H. Uzawa, K. Kobayashi, Bioorg. Med. Chem. 1999, 7,
2053.
[12] P. M. St. Hilaire, M. K. Boyd, E. J. Toone, Biochemistry 1994, 33,
14452.
[13] P. I. Kitov, H. Shimizu, S. W. Homans, D. R. Bundle, J. Am.
Chem. Soc. 2003, 125, 3284.
[14] P. Boerlin, S. A. McEwen, F. Boerlin-Petzold, J. B. Wilson, R. P.
Johnson, C. L. Gyles, J. Clin. Microbiol. 1999, 37, 497.
[15] R. L. Siegler, T. G. Obrig, T. J. Pysher, V. L. Tesh, N. D. Denkers,
F. B. Taylor, Pediatr. Nephrol. 2003, 18, 92.
Figure 3. A,B) Differential binding of Shiga toxin variants to synthetic
glycoconjugates. C) Ability of GC-2a to capture Stx1 in human stool.
D) Differential binding to GC-1. Biotinylated glycoconjugates were
added to streptavidin-coated microtiter wells and incubated with
decreasing concentrations of Stx. Binding was determined by using
anti-Stx polyclonal antibody and alkaline phosphatase conjugated goat
antirabbit secondary antibody. Results are the average of three
independent trials.
[16] N. W. Rutjes, B. A. Binnington, C. R. Smith, M. D. Maloney,
C. A. Lingwood, Kidney Int. 2002, 62, 832.
receptors. These glycoconjugates are highly robust, inexpen-
sive, require no refrigeration, can be scaled up, applied to any
sensor platform, and bind to toxins present in real-world
samples. Finally, the strategy described herein relies on
receptor-mediated recognition events, as opposed to anti-
gen-based recognition. Receptor-binding strategies are par-
ticularly useful to detect antigenic variants since amino acid
changes may change antibody binding without changing
function; however, receptor binding cannot be varied without
loss of function. Developing a focused glycoconjugate library
and correlating binding to toxicity is the subject of our current
endeavors and will be reported soon.
[17] R. Atkinson, G. Johnson, T. Root, T. Halse, D. Wroblewski, M.
Davies, A. Byrd, L. Long, L. Demma, F. Angulo, C. Bopp, P.
Gerner-Smidt, N. Strockbine, K. Greene, B. Swaminathan, P.
Griffin, J. Schaffzin, B. Goode in Morbidity and Mortality
Weekly Report, Vol. 55, CDC, Atlanta, GA, 2006, p. 1042.
[18] S. D. Gamage, C. M. McGannon, A. A. Weiss, J. Bacteriol. 2004,
186, 5506.
[19] M. R. Leslie, H. Parolis, L. A. Parolis, Carbohydr. Res. 1999, 323,
103.
[20] K. Hofmann, Y. Kiso, Proc. Natl. Acad. Sci. USA 1976, 73, 3516.
[21] S. L. Sabesan, Raymond U, Can. J. Chem. 1984, 62, 644.
[22] K. M. Koeller, M. E. Smith, C. H. Wong, Bioorg. Med. Chem.
2000, 8, 1017.
[23] R. R. Kale, C. M. Clancy, R. M. Vermillion, E. A. Johnson, S. S.
Iyer, Bioorg. Med. Chem. Lett. 2007, 17, 2459.
[24] S. Persson, K. E. Olsen, S. Ethelberg, F. Scheutz, J. Clin.
Microbiol. 2007, 45, 2020.
[25] C. Burk, R. Dietrich, G. Acar, M. Moravek, M. Bulte, E.
Martlbauer, J. Clin. Microbiol. 2003, 41, 2106.
[26] H. Ling, N. S. Pannu, A. Boodhoo, G. D. Armstrong, C. G. Clark,
J. L. Brunton, R. J. Read, Structure 2000, 8, 253.
Received: August 11, 2007
Revised: November 19, 2007
Published online: January 2, 2008
Keywords: biotin · differentiation · glycoconjugates · glycosides ·
.
Shiga toxin
1268
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1265 –1268