Glycan array on aluminum oxide-coated glass slides through phosphonate chemistry

Article abstract of DOI:10.1021/ja1046523

A new type of glycan array covalently or noncovalently attached to aluminum oxide-coated glass (ACG) slides has been developed for studies of enzymatic reactions and protein binding. To prepare the noncovalent array, glycans with a polyfluorinated hydrocarbon (-C₈F₁₇) tail are spotted robotically onto the ACG slide surface containing a layer of polyfluorinated hydrocarbon terminated with phosphonate. After incubation and washing, the noncovalent array can be characterized by MS-TOF via ionization/desorption at a low laser energy without addition of matrix. A representive cellotetraose array was developed to study the activity and specificity of different cellulases and to differentiate the exo-and endoglucanase activities. To prepare the covalent array, glycans with a phosphonic acid tail were synthesized and spotted robotically onto the ACG slide surface. After incubation, the slides can be used directly for quantitative protein binding analysis. Compared to the preparation of glycan arrays on glass slides and other surfaces, this method of arraying using phosphonic acid reacting with ACG is more direct, convenient, and effective and represents a new platform for the high-throughput analysis of protein-glycan interactions.

Full text of DOI:10.1021/ja1046523

Published on Web 09/07/2010
Glycan Array on Aluminum Oxide-Coated Glass Slides
through Phosphonate Chemistry
Shih-Huang Chang,†,‡ Jeng-Liang Han,† Susan Y. Tseng,† Hsin-Yu Lee,†
Chin-Wei Lin,† Yu-Chen Lin,†,§ Wen-Yih Jeng,⊥ Andrew H.-J. Wang,⊥
Chung-Yi Wu,*,† and Chi-Huey Wong*,†
Genomics Research Center, Academia Sinica, 128 Academia Road, Section 2, Nankang,
Taipei 115, Taiwan, Institute of Biochemical Sciences, National Taiwan UniVersity, Taipei,
Taiwan, Department of Chemistry, National Taiwan UniVersity, Taipei, Taiwan, and Institute of
Biological Chemistry, Academia Sinica, 128 Academia Road, Section 2, Nankang,
Taipei 115, Taiwan
Received May 28, 2010; E-mail: cyiwu@gate.sinica.edu.tw; chwong@gate.sinica.edu.tw
Abstract: A new type of glycan array covalently or noncovalently attached to aluminum oxide-coated glass
(ACG) slides has been developed for studies of enzymatic reactions and protein binding. To prepare the
noncovalent array, glycans with a polyfluorinated hydrocarbon (-C8F17) tail are spotted robotically onto
the ACG slide surface containing a layer of polyfluorinated hydrocarbon terminated with phosphonate.
After incubation and washing, the noncovalent array can be characterized by MS-TOF via ionization/
desorption at a low laser energy without addition of matrix. A representive cellotetraose array was developed
to study the activity and specificity of different cellulases and to differentiate the exo- and endoglucanase
activities. To prepare the covalent array, glycans with a phosphonic acid tail were synthesized and spotted
robotically onto the ACG slide surface. After incubation, the slides can be used directly for quantitative
protein binding analysis. Compared to the preparation of glycan arrays on glass slides and other surfaces,
this method of arraying using phosphonic acid reacting with ACG is more direct, convenient, and effective
and represents a new platform for the high-throughput analysis of protein-glycan interactions.
high affinity and specificity.6 Recently, glycan arrays have been
developed to mimic their cell-surface display for the study of
Introduction
carbohydrate-receptor interactions.7-16 Multiple proteins or
glycans can be quantitatively and qualitatively analyzed once
in a monovalent or multivalent manner.17-22 However, most
of glycan arrays used today are difficult to characterize and
Carbohydrates often exist on cell surfaces as glycoproteins
or glycolipid conjugates and play important structural and
functional roles in numerous biological recognition processes.1-4
In addition, natural products often require glycosylation for
biological activities, and many organisms diversify functions
and detoxify harmful exogenous xenobiotics through glycosy-
lation.5 However, the study of carbohydrate-protein interactions
still encounters much difficulty, largely because of the inac-
cessibility of such structurally complex molecules and their low-
affinity interactions with glycan-binding proteins (GBPs).
Typically the monomeric dissociation constant (KD) in a
carbohydrate-protein interaction is in the millimolar range; thus,
carbohydrate-mediated biological responses are often through
multivalent interactions on the cell surface in order to achieve
(6) Weis, W. I. Annu. ReV. Biochem. 1996, 65, 441–473.
(7) Park, S.; Shin, I. Angew. Chem., Int. Ed. 2002, 41, 3180–3182.
(8) Houseman, B. T.; Mrksich, M. Chem. Biol. 2002, 9, 443–454.
(9) Wang, D.; Liu, S.; Trummer, B. J.; Deng, C.; Wang, A. Nat.
Biotechnol. 2002, 20, 275–281.
(10) Fukui, S.; Feizi, T.; Galustian, C.; Lawson, A. M.; Chai, W. Nat.
Biotechnol. 2002, 20, 1011–1017.
(11) Love, K. R.; Seeberger, P. H. Angew. Chem., Int. Ed. 2002, 41, 3583–
3586.
(12) Willats, W. G. T.; Rasmussen, S. E.; Kristensen, T.; Mikkelsen, J. D.;
Knox, J. P. Proteomics 2002, 2, 1666–1671.
(13) Mellet, C. O.; Garcia´ Ferna´ndez, J. M. ChemBioChem 2002, 2, 819–
822.
† Genomics Research Center, Academia Sinica.
‡ Institute of Biochemical Sciences, National Taiwan University.
§ Department of Chemistry, National Taiwan University.
⊥ Institute of Biological Chemistry, Academia Sinica.
(1) Apweiler, R.; Hermjakob, H.; Sharon, N. Biochem. Biophys. Acta 1999,
1473, 4–8.
(14) Paulson, J. C.; Blixt, O.; Collins, B. E. Nat. Chem. Biol. 2006, 2,
238–248.
(15) Feizi, T.; Fazio, F.; Chai, W.; Wong, C.-H. Curr. Opin. Struct. Biol.
2003, 13, 637–645.
(16) Shin, I.; Park, S.; Lee, M. Chem.sEur. J. 2005, 11, 2894–2901.
(17) Liang, P. H.; Wu, C.-Y.; Greenberg, W. A.; Wong, C.-H. Curr. Opin.
Chem. Biol. 2008, 12, 86–92.
(2) Raman, R.; Raguram, S.; Venkataraman, G.; Paulson, J. C. Nat.
Methods 2005, 2, 817–824.
(18) Wu, C.-Y.; Liang, P.-H.; Wong, C.-H. Org. Biomol. Chem. 2009, 7,
2247–2254.
(3) Fukuda, M.; Hindsgaul, O. Molecular and Cellular Glycobiology;
Oxford, University Press: Oxford, 2000.
(19) Liang, C.-H.; Wu, C.-Y. Expert ReV. Proteomics 2009, 6, 631–645.
(20) Liu, Y.; Palma, A. S.; Feizi, T. Biol. Chem. 2009, 390, 647–656.
(21) Park, S.; Lee, M.-R.; Shin, I. Chem. Commun. 2008, 4389–4399.
(22) Oyelaran, O.; Gildersleeve, J. C. Curr. Opin. Chem. Biol. 2009, 13,
406–413.
(4) Seeberger, P. H. Nature 2005, 437, 1239.
(5) Thorson, J. S.; Vogt, T. In Glycosylated Natural Products in
Carbohydrate-Based Drug DiscoVery; Wong, C.-H., Ed.; Wiley-VCH,
Weinheim, 2002; pp 685-712.
9
10.1021/ja1046523  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 13371–13380 13371

A R T I C L E S
Chang et al.
Figure 1. Chemical structures of compounds used in this study. Aminopropyltriethoxysilane (APTES, 1); N-succinimidyl 4,4,5,5,6,6,7,7,8,8,9,9,10,
10-,11,11,11-heptadecafluoro-undecyl carbonate 2; (3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-heptadecafluorodecyl)phosphonic acid (HDFDPA, 3), 4-10 are
polyfluorinated derivatives of mannose (4), Gb5 (5), lactose (6), Globo H (7), cellobiose (8), cellolotriose (9), cellotetraose (10), and mannose with a
phosphonic acid group (11). Detailed synthetic procedures of compounds 3-11 and their 1H and 13C NMR spectra are shown in the Supporting Information.
quantify and thus lack of quality control. To tackle this problem,
a glycan array was developed using 2,6-diaminopyridine (DAP)
as a fluorescent linker to sugars and printed on epoxy-activated
slides.23-25 Another approach is to combine the array method
and mass spectrometry for label-free analysis. For example,
Mrksich’s group has developed a method using self-assembled
monolayer for matrix-assisted laser desorption ionization time-
of-flight mass spectrometry (SAMDI-TOF MS) to observe and
optimize synthesis of glycans on gold surface.26,27 This strategy
was further used by the Flitsch group to monitor glycosyltrans-
ferases involved in protein O-glycosylation.28,29 To avoid
nonspecific protein adsorption on the gold surface, a tri(ethylene
glycol) spacer was used. Since the nature of adducts and the
degree of fragmentation of the parent molecular ions were
dependent on the laser fluence rate (laser energy),30,31 addition
of matrix is often required to reduce fragmentation of the parent
molecular ion. However, the signal of matrix may interfere with
the analyst spectrum, especially when the analyte is a small
molecule. We have recently developed a desorption/ionization
on silicon mass spectrometry (DIOS-MS) method to characterize
the oligosaccharides that are covalently bound to porous silicon
with a built-in photocleavable linker.32 In addition, a technique
based on nanostructure-initiator mass spectrometry (NIMS) was
developed to characterize noncovalent glycan arrays using
fluorous tagged substrates that are physically adhered to the
fluorohydrocarbon-coated porous silicon surface.33 However,
incorporation of a photocleavable linker to individual glycans
is quite tedious, and preparation of porous silicon plates requires
corrosive acids which may not generate uniform results. Another
disadvantage is that covalent glycan arrays on conventional glass
slides cannot be characterized directly by mass spectrometry.
We have recently developed a new type of aluminum oxide-
coated glass slides (ACG) which can be functionalized by
(23) Xia, B.; Kawar, Z. S.; Ju, T.; Alvarez, R. A.; Sachdev, G. P.;
Cummings, R. D. Nat. Methods 2005, 2, 845–850.
(24) Song, X.; Jia, B.; Lasanajak, Y.; Smith, D. F.; Cummings, R. D.
Glycoconjugate J. 2008, 25, 15–25.
(25) Song, X.; Xia, B.; Stowell, S. R.; Lasanajak, Y.; Smith, D. F.;
Cummings, R. D. Chem. Biol. 2009, 16, 36–47.
(26) Ban, L.; Mrksich, M. Angew. Chem., Int. Ed. 2008, 47, 3396–3399.
(27) (a) Mrksich, M. ACS Nano 2008, 2, 7–18. (b) Gurard-Levin, Z. A.;
Mrksich, M. Annu. ReV. Anal. Chem. 2008, 1, 767–800.
(28) Laurent, N.; Voglmeir, J.; Write, A.; Blackburn, J.; Pham, N. T.; Wong,
S. C. C.; Gaskell, S. J.; Flitsch, S. L. ChemBioChem 2008, 9, 883–
887.
(30) Trevor, J. L.; Lykke, K. R.; Pellin, M. J.; Hanley, L. Langmuir 1998,
14, 1664–1673.
(31) Gong, W.; Elitzin, V. I.; Janardhanam, S.; Wilkins, C. L.; Fritxch, I.
J. Am. Chem. Soc. 2001, 123, 769–770.
(32) Lee, J.-C.; Wu, C.-Y.; Apon, J. V.; Siuzdak, G.; Wong, C.-H. Angew.
Chem., Int. Ed. 2006, 45, 2753–2757.
(29) (a) Laurent, N.; Voglmeir, J.; Flitsch, S. L. Chem. Commun. 2008,
4400–4412. (b) Laurent, N.; Haddoub, R.; Flitsch, S. L. Trends
Biotechnol. 2008, 26, 328–337.
(33) Northen, T. R.; Lee, J.-C.; Hoang, L.; Raymond, J.; Hwang, D. R.;
Yannone, S. M.; Wong, C.-H.; Siuzdak, G. Proc. Natl. Acad. Sci.
U.S.A. 2008, 105, 3678–3683.
9
13372 J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010

Glycan Array To Study Protein-Glycan Interactions
A R T I C L E S
Figure 2. Preparation of noncovalent glycan array on the Teflon-like ACG slides.
Scheme 1. Synthesis of Polyfluorinated Carbohydrates 4-7
reacting with an alkyl monoethoxysilane and linked to a glycan
containing the photocleavable linker.34 The glycan array with
a photocleavable linker on the ACG slide surfaces can be
characterized by time-of-flight mass spectrometry (MS-TOF)
without matrix. However, the array often gave a relatively low
signal-to-noise (S/N) ratio, most likely due to incomplete
photocleavage. Here, we reported a new method for the
preparation of stable polyfluorinated ACG slides through
phosphonate chemistry, which was then noncovalently arrayed
pH (4-6).36 In this study, the glycan array is subject to the
enzymatic hydrolysis at pH 4-6, and the hydrolyzed products
remaining on the slide surface can then be identified by MS-
TOF in situ. This study has led to the development of a new
assay to differentiate the exo- and endocellulase activities using
cellotetraose as a substrate. Moreover, a new covalent glycan
array using synthetic glycans with a phosphonic acid tail directly
reacting with the aluminum oxide surface has been developed.
with glycans for the study of cellulase activities by MS-TOF.
Cellulases are of current interest because of their potential
application to biofuel production.35 Cellobiose or cellotriose with
a fluorogenic or chromogenic group has been commonly used
as substrate in the investigation of cellulase activity and
specificity.36 However, the fluorogenic or chromogenic leaving
group generated in the enzymatic reaction only showed signal
at high pH, but cellulases exhibit their optimum activity at low
Results and Discussions
Preparation of Teflon-like ACG Slides. To prepare the Teflon-
like ACG slides, triethoxysilane 1 and phosphonic acid deriva-
tives 3 (Figure 1) were synthesized to react with the oxidized
aluminum surface. Fabrication of the silane-based slide involves
a two-step chemical reaction: functionalization of the aluminum
oxide surface using compound 1 as a grafting reagent under a
moisture-free environment, followed by the amide bond forma-
(34) Tseng, S. Y.; Wang, C.-C.; Lin, C.-W.; Chen, C.-L.; Yu, W.-Y.; Chen,
C.-H.; Wu, C.-Y.; Wong, C.-H. Chem. Asian J. 2008, 3, 1395–1405.
(35) (a) Zhang, Y. H. P.; Himmel, M. E.; Mielenz, J. R. Biotechnol. AdV.
2006, 24, 452–81. (b) Nieves, R. A.; Ehrman, C. I.; Adney, W. S.;
Elander, R. T.; Himmel, M. E. World J. Microbiol. Biotechnol. 1998,
14, 301–304.
(36) (a) Coleman, D. J.; Studler, M. J.; Naleway, J. J. Anal. Biochem. 2007,
371, 146–153. (b) Chernoglazov, V. M.; Jafarova, A. N.; Klyosov,
A. A. Anal. Biochem. 1989, 179, 186–189. (c) Deshpande, M. V.;
Eriksson, K. E.; Pettersson, L. G. Anal. Biochem. 1984, 138, 481–
487.
9
J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010 13373

A R T I C L E S
Chang et al.
tion between NHS-activated polyfluorohydrocarbon 2 and the
amine group on the surface of the slide. On the other hand,
since the phosphonic acid group could be conjugated to
aluminum oxide directly under mild conditions,37 a Teflon-like
ACG slide was made in a one-step chemical reaction, using an
aqueous solution of 3 and a cleaned aluminum oxide surface
under sonication to form a monolayer of perfluorophosphonate
on the surface.38 The covalent formation of phosphonic acid
was confirmed by FTIR (see Supporting Information (SI) Figure
S1.).35,38 These two types of slides were checked with MS-
TOF spectrometry for background test, and both showed a
cleaner baseline when compared to that with the previous
method.34 (see SI Figures S2-a and S3-a). The one-step method
using 3, however, was much more convenient and efficient.
Noncovalent Glycan Array on the Teflon-like ACG Slides. With
this encouraging result, we used the easily prepared phosphonate
slide to create a noncovalent glycan array (Figure 2). Mannose
with an amine linker 1234 was reacted with compound 2 to form
the polyfluorinated (-C8F17) derivative 439 (Scheme 1). A
solution of this sugar derivative was spotted robotically40 onto
the Teflon-like ACG slide surface. After incubation, the slides
were rinsed repeatedly with distilled water and subjected to MS-
TOF analysis. A very clean mass spectrum was obtained,
showing the peaks at 806 and 822 as the sodium and potassium
adducts, respectively (SI, Figure S3-b). The same slide was then
used for protein-binding analysis using Alexa 488-labeled
concanavalin A as the protein source (SI, Figure S3-c). Both
experiments showed satisfactory results. To further extend the
scope of this type of glycan array, compounds 13-1541 were
used for the synthesis of polyfluorinated lactose 5, SSEA-3
(Gb5) 6, and Globo H7 (Scheme 1), which were then im-
mobilized onto the Teflon-like ACG slide surface for mass
analysis and protein-binding evaluations.
In the MS analysis without adding additional matrix, no
fragmented signal was found, even with the use of such a labile
sugar, Globo H (Figure 3A), compared to the MALDI-TOF
analysis of Globo H where fragmentation, especially through
the cleavage of fucosidic linkage, is often observed.32 Parts B
and C of Figure 3 showed that these glycan arrays had retained
their sugar-protein binding patterns35 when VK-9 (a mouse
anti-Globo H monoclonal antibody) and anti-SSEA-3 mono-
clonal antibody were used for binding studies, respectively.
We further investigated the effect of laser fluence rate (laser
energy) and matrix on this new surface. In the absence of matrix,
we obtained a high S/N ratio (22) at a very low laser fluence
rate (9%), and no fragmented signal was found (Table 1). When
the fluence rate was increased to 10%, the S/N ratio was
enhanced to 40 without any fragmentation (original mass
spectrometry was shown in SI, Figure S4). We also added DHB
(2,5-dihydroxybenzoic acid) as a matrix to check the matrix
Figure 3. Two types of assays for noncovalent glycan array on the Teflon-
like ACG slide. (A) MALDI mass spectrometric analysis of polyfluorinated
Globo H 7 (MW: 1604.40), Gb5 6 (MW: 1458.39), and lactose 2 (MW:
932.21) immobilized on a Teflon-like ACG slide as their sodium adducts
[M + Na]+ at 1627.44, 1481.39, and 954.39 respectively. The fluence rate
is 12% without matrix addition. Protein-binding assay of (B) Globo H/VK9/
anti-VK9-Cy3 and (C) Gb5/anti-SSEA3-A488. The matrix was a 10 × 10
(100 spots) array of perfluorinated Globo H (left four columns), lactose
(fifth and sixth columns, served as the negative control), and Gb5 (right
four columns). VK9 is a monoclonal antibody against Globo H, and anti-
SSEA3 is a monoclonal antibody against Gb5. Both Cy3 and A488 are
fluorescent probes. The ACG with polyfluorinated phosphate surface was
used for arraying.
effect on this new surface. To our surprise, the S/N ratio is only
7.3 when the fluence rate is 11%. When the fluence rate was
increased to 13%, the S/N ratio was 61 with 6% fragmentation
signal (SI, Figure S5). Apparently, DHB matrix cannot enhance
the signal with a low laser fluence rate. However, it can enhance
the S/N ratio and reduce the fragmentation when the fluence
rate is over 25% (Table 1).
Cellulase Activity Studies. From the above studies, we have
found that polyfluorinated carbohydrates immobilized on the
Teflon-like ACG slide can be easily characterized by ionization/
desorption at low laser energy and the signals can be obtained
with high S/N ration without fragmentation. These results
encouraged us to further study enzyme specificity and activity:
enzymatic hydrolysis of the immobilized polyfluorinated cel-
lobiose 8 was conducted on the phosphonic acid slide surface.
Three commercially available cellulases, Aspergillus niger (A.
(37) Grundmeier, G.; Schmidt, W.; Stratmann, M. Electrochin. Acta 2000,
45, 2515–2533.
(38) McNatt, J. S.; Morgan, J. M.; Farkas, N.; Ramsier, R. D.; Young,
T. L.; Rapp-Cross, J.; Espe, M. P.; Robinson, T. R.; Nelson, L. Y.
Langmuir 2003, 19, 1148–1153.
(39) (a) Ko, K. S.; Jaipuri, F. A.; Pohl, N. L. J. Am. Chem. Soc. 2005,
127, 13162–13163. (b) Mamidyala, S. K.; Ko, K. S.; Jaipuri, F. A.;
Park, G.; Pohl, N. L. J. Fluorine Chem. 2006, 127, 571. (c) Jaipuri,
F. A.; Collet, B. Y.; Pohl, N. L. Angew. Chem., Int. Ed. 2008, 47,
1707–1710. (d) Chen, G. S.; Pohl, N. L. Org. Lett. 2008, 10, 785–
788.
(40) Blixt, O.; et al. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 17033–17038.
(41) Wang, C.-C.; Huang, Y.-L.; Ren, C.-T.; Lin, C.-W.; Hung, J.-T.; Yu,
J.-C.; Yu, A.-L.; Wu, C.-Y.; Wong, C.-H. Proc. Natl. Acad. Sci. U.S.A.
2008, 105, 11661–11666.
9
13374 J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010

Glycan Array To Study Protein-Glycan Interactions
A R T I C L E S
Table 1. S/N Ratio of Globo H under Different Laser Fluence Rate
enzyme solution (100 µL) was added to the solution of
fluorinated cellobiose (100 µL of 0.5 mM) in Eppendorfs. After
reaction, the solution was transferred to the empty wells of the
same slide. Each well was rinsed separately three times with
deionized water, and the slide was dried again under high
vacuum before MS-TOF assay.
with or without Matrix (DHB) Addition
matrix-assisted desorption/ionization
matrix-free desorption/ionization
fluence
rate (%)a
frag.
S/Nc
fluence
rate (%)a
frag.
S/Nc
GH S/Nb
GH S/Nb
11
13
14
15
18
19
25
35
7.3
61
157
0
6
13
25
87
108
168
125
9
10
12
16
18
19
25
35
22
40
0
0
5
10
17
64
356
373
In the MS-TOF analysis, the enzyme from T. Viride func-
tioned the best for the hydrolysis of cellobiose derivative in
solution, and that from T. reesei hydrolyzed disaccharide most
effectively among the three enzyme sources on the slide. In
solution, the enzyme from A. niger seemed to also act as a
synthetase, and the overall reaction produced 8% of cellotriose
(MW 1093) which was detected as sodium adduct at m/z of
1116.3 [M + Na]+ (SI, Figure S6A). The enzyme from A. niger
was characterized as a typical endo-type cellulose42 which
cleaved five glucose units in length at a time. It did not
hydrolyze either cellobiose or p-nitrophenyl-ꢀ-D-glucoside.43 To
further understand the mode of action, polyfluorinated (-C8F17)
cellotriose 9 was subjected to enzymatic hydrolysis on the
Teflon-like ACG slide. Using the same analytical procedure,
the results (Figure 4) indicated that the enzyme from T. reesei
218
275
375
741
2184
1512
316
1690
1956
3128
1445
a Fluence: the laser power (or fluence rate) applied to the slide
surface. b GH S/N: the signal/noise ratio for Globo H. c Frag. S/N: the
signal/noise ratio for the peak of fragmentation of Globo H.
niger), Trichoderma reesei (T. reesei), and Trichoderma Viride
(T. Viride) were prepared separately at 5 U/mL in a sodium
acetate (25 mM) buffer solution (pH 5), and aliquots were loaded
onto the functionalized slide divided into 16 wells using Fast
Frame reaction chambers. For comparison, an aliquote of
Figure 4. Study of cellulase activity using glycan array combined with mass spectrometry. (A) Hydrolyzed fragments of cellotriose derivatives remained
on the Teflon-like ACG slide surface. (B) MS-TOF data of enzymatic hydrolysis of polyfluorinated cellotriose of the control run without enzymes (a), and
with the cellulase proteins from A. niger (b), T. reesei (c), and A. Viride (d).
9
J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010 13375

A R T I C L E S
Chang et al.
Figure 5. MS-TOF data of enzymatic hydrolysis of different polyfluorinated oligo-cellulose in solution. Polyfluorinated cellobiose 8 (A), polyfluorinated
cellotriose 9 (B), and polyfluorinated cellotetraose 10 (C); the control run without enzymes (a), with exoglucanase (L3) (b), and with endoglucanase 44A (c).
hydrolyzed the cellotriose substrate most efficiently among the
enzymes from three different species. The results (Figure 4B-b
and SI, Figure S6) also indicated that A. niger does not function
well in hydrolyzing cellobiose or cellotriose on the Teflon-like
ACG slide surface. However, in solution, A. niger hydrolyzed
polyfluorinated cellobiose 8 and cellotriose 9 with one glucose
unit at a time at a very slow reaction rate (SI, Figure S6, A-b).
Since this commercial enzyme is not pure, these phenomena
may be due to the contamination of a small quantity of
ꢀ-glucosidase in the mixture.
The cellulase from T. Viride was known to effectively degrade
the newspaper material,44 and T. reesei45 was known to
hydrolyze the crystalline form of cellulose. In general, the
enzymatic hydrolysis on the ACG slide surface is more site-
specific but much slower than that in solution.
Study of Cellulase Specificity Using Glycan Array
Combined with Mass Spectrometry. Cellulases are usually
divided into several subclasses of isozymes on the basis of their
function: ꢀ-glucosidases (EC 3.2.1.21]) cleave cellobiose into
individual glucose units,46 exoglucanases (1,4-ꢀ-D-glucan cel-
lobiohydrolase [EC 3.2.1.91]) cleave cellobiose units from the
nonreducing end of the cellulose chain,47,48 and endoglucanases
(endo-1,4-ꢀ-D-glucanase [EC 3.2.1.4]) cleave the chain randomly
at internal positions46-48 creating new ends for exo-glucanases.48
HPLC analysis of the products of hydrolysis from MUF-
glucosides is often used to determine the hydrolytic specificity
of these purified enzymes.36b From the above results, the glycan
array appeared to be a convenient platform for studying the
specificity of various types of cellulases.
tions,49,50 and enzymatic hydrolysis reactions were conducted
in solution with the purified enzymes using substrates 8, 9 and
10. At the completion of the reaction, the solution mixtures were
transferred to the Teflon-like ACG slide and subjected to MS-
TOF assay after washing. Exoglucanase (L3) cleaved the
cellobiose unit from the end of cellobiose substrate 8, but the
reaction was very slow (Figure 5A-b). However, the enzyme
cleaved cellobiose quickly when cellotriose substrate 9 or
cellotetraose 10 was used as the substrate (Figure 5B-b and 5C-
b), consistent with the definition of exoglucanase. Endoglucanase
44A could not accept cellobiose substrate 8 or cellotriose
substrate 9 (Figure 5A-c and 5B-c) but could cleave the
trisaccharide or tetrasaccharide unit of cellotetraose substrate
10 (Figure 5C-c). For comparison, we also immobilized
substrates 8, 9, and 10 on the Teflon-like ACG surface and
carried out the enzymatic hydrolysis on the surface directly.
After the same washing procedures, these slides were subjected
to MS-TOF assay. Exoglucanase (L3) only weakly cleaved the
cellobiose substrate 8 after 24 h incubation (Figure 6A-b) but
would cleave the cellobiose unit quickly with 9 (Figure 6B-b)
or 10 (Figure 6C-b). The endoglucanase 44A could not accept
cellobiose 8 or cellotriose 9 as a substrate (Figures 6A-c and
6B-c). However, it cleaved cellotetraose quickly when 10 was
used as substrate (Figure 6C-c). In contrast to the hydrolysis
reaction in solution, endoglucanse 44A cleaved cellotetraose,
when enzymatic hydrolysis was carried out on the surface, but
could not cleave cellotriose, when cellotetraose 10 was used as
substrate. These results suggested that cellotetraose 10 can be
used to create an array on the Teflon-like ACG surface to
identify endo- and exocellulase activities. Comparing to the
HPLC method, this array method is more effective and
convenient and should facilitate the discovery of new endo- and
exocellulases.
In order to develop a better method to study enzyme
specificity, we purified the exoglucanase (L3) and endoglucanase
(44A) according to the literature methods with minor modifica-
(42) Uzcategui, E.; Ruiz, A.; Montesino, R.; Johansson, G.; Pettersson, G.
J. Biotechnol. 1991, 19, 271–285.
Creation of Covalent Glycan Array on ACG Slides. Several
types of functionalized glass slides are commercially available
and have been used for covalent glycan array preparation,
including glass slides coated with amine, carboxylate, N-
hydroxysuccinimide (NHS), avidin, epoxy, aldehyde, chelating
(43) Okada, G. Agric. Biol. Chem. 1985, 49, 1257–1265.
(44) van Wyk, J. P. H.; Mogale, M. A.; Seseng, T. A. Biotechnol. Lett.
2000, 22, 491–494.
(45) Fadel, M.; Kahil, T. Egypt. J. Biotechnol. 2004, 16, 74–94.
(46) Boschker, H. T. S.; Cappenberg, T. E. Appl. EnViron. Microbiol. 1994,
60, 3592–3596.
(47) Hilden, L.; Johannson, G. Biotechnol. Lett. 2004, 26, 1683–1693.
(48) Mun˜oz, I.; Ubhayasekera, W.; Henriksson, H.; Szabo, I.; Pettersson,
G.; Johansson, G.; Mowbray, S.; Stahlberg, J. J. Mol. Biol. 2001, 314,
1097–1111.
(49) Schubot, F. D.; Kataeva, I. A.; Chang, J.; Shah, A. K.; Ljungdahl,
L. G.; Rose, J. P.; Wang, B. C. Biochemistry 2004, 43, 1163–1170.
(50) Kitago, Y.; Karita, S.; Watanabe, N.; Kamiya, M.; Aizawa, T.; Sakka,
K.; Tanaka, I. J. Biol. Chem. 2007, 282, 35703–35711.
9
13376 J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010

Glycan Array To Study Protein-Glycan Interactions
A R T I C L E S
Figure 6. MS-TOF data of enzymatic hydrolysis of different polyfluorinated oligo-cellulose on Teflon-like ACG slide. Polyfluorinated cellobiose 8 (A),
polyfluorinated cellotriose 9 (B), and polyfluorinated cellotetraose 10 (C) immobilized on Teflon-like ACG slide; the control run without enzymes (a), with
exoglucanase (L3) (b), and with endoglucanase 44A (c) for 24 h incubation.
Scheme 2. Synthesis of Gb3 with Phosphonic Acid Linker Compound 22
nickel group, and others.19 When creating a glycan array on
these surfaces, a suitable buffer and repeated blocking and
washing steps are needed. Our previous studies showed that
phosphonic acid could react with the ACG slide directly and
the resulting array could tolerate repeating washing steps. We
therefore tried to use this property to develop covalent glycan
arrays using mannose phosphonic acid derivative 11 (Figure 1,
see Scheme S5 (SI), for detailed synthetic procedures) as an
example. It was found that the coupling reaction was much faster
than on the traditional N-succinimidyl glass slide surface, with
the incubation time reduced from 12 h to less than 15 min. In
addition, the ACG microarray showed great improvement on
signal-to-noise ratio since there was less light scattering to
interfere with the fluorescence signal (Figure S7 (SI)).
After the mannose compound with different concentrations
(Figure S8 (SI)) was spotted on the slides, they were rinsed
repeatedly with distilled water and used for protein binding
analysis, with Alexa 488-labeled concanavalin A. The incubation
process was continued for another hour; then, it was washed
again to prevent nonspecific binding between the lectin and the
slide surface.
Encouraged by this results, we then established a procedure
for the synthesis of complex oligosaccharides, including Gb3
and Globo H with a phosphonic acid group in the tail. As
illustrated in Scheme 2, condensation of the lactosyl thiocresol
16 with the phosphate linker 17 using NIS/TfOH as a promoter51
afforded ꢀ-lactosyl derivative 18 in 80% yield. Compound 18
was then glycosylated with galactosyl thiocresol donor 19 to
give the R-(1f4)-linked trisaccharide 20 in 50% yield. After
deprotection, Gb3 with phosphonic acid tail (22) was obtained
in 54% yield.
With the trisaccharide acceptor 21 in hand, a programmable
[1 + 2 + 3] one-pot approach52 was used to construct the entire
Globo H hexasaccharide compound 25 in 54% yield. Depro-
tection of 25 began with removal of the trichloroethoxycarbonyl
group by using activated Zn particles with acetic acid, followed
by acetylation of the amine group with acetic anhydride and
(51) Chao, C.-S.; Li, C.-W.; Chen, M.-C.; Mong, K.-K. T. Chem.sEur. J.
2009, 15, 10972–10982.
(52) Huang, C.-Y.; Thayer, D. A.; Chang, A. Y.; Best, M. D.; Hoffman,
J.; Head, S.; Wong, C.-H. Proc. Natl. Acad. Sci. U.S.A. 2006, 103,
15–20.
9
J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010 13377

A R T I C L E S
Chang et al.
Scheme 3. One-Pot Synthesis of Globo H with Phosphonic Acid Tail, 26a
a Reagents and conditions: (i) NIS, TfOH, -50 °C. (ii) NIS, TfOH, -30 °C, 54% from 23. (iii) (1) Zn/AcOH, (2) Pyridine/Ac2O/Cat. DMAP, (3)
NaOMe/MeOH, (4) H2/Pd(OH)2.
pyridine. Zemple´n debenzoation and deprotection of the ben-
zylidene acetal and benzyl group were accomplished by using
Pd-black in 5% formic acid/methanol under 1 atm hydrogen,
to give the fully deprotected Globo H hexasaccharide with
phosphonic acid tail (26) (Scheme 3).
equivalent to the sensitivity of a traditional glass slide.53 The
dissociation constant of Mbr1 binding to Globo H was deter-
mined17 (Figure 7B,C), and found to be consistent with the
number generated from the NHS slides. These experiments
demonstrated that glycans with a phosphonic acid tail can be
easily prepared and covalently arrayed on ACG for use in
binding analysis. Now, we are synthesizing more glycans with
phosphonic acid tail to create a glycan array with more structure
diversities on ACG slide.
Gb3 22 and Globo H 26 were then spotted on the aluminum
oxide-coated glass slide for Mbr1 (a mouse anti-Globo H
monoclonal antibody) binding experiments. The printing pro-
cedure was the same as for mannose compound 11. Since the
properties of solvents will affect the spot size of the printed
oligosaccharides, we tested several solvents for 26. We began
with water only; however, the coupling process and the
dehydration reaction proceeded slowly, resulting in large printed
spots. Then, more viscous glycerol was tested. Although the
spot size was small enough to gain strong signal intensity, the
high boiling point of glycerol made the printed spot impossible
to dry. Ethylene glycol was finally chosen due to its high
viscosity, medium boiling point, and small size of printed spots.
As expected, the monoclonal antibody MBr1 can only recognize
the Globo H glycan but not Gb3 (Figure 7A). To demonstrate
the sensitivity of this new type of glycan array, various
concentrations of Globo H 26 from 250 µM (1.5 × 10-13 mol)
to 0.4 µM (2.4 × 10-16 mol) were spotted onto the aluminum
oxide-coated glass slide, and the same binding experiment with
MBr1 was carried out to study the sensitivity of this new type
of glycan array. The array could detect the presence of antibody
down to the attomole level (a and b of Figure S8 (SI)) that is
For comparison, the dissociation constant of Mbr1 binding
to Globo H noncovalently bound on the fluorinated ACG was
also tested (Figure S9 (SI)). Due to the highly hydrophobic
property of fluorinated ACG slide surface, the printing condi-
tions should be optimized to print diluted solutions with small
volumes. To reduce the hydrophobicity, ethylene glycol was
used to rinse the surface and then was removed by compressed
dry air. After that treatment, the solution of compound 7 was
adjusted to 5 µM in ethylene glycol and printed on the
fluorinated ACG slide surface. From the surface KD determi-
nation, we found that the dissociation constant calculated from
the noncovalent binding type of fluorinated ACG slide is larger
than that of the covalent binding type. Consistent with the study
(53) Liang, P.-H.; Wang, S.-K.; Wong, C.-H. J. Am. Chem. Soc. 2007,
129, 11177–11184.
9
13378 J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010

Glycan Array To Study Protein-Glycan Interactions
A R T I C L E S
Figure 7. Binding of monoclonal antibody MBr1 to Gb3 and Globo H on the aluminum oxide-coated glass slide. (A) Scanometric image and corresponding
analysis of glycan array after probing with monoclonal antibody MBr1, followed by incubation with a fluorophore-tagged mouse anti-IgM secondary antibody.
The slide contained Globo H and Gb3 printed at 100 µM concentration. (B) and (C) show the dissociation constants of Mbr1 and Globo H. ACG slides were
printed with Globo H of 16 concentrations ranging from 10-12 to 10-15 mol. The KD,surf values were calculated from the corresponding plot of intensity
versus the concentrations of Globo H and Mbr1.
of mannose binding to ConA on various surfaces,53,54 the
immersed into another pure 2-propanol solution for 15 min. The
solution was also sonicated to remove excess phosphonic acid on
the slide surface. The slide was dried by nitrogen under reduced
pressure. Upon completion of the reaction, the slide was washed
thoroughly with IPA and purge dried by nitrogen. Water contact
angle (g115°) measurement was quickly checked for the completion
of the slide fabrication.
covalent array on ACG provides a very suitable platform for
protein binding study.
Conclusion
In summary, we have developed novel platforms of covalent
and noncovalent glycan arrays on aluminum oxide-coated glass
slides. Combining with mass spectrometry, the noncovalent
glycan array can be used to characterize glycans directly without
fragmentation and also to identify and differentiate various types
of cellulases and their efficiency. The unique properties of
aluminum oxide-coated glass slides allow for the very conve-
nient and direct preparation of covalent glycan arrays via
phosphonic acid chemistry. Both ACG-based glycan arrays,
especially the covalent glycan array, have been shown to be
particularly useful for the study of protein-sugar interactions,
and the noncovalent glycan array has been shown to be useful
for the study of enzymatic reactions using mass spectrometry
in the absence of matrix. Synthesis of glycans with a polyfluoro
or phosphonic acid tail are ongoing in our laboratory (Table
S1 (SI)).
MS-TOF Analysis and Glycan Array Preparation of 4. MS-
TOF Analysis of the Polyfluorinated Mannose Derivative Ad-
sorbed on the Teflon-like ACG Slides. Compound 4 was dissolved
in a methanol/water (6/4) solvent mixture at approximately 10 mM,
1 mM, and 100 µM in series. The slides were spotted manually
with 1 µL of solution and also microarrayed with the BioDot
AD3200 instrument (Agilent Technology) by robotic pin (Array
It, SMP4) with a deposition of approximately 1.1 nL of the solution
per spot of the array. The slides were stored in 30% humidity
chamber overnight and then analyzed by mass spectrometry. The
blank and polyfluorinated mannose slides, including both silane-
based and phosphonic acid-based ACG slides, were analyzed with
Bruker Ultraflex MALDI-TOF mass spectrometer equipped with a
nitrogen pulsed laser (355 nm). An equal volume of BSA trypsin
digested (1 pmol/µL) solution was mixed homogeneously with DHB
(dihydroxybenzoic acid, 10 mg in 1:1 acetonitrile/water) solution
and was used as the standard for MS-TOF mass calibration. Each
data point was collected at the average of 500 shots of the laser
beam with laser fluence between 2 to 20%. Most of the experiments
were carried out under positive polarized electrical field.
Materials and Methods
Compounds 1-26. The detailed procedures for the synthesis
of compound 1-26 are shown in the Supporting Information.
Fabrication of the Phosphonic Acid-Based Teflon-like
ACG Slides. The aluminum-coated glass slide was washed by
acetone and water for three times and then dried by dry clean air.
The clean slide was then activated and cleaned by oxygen plasma
(Harrick plasma, PDC-32G) for 15 min. After activation, the slide
was immersed into 3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-heptadeca-
fluorodecylphosphonic acid (HDFDPA) 3 solution (1 M, 65%
2-propanol in H2O, pH ) 6.17) immediately. The solution was
gently sonicated (50 W) for 15 min.8 Following the ultrasonic
treatment, the slide was removed from the solution and then
Fluorescence-Tagged Con A/Mannose Binding of the Polyflu-
orinated Mannose Adsorbed on the Teflon-like ACG Slides. A 100-
µL portion of Alex 488-tagged concanavalin A (Con A) in
phosphate-BSA buffer (25 µg/mL, pH 6.5) were applied on the
ACG slide surface immobilized with mannose derivative. These
slides with Con A solution were incubated at room temperature
for approximately 2 h. After incubation, the slides were washed
three times each with 12 mL of phosphate-BSA buffer, PBST
buffer, and deionized water in Petri dishes with gentle swirling and
then were purge dried by nitrogen and analyzed by Array WoRx
(Applied Precision) in reflective mode using the fluorescence light
scanner at 530 nm.
(54) Jaipuri, F. A.; Collet, B. Y.; Pohl, N. L. Angew. Chem., Int. Ed. 2008,
47, 1707–1710.
9
J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010 13379

A R T I C L E S
Chang et al.
On-Chip Analysis by MALDI-TOF (Ultra-Flex II) (Reaction
in Eppendorf). Cellulase was prepared (5 U/mL in 25 mM pH 5.05
NaOAc buffer solution), and a substrate was also dissolved in
NaOAc buffer solution (25 mM, pH 5.05) to give 0.5 mM substrate
solution. A cellulase solution (100 µL) was added to 100 µL
substrate solution in Eppendorf to give a 2.5 U/mL cellulase solution
with 0.25 mM substrate. This solution was incubated at 37 °C for
18 h. 100 µL of this incubated solution were added onto the slide
loaded in FAST Frame. The slide was put in a drybox to remove
water, and then trace water was removed by high vacuum. Each
well of the FAST Frame multislide plate was rinsed by 100 µL of
water to dissolve and remove the buffer solution. After removal of
any residual water on the plate by high vacuum, the slide was
analyzed by MALDI-TOF Ultra-Flex II.
plate was sealed, and the whole assembly was incubated at 37 °C
for 18 h. The slide was put in a drybox to remove the water, and
then high vacuum was applied to remove trace water. Each well of
the FAST Frame multislide plate was rinsed by 100 µL of water to
dissolve and remove the buffer solution. After removal of any
residual water on the plate by high vacuum, the slide was analyzed
by MALDI-TOF Ultra-Flex II.
Acknowledgment. We thank Mr. Cheng-Lung Chen for his help
to grow the anodized ACG slide, Mr. Yung-Fung Lin for his help
with NMR analysis of synthetic compounds, and Mrs. Jennifer Chu
for critically reading the manuscript. We thank the Academia Sinica,
Taiwan, and National Science Council, Taiwan, for financial
support. (Grant NSC 97-2113-M-001-009-MY2 to C.-Y.W).
On-Chip Analysis by MALDI-TOF (Ultra-Flex II) (Reaction
on Chip Directly). A solution of cellulase was prepared (5 U/mL
in 25 mM pH 5.05 NaOAc buffer solution), and a substrate was
dissolved in NaOAc buffer solution (25 mM, pH 5.05) to give 0.5
mM substrate solution. A 50-µL portion of the cellulase solution
was added to 50 µL of the substrate solution onto the glass slide
which was loaded in FAST Frame multislide plate to have a 2.5
U/mL cellulase solution and 0.25 mM substrate. The well of the
Supporting Information Available: Complete reference 40,
Figures S1-S9, Table S1, detailed experimental procedures and
1
characterization, and H and 13C NMR spectra of compounds
1-26. This material is available free of charge via the Internet
at http://pubs.acs.org.
JA1046523
9
13380 J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010