3D small-molecule microarrays

Article abstract of DOI:10.1039/b913665g

A PEG based 3D hydrogel slide was developed specifically for small-molecule microarraying purposes, which displayed improved loading capacity, signal sensitivity and spot morphology compared with a commercially available slide and comparative 2D slide. The Royal Society of Chemistry 2009.

Full text of DOI:10.1039/b913665g

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
3D small-molecule microarraysw
David M. Marsden,a Rebecca L. Nicholson,a Mark Ladlowb and David R. Spring*a
Received (in Cambridge, UK) 8th July 2009, Accepted 18th August 2009
First published as an Advance Article on the web 2nd September 2009
DOI: 10.1039/b913665g
A PEG based 3D hydrogel slide was developed specifically
for small-molecule microarraying purposes, which displayed
improved loading capacity, signal sensitivity and spot morphology
compared with a commercially available slide and comparative
2D slide.
purposes. Firstly PEG reduces the adsorption of proteins at
interfaces, which reduces non-specific interactions.8c Secondly
PEG is amphiphilic, and therefore is an ideal immobilisation
material for small-molecule microarray experiments, which
require immobilisation of small molecules using organic
solvents and protein incubation in aqueous buffer. 3D hydrogel
slides were made as described in Scheme 1. Ordinary glass
microscope slides were pre-functionalised with an acrylic ester
silane before treatment with the polymerisation mixture
containing the acrylamide PEG cross-linker 1,8a the hydroxy-
succinimide (NHS) active ester monomer unit 2,9 the
co-monomer 310 and the acetophenone UV radical initiator 4.
The polymerisation mixture was sandwiched with a hydro-
phobic cover slide, and polymerised upon irradiation with UV
light (l = 254 nm). The cover slide was removed to reveal an
apparent uniform gel layer, which was washed, dried and
analysed for slide thickness (Fig. 1) using an atomic force
microscope (AFM). The average gel thickness across two
different slides was calculated to be 470 nm.
Microarray technology has established itself as a versatile
tool for high throughput parallel investigation of molecular
interactions. This has included complementary oligonucleotide
hybridisations for gene expression profiling,1 protein–protein
interactions2 and small-molecule–protein interactions.3 The
latter are investigated using small-molecule microarrays, in
which small molecules, immobilised as spatially discrete spots
on a solid support, are probed with a fluorescently labelled
target protein. In this respect small-molecule microarrays have
potential as a high throughput screening tool for the discovery
of drug candidates or biological probes.3 The detection of
small-molecule–protein interactions is a significant challenge
using small-molecule microarrays, because the interactions are
often weak with a dissociation constant in the millimolar
range. As a result there is a requirement for microarray slides
with improved sensitivity for their detection.
For comparison purposes, 2D NHS ester slides were
made according to the literature procedure13 and commercial
Codelinkt slides were purchased. In order to compare loading
levels with the 3D hydrogel slide, the amine 5 was synthesised,
which contains an acid cleavable dimethoxytrityl group
(DMT).11 Upon immobilisation of 5, slides were treated with
perchloric acid to generate the DMT cation in solution
(Scheme 2), whose concentration was calculated using
UV-Vis spectroscopy, from which the average number of NHS
ester groups per cm2 was calculated for each slide. The average
number of NHS active ester groups on the 3D hydrogel slide
was 7.3 Â 1016 cmÀ2, which was over an order of magnitude
greater than the Codelinkt capacity (2.5 Â 1015 cmÀ2),
which in turn was more than twice the 2D slide capacity
(1.2 Â 1015 cmÀ2). We envisaged that the higher loading
capacity of the 3D hydrogel slides would have a significant
influence on the detected intensity level of fluorescence for a
given small-molecule–protein interaction.
Traditionally microarray slides are functionalised with a
relatively planar 2D surface for immobilisation purposes.
However 2D slides are often associated with low loading
capacities and weak signal detection.4 Slides that contain a
3D layer offer considerable advantages in terms of a higher
loading capacity. Polyacrylamide gel pads have been developed
for DNA and protein microarrays,5 and have been
commercialised by Perkin-Elmer (Hydrogelt slides).6 Slides
coated with a polymer containing active esters for the covalent
immobilisation of amines are also available commercially
(Codelinkt slides).7
We envisaged developing a 3D polyethylene glycol (PEG)
based microarray platform focused at the enhancement of
small-molecule–protein detection by increased small-molecule
loading capacity coupled with a reduction in non-specific
interactions. PEG is made up of repeating ethylene glycol
units (CH2CH2O)n and is an incredibly versatile material with
a plethora of chemical and biochemical uses. It has been used
to generate PEG based acrylamide resins for solid supported
chemical synthesis,8a and also to generate PEG based slides for
FRET based assays of peptides.8b PEG based surfaces offer
a number of advantages for small-molecule microarray
A standard small-molecule microarray experiment using
amino-modified biotin as the small molecule and Cy3-labelled
avidin as the protein was used to compare slides (Fig. 2A).
Biotin-amine 6 at 2.5, 1.25, 0.25 and 0.125 mM, in addition to
a biotin based negative control 7 at 5 mM, were printed onto
slides using a commercial microarrayer (Genetix QArray).
Printed slides were washed, incubated with Cy3 labelled avidin
and scanned (dry) for Cy3 fluorescence shown in Fig. 2B.
Slides were evaluated in terms of the following factors: how
strong the signal was, the signal-to-noise ratio (SNR) and the
spot morphology. The background corrected Cy3 intensity
was plotted against the concentration of biotin-amine 6 using
a log scale (Fig. 2C). All three slides displayed a linear
a Department of Chemistry, University of Cambridge, Lensfield Road,
Cambridge, UK CB2 1EW. E-mail: spring@cam.ac.uk
b Uniqsis Ltd, 29 Station Road, Shepreth UK SG8 6GB
w Electronic supplementary information (ESI) available: Full
experimental details. See DOI: 10.1039/b913665g
ꢀc
This journal is The Royal Society of Chemistry 2009
Chem. Commun., 2009, 7107–7109 | 7107

View Article Online
Scheme 1 Production of 3D hydrogel slides. 1 = bis-acrylamidoprop-1-yl-PEG1900, 2 = acryloyloxysuccinimide, 3 = acryloylmorpholine
and 4 = hydroxy-2-methylpropiophenone. Pre-functionalised glass slide generated by treatment of cleaned slide with bind-silane solution
(methylacryloxypropyltrimethoxysilane in acidic ethanol–water mixture). The pre-polymerisation mixture, which was optimised in terms of
loading capacity and microarray performance, consisted of 3% each of the cross-linker 1, monomer 2, co-monomer 3 and 1% initiator 4 in DMF.
The hydrophobic slide was used to cover the polymerisation mixture before irradiation with UV light.
fluorescent signal for a particular slide. The 3D hydrogel slides
produced the highest average SNR across the concentration
range of 6 printed, with a SNR of 24, compared to the
Codelinkt and 2D slides, which had values of 17 and 7,
respectively. Improvements in the SNR are likely in part due
to the presence of PEG in the hydrogel, which reduces the
non-specific adsorption of proteins.8c It should be noted that
at 2.5 mM 6, the 3D hydrogel slide generated particularly large
spots, which can be spaced further apart using a larger pitch
distance when printing. Alternatively smaller microarray pins
Fig. 1 AFM 3D surface image of a 3D hydrogel slide, where the gel
can be used to reduce the size of spots.
had been removed from part of slide by a scratch test. Average gel
The 3D hydrogel slide also produced a more consistent spot
morphology in terms of both spot size and shape compared
with the 2D and Codelinkt slides. In particular the 3D
hydrogel slide produced a lower variation in spot diameter
with a coefficient of variation (CV) range between 2 and 4%,
compared with 3–8% and 2–6% for the Codelinkt and 2D
slides, respectively.12 Regular spot morphology is particularly
important for accurate comparison of different spots within a
large-scale microarray screen.
thickness across two different slides = 470 nm, based on average of
three different cross-sectional measurements per slide (490 and
450 nm). See ESIw for further details.
relationship between fluorescence and concentration. The
3D hydrogel slide produced on average, a six-fold higher
Cy3 intensity over the Codelinkt slide, essentially because
of the 3D hydrogels superior loading capacity. The ability
to detect small-molecule–protein interactions at a higher
fluorescent signal is particularly relevant to their discovery
given that the interactions are often much weaker compared
with complementary oligonucleotide hybridisations and
protein–protein interactions. The requirement to scan the 2D
slide at a higher gain setting demonstrates its lower loading
capacity compared to the other two slides. The negative
control biotin 7 showed little sign of non-specific retention
across all three slides.
In summary, we have developed
a PEG based 3D
hydrogel microarray slide, which contains active esters for
the immobilisation of small-molecule amines. Loading studies
revealed our 3D hydrogel slide had over an order of magnitude
greater loading capacity compared to a commercial Codelinkt
slide and a comparative 2D slide. A standard small-molecule
microarray experiment also revealed that the 3D slides produced
a significantly higher fluorescent signal and improved spot
morphology across the concentration range of the small-
molecule biotin printed. We envisage that these key advantages
will enable us to use our 3D small-molecule microarray
The SNR is defined as the background corrected fluorescent
signal over the standard deviation of the background and
demonstrates how obtrusive the background noise is to the
Scheme 2 NHS ester slide loading capacity experiment using the dimethoxytrityl (DMT) linker 5.11 Following immobilisation of 5, slides were
treated with perchloric acid, which generatedÀa1deep orange colour corresponding to the cleaved DMT cation, whose concentration was calculated
using UV-Vis spectroscopy (e498 = 70 000 M cmÀ1). The average number of NHS ester groups for each type of slide was calculated across two
replicate slides: 3D hydrogel = 7.3 Â 1016 cmÀ2, Codelinkt = 2.5 Â 1015 cmÀ2 and 2D = 1.2 Â 1015 cmÀ2
.
ꢀc
This journal is The Royal Society of Chemistry 2009
7108 | Chem. Commun., 2009, 7107–7109

View Article Online
Fig. 2 (A) Small-molecule microarray experimental schematic in which biotin-amine 6 at 0.125, 0.25 and 1.25 mM and the biotin negative control
7 at 5 mM were both printed onto the three NHS ester slides prior to treatment with Cy3 labelled avidin. (B) Cy3 scanned images of the 3D
hydrogel (dry), Codelinkt and 2D slides. *2D slide did not display fluorescence at standard scanning settings so was rescanned at 40% PMT.12
(C) The background corrected Cy3 total intensity was plotted against the concentration of 6, using the log scale. Values were averaged over
25 spots and error bars correspond to the standard deviation.
focused platform to discover new small-molecule–protein
interactions, which would be beyond the detection capacity
of traditional microarray platforms.
6 J. C. Miller, H. Zhou, J. Kwekel, R. Cavallo, J. Burke,
E. B. Butler, B. S. Teh and B. B. Haab, Proteomics, 2003,
3, 56–63.
7 R. Ramakrishnan, D. Dorris, A. Lublinsky, A. Nguyen,
M. Domanus, A. Prokhorova, L. Gieser, E. Touma, R. Lockner,
M. Tata, X. Zhu, M. Patterson, R. Shippy, T. J. Sendera and
A. Mazumder, Nucleic Acids Res., 2002, 30, 30.
We gratefully acknowledge financial support from the
EPSRC, BBSRC, MRC, Newman Trust and GSK.
Notes and references
8 (a) M. Meldal, Tetrahedron Lett., 1992, 33, 3077–3080;
(b) M. Zourob, J. E. Gough and R. V. Ulijn, Adv. Mater., 2006,
1 M. Schena, D. Shalon, R. W. Davis and P. O. Brown, Science,
1995, 270, 467–470.
18, 655–659; (c) Y. S. Lin, V. Hlady and C. G. Golander,
Colloids Surf., B, 1994, 3, 49–62.
¨
2 G. MacBeath and S. L. Schreiber, Science, 2000, 289, 1760–1763.
3 (a) G. MacBeath, A. N. Koehler and S. L. Schreiber, J. Am. Chem.
Soc., 1999, 121, 7967–7968; (b) R. L. Nicholson, M. Ladlow and
D. R. Spring, Chem. Commun., 2007, 3906–3908. For recent reviews
see; (c) R. L. Nicholson, M. Welch, M. Ladlow and D. R. Spring,
ACS Chem. Biol., 2007, 2, 24–30; (d) A. J. Vegas, J. H. Fuller and
A. N. Koehler, Chem. Soc. Rev., 2008, 37, 1385–1394.
9 M. Mammen, G. Dahmann and G. M. Whitesides, J. Med. Chem.,
1995, 38, 4179–4190.
10 F. D’Agosto, M. T. Charreyre, F. Melis, B. Mandrand and
C. Pichot, J. Appl. Polym. Sci., 2003, 88, 1808–1816.
11 A. Avino, R. Guimil Garcia, F. Albericio, M. Mann, M. Wilm,
G. Neubauer and R. Eritja, Bioorg. Med. Chem., 1996, 4,
1649–1658.
4 D. W. Grainger, C. H. Greef, P. Gong and M. J. Lochhead,
Methods Mol. Biol. (Totowa, N. J.), 2007, 381, 37–57.
12 Fluorescent scanning was performed using a Genetix aQuire
scanner. 3D hydrogel and Codelinkt slides were scanned at 35%
PMT (photomultiplier tube setting), 2D at 40% PMT. The
coefficient of variation (CV) is the standard deviation expressed
as % of the mean.
5 (a) E. Timofeev, S. V. Kochetkova, A. D. Mirzabekov and
V. L. Florentiev, Nucleic Acids Res., 1996, 24, 3142–3148;
(b) J. Zlatanova and A. Mirzabekov, Methods Mol. Biol.
(Totowa, N. J.), 2001, 170, 17–38; (c) P. Arenkov, A. Kukhtin,
A. Gemmell, S. Voloshchuk, V. Chupeeva and A. Mirzabekov,
Anal. Biochem., 2000, 278, 123–131.
13 R. Benters, C. M. Niemeyer and D. Wohrle, ChemBioChem, 2001,
2, 686–694.
ꢀc
This journal is The Royal Society of Chemistry 2009
Chem. Commun., 2009, 7107–7109 | 7109