Microparticle matrix encoding of beads

Article abstract of DOI:10.1002/anie.200906563

Reloaded matrix: Facile optical encoding of polyethylene glycol (PEC) based resins provides a direct identification of compounds in combinatorial libraries, and the structures may be correlated with bioactivity. The new technique, microparticle matrix (MPM) encoding (see picture), circumvents some problems often encountered in solid-phase combinatorial chemistry and is extremely simple to implement. (Figure Presented).

Full text of DOI:10.1002/anie.200906563

Angewandte
Chemie
DOI: 10.1002/anie.200906563
Combinatorial Chemistry
Microparticle Matrix Encoding of Beads**
Morten Meldal* and Søren Flygering Christensen
The combinatorial approach remains one of the most power-
ful avenues for the discovery of new active compounds.[1]
Solid-phase screening of combinatorial libraries is associated
with: 1) millimolar rather than nanomolar concentrations,
b) nontrivial structural analysis, and c) a non-biocompatible
resin environment. In addition to chemical[2–4] and radio-
frequency[5] encoding, split/mix libraries[6,7] have been en-
coded by using several optical methods, which include
spherical encoding by dye diffusion,[8] laser-etched barco-
des,[9,10] in situ composite particle labeling,[11a] and infrared
tags in resins[12] (Figure 1). In another version of microparticle
composite labeling, Battersby et al. performed pre-synthesis-
coating of beads with microparticles for fluorescence and
light-scattering-based decoding.[11b] Despite these advances,
the number of codes used in libraries has remained quite low.
Efficient encoding requires simultaneous reliable produc-
tion of beads with a random code distribution.[13] In principle,
the method allows encoding of an infinite number of beads,[14]
although in practice the encoding is limited to the number of
beads(e.g., 30000), which have been handled in one experi-
ment. An encoding process that is independent of the
chemistry employed and allows a variety of assays is therefore
advantageous. We present herein a versatile, simple encoding
principle that may solve many problems in solid-phase
screening and decoding: optical microparticle matrix encod-
ing (MPM encoding) with fluorescent microparticles. The
utility of this technique has been demonstrated in the
identification of avidin ligands from a focused library, in
which compounds could not be distinguished with mass
spectrometric (MS) methods. The library design was based
upon previously reported l and d amino acid libraries for
avidin and streptavidin, thus indicating selectivity (R0) for
aromatic residues.[15–18] It is not clear whether histidine and
proline (in the histidine–proline–-glutamine (HPQ) sequence
binding to streptavidin) are required for avidin recognition.
MPM encoding (Figure 2) is efficient and satisfies the
prerequisites for functional encoding (see theoretical deter-
mination of encoding potential in the Supporting Informa-
tion). Uniform 10 mm Tentagel microparticles labeled with a
chemically stable fluorophore ATOTA (tris(dialkylamino)-
trioxatriangulenium ion)[19] were randomly distributed in a
bis(acrylamido) polyethylene glycol (PEG) macromono-
mer[20] (64000 cmÀ3) by ultrasound and homogenization.
Inverse suspension polymerization of the microparticle-con-
taining macromonomer in the presence of bis(diphenylme-
thylsilyl) PEG1500 stabilizer provided 500000 uniform,
Figure 1. Examples of optical encoding comprise a) diffusion of reac-
tive fluorophores,[8] b) laser bleaching,[9] and c) in situ encoding with
particles.[11a]
[*] Prof. Dr. M. Meldal
SPOCC Centre, Carlsberg Laboratory
Gamle Carlsberg Vej 10, 2500 Valby (Denmark)
Fax: (+45)3327-5301
E-mail: mpm@crc.dk
Homepage: http://www.crc.dk/SPOCC
Dr. S. F. Christensen
FeF Chemicals A/S
Københavnsvej 216, 4600 Køge (Denmark)
[**] This work was supported by the Danish National Research
Foundation. We are grateful to D. C. Tvermoes, P. Breddam, P. M.
St. Hilaire, R. Michael, and T. Reenberg for support and discussions.
T. Schulte-Herbrꢀggen assisted with the theoretical determination
of the encoding potential.
Figure 2. Images of one encoded bead, recorded using telecentric
optics with the instrument shown in Figure 4. The image coordinates
are converted into 3D-interparticle vector lengths (d) and angles (a)
between all vector pairs.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200906563.
Angew. Chem. Int. Ed. 2010, 49, 3473 –3476
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3473

Communications
amino-functionalized, uniquely encoded beads (72 mL;
ca 6.5 g dry weight), which contained an average of 8 micro-
particles per bead, with a bead size of (550 Æ 100) mm.
The 3D matrix presented by the random positions of
microparticles can be viewed as a set of unique vectors that
connect each microparticle with all other microparticles in the
bead. The vector lengths and angles between pairs of vectors
represent a large rotation- and translation-independent
parameter set. This dataset constitutes a unique fingerprint
that is easily compared by computational methods.
In order to record the codes, beads were gently sucked
onto 150 mm capillaries in a carousel controlled by a stepper
motor, and transported to three orthogonally positioned
cameras. The beads were illuminated using a laser and the
images recorded. Image analysis accurately provided the 2D
coordinates of centers of the microparticles in each image and
3D coordinates were determined by correlation as outlined in
Figure 3. A second station on the carousel was dedicated to
Figure 4. The core of the decoder consists of a rotating Kalrez disk
with pressure-drop capillaries. Output from bioassays may be read in-
line and directly correlated with the structure of compound.
Figure 3. Fitting procedures that compare axis projections from all
three images and optimize the indicated connectivity are used to
reliably transform the three sets of 2D coordinates into one set of 3D
coordinates.
and Table 1 at a loading of approximately 1 mmol ligand per
milliliter of resin. The split/mix synthesis was performed on
3000 encoded beads by splitting the resin with image record-
ing prior to each coupling. The progress of the peptide
couplings was monitored using the yellow coloration the
Dhbt-OH coupling catalyst developed with residual free
amino groups.
The library was incubated in bovine serum albumen
(BSA) buffer with ROX-labeled avidin, and 30 beads with
high fluorescence intensity were isolated and decoded. Of
these beads, 20 structures were determined and 9 were
recording the intensity of a biochemically derived fluorescent
signal, which is different from that of the microparticles.[21]
The carousel (Figure 4) was further connected to a gear pump
that provided a controlled pressure drop over the capillaries.
The equipment rapidly provided high-quality images and
bioassay readout in a short period of time (0.3 s per bead).
The short exposure eliminated the photobleaching often
observed under a fluorescence microscope.
Table 1: Amino acids used for synthesis of the library in Scheme 1.[a]
The images were analyzed with two computer pro-
grams:[22] ImToCoord provided 3D coordinates and H2S-
Compare transformed the coordinates and compared splits
with hits. Microparticle coordinates were converted into
interparticle distances and angles between particle-to-particle
vectors, and datasets were matched and scored. Selected hit/
split pairs were visually compared by simulated annealing
alignment of the microparticles.
R2
R1
R0
val
His
his
Tha
Ala(3-Pyr)
Ala(4-Pyr)
Gly(4-Pip)
Aze
Pro
pro
N-Me-Ala
Nca
Pca
Gln
gln
Phe(4-NO2)
phe(4-NO2)
Cit
Phe(3-CN)
Phe(4-CN)
Oic
For identification of avidin ligands, a focused library was
screened by using MPM encoding. Ligand concentration on
the solid support was reduced 10-fold to differentiate tight
from low-affinity binders.[23] A library of 343 compounds,
R2R1R0-G-PEGA (PEGA = polyethylene glycol-polyacryl-
amide copolymer), was assembled according to Scheme 1
[a] Ala=alanine, Aze=azetidine-2-carboxylic acid, Cit=citrulline, Gln=
glutamine, His=histidine, Nca=nipecotic acid, Pca=pipecolic acid,
Phe=phenylalanine, Pro=proline, Pyr=pyridyl, Oic=(2S,3aS,7aS)-
octahydroindole-2-carboxylic acid, Val=valine. Codes with an upper-
case letter represent the l-amino acid, while codes in lower case
represent the d-amino acid.
3474
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 3473 –3476

Angewandte
Chemie
Scheme 1. MPM-encoded synthesis of a split/mix library containing
several isobaric pairs of building blocks. The initial loading was
reduced by acylation with a mixture of Fmoc-Gly-OH and Boc-Gly-OH.
Boc=tert-butoxycarbonyl, Dhbt-OH=3-hydroxy-4-oxo-3,4-dihydro-
1,2,3-benzotriazine, Fmoc=9-fluorenylmethyloxycarbonyl, MSNT=1-
(2-mesitylenesulfonyl)-3-nitro-1H-1,2,4-triazole, NEM=N-ethyl mor-
pholine; Pip=piperidine, TBTU=O-(benzotriazol-1-yl)-N,N,N’,N’-tetra-
methyluronium tetrafluoroborate, TFA=trifluoroacetic acid.
Figure 5. Binding of ROX-avidin (ROX=6-carboxy-X-rhodamine) to
resynthesized hits. Compound 41 is HPQ and compounds 4, 19, and
22 are inactive controls. The numbers 1–41 represent compounds
tabulated in the Supporting Information.
Table 2: Data for the binding of peptides to immobilized avidin as
determined by SPR.
Compound
m/z (calcd)
Kd [m]
His-Pro-Gln-Gly-OH (41)
480.1641 (480.2206) 1.21ꢁ10À3
510.2091 (510.2101) 2.33ꢁ10À6
598.2611 (598.2625) 1.42ꢁ10À5
574.1982 (574.1971) 5.00ꢁ10À5
544.2144 (544.2156) 4.60ꢁ10À5
512.2263 (512.2258) 6.99ꢁ10À6
partially determined (2–3 possible structures; see Table 1 in
the Supporting Information).
his-Aze-Phe(4-CN)-Gly-OH (39)
his-Oic-phe(4-NO2)-Gly-OH (23)
Tha-Nca-phe(4-NO2)-Gly-OH (7)
his-Pro-phe(4-NO2)-Gly (5)
Tha-N-MeAla-Phe(4-CN)-Gly-OH
(34)
The identified hits were synthesized for solid-phase
binding assays and the binding affinity was investigated by
using surface plasmon resonance (SPR). The binding assay is
presented in Figure 5 and correlates well with the SPR results
(Table 2). Immobilization of avidin and binding of twofold
dilutions of ligands generally provided good binding curves
that show some nonstoichiometric binding. Accordingly,
fitting was performed on the rise of the association curves
and the latter section (15–300 s) of the dissociation curves.
The KD values ranged from 10À2 to 10À7 m (Table 2). Incorpo-
ration into cyclic peptides[24] (Figure 6) increased the affinity
by approximately one order of magnitude (Table 2).
The ligand preference was surprisingly variable for R1 and
R2, whereas a 4-substituted phenyl ring was clearly preferred
for R0. The relative configuration of the three amino acids
varied, with a d amino acid preferred for R0 and an l amino
acid for R1. Similarly, the structural R1 residue did not have a
preference for one particular ring size although Pro, Oic, and
Aze were predominant.
his-Aze-phe(4-NO2)-Gly-OH (13)
Tha-Aze-Gln-Gly-OH (25)
530.2002 (530.2000) 3.72ꢁ10À6
482.1696 (482.1710) 2.13ꢁ10À5
541.2048 (541.2047) 7.05ꢁ10À5
524.2251 (524.2258) 2.08ꢁ10À5
578.2719 (578.2727) 6.60ꢁ10À6
4-PyA-Aze-phe(4-NO2)-Gly-OH (37)
His-pro-Phe(4-CN)-Gly-OH (31)
his-Oic-Phe(4-CN)-Gly-OH (17)
4PipG-Oic-phe(4-NO2)-Gly-OH (38) 601.2972 (601.2986) 8.90ꢁ10À6
His-N-MeAla-gln-Gly-OH (3)
Tha-Pro-phe(4-NO2)-Gly-OH
Val-Aze-phe(4-NO2)-Gly-OH
468.2199 (468.2207) 2.70ꢁ10À5
560.1815 (560.1815) 8.73ꢁ10À5
492.2084 (492.2094) 7.80ꢁ10À5
Cyclic peptide ligands[a]
c-A-His-Pro-Gln-FPAEK-OH
c-A-his-Oic-Phe(NO2)-FPAEK-OH
c-A-his-Oic-phe(NO2)-FPAEK-OH
c-A-His-pro-Phe(4-CN)-FPAEK-OH
c-A-his-Oic-Phe(4-CN)-FPAEK-OH
503.7885 (503.7591) 6.70ꢁ10À2
562.7830 (562.7803) 2.44ꢁ10À6
562.7805 (562.7803) 1.78ꢁ10À6
525.7618 (525.7617) 5.34ꢁ10À6
562.7830 (562.7803) 5.18ꢁ10À7
c-A-His-Pca-Phe(4-NO2)-FPAEK-OH 542.79 (542.7644)
2.67ꢁ10À7
In conclusion, a concept and a practical method of
encoding beads with a matrix of fluorescent microparticles
have been presented. The encoding potential is large and
[a] c- represents cyclization between Glu side chain and N-terminal Ala.
Cyclopeptides were detected as m/2z. FPEAK=Phe-Pro-Glu-Ala-Lys.
These cyclopeptides are reported to significantly increase the binding of
their tripeptide inserts.
Angew. Chem. Int. Ed. 2010, 49, 3473 –3476
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3475

Communications
Keywords: automated screening · combinatorial chemistry ·
.
encoding · fluorescent tags · peptides
[1] J. P. Kennedy, L. Williams, T. M. Bridges, R. N. Daniels, D.
Weaver, C. W. Lindsley, J. Comb. Chem. 2008, 10, 345 – 354.
[2] Z.-J. Ni, D. Maclean, C. P. Holmes, M. M. Murphy, B. Ruhland,
J. W. Jacobs, E. M. Gordon, M. A. Gallop, J. Med. Chem. 1996,
39, 1601 – 1608.
[3] M. H. J. Ohlmeyer, R. N. Swanson, L. W. Dillard, J. C. Reader,
G. Asouline, R. Kobayashi, M. Wigler, W. C. Still, Proc. Natl.
Acad. Sci. USA 1993, 90, 10922 – 10926.
[4] S. Brenner, R. A. Lerner, Proc. Natl. Acad. Sci. USA 1992, 89,
5381 – 5383.
[5] K. C. Nicolaou, X.-Y. Xiao, Z. Parandoosh, A. Senyei, M. P.
Nova, Angew. Chem. 1995, 107, 2476 – 2479; Angew. Chem. Int.
Ed. Engl. 1995, 34, 2289 – 2291.
[6] K. S. Lam, S. E. Salmon, E. M. Hersh, V. J. Hruby, W. M.
Kazmierski, R. J. Knapp, Nature 1991, 354, 82 – 84.
[7] A. Furka, F. Sebestyen, M. Asgedom, G. Dibo, Int. J. Peptide
Protein Res. 1991, 37, 487 – 493.
[8] R. A. Farrer, G. T. Copeland, M. J. R. Previte, M. M. Okamoto,
S. J. Miller, J. T. Fourkas, J. Am. Chem. Soc. 2002, 124, 1994 –
2003.
[9] K. Braeckmans, S. C. d. Smedt, C. Roelant, M. Leblans, R.
Pauwels, J. Demeester, Nat. Mater. 2003, 2, 169 – 173.
[10] D. C. Pregibon, M. Toner, P. S. Doyle, Science 2007, 315, 1393 –
1396.
[11] a) B. J. Battersby, D. Bryant, W. Meutermans, D. Matthews,
M. L. Smythe, M. Trau, J. Am. Chem. Soc. 2000, 122, 2138 – 2139;
b) B. J. Battersby, G. A. Lawrie, A. P. R. Johnston, M. Trau,
Chem.Commun. 2002, 1435 – 1441.
[12] H. Fenniri, L. Ding, A. E. Ribbe, Y. Zyrianov, J. Am. Chem. Soc.
2001, 123, 8151 – 8152.
Figure 6. Examples of sensograms showing binding of a linear and a
cyclic ligand to avidin immobilized on a CM5 chip using a Biacore
3000 instrument.
[13] S. F. Christensen, J. H. Truelsen, M. Meldal, R. Michael, I.
Johannsen, US 127369, 2006.
[14] S. F. Christensen, I. Johannsen, J. M. Carstensen, L. Kuhlmann,
M. Meldal, WO 061094, 2005.
depends on the ability to record and analyze high-quality
images of the beads as well as the number of beads that can be
handled in practice, which is currently 30000 beads (ca. 4 mL
of resin with 550 mm beads). The recording method is rapid
and has been automated; fast algorithms that permit real-time
conversion of images to a bead signature have been devel-
oped. The structure identification can be combined with
direct in-line screening in the instrument; correlation of
structure with screening results is described in a separate
communication.[21] The method was validated by screening a
highly focused library of avidin ligands, and ligand structures,
which can not be by distinguished by using conventional MS-
MS techniques, were determined. The technology is promis-
ing for the screening of libraries of complex molecules and of
solid-phase-bound ligands at biologically relevant concentra-
tions. Practical microparticle detection and bead handling
limits the current scaling of the system, but the synthesis of
150 mm beads with 4 mm microparticles is feasible.
[15] K. S. Lam, M. Lebl, ImmunoMethods 1992, 1, 11 – 15.
[16] B. Gissel, M. Roland Jensen, K. Gregorius, H. Elsner, I.
Svendsen, S. Mouritsen, Pept. Sci. 1995, 1, 217 – 226.
[17] S. Østergaard, P. H. Hansen, M. Olsen, A. Holm, FEBS Lett.
1995, 362, 306 – 308.
[18] T. Gaja, S. C. Meyera, I. Ghosh, Protein Expression Purif. 2007,
56, 54 – 61.
[19] B. W. Laursen, F. C. Krebs, Angew. Chem. 2000, 112, 3574 – 3576;
Angew. Chem. Int. Ed. 2000, 39, 3432 – 3434.
[20] F.-I. Auzanneau, M. Meldal, K. Bock, J. Pept. Sci. J. Peptide Sci.
1995, 1, 31 – 44.
[21] J. E. Rasmussen, C. B. Schiødt, S. F. Christensen, L. Nørskov-
Lauritsen, M. Meldal, P. M. St. Hilaire, K. J. Jensen, Angew.
Chem. 2010, DOI: 10.1002/ange.200906602; Angew. Chem. Int.
Ed. 2010, DOI: 10.1002/anie.200906602.
[22] Available from the author upon request.
[23] M. Conza, H. Wennemers, Chem. Commun. 2003, 866 – 867.
[24] Y.-P. Chang, Y.-H. Chu, Anal. Biochem. 2005, 340 74 – 79.
Received: November 20, 2009
Revised: March 4, 2010
Published online: April 8, 2010
3476
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 3473 –3476