Pattern-Based Detection of Different Proteins Using an Array of Fluorescent Protein Surface Receptors

Article abstract of DOI:10.1021/ja039562j

A small library of porphyrin derivatives whose fluorescence emission changes on binding to protein surfaces has been developed as a protein "fingerprinting" array. When incubated with different proteins, the array yields a characteristic response, specifi

Full text of DOI:10.1021/ja039562j

Published on Web 04/16/2004
Pattern-Based Detection of Different Proteins Using an Array of Fluorescent
Protein Surface Receptors
Laura Baldini, Andrew J. Wilson, Jason Hong, and Andrew D. Hamilton*
Department of Chemistry, Yale UniVersity, P.O. Box 208107, 225 Prospect Street,
New HaVen, Connecticut 06520-8107
Received November 12, 2003; E-mail: andrew.hamilton@yale.edu
Scheme 1. (Top) Mixed Condensation Synthesis of Porphyrin
Receptors; (Bottom) The Five Peptidic Components
There is currently a pressing need for simple, robust, and high
throughput strategies for the detection of specific proteins in
different environments.1 For example, large-scale proteomics
projects require the facile detection of signaling proteins as they
rise and fall as a result of cellular stimulus.2 Similarly, in medical
diagnostics, it is important to determine the presence or absence
of the characteristic protein signature of a disease. Finally, current
concerns about bioterrorism have raised an immediate need for field
detectors that can identify airborn or dissolved pathogens and their
toxic protein byproducts.3 Any protein detectors must be responsive
and/or applicable to a wide range of different targets. As a result,
there has been extensive development of multiple protein binding
molecules in an array format, with the expectation that different
proteins will interact with the array in distinctive ways. Effective
protein-detecting arrays have been based on monoclonal antibodies,4
antibody fragments,5 or nucleic acid aptamers.6 However, these
strategies suffer from chemical instability of the biopolymer
recognition molecule and the need to label the protein target for
detection. Kodadek has described two principal hurdles to the
development of protein-detecting arrays: (1) the need for large
numbers of stable, easily prepared ligands that each bind to a protein
target with high affinity and selectivity, and (2) the incorporation
of a means for detecting the ligand/protein interaction into the array.1
In the present paper, we report a simplified approach to this
problem based on the use of solution arrays of fluorescent protein
surface receptors. We have previously shown7 that a synthetic
tetraphenylporphyrin (TPP) derivative containing four negatively
charged carboxamide substituents can bind to the positively charged
surface of cytochrome c with low nanomolar affinity. The TPP unit
is well-suited for protein surface recognition due to its large
hydrophobic surface area (>300 Å2) that can be easily function-
alized on its periphery (m, p-phenyl or â-pyrrole positions) to match
a complementary hydrophobic and charged domain on a protein
surface. More importantly, TPP derivatives are highly fluorescent
and can show emission intensity changes on binding to a protein
target. We therefore reasoned that a large number of TPP derivatives
peripherally functionalized with substituents of different charges,
size, hydrophobicity, and symmetry might form an array of
receptors having different binding characteristics. None of the
receptors need be either specific or highly efficient for any single
protein analyte, but when arranged in an array format they should
respond quite distinctively to proteins with varying surface char-
acteristics, yielding a composite response unique for each analyte.8-10
Moreover, changes (or lack of them) in the fluorescence of the
TPP derivatives would provide information on the surface charac-
teristics of the protein analyte.
PyBOP and an excess of two different peptidic derivatives chosen
from a pool of five (Scheme 1, bottom) led to a mixture of six
different products. Four of them and the mixture of the two
geometrical isomers, X2Y2, were easily separated, due to their
different polarities, and subsequently deprotected using 2.5% water,
2.5% triethylsilane, 47.5% trifluoroacetic acid, and 47.5% dichlo-
romethane, a mixture which minimized cleavage of the methyl esters
and alkylation of the amines by free cations. Iteration of this
porphyrin functionalization procedure starting with every possible
combination of the two peptidic components followed by separation
and deprotection resulted in the isolation of 35 unique fluorophores
comprising every possible charge combination from +8 to -8 and
from 4 to 8 hydrophobic groups.
Eight members of the library were then selected to form a
preliminary group of receptors having different functionalities
(Figure 1, left). Next, 5 µM solutions of these porphyrins in 5 mM
phosphate buffer (pH 7.4) containing 0.05% of Tween 20 and 0.5%
of DMSO were arrayed in the first five columns of a 96 well quartz
plate, a different porphyrin in each row (A-H). When observed
under UV light (302 nm), each well shows a bright and intense
red fluorescence. To test the protein recognition ability of this
preliminary array, we selected four proteins having different surface
characteristics, ranging from the very acidic ferredoxin (pI 2.7511)
to the highly basic cytochrome c (pI 10.6), and incubated 3 equiv
of each protein in columns 2-5 (column 1 as blank). When
irradiated with UV light (302 nm), the plate displayed a pattern of
fluorescent and nonfluorescent wells (Figure 1, right).
We have previously shown7 that proteins containing a metal
center with unpaired electrons efficiently quench the porphyrin
fluorescence upon binding, due to the proximity of the metal to
the porphyrin ring in the complex. The wells that show quenching
would then correspond to a binding interaction between the synthetic
receptor and the protein, while the wells that remain fluorescent
should contain species that do not form complexes. In this way,
“naked-eye” detection would enable the rapid screening of the
collection of fluorophores for the identification of high affinity
A library of TPP derivatives functionalized with different amino
acids or amino acid derivatives could be rapidly synthesized using
a mixed condensation strategy (Scheme 1, top). The “one-pot”
reaction of 1 equiv of meso-tetracarboxyphenylporphine with
9
5656
J. AM. CHEM. SOC. 2004, 126, 5656-5657
10.1021/ja039562j CCC: $27.50 © 2004 American Chemical Society

C O M M U N I C A T I O N S
Figure 2. Fingerprints of cytochrome c (2), ferredoxin (3), cytochrome
c551 (4), and myoglobin (5) based on the eight porphyrin array. Each bar
quantifies the extent of color attenuation measured as the quantity (Gref
prot)/Gref, where Gref is the average gray value for the blank wells and
Gprot is the average gray value for the protein-containing wells.
-
G
Figure 1. (Left) Schematic structure of porphyrin receptors A-H. (Right)
Picture of part of the 96 well quartz plate irradiated with UV light (302
nm). Each well contains 200 µL of a 5 µM solution of porphyrins A-H (5
mM phosphate buffer, pH 7.4, containing 0.05% of Tween 20 and 0.5% of
DMSO); in the wells of columns 2-5 were added 3 equiv (as 500-600
µM stock solutions in 5 mM phosphate buffer, pH 7.4) of the following
proteins: 2, cytochrome c (pI ) 10.6, MW ≈ 12 500); 3, ferredoxin (pI )
2.75,11 MW ≈ 5500); 4, cytochrome c551 (pI ) 4.7,12 MW ≈ 9000); 5,
myoglobin (pI ) 6.8, MW ≈ 18 000).
a region which favorably interacts with the acidic porphyrins A,
B, C, and, to a lesser extent, F and G. Myoglobin, which has a
fairly neutral surface, has some affinity for most of the porphyrins
except F, which has the highest charge and the lowest hydrophobic-
ity.
Thus, each column of the plate corresponds to a unique
fingerprint, characteristic of a specific protein. This allows for the
unambiguous visual identification of each protein within the four
investigated (Figure 2).
Table 1. Dissociation Constantsa and Structural Properties of
Selected Synthetic Receptors and Proteins
aryl
porphyrin
charge
groups
protein
pI
K (µM)
d
In conclusion, we have developed an approach to the generation
of an array of porphyrins with different groups on their periphery.
When incubated with different proteins, the array responds with a
unique pattern of fluorescent/nonfluorescent spots which represents
a characteristic fingerprint for the protein.
A
B
E
F
-8
-4
+6
-8
+6
8
7
8
4
4
myoglobin
cytochrome c
myoglobin
cytochrome c
ferredoxin
6.8
10.6
6.8
10.6
2.75
1.76 ( 0.08
1.59 ( 0.07
3.7 ( 0.1
0.67 ( 0.02
0.37 ( 0.05
H
a Determined at 5 mM sodium phosphate, 0.05% Tween-20, pH 7.4, 298
Acknowledgment. We thank the NIH (GM 35208) for financial
K.
support of this work.
ligands for protein surfaces. This hypothesis was confirmed by
individual titration experiments between porphyrin-protein pairs
that occupy nonfluorescent wells (Table 1); titration of the protein
into the porphyrin solution resulted in quenching of the receptor
fluorescence due to complex formation. The dissociation constants
calculated upon fitting the data to 1:1 binding isotherms are in the
low and sub-micromolar range, confirming that appropriately
functionalized porphyrins provide receptors suitable to interact with
a range of protein surfaces. As expected, titration of ferredoxin
into a solution of A did not result in a significant fluorescence
decrease (Supporting Information). The quenching in the different
wells cannot be directly correlated to the relative binding affinities,
because the mechanism of quenching may vary according to the
location of the porphyrin on the protein surface.
The response of the array reflects the charge complementarity
between the porphyrin receptors and the target protein. In fact,
cytochrome c is bound by the negatively charged fluorophores A,
B, C, F, and G, and ferredoxin interacts with the positively charged
porphyrins D, E, and H. Cytochrome c551 has a prevalence of
acidic groups on its surface, but the hydrophobic area containing
the heme edge is surrounded by a few basic residues, thus forming
Supporting Information Available: Synthetic procedures and
analytical data, titration data (PDF). This material is available free of
charge via the Internet at http://pubs.acs.org.
References
(1) For an insightful discussion of this problem, see: Kodadek, T. Chem.
Biol. 2001, 8, 105-115.
(2) Zhu, H.; Snyder, M. Curr. Opin. Chem. Biol. 2001, 5, 40-45.
(3) Bronze, M. S.; Greenfield, R. A. Drug DiscoVery Today 2003, 8, 740-
745.
(4) De Wildt, R. M. T.; Mundy, C. R.; Gorick, B. D.; Tomlinson, I. M. Nat.
Biotechnol. 2000, 18, 989-994.
(5) Boder, E. T.; Midelfort, K. S.; Wittrup, K. D. Proc. Natl. Acad. Sci. U.S.A.
2000, 97, 10701-10705.
(6) Hesselberth, J.; Robertson, M. P.; Jhaveri, S.; Ellington, A. D. ReV. Mol.
Biotechnol. 2000, 74, 15-25.
(7) Jain, R. K.; Hamilton, A. D. Org. Lett. 2000, 2, 1721-1723.
(8) Albert, K. J.; Lewis, N. S.; Schauer, C. L.; Sotzing, G. A.; Stitzel, S. E.;
Vaid, T. P.; Walt, D. R. Chem. ReV. 2000, 100, 2595-2626.
(9) Lavigne, J. J.; Anslyn, E. V. Angew. Chem., Int. Ed. 2001, 40, 3118-
3130.
(10) Rakow, N. A.; Suslick, K. S. Nature 2000, 406, 710-713.
(11) Dutton, J. E.; Rogers, L. J. Biochim. Biophys. Acta 1978, 537, 501-506.
(12) Chao, Y.-Y. H.; Bersohn, R.; Aisen, P. Biochemistry 1979, 18, 774-779.
JA039562J
9
J. AM. CHEM. SOC. VOL. 126, NO. 18, 2004 5657