Enzyme-artificial enzyme interactions as a means for discriminating among structurally similar isozymes

Article abstract of DOI:10.1021/jacs.5b02496

We describe the design and function of an artificial enzyme-linked receptor (ELR) that can bind different members of the glutathione-S-transferase (GST) enzyme family. The artificial enzyme-enzyme interactions distinctly affect the catalytic activity of the natural enzymes, the biomimetic, or both, enabling the system to discriminate among structurally similar GST isozymes.

Full text of DOI:10.1021/jacs.5b02496

Communication
pubs.acs.org/JACS
Enzyme−Artificial Enzyme Interactions as a Means for Discriminating
among Structurally Similar Isozymes
Karuthapandi Selvakumar, Leila Motiei, and David Margulies*
Department of Organic Chemistry, The Weizmann Institute of Science, Rehovot, Israel 76100
S
* Supporting Information
ABSTRACT: We describe the design and function of an
artificial enzyme-linked receptor (ELR) that can bind
different members of the glutathione-S-transferase (GST)
enzyme family. The artificial enzyme−enzyme interactions
distinctly affect the catalytic activity of the natural
enzymes, the biomimetic, or both, enabling the system
to discriminate among structurally similar GST isozymes.
he remarkable efficiency by which enzymes catalyze
Tchemical reactions has stimulated chemists to develop
myriad tools to mimic their function.1 These advancements have
led to the creation of biomimetics with catalytic turnovers, and
they also inspired the development of biosensors2 and stimuli-
responsive catalysts that utilize principles of cooperativity and
allostery,3 akin to natural enzymes. Although, to date, molecular-
scale artificial enzymes cannot compete with the catalytic
efficiency of natural proteins, these model systems significantly
Figure 1. (a) Operating principles of an artificial ELR integrating a
contribute to the realization of “supramolecular chemistry in
catalytic site (i.e., enzyme mimic) with a specific protein binder (i.e.,
water” 4 by advancing our understanding of the parameters
receptor). Binding of the biomimetic to isozymes with different surface
required to obtain synthetic receptors that can catalyze reactions
in aqueous medium at neutral pH and at ambient temperature.5
Beyond their ability to accelerate important biochemical
reactions, several groups of enzymes called enzyme-linked
receptors (ELRs)6 are also involved in signal transduction
processes in which binding of an enzyme to a protein partner
affects its catalytic activity and the consequent downstream signal
cascade. In addition to having a catalytic site, ELRs possess a
protein recognition domain7 that enables them to interact with
several proteins and regulate the transfer of chemically encoded
information across cells. Here we show that it is also possible to
endow artificial enzymes with the ability to interact with multiple
enzyme partners and that these interactions can distinctly affect
the catalytic activity of the natural enzymes, the biomimetic, or
both. We also show that information gained from the artificial
ELR-enzyme interactions enables the system to discriminate
among structurally similar isozymes.
Recently, we demonstrated that attaching a highly specific
protein receptor to a nonspecific protein surface binder results in
binding cooperativity, giving the latter enhanced affinity toward
the surface of the target protein.8 We also showed that, when
combined in arrays (the so-called “chemical noses/
tongues” 9,10), targeted protein surface sensors of this class can
generate unique optical “fingerprints” that enable them to
discriminate among structurally similar glutathione-S-transferase
(GST) isozymes.8 Activity-based protein profiling and related
methods, in which enzyme groups are discriminated by their
characteristics (isozymes i−iv) inhibits their activity and, at the same
time, distinctly affects the association of the synthetic catalyst with their
surfaces and, consequently, its catalytic efficiency. (b) Unique ID for
each isozyme can be obtained by following the formation of the products
(P,P′) from their colorimetric substrates (S,S′).
activities and specificities toward synthetic inhibitors and/or
substrates, have also been used to identify isozymes.11 GSTs, in
particular, have been effectively differentiated using arrays of
multiple inhibitors and substrates.12 It occurred to us that by
using an enzyme mimic as a nonspecific protein surface binder
(Figure 1a), one should be able to obtain an artificial ELR that,
similar to natural ELRs, can change its catalytic activity upon
binding to a small set of protein partners (Figure 1a, isozymes i
and iii). In addition, we expected that, by endowing the ELR
mimic with the ability to target an enzyme family (isozymes i−
iv), this monomolecular system could be used to differentiate
among multiple isozymes by measuring the changes in the
catalytic activity of both the synthetic and the natural enzyme
(Figure 1b). Four representative complexes, generated by
binding the ELR mimic to different isozymes, are shown in
Figure 1a. Differences in the reaction rates for each catalyst can
be followed optically by observing the formation of two different
products (P,P′) from their colorimetric substrates (S,S′).
Received: September 4, 2014
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.5b02496
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
Figure 2. Chemical structure of an artificial ELR consisting of an
esterase mimic (A), a broad-spectrum GST binder (B), and a
hydrophobic recognition element (C).
Figure 3. Colorimetric assays used to measure the activity of the natural
GST enzymes (channel 1) and artificial ELR (channel 2). Upon
formation of the 1-GST complex, the signal in each channel is distinctly
affected, enabling the system to generate a unique ID for each isozyme.
Binding of the ELR mimic to the active site of isozyme i, for
example, inhibits the catalytic activity of the enzyme and brings
the synthetic catalyst in the vicinity of the protein’s surface. As a
result of the “multivalency effect”,8 electrostatic and hydrophobic
interactions and hydrogen bonding between the artificial enzyme
and the protein’s surface are enhanced, blocking the biomimetic’s
catalytic site and inhibiting its activity. In contrast, isoform ii, with
distinct surface characteristics, does not interact with the catalytic
unit of the artificial ELR; thus, only the catalytic activity of the
natural enzyme is affected. Distinct catalytic responses are
obtained upon binding the ELR mimic to isozymes iii and iv.
Binding to isozyme iii inhibits the enzyme’s activity and enhances
the activity of the biomimetic due to allosteric effects on the
protein’s surface. The activity of isozyme iv, on the other hand,
which binds the biomimetic with the lowest affinity, is only
partially inhibited upon forming the biomimetic-enzyme
complex, whereas the activity of the synthetic catalyst remains
unchanged.
Based on these principles, we generated an artificial ELR
(Figure 2, 1) that can interact with different members of the GST
enzyme family. GSTs are detoxifying enzymes that assist in
removing a wide range of xenobiotics (e.g., 1-chloro-2,4-
dinitrobenzene, CDNB) from the body by catalyzing their
conjugation to glutathione (GSH) (Figure 3, channel 1). The
structure of 1 consists of a cis-amino proline scaffold appended
with an artificial esterase (A), a broad-spectrum GST inhibitor
(B), and a hydrophobic aromatic group (C). A biscyclen Zn(II)
complex was selected as the catalytic unit (A) for this biomimetic
because complexes of this type can generally act as esterase
mimics13,14 that accelerate the hydrolysis of fluorogenic
substrates, e.g., cleavage of fluorescent 1-hydroxypyrene-3,6,8-
trisulfonic acid (HPTS) from its ester precursor (Ac-HPTS)
(Figure 3, channel 2). Biscyclen metal catalysts have been shown
to be more efficient catalysts than the monometallic compounds,
due to the synergy between the two metal centers.14 For a specific
protein binder (B), we used a bisethacrynic amide (bis-EA)
inhibitor, which has been shown8,15 to simultaneously bind the
two active sites of these enzymes with nanomolar affinities. The
role of the acridine group (C) is to promote hydrophobic and π-
interactions, which will, together with the positively charged
Zn(II) complexes, facilitate the recognition of the aromatic and
negatively charged substrate (Ac-HPTS), as well as negatively
charged and hydrophobic patches on the protein’s surface.
Following its synthesis, apo-1 was incubated with Zn(NO3)2·
6H2O to afford the Zn(II) complex catalyst 1 (Supporting
Information (SI)). The catalytic activity of 1 was followed by
monitoring the hydrolysis of Ac-HPTS into the fluorescent
HPTS (Figure 4b), and the Michaelis−Menten model (Figure
Figure 4. (a) Enzymatic activity of a representative isozyme (GST M1).
(b) Catalytic activity of 1. Uncatalyzed background reactions are
denoted by a dashed line.
S1) was applied to determine the catalytic parameters (Table
S2), e.g., substrate binding constant KM = 18.8 1.4 μM, pseudo-
first-order rate constant kcat = 0.90 h−1, and rate enhancement
kcat/kuncat = 300. As expected from the cooperativity between the
metal centers, the apparent second-order rate constant, or the
specificity constant, k2 (= kcat/KM) for 1 (0.0479 μM−1 h−1) was
25 times higher than that of the monocyclen Zn(II) complex
(0.0019 μM−1 h−1). In parallel, we tested the conditions for
following the enzymatic activity of different GST isoforms
(Table S1) by following the formation of the GSH-CDNB
conjugate (Figure 4a). Eleven GST isoforms (GST A1, GST A2,
GST A3, GST M1, GST M2, GST P1, GST K1, GST S1, GST
T1, GST Z1, and GST O1) were screened for enzymatic activity.
As shown in Table S1, GST A1, GST A2, GST M1, GST M2, and
GST P1 exhibited good enzymatic activity at 10 nM
concentration, whereas GST A3, GST S1, and GST K1 required
a higher enzyme concentration (50 nM). No activity was
detected for GST Z1, GST T1, and GST O1 under these
concentrations, in agreement with previous reports.16 Although
these findings show that the CDNB-GSH assay can be used to
discriminate among isozyme groups (Table S1), i.e., highly active
(I), moderately active (II), and inactive (III), they also show that
enzymatic activity alone is not a sufficient parameter for multi-
isozyme differentiation.
After optimizing the colorimetric assays for measuring the
activity of 1 and GSTs, we used these assays to test the
interaction between them. By measuring the enzymatic activity of
the different isozymes in the presence and absence of a large
excess (1 μM) of the ELR mimic, we confirmed that 1 can bind to
different GSTs from groups I and II (Figure 5, dashed line). The
activity of most GSTs was strongly inhibited, indicating that 1
binds the active site of these enzymes, as expected from our
B
DOI: 10.1021/jacs.5b02496
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
Figure 5. Inhibition of GST isozymes by artificial ELR 1. Catalytic
activities of 20 nM (a) GST A1, (b) GST M1, (c) GST A2, and (d) GST
P1 were monitored at 340 nm in the absence (−) and presence of 100
nM (···) or 1 μM (---) 1.
Figure 7. (a) LDA mapping of catalytic activity in channels 1 and 2 for
different GSTs and their combinations (Table S4). (b) Analysis of urine
samples containing GST A1, GST P1, or both.
Figure 6. (a) Activity plot summarizing changes in the catalytic activity
of various GSTs by 1 (purple) and of 1 by GSTs (green). (b) Changes in
the catalytic activity of the artificial ELR 1 by various GST isozymes; λex
= 450 nm, λem = 510 nm.
the specific binder to the observed catalytic responses, we
performed two additional experiments. In the first (Figure S4),
we measured the effect of other proteins on the catalytic activity
of 1, and in the second, we tested how GSTs affect the activity of
a control compound lacking the bis-EA unit (Figure S5). The
facts that five different proteins did not significantly change the
activity of 1 and that the activity of the control compound was
only weakly affected by the presence of GST M2, S1, and K1
indicate the role of the bis-EA unit in inducing the observed
effects. GST P1 did inhibit the activity of the control catalyst;
however, this compound was also found to be an inhibitor of
GST P1 (Figure S6), most likely due to the high affinity of this
isozyme to metal complexes.17
design. Next, we followed the enzymatic activities of the different
GSTs with decreasing concentrations of 1 to obtain enzymatic
inhibition curves (Figure S2), which were used to identify the
appropriate concentrations for inhibiting different GSTs with
different magnitudes. As shown in Figures 5 (dotted line) and 6a
(purple bar), with 100 nM 1, the catalytic activities of GST P1
and GST A2 from group I were strongly inhibited, whereas the
enzymatic activities of GST M1 and GST A1 were decreased by
60 and 64%, respectively. GST M2 was inhibited by 20%. From
group II, GST S1 was inhibited by 53%, whereas GST A3 and
GST K1 were not inhibited under these conditions.
These results not only indicate the feasibility of obtaining an
ELR mimic that responds to the presence of specific proteins -
they also open the way for using it for GST identification. One
can couple the inhibition results obtained in channels 1 and 2 to
obtain a unique ID for most isozymes (Figure 6a). When these
data are combined with the activity of the enzymes in the absence
of 1 and using linear discriminant analysis (LDA), a pattern
recognition algorithm that can effectively classify unknown
samples (SI),10 isozyme differentiation could be further
improved. LDA converted these parameters into two factors
(F1,F2) that were plotted graphically (Figure 7a) to obtain well-
separated clusters of individual GSTs and combinations of two,
three, and even four isozymes (Table S4). To identify unknown
GST samples, the catalytic activity of each sample was followed in
channel 1, in the absence and presence of 1. In the next step, the
same sample was added to 1, and the change in its catalytic
activity was measured in channel 2. By associating the LDA
results with the corresponding clusters (Figure S8), 49 of 54
randomly selected samples could be identified, indicating 91%
accuracy. What distinguishes 1 from related systems used to
differentiate GSTs by arrays of various protein surface receptors8
or multiple inhibitors and substrates12 is that enzyme recognition
We then set out to determine how the different GSTs would
affect the catalytic activity of 1. We expected that, upon the
binding of 1 to the active site of GSTs, the interactions between
its catalytic unit and some of the isozymes’ surfaces would affect
the rate of cleavage of the fluorogenic substrate (Figure 1a).
Inspecting the catalytic activity of 1 at different concentrations
(Figure S3) revealed that 0.5 μM is the minimal amount of
catalyst required to obtain catalytic activity and that 2 μM is the
minimal concentration needed to attain a stable and reproducible
signal. By following the fluorescence signal generated by 2 μM 1
in the absence and presence of 1.5 μM GSTs, we could determine
which of the isozymes affects its catalysis. As shown in Figure 6b
(and 6a, green bar), GST P1 inhibited the catalytic activity of 1 by
90%, whereas GST S1 and GST M2 reduced the activity by about
65 and 74%, respectively. Interestingly, GST K1 induced 49%
enhancement in the catalytic activity, although this isozyme did
not seem to be inhibited by 1. This enhancement might thus
result from the use of much higher concentrations of the enzyme
and 1, which can enhance specific or nonspecific interactions
between them; both of them can change the configuration of 1
and, thereby, its catalytic response. To assess the contribution of
C
DOI: 10.1021/jacs.5b02496
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
is based on a unimolecular biomimetic device that can interact
with both the active site and the protein’s surface. A limitation of
this prototype is the lower catalytic activity of the artificial
esterase when compared with that of natural enzymes, requiring
higher concentrations of 1 and GST in channel 2.
REFERENCES
■
(1) Bjerre, J.; Rousseau, C.; Marinescu, L.; Bols, M. Appl. Microbiol.
Biotechnol. 2008, 81, 1. Kirby, A. J.; Hollfelder, F. From Enzyme Models to
Model Enzymes; RSC: Cambridge, 2009. Raynal, M.; Ballester, P.; Vidal-
Ferran, A.; van Leeuwen, P. W. N. M. Chem. Soc. Rev. 2014, 43, 1734.
Breslow, R. Acc. Chem. Res. 1995, 28, 146. Breslow, R. J. Biol. Chem.
2009, 284, 1337. Kofoed, J.; Reymond, J.-L. Curr. Opin. Chem. Biol.
2005, 9, 656. Breslow, R.; Dong, S. D. Chem. Rev. 1998, 98, 1997.
(2) Vial, L.; Dumy, P. New J. Chem. 2009, 33, 939.
The discriminatory ability of the system was further tested by
using it to differentiate among isozymes in human urine. GST P1
and GST A1 were selected as the analytes for this study because
elevated concentrations of these isozymes in urine have been
detected in kidney-related diseases.18,19 These urinary bio-
markers exhibit very similar activities, and hence, conventional
enzymatic assays, which can be used to diagnose high levels of
these isozymes, cannot determine their identity (Figure S7). In a
proof-of-principle experiment, human urine spiked with different
concentrations of GSTs, including medicinally relevant concen-
trations (0.4−0.8 μg/mL),18 was analyzed by our system. To
eliminate background reaction by urine esterases, the GST
content of each sample was enriched by a GSH column prior to
testing in channel 2 (SI). A LDA plot (Figure 7b) showed a clear
differentiation of different combinations and concentrations of
these isozymes, which enabled detection of 26 of 27 unknown
samples with 96% accuracy.
In conclusion, a novel enzyme mimic that integrates a catalytic
site and a protein recognition domain was created and used to
discriminate among structurally similar isozymes. In addition to
demonstrating a biomimetic approach to differential protein
sensing, this study highlights an important principle that could be
applied in future artificial enzyme design. The ability of 1 to be
engaged in enzyme−artificial enzyme interactions shows that
although enzyme mimics cannot yet compete with the catalytic
turnovers of natural enzymes, in terms of biomolecular
interactions, they can exhibit nanomolar binding affinities
comparable to those of natural proteins. Hence, beyond
modeling enzyme active sites, artificial enzymes of this class
could help to elucidate the parameters needed to couple protein
recognition with catalysis, a fundamental principle underlying the
function of signaling allosteric enzymes such as ELRs.
Considering the simplicity by which synthetic protein binders
can be attached to current artificial enzymes, we believe that
various other ELR mimics could be designed and contribute to
the development of stimuli-responsive biomimetics, biosensors,
and allosteric catalysts.
(3) Kovbasyuk, L.; Kramer, R. Chem. Rev. 2004, 104, 3161. Zhu, L.;
̈
Anslyn, E. V. Angew. Chem., Int. Ed. 2006, 45, 1190. Wiester, M. J.;
Ulmann, P. A.; Mirkin, C. A. Angew. Chem., Int. Ed. 2011, 50, 114.
(4) Oshovsky, G. V.; Reinhoudt, D. N.; Verboom, W. Angew. Chem.,
Int. Ed. 2007, 46, 2366.
(5) Breslow, R. Advances in Enzymology and Related Areas of Molecular
Biology; Wiley: New York, 2006; p 1.
(6) Alberts, B.; Johnson, A.; Lewis, J. Molecular Biology of the Cell;
Garland: New York, 2002.
(7) Lemmon, M. A.; Schlessinger, J. Cell 2010, 141, 1117.
(8) Motiei, L.; Pode, Z.; Koganitsky, A.; Margulies, D. Angew. Chem.,
Int. Ed. 2014, 53, 9289.
(9) Wright, A. T.; Griffin, M. J.; Zhong, Z.; McCleskey, S. C.; Anslyn, E.
V.; McDevitt, J. T. Angew. Chem., Int. Ed. 2005, 44, 6375. Zamora-
Olivares, D.; Kaoud, T. S.; Dalby, K. N.; Anslyn, E. V. J. Am. Chem. Soc.
2013, 135, 14814. Miranda, O. R.; Chen, H.-T.; You, C.-C.; Mortenson,
D. E.; Yang, X.-C.; Bunz, U. H. F.; Rotello, V. M. J. Am. Chem. Soc. 2010,
132, 5285. De, M.; Rana, S.; Akpinar, H.; Miranda, O. R.; Arvizo, R. R.;
Bunz, U. H. F.; Rotello, V. M. Nat. Chem. 2009, 1, 461. Margulies, D.;
Hamilton, A. D. Angew. Chem., Int. Ed. 2009, 48, 1771. Margulies, D.;
Hamilton, A. D. Curr. Opin. Chem. Biol. 2010, 14, 705.
(10) Collins, B.; Wright, A.; Anslyn, E. In Creative Chemical Sensor
Systems; Schrader, T., Ed.; Springer: Berlin, 2007; Vol. 277, p 181.
(11) Sieber, S. A. Activity-Based Protein Profiling; Springer: Heidelberg,
2012. Li, N.; Overkleeft, H. S.; Florea, B. I. Curr. Opin. Chem. Biol. 2012,
16, 227. Schmidinger, H.; Hermetter, A.; Birner-Gruenberger, R. Amino
Acids 2006, 30, 333. Sanman, L. E.; Bogyo, M. Annu. Rev. Biochem. 2014,
83, 249. Barglow, K. T.; Cravatt, B. F. Nat. Methods 2007, 4, 822.
Willems, L. I.; Overkleeft, H. S.; van Kasteren, S. I. Bioconjugate Chem.
2014, 25, 1181. Reymond, J.-L.; Wahler, D. ChemBioChem 2002, 3, 701.
Reymond, J.-L.; Fluxa, V. S.; Maillard, N. Chem. Commun. 2008, 46.
(12) Mannervik, B.; Alin, P.; Guthenberg, C.; Jensson, H.; Tahir, M. K.;
Warholm, M.; Jornvall, H. Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 7202.
̈
(13) Koike, T.; Kajitani, S.; Nakamura, I.; Kimura, E.; Shiro, M. J. Am.
Chem. Soc. 1995, 117, 1210.
(14) Subat, M.; Woinaroschy, K.; Anthofer, S.; Malterer, B.; Konig, B.
Inorg. Chem. 2007, 46, 4336.
̈
(15) Mahajan, S. S.; Hou, L.; Doneanu, C.; Paranji, R.; Maeda, D.;
Zebala, J.; Atkins, W. M. J. Am. Chem. Soc. 2006, 128, 8615.
(16) Eklund, B. I.; Moberg, M.; Bergquist, J.; Mannervik, B. Mol.
Pharmacol. 2006, 70, 747.
(17) Gasser, G.; Ott, I.; Metzler-Nolte, N. J. Med. Chem. 2010, 54, 3.
(18) Sundberg, A.; Appelkvist, E. L.; Dallner, G.; Nilsson, R. Environ.
Health. Perspect. 1994, 102, 293.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental procedures and kinetic experiments. This material
is available free of charge via the Internet at http://pubs.acs.org.
(19) Cawood, T. J.; Bashir, M.; Brady, J.; Murray, B.; Murray, P. T.;
O’Shea, D. Am. J. Nephrol. 2010, 32, 219. Westhuyzen, J.; Endre, Z. H.;
Reece, G.; Reith, D. M.; Saltissi, D.; Morgan, T. J. Nephrol. Dial.
Transplant. 2003, 18, 543.
AUTHOR INFORMATION
■
Corresponding Author
*david.margulies@weizmann.ac.il
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research was supported by the Minerva Foundation, the
International HFSP Organization, and a European Research
Council Starting Grant. Postdoctoral fellowship of Dr.
Selvakumar was partly funded by the Council for Higher
Education of Israeli Government under VATAT program.
D
DOI: 10.1021/jacs.5b02496
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX