1
02
R.N. Ezhov et al. / Journal of Photochemistry and Photobiology A: Chemistry 290 (2014) 101–108
outfitted with a compact ET-sensitizer, e.g. xanthone, whereas the
dithiane adduct is tethered to the second component (pink pen-
tagon labeled P). Binding brings the sensitizer into the vicinity of
the tag adduct, at which point irradiation commences. Xanthone
sensitizes fragmentation in the adduct, releasing the dithiane tag
into solution. The tags are then detected in a standardized analyt-
ical protocol revealing the identity of the bound pairs. A diverse
set of substituted dithiane tags is readily available for encoding.
Additionally, to further diversify the available variety of tags, we
have identified non-dithiane encoding tags, for example derivatives
of trithiabicyclo[2.2.2]octanes, which can be used in combinatorial
encoding [12].
Scheme 1. Externally sensitized photorelease of tags is contingent on molecular
recognition.
Tagged combinatorial libraries are well precedented. In the clas-
sical bead-tagging approach, for example the one developed by
Clark Still [13], each bead is encoded with the full set of tags needed
for subsequent identification of the ligand displayed on the surface
of the bead. Still’s strategy is not suitable for encoding individ-
ual molecules, which simply do not have enough “real estate” to
accommodate all the encoding tags without running into a risk of
multiple tags interfering with molecular recognition events. Our
strategy is to encode individual molecules one tag at a time. Every
library member can be encoded with a set of tags so that a certain
fraction of its molecules are encoded with the first tag, another frac-
tion – with second, etc., with the net result of each ligand being
present in the solution as several sub-populations cumulatively
encoded with all the tags necessary for its subsequent identifica-
tion. Irradiation in this case yields the desired result because all the
tags encoding individual molecules in the bound pairs are collec-
tively released into the solution, where they are analyzed revealing
the identity of the bound compound.
Initially we proved this general concept of tagging the individual
(unsupported) library members and screening for molecular recog-
nition using the known binding pair, avidin–biotin [14]. In these
experiments a mini library of compounds, which included biotin,
was encoded by tagging each compound with different dithianes.
Scheme 2 gives a general outline for screening of such library of lig-
ands (biotin is depicted by a pink “B” pentagon), encoded with the
tethered tags (pink circles 1–5). Avidin (the green octagon “A”) is
outfitted with a xanthone-based ET-sensitizer, and incubated with
the library. The binding event brings the sensitizer into the vicin-
ity of the tag/adducts (encircled with red), which, upon irradiation
triggers the release of tags 1–5, i.e. only the tags which encode
biotin.
So far our dithiane tags-encoding methodology was tested with
a model barbiturate-binding artificial receptor [15] and validated
for the avidin–biotin binding pair [14], known to have a very
tight KD. In this study we extend this approach onto biological
systems exhibiting low micro molar affinities and demonstrate
that it can be used to differentiate between very similar epi-
topes modified via post-translational modifications (PTMs), such as
unmodified and methylated lysine. Our choice of the protein sub-
strate is ING2 PHD (plant homeodomain) finger, which is known to
recognize a hexapeptide fragment of the histone H3 tail Ala-Arg-
Thr-Lys-Gln-Thr, but only when the lysine residue is trimethylated,
i.e. Ala-Arg-Thr-Lys(Me3)-Gln-Thr. The trimethylated epitope is
referred to as H3K4me3.
Nuanced aspects of molecular recognition play a pivotal role
in biological processes. This is especially important for the emerg-
ing field of epigenetics, where PTMs regulate complex signaling
cascades, including gene transcription, DNA repair, recombina-
tion, and replication and chromatin remodeling [16,17]. Epigenetic
misregulations have been linked to various human diseases includ-
ing cancer, premature aging and neurodegenerative disorders, and
thus development of experimental approaches to characterize PTM
recognition is essential in understanding the epigenetic mecha-
nisms, PTM-driven functional outcomes, and disease-associated
trimers exceeds 244 million and the number of tetramers exceeds
1
50 billion.
Various approaches are used to test for interactions between
biomolecular entities in solutions: differential scanning calorime-
try, titration calorimetry, analytical centrifugation, spectroscopy
including NMR, plasmon resonance, and mass spectrometry, of
which only the latter has enough “bandwidth” to deal with
modest combinatorial collections of biomolecules. Detection of
non-covalent interactions between biological molecules by ESI
mass-spectrometry was pioneered in early 90s, most prominently
by Ganem [5a,5b](for a review see [5c]). Later MALDI-TOF methods
followed [6]. It is still not clear whether or not gas phase affinities,
measured by these methods, reflect the actual binding in solution.
Yet, while current analysis is not ideal, the comprehensive
understanding of the fundamentals of peptide–peptide inter-
actions is at the core of a number of critical problems in
biomedical sciences. One specific problem concerns diseases
involving aggregation of peptides or small proteins, especially
such neurodegenerative disorders as Alzheimer’s disease, Hunt-
ington’s disease etc. [7]. It is not surprising then that the studies
directed toward better understanding of protein folding (for exam-
ple in the DeGrado [8] or Gellman [9] labs) are always intertwined
with studies of peptide–peptide interaction and oligomerization.
Unfortunately, the progress in the rational design of short peptides
capable of forming dimers or trimers is at best spotty. There are few
prominent examples of this in the literature, including one from the
Imperiali lab [10]. Combinatorial approaches were designed to aug-
ment rational design, especially for cases of great complexity where
theories are lacking or deficient. If there were a high throughput
method to test for peptide–peptide interactions one would obtain
invaluable information about the very basic motifs of such inter-
molecular binding. Regrettably, there are no methods currently
available to test for such interactions in a truly high-throughput
manner.
We have recently developed a method for chemical encoding of
combinatorial libraries with photolabile externally sensitized tags,
which will allow us to pre-screen solution phase libraries for bound
dimers or higher oligomers or, in more complex cases, to dramat-
ically narrow the possibilities to a manageable subset of potential
candidates. Our tagging methodology makes use of dithiane-based
photolabile molecular systems that are capable of photoinduced
fragmentation only when in the presence of an external sensitizer.
The mechanism of such fragmentation has been shown to involve
oxidative single electron transfer (ET) from the dithiane moiety to
the excited (most commonly, triplet) ET-sensitizer followed by a
mesolytic fragmentation in the generated cation radical [11]. In
these systems, photoinduced fragmentation is contingent upon
the occurrence of a molecular recognition event. In other words,
a molecular recognition event is needed to arm the system, after
which it becomes photolabile.
Scheme 1 outlines the general concept of such a system: one
component of the molecular recognition pair (green octagon), is