FlICk (fluorescent isoindole crosslinking) for peptide stapling

Article abstract of DOI:10.1016/bs.mie.2020.04.017

The rigidification of peptide secondary structure via stapling is an important and enduring goal in the development of functional peptides for biochemical and pharmaceutical applications. In addition, the incorporation of fluorophores and chromophores has

Full text of DOI:10.1016/bs.mie.2020.04.017

CHAPTER FOURTEEN
FlICk (fluorescent isoindole
crosslinking) for peptide stapling
Mihajlo Todorovic, David M. Perrin*
Department of Chemistry, University of British Columbia, Vancouver, BC, Canada
*Corresponding author: e-mail address: dperrin@chem.ubc.ca
Contents
1. Introduction
314
2. Materials
3. Peptide preparation
4. FlICk reaction
5. Modifications for bicyclic alpha amanitin analogue
6. Preparation of modified ortho-phthalaldehydes
6.1 Borane THF reduction
7. Swern oxidation
323
324
325
326
327
327
329
330
331
8. Conclusions
References
Abstract
The rigidification of peptide secondary structure via stapling is an important and endur-
ing goal in the development of functional peptides for biochemical and pharmaceutical
applications. In addition, the incorporation of fluorophores and chromophores has been
a sought-after application for creating peptidic probes of cellular function and localiza-
tion. The combined application of peptide stapling and fluorescent-readout is featured
by the reaction of ortho-phthalaldehydes to create isoindole staples, thus transforming
inactive linear and monocyclic precursors into fluorescent or visibly colored monocyclic
and bicyclic products with noted biological activity. Given its user-friendliness, we have
termed this approach FlICk (fluorescent isoindole crosslink) chemistry and we have
featured this application on an array of high-affinity macrocyclic α-MSH derivatives
as well as for late-stage intra-annular isoindole stapling furnished a bicyclic peptide
mimic of α-amanitin that is cytotoxic to CHO cells. The synthetic methods for preparing
substituted ortho-phthalaldehydes along with subsequent applications to FlICk stapling
are detailed herein.
#
Methods in Enzymology, Volume 639
ISSN 0076-6879
https://doi.org/10.1016/bs.mie.2020.04.017
2020 Elsevier Inc.
All rights reserved.
313

314
Mihajlo Todorovic and David M. Perrin
1. Introduction
The development of bio-orthogonal reactions for rapid, high-yielding,
site-selective installation of molecular functionality has wide-ranging impli-
cations in peptide chemistry. Noteworthy goals include constraining peptide
conformations and achieving structural rigidification, both of which can be
achieved with stapling. In addition, there is enduring interest in grafting of
reporter groups, most notably in the context of chromophores, for the con-
struction of absorbent and fluorescent probes. Rarely, a reaction that affords
stapling can be designed to provide an emissive or visible spectral signature
that not only confirms successful stapling but through the installation of
a chromophore, enables applications in screening and target identification.
Indeed, the very development of such a reaction and its methodological
implementation for use in peptide chemistry is the basis of this report.
On this subject, a classic analytical reaction used to quantify the concen-
tration of a primary amine present in a sample involves the condensation of
ortho-phthalaldehyde (OPA) with a primary amine and a thiol to produce
a fluorescent 1-thio-2H-isoindole. The OPA assay has found broad utility
in amino acid analysis (Roth, 1971) as well as in the food and beverage
industries (Dukes & Butzke, 1998) (Fig. 1).
To date, for most applications, OPA-mediated isoindole formation finds
its primary use in quantifying peptides that have at least one free amine (i.e.,
N-terminal, lys) (Go, Horikawa, Garcia, & Villarreal, 2008). Quantifying
the exact amount of peptide ligand to a high degree of certainty is essential
to accurately measure its Kd/IC50 values in a dose-dependent assay. Such is
particularly at issue in cases where the ligand, either produced or isolated in
minute quantities, renders gravimetric quantification unreliable due to the
inherent imprecision of weighing sub-milligram amounts of material as well
as the unaccountable presence of contaminating trifluoroacetic acid (TFA)
or formic acid (FA) molecules residually found in lyophilized samples.
Fig. 1 OPA condensation with a primary amine and a thiol to form a fluorescent
isoindole.

FlICk for peptide stapling
315
In contrast, using OPA to quantify peptides and amino acids results
in consistent extinction coefficients irrespective of amino acid identity
(Herna´ndez, Caman˜as, & Alvarez-Coque, 1990) leading to accurate qu-
antification of primary amine containing peptides. With excess OPA and
thiol (dissolved in aqueous buffer pH 8–10), isoindole formation is gen-
erally complete upon mixing enabling quantification of the amine of
interest. Absorbance of the resulting isoindole is acquired at 335nm
(λ-max) and concentration is calculated using an extinction coefficient
of 6800MÀ1 cmÀ1, found to be generally constant irrespective of the side
chain of the amino acid in question (Herna´ndez et al., 1990). This method
can be combined with HPLC analysis for even greater accuracy (Herna´ndez
et al., 1990). A variety of thiols can be employed in the OPA assay, with
β-mercaptoethanol being employed early on (Roth, 1971; Simons &
Johnson, 1977, 1978). However, use of this particular thiol has the draw-
back of producing an isoindole that autodegrades to an isoindolin-1-one on
standing in the reaction media (Fig. 2A). The rate of degradation appears to
be enhanced by super-stoichiometric OPA concentrations that are often
used in analysis (Simons & Johnson, 1977, 1978; Stobaugh et al., 1983).
This observation let to a proposed mechanism that involves the formation
of an unusual episulfonium intermediate (Stobaugh et al., 1983) that pre-
sumably undergoes hydrolysis by a different pathway (Fig. 2B). Use of a
A
B
Fig. 2 (A) Isoindole degradation mechanism with anchimeric assistance from
2-mercaptoethanol, originally proposed by Simons and Johnson (1977). (B) OPA-
enhanced degradation proposed by Stobaugh, Repta, Sternson, and Garren (1983).

316
Mihajlo Todorovic and David M. Perrin
longer alkyl chain that separates the thiol from the hydroxyl group greatly
reduces this problem (Stobaugh et al., 1983) as does the use of thiols lacking
an alcohol, such as N-acetyl cysteine. However, Jacobs found that isoindole
degradation was expedited by OPA even in cases where the thiol employed
lacked a pendant alcohol and proposed a mechanism with nucleophilic
attack on OPA by the isoindole (Leburg & Jacobs, 1986). A contemporary
review found no general agreement on the mechanism of degradation
(Garc´ıa Alvarez-Coque, Medina Herna´ndez, Villanueva Caman˜as, &
Mongay Ferna´ndez, 1989).
The rapidity of this reaction, along with the mild aqueous conditions
under which it proceeds, makes it highly compatible with applications in
biological chemistry and thus it has also found applications in enzymology.
To wit, as lysines and cysteines often play key catalytic roles within the
active site of enzymes, OPA naturally finds a role as an active site titrant.
Employing OPA, others showed that yeast hexokinase could be deactiva-
ted, implicating a lysine and cysteine in catalysis (Puri, Bhatnagar, &
Roskoski, 1988). In a related application, selective active site labeling was
achieved through the synthesis of a “Reporter Affinity Label,” an OPA-
β-funaltrexamine conjugate that could react specifically with the opioid
receptors to provide a fluorescent readout confirming their presence
(Le Bourdonnec et al., 2000) (Fig. 3A). The idea of exploiting the unique
reactivity of OPA for enzyme-selective bioconjugation was elegantly ex-
tended into the realm of kinases. By constructing an OPA-adenosine con-
jugate, others showed that kinases/substrate pairs could be covalently
tethered only upon substrate recognition and the resulting isoindole once
again serves as a fluorescent reporter (Maly, Allen, & Shokat, 2004)
(Fig. 3B). The labeling was facilitated by preparing “mutant” peptide sub-
strates where a cysteine is introduced in place of a serine or threonine res-
idue, which would be naturally phosphorylated (Maly et al., 2004).
In light of increasing pharmaceutical interest in biologic drugs, demand
for new peptide stapling methods has grown. Desirable attributes of these
reactions include mild aqueous conditions, rapid kinetics, chemoselectivity,
and access to readily available reagents. When considering these constraints,
OPA condensation emerges as a viable candidate for investigation. Indeed,
an extensive survey of the literature provided scant examples towards this
goal; the sole exception being crosslinking of glutathione for the purpose
of assessing the extinction coefficient of the resulting isoindole (Simons &
Johnson, 1978).

FlICk for peptide stapling
317
Fig. 3 Enzymatic labeling reactions of OPA; (A) Reaction of an opioid receptor with an
OPA-opioid conjugate; (B) three-component crosslinking of a kinase-substrate pair by
an OPA-Adenosine conjugate.
Notwithstanding such precedent, it was surprising that isoindole for-
mation had yet to be explored in the context of directed peptide stapling
despite decades of work on both intermolecular variants of the reaction
with amino acids and amine-containing peptides reacted with exogenous
thiols as well as intramolecular variants with enzymes and receptors. Thus,
we decided to explore this reaction and discovered the smooth application
of OPA condensation to prepare peptide mono- and bicyclic peptides
(Todorovic et al., 2019). Confident that this method would work rather gen-
erally across many peptides, we focused our efforts to develop this method-
ology directly on biologically relevant and potentially highly-active peptides,
avoiding preliminary investigations on sequences with limited functionality
or of non-evident pharmacological/biochemical significance. In so doing,
we sought important and meaningful biological readouts to provide the com-
munity with confidence, not only in the robustness of the chemical reaction,
but in its potential for imparting high biological activity to the resulting sta-
pled peptides. In the course of this work, we synthesized new OPA analogs
(vide infra), and in so doing, we termed this method: Fluorescent Isoindole
Crosslinking, or “FlICking” for short.

318
Mihajlo Todorovic and David M. Perrin
Before discussing the application of FlICk staples in test peptides of bio-
logical relevance, we appreciated that the area of substituted OPAs is
scantly explored. Hence, in broadening this application, we saw potential
for tuning the photophysical properties of isoindoles through the synthe-
sis of a series of substituted OPAs. To this end, we selectively reduced
a small set of phthalic acids to their corresponding diols with Borane-
THF complex, leaving nitro groups untouched where present (Fig. 4).
The use of a Swern-type oxidation was critical to achieving the desired
oxidation at both benzylic carbons, as in our hands, other oxidants which
produced a discrete aldehyde prior to quenching led to complex mixtures
(while oxidant was still present in the reaction pot), commonly including a
lactone as a major product. The products of the Swern oxidations were
isolated in the form of methylhemiacetals in case of electron-deficient
OPAs. The condensation of these phthalaldehydes with hexylamine and
N-Acetyl cysteine produced isoindoles in short order (<1min) in all cases
expect the nitro-containing OPAs, most notably, the 3-Nitro OPA which
required ꢀ20min to reach completion.
Fig. 4 Synthesis of Substituted OPAs. Compounds were isolated as hemiacetals with
methanol except in the case of the 4-Fluoro and 4-Bromo substituted OPAs.

FlICk for peptide stapling
319
In the case of the nitro-OPAs, the ʎmax was significantly red-shifted into
the visible range, leading to orange-red products, which were unfortunately
not fluorescent in aqueous media. Halogen substitution had little effect on
the ʎmax, but significantly affected the fluorescent output, with the 4-Fluoro
OPA being the most fluorescent of the modified OPAs in our model system.
Gratifyingly, when we measured the quantum yield of the peptides stapled
with fluorogenic OPAs, we found that fluoro-substitution significantly
increased fluorescent output (Table 1).
Hence, in applying FlICk synthesis to peptide stapling, we synthe-
sized isoindole stapled macrocyclic analogues of a melanoma drug lead
(Melanotan II) and of a bicyclic natural product peptide toxin (α-amanitin)
to provide two important comparators for valuing the application of
FlICks. Our choice for FlICking a heptapeptide MSH sequence was
informed by the fact that MSH has been heavily investigated over the years
and is known to be challenging in terms of delivering biological potency
by standard stapling reactions, particularly in the context of a short hepta-
peptide that, in the linear format, shows no measurable biological activity.
Nevertheless, at least one standard staple, an isopeptide linkage between
lysine and glutamate, afforded Melanotan-II (MTII), a clinically-trialed
peptide macrocycle that shows low nanomolar dissociation constant values
for the melanocortin-1 (MRC-1) receptor. Although MTII ultimately failed
as a treatment for melanoma, recently it was elaborated for 18F- and
68Ga-labeling and successfully used to PET-image melanoma xenografts in
mice thus providing new radiochemical leads for imaging and targeted ther-
apy (Zhang et al., 2018). As such, MTII represents a readily accessible peptide
macrocycle that may serve as a benchmark against which new peptide stables
can be evaluated in this context.
We arrayed this series of OPAs against four linear MSH precursors with
varying lengths of the amine-bearing side chain to afford a focused library
of 24 FlICked peptides (Fig. 5). Reaction in the context of this series of
linear heptapeptides was rapid and generally complete in under a minute
when performed in the presence of excess OPA (e.g., concentrations of
1–10mM). The cyclization yields, as assessed by HPLC, ranged from mod-
erate to roughly quantitative (as the linear precursors were ꢀ75%–80%
pure). All FlICked MSH peptides bound MCR-1 in a competitive binding
assay that employs a standard radio ligand. When diaminobutryic acid, or
an amino acid with a longer amine-bearing sidechain was used, an order
of magnitude of binding affinity was gained over the diaminopropionic acid

Table 1 Photophysical properties of substituted isoindoles.
4/7-Nitroisoindole 5/6-Nitroisoindole 4/7-Fluoroisoindole 5/6-Fluoroisoindole 5/6-Bromoisoindole Isoindole
λex (nm)
λem (nm)
487
444
329
415
86
338
454
116
345
459
114
338
452
117
a
a
Stokes Shift
(nm)
–
–
–
–
Quantum
Yield
0.04
0.30
<0.01
0.30–0.47 (Chen,
Scott, & Trepman,
1979)
0.02a
8200
0.31b
7400
0.13c
Ex.
coefficient (ε)
5600
5200
6700
6500–7000 (Herna´ndez
et al., 1990)
a3-FOPA derived peptide.
b4-FOPA derived peptide.
cOPA derived peptide.
R1 is N-Acetyl Cysteine and R2 is hexlyamine expect where starred.

FlICk for peptide stapling
321
A
B
C
Fig. 5 (A) Melanotan-II, (B) substituted OPAs employed in library synthesis; (C) FlICked
MSH synthesis, with yields in the table below (measured by HPLC with absorbance at
280nm); at right a sample of HPLC-peptide, directly eluted, as visualized by use of hand-
held UV lamp.
variant. Ornithine-derived FlICk-MSH exhibited the highest affinity with a
Ki ꢀ1nM, irrespective of isoindole substitution.
In order to prepare a bicyclic FlICk peptide macrocycle with the poten-
tial for biological activity, we turned our attention to alpha-amanitin, a
potent natural bicyclic peptide toxin found in A. phalloides, which is under
investigation as a payload for clinically emergent antibody-drug conjugates.

322
Mihajlo Todorovic and David M. Perrin
Amanitin is known to be highly rigid with a unique tryptathionine staple
contributing significantly to its structural definition and inherent toxicity.
Although there is limited structure activity relationship data due to limited
synthetic accessibility (Matinkhoo, Pryyma, Todorovic, Patrick, & Perrin,
2018), the natural product is cytotoxic against CHO cells at 1–2μM. As with
MTII, we supposed that amanitin could provide an important control for
evaluating the biological activity of FlICk staples in the context of an anal-
ogous cytotoxic bicyclic peptide. Notably, our choice of amanitin was also
based on the appreciation that the tryptathionine staple itself represents a
constitutional isomer of a FlICk staple thus making amanitin a highly
relevant model for testing aspects of potential bio-isosterism (Fig. 6).
Given the circularly permutable nature of peptide macrocycles, we rec-
ognized that two regioisomeric FlICked-amanitins could be prepared.
Molecular modeling of both (Fig. 6) led us to note not-so-subtle differences
in the orientation of the isoindole with respect to peptide macrocycle as well
as significant differences in terms of the orientation of the side chains on the
macrocycle that are constrained by virtue of this topology of the FlICk.
While it is known that the tryptathionine engages in both a pi-stacking
and pi-cation interactions at the active site of RNA Polymerase II, our
choice favored the FlICked amanitin that appeared to provide greater
Fig. 6 Modeling of FlICked amanitins. (A) The chemical structures of β-amanitin along
with the two possible cyclic topologies of a FlICked amanitin. (B) The crystal structure of
β-amanitin juxtaposed with energy-minimized molecular models of FlICked amanitins.

FlICk for peptide stapling
323
overall macrocyclic geometry compared to alpha-amanitin despite signifi-
cant alteration in orientation of the isoindole.
In contrast to FlICk synthesis on a linear heptapeptides that was essentially
complete on mixing, for the bicyclic amanitin-like structures, FlICking
required up to 30min at the same concentrations of OPA, likely reflecting
inherent ring strain in such bicylic peptides. Remarkably, despite different
orientations of the isoindole, FlICked-amanitin-1 exhibited an IC50 on
CHO cells of 70μM. Notably, the macrocyclic precursor showed no in-
hibitory properties. While we are unable to assert that FlICked amanitin
engaged RNA Pol II to promote cell death, were this not the case, then
it is equally interesting to contemplate the eventuality of discovering a
new target for cytotoxicity.
Finally, to demonstrate the generalizability of this method for other pep-
tide sequences, we also sought a route to synthesizing FlICk peptides con-
taining free amines. Hence, we found that we could FlICk peptides
containing amines as long as they were protected with either Dde or
Fmoc, which could then be subsequently removed with either hydrazine
or DBU respectively. The model peptides studied in our published work
showed that the reaction conditions were chemoselective and left other
potentially reactive peptide side chains untouched (alcohols, imidazoles,
indoles, guanidines, etc.). In additional unpublished work, we confirmed
that the reaction is absolutely specific to primary amines, leaving secondary
amines unmodified even when accomplished with excess OPA. Herein, we
describe the details of all practical aspects of FlICk chemistry relevant to car-
rying out the reaction in the contexts we have so far explored. Given the
interest in a rapid and robust peptide stapling method, herewith we wish
to report on the specific practical methodology of this reaction.
2. Materials
All reagents and solvents were acquired from vendors at the highest
purity specification available except where noted.
MilliQ Water: Tap water was purified using an EMD Millipore Milli-Q®
IQ 7000 water purification system equipped with an EMD Millipore
Progard® 2 cartridge and dispensed through a 0.22μm Millipak filter.
Borate Buffer: Dissolve 6.2g of boric acid and 2.83g of NaOH in 1L of
milliQ water. Check the pH, it should be around 9–9.5, if not adjust
accordingly.

324
Mihajlo Todorovic and David M. Perrin
0.1% formic acid H2O: Add 1mL of formic acid to milliQ H2O to make
a 1L solution. For use with HPLC, filter through a 0.2μm PAL filter
0.1% formic acid MeCN: Add 4mL formic acid to HPLC grade MeCN
to make a 4L solution. For use with HPLC no further manipulations
were done
40mM NH4OH(aq) (pHꢀ8): Add 4.7mL conc. Ammonium Hydroxide
to milliQ H2O to make a 1L solution. Add formic acid to reach a pH of
ꢀ8. For use with HPLC, filter through a 0.2μm PAL filter.
Pierce™ Centrifuge Columns (89896) 2mL, 5mL or 10mL variety.
SiliaFlash® Silica, 40–63μm (230–400 mesh).
˚
Merck Silica gel high-purity grade (9385), pore size 60A, 230–400 mesh
particle size.
Gilman Pipettes were used to measure μL volumes.
3. Peptide preparation
1. Cleave the peptide from the resin by incubation with 95:2.5:2.5 TFA/
TIS/H2O for a period of around 2h at room temperature in a pierce
centrifuge column.
2. Drain the contents of the column and collect the filtrate into a round
bottom flask. Concentrate the solution under reduced pressure to afford
a gummy residue.
3. Triturate the residue with Et2O. Add ꢀ5mL of Et2O to the flask, and
pipette the Et2O up and down and rinse the walls of the flask with a
pasteur pipette. Collect the resulting residue, resembling flakes, at the
bottom of the flask and transfer it to a 15mL falcon tube along with
the residual Et2O. Spin the contents of the falcon tube spun down in
a clinical centrifuge until a tight pellet is achieved. Pour out the super-
natant and add 3mL of fresh Et2O to the falcon tube. Vortex the tube
and then spin it down as described above. Repeat this once again for
a total of three triturations.
4. Take the cap off the tube and wrap a piece of filter paper around the top
of the tube with holes poked in it. Allow the residue to dry overnight.
5. Once dry, dissolve the peptide residue in a minimal amount of 0.1% formic
acid H2O. Note: Even if the peptide is incompletely soluble, do not add
MeCN at this stage as it may impair resolution on reverse phase silica.
6. Prepare a reverse phase silica column (SEP-PACK) by flushing the col-
umn with MeOH ꢀ5–10mL. Then load the column by adding the

FlICk for peptide stapling
325
crude peptide dissolved in 0.1% formic acid H2O. The rate of elution can
be aided by using a syringe with a rubber septum attached to the end. Run
an additional ꢀ5–10mL 0.1% formic acid H2O through the column.
7. Collect successive fractions of ꢀ10mL of 9:1 (v/v) 0.1% formic acid
H2O/MeCN, 8:2 (v/v) 0.1% formic acid H2O/MeCN, 7:3 (v/v) 0.1%
formic acid H2O/MeCN, 6:4 (v/v) 0.1% formic acid H2O/MeCN.
Check the fractions for the presence of product using either ESI Mass
Spectrometry or analytical reverse phase HPLC. If using ESI, the uncon-
centrated elute can be directly analyzed.
8. Pool the fractions containing the desired product in a 50mL Falcon
tube, freeze and lyophilize them. If quantifying large amounts of peptide
by gravimetry, pre-weigh the falcon tube.
4. FlICk reaction
1. Dissolve the quantified crude peptide in convenient amount of 0.1%
formic acid H2O (often 1mL). Note: It is very important to use an
acidic solution to dissolve the peptide at this stage, as it will have a free
thiol. The use of neutral H2O has been observed to result in significant
disulfide formation in the course of a few days. Crude linear peptides
dissolved in 0.1% formic acid H2O can be stored at 4°C for a period of
months and remain usable in FlICk reactions.
2. Pipette out a convenient amount of peptide into a 15mL falcon tube
(ꢀ0.5μmole was often used) and freeze and lyophilize it.
3. Add 80μL Borate Buffer (ꢀ1M HEPES buffer to pH 8–9 with
K3PO4(aq) can be substituted and led to cleaner reaction in some cases)
to the dry peptide in the falcon tube and vortex to dissolve. Add 20μL
0.05M XOPA dissolved in EtOH and vortex. Note: The reaction
tended to proceed faster and cleaner if the OPA solution was made prior
to immediate use. It can be stored at room temperature for weeks and
remain usable, or for at least a year if stored at 4°C. The reaction can be
deemed complete when a fluorescent precipitate forms. For unmodified
ortho-phthalaldehyde or the halo derivatives, this took only a few sec-
onds. For the nitro-derivatives, 5–20min was required. In general, most
reactions were simply allowed to proceed for 20min.
4. To redissolve the precipitate, add 20μL DMF or 0.5μL formic acid to
the solution and vortex. If desired, ESI mass spectrometry can be used
to check for reaction completion. If the acidic work up is done, this

326
Mihajlo Todorovic and David M. Perrin
solution can directly be analyzed, if not, adding formic acid to the sample to
reach pH ꢀ2.5–3 is ideal for observing a strong signal in positive ion mode.
5. Directly inject the entire reaction mixture onto reverse phase HPLC.
In the case of our MSH analogue peptides we used a gradient elution
with 0.1% formic acid H2O against 0.1% formic acid MeCN. 25%–
35% over 21min. In general, we found it best to use a more general
method with a wide range of MeCN proportion and then make a
focused method noting the MeCN proportion range at which the rel-
evant peaks elute. This reduced the time per run and solvent use. The
elution of the product was followed at 335nm for unmodified isoindoles
and at the respective ʎmax of the given modified isoindole. To evaluate
the reaction progress and ensure the consumption of the starting ma-
terial, channels at the ʎmax of useful sidechains (280 or 260nm for
Tryptophan or Tyrosine, respectively) and 230nm for general amide
bonds. Once the suspected product peak is collected, ESI mass spec-
trometry can be used to confirm the identity product by directly sam-
pling the column eluent.
6. To ensure no byproduct formation, cool the product immediately after
collecting the fraction. If possible, freeze and lyophilize the product
immediately. However, most isoindole peptides we have made can be
stored at À20°C for days to weeks in the acidic solution that comes
off the HPLC with minimal or no degradation. Note: storage at room
temperature, in solution, in the direct path of sunlight leads to bleaching
of the isoindole over time.
7. To accurately quantify the isolated peptides, dissolve the dry powder
in H2O and read at 335nm if it is an unmodified isoindole or at the
respective ʎmax of the given modified isoindole. A molar extinction
coefficient of 6800 was used for unmodified isoindoles or the extinction
coefficient we measured for the modified isoindoles.
5. Modifications for bicyclic alpha amanitin analogue
1. Despite the use of 0.1% formic acid H2O, we observed disulfide forma-
tion in the case of the monocyclic precursor in this case.
2. To enable smooth reaction progress use 50mM TCEP Borate Buffer
(pHꢀ8–9) in place of regular borate buffer if significant disulfide for-
mation has occurred. Note: It is critical that pH of the buffer be basic

FlICk for peptide stapling
327
at this step, or the reaction rate will be very slow or the product not form
at all. If TCEPÁHCl is used to make the buffer, check the pH and
add NaOH(aq) to reach pH 8–9. For the amanitin analogue, a reaction
time of around 30min was usually used, but the reaction is likely done
in less than 10.
3. The amanitin analogue was found to be sensitive to acid, so DMF was
used if necessary or nothing was added if no precipitate was observed.
4. In the case of the amanitin analogue, the use of acidic HPLC eluents
resulted in no product isolation. So, the use of 40mM NH4OH buffered
to pH ꢀ8 with formic acid was essential to the isolation of product.
Use MeCN with no additive as the other solvent. We suspect that
the increased sensitivity of the amanitin analogue was due to the high
degree of ring strain and that these precautions may not be necessary
for other bicyclic isoindole peptides.
6. Preparation of modified ortho-phthalaldehydes
6.1 Borane THF reduction
1. Flame dry a round bottom flask containing a stir bar. Alternatively, if
the precursor is made in house and is a gel on the inside of the flask,
dry the flask containing the precursor (and a stir bar) in a desiccator
under vacuum against P2O5 (housed in a beaker) for 2–3h. After this
period, quickly take the flask out of the desiccator and flush with
Argon 3 times after evacuating the flask of its previous atmosphere.
2. Add dry THF (freshly distilled) to the flask under Argon and stir until the
precursor is dissolved. Alternatively, if a commercial phthalic acid or
powder precursor is used, dissolve it in dry THF (also freshly distilled)
and add it to the reaction flask under Argon. Cool the contents of the
flask to À78°C by placing the flask in an aluminum dewar containing
a slurry of dry ice in acetone.
3. After allowing ample time for the flask to cool, add 10 equivalents of
1M Borane-THF complex (solution in THF) dropwise to the stirring
reaction mixture. Note: The use of fresh Borane-THF complex is critical
to reaction success. Old and compromised bottles result in a complex mix-
ture of partially reduced products in our experience. Note: Slow addition
at low temperature is critical as the reaction is highly exothermic, but with
appropriate precautions the reaction has been safely carried out at scale.

328
Mihajlo Todorovic and David M. Perrin
4. After adding the full volume of Borane-THF solution allow the reaction
to stir at À78°C for 30min. Then, remove the flask from the dewar and
allow the reaction to stir at room temperature overnight.
5. After overnight stirring cool the reaction mixture to 4°C by placing the
flask in an ice bath and ensure that the stirring continues. Quench the
reaction by slowly adding ꢀ2/3 reaction volume of 1:1 THF/H2O
dropwise to the reaction flask via syringe.
6. After opening the flask to the atmosphere, transfer the solution to a
separatory funnel and add solid K2CO3 to the funnel until two layers
form. Collect the THF layer and add another portion of THF to the
separatory funnel and extract the aqueous layer once more and pool
the resultant THF with the first THF layer. Dry the combined THF
layers over MgSO4 and concentrate the solution under reduced pressure
on rotovap to yield solids dispersed in a thick oil.
Alternate workup: After quenching add ꢀ2 total volume H2O and
transfer to separatory funnel, then extract with three successive portions
of EtOAc (ꢀ1/3 volume of total solution in funnel per portion). Pool
the organic layers. Note: If some product remains in the aqueous layer
can be acidified to pH ꢀ1 with conc. HCl(aq) and extracted with succes-
sive portions of EtOAc (ꢀ1/5 volume of total solution in funnel per
portion). Check the new layers by TLC to ensure no product degrada-
tion upon acidification, if clean, combine the organic layers and dry
over MgSO4 and concentrate the solution under reduced pressure on
rotovap to yield solids dispersed in a thick oil.
7. Purify the crude product by flash chromatography on silica. Silicycle
40–63μm silica can be used, Merck grade silica is unnecessary. Set up
the column by first adding sand to the glass tube then adding some of
the loading solvent. At this point carefully add the silicycle as a slurry
in the loading solvent at first, once a base of silica is established the rest
can be more quickly. Load the product (dissolved in a minimum amount
of 98:2 (v/v) DCM/EtOH) with 98:2 (v/v) DCM/EtOH and elute
with 96:4 DCM/EtOH. Note: Change the eluent composition if the
rf of the given substituted phenyl dimethanol deviates significantly from
this report. Evaluate the purity of each fraction by TLC with 92:8 (v/v)
DCM/EtOH. Pool the pure fractions and concentrate the solution
under reduced pressure on rotovap to a white or colored solid. Note:
MeOH can be used in place of EtOH.

FlICk for peptide stapling
329
7. Swern oxidation
1. Dissolve the starting material in dry DCM if possible. If it is insoluble
(as is the case with several substituted phenyl dimethanols) use a minimal
amount of dry THF or dry DMSO.
2. Flame dry a round bottom flask containing a stirbar. After allowing it to
cool add the appropriate volume of DCM (final concentration of the
reaction should be ꢀ0.19M). Cool the reaction to -78°C and carefully
add Oxalyl chloride (2.2eq) dropwise, to the stirring reaction mixture.
3. SlowlyaddDMSO(3.4eq.)dropwisetothestirringreactionmixture. Note:
This process is highly exothermic, take precautions to minimize the evolu-
tion of an exotherm. Allow the reaction to stir for 15min at À78°C.
4. Add the dissolved starting material slowly to the stirring reaction mix-
ture. Stir the reaction at À78°C for 2h.
5. Add Triethylamine (18eq) dropwise to the stirring reaction mixture and
allow it to proceed for 10min at À78°C. Then allow it to slowly warm
up to room temperature removing the dry ice/acetone bath. Allow the
reaction to stir at room temperature for 30min.
6. To quench the reaction there are two options:
(a) If the substituent on the phthalaldehyde is highly electron with-
drawing (in our experience, the 3-fluoro, 3-nito, and 4-nitro)
quench the reaction with approximately the same volume of
MeOH. Stir for 10min at room temperature and then concentrate
under reduced pressure.
(b) If the substituent on the phthalaldehyde is not highly electron with-
drawing (in our experience, the 4-fluoro and 4-bromo) quench the
reaction with H2O approximately the same volume as the DCM
used in the reaction. Collect the organic layer and concentrate
under reduced pressure.
7. (a) Purify the crude product by flash chromatography on silica. High-
˚
purity grade Silica gel, 60A pore size is necessary for adequate prod-
uct recovery in our hands. Set up the column by first adding sand to
the glass tube then adding some of the loading solvent. At this point
carefully add the Merck grade silica as a slurry in the loading solvent
at first, once a base of silica is established the rest can be more

330
Mihajlo Todorovic and David M. Perrin
quickly. Load the product (dissolved in a minimum amount of
1:39:60 (v/v/v) MeOH/EtOAc/Hexanes) with 1:39:60 (v/v/v)
MeOH/EtOAc/Hexanes and elute isocratically. Note: Change
the eluent composition if the rf of the given substituted phenyl
dimethanol deviates significantly from this report. Evaluate the
purity of each fraction by TLC with 1:39:60 (v/v/v) MeOH/
EtOAc/Hexanes. Pool the pure fractions and concentrate the
solution under reduced pressure on rotovap to a white or colored
solid. Note: As modified phthalaldehyde methyl hemiacetals elute
as a mixture of regio and stereoisomers, multiple spots will be the
desired product and when isolated will equilibrate back to the
mixture. All regioisomers are useful for FlICk reactions and
inconsequential for the outcome of reaction.
(b) Purify the crude product by flash chromatography on silica. High-
˚
purity grade Silica gel, 60A pore size is recommended for adequate
product recovery. Set up the column by first adding sand to the
glass tube then adding some of the loading solvent. At this point care-
fully add the Merck grade silica as a slurry in the loading solvent at
first, once a base of silica is established the rest can be more quickly.
Load the product (dissolved in a minimum amount of 4:6 (v/v)
EtOAc/Hexanes) with 4:6 (v/v) EtOAc/Hexanes and elute
isocratically. Note: Change the eluent composition if the rf of the
given substituted phenyl dimethanol deviates significantly from this
report. Evaluate the purity of each fraction by TLC with 4:6 (v/v)
EtOAc/Hexanes. Pool the pure fractions and concentrate the solu-
tion under reduced pressure on rotovap to a white or colored solid.
8. Conclusions
The development of any new chemical methodology should be
driven by unmet needs reflected in the desire to advance the state of the
art in a significant way. To that end, the method described herein satisfies
the initially stated motivation: to develop a useful bio-orthogonal stapling
reaction with the anticipation that beyond any novelty, it could readily
be anticipated to provide biological activity. Indeed, in two test cases, pre-
cursors lacking any measureable biological activity were converted to
FlICked macrocycles and bicycles that showed biological activity. When
applied to a linear peptide highly decorated with potentially competitive side
chains, FlICking is selective for a primary amine and a thiol and ignores

FlICk for peptide stapling
331
other potential nucleophiles. The reaction proceeds smoothly under biotic
conditions and has shown to be compatible with proteins in the literature.
Additionally, it adds useful functionality in the chemical moiety that is
incorporated into the macrocycle in the form of nascent fluorescence.
In a further testament to the robustness of this method, isoindole peptide
stapling was independently demonstrated simultaneously to our report
(Zhang, Zhang, Wong, & Li, 2019).
In concert with its demonstrable utility, FlICk resonates with the
concept of “click” chemistry as articulated by Sharpless (Kolb, Finn, &
Sharpless, 2001), namely: a reaction that ideally employs mild reagents
at room temperature which produces benign byproducts and incorporates
desired functionality into the starting materials in an atom-economical
manner. The FlICk reaction herein described neatly fits into this schema:
it employs commercially available OPA, runs at room temperature in
water, requires no workup, produces only water as a byproduct and is com-
plete on mixing. We hope that the operational ease of conducting these
reactions, especially given the detail with which they have been described
here, will motivate other researchers to employ this method when cross-
linking peptides. Current limitations of the method include a lack of
regioselectivity in FlICk reactions with some of the substituted OPAs.
Preliminarily, we believe this work demonstrates the fertility of isoindole
chemistry in the context of peptide stapling, particularly in the context of
combinatorial peptide arrays that can be converted into fluorescent peptide
macrocycles. We anticipate great potential in exploring this chemical space
and suggest that its operational simplicity will entice additional investigation
by others.
References
Chen, R. F., Scott, C.,
& Trepman, E. (1979). Fluorescence properties of
o-phthaldialdehyde derivatives of amino acids. Biochimica et Biophysica Acta, Protein
Structure, 576(2), 440–455. https://doi.org/10.1016/0005-2795(79)90419-7.
Dukes, B. C., & Butzke, C. E. (1998). Rapid determination of primary amino acids in grape
juice using an o-phthaldialdehyde/N-acetyl-L-cysteine spectrophotometric assay.
American Journal of Enology and Viticulture, 49(2), 125–134.
Garc´ıa Alvarez-Coque, M. C., Medina Herna´ndez, M. J., Villanueva Caman˜as, R. M., &
Mongay Ferna´ndez, C. (1989). Formation and instability of o-phthalaldehyde derivatives
of amino acids. Analytical Biochemistry, 178(1), 1–7. https://doi.org/10.1016/0003-2697
(89)90346-1.
Go, K., Horikawa, Y., Garcia, R., & Villarreal, F. J. (2008). Fluorescent method for detection
of cleaved collagens using O-phthaldialdehyde (OPA). Journal of Biochemical and
Biophysical Methods, 70(6), 878–882. https://doi.org/10.1016/j.jbbm.2007.05.004.

332
Mihajlo Todorovic and David M. Perrin
Herna´ndez, M. J. M., Caman˜as, R. M. V., & Alvarez-Coque, M. C. G. (1990). Determination
of total free amino acids with o-phthalaldehyde and N-acetyl-l-cysteine. Microchemical
Journal, 42(3), 288–293. https://doi.org/10.1016/0026-265X(90)90057-C.
Kolb, H. C., Finn, M. G., &Sharpless, K. B. (2001). Clickchemistry: Diverse chemical function
from a few good reactions. Angewandte Chemie, International Edition, 40(11), 2004–2021.
[https://doi.org/10.;1002/1521-3773(20010601)40:11<2004::AID-ANIE2004>3.0.
CO;2-5].
Le Bourdonnec, B., El Kouhen, R., Lunzer, M. M., Law, P. Y., Loh, H. H., &
Portoghese, P. S. (2000). Reporter affinity labels: An o-phthalaldehyde derivative of
β-naltrexamine as a fluorogenic ligand for opioid receptors. Journal of Medicinal
Chemistry, 43(13), 2489–2492. https://doi.org/10.1021/jm000138s.
Leburg, M. W., & Jacobs, W. A. (1986). Stability of o-phthalaldehyde-derived isoindoles.
Analytical Biochemistry, 156, 334–340.
Maly, D. J., Allen, J. A., & Shokat, K. M. (2004). A mechanism-based cross-linker for the
identification of kinase-substrate pairs. Journal of the American Chemical Society, 126(30),
9160–9161. https://doi.org/10.1021/ja048659i.
Matinkhoo, K., Pryyma, A., Todorovic, M., Patrick, B. O., & Perrin, D. M. (2018).
Synthesis of the death-cap mushroom toxin α-amanitin. Journal of the American
Chemical Society, 140(21), 6513–6517. https://doi.org/10.1021/jacs.7b12698.
Puri, R. N., Bhatnagar, D., & Roskoski, R. (1988). Inactivation of yeast hexokinase by
orthophthaldehyde—Evidence for the presence of a cysteine and a lysine at or near
the active site. Biochimica et Biophysica Acta, 957, 34–36. https://doi.org/10.1016/
0167-4838(88)90154-9.
Roth, M. (1971). Fluorescence reaction for amino acids. Analytical Chemistry, 43(7),
880–882. https://doi.org/10.1021/ac60302a020.
Simons, S. S., & Johnson, D. F. (1977). Ethanethiol: A thiol conveying improved properties
to the fluorescent product of o-phthalaldehyde and thiols with amines. Analytical
Biochemistry, 82(1), 250–254. https://doi.org/10.1016/0003-2697(77)90157-9.
Simons, S. S., & Johnson, D. F. (1978). Reaction of o-phthalaldehyde and thiols with primary
amines: Fluorescence properties of 1-alkyl(and aryl)thio-2-alkylisoindoles. Analytical
Biochemistry, 90(2), 705–725. https://doi.org/10.1016/0003-2697(78)90163-X.
Stobaugh, J. F., Repta, A. J., Sternson, L. A., & Garren, K. W. (1983). Factors affecting the
stability of fluorescent isoindoles derived from reaction of o-phthalaldehyde and
hydroxyalkylthiols with primary amines. Analytical Biochemistry, 135(2), 495–504.
https://doi.org/10.1016/0003-2697(83)90718-2.
Todorovic, M., Schwab, K. D., Zeisler, J., Zhang, C., Bꢀenard, F., & Perrin, D. M. (2019).
Fluorescent isoindole crosslink (FlICk) chemistry: A rapid, user-friendly stapling reac-
tion. Angewandte Chemie, International Edition, 58(40), 14120–14124. https://doi.org/
10.1002/anie.201906514.
Zhang, C., Zhang, Z., Lin, K. S., Lau, J., Zeisler, J., Colpo, N., et al. (2018). Melanoma
imaging using 18 F-labeled α-melanocyte-stimulating hormone derivatives with posi-
tron emission tomography. Molecular Pharmaceutics, 15(6), 2116–2122. https://doi.org/
10.1021/acs.molpharmaceut.7b01113.
Zhang, Y., Zhang, Q., Wong, C. T. T., & Li, X. (2019). Chemoselective peptide cyclization
and bicyclization directly on unprotected peptides. Journal of the American Chemical
Society, 141(31), 12274–12279.