Using 19F NMR and two-level factorial design to explore thiol-fluoride substitution in hexafluorobenzene and its application in peptide stapling and cyclisation

Article abstract of DOI:10.1002/pep2.24182

Hexafluorobenzene undergoes 1,4-selective thiol-fluoride disubstitution and is an attractive disulfide crosslinking reagent for peptide cyclisation and stapling. Little attention has been directed toward understanding the scope of this reaction. Traditional reaction optimisation relies on a one-variable-at-a-time approach, which can exclude important combined effects of reaction variables. This study initially explored base and solvent effects to inform a subsequent two-level factorial design approach to understand how to control the reactivity and product selectivity in a model reaction of hexafluorobenzene. We describe new conditions that selectively afford higher order substitution products for example, 1,2,4,5-tetrasubstitution, making hexafluorobenzene a possible suitable scaffold for future branched or multicyclic peptide systems. Moreover, our new conditions provide an improved rapid (<1 minute) and selective peptide disulfide stapling and cyclisation approach under peptide-compatible conditions.

Full text of DOI:10.1002/pep2.24182

Received: 30 April 2020
Revised: 21 June 2020
Accepted: 22 June 2020
DOI: 10.1002/pep2.24182
F U L L P A P E R
Using 19F NMR and two-level factorial design to explore
thiol-fluoride substitution in hexafluorobenzene and its
application in peptide stapling and cyclisation
Paolo Dognini1 | Patrick M. Killoran2 | George S. Hanson1,3 | Lewis Halsall1
|
Talhat Chaudhry1 | Zasharatul Islam1 | Francesca Giuntini1 | Christopher R. Coxon3
1School of Pharmacy and Biomolecular
Abstract
Sciences, Byrom Street Campus, Liverpool
John Moores University, Liverpool, UK
Hexafluorobenzene undergoes 1,4-selective thiol-fluoride disubstitution and is an
2Division of Structural Biology (STRUBI),
attractive disulfide crosslinking reagent for peptide cyclisation and stapling. Little
University of Oxford, Rutherford Appleton
Laboratory, Harwell Science and Innovation
Campus, Harwell, UK
attention has been directed toward understanding the scope of this reaction. Tradi-
tional reaction optimisation relies on a one-variable-at-a-time approach, which can
exclude important combined effects of reaction variables. This study initially explored
base and solvent effects to inform a subsequent two-level factorial design approach
to understand how to control the reactivity and product selectivity in a model reac-
tion of hexafluorobenzene. We describe new conditions that selectively afford higher
order substitution products for example, 1,2,4,5-tetrasubstitution, making
hexafluorobenzene a possible suitable scaffold for future branched or multicyclic
peptide systems. Moreover, our new conditions provide an improved rapid
(<1 minute) and selective peptide disulfide stapling and cyclisation approach under
peptide-compatible conditions.
3Institute of Chemical Sciences, School of
Engineering and Physical Sciences,
Heriot-Watt University, Edinburgh, UK
Correspondence
Christopher R. Coxon, Institute of Chemical
Sciences, School of Engineering and Physical
Sciences, Heriot-Watt University, Edinburgh
EH14 4AS, UK.
Email: c.coxon@hw.ac.uk
Funding information
Engineering and Physical Sciences Research
Council, Grant/Award Number: EP/R020299/
1; H2020 Marie Skłodowska-Curie Actions,
Grant/Award Number: 801604
K E Y W O R D S
19F NMR, design of experiments, factorial design, hexafluorobenzene, peptide stapling
1
|
INTRODUCTION
fluorinated arenes is a valuable method for constructing highly
functionalised heterocyclic aromatic molecules. Fluoride displacement
Hexafluorobenzene (HFB) is the most elementary perfluoroaromatic
system, prepared by reaction of hexachlorobenzene at high tempera-
ture and pressure with potassium fluoride.[1] It has been applied to
investigate tissue oxygenation in vivo,[2] but it can also be used as a
solvent in organic reactions,[3] in 1H/13C NMR, UV and IR spectros-
copy and as a reference standard for 19F NMR. The presence of six
inductively electron withdrawing fluorine atoms (the F atom has the
highest Pauling scale electronegativity value in the periodic table),
makes HFB highly electrophilic and promotes the ring reactivity to be
dominated by nucleophilic aromatic substitution (SNAr). SNAr on poly-
takes place with a range of nucleophiles,[4] including thiols (R-SH). The
SNAr reaction on electron-withdrawing group activated rings, usually
performed in a basic medium, generally proceeds via an addition-elim-
ination mechanism with the intermediate formation of a reactive neg-
atively charged adduct between the arene and the nucleophile
(Meisenheimer complex).[5] When HFB reacts with 2 M equivalents of
a thiol nucleophile, the anion stabilization of the first thiol increases
the rate of the second substitution affording exclusively disubstitution
products with a 1,4-regiosubstitution pattern. Subsequent substitu-
tions, generally observed with higher amounts of thiols and
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2020 The Authors. Peptide Science published by Wiley Periodicals LLC.
Peptide Science. 2020;e24182.
wileyonlinelibrary.com/peptidesci
1 of 12
https://doi.org/10.1002/pep2.24182

2 of 12
DOGNINI ET AL.
increasingly vigorous reaction conditions, follow the same
regioselectivity, resulting in 1,2,4,5-tetrasubstitution and possible
1,2,3,4,5,6-hexasubstitution. Mono-, tri- and penta-substitution prod-
ucts are rarely isolated in significant quantities. In principle, the extent
of substitution can be ordered by careful control of the reagents
nucleophilicity and the reaction conditions and is summarized in Fig-
ure 1.[6–20] In general, there are few approaches reported to selec-
tively afford mono-substitution of HFB, while disubstitution is
generally obtained under mild conditions at room temperature. Stron-
ger bases, longer reaction times, higher temperatures and a greater
concentration of thiols can orient the reaction toward tetra-
substitution while hexa-substitution can be achieved with even
harsher conditions and a larger excess of thiols. Aromatic thiolates (i.
e., thiophenol or derivatives) react more readily than aliphatic thiolates
and this difference influences the reaction conditions.
HFB has broad applications as a molecular scaffold for cysteine-
containing peptide stapling and multicyclisation. In the presence of a
deprotected cysteine-containing peptide, HFB undergoes a 1,4-dis-
ubstitution with no observable higher order substitution products
(Scheme 1). Such transformations are typically performed employing
DMF as solvent and either TRIS-base, Et3N or DIPEA as base
(Table 1).[15,21–28] In the work from Zhang et al,[22] the reaction is pro-
moted by a glutathione S-transferase (GST) enzyme that, after TCEP-
mediated deprotection of the Cys (StBu), catalyzes the
macrocyclisation. However, this is only applicable when the peptide
sequence contains a γ-Glu-Cys-Gly (GSH) motif. Examples of the same
reaction with different base-solvent combinations or alternative
methodologies are not reported.
The remaining four fluorine atoms are conceptually also available
for substitution, making this an excellent scaffold for stepwise
FIGURE 1 General approaches
to selectively access mono-,
di-, tetra- and hexa- substituted
fluorobenzenes

DOGNINI ET AL.
3 of 12
SCHEME 1 General reaction scheme for a SNAr on HFB with two-cysteine-containing peptides
The fluorine atoms of fluorobenzenes provide
a
diagnostic
TABLE 1 General stapling and cyclisation approaches that
reporter for product formation by 19F NMR. 19F NMR is a fast, easily
interpretable, quantifiable and reproducible reaction monitoring and
optimisation tool. The number of substituted fluorine atoms on the
aromatic ring is a direct indicator of the substitution degree on the
molecule. Fluorine signal integration is quantitative and, therefore,
allows for accurate quantification of the relative amount of each
product. Having no background fluorine, 19F NMR allows for easier
characterization of chemicals compared to 1H NMR. In this case, only
LC-MS analysis can be as much informative as NMR analysis is, but
solvent usage and long run times add value to NMR practicality.
Exploration of the SNAr reaction between HFB and N-acetyl cys-
teine will help peptide chemists to understand what drives the level of
HFB substitution when reacted with a thiol. Here, after initial screen-
ings of different base-solvent combinations, we apply a FD approach
coupled with 19F NMR to investigate the parameters that predomi-
nantly affect the product outcomes in the reaction of HFB with N-
acetyl cysteine as a model for upcoming peptide studies. This work
also introduces three new sets of conditions to execute the same
reaction for peptide stapling and cyclisation.
exploits the reaction between di-cysteine peptides and HFB
Conditions
Peptides
Ref
TRIS-base, DMF, rt,
4.5 hours
YCGGGCAL, YCERSCNMK,
ITFCDLLCYYGKKK,
[15]
CNLLCEAKKLNDAQAPK
TRIS-base, DMF, rt, 2 hours
From i,i + 1
[21]
[22]
[IKFTNGLCCLYESKR] to i,i
+ 14 [CIKFTNGLLYESKRC]
1. TRIS-base, DMF, rt,
30 minutes
γ-EC(StBu)G(GLKAG)xC
X = 2; 3; 4
2. GST, TCEP, 0.1 M
phosphate buffer (pH 8),
rt, 5 minutes
TRIS-base, DMF, rt,
18 hours
CDEETGEC, CEETGC,
[23]
[24]
[25]
CDPETGEC, CLDPETGEFC
TRIS-base, DMF, rt
HX1EGCX2TSCX3X4,
HX1EGTX2TCDX3X4C
Et3N, DMSO/DMF (1:1), rt,
14 hours
CGNKRTRGC
DIPEA, DMF, rt, 4 hours
DIPEA, DMF, rt, 4 hours
YCGGGCAL, FKACGKGCA
LTFCHYWCQLTS
[26]
[27]
[28]
TRIS-base, TCEP, DMF, rt,
2 hours
WGKGCGKGUGKGCW
2
|
METHODS AND MATERIALS
| General points
multicyclisation or poststapling introduction of additional functional-
ity. The literature indicates that further substitution is possible using
nonpeptide thiols and typically requires either higher temperature,
longer reaction times and/or stronger inorganic bases.
2.1
All the reagents were purchased from commercial suppliers and used
directly as indicated in the appropriate experimental procedures. All
Fmoc L-amino acids with standard side-chain protecting groups (Fmoc-
Ala-OH, Fmoc-Cys(Trt)-OH, Fmoc-Gly-OH, Fmoc-His(Trt)-OH, Fmoc-
Leu-OH, Fmoc-Gln(Trt)-OH, Fmoc-Ser(tBu)-OH, Fmoc-Thr(tBu)-OH,
Fmoc-Tyr(tBu)-OH), Rink Amide Pro-Tide resin and Oxyma Pure were
purchased from CEM UK Ltd (Buckingham, UK). TFA and DIC were pur-
chased from Fluorochem UK Ltd (Hadfield, UK). DMF was purchased
from Fisher Scientific UK Ltd (Loughborough, UK). DIPEA, piperidine,
HFB, triisopropylsilane, DMSO, MeCN, cesium carbonate, potassium
carbonate, THF, DBU, propylene carbonate (PC), Et2O were purchased
from Sigma Aldrich Company Ltd (Gillingham, UK).
Factorial design (FD) is a powerful experimental design approach
that, with a small number of experiments, provides high-quality infor-
mation in the whole experimental domain. The alternative One-Vari-
able-At-a-Time (OVAT) approach gives very specific information only
for the conditions in which the experiments have been performed.
However, a FD reflects the interactions between controllable vari-
ables (factors), while the OVAT would only be valid if the variables are
totally independent from each other.[29] Therefore, FD is advanta-
geous with respect to time, money and relevancy of the results and
can be usefully applied in the chemistry field. Time, temperature, vol-
ume of solvent(s) and amount of reagent(s) are just a few variables
that play a role within, or may affect the outcome of, a standard labo-
ratory reaction. Consequently, they are factors frequently considered
during optimisation processes and their mutual interaction is highly
likely. FD allows these to be studied individually, in addition to the
interactions between them to be observed.
2.2
|
19F NMR analysis
All proton-decoupled 19F NMR spectra where recorded on
Bruker Avance 600 MHz NMR, operating at 564.7 MHz at 300 K
a

4 of 12
DOGNINI ET AL.
TABLE 2 Factorial design space - high and low settings for each
factor
in high-resolution NMR tubes with 0.7 mL of each sample. Signals
were observed as singlets for the products of interest, which were
assigned based on known 19F NMR resonance data.[15,16] Data
analysis (integration, peak-picking) was carried-out using TopSpin
3.1, Bruker UK Ltd (Coventry, UK). All spectra are provided in
Figures S5-S35).
Controllable factor
Time
Low setting
4 hours
5 mL
High setting
168 hours
10 mL
Solvent volume
Temperature
21 ꢀ
4
C
50 ꢀ
10
C
Mol. equation N-acetyl cysteine
Mol. eq. Cs2CO3
Solvent
8
20
2.3
|
Liquid chromatography-mass spectrometry
DMSO
(high resolution) analysis
Characterization of crude peptides and reaction products were con-
ducted using an Agilent 1260 Infinity II LC system with Agilent 6530
Accurate-Mass QToF spectrometer, equipped with an Agilent
ZORBAX SB-C18 Stable Bond Analytical (5 μm particle size,
4.6 × 150 mm) from Agilent Technologies UK Limited (Cheadle, UK)
with a binary eluent system comprising MeCN/H2O (25 minutes gra-
dient: 1%-99% with 0.1% formic acid) as mobile phases. MS grade sol-
vents (MeCN, formic acid) were purchased from Fisher Scientific and
the employed ultrapure H2O was purified with a Milli-Q Reference
Water Purification System. Electrospray ionization mass spectrometry
was conducted in positive ion mode (m/z range: 50-3200) using a
fragmentor voltage of 150 V, gas temperature of 325 ꢀC (flow 10 L/
min) and sheath gas temperature of 400 ꢀC (flow 11 L/min).
raw data). Each reaction was performed in a 10 mL fritted syringe con-
taining a magnetic stirrer bar. N-acetyl cysteine (113 mg, 4 eq. or
282 mg, 10 eq.), Cs2CO3 (450 mg, 8 eq. or 1129 mg 20 eq.) were
added to the reaction vessel, followed by a solution of HFB (20 μL,
1 eq.) in DMSO (5 mL or 10 mL), leaving no headspace. The reac-
tion vessel was then capped and stirred for 4 or 168 hours at the
appropriate temperature (21 ꢀC or 50 ꢀC). For experiments that
were performed at an elevated temperature of 50 ꢀC the vessel
was stirred in a water bath at 50 ꢀC, maintained by using a tem-
perature probe. At the completion of the reaction, the contents of
the syringe were filtered through the frit and 700 μL was used for
19F NMR analysis (data provided in Figures S5-S35). Statistical
analysis, including ANOVA, factorial regression and main effects
plots were generated in Minitab.
2.4
|
Model reactions - OVAT approach
Reactions were performed in 14 mL screw top vials and stirred using
stirrer bars and magnetic stirrer-plates. N-acetyl L-cysteine (113 mg,
0.692 mmol, 4 eq.), the required solvent (5 mL), the required base
(3.46 mmol, 20 eq.) and HFB (20 μL, 0.173 mmol, 1 eq.) were added
to the reaction vessel and mixtures were stirred for 4 hours at 21 ꢀC.
The reaction outcomes were analyzed with 19F NMR in order to mea-
sure the percentage of unreacted HFB and various substitution prod-
ucts. The above general procedure was applied to combinations of
solvents (THF, MeCN, DMF, DMSO and PC) and bases (DIPEA,
cesium carbonate and DBU) as indicated. 1,4-disubstitution product
(3): HRMS [M + H]+ m/z for [C16H17F4N2O6S2]+ calculated
473.03859, found 473.04691; 1,2,4,5-tetrasubstitution product (4):
HRMS [M + H]+ m/z for [C26H33F2N4O12S4]+ calculated 759.08676,
found 759.09531. See Supporting Information for complete LCMS
characterization (Figure S3).
2.6
|
Solid phase peptide synthesis
Each linear di-cysteine peptide sequence was prepared using auto-
mated Fmoc-SPPS methods on a CEM Liberty Blue microwave-
assisted peptide synthesizer. Solid-phase synthesis was conducted
using Rink amide Pro-Tide resin (180 mg, 0.56 mmol/g loading;
0.1 mmol), employing the required Fmoc amino acids (0.2 M in DMF,
5 eq.); DIC (1 M stock solution in DMF; 10 eq.) as activator, Oxyma
Pure (1 M stock solution, 5 eq.) as racemisation suppressor, and piper-
idine (20% v/v in DMF; 587 eq., 4 mL) as deprotection agent. Standard
coupling procedures employed single coupling of each amino acid
(2.5 minutes, 90 ꢀC) and Fmoc-deprotection (2 minutes, 90 ꢀC). Race-
misation-prone amino acids bearing thermally-sensitive protecting
groups, for example, Fmoc-Cys(Trt)-OH were coupled under milder
conditions (10 minutes, 50 ꢀC). Following on-resin synthesis of the
appropriate sequence, the resin was transferred in 10 mL syringes
with frits and shrank with Et2O. Finally, peptides were cleaved from
the resin as the C-terminal amide by treatment with a cleavage cock-
tail comprising TFA, TIPS and H2O (8:1:1 v/v) with shaking at 21 ꢀC
for 4 hours. Peptides were precipitated from cleavage solutions by
dropwise addition into cold Et2O followed by centrifugation. The
resulting pellet was successively suspended in cold Et2O and cen-
trifuged twice further. The solids obtained after Et2O removal and its
complete evaporation were analyzed by LCMS. For LCMS analysis,
the peptide was dissolved in H2O, MeCN or MeOH with 0.1% formic
2.5
|
Factorial design
A two-level FD space was generated in Minitab (version 19) to assess
the influence of 5 factors: thiol (mol. eq.), base (mol. eq.), reaction time
(h), temperature (ꢀC) and reaction volume (mL) (Table 2). High (maxi-
mum) and low (minimum) values for each factor chosen based on ear-
lier observations, were combined in an array of 32 individual
experiments (See Supporting Information for complete details and

DOGNINI ET AL.
5 of 12
acid v/v (or a mixture of these solvents, depending on the solubility)
and analyzed by LCMS (as above). The crude peptides were used for
successive reactions without further chromatographic purification.
Peptide (5) H-YCGGGCAL-NH2: HRMS [M + H]+ m/z for
[C30H48N9O9S2]+ calculated 742.29382, found 742.30281; SPACE
peptide (7) Acetyl-ACTGSTQHQCG-NH2: MS [M + H]+ m/z for
[C42H67N16O17S2]+ calculated 1133.43898, found 1133.09. See
Supporting Information for complete LCMS characterization—
Figures S39, S43.
Cs2CO3 reaction: HRMS [M + H]+ m/z for [C36H46F4N9O9S2]+ calcu-
lated 888.27178, found 888.27962; peptide (6) from MeCN/DBU reac-
tion: HRMS [M + H]+ m/z for [C36H46F4N9O9S2]+ calculated
888.27178, found 888.27859. See Supporting Information for complete
LCMS characterization—Figures S40-S42.
2.8
|
General procedure for peptide stapling on
resin
After SPPS, the mass of dry resin corresponding to 1 eq. of peptide
was weighed directly in an empty 10 mL fritted-syringe and swollen
for 5 to 10 minutes in DMF. Trityl protected cysteine residues were
selectively deprotected on resin with a cleavage solution comprising
DCM:TFA:TIPS (35:4:1, 2 mL/10 mg of resin), for 1 hour at 21 ꢀC.[30]
The cleavage solution was discarded, and the resin washed succes-
sively with DCM and DMF. A solution of HFB (10 eq.) and DIPEA
(0.4 M in DMF; 1 mL/10 mg of resin) was drawn up into the syringe
containing the swollen peptide resin and shaken at 21 ꢀC for
18 hours. Subsequently, the reaction mixture was gently evaporated
under a stream of nitrogen gas, the resin washed with DCM and DMF
and shrunk with Et2O. The stapled/cyclic product was finally cleaved
2.7
solution
|
General procedure for peptide stapling in
The crude peptide (1 eq.) and any other solid reagent (e.g., Cs2CO3, 20
eq. or TCEP 2 eq.) were weighed and solubilized in small vials (from 1.5
to 3 mL, 1.5 mL of solvent for 50 mg of peptide) with magnetic stirring.
Any other liquid reagent (e.g., DBU, 20 eq.) and, last, HFB (10 eq. or 0.5
eq. as required) were added to a sealed vial that was stirred at 21 ꢀC for
24 hours, with sampling at intermediate timepoints. Twenty micro liter
of reaction mixture were sampled and analyzed by NMR and LCMS
after dilution with 1 mL H2O + 0.1% TFA. Peptide (6) from DMSO/
SCHEME 2 A, Model reaction between
HFB and N-acetyl cysteine. B, Typical 19F NMR
for this model reaction, showing characteristic
quantifiable resonances for substitution
products. These chemical shifts are in
accordance with previously reported literature
assignments.[15,16]

6 of 12
DOGNINI ET AL.
TABLE 3 Summary of solvent
properties and the effect of solvent-base
combinations on product outcome
Solvent
Dielectric constant (ε)
Polarisability (α)
Viscosity at 25 ꢀC (cP)
Boiling point (ꢀC)
THF
7.6
37
38
47
64
7.94
4.44
7.93
8.03
8.55
0.48
0.33
0.8
66
MeCN
DMF
DMSO
PC
82
153
189
240
1.99
22.4
pKa of conjugate acid (in H2O) = 10.75 [35]
DIPEA
Solvent
Unreacted
Monosubstituted
Disubstituted
Tetrasubstituted
THF
100
92
41
1
0
0
0
0
0
0
0
MeCN
DMF
DMSO
PC
7.34
22
0
0.66
37
99
75
20
5
pKa of conjugate acid (in H2O) = 11.5,[36] 11.9 [37]
DBU
Solvent
Unreacted
Monosubstituted
Disubstituted
Tetrasubstituted
THF
68
0
0
32
80
85
60
97
0
MeCN
DMF
DMSO
PC
10
12
0
10
3
0
0
40
3
0
0
pKa of conjugate acid (in H2O) = 10.33 [38]
Cs2CO3
Solvent
Unreacted
Monosubstituted
Disubstituted
Tetrasubstituted
THFa
n/a
0
0
0
0
100
93
0
DMFb
7
DMSO
0
63
37
Note: Darker colors provide a visual aid to indicate higher conversion.
aIn precipitated solid only.
bIn precipitated solid but no starting material remained in solution.
and analyzed following the procedure described above. Peptide (6):
HRMS [M + H]+ m/z for [C36H46F4N9O9S2]+ calculated 888.27178,
found 888.27915; SPACE-TFB peptide (8): MS [M + H]+ m/z for
[C48H67F4N16O17S2]+ calculated 1279.41694, found 1279.14. See
Supporting Information for complete LCMS characterization—
Figure S44.
catalysis and a large excess of thiols should ideally be avoided. In
order to narrow the scope of conditions that could be applied to a FD
space, the model reaction between N-acetyl cysteine and HFB was
studied using an initial OVAT approach (Scheme 2, A). As HFB and the
two major products both contain all equivalent fluorine nuclei, each
exhibited a characteristic singlet at approximately −165, −134 and
−96 ppm for unreacted HFB, disubstitution and tetrasubstitution
products, respectively (Scheme 2B).[15,16] This allowed relative quanti-
fication of the main reaction products by signal integration and divid-
ing by the number of fluorine nuclei, Figure S2.
3
|
RESULTS AND DISCUSSION
3.1
|
Initial condition scoping—OVAT
Preliminary studies (see Figure S4) showed that disubstitution
was obtained as the predominant product using the organic base
DIPEA, even with increasing equivalents of thiol and time. However,
the inorganic base K2CO3 generally encouraged a larger proportion of
higher-order substitution products, particularly when using a slightly
wasteful 10-fold excess of thiol. At least 2 mol. Equivalents of base
(with respect to HFB) were required as 1 equivalent afforded a mix-
ture of mainly unreacted, mono-substituted and a small proportion of
disubstituted fluorobenzene, while in the absence of base no reaction
The initial aim of this study was to explore how, by applying specific
combinations of reaction conditions, it is possible to selectively obtain
higher conversion to either di- or tetra-thiol-substituted fluoro-
benzenes, under mild and peptide-friendly conditions. A wide range of
procedures have been reported to afford similar products (Figure 1),
however, many of these are not well suited to peptides and amino
acids. For example, the use of temperatures higher than 50 ꢀC, metal

DOGNINI ET AL.
7 of 12
FIGURE 2 Two-Level factorial design analysis: A, Product distribution and relative % conversion by 19F NMR. B, Pareto plot to indicate
factors that influence reaction outcome. C, Main effects plot for 1,4-disubstitution (%), and D, 1,2,4,5-tetrasubstitution (%)
peptide synthesis.[32] For each solvent, the bases DIPEA, DBU and
cesium carbonate were probed. The reactions were performed in the
required solvent (5 mL) using the required base (20 mol. eq.) and HFB
(34.6 mM, 1 mol eq.), and the mixtures were stirred for 4 hours at
21 ꢀC prior to 19F NMR analysis. Using DIPEA as a base afforded little
or no conversion to substituted products in acetonitrile and THF, while
DMF provided a near-equal distribution of unreacted (41%), mono-
substituted (22%) and disubstituted (37%) products. Disubstitution
proceeded efficiently in both DMSO and propylene carbonate, despite
the latter also presenting some unreacted starting material. The stron-
ger organic base DBU generally seemed to promote higher conversions
to substituted products (significant variation when switching from
DIPEA to DBU in MeCN), following a similar solvent-trend as for
DIPEA. However, of particular note in this case was the moderate pro-
duction (40%) of the tetrasubstituted product in DMSO. Replacement
of the organic bases with the inorganic base Cs2CO3 presented an addi-
tional challenge: the base was mostly insoluble in each solvent and
most reactions afforded an additional precipitate. Redissolution and
acidification of the precipitate confirmed by 19F NMR that the substitu-
tion products were precipitated as presumed insoluble cesium salts. In
THF the precipitate was entirely of the disubstituted product, however,
it should be noted that this cannot be accurately quantified due to the
presence of some unreacted starting material remaining in solution. In
DMF, no unreacted material remained in solution and afforded a
TABLE 4 Perfluoroaryl stapling of model peptides under new
conditions
Conditions
C-terminus
Peptide sequences
Cs2CO3, DMSO, 21 ꢀC,
5 minutes
H (free
YCGGGCAL (6)
peptide)
DBU, MeCN, 21 ꢀC,
1 hour
H (free
YCGGGCAL (6)
peptide)
DIPEA, DMF, 21 ꢀC,
18 hours
Resin
YCGGGCAL (6),
ACTGSTQHQCG (SPACE
peptide)
took place. Gentle heating (40 ꢀC) afforded lower conversion when
compared to the same reaction at RT, likely due to the relative volatil-
ity of HFB.
We have previously reported that careful selection of solvents
can increase chemoselectivity in thiol-fluoride substitutions in reac-
tions of peptide amines/thiols with pentafluoropyridine.[26,31] Subse-
quently, additional bases were studied in a range of polar aprotic
solvents to try to gauge the importance of properties such as dielec-
tric constant and polarizability (Table 3) upon promoting substitution
at the fluorinated aromatic ring. Several solvents were compared,
including DMSO, DMF, MeCN, THF and propylene carbonate, which
has recently shown promise as a “green” alternative to DMF for

8 of 12
DOGNINI ET AL.
SCHEME 3 Perfluoroaryl stapling of model peptides
precipitate as a mixture of disubstituted (major) and tetrasubstituted
(minor) products. When DMSO was used as the solvent, all products
remained in solution and afforded a similar product distribution as was
the case for DBU in DMSO, with around 40% tetrasubstituted product.
In summary, there appears to be an overall trend for increasing solvent
dielectric and polarisability to afford higher conversion to disubstitution
and in some cases (DMSO as solvent) the tetrasubstituted product.
Therefore, the higher polarity solvent may be a better insulator of the
charges of the thiolate nucleophile, and the Meisenheimer complex
formed in the reaction.[33] The outlier to this trend is propylene carbon-
ate which has the highest polarisability and dielectric properties of the
solvents investigated, yet seems to be less efficient than DMSO in pro-
moting substitution. This may be due to its higher viscosity affecting
efficient mixing or perhaps another property not considered here. In
propylene carbonate, the fluoride leaving group, which is normally a
very poor nucleophile in solution,[34] may be now strong enough to per-
form the reverse reaction or at least compete with the thiolate. In gen-
eral, we observed that selective high conversion to disubstitution or
tetrasubstitution could be controlled by the choice of solvent-base
combinations that is, 1,4-disubstitution can be obtained cleanly using
DMSO/DIPEA or THF/Cs2CO3 or propylene carbonate/DBU. Interest-
ingly, DMF (used in perfluoroaryl peptide stapling, Table 1) does not
appear to be optimal for this transformation, especially if in combina-
tion with DIPEA; and MeCN/DBU could be a valuable alternative.
However, while 1,2,4,5-tetrasubstitution could be obtained using
DMSO-DBU/Cs2CO3, this is only afforded as a moderate product
(approx. 40%) and further optimisation was required.
to control di- and tetra- fluoride-thiol substitution. This employed a
two-level, five factorial full design space (Table 2) varying reaction
time, temperature, solvent volume, molar equivalents of thiol and
molar equivalents of base. A DMSO/Cs2CO3 system was selected as
the combination of a polar aprotic solvent and a strong base (e.g.,
Cs2CO3, K2CO3 or DBU) increased the formation of substitution
products; and the selection of a highly polar solvent would enable all
reagents and products to remain in solution, which would expedite
reaction profiling through 19F NMR.
Specific combinations of each factor were combined, such that
the relationship between each factor and the formation of specific
substitution products (3 or 4) could be established with relatively
few individual experiments and using simple statistical analyses. A
total of 32 individual experiments were performed in disposable
fritted syringes to minimize vessel headspace to avoid possible
evaporation of HFB. After the completion of the reaction time (4 or
168 hours), samples were filtered, removing cesium salts and
unreacted cesium carbonate and the reaction outcomes were mea-
sured using 19F NMR (Figure 2A). In each case, only the major
products or starting material were integrated, yet in some cases,
some other minor 19F resonances were observed (mostly mono-sub-
stitution but some others unidentified were generally observed with
higher temperature and longer reaction time for example, reactions
9 and 31—see Supporting Information for raw data (Table S1) and
19F NMR spectra (Figures S5-S35).
Analysis of variance (ANOVA, Figure 2B) indicated that the major
factors that influence the product outcome are time, base and combi-
nations of base-time and thiol-temperature; and with P-values <.05,
these are unlikely to be random observations. Main effects analysis
(Figure 2C,D) also indicated that both the amount of base and the
time allotted for the reaction exert significant influence on the substi-
tution outcome, as indicated by the steep gradient. It was clear that a
greater amount of base generally afforded a higher proportion of tet-
rasubstitution over disubstitution, while longer reaction times also
tended to afford higher conversion to higher substituted products.
Unsurprisingly, the exact opposite was true for each factor in
3.1.1
|
Two-level FD—screening experiment
The above analyses indicated that the reaction outcome may be sensi-
tive to thiol concentration and the nature of the base and solvent. A
more detailed FD screening experiment was used to further explore
the combined roles of reaction time, temperature, concentration and
reagent molar equivalence on product outcome and that may be used

DOGNINI ET AL.
9 of 12
FIGURE 3 UV chromatograms (response at λ = 280 nm vs acquisition time in minutes) from LCMS analysis of: A, the model peptide (5) and
reactions B-D sampled after 1, 15, 30, 45 minutes, 1, 2, 4, 8 and 24 hours under different conditions. The red boxes highlight the first timepoints
for each condition where complete conversion of the linear peptide (5) into the stapled peptide (6) was observed
promoting 1,4-disubstitution and affords a means of discrimination
between a disubstituted (stapled/cyclic) peptide and higher-substitu-
tion. Perhaps surprisingly, on average, the molar equivalents of thiol,
reaction temperature and reaction concentration had comparatively
lower influences over product selectivity and these variables alone
were not a strong indicator of reaction outcome. In fact, there are
examples of both high and low values for thiol, temperature and
volume in both disubstitution favoring and tetrasubstitution favor-
ing reactions. This is not to suggest that such factors do not influ-
ence the reaction, but likely indicates a more complex interaction
with another factor (Figures S36-S38) that together have a more
profound influence, which underlines the value of using a multifac-
torial analysis approach.
3.2
|
Perfluoroaryl peptide-stapling
These outcomes of the base-solvent combination screenings and FD
studies were applied to the exploration of cysteine-containing peptide
stapling around HFB. Despite this transformation being typically

10 of 12
DOGNINI ET AL.
FIGURE 4 UV chromatogram
(λ = 280 nm) and mass spectrum (ESI) of
the crude cleavage product following on-
resin perfluoroaryl-stapling reaction with
DMF/DIPEA
performed using DMF (Table 1), our exploration indicated that MeCN
and DMSO may provide the best combination of solubility (peptide
and base) and potential for clean and rapid reaction (Table 4).
and this is possibly a proximity effect (second substitution step is
intramolecular).
In each case (Figure 3B,D), the presumed degradation products
could not be ascribed to any identifiable products and no higher order
substitution (i.e., tetrasubstitution - “double stapling”) products were
observed after prolonged reaction times. We hypothesized that a
larger amount of peptide might favor the formation of such multicyclic
systems. Therefore, the fast and clean reaction in DMSO/Cs2CO3 was
repeated with a lower amount of HFB (around 0.5 eq., Figure 3C) and
TCEP. We could verify that the combination of these factors afforded
the stapled product rapidly and cleanly as before, but in this case, 6
remained detectable/stable for at least 2 hours and no tetra-substitu-
tion products were observed. The reaction was complete even with a
stoichiometric or sub-stoichiometric amount of HFB and TCEP may
play and important role in maintaining the stability of the product.
There are many examples of single-component stapling/
cyclisation (e.g., alkene metathesis, lactamisation) carried out on solid-
supported peptides, whereas similar two-component reactions are
generally performed in solution because of potential site-isolation and
by-products formation on resin.[39] We also wanted to determine
whether the perfluoroaryl-stapling of a di-cysteine peptide was
As such, the combinations MeCN/DBU and DMSO/Cs2CO3 were
applied to a model di-cysteine peptide system to explore their applica-
tion in peptide stapling and cyclisation (Scheme 3). Due to the
increased complexity of the peptide reaction compared to the model
reaction, and because 19F NMR peak intensities are affected by the
high amount of residual HFB (in the reaction mixture) and TFA (from
peptide cleavage), LCMS analysis was used as the main tool for reac-
tion monitoring and product characterization. By sampling the reac-
tion mixtures at different time points, we produced a kinetic profile of
the product formation (Figure 3B-D). In the reaction performed in
DMSO and using Cs2CO3, the complete conversion of the linear pep-
tide (5) into the stapled form (6) occurred almost instantaneous
(<1 minute, Figure 3 B). Nevertheless, it appeared that little of 6
remained after 15 minutes under these conditions. Performing the
same reaction in MeCN and using DBU as base gave a slower but
complete conversion of the starting peptide into a cleaner stapled
product (within approx. 1 hour, Figure 3D). These reactions appear to
happen faster than those of the model system with N-acetyl cysteine

DOGNINI ET AL.
11 of 12
possible on resin. Solid phase reactions offer undeniable advantages
such as user-friendly handling, no laborious workup, easier purification
and possible automation, resulting in a greener, faster, cheaper and
possibly higher yielding process. Unfortunately, our optimal stapling
conditions were deemed unsuitable for polystyrene resins, therefore,
we investigated the efficiency of the standard solution-phase pro-
cedure (DMF, DIPEA) for stapling a 2-Cys peptide on-resin.[26,27]
After selective on-resin trityl deprotection of cysteine residues, the
stapled product (6) was successfully obtained after overnight shak-
ing of the resin with a reaction mixture made of HFB and DIPEA
in DMF (Figure 4).
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ORCID
Christopher R. Coxon
https://orcid.org/0000-0002-3375-3901
REFERENCES
[1] N. N. Vorozhtsov, V. E. Platonov, G. G. Yakobson, Bull. Acad. Sci.
1963, 12, 1389. https://doi.org/10.1007/BF00847820.
[2] R. P. Mason, W. Rodbumrung, P. P. Antich, NMR Biomed. 1996, 9(3),
125.
https://doi.org/10.1002/(SICI)1099-1492(199605)9:3<125::
AID-NBM405>3.0.CO;2-F.
Finally, to demonstrate that the HFB-mediated cyclisation on
solid-phase could be applied to a longer and more complex pep-
tide sequence, we successfully cyclised the Skin Penetrating and
Cell Entering (SPACE) disulfide peptide[40,41] (peptide 7) on-resin
using DMF/DIPEA. The cyclised product (peptide 8) was
[3] A. Lattanzi, C. De Fusco, A. Russo, A. Poater, L. Cavallo, Chem.
Commun. 2012, 48(11), 1650. https://doi.org/10.1039/c2cc17488j.
[4] L. A. Wall, W. J. Pummer, J. E. Fearn, J. M. Antonucci, J. Res. Natl. Bur.
Stand. Sect. A Phys. Chem. 1963, 67A(5), 481. https://doi.org/10.
6028/jres.067a.050.
[5] A. J. J. Lennox, Angew. Chem. Int. Ed. 2018, 57(45), 14686. https://
doi.org/10.1002/anie.201809606.
obtained with clean conversion and only
a small amount of
[6] S. E. Denmark, A. J. Cresswell, J. Org. Chem. 2013, 78(24), 12593.
https://doi.org/10.1021/jo402246h.
starting material remaining after 18 hours reaction time and fol-
lowing cleavage from the resin (Scheme S1 and Figures S43,
S44). In addition to the procedures reported in Table 1, this is
further evidence that this peptide stapling technique is applicable
to changes in the peptide sequence components, length or inter-
thiol spacing.
[7] I. P. Beletskaya, G. A. Artamkina, A. Y. Mil'Chenko, P. K. Sazonov,
M. M. Shtern, J. Phys. Org. Chem. 1996, 9(6), 319. https://doi.org/10.
1002/(SICI)1099-1395(199606)9:6<319::AID-POC786>3.0.CO;2-7.
[8] H. Wu, W. Chi, G. Baryshnikov, B. Wu, Y. Gong, D. Zheng, X. Li, Y.
Zhao, X. Liu, H. Ågren, L. Zhu, Angew. Chem. Int. Ed. 2019, 58(13),
4328. https://doi.org/10.1002/anie.201900703.
[9] L. Testaferri, M. Tingoli, M. Tiecco, J. Org. Chem. 1980, 45(22), 4376.
https://doi.org/10.1021/jo01310a022.
[10] B. Gierczyk, G. Schroeder, Mendeleev Commun. 2009, 19(2), 75.
https://doi.org/10.1016/j.mencom.2009.03.007.
4
|
CONCLUSIONS
[11] M. Villa, M. Roy, G. Bergamini, M. Gingras, P. Ceroni, Dalton Trans.
2019, 48(12), 3815. https://doi.org/10.1039/c9dt00251k.
[12] J. N. Lowe, D. A. Fulton, S. H. Chiu, A. M. Elizarov, S. J. Cantrill, S. J.
Rowan, J. F. Stoddart, J. Org. Chem. 2004, 69(13), 4390. https://doi.
org/10.1021/jo030283o.
In conclusion, the application of a two-level FD study in combina-
tion with 19F NMR, has provided a more detailed understanding of
the reactivity of HFB toward N-acetylcysteine and how this is par-
ticularly sensitive to base and solvent effects. This work also pro-
vided new conditions that afford the selective preparation of either
1,4-di or 1,2,4,5-tetra thiol-fluoride substitution products, principally
controlled by the above factors. It is envisaged that these
approaches can be exploited in future for the synthesis of
branched or multicyclic peptide systems and poststapling modifica-
tions. Finally, new conditions (DMSO/Cs2CO3 and MeCN/DBU)
that permit rapid (<1 minute and < 1 hour, respectively), clean and
selective peptide stapling under peptide-compatible conditions were
introduced. The products obtained using our procedure (without
purification) are of equal if not better crude purity than previously
reported peptide stapling approaches.[15,26,27] We also demonstrate
that the 2-component on-resin perfluoroaryl stapling is achievable
with high crude conversion using conditions that were previously
only used in solution.
[13] J. Ponce González, M. Edgar, M. R. J. Elsegood, G. W. Weaver, Org.
Biomol. Chem. 2011, 9(7), 2294. https://doi.org/10.1039/
c0ob00790k.
[14] O. Jacobson, X. Yan, Y. Ma, G. Niu, D. O. Kiesewetter, X. Chen, Bio-
conjug. Chem. 2015, 26(10), 2016. https://doi.org/10.1021/acs.
bioconjchem.5b00278.
[15] A. M. Spokoyny, Y. Zou, J. J. Ling, H. Yu, Y. S. Lin, B. L. Pentelute, J.
Am. Chem. Soc. 2013, 135(16), 5946. https://doi.org/10.1021/
ja400119t.
[16] S. Niembro, A. Vallribera, M. Moreno-Mañas, New J. Chem. 2008, 32
(1), 94. https://doi.org/10.1039/b707776a.
[17] R. D. Kennedy, C. W. MacHan, C. M. McGuirk, M. S. Rosen, C. L.
Stern, A. A. Sarjeant, C. A. Mirkin, Inorg. Chem. 2013, 52(10), 5876.
https://doi.org/10.1021/ic302855f.
[18] T. Umemoto, L. M. Garrick, N. Saito, Beilstein J. Org. Chem. 2012, 8(1),
461. https://doi.org/10.3762/bjoc.8.53.
[19] M. Arisawa, T. Suzuki, T. Ishikawa, M. Yamaguchi, J. Am. Chem. Soc.
2008, 130(37), 12214. https://doi.org/10.1021/ja8049996.
[20] M. Mayor, J.-M. Lehn, Helv. Chim. Acta 1997, 80(8), 2277. https://
doi.org/10.1002/hlca.19970800802.
ACKNOWLEDGMENT
[21] Y. Zou, A. M. Spokoyny, C. Zhang, M. D. Simon, H. Yu, Y. S. Lin, B. L.
Pentelute, Org. Biomol. Chem. 2014, 12(4), 566. https://doi.org/10.
1039/c3ob42168f.
This work was supported by an EPSRC First Grant (EP/R020299/1)
that employed PK and a University Alliance Doctoral Training Alli-
ance COFUND/Marie Skłodowska-Curie (European Union's Horizon
2020 research and innovation programme, grant agreement No
801604) PhD Fellowship Programme in Applied Biosciences for
Health (PD).
[22] C. Zhang, P. Dai, A. M. Spokoyny, B. L. Pentelute, Org. Lett. 2014, 16
(14), 3652. https://doi.org/10.1021/ol501609y.
[23] R. J. Steel, M. A. O'Connell, M. Searcey, Bioorg. Med. Chem. Lett.
2018, 28(16), 2728. https://doi.org/10.1016/j.bmcl.2018.03.003.

12 of 12
DOGNINI ET AL.
[24] J. E. Swedberg, C. I. Schroeder, J. M. Mitchell, T. Durek, D. P. Fairlie,
D. J. Edmonds, D. A. Griffith, R. B. Ruggeri, D. R. Derksen, P. M.
Loria, S. Liras, D. A. Price, D. J. Craik, Eur. J. Med. Chem. 2015, 103,
175. https://doi.org/10.1016/j.ejmech.2015.08.046.
[35] D. Perrin, Dissociation Constants of Organic Bases in Aqueous Solution,
Butterworths, London 1965.
[36] F. Ravalico, S. L. James, J. S. Vyle, Green Chem. 2011, 13(7), 1778.
https://doi.org/10.1039/c1gc15131b.
[25] A. C. Conibear, S. Chaousis, T. Durek, K. Johan Rosengren, D. J.
Craik, C. I. Schroeder, Biopolymers 2016, 106(1), 89. https://doi.org/
10.1002/bip.22767.
[37] R. Srivastava, J. Mol. Catal. A Chem. 2007, 264(1–2), 146. https://doi.
org/10.1016/j.molcata.2006.09.012.
[38] D. Perrin, Dissociation Constants of Inorganic Acids and Bases in Aque-
ous Solution, Butterworths, London 1969.
[26] D. Gimenez, C. A. Mooney, A. Dose, G. Sandford, C. R. Coxon, S. L.
Cobb, Org. Biomol. Chem. 2017, 15(19), 4086. https://doi.org/10.
1039/c7ob00283a.
[39] Y. H. Lau, P. De Andrade, Y. Wu, D. R. Spring, Chem. Soc. Rev. 2015,
44(1), 91. https://doi.org/10.1039/c4cs00246f.
[27] S. J. M. Verhoork, C. E. Jennings, N. Rozatian, J. Reeks, J. Meng, E. K.
Corlett, F. Bunglawala, M. E. M. Noble, A. G. Leach, C. R. Coxon,
Chem. - A. Eur. J. 2019, 25(1), 177. https://doi.org/10.1002/chem.
201804163.
[40] T. Hsu, S. Mitragotri, Proc. Natl. Acad. Sci. U. S. A. 2011, 108(38),
15816. https://doi.org/10.1073/pnas.1016152108.
[41] M. Chen, M. Zakrewsky, V. Gupta, A. C. Anselmo, D. H. Slee, J. A.
Muraski, S. Mitragotri, J. Control. Release 2014, 179(1), 33. https://
doi.org/10.1016/j.jconrel.2014.01.006.
[28] Y. Yin, Q. Fei, W. Liu, Z. Li, H. Suga, C. Wu, Angew. Chem. Int. Ed.
2019, 58(15), 4880. https://doi.org/10.1002/anie.201813827.
[29] R. Leardi, Anal. Chim. Acta 2009, 652(1–2), 161. https://doi.org/10.
1016/j.aca.2009.06.015.
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Supporting Information section at the end of this article.
[30] E. Marcucci, N. Bayó-Puxan, J. Tulla-Puche, J. Spengler, F. Albericio,
J. Comb. Chem. 2008, 10(1), 69. https://doi.org/10.1021/cc7001588.
[31] D. Gimenez, A. Dose, N. L. Robson, G. Sandford, S. L. Cobb, C. R.
Coxon, Org. Biomol. Chem. 2017, 15(19), 4081. https://doi.org/10.
1039/c7ob00295e.
How to cite this article: Dognini P, Killoran PM, Hanson GS,
et al. Using 19F NMR and two-level factorial design to explore
thiol-fluoride substitution in hexafluorobenzene and its
application in peptide stapling and cyclisation. Peptide Science.
2020;e24182. https://doi.org/10.1002/pep2.24182
[32] S. B. Lawrenson, R. Arav, M. North, Green Chem. 2017, 19, 1685.
https://doi.org/10.1039/c7gc00247e.
[33] M. Gazitúa, R. A. Tapia, R. Contreras, P. R. Campodónico, New J.
Chem. 2018, 42(1), 260. https://doi.org/10.1039/c7nj03212a.
[34] V. M. Vlasov, J. Fluor. Chem. 1993, 61, 193. https://doi.org/10.1016/
S0022-1139(00)80104-9.