Aza-amino acids disrupt β-sheet secondary structures

Article abstract of DOI:10.3390/molecules24101919

Cα to N substitution in aza-amino acids imposes local conformational constraints, changes in hydrogen bonding properties, and leads to adaptive chirality at the nitrogen atom. These properties can be exploited in mimicry and stabilization of peptide secondary structures and self-assembly. Here, the effect of a single aza-amino acid incorporation located in the upper β-strand at a hydrogen-bonded (HB) site of a β-hairpin model peptide (H-Arg-Tyr-Val-Glu-Val-d-Pro-Gly-Orn-Lys-Ile-Leu-Gln-NH₂) is reported. Specifically, analogs in which valine³ was substituted for aza-valine³ or aza-glycine³ were synthesized, and their β-hairpin stabilities were examined using Nuclear Magnetic Resonance (NMR) spectroscopy. The azapeptide analogs were found to destabilize β-hairpin formation compared to the parent peptide. The aza-valine³ residue was more disruptive of β-hairpin geometry than its aza-glycine³ counterpart.

Full text of DOI:10.3390/molecules24101919

molecules
Article
Aza-Amino Acids Disrupt β-Sheet
Secondary Structures
†
†
Michael A. McMechen , Evan L. Willis
, Preston C. Gourville and Caroline Proulx *
Department of Chemistry, North Carolina State University, Raleigh, NC 27695-8204, USA;
mamcmech@ncsu.edu (M.A.M.); elwilli5@ncsu.edu (E.L.W.); pcgourvi@ncsu.edu (P.C.G.)
*
Correspondence: cproulx@ncsu.edu; Tel.: +1-919-515-9534
†
These authors contributed equally to this work.
ꢀ
ꢁꢂꢀꢃꢄꢅꢆꢇ
Academic Editors: Henry Mosberg, Tomi Sawyer and Carrie Haskell-Luevano
Received: 1 May 2019; Accepted: 17 May 2019; Published: 18 May 2019
ꢀ
ꢁꢂꢃꢄꢅꢆ
Abstract:
Cα to N substitution in aza-amino acids imposes local conformational constraints, changes
in hydrogen bonding properties, and leads to adaptive chirality at the nitrogen atom. These properties
can be exploited in mimicry and stabilization of peptide secondary structures and self-assembly. Here,
the effect of a single aza-amino acid incorporation located in the upper
HB) site of a -hairpin model peptide (H-Arg-Tyr-Val-Glu-Val-d-Pro-Gly-Orn-Lys-Ile-Leu-Gln-NH )
is reported. Specifically, analogs in which valine was substituted for aza-valine or aza-glycine were
synthesized, and their -hairpin stabilities were examined using Nuclear Magnetic Resonance (NMR)
β-strand at a hydrogen-bonded
(
β
2
3
3
3
β
spectroscopy. The azapeptide analogs were found to destabilize
β
-hairpin formation compared to
3
the parent peptide. The aza-valine residue was more disruptive of
β-hairpin geometry than its
3
aza-glycine counterpart.
Keywords: peptidomimetics; azapeptides; aza-amino acids; β-hairpin; β-sheet
1
. Introduction
β
-sheets are common protein secondary structures that are often involved in protein-protein
interactions and protein aggregation [ ]. To understand factors governing folding, stability,
and molecular recognition events involving -sheets, water soluble -hairpin model systems that
are partially folded have been developed, consisting of two antiparallel -strands connected by a
reverse turn unit [ ]. In the pursuit of protein-protein interaction (PPI) inhibitors, -strand and -sheet
peptidomimetics have been explored using a variety of unnatural scaffolds [ ]. When mimicking
-strands in particular, the hydrogen bonding pattern of a natural peptide is ideally replicated by the
artificial templates to maintain cross-strand interactions, yet surface exposed hydrogen bonding sites
1,2
β
β
β
3
β
β
4
β
should be minimized to prevent uncontrolled aggregation, creating a so-called “blocking strand” [
5
–
7
].
3
Here, we describe the effect of a single C
α
substitution for a nitrogen at the Val position of a model
β
-hairpin peptide. The aza-amino acid residue was envisioned to provide two distinct faces with
divergent hydrogen bonding capabilities, while retaining side chain position and chemistry (Figure 1).
Aza-amino acids [ ] have been used to promote -turn secondary structures and hyperstable collagen
peptide mimics [ 10] via a combination of hydrogen bonding and backbone dihedral angle modulation.
Their effect on -hairpin stability has yet to be quantified within -strand regions. In aza-amino acids,
perpendicular lone pair orientation between adjacent nitrogen and urea planarity typically reinforce
backbone dihedral angles that fall within the range of -turns and polyproline type II secondary
structures [ ]. Although this may prevent adoption of extended
8
β
9,
β
β
β
8
β
-strand geometry [11], computational
◦
studies have set the energy barrier for rotation from the twisted conformer (
ϕ
= 90 ) to the extended
◦
conformer (
ϕ
= 180 ) to be < 1 kcal/mol within 1,2-diformylhydrazine in the Z,Z configuration as an
◦
azaGly model [12]. Two favorable intramolecular hydrogen bonds are proposed to stabilize the 180
Molecules 2019, 24, 1919; doi:10.3390/molecules24101919
www.mdpi.com/journal/molecules

Molecules 2019, 24, 1919
2 of 12
orientation. Although addition of side chain chemistry in aza-amino acids removes an intramolecular
hydrogen bond with a neighboring residue, cross-strand hydrogen bonding and side chain; side chain
interactions in a
Val for aza-valine was thus examined in a well-studied
Magnetic Resonance (NMR)-based quantification of
β
3
-hairpin may provide added stability to the extended conformer. Substitution of
3
β
-hairpin model system for which Nuclear
-hairpin folding have been established, first
,14]: H-Arg-Tyr-Val-Glu-Val-d-Pro-Gly-Orn-Lys-Ile-Leu-Gln-NH (1a).
β
reported by Gellman et al. [13
2
3
Valine is situated in the middle of the upper
β
-strand at a hydrogen bonded (HB) site (Figure 2, in blue),
3
equally removed from the d-Pro-Gly turn region and the N-terminus. In addition to azaVal analog 1b,
3
3
d-Val control 1c, azaGly analog 1d, and previously described [14] positive control cyclic peptide 1e
,
6
7
and negative control l-Pro -Gly peptide 1f, all were synthesized and studied (Figure 2). Since the first
incorporation of azaVal to give an azapeptide angiotensin analog [15], considerable interest has been
generated in the synthesis and conformational properties of this aza-amino acid.
Figure 1. Aza-amino acid (in blue) creates one face with increased hydrogen bonding properties and
one with reduced hydrogen bonding capacity.
Figure 2. β-hairpin model peptide 1a [8], derivatives 1b–d, and controls 1e–f [8] under study.

Molecules 2019, 24, 1919
3 of 12
2
. Results and Discussion
2
.1. Synthesis
Solid-phase peptide synthesis was performed on Rink amide resin, using standard Fmoc
procedures for all l- and d-amino acid coupling and deprotection cycles [16]. Incorporation of
3
azaVal for the synthesis of 1b was accomplished via activation of N’-isopropyl-fluorenylmethyl
carbazate [17] using bis-(trichloromethyl)carbonate (BTC) as the activating agent (Scheme 1) and as
a safer alternative to phosgene. Although submonomer azapeptide synthesis protocols [18] were
considered for the installation of aza-valine to circumvent the solution-phase synthesis and activation
of hydrazine building blocks, N-alkylation of resin-bound semicarbazone intermediates with sterically
hindered secondary alkyl halides has typically resulted in lower conversions [19]. In our hands,
incorporation of aza-valine via N’-alkyl fluoren-9-ylmethyl carbazate activation with triphosgene
afforded azapeptide
reaction. In contrast, in accordance with previous literature, coupling to the aza-amino acid to give
was challenging due to the decreased nucleophilicity of semicarbazides [ ] and steric hindrance
6 as the major product in 77% crude purity after a single overnight coupling
9
8
from the branched aza-valine side chain. In the past, coupling to aza-valine and other aza-amino acids
with branched side chains has been accomplished using N,N’-dicyclohexylcarbodiimide (DCC) [15],
t
diisoproylcarbodiimide (DIC) [19
,
20], and BTC [21]. Here, the coupling of Fmoc-Tyr( Bu) to resin-bound
azaV-EVpGOKILQ using BTC and 2,4,6-collidine in tetrahydrofuran (THF) provided 9 in only 33%
conversion by liquid chromatography-mass spectrometry (LC-MS) analysis, with 21% uncoupled
semicarbazide and 28% isocyanate or hydantoin byproduct [22] (Table 1, entry 1). In the context of
β-hairpin synthesis, the already difficult coupling to aza-Val may be rendered even more challenging
by folding or aggregation on-resin [23]. Attempts employing a LiCl salt additive to the BTC coupling
solution to disrupt possible on-resin aggregation [24] were less successful than efforts to optimize the
coupling reagent (Table 1, entries 2–5).
Scheme 1. Synthesis of azapeptide 1b.
Recent successes using benzotriazol-1-yloxytris(pyrrolidinophosphonium-hexafluorophosphate
(
PyBOP)/N,N-diisopropylethylamine (DIEA) [25] and (1-cyano-2-ethoxy-2-oxoethylidenaminooxy)
dimethylaminomorpholino-carbenium hexafluorophosphate (COMU)/DIEA [26 27] in difficult couplings
to aza-amino acids led us to try these activating agents; however, was obtained in low conversions in both
,
9

Molecules 2019, 24, 1919
4 of 12
0
0
cases (6–13%, Table 1, entries 2 and 3). We next investigated N,N,N ,N -tetramethylchloroformamidinium
hexafluorophosphate (TCFH)/N-methylimidazole (NMI) (Table 1, entry 4), which was recently
developed [28] to couple sterically hindered carboxylic acids with non-nucleophilic aniline derivatives
and sterically hindered amino acids, including valine. Gratifyingly, these conditions provided 9 in high
conversion (89%) with little byproduct formation and no detectable starting material (Table 1, entry 4),
although appearance of two peaks with identical masses by LC-MS suggested possible epimerization.
Alternatively, the use of 1-chloro-N,N,2-trimethyl-1-propenylamine (Ghosez’s reagent) [29] (Table 1,
entry 5) provided the desired elongated azapeptide
9 as a single peak by LC-MS analysis in 67%
conversion, with no uncoupled semicarbazide detected. Using this procedure for coupling onto azaVal,
azapeptide 1b was obtained in 53% crude purity, following deprotection, elongation, and cleavage
from the resin. Purification of the crude azapeptide by reverse phase – high performance liquid
chromatography (RP-HPLC) afforded 1b in 8% yield (see Supporting Information).
Table 1. Optimization of coupling to resin-bound aza-VEVpGOKILQ.
%
Isocyanate/
Coupling
Reagent
Entry
Equiv.
Base
Equiv.
Solvent
% Coupled
% Uncoupled
Hydantoin
Byproduct [22]
1
2
3
4
5
BTC
1
3
5
3
4.2
2,4,6-collidine
DIEA
DIEA
NMI
NaHCO3
10
6
10
9
THF
DMF
DMF
33
6
21
35
17
0
28
41
53
10
22
PyBOP
COMU
TCFHa
Ghosez
13
a
CH Cl2
89
2
30
CH Cl2
67
0
2
a
Possible epimerization was detected by LC-MS analysis.
Activation of N-Fmoc-hydrazine with phosgene equivalents was previously shown to lead to the
formation of a cyclic oxadiazolone [30]. Although ring-opening of this heterocycle to afford the desired
Fmoc-protected azaGly-terminated peptides may be possible [30], activation and coupling of aza-glycine
using these conditions has been reported to be problematic in certain cases [31]. Here, aza-glycine
0
installation was accomplished by activating fluoren-9-methyl carbazate with N,N -disuccinimidyl
carbonate (DSC) as the activating agent instead of BTC, which afforded the desired azapeptide in higher
t
yields [32]. Fmoc-Tyr( Bu)-OH activation and coupling to the resin-bound aza-glycine-terminated
peptide was accomplished using BTC/2,4,6-collidine in THF and required no further optimization.
2
.2. Conformational Analysis
-Hairpin folding was first assessed by measuring the extent of diastereotopic Gly H
in the turn region for each analog and comparing it to the fully folded cyclic control (Equation (1)),
as previously described [33]. Using this method, the relative stability of each -hairpin analog can be
quickly compared as the glycine signals lie in a distinct region of the spectra (Figure 3, Table 2).
7
β
α splitting
β
Fraction folded = (∆δGly Obs)/(∆δGly 100
)
(1)
3
Replacing valine for an aza-valine residue in analog 1b caused a significant decrease in
β
-hairpin
stability, reflected by the much smaller Gly7
hydrogen bonding and side chain - side chain interactions between azaVal and Ile appear to be
insufficient to stabilize the extended conformation in aza-valine , which likely adopts a twisted
δHα–δHα’ value (Figure 3). Potential cross-strand
3
10
3
◦
conformation (
ϕ
= 90 ) that disrupts
β-hairpin formation. Because aza-amino acids may exhibit
adaptive chirality at N
which could in turn affect
α
, the aza-valine residue could be achiral or exhibit either l or d-like chirality [34],
3
β
-hairpin stability. We synthesized d-Val analog 1c as a control and found its
7
3
Gly
δHα
–
δHα
’ value to be larger than that of aza-Val analog 1b, yet smaller than the parent peptide
3
1
a
. While changing the stereochemistry of the valine residue from l to d decreased
β-hairpin stability
overall, this substitution was found to be less disruptive than the backbone C
α
to N substitution
α

Molecules 2019, 24, 1919
5 of 12
in the aza-valine residue. As such, the destabilizing effect of the aza-valine residue cannot be solely
attributed to the loss of a defined stereocenter.
7
Figure 3. Diastereotopic splitting of Gly Hα chemical shifts.
Given that the extended conformation in aza-glycine vs. aza-valine residues may be more
favorable due to additional intramolecular hydrogen-bonding interactions (vide supra), the synthesis
3
and conformational analysis of azaGly analog 1d was also pursued. Despite the absence of an
isopropyl side chain, the aza-glycine substitution was found to be less detrimental to
β
-hairpin stability
3
compared to the aza-Val analog 1b; however,
β
-hairpin folding was still decreased relative to the
3
parent peptide 1a. The disruptive effects of the azaGly substitution are similar those observed in the
3
7
d-Val analog using Gly diastereotopic Hα splitting measurements (Figure 3) [35].
It should be noted that glycine splitting measurements have been found to overestimate
folding in sequences that use d-Pro-Gly turn units [36]. Downfield chemical shifts in CH
β-hairpin
α
protons
relative to unfolded peptide for residues away from the turn region, especially those situated at HB
sites, can be used to quantify the degree of -hairpin folding more reliably according to Equation (2),
where and are the chemical shift values for the unfolded analog 1f and the cyclic control
β
δ
δ
100
0
1
e, respectively.
Fraction folded = (δObs − δ )/(δ − δ0)
Specifically, the downfield chemical shifts of Val , Ile , and Orn have been found to be more
(2)
0
100
3
10
8
reliable and are typically used to measure folded fractions for this
β
-hairpin model peptide and its
3
derivatives [14]. Here, because the C
α
of Val is substituted by a nitrogen in our azapeptide analogs,
we omitted this residue from our studies (Table 2). The CH
α
chemical shift values for Ile10 are
particularly informative because this residue sits in the middle of the lower strand and is positioned
8
to form hydrogen bonds with the variable residue in the third position; Orn is situated closer to the
1
0
3
turn region. Based on the chemical shift values for Ile , aza-Val peptide 1b is completely unfolded,
7
in contrast to the partially folded structure predicted based on glycine H
Figure 3, Table 2). Considering that diastereotopic splitting of Gly H
-hairpin model peptides that have d-Pro-Gly turn units, incorporation of azaVal is likely completely
α
splitting measurements
7
(
β
α
overestimate folding in
3
3
disrupting the
β
-hairpin fold. In contrast, the azaGly analog 1d which lacks side chain functionality
was ~50% folded based on the Ile10 downfield H
α
chemical shifts, and exhibited a greater percentage

Molecules 2019, 24, 1919
6 of 12
of folding relative to both the aza-valine (1a) and d-valine (1c) analogs. The additional potential for
azaGly to intra- and/or inter-strand hydrogen-bond may stabilize the extended conformer. The %
8
folding values calculated from Orn chemical shifts reinforced the conclusion that the aza-Valine analog
1
b was the least folded of all analogs, yet suggested similar degrees of folding for the d-Val (1c) and the
azaGly (1d) analogs.
7
8
10
Table 2. % folding based on glycine splitting and Hα chemical shifts for Orn and Ile .
%
Folding
Peptide
Gly δHα–δHα’ Orn8 Ile10
7
Average from Orn /Ile
8
10
1
a
75
17
54
50
72
36
56
50
64
0
15
49
68
18
35.5
49.5
1
b
1
c
1
d
3
. Materials and Methods
3
.1. General
Polystyrene Rink Amide resin (0.78 mmol/g) and Fmoc-glycine-2-chlorotrityl resin (200–400 mesh,
.493 mmol/g) were purchased from Protein Technology Inc and Chem-Impex Int’l, Inc, respectively,
0
™
and the manufacturer’s reported loading of the resin was used in the calculation of the yields of the
TM
final products. Solid phase peptide synthesis was performed using an automated Biotage Syro Wave
peptide synthesizer in 10 mL parallel reactors with polytetrafluoroethylene (PTFE) frits. Incorporation
of aza-amino acids and further elongations were performed manually in disposable filter columns with
20 µM polyethylene (PE) frit filters and caps purchased from Applied Separations (cat # 2413 and 2416
for 3 mL and 6 mL filter columns, respectively) with gentle agitation on a Thermo Fisher vortex mixer
equipped with a microplate tray. Solution draining and washing of the resin was accomplished by
connecting the filter columns to a water aspirator vacuum via a waste trap. Analytical LC-MS analyses
were performed using an Agilent Technologies 1260 Infinity II series LC-MS Single Quad instrument
with ESI ion-source and positive mode ionization, equipped with a 5
column purchased from Mac-Mod Analytical, Inc (cat # 218TP5415). A flow rate of 0.5 mL/min and
–95%, 20–80%, or 12–60% gradients of CH CN [0.1% trifluoroacetic acid (TFA)] in water (0.1% TFA)
µM, 150
×
4.6 mm C18 Vydac
5
3
over 12 min (total run time = 22 min) were used for all LC-MS analyses. Peptides were purified on a
preparative HPLC (Agilent 218 purification system) using a preparative column (10–20 M, 250 mm
2 mm, C18 Vydac column, cat # 218TP101522) at a flow rate of 10 mL/min with gradients of CH CN
µ
×
2
3
[0.1% trifluoroacetic acid (TFA)] in water (0.1% TFA) over 30 min (total run time = 60 min).
3
.2. Reagents
Amino acids, N,N -diisopropylethylamine (DIEA), N,N -Disuccinimidyl carbonate (DSC),
Fmoc chloride, (1-cyano-2-ethoxy-2-oxoethylidenaminooxy)dimethylamino-morpholino-carbenium
0
0
0
0
hexafluorophosphate (COMU), chloro-N,N,N ,N -tetramethylformamidinium hexafluorophospate
TCFH) and triphosgene (BTC) were purchased from Chem Impex Int’l, Inc. Reagents such as
piperidine, lithium chloride, sodium cyanoborohydride (NaBH CN), triisopropylsilane (TIPS), sodium
(
3
acetate-d , acetic acid-d , and D O were purchased from Sigma Aldrich. Reagents such as hydrazine
3
4
2
monohydrate and 2,4,6-collidine were purchased from Alfa Aesar. Trifluoroacetic acid, glacial
acetic acid, sodium bicarbonate, and solvents were purchased from Fisher. Reagents including
HBTU and PyBOP were purchased from Oakwood Chemical. 1-Methylimidazole (NMI) and
1
-chloro-N,N,2-trimethylpropenylamine (Ghosez’s Reagent) were purchased from Acros Organics.

Molecules 2019, 24, 1919
7 of 12
3
.3. 9-Fluorenylmethoxycarbonyl (Fmoc)-Based Solid Phase Peptide Synthesis (SPPS): Fmoc Deprotection and
HBTU Couplings
Peptide syntheses were performed under standard manual conditions on an automated shaker or
TM
using a Biotage Syro Wave peptide synthesizer using Polystyrene Rink Amide resin (0.78 mmol/g).
Couplings of amino acids (3 equiv) were performed in DMF using HBTU (3 equiv) as coupling reagent
and DIEA (6 equiv) as base. Fmoc deprotections were performed initially by treating the resin with
2
0% piperidine in DMF (v/v) for 5 min, followed by treatment with a fresh solution of 20% piperidine
in DMF (v/v) for 15 min. Resin was washed after each coupling with DMF (3 1 mL) and deprotection
reaction with DMF, MeOH, and CH Cl (2 1 mL). Prior to cleavage from the resin or storage, resin was
×
×
2
2
washed with CH Cl (3 × 1 mL).
2
2
3
.4. Test Cleavages of Resin-Bound Peptides
A small amount of resin (1–5 mg) was washed with CH Cl to remove traces of DMF, drained,
2
2
and treated with a freshly made solution of TFA/H O/TIPS (90:5:5, v/v/v, 0.5 mL) for 2 h at room
2
temperature. The cleavage solution was collected by filtering the resin through a disposable, PE fritted
cartridge. The filtrate was evaporated to dryness and the crude peptide was precipitated twice with
cold ether (4 mL) followed by decanting. The pellet (crude peptide sample) was dissolved in 1:1
MeCN/H O v/v (1 mg/mL) and subjected to LC-MS analysis.
2
0
3
.5. N -Isopropyl-fluoren-9-ylmethyl Carbazate (4)
Synthesized according to literature procedures [12
,
16]. Briefly, a solution of Fmoc-Cl (2.00 g,
◦
7
.73 mmol) in 120 mL of CH CN was added dropwise over 2 h at 0 C to a solution of excess hydrazine
3
hydrate (3.8 mL, 78.0 mmol, 10 equiv) in 26 mL of CH CN/H O (1:1, v/v). The solution was warmed
3
2
to room temperature and left stirring for 12 h, prior to being concentrated in vacuo and filtered.
The resulting solid was washed with water, followed by hexanes, to give 1.92 g of white solid in 98%
yield. The resulting 9H-fluoren-9-ylmethyl hydrazinecarboxylate
3 (1.92 g, 7.55 mmol) was suspended
in acetone (50 mL) and heated at reflux for 2 h. Acetone was evaporated in vacuo and the hydrazone
intermediate was dissolved in THF (50 mL), treated with acetic acid (0.48 mL, 8.39 mmol, 1.1 equiv) and
NaBH CN (521 mg, 8.31 mmol, 1.1 equiv), and stirred for 2 h. The volatiles were removed and the crude
3
was dissolved in EtOAc (100 mL), washed with 1 M aqueous KHSO (4
×
50 mL) and brine (2 × 50 mL),
4
dried over Na SO , and concentrated to give a white solid. The obtained product was dissolved in
2
4
EtOH and heated for 1 h, followed by evaporation of EtOH to yield N’-isopropyl-fluoren-9-ylmethyl
carbazate as a white solid (1.32 g, 59% yield) after column chromatography using a 20–100% gradient
of EtOAc in hexanes. NMR (CDCl ) spectra matched literature values [12].
4
3
3
.6. Incorporation of Aza-Valine on the Solid Phase
N’-isopropyl-fluoren-9-ylmethyl carbazate
4 (138 mg, 0.466 mmol, 3 equiv) was dissolved in
◦
dry CH Cl (1 mL) and cooled to 0 C. A solution of bis(trichloromethyl)carbonate (BTC, 46.4 mg,
2
2
◦
0
.156 mmol, 1 equiv) dissolved in dry CH Cl (1 mL) and cooled to 0 C was added dropwise to the
2
2
◦
N’-isopropyl-fluoren-9-ylmethyl carbazate suspension. The reaction mixture was left stirring at 0 C
for 5 min, allowed to warm to room temperature, and stirred for another 25 min. DIEA (0.163 mL,
0
.930 mmol, 6.0 equiv) was added to the solution and the solution was stirred for 5 min prior to being
transferred to a SPPS cartridge containing resin-bound peptide (200 mg, 0.156 mmol) swollen in
CH Cl . The SPPS cartridge was left shaking for 12 h, after which the solution was drained, and the
2
2
2
resin was washed with CH Cl (3 × 2 mL).
2
2

Molecules 2019, 24, 1919
8 of 12
3
.7. Coupling to AzaV-EVpGOKILQ
3
.7.1. BTC Coupling
t
Fmoc-Tyr( Bu)-OH (53.8 mg, 0.117 mmol, 3.0 equiv) was dissolved in dry THF (0.3 mL), cooled
◦
to 0 C, and treated dropwise with a cold solution of BTC (11.5 mg, 0.039 mmol, 1.0 equiv) in dry
◦
THF (0.3 mL). The reaction mixture was left stirring at 0 C for 15 min, before adding 2,4,6-collidine
0.052 mL, 0.39 mmol, 10.0 equiv). The solution was stirred at 0 C for 2 min, allowed to warm to
◦
(
room temperature, and stirred another 5 min. The activated amino acid was then transferred to a SPPS
cartridge containing the resin-bound azapeptide (50 mg, 0.039 mmol), previously swollen in dry THF.
The resin was left shaking at room temperature overnight, then washed with DMF (2
2 × 0.5 mL), and THF (2 × 0.5 mL).
×
0.5 mL), MeOH
(
3
.7.2. PyBOP Coupling
t
Fmoc-Tyr( Bu)-OH (53.8 mg, 0.117 mmol, 3.0 equiv) and PyBOP (60.9 mg, 0.117 mmol, 3.0 equiv)
◦
were each dissolved in DMF (0.3 mL) and cooled to 0 C separately. The PyBOP solution was then
t
◦
added dropwise to the solution of Fmoc-Tyr( Bu)-OH and the reaction mixture was stirred at 0 C
for 5 min. DIEA (0.0408 mL, 0.234 mmol, 6.0 equiv) was added to the solution and the reaction was
allowed to warm to room temperature before being transferred to a SPPS cartridge containing the
resin-bound azapeptide (50 mg, 0.039 mmol), previously swollen in DMF. The resin was shaken at
room temperature overnight, then washed with DMF (2
2 × 0.5 mL).
×
0.5 mL), MeOH (2
×
0.5 mL), and CH Cl
2 2
(
3
.7.3. COMU Coupling
t
Fmoc-Tyr( Bu)-OH (89.6 mg, 0.195 mmol, 5.0 equiv) and COMU (83.5 mg, 0.195 mmol, 5.0 equiv)
◦
were each dissolved in 0.3 mL of DMF and cooled to 0 C separately. The COMU solution was added
t
◦
dropwise to the solution of Fmoc-Tyr( Bu)-OH and the reaction mixture was stirred at 0 C for 5 min.
DIEA (0.0504 mL, 0.39 mmol, 10.0 equiv) was added and the reaction was warmed to room temperature
before being transferred to a SPPS cartridge containing the resin-bound azapeptide (50 mg, 0.039 mmol),
previously swollen in DMF. The resin was shaken at room temperature overnight, then washed with
DMF (2 × 0.5 mL), MeOH (2 × 0.5 mL), and CH Cl (2 × 0.5 mL).
2
2
3
.7.4. NMI/TCFH Coupling
t
Fmoc-Tyr( Bu)-OH (53.8 mg, 0.117 mmol, 3.0 equiv) was dissolved in dry CH Cl (0.3 mL), to which
2
2
◦
N-methyl imidazole (0.028 mL, 0.351 mmol, 9 equiv) was added before cooling the solution to 0 C.
A solution of TCFH (32.8 mg, 0.117 mmol, 3 equiv) in dry CH Cl (0.3 mL), cooled to 0 C, was added
to the Fmoc-Tyr( Bu)-OH and NMI solution in a single portion and the reaction was immediately
transferred to a SPPS cartridge containing the resin-bound azapeptide (50 mg, 0.039 mmol), previously
◦
2
2
t
swollen in dry CH Cl . The resin was shaken at room temperature overnight, then washed with DMF
2
2
(
2 × 0.5 mL), MeOH (2 × 0.5 mL), and CH Cl (2 × 0.5 mL).
2
2
3
.7.5. Ghosez Coupling
NaHCO (98.3 mg, 1.17 mmol, 30 equiv) was added to the resin-bound azapeptide (50 mg,
3
0
.039 mmol) pre-swelled in dry CH Cl (0.2 mL) and shaken for 15 min. During that time,
2 2
t
Fmoc-Tyr( Bu)-OH (53.8 mg, 0.117 mmol, 3.0 equiv) was dissolved in dry CH Cl (0.4 mL) and
cooled to 0 C. Ghosez’s reagent (0.0217 mL, 0.164 mmol, 4.2 equiv) was added dropwise to the cooled
solution of Fmoc-Tyr( Bu)-OH and stirred for 5 min. The reaction was allowed to warm to room
2
2
◦
t
temperature for 5 min, prior to being transferred to the SPPS cartridge containing NaHCO in CH Cl .
3
2
2
The resin was shaken at room temperature overnight. The resin was washed with DMF (2
×
0.5 mL),
MeOH (2 × 0.5 mL), and CH Cl (2 × 0.5 mL).
2
2

Molecules 2019, 24, 1919
9 of 12
3
.8. Incorporation of Aza-Glycine on the Solid Phase
0
N,N -Disuccinimidyl carbonate (DSC) (120 mg, 0.466 mmol, 3.0 equiv) was added to a solution of
9
5
H-fluoren-9-ylmethyl hydrazinecarboxylate
min. The solution was then transferred to resin-bound peptide
3
(123 mg, 0.484 mmol, 3.1 equiv) in DMF and stirred for
(200 mg, 0.156 mmol) swollen in DMF
2
and left shaking for 12 h. The solution was drained and the resin was washed with DMF (3 × 2 mL).
3
.9. Coupling to azaG-EVpGOKILQ
t
Fmoc-Tyr( Bu)-OH (22.2 mg, 0.0484 mmol, 3.1 equiv) was dissolved in dry THF (0.3 mL), cooled
◦
to 0 C, and treated dropwise with a cold solution of BTC (4.63 mg, 0.0156 mmol, 1.0 equiv) in dry
◦
THF (0.2 mL). The reaction mixture was left stirring at 0 C for 15 min before adding 2,4,6-collidine
0.0206 mL, 0.156 mmol, 10.0 equiv). The solution was stirred at 0 C for 5 min. The reaction was
◦
(
allowed to warm to room temperature prior to transferring the activated amino acid to a SPPS cartridge
containing the resin-bound azapeptide (20 mg, 0.0156 mmol) previously swollen with dry THF.
The reaction was left shaking at room temperature overnight, then washed with DMF (2 × 0.5 mL),
MeOH (2 × 0.5 mL), and THF (2 × 0.5 mL).
3
.10. Synthesis of Cyclic Peptide 1e
t
t
The linear precursor H N-Arg(Pbf)-Tyr( Bu)-Val-Glu(O Bu)-Val-pro-Gly-Orn(Boc)-Lys(Boc)-Ile-
2
Leu-Gln(Trt)-pro-Gly-OH was synthesized on Fmoc-glycine-2-chlorotrityl resin (200–400 mesh,
.493 mmol/g loading) according to standard SPPS protocols described above. Liberation of the
0
side chain protected peptide from solid support was conducted using a freshly prepared solution of
hexafluoroisopropanol in CH Cl (20% v/v), the procedure was repeated, solutions were combined,
2
2
and volatiles were removed in vacuo. The protected linear peptide was dissolved in DMF (15 mL) and
ꢀ
a solution of HBTU (41.2 mg, 0.109 mmol, 6 equiv) and HOBt H O (16.6 mg, 0.109 mmol, 6 equiv)
2
in DMF (1 mL) was added dropwise, followed by an addition of DIEA (0.0494 mL, 0.283 mmol,
1
5 equiv) in DMF (1 mL). The reaction was allowed to proceed overnight under N atmosphere. DMF
2
was removed in vacuo and the cyclic side chain protected peptide was carried forward to global
deprotection. A fresh 90:5:5 (v/v/v) solution of TFA: H O: TIPS was prepared and ~10 mL were added
2
to the dried peptide. The reaction was stirred for 2 h at room temperature, followed by the removal of
all volatiles in vacuo. Cold ether precipitation was performed followed by dissolution of the cyclic
peptide in 20% MeCN in water (3 mL). The crude peptide sample was purified by RP-HPLC according
to protocols described in General Methods.
3
.11. Full Cleavage and Purification of (Aza)Peptides
The resin-bound (aza)peptide was washed with CH Cl to remove traces of DMF, drained,
2
2
transferred into a 20 mL scintillation glass vial, and treated a freshly made solution of TFA/H O/TIPS
2
(
90:5:5, v/v/v, 5 mL). The vial was capped and agitated for 2 h at room temperature on an orbital shaker.
The cleavage mixture was filtered through a disposable, PP fritted cartridge into a 50 mL falcon tube
containing cold ether (~20 mL). The ether was decanted following centrifugation at 8000 rpm for
2
min. The peptide pellet was suspended in 1:9 MeCN/H O v/v (5 mL). The sample was frozen and
2
lyophilized. The crude (aza)peptpide was redissolved in a MeCN/H O solvent mixture and purified
2
using reverse-phase HPLC. See Supporting Information for characterization data.
3
.12. NMR Spectroscopy
Lyophilized (aza)peptides were diluted to concentrations of 3–4 mM in D O buffered with 50 mM
2
NaOAc-d3 buffer at pH 4.2 (uncorrected), pH adjusted with AcOH-d3 [36]. All NMR spectra were
collected at the Molecular Education, Technology, and Research Innovation Center (METRIC) at NC
◦
State University on a Bruker NEO 700 MHz instrument with a TCI cryoprobe at 5 C. 1D spectra
were collected with 32K data points and at least 32 scans and water suppression with a pre-saturation

Molecules 2019, 24, 1919
10 of 12
pulse or with a 1D-1H NOESY. 2D-experiments were collected on the same instrumentation, using the
standard Bruker ROESY, COSY, and TOCSY pulse sequences with presaturation. All 2D experiments
were done with eight scans, 2048 points in the first dimension, and 256 or 512 in the indirect dimension.
TOCSY spectra were collected with 80 ms spin-lock and ROESY spectra were collected with a mixing
time of 200 ms. Spectral analysis was performed with TopSpin 4.0.5 software.
4
. Conclusions
Aza-amino acids have previously been shown to stabilize
hyperstability to self-assembling collagen peptide mimics. Here, we expand the folding rules for
aza-amino acids beyond turns and polyproline type II helices and characterize their effect on -hairpin
stability when incorporated into the -strand region. Using NMR spectroscopy, we demonstrate that
a valine to aza-valine substitution significantly disrupts
β
-turn secondary structures and impart
β
β
β
8
-hairpin folding, based on both glycine
10
splitting measurements and CH
α
proton shifts at the Orn and Ile residues relative to cyclic and
unfolded controls. Aza-glycine incorporation at the same position was found to retain some folding,
despite loss of side chain chemistry, which may be due to additional hydrogen bonding. Overall,
our findings support earlier computational hypothesis that aza-amino acids are not well tolerated
within β-sheet secondary structures.
Supplementary Materials: The following are available online: peptide and aza(peptide) characterization data;
1
LC-MS chromatograms; tabulated H NMR data for all peptide and aza(peptide) analogs.
Author Contributions: Conceptualization, C.P. and M.A.M.; Methodology and Analysis, M.A.M., E.L.W., P.C.G.;
Writing-Original Draft Preparation, C.P.; Writing-Review and Editing, C.P., M.A.M., E.L.W., P.C.G.; Supervision,
C.P.; Project Administration, C.P.
Funding: This research received no external funding.
Acknowledgments: The authors thank Peter Thompson for assistance with NMR spectroscopy and North Carolina
State University for startup support. All NMR measurements were made in the Molecular Education, Technology,
and Research Innovation Center (METRIC) at North Carolina State University.
Conflicts of Interest: The authors declare no conflict of interest.
References and Notes
1.
2.
3.
4.
5.
6.
Cheng, P.N.; Pham, J.D.; Nowick, J.S. The Supramolecular Chemistry of beta-Sheets. J. Am. Chem. Soc. 2013
35, 5477–5492. [CrossRef] [PubMed]
Watkins, A.M.; Arora, P.S. Anatomy of beta-Strands at Protein-Protein Interfaces. ACS Chem. Biol. 2014, 9,
747–1754. [CrossRef]
Hughes, R.M.; Waters, M.L. Model systems for beta-hairpins and beta-sheets. Curr. Opin. Struct. Biol. 2006
6, 514–524. [CrossRef]
Loughlin, W.A.; Tyndall, J.D.A.; Glenn, M.P.; Fairlie, D.P. Beta-strand mimetics. Chem. Rev. 2004, 104,
085–6117. [CrossRef]
Doig, A.J. A three stranded beta-sheet peptide in aqueous solution containing N-methyl amino acids to
prevent aggregation. Chem. Commun. 1997, 22, 2153–2154. [CrossRef]
Nowick, J.S.; Chung, D.M.; Maitra, K.; Maitra, S.; Stigers, K.D.; Sun, Y. An unnatural amino acid that mimics
a tripeptide beta-strand and forms beta-sheet like hydrogen-bonded dimers. J. Am. Chem. Soc. 2000, 122,
,
1
1
,
1
6
7
654–7661. [CrossRef]
7
8
9
1
.
.
.
Sarnowski, M.P.; Kang, C.W.; Elbatrawi, Y.M.; Wojtas, L.; Del Valle, J.R. Peptide N-Amination Supports
beta-Sheet Conformations. Angew. Chem. Int. Ed. 2017, 56, 2083–2086. [CrossRef]
Proulx, C.; Sabatino, D.; Hopewell, R.; Spiegel, J.; Ramos, Y.G.; Lubell, W.D. Azapeptides and their therapeutic
potential. Future Med. Chem. 2011, 3, 1139–1164. [CrossRef]
Zhang, Y.T.; Malamakal, R.M.; Chenoweth, D.M. Aza-Glycine Induces Collagen Hyperstability. J. Am.
Chem. Soc. 2015, 137, 12422–12425. [CrossRef] [PubMed]
0. Zhang, Y.T.; Herling, M.; Chenoweth, D.M. General Solution for Stabilizing Triple Helical Collagen. J. Am.
Chem. Soc. 2016, 138, 9751–9754. [CrossRef]

Molecules 2019, 24, 1919
11 of 12
1
1. Thormann, M.; Hofmann, H.J. Conformational properties of azapeptides. J. Mol. Struct. THEOCHEM 1999
69, 63–76. [CrossRef]
2. Reynolds, C.H.; Hormann, R.E. Theoretical study of the structure and rotational flexibility of diacylhydrazines:
Implications for the structure of nonsteroidal ecdysone agonists and azapeptides. J. Am. Chem. Soc. 1996
18, 9395–9401. [CrossRef]
,
4
1
,
1
1
3. Stanger, H.E.; Gellman, S.H. Rules for antiparallel beta-sheet design: D-Pro-Gly is superior to L-Asn-Gly for
beta-hairpin nucleation. J. Am. Chem. Soc. 1998, 120, 4236–4237. [CrossRef]
1
4. Syud, F.A.; Espinosa, J.F.; Gellman, S.H. NMR-based quantification of beta-sheet populations in aqueous
solution through use of reference peptides for the folded and unfolded states. J. Am. Chem. Soc. 1999, 121,
1
1577–11578. [CrossRef]
1
5. Hess, H.J.; Moreland, W.T.; Laubach, G.D. N- [2-Isopropyl-3-(L-Aspartyl-L-Arginyl)-Carbazoyl]-L-Tyrosyl-L-
Valyl-L-Histidyl-L-Prolyl-L-Phenylalanine, an Isostere of Bovine Angiotensin Ii. J. Am. Chem. Soc. 1963, 85,
4
040–4041. [CrossRef]
1
6. Meienhofer, J.; Waki, M.; Heimer, E.P.; Lambros, T.J.; Makofske, R.C.; Chang, C.D. Solid-Phase Synthesis
without Repetitive Acidolysis. Int. J. Pept. Prot. Res. 1979, 13, 35–42. [CrossRef]
1
7. Boeglin, D.; Lubell, W.D. Aza-amino acid scanning of secondary structure suited for solid-phase peptide
synthesis with Fmoc chemistry and aza-amino acids with heteroatomic side chains. J. Comb. Chem. 2005, 7,
8
64–878. [CrossRef]
1
1
2
2
2
8. Chingle, R.; Proulx, C.; Lubell, W.D. Azapeptide Synthesis Methods for Expanding Side-Chain Diversity for
Biomedical Applications. Acc. Chem. Res. 2017, 50, 1541–1556. [CrossRef]
9. Sabatino, D.; Proulx, C.; Klocek, S.; Bourguet, C.B.; Boeglin, D.; Ong, H.; Lubell, W.D. Exploring Side-Chain
Diversity by Submonomer Solid-Phase Aza-Peptide Synthesis. Org. Lett. 2009, 11, 3650–3653. [CrossRef]
0. Kurian, L.A.; Silva, T.A.; Sabatino, D. Submonomer synthesis of azapeptide ligands of the Insulin Receptor
Tyrosine Kinase domain. Bioorg. Med. Chem. Lett. 2014, 24, 4176–4180. [CrossRef]
1. Spiegel, J.; Mas-Moruno, C.; Kessler, H.; Lubell, W.D. Cyclic Aza-peptide Integrin Ligand Synthesis and
Biological Activity. J. Org. Chem. 2012, 77, 5271–5278. [CrossRef] [PubMed]
2. Isocyanate and/or hydantoin byproducts likely arise from activation of the resin-bound semicarbazide with
the carbonyl donor, followed by intramolecular attack by the nitrogen of the preceding amino acid residue.
See: Quibell, M.; Turnell, W.G.; Johnson, T., Synthesis of Azapeptides by the Fmoc Tert-Butyl Polyamide
Technique. J. Chem. Soc. Perk. Trans 1 1993, 22, 2843–2849. [CrossRef]
2
3. Paradis-Bas, M.; Tulla-Puche, J.; Albericio, F. The road to the synthesis of “difficult peptides”. Chem. Soc. Rev.
2
016, 45, 631–654. [CrossRef]
2
4. Murray, J.K.; Farooqi, B.; Sadowsky, J.D.; Scalf, M.; Freund, W.A.; Smith, L.M.; Chen, J.D.; Gellman, S.H.
Efficient synthesis of a beta-peptide combinatorial library with microwave irradiation. J. Am. Chem. Soc.
2
005, 127, 13271–13280. [CrossRef]
2
2
2
2
5. Chingle, R.; Ratni, S.; Claing, A.; Lubell, W.D. Application of Constrained aza-Valine Analogs for Smac
Mimicry. Biopolymers 2016, 106, 235–244. [CrossRef] [PubMed]
6. Arujoe, M.; Ploom, A.; Mastitski, A.; Jarv, J. Comparison of various coupling reagents in solid-phase
aza-peptide synthesis. Tetrahedron Lett. 2017, 58, 3421–3425. [CrossRef]
7. Arujoe, M.; Ploom, A.; Mastitski, A.; Jarv, J. Influence of steric effects in solid-phase aza-peptide synthesis.
Tetrahedron Lett. 2018, 59, 2010–2013. [CrossRef]
8. Beutner, G.L.; Young, I.S.; Davies, M.L.; Hickey, M.R.; Park, H.; Stevens, J.M.; Ye, Q.M. TCFH-NMI: Direct
Access to N-Acyl Imidazoliums for Challenging Amide Bond Formations. Org. Lett. 2018, 20, 4218–4222.
[CrossRef] [PubMed]
2
9. Devos, A.; Remion, J.; Frisquehesbain, A.M.; Colens, A.; Ghosez, L. Synthesis of Acyl Halides under Very
Mild Conditions. J. Chem. Soc. Chem. Comm. 1979, 24, 1180–1181. [CrossRef]
3
0. Gibson, C.; Goodman, S.L.; Hahn, D.; Ho
and Azapeptoides via Fmoc-Strategy and Its Application in the Synthesis of RGD-Mimetics. J. Org. Chem.
999, 64, 7388–7394. [CrossRef]
¨lzemann, G.; Kessler, H. Novel Solid-Phase Synthesis of Azapeptides
1
3
1. Bourguet, C.B.; Sabatino, D.; Lubell, W.D. Benzophenone Semicarbazone Protection Strategy for Synthesis of
Aza-Glycine Containing Aza-Peptides. Biopolymers 2008, 90, 824–831. [CrossRef] [PubMed]

Molecules 2019, 24, 1919
12 of 12
3
3
2. Chingle, R.; Mulumba, M.; Chung, N.N.; Nguyen, T.M.; Ong, H.; Ballet, S.; Schiller, P.W.; Lubell, W.D.
Solid-Phase Azopeptide Diels-Alder Chemistry for Aza-Pipecolyl Residue Synthesis to Study Peptide
Conformation. J. Org. Chem. 2019, 84, 6006–6016. [CrossRef] [PubMed]
3. Searle, M.S.; Griffiths-Jones, S.R.; Skinner-Smith, H. Energetics of weak interactions in a beta-hairpin peptide:
Electrostatic and hydrophobic contributions to stability from lysine salt bridges. J. Am. Chem. Soc. 1999, 121,
1
1615–11620. [CrossRef]
3
3
3
4. Bouayad-Gervais, S.H.; Lubell, W.D. Examination of the potential for adaptive chirality of the nitrogen chiral
center in aza-aspartame. Molecules 2013, 18, 14739. [CrossRef]
5. It should be noted that replacement of valine for an aza-glycine residue results in an overall decrease in
chiral environment, which may affect diastereotopic glycine splitting.
6. Kiehna, S.E.; Waters, M.L. Sequence dependence of beta-hairpin structure: Comparison of a salt bridge and
an aromatic interaction. Protein Sci. 2003, 12, 2657–2667. [CrossRef] [PubMed]
Sample Availability: Samples of the compounds are available from the authors.
©
2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).