Article
Reference
Sidechain Engineering in Cell-Penetrating Poly(disulfide)s
MORELLI, Paola, MATILE, Stefan
Abstract
Cell-penetrating poly(disulfide)s (CPDs) have been introduced recently to explore new ways
to enter into cells. In this report, we disclose a general method to covalently modify the
sidechains of CPDs. Compatibility of copper-catalyzed alkyne-azide cycloaddition (CuAAC)
with the addition of either strained cyclic disulfides of varied ring tension or increasing
numbers of guanidinium and phosphonium cations is demonstrated. Reloading CPDs with
disulfide ring tension results in an at least 20-fold increase in activity with preserved sensitivity
toward inhibition with the Ellman's reagent. The cumulation of permanent positive charges by
sidechain engineering affords Ellman-insensitive CPDs with similarly increased activity.
Co-localization experiments indicate that the CPDs reach endosomes, cytosol and nucleus,
depending on their nature and their concentration. Supported by pertinent controls, these
trends confirm that CPDs operate with combination of counterion- and thiol-mediated uptake,
and that the balance between the two can be rationally controlled. For the most active CPDs,
uptake can be observed at substrate [...]
MORELLI, Paola, MATILE, Stefan. Sidechain Engineering in Cell-Penetrating Poly(disulfide)s.
Helvetica chimica acta, 2017, vol. 100, no. 3, p. e1600370
DOI : 10.1002/hlca.201600370
Available at:
http://archive-ouverte.unige.ch/unige:93861
Disclaimer: layout of this document may differ from the published version.
1 / 1
HELVETICA
1
Sidechain Engineering in Cell-Penetrating Poly(disulfide)s
Paola Morelli and Stefan Matile*
Department of Organic Chemistry, University of Geneva, Quai Ernest Ansermet 30, CH-1211 Geneva 4, Switzerland, e-mail: stefan.matile@unige.ch
Cell-penetrating poly(disulfide)s (CPDs) have been introduced recently to explore new ways to enter into cells. In this report, we disclose a general
method to covalently modify the sidechains of CPDs. Compatibility of copper-catalyzed alkyne-azide cycloaddition (CuAAC) with the addition of either
strained cyclic disulfides of varied ring tension or increasing numbers of guanidinium and phosphonium cations is demonstrated. Reloading CPDs with
disulfide ring tension results in an at least 20-fold increase in activity with preserved sensitivity toward inhibition with the Ellman’s reagent. The
cumulation of permanent positive charges by sidechain engineering affords Ellman-insensitive CPDs with similarly increased activity. Co-localization
experiments indicate the CPDs reach endosomes, cytosol and nucleus, depending on their nature and their concentration. Supported by pertinent
controls, these trends confirm that CPDs operate with combination of counterion- and thiol-mediated uptake, and that the balance between the two
can be rationally controlled. For the most active CPDs, uptake can be observed at substrate (fluorophore) concentrations as low as 5 nM.
Keywords: Cell-Penetrating Poly(disulfide)s • Thiol-Mediated Uptake • Disulfide Ring Tension • Cell-Penetrating Peptides • Membranes •
Introduction
Most cationic cell-penetrating peptides (CPPs) are guanidinium-rich
oligomers and polymers (e.g. 1, Fig. 1).
[1–6]
The ability of CPPs to cross
lipid bilayer membranes originates from the poor acidity of the
guanidinium cation.
[2]
They bind to cell membranes by repulsion-driven
ion-pairing interactions with anionic lipids or activators,
[2]
cross the
membrane through dynamic micellar defects,
[3]
and detach from the
membrane by ion exchange with intracellular polyanions (Fig. 1a). This
productive, counterion-mediated delivery into the cytosol (Fig. 1aA) is in
kinetic competition with endocytic uptake and endosomal capture (Fig.
1aB).
In cell-penetrating poly(disulfide)s (CPDs) such as 2,
[7]
the peptide
backbone of CPPs is replaced by a disulfide polymer.
[8–13]
This is of
interest to a) prepare the transporters in situ by ring-opening disulfide-
exchange polymerization,
[7]
b) destroy the transporters upon arrival in the
cytosol by reductive depolymerization to minimize toxicity and liberate
the native substrate,
[8]
and c) integrate new uptake mechanisms (Fig.
1a).
[8]
Namely, disulfide exchange with exofacial thiols attaches CPDs
covalently to the cell surface, disulfide exchange with glutathione
releases them into the cytosol. Existence and significance of
contributions from this thiol-mediated uptake mechanism
[9]
has been
demonstrated by partial CPD inhibition upon removal of exofacial thiols
with Ellman’s reagent.
[8]
Moreover, cellular uptake of monomeric
disulfides has been shown to increase with increasing ring tension from
lipoic acid with a CSSC dihedral angle of 35º in 3 to asparagusic acid with
27º in 4 (Fig. 1b).
[14]
Uptake of activated acyclic disulfides 5 without ring
tension is clearly less efficient.
[14]
Sidechain modification of polymers is of general interest to avoid
tedious optimization of polymerization conditions with every structural
modification and to produce comparable functional systems with
identical scaffold.
[4-6,12,15-21]
Synthetic strategies that in part have been
Figure 1. General structure of CPPs (1), CPDs (2) and monomeric transporters 3–5
with activated disulfides (T = terminator; F = fluorophore / initiator. a) CPDs
combine counterion- and thiol-mediated uptake: Binding to cell membranes
repulsion-driven ion-pairing and disulfide exchange with exofacial thiols is followed
by translocation through micellar defects and release and destruction in the cytosol
by counterion and disulfide exchange. This direct cytosolic delivery (A) is in kinetic
competition with endocytic delivery to endosomes (B). b) Strain-promoted thiol-
mediated delivery of monomeric disulfides.
applied previously to synthetic transport systems include the formation
of hydrazones,
[4,15,16]
sulfonium cations,
[6]
boronate esters,
[17]
amides,
[12]
thioesters,
[18]
disulfides,
[19]
diselenides,
[20]
triazoles,
[5,21]
and so on. In the
following, we introduce a general method for sidechain modification of
endosomes
S
–
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
HS
S
S
S
S
S
S
GS
–
O
HN
O
O
S
S
NH
O
F
S
S
NH
O
F
OHN
O
S
S
O
NO
2
-
O
F
N
N
N
H
H
H
H
H
O
N
N
N
HH
H
H
H
N
H
S
S
S
n
T
F
n
1
2
5
3
4
(CPP)
(CPD)
T
F
S
S
S
–
S
S
F
F
S
S
SR
F
S
S
F
a) b)
cytosol
A
B
HELVETICA
2
Scheme 1. a) Synthesis and b) sidechain modification of representative CPDs (n =
18), with indication of CSSC dihedral angles for c) high (red) and d) low tension
(purple). a) Release of ring tension during substrate-initiated ring-opening
disulfide-exchange polymerization (DMF, TEOA buffer, pH 7.0, 70 ºC, 30 min). b)
Reloading of CPDs with ring tension by sidechain modification with CuAAC (CuSO
4
,
sodium ascorbate, TBTA, 25 ºC, 4 h).
CPDs, apply the secured synthetic access to comparable systems to tackle
the intriguing challenge to reload poly(disulfide)s from ring-opening
disulfide-exchange polymerization with disulfide ring tension (Scheme 1,
TOC graphic) and dissect contributions from counterion-mediated (Fig.
1a)
[2]
and thiol-mediated (Fig. 1b)
[8,9,14]
uptake to end up with the most
active CPDs reported so far.
Results and Discussion
The challenge to reload CPDs with tension is of conceptual interest
because during the preparation of polymers 6 by ring-opening disulfide-
exchange polymerization from initiators 7 to terminators 8, all tension
present in monomers 9 is naturally released (Scheme 1). The idea to
synthesize CPDs with ring tension was thus chemically intriguing.
However, the availability of reloaded CPDs such as 10 was also of
practical interest because increasing uptake activity of monomers 3–5
with increasing tension suggested that similar increases in activity could
be achieved on the level of the intrinsically much more active polymers.
Because of much experience with orthogonal dynamic covalent
chemistry in different contexts,
[15]
hydrazone exchange was initially
envisioned for sidechain modification of CPDs. However, the results were
not convincing (not shown). CuAAC (copper(I)-catalyzed alkyne-azide
cycloaddition),
[5,21,22]
compatible with CPDs,
[23]
was chosen next. In
monomers 9, azides were added without changing the arginine motif in
original CPD 2 (Scheme 1; monomers with alkynes instead of azides gave
poorly soluble CPDs, not shown). The synthesis of monomers 9 was
straightforward, details can be found in the Supporting Information
(Scheme S1). Ring-opening disulfide-exchange polymerization
[7]
of azide
monomers 9 under modified conditions in DMF at 70 °C was initiated
with the previously reported, thiolated and red-emitting
tetramethylrhodamine (TAMRA) fluorophore 7 (Fig. 2), and terminated
with iodoacetamide 8 (Scheme 1). Characterization of the obtained azide
CPDs 6 by GPC gave reproducibly an average M
w
= 8.6 ± 0.8 kDa, an M
n
=
9.5 ± 1.0 kDa, and an excellent PDI = 1.03 ± 0.3 (Fig. 2b). An average of M
w
= 8.6 ± 0.8 kDa obtained for CPDs 6 corresponded to 18 monomers per
polymers (n = 18).
For sidechain modification, alkyne 11 with a less strained lipoic acid
was prepared first (Schemes 1, S2). CuAAC conditions and product
analysis by RP-HPLC after disulfide reduction were established with
monomer 9 (Schemes 1, S5; Table S1). Application of the lessons learned
called for sidechain modification of azide CPDs 6 in H
2
O/THF 9:1 in the
presence of sodium ascorbate, CuSO
4
, tris(benzyltriazoylmethyl)amine
(TBTA), and an excess of alkyne 11 (Scheme 1). RP-HPLC analysis after
reductive depolymerisation of the reloaded CPD 10 revealed triazole 12
as main product together with traces of unreacted azide 13 (Fig. 2a). The
yield of sidechain modification determined by this method calculated to
80-95% for this and most CPDs described in the following.
Figure 2. (a) RP-HPLC of CPD 10 after reductive depolymerization with DTT, with ESI
MS of pertinent peaks. (b) GPCs of several, independently prepared CPDs 6.
Following the developed procedure, the synthesis of CPD 14 with
increased tension in the sidechain was unproblematic (Fig. 3, S2, S7;
Schemes S2, S5; Table S1). As expected from strained disulfide
monomers,
[14]
the introduction of ring tension by sidechain modification
resulted in significant quenching of the TAMRA fluorescence in CPDs 10
and 14 by a factor of 40 and 52, respectively (Fig. 4). Fluorescence
recovery in response to disulfide reduction confirmed that the strained
cyclic disulfides in CPDs 10 and 14 are intact (Fig. 4).
S
90º
S
HN
HN
O
O
N
N
N
H
H
H
H
H
S
S
S
n
10
N
N
N
S
S
HN
O
6
HN
HN
O
O
N
N
N
H
H
H
H
H
N
3
S
S
9
11
7
8
S
S
HN
O
a)
b)
S
S
35º
n
S
–
F
–
OOC
NH
O
O
N
+
N
O
O
NH
2
O
I
H
2
N
O
HN
HN
O
O
N
N
N
H
H
H
H
H
S
S
S
n
N
3
–
OOC
NH
O
O
N
+
N
O
O
NH
2
O
c)
d)
HN
HN
O
O
N
N
N
H
H
H
H
H
HS
SH
12
N
N
N
HN
O
SH
HS
7
COO
-
HN
O
O
N
+
N
SH
O
O
HN
HN
O
O
N
N
N
H
H
H
H
H
HS
SH
13
N
3
a)
b)
1.5
2.0 3.0
0
10
20
30
t / min
40
100
400
700
t / min
m / z
100
300
500 700
m / z
2.5
[M+H]
+
[M+2H]
2+
[M+H]
+
[M+2H]
2+
HELVETICA
3
Figure 3. Structure of sidechain modified CPDs (n = 18). T = CONH
2
, F from TAMRA
7.
Figure 4. Normalized emission intensity of TAMRA 7 (a, black), CPDs 15 (b and e,
blue), 10 (c and f, green) and 14 (d and g, red) before (e, f and g, solid) and after
depolymerization (b, c and d, dashed).
Complementary to the reloading of ring tension in CPDs 10 and 14,
sidechain engineering in CPDs 15–18 was used to explore the
cumulation of positive charges on a constant, comparable polymer
backbone. CPDs 15–17 carry increasing numbers of arginine residues in
their sidechains, and CPD 18 a more hydrophobic triphenyl
phosphonium cation (Fig. 3). The synthesis of the respective sidechain
modifiers was straightforward (Schemes S3, S4). The preparation of alkyne
19 is shown as a representative example (Scheme 2). The synthesis of
CPDs 15 and 18 by CuAAC sidechain engineering as outlined above
(Scheme 1) occurred quantitatively (Figs. S8, S11; Scheme S5; Table S1).
Consistent with the literature,
[24]
CuAAC yields for sidechain engineering
in CPD 6 decreased gradually with increasing bulk of the sidechain
modifiers. Namely, CPD 16 was obtained in 75% yield by reacting the
alkyne 19 containing two more arginines with the azides in CPD 6, CPD
17 with three arginines in only 37% yield (Figs. 3, S9, S10; Scheme S5; Table
S1). Contrary to the situation with strained disulfides in 10 and 14,
sidechain engineering with cumulative charges did not cause significant
quenching of the fluorophores in 15–18 (Figs. 4, S12).
Scheme 2. Representative synthesis of sidechain modifiers: synthesis of alkyne 19.
a) Propargylamine, HATU, DIPEA, DMF, rt, 3 h, 71%; b) piperidine, DMF, rt, 0.2 h,
quant.; c) 20, HATU, DIPEA, DMF, 0.5 h, 93%; d) piperidine, DMF, rt, 0.2 h, 74%; e) TFA,
CHCl
3
, rt, 0.4 h, quant.
Fluorescence quenching up to a factor of 52 in CPDs with reloaded
ring tension but not with cumulated charges complicated quantitative
studies on cellular uptake. Studies on uptake into HeLa Kyoto cells by
flow cytometry
[10]
were less meaningful under these circumstances,
particularly because the extent of intracellular fluorescence recovery by
disulfide reduction, although expected to be complete, was impossible to
assess with full confidence. Studies on cellular uptake therefore had to
rely on the more qualitative but also more informative confocal laser
scanning microscopy (CLSM). CPD concentrations were estimated from
the absorbance of TAMRA in solution. At concentrations as low as 100
nM, incubation with HeLa Kyoto cells in Leibovitz medium for four hours
at 37 ºC revealed uptake for all polymers (Figs. 5, S13). With relatively
short CPDs at very low concentrations, the resulting images were
naturally dominated by puctate emission, usually associated with
inefficient delivery to endosomes (see below). This was consistent with
previous results, delivery into cytosol and nucleoli is observed at higher
concentration and/or with longer polymers.
[25]
However, most images
also contained diffuse emission from the cytosol. At high dilution near
detection limit, this diffuse cytosolic emission can be difficult to see
against background from outside the cells. It is best appreciated from the
contrast provided by the nuclei, appearing as large dark circles. This
clean contrast provided by the “black” nuclei also supported that neither
diffuse nor puctate emission originate from CPD absorbed outside of the
cells,
[26]
possibly also resisting the routinely applied heparin extraction
before imaging.
[27]
The presence of the “black” nuclei naturally confirmed
not only the presence of CPDs in cytosol but also their absence in the
nucleus when delivered at these very low concentrations.
HN
HN
O
O
N
N
N
H
H
H
H
H
S
S
S
n
T
F
N
N
N
HN
H
2
N
O
N
N
N
H
H
H
H
H
HN
HN
O
O
N
N
N
H
H
H
H
H
S
S
S
n
T
F
18
N
N
N
HN
O
P
HN
HN
O
O
N
N
N
H
H
H
H
H
S
S
S
n
T
F
14
N
N
N
S
S
HN
O
m
15: m = 1
16: m = 2
17: m = 3
600 650
700
λ
/ nm
(a, b, c, d)
(e)
(f, g)
575 625
675
HO
O
FmocHN
N
H
NHPbf
NH
a)
FmocHN
O
NH
H
N
NHPbf
NH
H
2
N
O
NH
H
N
NHPbf
NH
d)
N
H
O
NH
H
N
NHPbf
NH
O
NH
NHPbf
HN
FmocHN
N
H
O
NH
H
N
NHPbf
NH
O
NH
NHPbf
HN
H
2
N
N
H
O
NH
H
N
NH
2
+
TFA
–
NH
2
O
NH
NH
2
+
TFA
–
H
2
N
H
2
N
20
21
22
23
24 19
b)
c)
e)
HELVETICA
4
Figure 5. CLSM images of HeLa Kyoto cells after 4 h incubation with 100 nM (left,
laser power: LP = 10%) and 50 nM (right) 2 (a, LP = 8%) and 10 (b, LP = 15%) at 37 °C
in Leibovitz's medium. Scale bar 10 µm.
At 100 nM, similar results were obtained for all CPDs tested. At 50
nM, however, the original CPD 2 and the new azide precursor 6 could not
enter the cells anymore (Figs. 5a, S14a-b). In clear, consistent and
reproducible contrast, all sidechain-engineered CPDs tested remained
active at 50 nM (Figs. 5b, S14c-e). Continuing dilution revealed activity
down to 5 nM of CPD 15 with additional guanidinium cations from
sidechain engineering (Fig. 6b). Similarly intriguing activity was obtained
for CPDs 16 and 17 with additional guanidiniums as well as reloaded ring
tension in their sidechain for CPDs 10 (Fig. 6a) and 14. Even at 1 nM
concentration, uptake of sidechain-engineered CPDs 14–17 was
detectable, although image quality at detection limit became quite poor
(not shown). Uptake activity of CPD 10 with reloaded tension at high
dilution appeared inferior to that of CPD 15 with cumulated charges (Fig.
6a vs 6b). However, these measurements were performed near detection
limit, ultrahigh dilution, and emission intensities depend much on the
release of ring tension (Figs. 4, S12). Apparent differences between CPD
15 with additional guanidiniums and CPD 10 with reloaded ring tension
should thus not be overestimated. What can be said with certainty is that
sidechain-engineered CPDs, either with reloaded tension or cumulated
charges, are at least 20 times more active than the original CPDs with
regard to minimal deliverable substrate concentration, with activity being
detectable down toward the detection limit of the fluorophore used.
Inhibition by the Ellman’s reagent DTNB is a hallmark of all thiol-
mediated uptake.
[8,9,14]
Oxidation of exofacial thiols will destroy their
ability to react with cell-penetrating poly(disulfide)s. Preincubation of the
HeLa cells with DTNB resulted in complete inhibition of the original CPD 2
(Fig. 7a). This result confirmed the validity of previous results from flow
cytometry by fluorescence imaging at higher dilution.
[8]
DTNB inhibition
of the more active CPD 10 with reloaded tension was as powerful as
inhibition of the original CPD 2 (Fig. 7b). The same was true for the azide
precursor 6 and CPD 14 with increased tension. However, DTNB failed to
inhibit the uptake of CPDs 15 (Fig. 7c), 16 (Fig. 7d) and 17 with one to
three additional guanidiniums in their sidechain. These complementary
trends suggested that the Ellman-sensitive CPDs with reloaded tension
enter cells preferably by thiol-mediated uptake,
[8,9,14]
whereas the Ellman-
insensitive CPDs with cumulated charges enter cells preferably by
counterion-mediated uptake (Fig. 1).
[2]
This interpretation supported that
a dual mechanism accounts for the entry of CPDs into cells, and that the
balance between entering cells by thiol-mediated and counterion-
mediated uptake depends on their structure and, presumably, also other
Figure 6. CLSM images of HeLa Kyoto cells after 4 h incubation with 10 (a) and 15 (b) at, from left to right, 50 nM, 25 nM, 10 nM, and 5 nM concentration, 37 °C, Leibovitz's
medium. LP = 15%. Scale bar 10 µm.
a)
b)
10, 50 nM
15, 50 nM
10, 25 nM
15, 25 nM
10, 10 nM
15, 10 nM
10, 5 nM
15, 5 nM
a)
b)
2, 100 nM
10, 100 nM
2, 50 nM
10, 50 nM
