Dynamic amphiphile libraries to screen for the "fragrant" delivery of siRNA into HeLa cells and human primary fibroblasts

Article abstract of DOI:10.1021/ja404153m

Dynamic amphiphiles are amphiphiles with dynamic covalent bridges between their hydrophilic heads and their hydrophobic tails. Their usefulness to activate ion transporters, for odorant release, and for differential sensing of odorants and perfumes, has been demonstrated recently. Here, we report that the same "fragrant" dynamic amphiphiles are ideal to screen for new siRNA transfection agents. The advantages of this approach include rapid access to fairly large libraries of complex structures, and possible transformation en route to assist uptake and minimize toxicity. We report single-component systems that exceed the best commercially available multicomponent cocktails with regard to both efficiency and velocity of EGFP knockdown in HeLa cells. In human primary fibroblasts, siRNA-mediated enzyme knockdown nearly doubled from >30% for Lipofectamine to >60% for our best hit. The identified structures were predictable neither from literature nor from results in fluorogenic vesicles and thus support the importance of conceptually innovative screening approaches.

Full text of DOI:10.1021/ja404153m

Communication
pubs.acs.org/JACS
Dynamic Amphiphile Libraries To Screen for the “Fragrant” Delivery
of siRNA into HeLa Cells and Human Primary Fibroblasts
Charlotte Gehin,†,§ Javier Montenegro,†,§,∥ Eun-Kyoung Bang,† Ana Cajaraville,†,∥ Shota Takayama,‡,⊥
Hisaaki Hirose,‡,# Shiroh Futaki,‡ Stefan Matile,*,† and Howard Riezman*,†
†School of Chemistry and Biochemistry, National Centre of Competence in Research (NCCR) Chemical Biology, University of
Geneva, Geneva, Switzerland
‡Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan
S
* Supporting Information
their delivery into cells remain, particularly with regard to in vivo
ABSTRACT: Dynamic amphiphiles are amphiphiles with
dynamic covalent bridges between their hydrophilic heads
and their hydrophobic tails. Their usefulness to activate
ion transporters, for odorant release, and for differential
sensing of odorants and perfumes, has been demonstrated
recently. Here, we report that the same “fragrant” dynamic
amphiphiles are ideal to screen for new siRNA transfection
agents. The advantages of this approach include rapid access
to fairly large libraries of complex structures, and possible
transformation en route to assist uptake and minimize
toxicity. We report single-component systems that exceed
the best commercially available multicomponent cocktails
with regard to both efficiency and velocity of EGFP
knockdown in HeLa cells. In human primary fibroblasts,
siRNA-mediated enzyme knockdown nearly doubled from
>30% for Lipofectamine to >60% for our best hit. The
identified structures were predictable neither from literature
nor from results in fluorogenic vesicles and thus support the
importance of conceptually innovative screening approaches.
applications. They include the frequent need of multicomponent
cocktails in practice, variable efficiencies with different cell lines
depending on the specific properties of their cellular barriers,
and, most importantly, the scarcity of general structure−activity
guidelines for rational design. This situation calls for screening
approaches.1d Dynamic amphiphiles appeared most attractive for
rapid access to fairly large libraries. They are amphiphiles with
dynamic covalent bonds between their hydrophilic heads and
their hydrophobic tails. Dynamic covalent bonds are interesting
tools because they can combine the advantages of strong covalent
bonds and weak non-covalent bonds, depending on con-
ditions.6−11 Dynamic amphiphiles have been introduced recently
as activators of ion transporters,7 for slow release of odorants,8
dynamic micelles, vesicles and gels,9 and as biosensors7 or dif-
ferential sensors10 that function in lipid bilayers. Here, we show
that the same odorant libraries used in “artificial noses” 10 can be
used to screen for the “fragrant” delivery of siRNA. Facile access
to expanded libraries with hydrazone, oxime and disulfide
bridges11 is used to screen up to 900 amphiphiles, either in model
vesicles,10,11 cells, or both. The screening results for GFP knock-
down in HeLa cells disclosed in the following include hits with
fairly complex, totally unpredictable structures that can compete
with multicomponent systems on the market. They operate as
single-component systems, with ability to deliver siRNA into
hard-to-transfect human primary fibroblasts, high reproducibility
and low toxicity.
Our formal library of 900 dynamic amphiphiles was con-
structed from 18 heads and 50 tails (Figures 1 and 2). The heads
contain one to two ammoniums (A) or guanidinium (G) cations
plus one to six reactive groups to form hydrazone (H) or oxime
(O) bridges with aldehyde or ketone containing tails (T1−T50).
For doubly bridged amphiphiles, preformed disulfide bridges (S)
were placed in the head part before amphiphile formation.
The synthesis of all peptide dendrons used as scaffolds in the
head groups has been reported.10,11 Most tails were commercially
available, some had to be prepared following or adapting straight-
forward procedures (Scheme S1).12 Dynamic amphiphiles were
prepared by incubation of heads (e.g., G2H4) and tails (e.g.,
T20) in DMSO, usually for 1 h at 60 °C (Figure 1). The for-
mation of hydrazone-, oxime-, or disulfide-bridged amphiphiles
NA interference (RNAi) has emerged as a powerful method
R
to inhibit the biosynthesis of specific proteins with their
complementary small interfering RNA (siRNA).1 Because of the
great potential for applications in biology and medicine, the
question of how to deliver siRNA to the cytosol has attracted
significant scientific attention.1 Extensive experience with gene
transfection could be adapted partially to yield a high number of
nonviral siRNA transporters, including cationic amphiphiles, peptides,
polymers, dendrimers, nucleobase-lipid hybrids, and more complex
architectures.1,2 Triggerable amphiphiles3a that transform in response
to the pH drop in the endosome include Thompson’s beautiful
vinyl ethers,3b acetals,3c activated esters,3d−f or very elegant ortho-
esters3g and phosphotriesters.3h Pioneering examples for gene
transfection with acid-labile hydrazone bridges have suggested
that degradation en route into cells could be advantageous to
minimize cytotoxicity and maximize oligonucleotide release.4a
However, other studies have suggested that hydrazone-bridged
amphiphiles4 are inferior to disulfide-bridged amphiphiles.4b The
unique power of cytosolic disulfide reduction to minimize
toxicity and release the substrate is also attracting much attention
with cell-penetrating poly(disulfide)s.5
Despite these impressive efforts and the enormous potential
for applications in biology and medicine, problems concerning
Received: April 26, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja404153m | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
transport DNA across fluorogenic vesicles10 but this property
was weaker in the HeLa GPI-EGFP assay (Figure 2). Whereas
the oleyl tails in G2H4T20 are well precedented,1 lauryl tails have
been implied as less suitable for cellular uptake.11,13 Although
particularly promising in vesicles,10,11 fragrant amphi-
philes with tails from jasmine and cyclamen T44 and T45 did not
perform well. We also noticed that minor structural changes down to
single carbon homologues could cause large differences in activity.
This overall poor predictability confirmed the importance of facile
access to large amphiphile libraries as well as methods development
for high-throughput robotic screening of siRNA delivery.
Figure 1. In dynamic amphiphiles (e.g., G2H4T20), heads (e.g., G2H4)
and tails (e.g., T20) are connected with dynamic hydrazone (H), oxime
(O) and/or disulfide bridges (S, compare Figure 2).
The enhanced detergent activity observed in vesicles for the
lengthened oxime and disulfide amphiphiles11 was translated into
an increase of the toxicity in cells (Figures 2, S2, and S3). Replacement
of the guanidinium head by an ammonium decreased the transfection
efficiency of our cationic amphiphiles (Figures 2, S2, and S3).
Screening a panel of increasing concentrations for each amphiphiles
also allowed overcoming the issue of false negative results due to
cytotoxicity. Indeed, some amphiphiles (e.g., G2H4T24) were even
more active when lowering their concentration (Figures 2 and S2).
After validation and optimization of siRNA transfection con-
ditions in HeLa GPI-EGFP (Figures S4−S6), a time-course assay
was performed in order to compare kinetics between active dynamic
amphiphiles (i.e., G2H4T12 and G2H4T20) and Lipofectamine
RNAiMax. In cells whose siEGFP was transfected with dynamic
amphiphiles, a strong fluorescence decrease was already observed
24 h after transfection, well before Lipofectamine RNAiMax whose
action started 48 h post transfection. These different kinetics
suggested that dynamic amphiphiles take a different or more rapid
pathway than Lipofectamine RNAiMax to enter cells (Figure 3a).
siRNA transfection in hard-to-transfect primary human skin
fibroblasts was then evaluated using a custom siRNA sequence
targeting glyceraldehyde 3-phosphate dehydrogenase expression
(siGAPDH).12 Comparison of a pair of our hits (G1H3T12 and
G2H4T20) with Lipofectamine RNAiMax showed a clearly
improved knockdown of GAPDH activity, particularly for the
dynamic amphiphile G2H4T20 (Figure 3b). Cell viability of
fibroblasts transfected with amphiphile/siGAPDH complexes
was confirmed in the same time by measuring LDH activity in
cell supernatant (Figure 3b).
The formation of complexes between siRNA and the best
dynamic amphiphiles was monitored by routine gel shift assays
(Figure S9). According to dynamic light scattering, they have a
diameter around 100 nm, and inversion of their ζ potentials
occurs at a molar ratio of ∼200:1 (Figure S8).12 The lability of
the hydrazone bridges at pH 5.5 was consistent with expectations
from the literature.11,14 In preliminary mechanistic studies, HeLa
cells were incubated for 1 h with 40 μM G1H3T12 and 2 μM of
an FITC-labeled DNA oligomer in serum at 37 °C. After
washing, mostly punctate fluorescence was observed (Figure S7).
This finding was characteristic for the accumulation of DNA in
the endosome, although contributions from small aggregates on
cell surfaces could not be fully excluded. Weaker diffuse emission
from cytosolic areas might suggest that a minor fraction of FITC-
DNA can escape from the endosomes or enter cells directly by
passive diffusion. Clearly reduced uptake at 4 °C further supported
endocytosis as main pathway (Figure S7).
(e.g., G2H4T20) was confirmed by ESI-MS as described
previously.10,11 Some amphiphiles were eliminated early on
because of poor physical properties. The characterization of most
amphiphiles as activators of DNA as cation transporters in fluo-
rogenic vesicles has been reported.10,11 Several underperforming
amphiphiles were also eliminated at this level. The remaining
focused library of 160 amphiphiles was tested for cellular uptake.
For robotic library screening, a GFP silencing assay was
performed in HeLa GPI-EGFP cells (genetically engineered
HeLa cells that stably express GPI-anchored green fluorescent
proteins at their plasma membrane). A liquid-handler (Biomeck
FX) automatically mixed amphiphiles in serum-containing medium,
with a custom siRNA sequence that when properly delivered, has
the ability to knockdown EGFP expression (siEGFP). The HeLa
GPI-EGFP cells were incubated with this mixture for 72 h and
EGFP expression was quantified with a fluorimeter. All experiments
were carried out at constant concentrations of siRNA (33.8 nM)
and DMSO (0.25% (v/v)) and increasing concentrations of
dynamic amphiphiles (Figure 2). No pre-incubation step of siRNA
and dynamic amphiphiles in low serum media (i.e., Opti-MEM) was
required before transfection.
For comparison, parallel experiments were performed with
scrambled siRNA as a negative control. G2H4T20 was used as
the reference in positive control experiments due to its similar
transfection efficiency compared with Lipofectamine RNAiMax
(Figure S1). EGFP knockdown efficiency was calculated as the
percentage of fluorescence decrease observed in cells transfected
with siEGFP compared to transfection with scrambled siRNA.
To facilitate high-throughput screening, cell viability was first
evaluated as the percentage of fluorescence decrease in samples
transfected with complexes made of scrambled siRNA and
amphiphiles compared to untreated cells in medium supple-
mented with 0.25% (v/v) DMSO (see SI for experimental
details).12 Only amphiphiles with cell viability up to 70% were
retained for further validation and optimization. In further
experiments, nontoxicity of selected amphiphiles was confirmed
using a commercially available kit (cytotoxicity detection kit, Roche)
that detects cell death by measuring activity of lactate dehydrogenase
(LDH) released from cells whose plasma membrane was damaged.
The results from robotic high-throughput screening of our
library confirmed that heads and tails alone (33.8 μM final
concentration) were neither toxic nor able to transfect siRNA
(Figure 2g, ø). The best results were generally obtained for
hydrazones with one or two guanidinium heads linked to three or
four aliphatic tails, either long unsaturated (T19, T20, T21, T22,
and T24) or shorter saturated alkyl chains (T12, Figures 2, S2, and
S3). The structures of most of the hits were predictable neither
from literature nor from activities in vesicles.10,11 G2H4T31 and
A2H4T31 have been previously described to be very active to
In summary we have developed an easy and straightforward
methodology toward the preparation of controlled libraries of
dynamic amphiphiles for the screening of single component
siRNA transfecting agents. The reported methodology allows the
rapid identification of unpredictable hits with similar perform-
ance than the actual commercially available cocktails in different
B
dx.doi.org/10.1021/ja404153m | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
Figure 2. (a) Structure of 18 heads and 50 tails used to build a formal library of 900 amphiphiles. (b−g) In vitro screening of 160 dynamic amphiphiles for
siRNA delivery. HeLa cells expressing GPI-EGFP were treated with EGFP-targeting siRNA mixed with dynamic amphiphiles. The average percent
reduction in EGFP expression is shown after treatment with siRNA (33.8 nM) in the presence of (b) 3.4, (c) 6.8, (d) 13.5, (e) 20.3, (f) 27.0, and
(g) 33.8 μM dynamic amphiphiles in the medium (in triplicate). Gray: >50% of cytotoxicity; from yellow to red: 0−20, 20−40, 40−60, 60−80, and
80−100% GFP knockdown.
cells lines. siRNA-mediated enzyme knockdown in human
primary fibroblasts almost doubled to >60% (compared to >30%
for Lipofectamine RNAiMax). Endocytosis of the siRNA seems
to be a major pathway of internalization, but kinetics experiments
C
dx.doi.org/10.1021/ja404153m | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
Jeffs, L. B.; Nechev, L.; Eisenhardt, M. L.; Jayaraman, M.; Kazem, M.;
Maier, M. A.; Srinivasulu, M.; Weinstein, M. J.; Chen, Q.; Alvarez, R.;
Barros, S. A.; De, S.; Klimuk, S. K.; Borland, T.; Kosovrasti, V.; Cantley,
W. L.; Tam, Y. K.; Manoharan, M.; Ciufolini, M. A.; Tracy, M. A.; de
Fougerolles, A.; MacLachlan, I.; Cullis, P. R.; Madden, T. D.; Hope, M. J.
Nat. Biotechnol. 2010, 28, 172. (b) Mevel, M.; Kamaly, N.; Carmona, S.;
́
Oliver, M. H.; Jorgensen, M. R.; Crowther, C.; Salazar, F. H.; Marion, P.
L.; Fujino, M.; Natori, Y.; Thanou, M.; Arbuthnot, P.; Yaouanc, J.-J.;
̀
Jaffres, P. A.; Miller, A. D. J. Controlled Release 2010, 143, 222. (c) Davis,
M. E.; Zuckerman, J. E.; Choi, C. H. J.; Seligson, D.; Tolcher, A.; Alabi,
C. A.; Yen, Y.; Heidel, J. D.; Ribas, A. Nature 2010, 464, 1067. (d) Akinc,
A.; Zumbuehl, A.; Goldberg, M.; Leshchiner, E. S.; Busini, V.; Bacallado,
S. A.; Hossain, N.; Alvarez, R.; Borodovsky, A.; Borland, T.; Constien,
R.; de Fougerolles, A.; Dorkin, J. R.; Jayaprakash, K. N.; Jayaraman, M.;
John, M.; Koteliansky, V.; Manoharan, M.; Nechev, L.; Qin, J.; Racie, T.;
Raitcheva, D.; Rajeev, K. G.; Sah, D. W.; Soutschek, J.; Toudjarska, I.;
Vornlocher, H.-P.; Zimmermann, T. S.; Langer, R.; Anderson, D. G. Nat.
Biotechnol. 2008, 26, 561. (e) Zhang, X.-X.; LaManna, C. M.; Kohman,
R. E.; McIntosh, T. J.; Han, X.; Grinstaff, M. W. Soft Matter 2013, 9,
4472. (f) McCarthy, J.; O’Neill, M. J.; Bourre, L.; Walsh, D.; Quinlan, A.;
Hurley, G.; Ogier, J.; Shanahan, F.; Melgar, S.; Darcy, R.; O’Driscoll, C.
M. J. Controlled Release 2013, 168, 28. (g) Chen, D.; Love, K. T.; Chen,
Y.; Eltoukhy, A. A.; Kastrup, C.; Sahay, G.; Jeon, A.; Dong, Y.;
Whitehead, K. A.; Anderson, D. G. J. Am. Chem. Soc. 2012, 134, 6948.
(2) Bhattacharya, S.; Bajaj, A. Chem. Commun. 2009, 45, 4632.
Figure 3. (a) GFP knockdown (top) and cell viability (bottom) 24, 48,
and 72 h after forward transfection of EGFP-targeting siRNA in HeLa
GPI-EGFP with RNAiMax (blue, bar 1), G2H4T12 (green, bar 2),
G2H4T20 (light green, bar 3) and DMSO (gray, bar 4). (b) Decrease in
GAPDH activity (bars) and cell viability ( ) 72 h after forward
transfection of GAPDH-targeting siRNA in human primary skin fibroblasts
with RNAiMax, G1H3T12, G2H4T20, and DMSO (left to right).
■
indicate that siRNA uptake is much faster than with Lipofect-
amine RNAiMax. Therefore, it is possible that the effective
amphiphile population enters the cell by a more direct route than
endocytosis. The effortless library preparation and the simplicity
of transfection screening certainly certify the potential of
dynamic amphiphiles as controlled delivery vehicles.
(3) (a) Martin, B.; Sainlos, M.; Aissaoui, A.; Oudrhiri, N.;
Hauchecorne, M.; Vigneron, J.-P.; Lehn, J.-M.; Lehn, P. Curr. Pharm.
Design 2005, 11, 375. (b) Boomer, J. A.; Qualls, M. M.; Inerowicz, H. D.;
Haynes, R. H.; Patri, V. S.; Kim, J. M.; Thompson, D. H. Bioconjugate
Chem. 2009, 20, 47. (c) Cui, L.; Cohen, J. L.; Chu, C. K.; Wich, P. R.;
́
Kierstead, P. H.; Frechet, J. M. J. J. Am. Chem. Soc. 2012, 134, 15840.
ASSOCIATED CONTENT
* Supporting Information
Detailed experimental procedures. This material is available free
of charge via the Internet at http://pubs.acs.org.
■
(d) Pijper, D.; Bulten, E.; Smisterova, J.; Wagenaar, A.; Hoekstra, D.;
Engberts, J. B. F. N.; Hulst, R. Eur. J. Org. Chem. 2003, 4406. (e) Prata, C.
A. H.; Zhao, Y. X.; Barthelemy, P.; Li, Y. G.; Luo, D.; McIntosh, T. J.;
Lee, S. J.; Grinstaff, M. W. J. Am. Chem. Soc. 2004, 126, 12196.
(f) Barnard, A.; Posocco, P.; Pricl, S.; Calderon, M.; Haag, R.; Hwang,
M. E.; Shum, V. W. T.; Pack, D. W.; Smith, D. K. J. Am. Chem. Soc. 2011,
133, 20288. (g) Zhu, J.; Munn, R. J.; Nantz, M. H. J. Am. Chem. Soc.
2000, 122, 2645. (h) Pierrat, P.; Kereselidze, D.; Wehrung, P.; Zuber,
G.; Pons, F.; Lebeau, L. Pharm. Res. 2013, 30, 1362. (i) Aytar, B. S.;
Muller, J. P. E.; Kondo, Y.; Talmon, Y.; Abbott, N. L.; Lynn, D. M. J. Am.
Chem. Soc. 2013, DOI: 10.1021/ja403546b. (j) Tang, F.; Hughes, J. A.
Biochem. Biophys. Res. Commun. 1998, 242, 141.
S
AUTHOR INFORMATION
Corresponding Author
stefan.matile@unige.ch; howard.riezman@unige.ch
■
Present Addresses
∥J.M., A.C.: Departamento de Quimica Organica y Centro
Singular de Investigacion en Quimica Biologica y Materiales
Moleculares, University of Santiago de Compostela, Santiago de
Compostela, Spain
(4) (a) Aissaoui, A.; Martin, B.; Kan, E.; Oudrhiri, N.; Hauchecorne,
M.; Vigneron, J.-P.; Lehn, J.-M.; Lehn, P. J. Med. Chem. 2004, 47, 5210.
(b) Xia, W.; Low, P. S. J. Med. Chem. 2010, 53, 6811. (c) Zhang, Y.;
Satterlee, A.; Huang, L. Mol. Ther. 2012, 20, 1298. (d) Rehman, Z.;
Zuhorn, I. S.; Hoekstra, D. J. Controlled Release 2013, 166, 46. (e) Chan,
C.-L.; Majzoub, R. N.; Shirazi, R. S.; Ewert, K. K.; Chen, Y.-J.; Liang, K.
S.; Safinya, C. R. Biomaterials 2012, 33, 4928.
(5) (a) Bang, E.-K.; Gasparini, G.; Molinard, G.; Roux, A.; Sakai, N.;
Matile, S. J. Am. Chem. Soc. 2013, 135, 2088. (b) Bang, E.-K.; Lista, M.;
Sforazzini, G.; Sakai, N.; Matile, S. Chem. Sci. 2012, 3, 1752.
(6) Otto, S. Acc. Chem. Res. 2012, 45, 2200.
⊥S.T.: Department of Medicinal Chemistry, Tokyo University of
Pharmacy and Life Sciences, Hachioji, Tokyo 192-0392, Japan
#H.H.: Institut de Pharmacologie Molec
́
ulaire et Cellulaire,
CNRS, Valbonne, France
Author Contributions
§C.G. and J.M. contributed equally.
Notes
The authors declare no competing financial interest.
(7) Butterfield, S. M.; Miyatake, T.; Matile, S. Angew. Chem., Int. Ed.
2009, 48, 325.
ACKNOWLEDGMENTS
■
(8) Herrmann, A. Chem.Eur. J. 2012, 18, 8568.
We thank A. Fin, D.-H. Tran, Q. Verolet, S. Soleimanpour, and O.
Kucherak for assistance with synthesis and sample preparation;
NMR and MS Platforms and the Biomolecular Screening Facility
(EPFL) for services; and the University of Geneva, the European
Research Council (ERC Advanced Investigator), the NCCR
Chemical Biology, and the Swiss NSF for financial support. H.H.
is grateful for a JSPS Research Fellowship for Young Scientists.
(9) Minkenberg, C. B.; Li, F.; van Rijn, P.; Florusse, L.; Boekhoven, J.;
Stuart, M. C. A.; Koper, G. J. M.; Eelkema, R.; van Esch, J. H. Angew.
Chem., Int. Ed. 2011, 50, 3421.
(10) Takeuchi, T.; Montenegro, J.; Hennig, A.; Matile, S. Chem. Sci.
2011, 2, 303.
(11) Montenegro, J.; Bang, E.-K.; Sakai, N.; Matile, S. Chem.Eur. J.
2012, 18, 10436.
(12) See SI.
(13) (a) Nishihara, M.; Perret, F.; Takeuchi, T.; Futaki, S.; Lazar, A. N.;
Coleman, A. W.; Sakai, N.; Matile, S. Org. Biomol. Chem. 2005, 3, 1659.
(b) Som, A.; Reuter, A.; Tew, G. N. Angew. Chem., Int. Ed. 2012, 51, 980.
(14) Kalia, J.; Raines, R. T. Angew. Chem., Int. Ed. 2008, 47, 7523.
REFERENCES
■
(1) (a) Semple, S. C.; Akinc, A.; Chen, J.; Sandhu, A. P.; Mui, B. L.;
Cho, C. K.; Sah, D. W. Y.; Stebbing, D.; Crosley, E. J.; Yaworski, E.;
Hafez, I. M.; Dorkin, J. R.; Qin, J.; Lam, K.; Rajeev, K. G.; Wong, K. F.;
D
dx.doi.org/10.1021/ja404153m | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX