Synthesis of new anionic carbosilane dendrimers via thiol-ene chemistry and their antiviral behaviour

Article abstract of DOI:10.1039/c4ob00162a

A synthetic strategy has been developed for the preparation of anionic carbosilane dendrimers bearing sulfonate or carboxylate groups at their periphery by means of thiol-ene chemistry. It offers significant advantages, such as milder reaction conditions,

Full text of DOI:10.1039/c4ob00162a

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
Synthesis of new anionic carbosilane dendrimers
via thiol–ene chemistry and their antiviral
behaviour†
Cite this: Org. Biomol. Chem., 2014,
12, 3222
Marta Galán,a,b Javier Sánchez Rodríguez,b,c José Luis Jiménez,b,c Miguel Relloso,b,d
Marek Maly,e,f F. Javier de la Mata,*a,b M. A. Muñoz-Fernández*b,d and
Rafael Gómez*a,b
A synthetic strategy has been developed for the preparation of anionic carbosilane dendrimers bearing
sulfonate or carboxylate groups at their periphery by means of thiol–ene chemistry. It offers significant
advantages, such as milder reaction conditions, shorter reaction times and more facile purification
methods, when compared with other synthetic protocols used previously, e.g. hydrosilylation followed by
a Michael-type addition or azide–alkyne coupling reactions. Molecular dynamics simulations of the
second generation anionic dendrimers addressing shape and size effects of the terminal groups and con-
formational variability indicated that the core eccentricity and flexibility might need to be taken into
account for toxicity and interaction with viral and/or cellular receptors, respectively. The biocompatibility
of anionic carbosilane dendrimers has been explored showing differences between silicon-cored and
polyphenoxo-cored dendrimers. In addition, silicon-cored dendrimers achieved 85–90% of HIV inhibition
without inducing inflammation or vaginal irritation in mice, which makes them likely candidates for readily
available, good and safe topical vaginal microbicides against HIV.
Received 21st January 2014,
Accepted 12th March 2014
DOI: 10.1039/c4ob00162a
www.rsc.org/obc
capable of controlling the spread of the disease by interfering
with earlier steps of the viral infection mechanism.3 Microbi-
Introduction
Since the discovery of AIDS in 1981, no effective treatment has cides are chemical entities that can prevent or reduce the
been developed despite all the efforts of the scientific commu- transmission of AIDS and other STIs when applied to the
nity. Nowadays, more than 35 million people live with AIDS vagina (or rectum), which allows for better disease control in
with 2.3 million of new infections in 2012.1 As for the current women, who are the most vulnerable to infection. Many micro-
antiretroviral therapy (HAART), people’s longevity has been bicide candidates are based on polyanionic compounds, which
increased but infection has not been eradicated.2 Therefore, have the capacity to interact with the gp120 protein which is
there is an urgency to find a way to reduce viral spreading. anchored to the viral envelope and is responsible for the
This is the reason why many researchers have focused their binding to host cells, thus inhibiting the viral fusion process,
efforts on the synthesis and preparation of microbicides a vital step in the viral replication cycle.4 Several examples of
this kind of chemicals can be found in the literature,5
although VivaGel®,
a
polylysine-based dendrimer with
naphthylsulfonate groups at the periphery, is currently the
only agent to have made it as far as Phase III trials for HSV
and Phase II trials for HIV.5e,6
aDepartamento de Química Inorgánica, Universidad de Alcalá, Campus
Universitario, E-28871 Alcalá de Henares, Spain
bNetworking Research Center on Bioengineering, Biomaterials and Nanomedicine
(CIBER-BBN), Spain. E-mail: rafael.gomez@uah.es; Fax: (+34) 91 885 4683;
Tel: (+34) 91 885 4685
cPlataforma de Laboratorio, Hospital General Universitario Gregorio Marañón,
Madrid, Spain
dLaboratorio de Inmunobiología Molecular, Hospital General Universitario Gregorio
Anionic dendrimers have been used as an alternative to
polymers in many medical applications for their ability to
present a highly functionalized surface without losing struc-
tural control.7 Over the last few years, our research group has
developed a number of anionic carbosilane dendrimers with
sulfonate, sulfate or carboxylate groups at the periphery that
have been proven to prevent infection of free HIV-1 in vitro.
These dendrimers have been prepared via hydrosilylation reac-
tion followed by a Michael-type addition reaction8 or using a
Marañón, Madrid, Spain
eFaculty of Science, J. E. Purkinje University in Usti n. L., Czech Republic
fUniversity of Applied Science of Southern Switzerland, Department of Innovative
Technologies, SUPSI-DTI Galleria 2, CH-6928 Manno, Switzerland
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
c4ob00162a
3222 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
click chemistry approach based on the azide–alkyne coupling 1-propanesulfonate to afford high yields of {GnSi-
reaction.9 However, in the case of the hydrosilylation reaction [S(CH2)3(SO3Na)m]} (where n = 1, m = 8 (1); n = 2, m = 16 (2) and
the synthesis of these nanomolecules is time-consuming, n = 3, m = 32 (3)) and {GnO3[S(CH2)3(SO3Na)m]} (where n = 1,
requires drastic reaction conditions which allow side reactions m = 6 (4); n = 2, m = 12 (5) and n = 3, m = 24 (6)), respectively.
to happen and reaction conditions which are sometimes The reactions were performed by stepwise addition of the com-
difficult to transfer from one generation to another. An mercial mercaptosulfonate sodium salt along with a photo-
additional problem is the use of platinum as a catalyst in the initiator (2,2-dimethoxy-2-phenylacetophenone, DMPA) in a
process, since it is difficult to remove from the dendritic solvent mixture of THF–MeOH–H2O under 365 nm UV-light
structure afterwards, especially in higher generations. The for 4 hours. In each case, the resulting solutions were purified
azide–alkyne coupling reaction presents a similar problem by nanofiltration through appropriate molecular weight cut-off
with the elimination of copper used as catalyst, which might membranes (MWCO = 500 or 1000 Da), and the solvent elimi-
become a drawback from a toxicological point of view.
nated by evaporation obtaining the desired dendrimers in
Therefore, we have focused our work on the development of moderate yields as white-yellow solids. In cases where the
new synthetic strategies bypassing the use of metals as cata- photoinitiator was retained inside the dendritic structure,
lysts and reducing the number of reaction steps and the reac- nanofiltration was not efficient enough and products had to be
tion time whilst maintaining antiviral capacity. One of these resuspended in a mixture of MeOH–Et2O and subsequently fil-
protocols is based on thiol–ene chemistry which has the tered for complete removal of DMPA. All products were soluble
advantages of click reactions (i.e. no by-products, compatible in water and low generations where also soluble in MeOH.
with many functional groups,10 etc.). No metal catalyst is used When considering vinyl-terminated dendrimers, reactions
in these reactions, but an organic radical photoinitiator that took place under milder conditions, i.e. at lower photoinitiator
can be easily removed by conventional methods (i.e. by concentration.
washing or nanofiltration). Herein, we present the synthesis
The different solubility of the reagents and the products led
and structural behaviour of a new family of anionic carbo- to the formation of carbosilane dendrimer aggregates sub-
silane dendrimers synthesized via thiol–ene methodology as a sequently affording lower yields, when the reaction conditions
metal-free approach, which needs few steps and short reaction were not specifically adjusted. These problems were solved by
times. Their application as antiviral agents against HIV is also adding small amounts of THF to the turbid mixture until com-
discussed.
plete dissolution and by adding reagents in four steps, in
order to avoid aggregate formation due to an increase in ionic
strength or a change of pH.13 The analytical and NMR spectro-
scopic data of compounds 1–6 are consistent with the pro-
posed structures (see Scheme 1 and Fig. 3).
Results and discussion
The synthesis of the allyl-terminated carbosilane dendrimers
has been described elsewhere.11 These precursors were treated
with commercially available thiols under UV-light to provide
the functionalized dendrimers in a 4 hour reaction time, in a
similar way to that reported by Rissing et al.10c,d for the syn-
thesis of small ionic compounds from tetravinylsilane. The
Herein, only NMR data of the outer sphere will be dealt
with, since the carbosilane skeleton has already been
described elsewhere8c,12 and did not undergo modification
upon peripheral functionalization. For compounds 1–3, the
three signals due to the presence of allyl groups disappeared
indicating that the reaction was complete. The 1H-NMR
spectra in D2O, however, showed a new chain assigned to three
methylene groups at ca. 2.60–2.65 ppm for the sulfur-bonded
methylene, 1.60–1.66 ppm for the one in β to the sulfur atom
and 0.60–0.65 ppm for the methylene group bonded to the
silicon atom which can only be assigned with further exper-
1
obtained dendrimers where characterized by H, 13C and 29Si
NMR, as well as by elemental analysis and mass spectroscopy
when possible.
Sulfonate-terminated dendrimers
A new family of sulfonate-terminated dendrimers has been iments such as TOCSY and HSQC, since the peaks overlap
designed, attaching this functionality to the carbosilane den- with the carbosilane scaffold signals. Along with this new
drimer scaffold. The synthetic procedure is based on the chain, three new methylene groups at ca. 3.00, 2.70 and
radical addition of the thiol derivative to the double bond 2.00 ppm confirmed the presence of propanesulfonate units,
function of the spherical dendrimer. Allyl-terminated carbo- corresponding to the methylene groups in α, γ and β to the sul-
silane dendrimers with a GnSiAm-type silicon core (n = 1, m = 8; fonate group, respectively. 13C-NMR data were also consistent
n = 2, m = 16 and n = 3, m = 32, where “Si” means a silicon- with the proposed structures. The most significant resonances
cored structure and Am is the number of terminal allyl were those attributed to the methylene group attached to sulfo-
groups)11 or vinyl-terminated dendrimers, analogue to the nate units (f) that were detected at ca. 50 ppm and those
allyl-terminated dendrimers previously described by our bonded to the inner sulfur atom (c) and (d) located at ca. 35
group12 with a polyphenoxo core GnO3Vm (n = 1, m = 6; n = 2, and 30 ppm, respectively. As an example, 1H- and 13C-NMR
m = 12 and n = 3, m = 24, where “O3” stands for the polyphenoxo- spectra of compound 1 in D2O are shown in Fig. 1.
cored structure and Vm is the number of terminal vinyl groups)
For compounds 4–6, only two groups of signals disappeared
were treated with commercially available sodium 3-mercapto- due to the presence of vinyl groups indicating reaction com-
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3223

View Article Online
Paper
Organic & Biomolecular Chemistry
Scheme 1 Synthetic pathway and proposed dendritic skeletons for dendrimers 1–6, where “Si” or “O3” stand for silicon-cored or polyphenoxo-
cored carbosilane structures, respectively.
Fig. 1 NMR data of the first generation dendrimer {G1Si[S(CH2)3(SO3Na)8]} (1), (A) 1H-NMR (D2O) and (B) 13C-NMR (D2O).
pletion. Analogously, in the 1H-NMR spectra, a new chain of higher field at ca. 14 ppm. The chemical shifts hardly differed
two methylene groups appeared at 2.68 and 0.92 ppm. The from one generation to another, though a broadening of the
lower field location of the methylene group bonded to the signals was observed in both types of anionic dendrimers.
silicon atom compared with that of the analogue group for the
Methylacrylate-terminated dendrimers
allyl derivatives was due to the presence of the sulfur atom,
drawing the chemical shifts to higher ppm values. Three new
methylene groups at ca. 3.00, again 2.68 and 2.01 ppm con-
firmed the presence of propanesulfonate units. The 13C-NMR
data were also consistent with the proposed structures, with
similar shifts to those observed in derivatives 1–3. Only the
methylene group bonded to the silicon atom shifted to a
In a synthetic procedure similar to the one described above,
the addition of methyl thioglycolate to the same double bonds
of functionalized carbosilane dendrimers generated methyl-
acrylate-terminated structures. The reaction was carried out in
the presence of THF as solvent and no photoinitiator was
needed. Completion of the reaction occurred under UV-light in
3224 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
using GnSiAm and GnO3Vm as precursors, respectively (see
Scheme 2 and Fig. 3).
In this case, the 1H-NMR spectra in CDCl3 presented two
new singlets, one for the methylene group located between the
sulfur atom and the carbonyl group SCH2CO at ca. 3.2 ppm
and another one as a result of the methylester located at
ca. 3.7 ppm. The disappearance of the double bond multiplet
was used to monitor the reaction’s progress. In the 13C-NMR
spectra, the most significant signals were the methoxy group
at ca. 52.2 ppm and the methylene attributed to the SCH2CO
group at ca. 33 ppm for all derivatives while the carbonyl
group was located at 171 ppm. The other methylene group
Scheme 2 Synthetic pathway for dendrimers 7–12.
shorter times (less than 2 hours) but was highly dependent on bonded to a sulfur atom appeared at 36 and 28 ppm for the
the amount of solvent used. A stepwise addition was not silicon and polyphenoxo core derivatives, respectively. The
necessary in this case. Thus, quantitative yields of [GnSi- higher field signal observed in the latter ones is attributed to
(SCH2CO2Me)m] (where n = 1, m = 8 (7); n = 2, m = 16 (8) and the presence of a silicon atom in β for these compounds. Fig. 2
n = 3, m = 32 (9)) and [GnO3(SCH2CO2Me)m] (where n = 1, m = 6 shows 1H- and 13C-NMR spectra in CDCl3 recorded for com-
(10); n = 2, m = 12 (11) and n = 3, m = 24 (12)) were synthesized pound 7.
Fig. 2 NMR data of the first generation dendrimers. [G1Si(SCH2CO2Me)8] (7) in CDCl3: (A) 1H-NMR and (B) 13C-NMR. [G1Si(SCH2CO2Na)8] (13) in
D2O: (C) 1H-NMR and (D) 13C-NMR.
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3225

View Article Online
Paper
Organic & Biomolecular Chemistry
Carboxylate-terminated dendrimers
Carboxylate-terminated dendrimers were easily made from the
methylacrylate-terminated systems 7–12 by addition of an
excess of NaOH in MeOH leading to precipitation of the
desired compounds as white solids. The solvent was removed
by filtration and the products were dissolved in water and puri-
fied by means of nanofiltration.
The corresponding dendrimers bearing a silicon core [GnSi-
(SCH2CO2Na)m] (where n = 1, m = 8 (13); n = 2, m = 16 (14) and
n = 3, m = 32 (15)) or a polyphenoxo core [GnO3(SCH2CO2Na)m]
(where n = 1, m = 6 (16); n = 2, m = 12 (17) and n = 3,
m = 24 (18)) were synthesized in high yields (see Scheme 3 and
Fig. 3).
Scheme 3 Synthetic pathway for dendrimers 13–18.
resonance in both 1H and 13C-NMR confirmed product for-
mation (see Fig. 2 for an example).
Analysing the NMR data of dendrimers 13–18, a shift to a
lower field was observed for the signals of the methylacrylate
precursors due to the presence of a negative charge located
next to the carboxylate unit. Disappearance of the methylester
Molecular modelling
Computer models of the second generation anionic dendri-
mers (sulfonate 2 and 5, and carboxylate 14 and 17) were
Fig. 3 Molecular representation of second generation dendrimers 2, 5, 14 and 17.
3226 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 4 The five most representative conformations from the last 50 ns of simulation for each dendrimer, including visualization of the molecular
surface. Dendrimers are 14 (A), 2 (B), 17 (C) and 5 (D). The numbers indicate the percentage of all analyzed conformations (5000 conformations
obtained from the last 50 ns of simulation) which render best the given representative structures in terms of average RMSD distance (see ESI,† for
details). The Si core atom (in 14 and 2) and phenol ring carbon atoms (in 17 and 5) are colored in magenta. Hydrogen atoms are omitted for clarity.
Atoms are color coded as follows: C – grey, O – red, Si – beige, S – yellow.
created and simulations in explicit salt water were conducted core, due to the presence of a longer spacer between the inner
using molecular dynamics in order to get detailed information sulfur atoms and the terminal anionic groups in the case of
about their conformation in a water environment at T = 310 K sulfonate structures (3 C versus 1 C-long spacer). The difference
and P = 0.1 MPa. In general, these studies revealed spherical in size might also be partly attributed to the different inter-
−
−
shapes composed of rather shrunken hydrophobic carbosilane action of the CO2 and SO3 groups with Na+ ions and water
interiors surrounded by the hydrophilic terminal groups on molecules. From Fig. 6E, the average density of the Na+ cations
the molecular surface (see Fig. 4 and 5).
in close proximity to the terminal oxygen atoms (in ca. 2.3 Å
Equilibrated dendrimer structures were subsequently ana- distance) was significantly higher for carboxylate groups than
lyzed. The maximal distance of the dendrimer atoms from the for SO3. This fact is connected not only with the different
dendrimer center of geometry (Rmax), which might be inter- number of oxygen atoms (2 in CO2 versus 3 in SO3) but also
preted as the highest estimate of the molecular radius, radius with the differences in the nature of the C–O and S–O bonds
of gyration (Rg), solvent-accessible surface area (SASA), solvent- (electronic structure, length) and in the whole spatial structure
excluded surface area (SESA), solvent-excluded volume (SEV), of these groups, all of which define an electrostatic field in
sphericity (SPHER) and conformational variability (CV) were their close proximity. These aspects determined the way these
calculated and are shown in Table 1.
anionic groups interacted with Na+ cations but also with water
From the data presented in Table 1, Fig. 4 and 5 it can be molecules (see Fig. 7 and from ESI 5†). Therefore, Na+ ions,
−
deduced that the structures with the central Si core atom (14 which interacted better with CO2 groups, played a better
and 2) are slightly bigger than those with a polyphenoxo central “shielding-glueing” role in carboxylated dendrimers than in
unit (17 and 5). This was particularly clear when structures sulfonated dendrimers, thus contributing to a more compact/
with different cores but the same terminal groups were com- flatter surface (see SASA, SESA surface areas, where a flatter/
pared, which may be mainly attributed to differences in mole- more compact surface should give a smaller area) and also to a
cular weight (or the total number of atoms present). Regarding smaller molecular size. The tighter spatial packing of CO2
the size effect of the terminal groups, the sulfonate structures groups on the molecular surface was also observed from the
were larger than the carboxylate ones though bearing the same radial distribution function of the terminal oxygen atoms (see
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3227

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 6F and ESI† for more data about these interactions). This
fact has also been noted in related dendrimers prepared via
hydrosilylation, as reported elsewhere.8c
Analogously, the simulations led us to the conclusion that
−
CO2 groups also interact better with water molecules (see
Fig. ESI 5†) affecting the structure of the first/second water
shell and its impact on the dendrimer surface morphology.
As shown in Table 1, the largest molecule was dendrimer 2,
with 17 being the smallest. Moreover, dendrimer 17 showed
the lowest CV value which means the smallest conformational
variability (i.e., the highest stability/rigidity) of its 3D structure
(shape) (see also Fig. 4 for corroboration). On the other hand,
when comparing dendrimers with the same core but different
terminal groups, the sulfonate-terminated systems were more
flexible with 5 being the most variable (flexible) structure in
salt water but the most spherical one on average. In con-
clusion, the conformational variability of a given dendrimer
depends mainly on the terminal group, with sulfonated den-
drimers being conformationally more variable (flexible) than
carboxylated ones. This finding might be directly connected
with the differences in terminal spacer lengths and inter-
actions with Na+ ions and water mentioned above.
When considering the conformational variability between
dendrimers with the same terminal groups (see CV values in
Table 1), a different behavior was observed depending on their
nature. In the case of sulfonate dendrimers, the biggest system
2 had a less variable (i.e. more stable) structure than the
smaller dendrimer 5. This was probably due to greater steric
congestion in 2 resulting from the higher number of available
terminal hydrophobic groups. Interestingly, in the case of car-
boxylate dendrimers the reverse situation was true, though the
difference in CV values here is largely negligible.
In order to determine the spatial distribution of different
parts in the studied dendrimers (dendrimer atoms, water
atoms, C or S atoms from terminal groups or ions) with
respect to the dendrimer core, as well as to characterize
mutual distributions of some other important atom groups
(e.g. distribution of Na+ ions with respect to the dendrimer
terminal oxygen atoms), radial density distribution profiles
Fig. 5 Computer models of dendrimers 14 (top-left), 2 (top-right), 17
(bottom-left), 5 (bottom-right) simulated in salt water (A) and their visu-
alized molecular surfaces (B). The Si core atom (in 14 and 2) and phenol
ring carbon atoms (in 17 and 5) are colored in magenta. Hydrogen
atoms are omitted for clarity. The most representative conformations
from the last 50 ns of simulation are shown. Atoms are color coded as
follows: C – grey, O – red, Si – beige, S – yellow.
Table 1 Calculated structural parameters. Rmax – the maximal distance of the dendrimer atoms from the dendrimer center of geometry. Rg
–
radius of gyration. SASA – solvent-accessible surface area. SESA – solvent-excluded surface area. SEV – solvent-excluded volume. SPHER – spheri-
city. CV – conformational variability, average distance (RMSD based) between pairs of structures (of the given dendrimer type) (see ESI for details†)
Si atom core
Polyphenoxo core
Carboxylate
14
Sulfonate
2
Carboxylate
17
Sulfonate
5
Dendrimer
Rmax [nm]
Rg [nm]
1.7 0.132
0.9 0.0277
29.703 1.081
21.944 0.721
3.467 0.027
0.933 0.02
1.782 0.125
0.941 0.022
35.137 1.111
26.26 0.884
4.4036 0.044
0.948 0.023
0.745 0.157
1.503 0.133
0.793 0.032
23.518 0.712
16.945 0.523
2.489 0.042
0.934 0.031
0.6983 0.1571
1.621 0.144
0.857 0.026
29.099 1.581
21.031 1.125
3.132 0.037
0.964 0.017
0.8075 0.1654
SASA [nm2]
SESA [nm2]
SEV [nm3]
SPHER
CV [nm]
0.713 0.144
3228 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 6 Radial distribution density profiles of all dendrimer atoms (A), non-terminal (non-SO3) sulfur atoms (left graph in B) and terminal C or S
atoms (from carboxylate or sulfonate groups) (right graph in B), water atoms (C) and Na+ ions (D), in all cases with respect to selected dendrimer
core atoms (central Si atom or six C atoms of the polyphenol ring). Radial distribution density profiles of Na+ ions with respect to terminal oxygen
atoms (E) and terminal oxygen atoms with respect to themselves (F). Radial distance r measured in Å.
were calculated. The first series of density profiles were calcu-
lated taking into account the dendrimer core (central Si atom
or six C atoms of the polyphenol ring – colored in magenta in
Fig. 4 and 5). These profiles comprised the distribution of all
dendrimer atoms, terminal C atoms (of the carboxylate
−
groups) or S atoms (of the SO3 groups but also the non-term-
inal ones), water atoms and Na+ ions, and results are shown in
Fig. 6.
Fig. 6A shows that the dendrimers with the same core pro-
duced similar RDF profiles which means that the overall con-
formation of solvated dendrimers mainly depends on the
architecture of their core and also on associated structural
−
differences. The variation in terminal groups (CO2−, SO3
)
induced only minor differences here, which were nevertheless
significant in structures bearing the central Si atom at a dis-
tance ranging 5–10 Å. The latter might be connected with the
fact that 14 preferred a slightly more eccentric position of the
central Si atom when compared with 2 (see Fig. 1A and ESI
1†). This is clearly consistent with the water density profiles
(see Fig. 3C) where the better accessibility of the solvent to the
more eccentric 14 central atom was clearly demonstrated by
distinctly higher water density values. In the case of 5 and 17,
the values of water density profiles near the core were notice-
ably higher thanks to the significantly eccentric core position
(providing good accessibility for the solvent).
The differences in the core position were also confirmed by
the profiles of the terminal carbon or sulfur atoms (see right
graph of Fig. 6B and ESI 1 and 2†). Interestingly, the sharp
peak of the non-terminal sulfur atoms profile in dendrimer 2,
centered at about 7.5 Å (see left part of Fig. 6B), revealed
Fig. 7 Top – electrostatic potential around the CO2 (A) and SO3 (B, C)
groups. In case A, the plane visualizing the electrostatic potential is
defined by the 3 atoms of the CO2 group. In case B this plane is defined
by the last carbon atom, sulfur atom and one oxygen atom (the top one
here). In case C the two oxygen atoms (the top ones) and the sulfur atom
define the given plane. Atoms are color coded with C in “dark cyan”, O in
red, S in yellow and H in white. Regarding potential, the red color denotes
low values (−2.57 V and lower) and the blue color means high potential
values (+2.57 V and higher). The effect of water was implicitly taken into
account in this electrostatic calculation. Bottom – the most favorable/fre-
quent position of the Na+ (in purple) cation with respect to “isolated” CO2
(D) and SO3 (E) groups. The cyan lines denote H-bonds and the green
dashed lines together with the green numbers denote O⋯Na+ distances.
The term “isolated” is used here to indicate a terminal group which is not
part of some terminal group cluster sharing one or more Na+ ions.
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3229

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 8 Viability studies of dendrimers 2 and 14. MTS assay on HEC-1A, TZM.bl, HeLa and VK2/E6E7 cells lines and PBMC after 24 h of incubation.
partial backfolding of some branches. However, this pheno-
Biocompatibility of carbosilane dendrimers. The biocom-
menon did not significantly affect the radial distribution of the patibility assays testing the anionic (carboxylate and sulfonate)
hydrophilic terminal groups with respect to the core Si atom carbosilane dendrimers containing both silicon and polyphe-
due to the relatively long spacer between the non-terminal and noxo cores were carried out using the endometrial epithelial
the terminal sulfur atoms.
cell line HEC-1A and the human vaginal mucosa cell line VK2/
The differences in core eccentricity could be a feature to E6E7 as models for the first barrier of protection against HIV
take into consideration, among others, for the explanation of infection in the viral transmission process for in vitro tests.
the different toxicity shown by these systems (vide infra). In Other cell lines, like human epithelial HeLa or TZM.bl and
dendrimers 5 and 17, the higher accessibility of the polyphe- peripheral blood mononuclear cells (PBMC) have also been
noxo core to the biological medium could partially explain the used.
lower biocompatibility when compared with the silicon-cored
For silicon-cored carbosilane dendrimers, the toxic effect
dendrimers. In addition, the same could be true for the subtle was studied in MTS assays using only the second generation
differences observed between the carboxylate dendrimer 14 dendrimers, 2 and 14, as the most representative examples,
and the sulfonate system 2.
and compared with those prepared via hydrosilylation/
Regarding the differences between dendrimers prepared by Michael-type addition procedures reported elsewhere.8a,c The
click thiol–ene or Michael-type addition8 reactions, although sulfonate-terminated dendrimer 2 was biocompatible at con-
the structural parameters like SASA, SESA or SPHER are basi- centrations up to 100 mM in all cell lines tested, while the car-
cally the same, the SEV parameter is always bigger for dendri- boxylate-ended dendrimer 14 showed toxicity in VK2/E6E7 and
mers with a sulfur atom than for those reported via Michael HeLa at this concentration (see Fig. 8). These data are similar
type addition. The bigger solvent-excluded volume can be to those observed in dendrimers prepared via hydrosilylation/
related to a higher hydrophobicity of these compounds lacking Michael-type addition protocols.8a,c
N atoms. Higher hydrophobicity can also be explained by the
In the case of the polyphenoxo-cored dendrimers contain-
lower density profile of water molecules in the internal cavities ing sulfonate, 4–6, or carboxylate, 16–18, units as terminal
of dendrimers with S atoms, both for polyphenoxo or silicon groups the toxicity profiles were studied by an MTT assay and
cores. The hydrophobicity degree could be an important factor showed that all dendrimers were toxic in epithelial cell lines
to tune the interaction of dendrimers with biological mem- HEC-1A and VK2/E6E7 at 10 mM, but biocompatible in PBMC
branes and therefore with viral or cell receptors. In addition, if at all concentrations used (see Fig. 9). The results observed in
radial distribution density profiles are compared between click the epithelial cell lines contrasted with those shown by analo-
thiol–ene and Michael-type addition reactions, differences on gous polyphenoxo-cored dendrimers obtained via hydrosilyl-
the eccentricity were observed for dendrimers with polyphe- ation/Michael-type addition procedure.8a,c The higher toxicity
noxo cores (regardless of the synthetic route employed). Also, detected for dendrimers 4–6 or 16–18 may be tentatively
some eccentricity appears for dendrimer 14 as mentioned ascribed, along with other factors, to the presence of a poly-
above, and not for the equivalent dendrimer obtained through phenoxo core as predicted by molecular modelling (vide
Michael-type addition, which could be explained in terms of supra). Therefore, the observed toxicity prevents their use in
the smaller flexibility of the branches of dendrimer 14. There- subsequent biomedical experiments and only dendrimers 2
fore, these data would contribute to understanding the and 14 were chosen to study their potential anti-HIV
different results observed in terms of toxicity with the different properties.
dendrimers used in this work and those reported elsewhere.8
Anti-HIV activity in epithelial cell lines
Biomedical assays
In spite of the fact that urogenital epithelial cells show a slight
Preliminary biomedical experiments were carried out to deter- level of HIV infection, they are the first line of defense against
mine whether the synthesized dendrimers behave as potential HIV entry into the organism and the first cells encountered
candidates for their use in an anti-HIV microbicide during sexual intercourse. VK2/E6E7 and HEC-1A cell lines
formulation.
were pre-treated with dendrimers 2 or 14 at 10 µM for 1 h,
3230 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 9 Biocompatibility studies of dendrimers 4–6 and 16–18 by MTT assay.
Fig. 10 Anti-HIV activity of dendrimers 2 and 14. VK2/E6E7 (left) and HEC-1A (right) cells were treated with compounds at 10 µM, 1 h before
R5-HIV-1NL(AD8) or X4-HIV-1NL4.3 infection. Supernatants were collected after 24 h and p24gag antigen levels quantified by ELISA. [Suramin (Sur)] =
10 µM. *p-value ≤ 0.01; **p-value ≤ 0.001; ***p-value ≤ 0.0001 compared with HIV-1 infected non-treated cells (control: C+).
then infected with X4-HIV-1NL4.3 or R5-HIV-1NL(AD8) for 3 h.
Suramin was used as positive control of HIV inhibition. Infec-
tion of VK2/E6E7 and HEC-1A cell lines was followed by
quantification of p24gag antigen production in the supernatant
of the culture cells after 24 h. An inhibitory effect of p24gag
antigen production was observed in VK2/E6E7 and HEC-1A
when cells were pre-treated with either dendrimer (Fig. 10).
Results for VK2/E6E7 showed a significant HIV-1 inhibition
when cells were pre-treated with dendrimers 2 or 14 (about
85–90% inhibition) in both viral strains compared with control
samples and suramin.
Table 2 Profile of cytokines secreted by VK2/E6E7. Cells were treated
with dendrimer 2 or 14 (10 µM) for 24 h. Cytokine expression levels
were determined in cell culture supernatants by flow cytometry using
the Diaplex Human Th1/Th2/inflammation kit. ND: not detected. Poly
(inosinic acid–cytidylic acid), (Poly I–C) was used as positive control of
cytokine production. Values are expressed in pg mL−1
Profile of cytokines secreted by VK2/E6E7
VK2/E6E7
Basal
Poly I–C
Dendrimer 2
Dendrimer 14
IL-12
TNF-α
IL-4
2.75
17.46
6.31
ND
ND
ND
2.75
17.46
6.87
2.75
17.46
6.87
IL-6
IL-8
16.75
178.05
265.14
941.90
16.75
351.36
16.75
560.36
Non-inflammatory response: Th1/Th2 cytokine profiles
In order to determine if the treatment with 2 or 14 resulted in
changes in the cytokine expression profile that could sub-
sequently activate an inflammatory pathway in neighboring
cells, the cytokine profiles of treated VK2/E6E7 cells were ana-
lyzed at different times using the Diaplex Human Th1/Th2/
inflammation kit. The treatment with dendrimers 2 or 14 did
not increase expression levels of any cytokine, compared with
the medium alone (control cells) (see Table 2). The results
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3231

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 11 Histopathological examination of vaginal tissues after intravaginal application of dendrimers 2 or 14. Slides were prepared from vaginal
tissues treated with 14 LD 2 h (a), 14 HD 2 h (b), 2 LD 2 h (c), 2 HD 2 h (d), vehicle (PBS) 2 h (e), 14 LD 24 h (f), 14 HD 24 h (g), 2 LD 24 h (h), 2 HD
24 h (i), vehicle (PBS) 24 h ( j). Hematoxylin-eosin staining was used for all slides. Magnification, ×100. LD (low dose, 10 µM), HD (high dose, 100 µM).
showed that neither of the dendrimers was able to activate a
Conclusions
cellular inflammatory pathway (in case of IL-8, values were not
statistically significant)9 which is essential to make a safe A synthetic strategy has been developed for the preparation of
topical microbicide available.
anionic carbosilane dendrimers bearing sulfonate or carboxy-
late groups at their periphery. It is based on the use of thiol–
ene chemistry which offers several advantages over other
different synthetic protocols used for preparing similar dendri-
mers, e.g. hydrosilylation reaction followed by a Michael-type
addition8 or alkyne-azide coupling.9 First of all, this procedure
entails fewer reaction steps and milder reaction conditions.
While the hydrosilylation protocol requires temperatures
ranging between 90–120 °C and long reaction times, normally
2–3 days, and the azide–alkyne procedure is carried out at
40–60 °C and takes at least 24 h, the thiol–ene reactions work
at room temperature and require 4 h for completion. In
addition, no metal catalyst is used in these reactions, but an
organic radical photoinitiator that can easily be removed by
conventional methods (i.e. by washing or by nanofiltration).
Therefore, this synthetic technology offers milder reaction
conditions, shorter reaction times and easier purification
methods.
Topical vaginal safety in an animal model
The effect of dendrimers 2 and 14 on topical mucosa was
studied in a mouse model for in vivo evaluation, analyzing the
integrity of the mucosal tissue. Histopathological examination
showed that dendrimer-treated mice displayed neither damage
nor inflammatory response in cervicovaginal tissues (see
Fig. 11). No mortality or signs of vaginal discharge, erythema
or edema were observed in mice treated with the lowest dose
(10 µM) and higher dose (100 µM) of dendrimers 2 and 14 (see
ESI Table 1†). In post-mortem and histopathological examin-
ations of the vaginal tissues, no significant abnormalities were
observed in the mucosa in the presence of the dendrimers
compared with the vehicle (negative control). The vaginal
irritation index was calculated by scoring the microscopic
observations (irritation of epithelium, leukocyte infiltration,
vascular congestion and edema) and achieved values of 1.7
and 0.3 after 2 h, and 0.3 and 1.3 after 24 h for dendrimer 14
at low and high dose, respectively, and values of 1 and 2 at 2 h,
and 2.3 and 2.67 at 24 h for dendrimer 2 at low and high dose,
respectively. According to Eckstein et al., a vaginal irritation
index <8 is acceptable, 9–10 is marginal, and ≥11 unaccepta-
ble.14 These data confirmed the biocompatibility of both
dendrimers applied to the vaginal mucosa of mice. Thus, den-
drimers 2 and 14 did not induce vaginal irritation and seem to
be a safe product for topical application.
Molecular dynamics simulations of the second generation
anionic dendrimers bearing a silicon core, 2 and 14, or a poly-
phenoxo core, 5 and 17, were performed in explicit salt water.
In general, these studies revealed spherical shapes composed
of rather shrunken hydrophobic carbosilane interiors sur-
rounded by the hydrophilic terminal groups on the molecular
surface. Regarding the size effect of the terminal groups, the
sulfonate structures are bigger and more flexible than the car-
boxylate ones containing the same core. This can be ascribed
to a longer spacer between the inner sulfur atoms and the
terminal anionic groups, but also partly attributed to the
−
−
different interactions of the CO2 and SO3 groups with the
3232 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Na+ ions and water molecules. With regard to conformational good, safe and readily available candidates for topical vaginal
variability, polyphenoxo-cored dendrimers 5 and 17 show a microbicides against HIV. Deeper biomedical studies are in
high core eccentricity which means that in these cases the core progress in order to further investigate their potential use as
is rather close to the molecular surface and therefore more microbicide agents.
accessible to the solvent. This difference could be linked to the
different toxicity shown by these systems when compared with
silicon-cored dendrimers. The same may apply for the subtle
Experimental section
differences observed between the carboxylate dendrimer 14
General methods
and the sulfonate system 2. From the theoretical analysis, den-
Unless otherwise stated, reagents were obtained from commer-
cial sources and used as received. Compounds GnXYm (X = Si,
Y = allyl and X = O3, Y = vinyl) were synthesized as pub-
lished.11,12,16 Thiol–ene reactions were carried out employing a
HPK 125W Mercury Lamp from Heraeus Noblelight with
maximum energy at 365 nm, in normal glassware under inert
atmosphere. NMR spectra were recorded on a Varian Unity
VXR-300 (300.13 (1H), 75.47 (13C) MHz) or on a Bruker AV400
(400.13 (1H), 100.60 (13C), 79.49 (29Si) MHz). Chemical shifts
(δ) are given in ppm. 1H and 13C resonances were measured
relative to solvent peaks considering TMS = 0 ppm, while 29Si
resonances were measured relative to external TMS. When
necessary, assignment of resonances was done from HSQC
and TOCSY NMR experiments. Elemental analyses were per-
formed on a Perkin-Elmer 240C. Mass Spectra were obtained
from an Agilent 6210 (ESI).
drimer 2 presents structural differences that make it more flexi-
ble than the corresponding analogous carboxylate system 14.
This property may be very useful for an antiviral agent, since
more flexibility means more effectiveness of the surface when
interacting with important domains of viral or cellular
receptors.
The biocompatibility of anionic carbosilane dendrimers
containing either silicon or polyphenoxo cores has been tested
on endometrial epithelial cell lines HEC-1A and human
vaginal mucosa VK2/E6E7 as models of the first barrier of pro-
tection against HIV infection as well as on other cell lines like
HeLa or TZM.bl and primary culture PBMC. Silicon-cored den-
drimers are non-toxic up to 100 mM, while the polyphenoxo-
cored systems show reduced biocompatibility as predicted by
molecular modelling studies. For this reason, only silicon-
cored second generation dendrimers 2 and 14 were used for
HIV antiviral assays, showing 85–90% of inhibition compared
with the control and higher values than the positive control
Synthesis of compounds
suramin in both viral strains, VIH-X4 and HIV-R5. In addition, Dendrimers are named as GnXYm, where n = 1, 2, 3 for first,
both dendrimers did not trigger inflammation nor vaginal irri- second and third generation dendrimers respectively; X = Si or
tation in mice. These features are important and need to be O3 for silicon or polyphenoxo-cored dendrimers and Ym = allyl
considered, since creating an inflammatory state may fatally or vinyl groups in the periphery.
increase viral transmission, as described in the literature on
the failure of nonoxinol-9 (N-9) as an anti HIV vaginal mers are presented here, as they were chosen for molecular
microbicide.15
modeling and biomedical assays. Complete characterization of
Only characterization data for second generation dendri-
Regarding the differences between dendrimers prepared by all compounds along with synthetic procedure and selected
click thiol–ene or Michael-type addition8 reactions, the former NMR spectra can be found in the ESI.†
introduces a sulfur atom in the structure, while a nitrogen
G2Si[S(CH2)3(SO3Na)16] (2). C144H300Na16O48S32Si13 (4558.92).
atom is present in the latter. The basicity of the inner nitrogen White powder solid (1.08 g, 80%). Reagents: G2SiA16 (0.51 g,
creates zwitterionic species in the case of the carboxylate den- 0.29 mmol), sodium 3-mercapto-1-propanesulfonate (1.02 g,
1
drimers at physiological pH, a situation that cannot occur in 5.73 mmol), DMPA (0.15 g, 0.57 mmol). H-NMR (D2O): δ 2.99
analogous dendrimers from the thiol–ene protocol. The exist- (32
H,
t,
SCH2CH2CH2SO3Na),
2.66
(32
H,
t,
ence of non-protonated structures and the conformation SCH2CH2CH2SO3Na), 2.59 (32 H, t, SiCH2CH2CH2S), 2.02 (32
adopted by the carboxylate dendrimers and confirmed by H, m, SCH2CH2CH2SO3Na), 1.61 (32 H, m, SiCH2CH2CH2S),
molecular modeling may explain the different biomedical be- 1.41 (24 H, m, SiCH2CH2CH2Si), 0.65 (80 H, m,
havior observed. The fact that no big differences were found SiCH2CH2CH2Si(Me)CH2CH2CH2S), 0.04 (36 H, s, SiMe).
for the sulfonate dendrimers prepared from the two synthetic 13C-NMR (D2O):
protocols corroborates the similar biomedical behavior. In (SiCH2CH2CH2S),
δ
50.04 (SCH2CH2CH2SO3Na), 35.23
30.27 (SCH2CH2CH2SO3Na), 24.44
addition, the different hydrophobicity and eccentricity values (SCH2CH2CH2SO3Na), 23.93 (SiCH2CH2CH2S), 18.58 (SiCH2-
found in dendrimers synthesized by the two protocols should CH2CH2Si, SiCH2CH2CH2Si), 13.11 (SiCH2CH2CH2S), −4.26
be taken into consideration for biomedical purposes.
(SiMe), −5.04 (SiMeCH2CH2CH2S). 29Si-RMN (D2O): δ 2.52
Therefore, with these very promising data concerning con- (SiMeCH2CH2CH2S), 1.21 (SiMeCH2CH2CH2Si). Elemental
formation, biocompatibility, inhibition, non-inflammatory analysis: Calc. %: C, 37.94; H, 6.63; S, 22.51; exp %: C, 36.54;
response and integrity of the mucosal tissues against irri- H, 6.39; S, 23.26.
tation, combined with a clean, simple and metal-free synthetic
G2O3[S(CH2)3(SO3Na)12] (5). C105H210Na12O39S24Si9 (3394.97).
method based on thiol–ene chemistry, dendrimer 2 and den- White powder solid (0.43 g, 80%). Reagents: G2O3V12 (0.20 g,
drimer 14, though perhaps to a lesser degree, seem to be 0.16 mmol), sodium 3-mercapto-1-propanesulfonate (0.42 g,
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3233

View Article Online
Paper
Organic & Biomolecular Chemistry
1
1.92 mmol), DMPA (0.05 g, 0.19 mmol). H-NMR (D2O): δ 5.98
G2O3[(SCH2CO2Na)12] (17). C93H162Na12O27S12Si9 (2625.69).
(3 H, s, ArH), 3.77 (6 H, t, CH2O), 2.97 (24 H, White solid powder (0.30 g, 60%). Reagents: 11 (0.45 g,
1
t, SCH2CH2CH2SO3Na), 2.65 (48 H, m, SiCH2CH2S, 0.18 mmol), NaOH (0.26 g, 6.6 mmol). H-NMR (D2O): δ 5.94
SCH2CH2CH2SO3Na), 2.00 (24 H, m, SCH2CH2CH2SO3Na), 1.37 (3 H, s, Ar-H), 3.76 (6 H, t, CH2O), 3.30 (24 H, s, SCH2CO), 2.69
(18 H, m, OCH2CH2CH2CH2Si, SiCH2CH2CH2Si), 0.92 (24 H, t, (24 H, t, SiCH2CH2S), 1.69 (6 H, m, OCH2CH2CH2CH2Si), 1.43
SiCH2CH2S), 0.60 (30 H, m, OCH2CH2CH2CH2Si, SiCH2- (18 H, m, OCH2CH2CH2CH2Si, SiCH2CH2CH2Si), 0.69 (54 H,
CH2CH2Si), 0.06 (18 H, s, SiMeCH2CH2S), −0.04 (9 H, s, SiMe). m, SiCH2CH2S, OCH2CH2CH2CH2Si, SiCH2CH2CH2Si), 0.13
13C-NMR (D2O): δ 67.53 (OCH2), 50.12 (SCH2CH2CH2SO3Na), (18 H, s, SiMeCH2CH2S), 0.03 (9 H, s, SiMe). 13C-NMR (D2O): δ
30.13 (SiCH2CH2S), 26.96 (SCH2CH2CH2SO3Na), 24.34 (SCH2- 177.98 (CO), 160.52 (Cipso), 93.91 (ArC), 67.18 (OCH2),
CH2CH2SO3Na), 18.52 (SiCH2CH2CH2Si), 14.33 (SiCH2CH2S, 37.08
SiCH2CH2CH2Si), −4.58 (SiMe), −5.48 (SiMeCH2CH2S). (SiCH2CH2S), 18.43 (OCH2CH2CH2CH2Si), 14.07 (SiCH2CH2S,
29Si-RMN (D2O):
2.10 (SiMeCH2CH2S), 1.00 (SiMeCH2- OCH2CH2CH2CH2Si), −4.67 (SiMeCH2CH2S), −5.16 (SiMe).
CH2CH2Si). Elemental analysis: Calc. %: C, 37.15; H, 6.23; S, 29Si-RMN (D2O):
2.10 (SiMeCH2CH2S), 1.70 (SiMeCH2-
22.67; exp %: C, 37.09; H, 6.64; S, 25.51. CH2CH2Si). Elemental analysis: Calc. %: C, 42.54; H, 6.22; S,
(SCH2CO),
32.92
(OCH2CH2CH2CH2Si),
27.83
δ
δ
G2Si[(SCH2CO2Me)16] (8). C144H284O32S16Si13 (3405.93). Oil 14.65; exp %: C, 40.25; H, 6.05; S, 12.29.
(0.25 g, 60%). Reagents: G2SiA16 (0.22 g, 0.13 mmol), methyl
1
Computational details
thioglycolate (0.2 mL, 2.02 mmol). H-NMR (CDCl3): δ 3.71 (48
H, s, COOCH3), 3.19 (32 H, s, SCH2CO), 2.62 (32 H, t, 3D computer models of dendrimer structures were created
SiCH2CH2CH2S), 1.55 (32 H, m, SiCH2CH2CH2S), 1.23 (24 H, using dendrimer builder as implemented in Materials Studio
m, SiCH2CH2CH2Si), 0.54 (80 H, m, SiCH2CH2CH2Si(Me)- software package from Accelrys Inc. The individual dendrimer
CH2CH2CH2S), −0.05 (24 H, s, SiMeCH2CH2CH2S), −0.09 (12 residues were parametrised using Antechamber tool which is
H, s, SiCH2CH2CH2SiMe). 13C-NMR (CDCl3):
δ
170.89 part of Amber12 software.17 For calculation of partial charges
(COOCH3), 52.30 (COOCH3), 36.34 (SiCH2CH2CH2S), 33.33 AM1-BCC approach was used.18 GAFF force field (Generalized
(SCH2COOCH3), 23.66 (SiCH2CH2CH2S), 13.30 (SiCH2- Amber Force Field) was used for parametrisation of our den-
CH2CH2S), 18.84–18.36 (SiCH2CH2CH2Si, SiCH2CH2CH2Si), drimer models.19 The missing bond/angle/torsion terms con-
−5.25 (SiMe). 29Si-NMR (CDCl3): δ 2.69 (SiMeCH2CH2CH2S), taining silica atoms were derived from QM calculations. These
0.81 (SiCH2CH2CH2SiMe). Elemental analysis: Calc. %: C, ff parameters were fitted by minimizing the differences
50.78; H, 8.40; S, 15.06; exp %: C, 51.41; H, 7.84; S, 15.09.
between QM and force field based relative energies of properly
G2O3[(SCH2CO2Me)12] (11). C105H198O27S12Si9 (2526.87). Oil chosen molecular fragments using paramfit routine from
(0.50 g, 80%). Reagents: G2O3V12 (0.31 g, 0.25 mmol), methyl AMBER12 software. 100 conformations of each molecular frag-
thioglycolate (0.30 mL, 3.00 mmol). 1H-NMR (CDCl3): δ 6.02 ment were used for the force field parameters fitting. QM ener-
(3 H, s, Ar-H), 3.86 (6 H, t, CH2O-Ar), 3.69 (36 H, s, COOCH3), gies were calculated at MP2/HF/6-31G** level of theory using
3.21 (24 H, s, SCH2CO), 2.63 (24 H, t, SiCH2CH2S), 1.74 (6 H, GAMESS software.20 Van der Waals parameters for Si were
m, OCH2CH2CH2CH2Si), 1.40–1.20 (18 H, m, OCH2CH2- taken from the MM3 force field.21 The dendrimer structures
CH2CH2Si, SiCH2CH2CH2Si), 0.87 (24 H, m, SiCH2CH2S), 0.55 were solvated in explicit water (TIP3P model) with proper
(30 H, m, OCH2CH2CH2CH2Si, SiCH2CH2CH2Si), −0.01 (18 H, number of counterions and additional salt to preserve neu-
s, SiMeCH2CH2S), −0.08 (9 H, s, SiMe). 13C-NMR (CDCl3): δ trality of the whole system and to ensure proper ionic strength
170.73 (COOCH3), 160.75 (Cipso), 93.54 (ArC), 67.55 (OCH2), (0.15 M).22 All molecular systems were then minimised in two
52.23 (COOCH3), 33.18 (SCH2COOCH3), 32.55 (OCH2CH2- stages. During the first stage, which served for elimination of
CH2CH2Si), 28.11 (SiCH2CH2S), 20.46 (OCH2CH2CH2CH2Si), all bad molecular contacts, solvated molecules were restrained
18.58 (SiCH2CH2CH2Si), 13.85 (SiCH2CH2S, OCH2CH2- (restraint constant 5 kcal (mol Å2)−1) and systems were sub-
CH2CH2Si), −5.21 (SiMe), −5.38 (SiMeCH2CH2S). 29Si-NMR jected to 500 steps of steepest descent and 4500 conjugate gra-
(CDCl3): δ 2.40 (SiMeCH2CH2S), 1.71 (SiCH2CH2CH2SiMe). dient minimization. This first minimization stage was
Elemental analysis: Calc. %: C, 49.84; H, 7.89; S, 15.21; exp %: followed by a second one without any restraint applied to let
C, 51.43; H, 7.84; S, 15.09. ESI-MS: [M + (NH4)]+ = 2546.99.
the dendrimer structures relax in a water environment (2000
G2Si[(SCH2CO2Na)16] (14). C128H236Na16O32S16Si13 (3533.21). steepest descent steps + 5000 steps of conjugate gradient).
White solid powder (0.21 g, 84%). Reagents: 8 (0.24 g, Then all systems were slowly heated using NVT conditions to
1
0.07 mmol), NaOH (0.13 g, 3.36 mmol). H-NMR (D2O): δ 3.21 achieve the desired temperature (310 K). After heating, all
(32 H, s, SCH2CO), 2.59 (32 H, t, SiCH2CH2CH2S), 1.62 (32 H, molecular systems were simulated using molecular dynamics
m, SiCH2CH2CH2S), 1.40 (24 H, m, SiCH2CH2CH2Si), 0.65 in NPT ensemble (T = 310 K and P = 0.1 MPa) for 150 ns.
(80 H, m, SiCH2CH2CH2Si(Me)CH2CH2CH2S), 0.04 (36 H, Hydrogens were constrained with the SHAKE algorithm23 to
s, SiMe). 13C-NMR (D2O): δ 178.12 (CO), 36.93 (SCH2CO), allow 2 fs time step. Langevin thermostat with collision fre-
35.93 (SiCH2CH2CH2S), 23.59 (SiCH2CH2CH2S), 18.55 (SiCH2- quency 2 ps was used for all molecular dynamics runs.24 The
CH2CH2S, SiCH2CH2CH2Si, SiCH2CH2CH2Si), 13.11 (SiCH2- pressure relaxation time for weak-coupling barostat was 2 ps.
CH2CH2S), −5.10 (SiMe). Elemental analysis: Calc. %: C, 43.51; Particle mesh Ewald method (PME)25 was used to treat long
H, 6.73; S, 14.52 exp %: C, 41.38; H, 6.73; S, 12.81.
range electrostatic interactions under periodic conditions with
3234 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
a direct space cutoff of 10 angstroms. The same cutoff was adenocarcinoma (NIH AIDS Research and Reference Reagent
used for van der Waals interactions. The pmemd.cuda module Program), and was grown in DMEM (Life Technologies, Spain)
from Amber12 was used for all the above described simulation supplemented with 10% FBS, 1% L-glutamine and antibiotic
steps.26 For the structural and clustering analyses (Rg, Rmax
,
cocktail. TZM.bl cell line (NIH AIDS Research and Reference
RDF, CV etc.) the last 5000 frames (which span the final 50 ns Reagent Program), which contains integrated copies of lucifer-
of the whole simulation) were used. Amber modules ptraj, ase gen under control of the HIV-1 promoter31 were cultured
cpptraj were used to accomplish these analyses. The UCSF in DMEM supplemented with 5% FBS, 1% L-glutamine and
Chimera software was used for final visualizations and for cal- antibiotic cocktail.
culations of molecular surfaces and volumes.27 Calculations
Virus production. Virus stocks were prepared by amplifica-
and visualizations of electrostatic potential around CO2 and tion of X4-HIV-1NL4.3 in MT-2 cell line (ATCC) and HIV-1NL(AD8)
SO3 groups were done using Adaptive Poisson–Boltzmann by transient transfection of pNL(AD8) in 293 T cell line (ATCC).
Solver as implemented in APBS28 plugin of VMD.29 Molecular Physical titers of all HIV viral stocks were evaluated by quanti-
sphericity was expressed using formula (1) where Ix, Iy, Iz are fication of HIV p24gag by ELISA kit (Innogenetics, Belgium).
principal moments of inertia which were calculated using sym-
metry tool in VMD.
Reagents. Reagents used as controls for inhibition of viral
replication were: Dextran (Sigma-Aldrich), a harmless molecule
that is used as a negative control for toxicity; and Suramin
(Sigma-Aldrich), an inhibitor of electrostatic attachment of
the V3 region of HIV-1 envelope glycoprotein gp120 to
galactosylceramide.32
Cell viability assays. Cell viability was determined by MTS
(Promega, Madison, WI, USA) assay following manufacturer’s
instructions. We included Dextran to evaluate its innocuous
effect in cell cultures, DMSO 10% (Sigma-Aldrich) as positive
control of cell death and PBS as negative control. Each experi-
ment was performed in triplicate.
Inhibition of HIV replication. VK2/E6E7 and HEC-1A epi-
thelial cells were treated with Suramin (10 µM) as positive con-
trols of HIV-inhibition, 2 or 14 (10 µM). Cells were then
infected for 3 h with 100 ng per 106 cells of X4-HIV-1NL4.3 or
R5-HIV-1NL(AD8) and further extensively washed. After 24 h,
supernatant of infected epithelial cells were collected and HIV
was measured using the p24gag ELISA kit.
P
IiIj
i=j
P
SPHER ¼
i; j ¼ x; y; z
ð1Þ
2
Ii
i
The SASA, SESA, SEV and SPHER values reported in Table 1
were evaluated only for each of the 5 most representative struc-
tures (see Fig. 5) and the reported values were then calculated
as the weighted average (k1X1 + k2X2 + k3X3 + k4X4 + k5X5)
where the weights k1, k2, k3, k4, k5 (statistical relevancy) are
listed in Fig. 5 for each dendrimer. The reported value of con-
formational variability CV was calculated as an average value
of the root mean square deviations (RMSD of atom coordi-
nates) of all possible pairs of conformations considering the
above mentioned set of 5000 conformations.
Biomedical methods
Cells. Blood samples were obtained from healthy anony-
Th1/Th2 cytokines profile. VK2/E6E7 were treated for 24 h
mous donors from the transfusion centers of Albacete and with 2 or 14 (10 µM) and levels of 10 cytokines (IL-8; IL-6; IL-4;
Madrid following national guidelines. Peripheral blood mono- TNF-α; IL-12p70) in cell-free supernatants were quantified by
nuclear cells (PBMC) were isolated on a Ficoll-Hypaque density using the DIAplex kit (Th1/Th2/inflammation, GenProbe).
gradient (Rafer, Spain) following the current procedures of Cytokine concentration (represented in pg mL−1) was calcu-
Spanish HIV HGM BioBank.30 PBMC and MT2 were cultured lated from a standard curve of the corresponding recombinant
in RPMI 1640 medium (Gibco, UK) supplemented with 10% human cytokine. All the cytometric analyses were performed
heat-inactivated FBS, 1% L-glutamine and antibiotic cocktail using FACS Gallios™ Beckman Coulter.
(125 mg mL−1 ampicillin, 125 mg mL−1 cloxacillin and 40 mg
CD-1 (ICR) mice vaginal irritation study. We analyzed the
mL−1 gentamicin; Sigma, St-Louis, MO, USA). PBMC were cul- effect of dendrimers in CD-1 strain mice (Crl: CD-1 (ICR) BR)
tured with 60 IU mL−1 of interleukin-2 (IL-2, Bachem, Switzer- (Charles River, France). Vaginal smears were made on the
land) and stimulated with phytohemagglutinin (PHA, 1 µg mice; mice in estrus cycle were selected and were inoculated
mL−1, Remel, Santa Fe, USA) for 48 h. VK2/E6E7 (ATCC) is an with 30 µL of each of the dendrimers in sterile PBS at concen-
epithelial cell line derived from normal vaginal mucosa tissue trations of 10 µM (low dose) and 100 µM (high dose) in the
and was grown in serum-free keratinocyte medium (Gibco) vaginas. At 2 and 24 h mice were sacrificed, the vaginas were
supplemented with recombinant human epidermal growth taken and fixed in 10% formaldehyde. Three mice per con-
factor (rEGF, 0.2 ng mL−1
,
Immunotools, Friesoythe, dition were used. The pathological examination was made by
Germany), bovine pituitary extract (BPE, 30 µg mL−1, Sigma- the company anaPath (Granada, Spain) accredited by ISO
Aldrich), 1% L-glutamine and antibiotics cocktail. HEC-1A 9001:2008. Epithelial lesions were evaluated considering the
(ATCC) is a epithelial cell line derived from a human endo- existence of hyperplasia or hyperkeratosis, ulceration and/or
metrial carcinoma (uterus mucosa carcinoma) and was grown inflammatory infiltrate in the epithelium. The values assigned
in McCoy’s 5A Medium Modified (Biochrom AG, Germany) were 0: no lesions; 1: minimum lesion; 2: slight injury; 3: mod-
supplemented with 10% FBS and antibiotics cocktail. HeLa erate injury; 4: severe injury. The procedures were performed
cell line is a human epithelial cell line derived from a cervical in accordance with Directive 2010/63/EU of the European
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3235

View Article Online
Paper
Organic & Biomolecular Chemistry
Parliament and Royal Decree 1201/2005, and were approved by
the ethics committee on animal experimentation of Hospital
General Universitario Gregorio Marañon.
7 (a) O. Rolland, C.-O. Turrin, A.-M. Caminade and
J.-P. Majoral, New J. Chem., 2009, 33, 1809; (b) R. Duncan
and L. Izzo, Adv. Drug Delivery Rev., 2005, 57, 2215–2237.
8 (a) L. Chonco, M. Pion, E. Vacas, B. Rasines, M. Maly,
M. J. Serramía, L. López-Fernández, J. de la Mata,
S. Alvarez, R. Gómez and M. A. Muñoz-Fernández, J. Con-
trolled Release, 2012, 161, 949–958; (b) J. L. Jiménez,
M. Pion, F. J. D. L. Mata, R. Gomez, E. Muñoz, M. Leal and
M. A. Muñoz-Fernandez, New J. Chem., 2012, 36, 299–309;
(c) B. Rasines, J. Sánchez-Nieves, M. Maiolo, M. Maly,
L. Chonco, J. L. Jiménez, M. Á. Muñoz-Fernández, F. J. de
la Mata and R. Gómez, Dalton Trans., 2012, 41, 12733–
12748.
9 (a) E. Vacas-Cordoba, E. Arnaiz, M. Relloso, C. Sanchez-
Torres, F. Garcia, L. Perez-Alvarez, R. Gomez, F. J. de la
Mata, M. Pion and M. A. Munoz-Fernandez, AIDS, 2013, 27,
1219–1229; (b) E. Arnaiz, E. Vacas-Cordoba, M. Gala,
M. Pion, R. Gómez, M. A. Muñoz-Fernandez and
F. J. de la Mata, J. Polym. Sci., Part A. Polym. Chem., 2014,
52, 1099–1112.
10 (a) A. Dondoni, Angew. Chem., Int. Ed., 2008, 47, 8995–8997;
(b) K. L. Killops, L. M. Campos and C. J. Hawker, J. Am.
Chem. Soc., 2008, 130, 5062–5064; (c) C. Rissing and
D. Y. Son, Organometallics, 2008, 27, 5394–5397;
(d) C. Rissing and D. Y. Son, Organometallics, 2009, 28,
3167–3172; (e) R. K. Roy and S. Ramakrishnan, J. Polym.
Sci., Part A: Polym. Chem., 2011, 49, 1735–1744.
11 (a) A. W. van der Made, P. W. N. M. van Leeuwen, J. C. De
Wilde and R. A. C. Brandes, Adv. Mater., 1993, 5, 466–468;
(b) A. W. van der Made and P. W. N. M. van Leeuwen,
J. Chem. Soc., Chem. Commun., 1992, 1400–1401;
(c) J. W. J. Knapen, A. W. van der Made, J. C. de Wilde,
P. N. M. W. Van Leeuwen, P. Wijkens, D. M. Grove and
G. van Koten, Nature, 1994, 372, 659–663; (d) D. Seyferth,
D. Y. Son, A. L. Rheingold and R. L. Ostrander, Organo-
metallics, 1994, 13, 2682–2690; (e) G. R. Newkome and
C. Shreiner, Chem. Rev., 2010, 110, 6338–6442.
Acknowledgements
This work has been supported by grants from CTQ2011-23245
(MINECO) and Consortium NANODENDMED ref S2011/
BMD-2351 (CM) to R. G. and UAH2011/EXP-037 (University of
Alcalá) to F. J. M. This work was supported by grants from
Fondos de Investigación Sanitaria ISCIII (INTRASALUD PI09/
02029, P509102669), Fundación Eugenio Rodríguez Pascual
Red Temática de Investigación Cooperativa Sanitaria ISCIII
(RETIC RD06/0006/0035 and RD12/0017/0037), Red Nacional
de Biobancos (RD09/0076/00103), INDISNET S-2011-BMD2332,
and FIPSE to M. A. M. F. Programa de Investigacion de la Con-
sejeria de Sanidad de la Comunidad de Madrid (PI11/00888)
to JLJF. Marek Maly gratefully acknowledges project GACR
13-06989S. This study was also supported by CIBER-BBN
financed by the Instituto de Salud Carlos III, with assistance
from the European Regional Development Fund.
References
1 WHO/UNAIDS Global Report 2013.
2 N. S. Padian, A. Buvé, J. Balkus, D. Serwadda and W. Cates Jr.,
Lancet, 2008, 372, 585–599.
3 (a) K. K. Ariën, V. Jespers and G. Vanham, Rev. Med. Virol.,
2011, 21, 110–133; (b) L. C. du Toit, V. Pillay and
Y. E. Choonara, Adv. Drug Delivery Rev., 2010, 62, 532–546;
(c) J. Nuttall, Drugs, 2010, 70, 1231–1243.
4 M. Lüscher-Mattli, Antivir. Chem. Chemother., 2000, 11,
249–259.
5 (a) B. Cutler and J. Justman, Lancet Infect. Dis., 2008, 8,
685–697; (b) V. Dzmitruk, D. Shcharbin, E. Pedziwiatr-Wer-
bicka and M. Bryszewska, Adv. Nanocompos. Technol., 2011,
361–374; (c) R. E. Haaland, T. Evans-Strickfaden, A. Holder, 12 J. Sánchez-Nieves, P. Ortega, M. Á. Muñoz-Fernández,
C. P. Pau, J. M. McNicholl, S. Chaikummao,
R. Gómez and F. J. de la Mata, Tetrahedron, 2010, 66, 9203–
W. Chonwattana and C. E. Hart, Antimicrob. Agents Che-
9213.
mother., 2012, 56, 3592–3596; (d) A. S. Ham, L. C. Rohan, 13 B. Korthals, M. C. Morant-Miñana, M. Schmid and
A. Boczar, L. Yang, K. W. Buckheit and R. W. Buckheit, S. Mecking, Macromolecules, 2010, 43, 8071–8078.
Pharm. Res., 2012, 29, 1897–1907; (e) S. Telwatte, K. Moore, 14 P. Eckstein, M. C. N. Jackson, N. Millman and A. J. Sobrero,
A. Johnson, D. Tyssen, J. Sterjovski, M. Aldunate, J. Reprod. Fertil., 1969, 20, 85–93.
P. R. Gorry, P. A. Ramsland, G. R. Lewis, J. R. A. Paull, 15 L. Van Damme, G. Ramjee, M. Alary, B. Vuylsteke,
S. Sonza and G. Tachedjian, Antiviral Res., 2011, 90, 195–
199; (f) K. J. Whaley, J. Hanes, R. Shattock, R. A. Cone and
D. R. Friend, Antiviral Res., 2010, 88, S55–S66.
V. Chandeying, H. Rees, P. Sirivongrangson, L. Mukenge-
Tshibaka, V. Ettiègne-Traoré, C. Uaheowitchai, S. S. Karim,
B. Mâsse, J. Perriëns and M. Laga, COL-1492 Study Group,
Lancet, 2002, 360, 1892–1892.
6 (a) P. Kissinger, C. R. Cohen, J. Brown, A.-B. Moscicki,
E. A. Bukusi, J. R. A. Paull, C. F. Price and S. Shiboski, PLoS 16 A. W. Bosman, H. M. Janssen and E. W. Meijer, Chem. Rev.,
One, 2011, 6, e16258; (b) C. F. Price, D. Tyssen, S. Sonza, 1999, 99, 1665–1688.
A. Davie, S. Evans, G. R. Lewis, S. Xia, T. Spelman, 17 D. A. Case, T. A. Darden, C. T. E. III, C. L. Simmerling,
P. Hodsman, T. R. Moench, A. Humberstone, J. R. A. Paull
and G. Tachedjian, PLoS One, 2011, 6, 1–12; (c) R. Rupp,
S. L. Rosenthal and L. R. Stanberry, Int. J. Nanomedicine,
2007, 2, 561–566.
J. Wang, R. E. Duke, R. Luo, R. C. Walker, W. Zhang,
K. F. Merz, B. Roberts, S. Hayik, A. Roitberg, G. Seabra,
J. Swails, A. W. Goetz, I. Kolossváry, K. F. Wong, F. Paesani,
J. Vanicek, R. M. Wolf, J. Liu, X. Wu, S. R. Brozell,
3236 | Org. Biomol. Chem., 2014, 12, 3222–3237
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
T. Steinbrecher, H. Gohlke, Q. Cai, X. Ye, J. Wang, 25 T. Darden, L. Perera, L. Li and L. Pedersen, Structure, 1999,
M.-J. Hsieh, G. Cui, D. R. Roe, D. H. Mathews, M. G. Seetin, 7, R55–R60.
R. Salomon-Ferrer, C. Sagui, V. Babin, T. Luchko, 26 A. W. Götz, M. J. Williamson, D. Xu, D. Poole, S. Le Grand
S. Gusarov, A. Kovalenko and P. A. Kollman, University of
California, San Francisco, 2012.
and R. C. Walker, J. Chem. Theory Comput., 2012, 8, 1542–
1555.
18 (a) A. Jakalian, D. B. Jack and C. I. Bayly, J. Comput. Chem., 27 E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch,
2002, 23, 1623–1641; (b) A. Jakalian, B. L. Bush, D. B. Jack
and C. I. Bayly, J. Comput. Chem., 2000, 21, 132–146.
D. M. Greenblatt, E. C. Meng and T. E. Ferrin, J. Comput.
Chem., 2004, 25, 1605–1612.
19 J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman and 28 N. A. Baker, D. Sept, S. Joseph, M. J. Holst and
D. A. Case, J. Comput. Chem., 2004, 25, 1157–1174.
20 (a) M. S. Gordon and M. W. Schmidt, Theory and Appli-
J. A. McCammon, Proc. Natl. Acad. Sci. U. S. A., 2001, 98,
10037–10041.
cations of Computational Chemistry: the first forty years, Else- 29 W. Humphrey, A. Dalki and K. Schulten, J. Mol. Graphics,
vier, Amsterdam, 2005; (b) M. W. Schmidt, K. K. Baldridge, 1996, 14, 33–38.
J. A. Boatz, S. T. Elbert, M. S. Gordon, J. H. Jensen, 30 (a) I. García-Merino, N. de las Cuevas, J. L. Jimenez,
S. Koseki, N. Matsunaga, K. A. Nguyen, S. J. Su,
T. L. Windus, M. Dupuis and J. A. Montgomery, J. Comput.
Chem., 1993, 14, 1347–1363.
J. Gallego, C. Gomez, C. Prieto, M. J. Serramia, R. Lorente,
M. A. Munoz-Fernandez and S. H. BioBank, Retrovirology,
2009, 6, 27–37; (b) I. García-Merino, N. de las Cuevas,
J. L. Jimenez, A. Garcia, J. Gallego, C. Gomez, D. Garcia and
M. A. Muñoz-Fernandez, AIDS Res. Hum. Retroviruses, 2010,
26, 241–244.
21 J.-H. Lii, A. Norman and L. Allinger, J. Comput. Chem.,
1991, 12, 186–199.
22 W. L. Jorgensen, J. Chandrasekhar, J. Madura and
M. L. Klein, J. Chem. Phys., 1983, 79, 926.
23 J.-P. Ryckaert, G. Ciccotti and H. J. C. Berendsen, J. Comput.
Phys., 1977, 23, 327–341.
24 X. Wu and B. R. Brooks, Chem. Phys. Lett., 2003, 381,
512–518.
31 E. J. Platt, M. Bilska, S. L. Kozak, D. Kabat and
D. C. Montefiori, J. Virol., 2009, 83, 8289–8292.
32 N. Yahi, J. M. Sabatier, P. Nickel, K. Mabrouk,
F. Gonzalezscarano and J. Fantini, J. Biol. Chem., 1994, 269,
24349–24353.
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 3222–3237 | 3237