Structure and organization of phospholipid/polysaccharide nanoparticles

Abstract

In recent years nanoparticles and microparticles composed of polymeric or lipid material have been proposed as drug carriers for improving the efficacy of encapsulated drugs. For the production of these systems different materials have been proposed, among them phospholipids and polysaccharides due to their biocompatibility, biodegradability, low cost and safety. We report here a morphological and structural investigation, performed using cryo-TEM, static light scattering and small angle neutron and x-ray scattering, on phospholipid/saccharide nanoparticles loaded with a lipophilic positively charged drug (tamoxifen citrate) used in breast cancer therapy. The lipid component was soybean lecithin; the saccharide one was chitosan that usually acts as an outer coating increasing vesicle stability. The microscopy and scattering data indicate the presence of two distinct nanoparticle families: uni-lamellar vesicles with average radius 90 Å and multi-lamellar vesicles with average radius 440 Å. In both families the inner core is occupied by the solvent. The presence of tamoxifen gives rise to a multi-lamellar structure of the lipid outer shell. It also induces a positive surface charge into the vesicles, repelling the positively charged chitosan molecules which therefore do not take part in nanoparticle formation.

Full text

IOP PUBLISHING JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 20 (2008) 104211 (8pp) doi:10.1088/0953-8984/20/10/104211
Structure and organization of
phospholipid/polysaccharide nanoparticles
Y Gerelli1,MTDiBari
1, A Deriu1,LCant`
u2,PColombo
3,
CComo
3, S Motta3,FSonvico
3and R May4
1Dipartimento di Fisica and CNISM, Universit`a degli Studi di Parma and CRS SOFT,
INFM-CNR, Italy
2Dipartimento di Chimica, Biochimica e Biotecnologie per la Medicina—LITA, Universit`adi
Milano, Italy
3Dipartimento Farmaceutico, Universit`a degli Studi di Parma, Italy
4Institut Laue-Langevin, Grenoble, France
E-mail: Antonio.Deriu@fis.unipr.it
Received 16 July 2007, in final form 1 November 2007
Published 19 February 2008
Online at stacks.iop.org/JPhysCM/20/104211
Abstract
In recent years nanoparticles and microparticles composed of polymeric or lipid material have
been proposed as drug carriers for improving the efficacy of encapsulated drugs. For the
production of these systems different materials have been proposed, among them phospholipids
and polysaccharides due to their biocompatibility, biodegradability, low cost and safety. We
report here a morphological and structural investigation, performed using cryo-TEM, static light
scattering and small angle neutron and x-ray scattering, on phospholipid/saccharide
nanoparticles loaded with a lipophilic positively charged drug (tamoxifen citrate) used in breast
cancer therapy. The lipid component was soybean lecithin; the saccharide one was chitosan that
usually acts as an outer coating increasing vesicle stability.
The microscopy and scattering data indicate the presence of two distinct nanoparticle
families: uni-lamellar vesicles with average radius 90 ˚
A and multi-lamellar vesicles with
average radius 440 ˚
A. In both families the inner core is occupied by the solvent. The presence
of tamoxifen gives rise to a multi-lamellar structure of the lipid outer shell. It also induces a
positive surface charge into the vesicles, repelling the positively charged chitosan molecules
which therefore do not take part in nanoparticle formation.
1. Introduction
Nanoscience has been recently indicated as one of the
most promising fields for technological innovation. Among
other applications, nanotechnologies are expected to provide
breakthrough improvements in healthcare and medicine. In
particular therapy, regenerative medicine, diagnosis and drug
delivery [1,2] are expected to benefit from new nanomaterials.
The need of new formulations for biotech drugs, as well
as the opportunity of revaluing old drugs through new
administration routes and innovative therapeutic applications,
have given great impulse to the development of pharmaceutical
nanotechnologies and microtechnologies. Indeed, their
application allows the modification of the biopharmaceutic
properties of drugs without changing their molecular structure;
it is therefore an interesting approach for the administration
of drugs presenting some hindering characteristics, such
as low water solubility, rapid metabolism or elimination,
narrow therapeutic index, toxic side effects. Several colloidal
drug delivery systems have been proposed, as liposomes,
micelles, nanoemulsions and nanoparticles. Polymeric and
lipidic nanosystems improve drug bioavailability, modify
pharmacokinetics or protect the encapsulated drug from
enzymatic degradation. Some liposomal drugs have been
approved or are under evaluation for clinical application,
in particular for cancer treatment [3,4]. Polymeric
nanoparticles built up from biocompatible and biodegradable
materials, such as poly(lactic acid) or poly(lactic-co-glycolic
acid) have been investigated [5]. Synthetic polymers
however are often insoluble in water and require the use
of organic solvents for the manufacture of colloidal drug
carriers. For this reason, great interest has been devoted
0953-8984/08/104211+08$30.00 ©2008 IOP Publishing Ltd Printed in the UK1

J. Phys.: Condens. Matter 20 (2008) 104211 Y Gerelli et al
Figure 1. Cryo-TEM image of nanoparticles containing tamoxifen (panel (a)). The presence of two families (uni- and multi-lamellar vesicles)
is evident, as well as a high overall polydispersity. In the multi-lamellar vesicles, the thickness tcof the region between bilayers is also
polydisperse. For comparison a cryo-TEM micrograph for nanoparticles loaded with a lipophilic non-charged drug (progesterone) is also
shown (panel (b)). Their structure is similar to that of empty nanoparticles (data to be published).
to colloidal preparations obtained from biomaterials like
polysaccharides and lipids that are considered biocompatible,
biodegradable and safe. Recently, we have started a
systematic investigation of the pharmaceutical [6]and
physical–chemical [7] characteristics of new types of
nanosystems obtained by self-assembling of phospholipids
(soybean lecithin) and chitosan. Lecithin is a natural
mix of phospholipids, obtained from soybean, composed
mainly of phosphatidylcholine and phosphatidylethanolamine,
that has been frequently used for liposome and micelle
preparation [8,9]. Chitosan, a cationic polysaccharide
[poly-β-(1,4) glucosamine, (C6H11O4N)n] derived from chitin
by deacetylation, is a versatile biomaterial with favourable
biochemical properties: no toxicity, biocompatibility,
biodegradability, bioadhesion [10,11]. It has been proposed
for a number of clinical and biomedical applications, ranging
form scaffolds for tissue engineering to the production of
colloidal drug carriers obtained by ionotropic gelation of the
polymer with tripolyphosphate [12,13]. Additionally, chitosan
has been investigated for stabilization of microemulsions in
which lecithin was one of the emulsifying agents [14]. In
most of the cases reported in the literature, it turns out that the
saccharide component acts as an outer coating that stabilizes
the lipid vesicle structure. The lecithin/chitosan nanoparticles
obtained with our preparation method are characterized by an
hollow core surrounded by a lipid bilayer with a thin chitosan
outer coating. This structure is observed in nanoparticles
not containing drugs (void nanovectors), as well as in those
loaded with a lipophilic non-charged drug (cryo-TEM image
reported in figure 1; small angle neutron scattering data to
be published). Here we present a structural investigation,
performed by cryo-TEM (cTEM), static light scattering (SLS)
and small angle scattering of neutrons (SANS) and x-rays
(SAXS) on lecithin/chitosan nanoparticles in the presence of
tamoxifen citrate, a lipophilic positively charged drug largely
used in breast cancer therapy. Tamoxifen is a drug with a very
low water solubility and a poor and erratic bioavailability after
oral administration and therefore is expected to benefit from
its formulation in biocompatible nanocarriers. Considering
the lipophilic nature of this drug is it possible to hypothesize
that its molecules will be mostly present into the lipid part
of the investigated nanocapsules. This is a common feature
of different kind of lipid–drug delivery systems [15–17]. A
detailed knowledge of the changes in the nanoparticle structure
due to the presence of the drug is important in order to optimize
the loading efficiency and the kinetics of drug release.
2. Experimental details
2.1. Sample preparation and preliminary characterization
The main component of our system is Lipoid S45 (a commer-
cial lecithin manufactured by Lipoid AG, Switzerland): a mix-
ture of lipids, phospholipids and 30% of fatty acids with an
overall negative charge. The second component is highly pu-
rified chitosan, a polysaccharide obtained by deacetylation of
chitin which is positively charged at acid pH. The third compo-
nent is tamoxifen citrate [18], a lipophilic drug used in breast
cancer therapy, positively charged at acid pH. We started the
preparation from two solutions: a first one composed by 200
mg of lecithin plus 60 mg of TAM all dissolved in 8 ml of
methanol, a second one made up by 10 mg of chitosan di-
luted in 92 ml of water (H2OorD
2O according to custom).
Self-assembled nanoparticles were obtained by rapid injection
(nozzle diameter 0.75 mm, injection rate 40 ml min−1), under
mechanical stirring, of the methanol solution into the aqueous
one. The self-assembling process gave rise to a suspension of
drug-loaded nanoparticles with concentration c=0.2% w/w
and pH =2.7[6]. It is worth noticing that TAM is already
admixed with lecithin in the methanol solution while chitosan
interacts whit lecithin only upon injection. For cTEM, SAXS
and SLS measurements, the solvent was a mixture of pure
H2O (92%) and methanol (8%). For the SANS experiments
two buffers were used: a perdeuterated one (D2O 92% and
D-methanol 8%), and a D2O/H2O mixture corresponding to
the index matching of lecithin (73.7% H2O, 18.3% D2O, 8%
H-methanol). In order to determine the mean hydrodynamic
radius and the Z-potential (connected to the charge per unit
surface area) a preliminary characterization was carried out
by dynamic light scattering. Samples containing lecithin and
lecithin/chitosan nanoparticles were also measured for com-
parison. The results are summarized in table 1. The particle
morphology was investigated by cTEM experiments. cTEM is
based on electron transmission microscopy on thin (≈5000 ˚
A)
vitrified aqueous films [19] obtained by cooling to liquid nitro-
gen temperature. The limiting factor, when using electrons as
probes, is the small difference between the electronic density
of amphiphilic molecules and that of the surrounding water; the
experimental resolution can not therefore be higher than ∼40–
50 ˚
A. This corresponds to a typical bilayer thickness, therefore
the internal bilayer structure can not be observed. Figure 1
shows cTEM micrographs for a typical suspension of TAM-
loaded nanoparticles (panel (a)), and nanoparticles loaded with
2

J. Phys.: Condens. Matter 20 (2008) 104211 Y Gerelli et al
Table 1. Comparison between average hydrodynamic radius (RH)
and surface potential of lecithin vesicles (LEC), chitosan/lecithin
nanoparticles (NCL) and those with TAM.
System RH
(˚
A)
Polydispersity
index (%)
Z-potential
(mV)
LEC 350 ±20 20 −48 ±2
NCL 1140 ±20 20 45 ±2
NCL+TAM 670 ±20 20 45 ±3
progesterone, a non-charged lipophilic drug (panel (b)). The
difference in structure is very evident: nanoparticles with pro-
gesterone are characterized by an hollow core surrounded by a
lipid bilayer with a thin chitosan outer coating; this structure
is very close to that observed in empty nanoparticles obtained
with the same preparation method. In nanoparticles with TAM
a multi-layered structure is clearly visible.
2.2. Static light scattering
The measurements were performed with a 677 nm incident
wavelength covering a range of momentum Qfrom 3.0×10−4
to 2.4×10−3˚
A−1. Sample suspensions for SLS were diluted to
a concentration c=0.04% (w/w), adding water and 0.0001%
of HCl in order to keep a constant pH. This was essential to
avoid multiple scattering contributions.
2.3. Small angle neutron scattering
SANS experiments were performed at the Institut Laue-
Langevin, ILL (Grenoble) using the D11 diffractometer. Three
different wavelength, λ, and sample-to-detector distance, D,
configurations were adopted (λ=5˚
A, D=5.5m;λ=
5˚
A, D=20.5m;λ=10 ˚
A, D=34 m) in order to
cover a Q-range of about two decades from 9.4×10−4to
0.1˚
A−1. Sample holders were standard quartz flat cells (1 mm
thick). Standard corrections, cell subtraction and normalization
to absolute scattering units were performed using ILL SANS
routines.
2.4. Small angle x-ray scattering
SAXS data were collected at the ID02 beamline at the
European Synchrotron radiation Facility, ESRF (Grenoble).
Scattered x-rays (λ=1˚
A) were collected using a charge-
coupled device (CCD). The sample-to-detector distance, D was
1.2 m; this configuration allows to explore a Q-range from
8×10−3to 4 ×10−1˚
A−1. Calibration was performed using
a silver behenate standard. Collected images were processed
and corrected with the ID02 software; accurate background
subtraction was performed by us at a later stage.
3. Data analysis
3.1. SLS and SANS
AsimplelowQGuinier analysis was not feasible for both
SANS and SLS data due to the high overall nanoparticle
polydispersity (see figure 1). We therefore decided to assume
Figure 2. SANS and SLS profiles for the investigated nanoparticle
solutions. SANS measurements in deuterated solvent (◦)arein
excellent agreement with the SLS ones (). SANS data at the
lecithin matching point are also reported ().
an analytical model for the scattering cross-section d(Q)/d
suitable for the whole Q-range explored. Nanoparticle
suspensions for the SLS experiments required a dilution higher
than that used in SANS measurements (≈50 times). In order
to properly match SLS and SANS data, we performed a
preliminary SANS measure at the SLS concentration (0.04%,
data not shown); these low concentration SANS data set can
be perfectly superimposed to the higher concentration ones by
re-scaling them according to their dilution ratio. This indicates
that the overall size and structure of the nanoparticles are not
appreciably influenced by dilution. The SANS curves were
expressed in absolute scattering units and the SLS data were
then normalized to the SANS ones obtaining a continuous
curve covering almost three decades in Q(figure 2). The
extremely good agreement of the SANS and SLS data in the
overlap region has to be remarked. For a dilute solution of
homogeneous and identical particles the scattered intensity
I(Q)is proportional to
d(Q)
d=n
Vρ 2
p|F(Q)|2(1)
where n/Vis the particle number density, ρ p≡ρp−ρsis
the contrast between the scattering length density, SLD, of the
particles (ρp) and of the solvent (ρs)[20]. In order to take into
account the nanoparticle polydispersity the form factor F(Q)
has been weighted according to a Schultz distribution [21,22],
D(σ, R), with polydispersity index σand centroid R0.The
differential scattering cross-section becomes therefore
d(Q)
d=φ
Vρ 2
p+∞
0
dRD(σ, R0)|F(Q,R)|2.(2)
Here the n/Vin equation (1) is replaced by φ/V,where
Vis the mean particle volume according to the adopted
distribution (for a Schultz distribution R3=(σ 2+1)
(2σ2+1)R3)andφis the particle volume fraction. The ratio
φ/Vis then equal to the average particle number density.
Guided by the morphological cTEM analysis, we adopted a
model that takes into account both uni-lamellar(ULV) vesicles,
3

J. Phys.: Condens. Matter 20 (2008) 104211 Y Gerelli et al
and multi-lamellar ones (MLV). The resulting differential
cross-section is then the sum of two contributions shown in
equation (3) and explained in detail in the following sections.
AQ-independent background is also added to account for the
incoherent scattering contribution from the hydrogens present
in the sample
d(Q)
d=p×φ
Vu∞
0
Du(σu,Ru
0)F2
u(Q,R)dR
+(1−p)φ
Vm∞
0dND
m(σm,N)
∞
0
dtD(σc,tc)F2
m(Q,R(N,t))+bkg.(3)
The subscripts u and m indicate the type of particles involved
in each scattering contribution (ULV and MLV, respectively)
and pis the relative population of ULV particles.
3.1.1. ULV particles. The model takes into account
a polydisperse core (mean radius Ru
0), described by a
Schultz distribution (Du(σu,Ru
0)), surrounded by a shell with
thickness tu
B[23]. The total particle radius, Ru
0+tu
B,has
therefore the same polydispersity index (σu) as the core. The
resulting form factor is
Fu(Q,R)=3ρu
BV(R+tu
B)j1(Q(R+tu
B))
Q(R+tu
B)
−V(R)j1(QR)
QR +3ρ0V(R)j1(QR)
QR (4)
where j1is the spherical Bessel function of the first kind. The
parameters of the ULV model are therefore the particle number
density p×φ
Vu, the polydispersity index σu, the inner radius Ru
0,
the lipid bilayer thickness tu
Band the two contrasts, ρ0and
ρ u
Bfor core and bilayer respectively.
3.1.2. MLV particles. In agreement with the morphological
cTEM results, we assumed the contrast profile schematized
in figure 3. It originates from an hollow core with radius
Rm
0and contrast ρ m
0, a first bilayer with thickness tm
Band
contrast ρ m
Bfollowed by a repetition of N−1 units, each
composed by a solvent layer (thickness tcand contrast ρm
0)
plus a lipid bilayer. The particle polydispersity is due to
the variable number of bilayers (described by the distribution
Dm(σm,N)) and to the polydispersity of the inter-layer
thickness, tc, described by a further Schultz distribution
D(σc,tc). The two above distributions (over Nand over
tc) play a totally different role. The first one affects the
external radius value and gives rise to a smearing of the SANS
curve mostly near the Guinier region. The latter is needed
to account for the width of the broad peak centred at Q≈
7×10−2˚
A−1(see figure 2) and does not affect the mean total
radius Rtot. The form factor for the MLV particles can then
be expressed as
Fm(Q,N,t)=4πρm
0Rm
0
0
dRR
2sin(QR)
QR
+4πρm
BRm
0+tm
B
Rm
0
dRR
2sin(QR)
QR
Figure 3. Radial SANS contrast profile adopted for multi-lamellar
vesicle population (MLV).
+4π
N−1

j=0ρ m
0Rc
j
R0
j
dRR
2sin(QR)
QR
+ρ m
BR0
j+1
Rc
j
dRR
2sin(QR)
QR (5)
where the integration limits are
R0
j=Rm
0+(j+1)tm
B+jt
⇒outer radius of the (j+1)th bilayer
Rc
j=Rm
0+(j+1)(tm
B+t)
⇒outer radius of the (j+1)th solvent layer.
Equation (5) can be explicitly expressed in terms of spherical
Bessel functions of the first kind, j1, recalling that
4πB
A
dxx
2sin(Qx)
Qx
=34
3πB3j1(QB)
QB −4
3πA3j1(QA)
QA .(6)
For this model the parameters are: two polydispersity indices,
σmand σc, the number of bilayers, N, the core radius Rm
0,the
thickness of the lipid layer and of the inter-layer (tm
Band tc
respectively) and the contrasts ρ m
0and ρ m
B. During the fit
runs they were not all kept free at the same time. As explained
in the following section, the results of the cTEM analysis and
of first test of the model enabled us to fix some of them and to
select well defined limits for other ones.
3.2. SAXS
SAXS experiments have been performed to exploit the
different x-ray contrasts between lipid, saccharide and drug
components of the nanoparticles and to extend the SANS upper
Q-limit (0.1) to 0.4˚
A−1. In this way one can also investigate
the internal structure of the lipid bilayers. In order to do this,
we write the bilayer form factor as
F(Q,d)=+d/2
−d/2
dxρ ( x)cos(Qx)(7)
where d=2(th+tt)is the bilayer thickness and ρ(x)
is the local electronic density across the bilayer. The x-
dependent contrast ρ (x)is described by a simplified head–
tail model (see figure 4)[24], implying two uniform scattering
4

J. Phys.: Condens. Matter 20 (2008) 104211 Y Gerelli et al
Figure 4. Electron density profile for the adopted head–tail model
(——) and of the solvent (······). The head and tail thicknesses, th
and tt, and the total bilayer thickness, d, are also indicated.
length densities for the tail and for the head respectively. A
more detailed description of ρ ( x)is not justified since the
composition of the adopted commercial lipid (Lipoid S45) is
partially unknown (see section 2.1). The SAXS intensity can
therefore be expressed as
I(Q)=A
Q2F2(Q,d,ρ(x)). (8)
Fitting parameters are: the scaling factor Aand the thickness
thand tt(see figure 4).TheroutineusedfortheSAXS,SANS
and SLS data is based upon an least square minimization from
the MINUIT package [25].
4. Results and discussion
4.1. Nanoparticle structure
The cTEM micrographs (panel (a)) in figure 1show the
presence of two clearly different nanoparticle populations.
Smaller nanoparticles (ULV, typical size about 20 nm) have a
vesicular structure with a large hollow core and a relatively thin
outer shell attributable to a single bilayer structure. From the
micrographs one can infer a relatively large size polydispersity.
Larger nanoparticles (MLV, typical size above 80 nm) have
a clear multi-layer structure. They are as well polydisperse
both in size and in number of layers. For MLV particles
we have adopted the multi-layer model described previously
and depicted in figure 3. Electronic contrast between water
and saccharides is ten times bigger than that between water
and lipids; cTEM can therefore provide information on the
arrangement of chitosan. The cTEM micrographs suggest that
chitosan is almost absent from the nanoparticles. This can be
understood in terms of the nanocapsule preparation method:
due to hydrophobic attraction, the lipophilic cationic TAM
and the amphiphilic anionic lecithin, already admixed into the
methanol solution, form mixed aggregates upon injection into
the water/chitosan solution. The resulting positive charge of
the resulting nanoparticles (see Z-potential values in table 1)
inhibits surface adhesion of the positively charged chitosan
molecules. Because of the acidity of the final solution (pH =
2.7), chitosan molecules in solution do not aggregate and keep
elongated chain shapes [26].
Figure 5. SANS data: (◦) nanoparticles in deuterated solvent,
() nanoparticles in lipoid-matched solvent, SLS (•) data. The
continuous lines are the fits on both data sets to equation (9). The
contributionsofULV(----)andMLV(
······) populations are also
shown for nanoparticles in deuterated solvent.
Figure 5shows the two SANS curves in absolute units for
nanoparticle suspensions with concentration c=0.2% (w/w)
in both perdeuterated and lipoid-matched solvents. In the same
figure the SLS data for a similar sample with c=0.04% are
reported after re-scaling to the SANS data. In the overlap
region the agreement between the two curves is remarkable.
It has also to be noted that a very small amount of much larger
particles is likely to be present in the solution, as suggested by
the discrepancy at very low Qbetween experimental SLS data
and the fit. Their presence cannot be appreciated by SANS,
restricted to higher Q, and they have therefore been neglected
in the adopted model.
The two curves shown in figure 5(SANS+SLS data for
the sample in deuterated solvent and SANS data for the one in a
solvent at the lipoid matching point) have been simultaneously
fitted adopting the two-family model described in section 3.1.1
(see equation (3)). Preliminary fits, performed varying all the
free parameters of the model, lead to the conclusion that the
scattering contrasts ρ0and ρ m
0are always very close to
zero (<10−14 ˚
A−2). This indicates that the core regions of
both families and the inter-layer region of MLV nanoparticles
are occupied by the solvent; these two contrasts were then set
to zero in the subsequent analysis. Under this zero-contrast
hypothesis the simultaneous fit of the data in figure 5implies a
system of two equations which differ only for the SLDs of the
two solvents:
d(Q)
d=16π2φ
Q2[p(ρu
B−ρs)2fu(Q)
+(1−p)(ρm
B−ρs)2fm(Q)]+bkg
d(Q)
d=16π2φ
Q2[p(ρu
B−ρs)2fu(Q)
+(1−p)(ρm
B−ρs)2fm(Q)]+bkg
(9)
where, apart from the contrast factors, fu(Q)and fm(Q)
represent the form factors of the ULV and MLV families,
5

J. Phys.: Condens. Matter 20 (2008) 104211 Y Gerelli et al
respectively,
fu(Q)=1
Vu∞
0
dRD
u(σu,Ru
0)R
0
drr sin(Qr)
2
fm(Q)=1
Vm∞
0dND
m(σm,N)∞
0dtD(σc,tc)

N−1

j=0Rm
0+j(t+tm
B)+tm
B
Rm
0+j(t+tm
B)
drr sin(Qr)
2.
It has to be noted that equation (9) contains the product
of the scattering contrast and the particle number density
for both ULVs and MLVs. Nevertheless, since we work in
absolute units and the SLD of both solventsare known, we can
unambiguously determine the values of pand of the contrasts
ρ m
Band ρ u
B. From the cTEM micrographs the thickness
parameters, tb,tm
band tccould be estimated; we could
therefore put narrow limits to their range of variation, namely
from 35 to 60 ˚
A. cTEM provided also an initial guess for the
relative population number, pand for the two polydispersity
indices σmand σu. The curves resulting from the overall fit
are shown as continuous lines in figure 5together with the
contributions of the ULV and MLV families. The results are
summarized in table 2: they confirm that both ULV and MLV
families are highly polydisperse. The values obtained for the
lipid bilayer thicknesses tu
Band tm
B(≈50 ˚
A) are in agreement
with those typical for extruded pure lipid vesicles [24,27,28].
The SLD values indicate that in ULV particles the lipid bilayer
is mostly composed by Lipoid (ρlipoid ≡ρs=(0.59 ±0.06)×
10−6˚
A−2as measured by index matching, ρtam =1.5×
10−6˚
A−2calculated),the amount of drug is zero within the
experimental accuracy (≈0.08) in this case. On the other hand,
from the SLD values obtained for MLV bilayers, it is possible
to conclude that the drug/lipid ratio (w/w) is 0.20 ±0.09.
Overall, MLV nanoparticles are characterized by an average
bilayer number N=5 with individualthickness tm
bseparated
by inter-layer regions, with thickness tc, that are occupied by
the solvent. This repetition leads to the broad peak observed in
figure 5at Q7×10−2˚
A−1, its broadening being mainly due
to the width of the D(σc,tc)distribution. Indeed the average
thickness of the multi-layer repeating unit ttot =tc+tm
b=
90 ±24 ˚
A corresponds to Q=2π/ttot =(7±2)×10−2˚
A−1.
The average bilayer number N, together with Rm
0,tm
b,and
tcdetermine the mean external nanoparticle radius Rtot
according to
Rtot=Rm
0+tm
B+[N−1] tc+tm
B=440 ˚
A.
The radius polydispersity index σtot =48% is determined by
the distribution in the number of bilayers N.
4.2. Drug encapsulation
On the basis of the obtained SLD values, we can confirm that
TAM is located within the lipid bilayers; this is not surprising
if we consider the lipophilic nature of this drug. As explained
before, TAM is mostly present into the MLV nanoparticles,
with a 0.2–1 TAM to lecithin ratio (w/w). This ratio can
Figure 6. SANS curves for nanoparticles in deuterated (◦)and
lecithin-matched solvents (). In both data sets a flat incoherent
background has been subtracted. The latter data are also shown after
re-scaling by a factor [ρ m
B(deuterated)/ρm
B(matched)]2=1500
().
Table 2. Fitting parameters obtained by equation (3) applied to
SANS and SLS data.
ULV MLV
Core radius R0(˚
A) 40 ±220±7
Polydispersity σ(%) ∼50 —
Inter-layer thickness tc(˚
A) — 37 ±1
Polydispersity σ(%) — ∼23
Bilayer thickness tB(˚
A) 50 ±553±2
Bilayer SLD ρB(×10−6˚
A−2)0.59 ±0.05 0.78 ±0.08
Population in % 38 62
be compared to the TAM/LEC one in the starting preparation
(see section 2.1): 0.34–1. Taking into account the volume
fraction of MLV nanoparticles, the effective loading efficiency
(encapsulated drug/drug present in the starting solution) is
(60 ±10)%. This is in agreement with the loading capacity
derived from the measured concentration of non-encapsulated
drug extracted from the final solution by ultracentrifugation
which is ≈60% [29]. This result is also confirmed by the
contrast values resulting from the fit of the data described in
section 4.1. After the flat incoherent background subtraction,
the data at the lipoid matching point, are significantly different
from zero only below Q0.025 ˚
A−1.Therethe
scattering signal is dominated by the MLV contribution; in
fact ρ m
B(matched)0.16 and ρ u
B(matched)0.040
(see table 2); the scattered intensity is proportional to (ρ)2
therefore the scattering signal arising from ULVs is negligible.
The ratio between the scattering signals from the sample in
deuterated solvent and from the one at the lipoid matching
point is then: [ρ m
B(deuterated)/ρm
B(matched)]2=1500.
This result is shown in figure 6which shows a good
superposition between data in deuterated solvent and the ones
in the lipoid-matched solvent, once they are re-scaled by the
above ratio. We have to remark that this comparison can be
performed only below Q=0.025 ˚
A−1i.e. in the region where
the signal from the lecithin-matched samples, after subtraction
of the flat incoherent background, is still appreciably = 0.
6

J. Phys.: Condens. Matter 20 (2008) 104211 Y Gerelli et al
Figure 7. SAXS (•) data together with the fit according to
equation (8)(——). The position expected for the low intensity peak
due to the multi-lamellar periodicity is indicated.
Table 3. Fitting parameters obtained by equation (7)appliedtoID02
data.
Zone
SLD
(×10−6˚
A−2)
Thickness
(˚
A)
Head 11 ±111±2a
Tail 9 ±113.0±0.5a
aTotal bilayer thickness
d=2×(th+tt)=48 ±5˚
A.
4.3. Bilayer thickness
The SAXS analysis, performed in the range 8 ×10−3–4 ×
10−1˚
A−1, enabled us to investigate in more detail the structure
of the lipid bilayer (see figure 7). In comparing the SANS
and SAXS data one has to note that the low intensity peak at
Q∼0.07 ˚
A−1observed in the SANS profile and due to the
multi-lamellar periodicity is almost not visible in the SAXS
curve. This fact is just accidental: the peak occurs close to
the first minimum of the bilayer form factor in the x-ray data
(Q∼0.05 ˚
A−1), while in the neutron case this minimum is
expected at higher Q-values outside the range we investigated.
Owing to the partially unspecified composition of Lipoid
S45, we have adopted a simplified description of the bilayer
form factor leading to equation (7). The parameter values
obtained from the fit are shown in table 3. The total bilayer
thickness, d48 ˚
A agrees well with the one obtained from
the SANS measurements and with literature values for the main
lipid components of our lecithin [24,27,28].
5. Conclusion
The adopted nanoparticle preparation methodology (self-
assembling following rapid injection of a lecithin+TAM
methanol solution into a diluted aqueous solution of chitosan)
gives rise to a polydisperse population characterized by two
distinct families: uni-lamellar vesicles (ULV) with average
radius 90 ˚
A and multi-lamellar vesicles (MLV) with average
radius 440 ˚
A. In both families the inner core is occupied by the
solvent. Tamoxifen is almost absent in ULV particles, while
it is present in the MLV ones which have a loading efficiency
around 60%. TAM molecules are located within each bilayer
as expected for lipophilic molecules.
The presence of tamoxifen changes structure of the
lipid outer shell from the uni-lamellar one observed in ULV
particles, to a multi-lamellar with an average number of
bilayers N=5 and a polydispersity index σm=48%.
The thickness of a single lipid bilayer is similar in ULVs and
in MLVs with a value close to that of pure DMPC [30], the
main lipid component of lecithin. It seems therefore not to be
affect by the presence of a charged drug and of a not negligible
amount of fatty acids.
In lipid-based nanoparticles, chitosan acts usually as
an outer coating that increases the vesicle stability. The
introduction of a charged drug like tamoxifen induces a
positive surface charge in the vesicles that repels the positively
charged chitosan molecules which therefore do not take
part in nanoparticle formation. Even in the absence of
chitosan, the loaded nanoparticles show a stability sufficient
for pharmaceutical applications [29]; this is probably due to
the multi-lamellar assembly induced by the charged drug.
The choice of a multi-component commercial lipid
mixture like Lipoid S45 was determined by the need of
studying systems close to the ones used in pharmacological
applications. The presence of 30% of fatty acids in Lipoid
S45 can in principle additionally complicate the interpretation;
we remark however that the matched solvent used in this
experiment refers precisely to Lipoid S45 therefore, as
concerns the scattering length densities, the presence of fatty
acids is to some extent taken into account. The present
analysis shows that careful morphological and structural
characterization by complementary techniques like cTEM,
SLS, SANS and SAXS can provided enough information to
describe the main structural features of the particles and the
effect of drug–particle interactions even when commercial
components are adopted. In the next future we plan to
extend this work towards nanoparticles in which commercial
lecithin is replaced by a better-controlled binary lipid mixture.
This should enable us to investigate in greater detail the
lipid–solvent–drug interactions, and to better tailor these self-
assembled systems for specific applications.
Acknowledgments
Many thanks to Mr S Barbieri for valuable help in sample
preparation and for helpful discussions. The authors are
grateful to Dr G¨oran Karlsson (University of Uppsala) for
the cryo-TEM measurements. We also thank Dr E DiCola
for her valuable technical assistance on the ID02 beamline
at ESRF. Financial support by the Ministry of University
and Research (PRIN founding framework) is gratefully
acknowledged. Y Gerelli also acknowledges NMI3 for
supporting his participation in ECNS 2007.
References
[1] Sonvico F, Dubernet C, Colombo P and Couvreur P 2005
Curr. Pharm. Des. 11 2091–105
[2] Hughes G A 2005 Nanomedicine 122–30
7

J. Phys.: Condens. Matter 20 (2008) 104211 Y Gerelli et al
[3] Torchilin V P 2005 Nat. Rev. Drug Discovery 4145–60
[4] Kim K Y 2007 Nanomedicine 3103–10
[5] Csaba N, Sanchez A and Alonso M J 2006 J. Control. Release
113 164–72
[6] Sonvico F, Cagnani A, Rossi A, Motta S, DiBari M T,
Cavatorta F, Alonso M J, Deriu A and Colombo P 2006 Int.
J. Pharm. 324 67–73
[7] Sonvico F, DiBari M T, Bove L, Deriu A, Cavatorta F and
Albanese G 2006 Physica B385 725–7
[8] Batzri S and Korn E 1973 Biochim. Biophys. Acta 298 1015–9
[9] Felt O, Buri P and Gurny R 1999 Drug Dev. Ind. Pharm. 24
979–93
[10] Illum L 1998 Pharm. Res. 15 1326
[11] Singla A W and Chawla M 2001 J. Pharm. Pharmacol.
53 1047
[12] Agnihotri S, Mallikarjuna N and Aminabhavi T 2004
J. Control. Release 100 5–28
[13] Janes K, Calvo P and Alonso M 2001 Adv. Drug Delivery Rev.
47 83–97
[14] Calvo P, Remunan-Lopez C, Vila-Jato J and Alonso M 1997
J. Appl. Polym. Sci. 63 125–32
[15] Reddy L H, Vivek K, Bakshi H and Murthy R S 2006 Pharm.
Dev. Technol. 11 167–77
[16] Bond`ı M L, Craparo E F, Giammona G, Cervello M,
Azzolina A, Diana P, Martorana A and Cirrincione G 2007
Drug Delivery 14 61–7
[17] Riske K E, Cho S H, Lee H B, Jeong S Y and Yuk S H 2003
J. Microencapsulation 20 489–96
[18] PubChem 2004 http://pubchem.ncbi.nlm.nih.gov/summary/
summary.cgi?sid=17389724&viewopt=Deposited
[19] Almgren M, Edwards K and Karlsson G 2000 Colloids Surf. A
174 3–21
[20] Sears V F 1992 Neutron News 326–37
[21] Lee J H, Agarwal V, Bose A, Payne G F and Raghavan S R
2006 Phys.Rev.Lett.96 048102–1
[22] Kiselev M A, Wartewig S, Janich M, Lesieur P, Kiselev A M,
Ollivon M and Neubert R 2003 Chem. Phys. Lipids
123 31–44
[23] Bartlett P and Ottewill R H 1992 J. Chem. Phys. 96 3306–18
[24] Riske K A, Amaral L Q and Lamy-Freund M T 2001 Biochim.
Biophys. Acta 1511 297–308
[25] Cern program library: http://cernlib.web.cern.ch/cernlib/
[26] Guo H F, Lin H L and Yu T L 2002 J. Macromol. Sci. A
39 837–52
[27] Kuˇcerka N, Pencer J, Sachs J N, Nagle J F and Katsaras J 2007
Langmuir 23 1292–9
[28] Ristori S, Oberdisse J, Grillo I, Donati A and Spalla O 2005
Biophys. J. 88 535–47
[29] Como C, Sonvico F, Rossi A, Corradini E and Colombo P 2006
AAPS J. S2 8W4163
[30] Zemlyanaya E V, Kiselev M A and Aswal V K 2004 Preprint
0411029v2 [physics.chemph]
8