Journal of Controlled Release 73 (2001) 255–267

www.elsevier.com / locate / jconrel

Chitosan nanoparticles as delivery systems for doxorubicin

Kevin A. Janes a , Marie P. Fresneau a , Ana Marazuela b , Angels Fabra b ,
´ Jose´ Alonso a , *
Marıa
a
Department of Pharmacy and Pharmaceutical Technology, School of Pharmacy, The University of Santiago de Compostela,
15706 Santiago de Compostela, Spain
b
`
Institut de Recerca Oncologica, Hospital Duran i Reynals, 08907 Barcelona, Spain
Received 17 August 2000; accepted 13 March 2001

Abstract

The aim of this paper was to evaluate the potential of chitosan nanoparticles as carriers for the anthracycline drug,
doxorubicin (DOX). The challenge was to entrap a cationic, hydrophilic molecule into nanoparticles formed by ionic
gelation of the positively charged polysaccharide chitosan. To achieve this objective, we attempted to mask the positive
charge of DOX by complexing it with the polyanion, dextran sulfate. This modification doubled DOX encapsulation
efficiency relative to controls and enabled real loadings up to 4.0 wt.% DOX. Separately, we investigated the possibility of
forming a complex between chitosan and DOX prior to the formation of the particles. Despite the low complexation
efficiency, no dissociation of the complex was observed upon formation of the nanoparticles. Fluorimetric analysis of the
drug released in vitro showed an initial release phase, the intensity of which was dependent on the association mode,
followed by a very slow release. The evaluation of the activity of DOX-loaded nanoparticles in cell cultures indicated that
those containing dextran sulfate were able to maintain cytostatic activity relative to free DOX, while DOX complexed to
chitosan before nanoparticle formation showed slightly decreased activity. Additionally, confocal studies showed that DOX
was not released in the cell culture medium but entered the cells while remaining associated to the nanoparticles. In
conclusion, these preliminary studies showed the feasibility of chitosan nanoparticles to entrap the basic drug DOX and to
deliver it into the cells in its active form.  2001 Elsevier Science B.V. All rights reserved.

Keywords: Chitosan; Dextran sulfate; Nanoparticles; Doxorubicin; Adriamycin

1. Introduction related to the development of multidrug resistance

Doxorubicin (DOX) and its bioactive derivatives
are among the most widely used anticancer drugs in
chemotherapy treatment [1]. However, problems
*Corresponding author. Tel.: 134-981-594-627; fax: 134-981-
547-148.
E-mail address: ffmjalon@usc.es (M.J. Alonso).
[2] and acute cardiotoxicity [3] have led researchers
to investigate alternative forms of administering
DOX for the treatment of cancer, with both prodrug
[4] and particulate [5] methods involved as active
fields of DOX research for the past two decades.
DOX microencapsulation has shown some appli-
cations for the controlled release of DOX over
extended periods of time [6]. Though relevant for

0168-3659 / 01 / $ – see front matter  2001 Elsevier Science B.V. All rights reserved.
PII: S0168-3659( 01 )00294-2
256 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
solid, accessible tumors, these particles are too large
to be endocytosed by most cells or circulate freely in
the bloodstream. Consequently, the association of
DOX to submicron carriers, such as liposomes [7],
nanoparticles [8], or micelles [9], has drawn greater
interest.
The majority of attempts to associate DOX to
nanoparticulate carriers have used anionic or neutral
polymers. Akiyoshi et al. [10] achieved DOX en-
capsulation in cholesterol-bearing pullulan hydrogel
nanoparticles, though loading levels were very low
(,0.1 wt.%) and cytotoxic effects of the nanoparti-
cles were lower than that of free DOX. Combining
prodrug and encapsulation strategies, Yoo et al. [11]
covalently linked DOX to the terminus of poly( D,L-
lactic-co-glycolic acid) (PLGA), then formed
nanoparticles with the conjugate by an emulsion-
solvent diffusion method. The group was able to
obtain considerable loadings (3.45 wt.%), achieve a
controlled release of DOX over nearly 1 month, and
maintain antiproliferative activity relative to free
DOX, though these results were possible only by
forming the covalent linkage between the polymer
and the drug.
The vast majority of work involving nanoparticu-
late DOX association, however, has been with
polyacrylates, exploiting charge interactions of the
polymer with the drug to achieve high association
efficiencies. Polymethacrylate nanoparticles with ad-
sorbed DOX were administered intravenously to
hepatoma patients and demonstrated prolonged plas-
ma levels, as well as reduced total clearance of DOX
relative to a control DOX solution [12]. DOX
associated to polyalkylcyanoacrylate nanoparticles
[13] have demonstrated reduced cardiotoxicity fol-
lowing intravenous administration in mice [14] as
well as increased cytoxicity against multidrug resis-
tant cell lines in vitro [15]. Later work showed that
coating of these particles with polysorbate 80 sig-
nificantly increased DOX accumulation in brain
tissue [16]. However, these DOX loaded particles
have demonstrated acute renal toxicity [17] as well
as decreased permeability of the drug across artificial
membranes with respect to free DOX [18].
An alternative approach would be to entrap DOX
into a positively charged carrier. Cell adhesion and
potentially cell uptake of such particles should be
favored due to their attraction to negatively charged
cell membranes, an attractive feature for the treat-
ment of solid tumors. From the perspective of
intravenous administration, positively charged par-
ticles would interact with different blood components
as compared to negatively charged particles. These
changes could potentially create a different biodistri-
bution and / or organ accumulation pattern following
intravenous administration. Additionally, a positively
charged system that would be expected to interact
with cells and / or membranes would be desirable for
testing alternative modes of administration of DOX,
i.e. mucosal administration.
We believed that an interesting candidate with
which to test these hypotheses was the cationic
polysaccharide, chitosan. This biopolymer has shown
favorable biocompatibility characteristics [19] as
well as the ability to increase membrane permeabili-
ty, both in vitro [20] and in vivo [21], and be
degraded by lysozyme in serum [22]. Consequently,
the aim of this paper was to encapsulate appreciable
quantities of DOX in chitosan nanoparticles made by
ionotropic gelation with sodium tripolyphosphate
(TPP) and test the effects of DOX encapsulation
and / or release on cytotoxic activity relative to free
DOX. To achieve this aim, we tried two approaches:
ionic bridging with a coincorporated polyanion and
polymer / drug complexation.
2. Experimental section
2.1. Materials
Chitosan hydrochloride salt, Protasan CL 110
(Mw .100 kDa), was purchased from Pronova Bio-
polymers (Oslo, Norway). Doxorubicin hydrochlo-
ride was obtained as a 2 mg / ml solution in 0.9%
(w / v) sodium chloride from Tedec-Meiji Farma
(Madrid, Spain). TPP, type B gelatin (75 bloom),
polyphosphoric acid, and dextran sulfate (Mw 510
kDa) were all purchased from Sigma-Aldrich S.A.
(Madrid, Spain). Phosphorylated glucomannan was a
´
gift from Industrial Farmaceutica Cantabria (Madrid,
Spain). Unless otherwise mentioned, water was
´
ultrapure grade (Milli-Q plus, Millipore Iberica,
Spain). All other chemicals were reagent grade.
 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
2.2. Spectrophotometric analysis
Reagent concentrations were fixed for all spectro-
scopic studies. Chitosan was maintained constant at
400 mg / ml and polyphosphoric acid, dextran sulfate,
and DOX concentrations all at 40 mg / ml. Spectra
were recorded from 350 to 600 nm using a UV-VIS
spectrophotometer (Model UV-1603, Shimadzu,
Columbia, MD) with a 2 nm slit width and a 1 cm
path length at intervals of 0.5 nm using water as the
baseline reference.
2.3. Formation of chitosan–DOX complex
DOX was added to a solution of 0.2% (w / v)
chitosan in water to a final concentration of 30%
(w / w) with respect to chitosan at pH 5.5. The
solution was left under magnetic stirring 24 h at
room temperature, dialyzed against distilled water
lowered to pH 5 with 1 N HCl for 36 h, and
lyophilized. To avoid photodegradation of DOX
during the complexation and purification, all pro-
cedures were performed in the absence of light.
DOX loading was calculated spectrophotometrically
at 487 nm (´ 51.74310 25 l / g cm).
2.4. Preparation of chitosan nanoparticles
Chitosan particles incorporating polyanions and
chitosan–DOX complexed particles were prepared as
described previously [23]. Briefly, chitosan or
chitosan–DOX complex was dissolved at 0.175%
(w / v) with 1% (v / v) acetic acid and then raised to
pH 4.7–4.8 with 10 N NaOH. For DOX-loaded
nanoparticles incorporating polyanions, 88–363 mg
DOX (10–30% loading) was first incubated for 20–
30 s with 88 mg of the polyanion. The mixture was
then added to a chitosan solution giving a final
chitosan concentration of 0.175% (w / v). To 500 ml
of this polymer solution, 100 ml of 0.291% (w / v)
TPP in water were added under magnetic stirring,
leading to the immediate formation of the nanoparti-
cles.
2.5. Physicochemical characterization of
nanoparticles
Size and zeta potential measurements were per-
257
formed by photon correlation spectroscopy and laser
Doppler anemometry, respectively, with a Zetasizer 
3000HS (Malvern Instruments, UK). For size mea-
surements, samples were diluted in water and mea-
sured for a minimum of 180 s. Raw data were
subsequently correlated to mean hydrodynamic size
by cumulants analysis. For zeta potential measure-
ments, samples were diluted in 0.1 mM KCl and
measured in automatic mode. All measurements
were performed in triplicate.
Particle morphology was examined by transmis-
sion electron microscopy (CM12 Philips, Eindhoven,
Netherlands). Samples were stained with 2% phos-
photungstic acid for 10 min, immobilized on copper
grids, and dried overnight for viewing.
2.6. Evaluation of DOX encapsulation
Encapsulation efficiency and nanoparticle yield of
the different formulations were determined by cen-
trifugation of the samples at 24,0003g for 30 min.
Pellets were incubated at 808C overnight and
weighed. Supernatant DOX concentrations were
calculated by fluorimetry (Model LS-50B, Perkin-
Elmer, Norwalk, CT) with a slit width of 5 nm and
excitation and emission wavelengths at 480 nm and
590 nm, respectively. Dilutions of samples and
calibration curves were performed in water, and all
measurements were performed in triplicate. Encapsu-
lation efficiency was calculated as follows:
DOX encapsulation
total DOX 2 free DOX
efficiency 5 ]]]]]]].
total DOX
2.7. Evaluation of in vitro DOX release
The nanoparticles were collected by centrifugation
at 90003g for 40 min on a glycerol bed. The pellets
were resuspended and incubated in 4 ml of 100 mM
acetate buffer (pH 4) or phosphate buffered saline
(pH 7.4) at 378C under light agitation. The quantity
of nanoparticles was adjusted to obtain a maximum
DOX concentration of 1 mg / ml. At varying time
points, supernatants were isolated by centrifugation
at 24,0003g for 30 min and measured by
fluorimetry as described earlier. Following superna-
tant extraction, pellets were discarded (destructive
258 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
sampling). Calibration curves were made with the
incubation medium, and all measurements were
performed in triplicate.
2.8. Fluorimetric analysis
DOX emission spectra were recorded in water
from 500 to 750 nm at a fixed excitation of 480 nm
with excitation and emission slit widths of 5 nm.
Spectra were read at a scan speed of 200 nm / min
and normalized with respect to peak emission.
2.9. In vitro cytostasis assays
Human melanoma A375 cells (ATCC, Rockville,
MD) and C26 murine colorectal carcinoma cells
were grown in Ham F-12 medium (GIBCO, Grand
Island, NY) supplemented with 10% fetal bovine
serum, sodium pyruvate, non-essential amino acids,
L-glutamine, and twofold vitamin solution (GIBCO).
The cultures were maintained in plastic flasks and
incubated in 5% CO 2 / 95% air at 378C in a
humidified incubator. The cell lines were examined
and found to be free of mycoplasma, as assayed by
the Gen-Probe Mycoplasma T.C. (Gen-Probe Inc.,
San Diego, CA).
The antiproliferative effect of DOX was analyzed
by the MTT method [24]. Cells from exponential
cultures were seeded onto 96-well tissue culture
plates (TPP, Switzerland) at a density of 5310 3
cells / well for a 0.36-cm 2 well (optimal seeding
density that avoids full confluency for the length of
the 4-day experiment). One day later, the cultures
were washed and incubated for 2 h with the different
samples at various dilutions in serum-free, Ham F-12
medium. The different preparations were adjusted to
maintain the same drug concentration, and experi-
ments were performed in parallel wells with increas-
ing DOX concentrations of 0.1 mg / ml to 100 mg / ml.
Following incubation, the cell monolayers were
washed five times with PBS and left 3 days in
complete media. After this time, 50 ml / well of PBS
containing 1 mg / ml MTT (tetrazolium salt, Sigma-
Aldrich S.A., Madrid) was added, and the plates
were incubated an additional 4 h. The intracellular
formazan crystals resulting from the reduction of the
tetrazolium salts present only in metabolically active
cells were solubilized with DMSO (Sigma-Aldrich).
The number of active cells was estimated by measur-
ing the absorbance at 540 nm (Titertek Multiscan,
ICN, Costa Mesa, CA). The percentage of cytostasis
was calculated as follows:
A2B
Cytostasis 5 ]]
A
where A is the absorbance of cells incubated with
culture medium and B is the absorbance of cells
incubated with the different nanoparticle formula-
tions. All samples were made in sextuplicate.
2.10. Confocal microscopy analysis
DOX accumulation in treated cells was localized
by confocal microscopy. Briefly, cells (Human
melanoma A375 cells (ATCC)) from exponential
cultures were grown on 1.13-cm 2 glass coverslips
¨
(Objekttrager, Menzol Glaser, Germany) at a density
of 5310 4 cells / coverslip. One day later, the cultures
were washed and treated with serum-free, Ham F-12
medium containing either DOX-loaded chitosan
nanoparticles incorporating dextran sulfate (30 min
to 24 h incubation) or free DOX (30 min incubation).
All incubations were carried out at 378C with an
equivalent final concentration of DOX (5 mg / ml).
Alternatively, cells were incubated using a two-
compartment Boyden chamber. Briefly, either DOX-
loaded chitosan nanoparticles containing dextran
sulfate or free DOX solutions were diluted in serum-
free culture medium and incorporated into the upper
compartment of polycarbonate transwell filters (0.4
mm pore diameter, Costar). The cells seeded on
coverslips for 24 h were placed in the lower com-
partment, thereby receiving the DOX solution fil-
tering from the upper compartment but excluding the
DOX associated to the nanoparticles (a control
experiment was performed demonstrating that the
particles did not cross the filter). Following incuba-
tion, cells were washed five times with PBS, and the
coverslips were mounted on slides. Fluorescence
observation was carried out with a confocal micro-
scope (TCS 4D, Leica Instruments) using an argon /
krypton laser (75 mW) at 488 nm for excitation and
an LP filter of 590 nm for DOX detection. Contrast
images were simultaneously observed using the
inverted microscope equipment with a PL Apo 633 /
 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267 259

1.4-oil objective (DMIRBE, Leitz). All experiments

were performed in sextuplicate.

2.11. Statistical analysis

The statistical significance of all results was

determined using the two-tailed Student’s t-test.

3. Results

The molecular structures of DOX and the com-

plexing agents, polyphosphoric acid and dextran
sulfate, are shown in Fig. 1. The protonable groups
in the DOX molecule were expected to interact with
the deprotonable groups of polyphosphoric acid and
dextran sulfate.
The interaction of DOX with different polyanions
and chitosan was first investigated spectrophotomet-
rically. As seen in Fig. 2A and B, the DOX peak at
480 nm was reduced by |53% upon incubation with
either polyphosphoric acid (Fig. 2A) or dextran
sulfate (Fig. 2B). Spectral changes in the DOX peak
were also observed for the samples, with polyphos-
phoric acid and dextran sulfate inducing red shifts of
9 and 14 nm for DOX solutions, respectively.
Subsequent addition of chitosan increased the in-
tensity of the 480 nm peak with polyphosphoric acid
and dextran sulfate to 110 and 83% of the original
DOX absorbance. Spectral peaks for both samples
returned to 480 nm. In control studies, no detectable
absorbance was noted for individual chitosan, poly-
phosphoric acid, or dextran sulfate solutions over the
chosen wavelength range at the concentrations tested
(data not shown). No significant differences in pH
were noted among any of the formulations shown Fig. 1. Chemical structure of: (A) DOX, (B) polyphosphoric acid,
(C) dextran sulfate. *Deprotonable functional group, **protonable
(DOX, DOX1polyanion, DOX1polyanion1
functional group.
chitosan).
A comparison of the DOX encapsulation efficien-
cies for different chitosan–TPP nanoparticle formu- polyanion concentration, forming visible agglomer-
lations is shown in Table 1. At 10% (w / w) polyanion ates. Incorporation of dextran sulfate increased DOX
with respect to chitosan, no significant differences in encapsulation efficiency approximately twofold with
DOX encapsulation efficiencies were observed with respect to the control formulation.
the coincorporation of gelatin, glucomannan, or The effect of initial DOX loading on encapsulation
polyphosphoric acid relative to the control formula- efficiency for chitosan–TPP nanoparticles incor-
tion, with all encapsulation efficiencies between 8 porating dextran sulfate can be seen in Table 2, with
and 13%. Also, the addition of polyphosphoric acid values ranging from 19 to 23%. Mean encapsulation
caused a destabilization of the suspension at this efficiencies decreased only slightly with increases in
260 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267

Table 2
Effect of DOX loading on encapsulation efficiency for chitosan
nanoparticles incorporating dextran sulfate (n53)
Theoretical Encapsulation
loading (%) efficiency (%)
5 22.561.2
10 21.962.5
20 19.361.8

particles showed a dense, spherical structure, which

was consistent with previous observations [25],
though size dispersion did appear to be greater with
the addition of dextran sulfate. Particle size and zeta
potential values for the unloaded nanoparticles were
259615 nm and 133.460.8 mV, respectively. DOX
loading (10% theoretical) did not significantly alter
these values (292642 nm and 133.260.1 mV).
Blank nanoparticle yield was 5161% which was
unchanged following 10% DOX loading (5262%),
resulting in a real DOX loading of 4 wt.% with
respect to the polymer.
A similar morphology was observed for DOX–
chitosan complexed nanoparticles (data not shown).
DOX complexation efficiency to chitosan was de-
termined to be 1.4% (w / w). The total amount of
Fig. 2. (A) UV-VIS spectra of: (———) DOX solution,
(— — —) DOX1polyphosphoric acid solution, (- - -) DOX1
DOX complexed to chitosan was incorporated in the
polyphosphoric acid1chitosan solution. (B) UV-VIS spectra of: nanoparticles structure, resulting in a real loading of
(———) DOX solution, (— — —) DOX1dextran sulfate solu- 0.43% (w / w). DOX–chitosan complexed nanoparti-
tion, (- - -) DOX1dextran sulfate1chitosan solution. cles possessed a mean nanoparticle size of 21363
nm and zeta potential of 133.760.6 mV. The low
DOX loading, and differences were only marginally degree of complexation was not expected to sig-
significant (P,0.1) between the highest and lowest nificantly alter the characteristics of the formulation,
loading levels. and the yield was assumed to be |100%, in accord-
TEM images of chitosan–TPP nanoparticles incor- ance with previously reported yields [23] for this
porating dextran sulfate are shown in Fig. 3. The chitosan–TPP formulation.
The in vitro release profile of chitosan–TPP
nanoparticles in acetate buffer (pH 4) is shown in
Table 1
Fig. 4. The particles incorporating dextran sulfate
Encapsulation efficiencies for different chitosan nanoparticle
formulations. All polyanions were incorporated at 10% (w / w) showed a burst release of 17% at 2 h, followed by an
with respect to chitosan. Theoretical DOX loading: 10% (n53) additional release of 4.5% over the next 2 days. For
Polyanion Encapsulation nanoparticles containing DOX complexed with
incorporated efficiency (%) chitosan, a small release of 4.5% was noted at 2 h,
with negligible DOX increases in release detected
No polyanion 9.162.2
Type B gelatin 8.461.5 over the proceeding 5 days.
Glucomannan 9.363.3 Fluorescence profiles of DOX emission spectra are
Polyphosphoric acid 12.264.1 shown in Fig. 5. Relative fluorescence was nearly
Dextran sulfate 21.962.5* identical for the DOX solution and DOX released
* P,0.01. from nanoparticles, with an average difference of
 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267 261

Fig. 3. TEM images of chitosan–TPP nanoparticles incorporating dextran sulfate.

0.260.8% between the normalized profiles. Con-
versely, encapsulated DOX exhibited an additional
emission band at 630 nm which was evident over the
range of 600–700 nm.
Fig. 6 shows the cytostasis of the different
chitosan formulations in vitro for the C26 cell line.
No significant differences were noted between the
DOX loaded chitosan–TPP nanoparticles incorporat-
ing dextran sulfate and the control DOX solution
over drug concentrations from 0.1 mg / ml to 100
mg / ml in any of the cell lines tested. Nanoparticles
made from DOX complexed to chitosan showed
slightly decreased cytostatic activity relative to the
DOX solution over this same concentration range.
However, for most concentrations assayed, the dif-
ferences were not statistically significant. The blank
nanoparticle suspension showed no significant cyto-
stasis. Very similar results were obtained using
human melanoma A375 cells (results not shown).
DOX localization of different nanoparticle / control
formulations in human melanoma A375 cells is
shown in Fig. 7. Within 30 min incubation, free
DOX was localized within the cell nucleus (Fig. 7A),

Fig. 4. In vitro release profile for: (– + –) DOX loaded chitosan– Fig. 5. Fluorescence emission spectra of: (———) DOX solution
TPP nanoparticles incorporating dextran sulfate (DOX loading: in water, (— — —) DOX encapsulated in chitosan–TPP
4%) and (– j –) DOX complexed nanoparticles (DOX loading: nanoparticles, (- - -) DOX released from chitosan–TPP nanoparti-
0.4%) (n53). cles.
262 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
Fig. 6. In vitro cytostasis in C60 cells for: (9) DOX loaded
chitosan nanoparticles incorporating dextran sulfate, ( ) chitosan
nanoparticles containing DOX complexed to chitosan, ( ) blank
chitosan nanoparticles, (h) DOX solution (n56).
whereas intracellular localization of DOX loaded
into chitosan nanoparticles containing dextran sulfate
could be visualized only after 24 h incubation. At
this point, no differences were seen in the intracellu-
lar localization of free DOX versus DOX previously
associated to chitosan nanoparticles (Fig. 7A and B).
To determine if DOX might be released from
chitosan nanoparticles before entering the cells, we
used a two compartment set-up where cells and DOX
samples were separated by a 0.4 mm polycarbonate
filter. As shown in Fig. 7C, no DOX was observed in
cells incubated for 24 h with DOX-loaded nanoparti-
cles in the upper compartment. A control mixture of
free DOX and blank nanoparticles in the upper
compartment showed an intracellular distribution
comparable to free DOX in solution (Fig. 7D).
4. Discussion
The major goal of this work was to develop a
chitosan nanoparticulate system as a novel, positive-
ly charged, colloidal carrier for DOX. The greatest
challenge was to encapsulate appreciable quantities
of DOX, overcoming the charge repulsion between
the cationic polymer (pKa 56.5) [26] and the pre-
dominantly positively charged anthracycline drug
(pKa 58.2) [27]. To begin, we selected a chitosan–
TPP nanoparticle formulation [23] which could
accommodate a large quantity of TPP (only |25%
molar excess of chitosan), thus minimizing the
polymer–drug repulsion by pairing a large fraction
of the positive amino groups with negatively charged
phosphates.
This formulation in itself did allow small quan-
tities of DOX to be retained within the particles
(Table 1), likely by physical entrapment. However,
the efficiency was quite low (9.1%), and since the
nanoparticles were still positively charged, we feared
that there would be a repulsion between chitosan and
DOX, owing to the positive charge of the drug. In an
attempt to remedy these concerns, we tested a variety
of polyanions which could simultaneously form a
strong ionic interaction between both chitosan and
DOX, increasing the encapsulation efficiency and
binding the molecule tightly to the particles. Type B
gelatin (pKa 54.7–5.0, as stated by the manufacturer)
and phosphorylated glucomannan (763% phos-
phorylation, as stated by the manufacturer) fit these
criteria as macromolecules, but no effect on the
spectroscopic profile of DOX, potentially indicating
intermolecular interactions, was observed (data not
shown). We concluded that neither gelatin nor gluco-
mannan possessed a sufficiently high negative charge
density to complex appreciable quantities of both
chitosan and DOX. These observations were cor-
roborated by failed attempts to augment DOX en-
capsulation with the incorporation of these molecules
into chitosan–TPP nanoparticles, shown in Table 1.
From this point, we shifted our attention to
polymers with higher anionic charge densities (at
least one negative charge per monomer). We first
tested polyphosphoric acid, which did appear to
interact strongly with DOX in solution. This inter-
action, however, was completely eradicated by the
addition of chitosan, as evidenced by the reversion of
the spectroscopic profile in Fig. 2A. The overall
absorption profile of this system was, in fact, higher
than the original DOX spectrum, perhaps due to an
increase in sample turbidity as polyphosphoric acid
formed particulated complexes with chitosan. Never-
theless, a mixture of chitosan and polyphosphoric
acid at the same concentrations but in the absence of
DOX showed no significant absorption over the
range tested (data not shown). Corroborating the
absorption spectra, the incorporation of polyphos-
phoric acid did not increase DOX encapsulation.
Moreover, the presence of this polymer in the
 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267 263

Fig. 7. Confocal images of (A) free DOX (30 min incubation), (B) DOX-loaded nanoparticles (overnight incubation), (C) DOX-loaded
nanoparticles in the upper compartment of a Boyden chamber (overnight incubation), and (D) DOX1blank nanoparticle mixture in the
upper compartment of a Boyden chamber (overnight incubation). All confocal studies were performed in A375 melanoma cells with 5
mg / ml equivalent DOX concentration. Magnifications: 3630 (A, B, D), 3400 (C).

chitosan solution led to the formation of aggregates,
rather than nanoparticles, upon addition of TPP.
Sulfonic acid groups have been shown to bind
DOX in considerable quantities when incorporated in
ion-exchange resins [28]. Additionally, dextran sul-
fate has been used successfully to augment the
encapsulation of DOX in albumin microspheres [29].
With these reports in mind, we decided to test the
feasibility of incorporating dextran sulfate into
chitosan nanoparticles for the encapsulation of DOX.
As could be anticipated, effects of the polyanion on
the UV-VIS spectrum of DOX were readily visible.
More importantly, however, the DOX–dextran sul-
fate complex appeared to only be partially disso-
ciated by the addition of chitosan (Fig. 2B), opening
the possibility of using the complex to draw more
DOX into the nanoparticles. Indeed, this entrapment
did occur, as the formulation incorporating dextran
264 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
sulfate was the only one to achieve encapsulation
efficiencies significantly above the control formula-
tion. DOX encapsulation was also visibly apparent
with these nanoparticles, which formed a dense red
pellet upon centrifugation.
The difference between DOX–dextran sulfate and
DOX–polyphosphoric acid interactions is intriguing,
in that one would expect coulombic forces to be
stronger with the acid, owing to its higher charge
density on a per weight basis. However, another
important consideration is the interaction of these
polyanions with chitosan. Using the same argument
of charge density, polyphosphoric acid should ex-
hibit far greater avidity to chitosan relative to dextran
sulfate. Hence, more DOX would be displaced from
polyphosphoric acid than from dextran sulfate upon
addition of chitosan.
Interestingly, DOX encapsulation efficiency in the
formulation containing dextran sulfate appeared
minimally dependent upon theoretical DOX loading
over the range of 5–20% (w / w). At 10% DOX
loading, there remains a 3.7-fold molar charge excess
of negatively-charged sulfonic acid groups relative to
DOX amino groups, so it is quite feasible that, due to
this excess, the saturation level of DOX association
is not reached. Under these conditions, therefore, it is
likely that the formation of DOX–dextran sulfate
complexes is favored by higher quantities of DOX
incubated, with real loading increasing almost linear-
ly with theoretical loading.
An entirely different approach was taken with
nanoparticles containing DOX complexed to
chitosan. As an amphoteric drug (protonable amino
group and deprotonable phenolic group), there con-
tinually exists an equilibrium between the positively
charged, negatively charged, neutral, and zwit-
terionic species of DOX (Fig. 1). Additionally, there
are other factors (hydrophobic / hydrophilic interac-
tions, resonance effects, etc.) which could allow
small quantities of DOX to complex with chitosan,
despite the overwhelming charge repulsion between
the two molecules, as has been noted previously
between DOX and positively charged transition
metal ions [30,31]. We tested the extent of this
association by incubating DOX and chitosan in
solution, dialyzing to remove non-associated DOX,
and lyophilizing to promote polymer–drug interac-
tions. We did not expect considerable association to
occur, but that the DOX which did associate would
be tightly bound to chitosan. Indeed, this phenom-
enon did take place. While there was only a minimal
yield for the complexation (0.43 wt.%), all of the
DOX that was complexed remained incorporated
within the nanoparticles. Therefore, it could be
induced that the entrapment efficiency of DOX
previously associated to chitosan was 100%. The
possibility that the high entrapment efficiency is
merely caused by low initial DOX loading was also
considered. As a control study, we prepared chitosan
nanoparticles with the same DOX initial 0.43 wt.%
loading, but adding DOX to the chitosan solution
prior to the nanoparticles formation. In this case the
encapsulation efficiency was far lower (23.061.0%)
than that observed for chitosan–DOX complexes.
In vitro release studies were performed in acetate
buffer (pH 4). We chose this medium because DOX
is maximally stable at pH values between 3 and 5
and also because, at higher pHs we encountered
problems of fluorescence quenching or interference
for the quantification of released DOX. Obviously, in
these experiments, we did not expect to predict the
release behavior of these particles in vivo but to
compare the formulations developed and to gain
some insight about the mechanism of release.
Nanoparticles incorporating dextran sulfate showed a
burst release of 17% at 2 h, followed by an addition-
al release of 4.5% over the next 2 days. This slow
release was quite distinct from the profiles obtained
from similar chitosan nanoparticles encapsulating
insulin, where 100% release was observed within 15
min [21]. Even less DOX was detected in PBS at 5
days (data not shown), probably due to DOX degra-
dation in the near neutral medium. The initial phase
of release is logically attributed to the DOX located
at the surface of the particles while the remainder of
the unreleased DOX was assumed to be well en-
trapped within the chitosan nanoparticles and tightly
associated to the chitosan molecules, probably as an
ionic complex with dextran sulfate. Therefore, the
degradation of chitosan would be required for ac-
complishing the release process. Unfortunately, di-
rect confirmation of this hypothesis by enzymatic
digestion of the nanoparticles was not possible since
chitosanase treatment of the nanoparticle suspension
also degraded DOX solutions and / or quenched
fluorescence in control studies. Nevertheless, the
effect of dextran sulfate on DOX release was appar-
ent, as chitosan nanoparticles without the polyanion
 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
showed over twice the burst effect after 2 h under the
same conditions (36.760.3%).
Nanoparticles containing DOX complexed with
chitosan displayed an even smaller release over the
same period, an observation that is easily explained
by the aforementioned interactions binding the drug
to the polymers. Indeed, since the interaction DOX–
chitosan seems to be quite stable, any drug released
would be a result of degradation of chitosan or by
release of DOX complexed on the particle surface.
Notwithstanding, spectral analysis of the DOX
released from chitosan nanoparticles incorporating
dextran sulfate showed that the released DOX was
fluorimetrically identical to that of the native DOX
solution, as seen in Fig. 5. This preservation of the
fluorescence signature supports the claim that DOX
structure is retained following encapsulation in
chitosan nanoparticles, though it is not a definitive
indication in itself. Conversely, the spectrum of
associated DOX showed an entirely new, longer
wavelength fluorescence band. This same band has
been previously reported for DOX in environments
with high dielectric constant [32], and suggests that
DOX is mostly associated with the nanoparticles (via
encapsulation, adsorption, or both), rather than
nanoprecipitated outside of the particles.
The retention of DOX bioactivity was best demon-
strated by the in vitro cytostasis assays, shown in
Fig. 6. Despite that only about one fifth of the
encapsulated DOX was released from chitosan–TPP
nanoparticles in vitro, this formulation was equally
able to slow tumor cell proliferation relative to DOX
solutions for the C26 and human melanoma A375
cell lines, indicating that DOX must maintain its
bioactivity within these nanoparticles. The same was
the case with the nanoparticles using DOX–chitosan
complexes, though the formulation did show lesser
cytostasis at certain drug concentrations relative to
the control DOX solution. This could be due to an
excessively tight interaction between the drug and
chitosan, which might impede its transit to the
nucleus. However, one cannot discard the possibility
that partial damage to the molecular structure of
DOX occurred during its complexation with
chitosan.
Given the limited DOX release exhibited by the
two formulations over this period, we hypothesized
that the cytotoxic action exhibited by these
nanoparticles would be due to endocytosis, rather
265
than to the release of free drug in the cell culture
medium. To confirm this hypothesis we investigated
the mechanism of in vitro cytotoxicity for DOX-
loaded chitosan nanoparticles, using human
melanoma A375 cells, via confocal microscopy.
Comparable fluorescence localization, visualized as
red, was seen for DOX-loaded nanoparticles relative
to free DOX (Fig. 7A and B), however, a sig-
nificantly longer incubation time was required for the
nanoparticles to display the intracellular fluorescence
signals. This suggests that these particles might enter
the cells, and that this internalization process occurs
over a much longer time than the diffusion of the
free drug. Indeed, from these observations it could be
inferred that, after short incubation times, the
nanoparticles are not sufficiently associated with the
cells and are thus easily removed in the subsequent
washes.
An alternative explanation of these results, how-
ever, could be that the longer incubation time is
needed simply to allow the particles to release an
amount of DOX in the culture medium comparable
to that of the control DOX solution. To exclude this
possibility, we used a two-compartment in vitro set-
up where the DOX-loaded nanoparticles were placed
in the donor compartment and the cells in a receptor
compartment, separated via a polycarbonate mem-
brane. After 24 h incubation, no DOX could be
detected in the cell culture compartment (Fig. 7C).
This led us to conclude that no significant amounts
of DOX were released from the nanoparticles into
the cell culture medium. This was also confirmed by
the fact that a control mixture of free DOX and blank
nanoparticles under the same experimental condi-
tions showed significant DOX accumulation (Fig.
7D), indicating that the drug can freely pass through
the membrane. The results of the in vitro release and
confocal studies combined strongly support our
hypothesis that DOX-loaded chitosan nanoparticles
are internalized by cells and degraded intracellularly
to release the drug. However, more thorough con-
focal studies are needed to ultimately prove this
mechanism.
5. Conclusion
In this paper, we describe the feasibility of using
chitosan nanoparticles as colloidal carriers for the
266 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
delivery of the small, cationic anthracycline drug,
doxorubicin (DOX). By incorporating the polyanion,
dextran sulfate, we were able to encapsulate con-
siderably high quantities of DOX considering the
inherent polymer–drug charge repulsion. These par-
ticles demonstrated a minimal burst release and
retained the cytotoxic activity of DOX in vitro.
Additionally, we showed that DOX can be com-
plexed to chitosan before particle formation by
incubation and separation of the complexes. This
approach appeared to bind drug even more tightly to
the particles, though the complexation reduced the
anti-proliferative activity of DOX. Confocal images
appeared to indicate that these particles enter the
cells via an endocytic mechanism and release DOX
intracellularly. Further experimentation, most notably
the testing of these formulations in vitro with DOX
resistant cell lines and in vivo, is necessary to clarify
the potential of chitosan nanoparticles for the deliv-
ery of DOX.
Acknowledgements
´
We would like to thank Monica ´
Hombreiro Perez
for her valuable help in the preparation of the
samples for the confocal studies. This work was
supported by a grant from the Spanish government
(SAF97-0169). K.A.J. and M.P.F. would like to
additionally thank the USIA Fulbright Association
and the UE Socrates Program, respectively, for
financial support.
References
[1] F. Arcamone, Doxorubicin, Academic Press, New York,
1981.
[2] L.J. Goldstein, H. Galski, A. Fojo, M. Willingham, S.L. Lai,
A. Gazdar, R. Pirker, A. Green, W. Crist, G.M. Brodeur, M.
Lieber, J. Cossman, M.M. Gottesman, I. Pastan, Expression
of multidrug resistance gene in human cancers, J. Natl.
Cancer Inst. 81 (1989) 116–124.
[3] D. Von Hoff, M. Rozencweig, M. Piccart, The cardiotoxicity
of anticancer agents, Semin. Oncol. 9 (1982) 23–33.
[4] W.A.R. Van Heeswijk, T. Stoffer, M.J.D. Eenick, W. Potman,
W.J.F. Van der Vijgh, J. Van der Poort, H.M. Pinedo, P.
Lelieveld, J. Feijen, Synthesis, characterization and anti-
tumor activity of macromolecular prodrugs of adriamycin,
in: J.M. Anderson, S.W. Kim (Eds.), Recent Advances in
Drug Delivery Systems, Plenum, New York, 1984, pp. 77–
100.
[5] D.J. Kerr, Microparticulate drug delivery systems as an
adjunct to cancer treatment, Cancer Drug Deliv. 4 (1987)
55–61.
[6] C. Jones, M.A. Burton, B.N. Gray, Albumin microspheres as
vehicles for the sustained and controlled release of doxorubi-
cin, J. Pharm. Pharmacol. 41 (1989) 813–816.
[7] A. Gabizon, R. Shiota, D. Papahadjopoulos, Phar-
macokinetics and tissue distribution of doxorubicin encapsu-
lated in stable liposomes with long circulation times, J. Natl.
Cancer Inst. 81 (1989) 1484–1488.
[8] P. Couvreur, L. Roblot-Treupel, M.F. Poupon, F. Brasseur, F.
Puisieux, Nanoparticles as microcarriers for anticancer drugs,
Adv. Drug Deliv. Rev. 5 (1990) 209–230.
[9] G.S. Kwon, M. Naito, Y. Masayuki, T. Okano, Y. Sakurai, K.
Kataoka, Physical entrapment of adriamycin in AB block
copolymer micelles, Pharm. Res. 12 (1995) 192–195.
[10] K. Akiyoshi, I. Taniguchi, H. Fukui, J. Sunamoto, Hydrogel
nanoparticle formed by self-assembly of hydrophobized
polysaccharide. Stabilization of adriamycin by complexation,
Eur. J. Pharm. Biopharm. 42 (1996) 286–290.
[11] H.S. Yoo, J.E. Oh, K.H. Lee, T.G. Park, Biodegradable
nanoparticles containing doxorubicin–PLGA conjugate for
sustained release, Pharm. Res. 16 (1999) 1114–1118.
[12] A. Rolland, Clinical phamacokinetics of doxorubicin in
hepatoma patients after a single intravenous injection of free
or nanoparticle-bound anthracycline, Int. J. Pharm. 54 (1989)
113–121.
[13] C. Verdun, P. Couvreur, H. Vranckx, V. Lenaerts, M. Roland,
Development of a nanoparticle controlled-release formula-
tion for human use, J. Control. Release 3 (1986) 205–210.
[14] P. Couvreur, B. Kante, L. Grislain, M. Roland, P. Speiser,
Toxicity of polyalkylcyanoacrylate nanoparticles. II. Doxoru-
bicin loaded nanoparticles, J. Pharm. Sci. 71 (1982) 790–
793.
[15] C. Cuvier, L. Roblot-Treupel, J.M. Millot, G. Lizard, S.
Chevillard, M. Manfait, P. Couvreur, M.F. Poupon, Doxoru-
bicin-loaded nanospheres bypass tumor cell multidrug resist-
ance, Biochem. Pharmacol. 44 (1992) 509–517.
[16] A.E. Gulyaev, E.G. Svetlana, I.N. Skidan, A.S. Antropov,
G.Y. Kivman, J. Kreuter, Significant transport of doxorubicin
into the brain with polysorbate 80-coated nanoparticles,
Pharm. Res. 16 (1999) 1564–1569.
[17] L. Manil, P. Couvreur, P. Mahieu, Acute renal toxicity of
doxorubicin (adriamycin)-loaded cyanoacrylate nanoparti-
cles, Pharm. Res. 12 (1995) 85–87.
[18] ¨
B. Muller, J. Kreuter, Enhanced transport of nanoparticle
associated drugs through natural and artificial membranes —
a general phenomena?, Int. J. Pharm. 178 (1999) 23–32.
[19] S. Hirano, H. Seino, Y. Akiyama, I. Nonaka, Biocompatibili-
ty of chitosan by oral and intravenous administration, Polym.
Eng. Sci. 59 (1989) 897–901.
[20] P. Artursson, T. Lindmark, S.S. Davis, L. Illum, Effect of
chitosan on the permeability of monolayers of intestinal
epithelial cells (Caco-2), Pharm. Res. 11 (1994) 1358–1361.
[21] ´
R. Fernandez-Urrusuno, ˜´ ´
P. Calvo, C. Remunan-Lopez, J.L.
Vila-Jato, M.J. Alonso, Enhancement of nasal absorption of
 K. A. Janes et al. / Journal of Controlled Release 73 (2001) 255 – 267
insulin using chitosan nanoparticles, Pharm. Res. 16 (1999)
1576–1581.
[22] ˚
K.M. Varum, M.M. Myhr, R.J.N. Hjerde, O. Smidsrød, In
vitro degradation rates of partially N-acetylated chitosans in
human serum, Carbohydr. Res. 299 (1997) 99–101.
[23] K.A. Janes, M.J. Alonso, Depolymerized chitosan nanoparti-
cles for protein delivery: preparation and characterization,
submitted.
[24] B.T. Mossman, In vitro approaches for determining mecha-
nisms of toxicity and carcinogenicity by asbestos in the
gastrointestinal and respiratory tracts, Environ. Health Per-
spect. 53 (1983) 155–161.
[25] ˜
P. Calvo, C. Remunan-Lopez,´ J.L. Vila-Jato, M.J. Alonso,
Novel hydrophilic chitosan-polyethylene oxide nanoparticles
as protein carriers, J. Appl. Polym. Sci. 63 (1997) 125–132.
[26] M.W. Anthonsen, O. Smidsrød, Hydrogen ion titration of
chitosans with varying degrees of N-acetylation by moni-
toring induced H-NMR chemical shifts, Carbohydr. Polym.
26 (1995) 303–305.
[27] R.J. Sturgeon, S.G. Schulman, Electronic absorption spectra
and protolytic equilibria of doxorubicin: direct spectrophoto-
267
metric determination of microconstants, J. Pharm. Sci. 66
(1977) 958–961.
[28] Y. Chen, M.A. Burton, J.P. Codde, S. Napoli, I.J. Martins, B.
Gray, Evaluation of ion-exchange microspheres as carriers
for the anticancer drug doxorubicin: in vitro studies, J.
Pharm. Pharmacol. 4 (1992) 211–215.
[29] Y. Chen, R.K. McCulloch, B.N. Gray, Synthesis of albumin–
dextran sulfate microspheres possessing favorable loading
and release characteristics for the anticancer drug doxorubi-
cin, J. Control. Release 31 (1994) 49–54.
[30] H. Beraldo, A. Garnier-Suillerot, L. Tosi, F. Lavelle,
Iron(III)–adriamycin and iron(III)–daunorubicin complexes:
physiochemical characteristics, interaction with DNA and
antitumor activity, Biochemistry 24 (1985) 284–289.
[31] M. Tachibana, M. Iwaizumi, S.T. Kubota, EPR studies of
copper(III) and cobalt(II) complexes of adriamycin, J. Inorg.
Biochem. 30 (1987) 133–140.
[32] K.K. Karukstis, E.H.Z. Thompson, J.A. Whiles, R.J. Rosen-
field, Deciphering the fluorescence signature of daunomycin
and doxorubicin, Biophys. Chem. 73 (1998) 249–263.