100

Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86
Contents lists available at ScienceDirect
Colloids and Surfaces A: Physicochemical and
Engineering Aspects
j our nal homepage: www. el sevi er . com/ l ocat e/ col sur f a
Enhancing encapsulation efficiency of highly water-soluble antibiotic in
poly(lactic-co-glycolic acid) nanoparticles: Modifications of standard
nanoparticle preparation methods
Wean Sin Cheow, Kunn Hadinoto
∗
School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637459, Singapore
a r t i c l e i n f o
Article history:
Received 30 June 2010
Received in revised form 11 August 2010
Accepted 30 August 2010
Available online 9 September 2010
Keywords:
Nanoparticle
Encapsulation
Nanoprecipitation
Emulsification
Polymer
Biofilm
a b s t r a c t
Effective encapsulations of drugs that are highly soluble in both water and organic solvents are notori-
ously difficult to achieve using standard nanoparticle preparation methods, such as nanoprecipitation
(NPC), single (ESE), and double (DESE) emulsification-solvent-evaporation methods. Modifications of the
standard preparation methods are therefore needed to enhance encapsulation efficiency of this group of
drugs. The present work investigates the feasibility of enhancing the encapsulation efficiency of highly
water and solvent-soluble levofloxacin, which is widely used in antibiotic therapy against pulmonary
biofilm infections, into poly(lactic-co-glycolic acid) (i.e. PLGA) nanoparticles. In addition, the nanoparti-
cles are evaluated in terms of their drug loading, size, production yield, and in-vitro drug release profile.
Lecithin inclusion into the aqueous phase in the ESE method results in two-fold improvements in the
encapsulation efficiency and drug loading (i.e. 23%and 2.3%, w/w, respectively) compared to the standard
method. Similar results are obtained when the water-miscibility level of the oil phase is increased in the
DESE method. Importantly, these nanoparticles possess size ≈200–300nm and biphasic extended drug
release profiles suitable for anti-biofilmtherapy. Lastly, the nanoparticles can be readily transformed into
inhalable solid dosage forms without jeopardizing their drug loadings and antibacterial activities.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
The use of nano-scale formulations inthe deliveryof therapeutic
agents through the pulmonary route has recently gained signifi-
cant interest [1]. While nanoparticles can also be delivered through
the oral and parenteral routes, inhalation of nanoparticles repre-
sents a more attractive delivery route, because of the avoidance
of the first-pass metabolism, ease of administration, and fast drug
absorptions due to the lungs’ high surface area and thin epithelial
layer. Specifically, polymer-based nanoparticle formulations have
been actively researched as inhaled therapeutic carriers owed to
the prospects of having (i) targeted drug deliveries by surface func-
tionalization of the polymer and (ii) sustained drug deliveries by
drug encapsulation in the polymer.
In this regard, antibiotic-loaded polymeric nanoparticles are
highly applicable for the treatment of persistent pulmonary infec-
tions, suchas cystic fibrosis (CF) andchronic obstructive pulmonary
diseases (COPD), a disease suffered by more than 50 million peo-
ple worldwide [2]. Bacterial biofilmcolonies present in the lungs of
CF and COPD patients are surrounded by thick stationary sputum
notorious for binding and deactivating antibiotics, thereby negat-
∗
Corresponding author. Tel.: +65 6514 8381; fax: +65 6794 7553.
E-mail address: [email protected] (K. Hadinoto).
ing the therapeutic effects of inhaled antibiotics. Encapsulation of
antibiotics into polymeric nanoparticles overcomes the problem
of antibiotic deactivations as it prevents interactions between the
antibiotics and the sputum contents.
Even though the encapsulation can also be achieved using
micro-scale polymeric carriers, nano-scale formulations are
favored because the average spacing of sputum mesh is
≈60–300nm[3], such that only nanoparticles can pass through the
sputumtoreachembeddedbiofilmcolonies. Byusingnanoparticles
as the carriers, the antibiotics canbe releasedina sustainedmanner
in the vicinity of biofilm colonies resulting in enhanced antibacte-
rial efficacies compared to that obtained from antibiotic solutions
[4,5]. In addition, by virtue of their small size, nanoparticles can
effectively evade the pulmonary phagocytic clearance resulting in
an increased antibiotic bioavailability in the lungs [6].
To this end, the nanoparticles are ideally delivered to the lungs
using the dry powder inhaler (DPI) platform. The adoption of the
DPI platformtranslates to portability, improved shelf-life, and high
delivery efficiency of the inhaled antibiotics, in contrast to the rel-
ative inconvenience (e.g. lengthy treatment time, non-portability)
and low delivery efficiency associated with the conventional neb-
ulization delivery platform [1]. However, nanoparticles in the
dry-powder form exhibit a strong tendency to severely aggregate
due totheir highsurface energy, thus resultinginnon-aerosolizable
powders. Furthermore, nanoparticles, with the exception of par-
0927-7757/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.colsurfa.2010.08.050
80 W.S. Cheow, K. Hadinoto / Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86
ticles with aerodynamic diameter (d
A
) <50nm, are predominantly
exhaled fromthe lung without depositing [6]. To effectively deliver
the nanoparticles to the lungs, they are typically transformed into
micro-scale dry-powder aggregates in the inhalable size range
(1≤d
A
≤5␮m) by spray-drying technique [7,8].
To be therapeutically effective, good aerosolization proper-
ties of the nanoparticle aggregates must be complemented by
having a potent antibiotic encapsulatedinto the nanoparticles. Lev-
ofloxacin, a fluoroquinolone antibiotic, is highly efficacious against
pulmonary biofilm cells as it does not have the propensity for spu-
tum binding, unlike aminoglycoside antibiotics [9,10]. However,
the high solubilities of levofloxacin in both organic solvents (e.g.
dichloromethane, ethanol, acetone) and water (≈200–300mg/mL
at pH≈2–8) render its encapsulation in polymeric nanoparticles
highly inefficient [11].
The low encapsulation efficiency of levofloxacin is particularly
true when the nanoparticles are prepared by standard methods,
such as nanoprecipitation and emulsification-solvent-evaporation.
For these methods, high drug solubility in either the organic phase
or the aqueous phase, but not both, is required to achieve high drug
encapsulation efficiency, as the drug must not leak out from its
initial dissolved phase upon precipitations of the polymer chains
to form the nanoparticles.
In this regard, several investigators have employed standard
nanoparticle preparation methods to encapsulate drugs that are
highly soluble in both water and organic solvents (∝10
2
mg/mL),
such as procaine hydrochloride, sodiumcromoglycate, and pheno-
barbital, into polymeric nanoparticles [12–14]. Not unexpectedly,
low drug encapsulation efficiency (<20%, w/w) and low drug load-
ings (≈1%, w/w) were reported. For this reason, modifications to
the standard method were subsequently performed, including (i)
incorporation of salt additives, (ii) pH variation, and (iii) using
alternative solvents, which led to significant improvements in the
encapsulation efficiency (≈40–50%, w/w) and consequently the
drug loading.
The main objective of the present work is to investigate the
effects of modifications of the standard nanoparticle preparation
methods on the encapsulation efficiency and drug loading of lev-
ofloxacin in poly(lactic-co-glycolic acid) nanoparticles (i.e. PLGA).
Threedifferent nanoparticlepreparationmethods (i.e. nanoprecipi-
tation, single, and double emulsification-solvent-evaporations) are
examined. PLGA is selected as the polymeric carrier because of its
well-establishedbiodegradability andbiocompatibility manifested
by its widespread applications as therapeutic carrier materials.
In addition to the encapsulation efficiency and drug loading, the
nanoparticles are evaluated in terms of their size, production yield,
and in-vitro drug release profile. In this regard, small nanoparticle
sizes (<300nm) are desirable tofacilitate aneffective sputumpene-
tration, whereas sustained antibiotic release profiles are favored as
they canpotentially reduce the dosing frequency andlower the risk
of systemic toxicity. Moreover, the feasibility of transforming the
levofloxacin-loaded PLGA nanoparticle suspension into inhalable
dry-powder nanoparticle aggregates by spray drying is investi-
gated. Lastly, the antibacterial efficacy of the solid dosage form
formulation is examined against Pseudomonas aeruginosa (P. aerug-
inosa) biofilm cells, which constitute a majority of the bacterial
pathogens present in the lungs of CF and COPD patients [15].
2. Materials and methods
2.1. Materials
PLGA polymer (Purasorb 5004A) is provided by PURAC Bioma-
terials (Netherlands). Chemicals usedinthe nanoparticle synthesis,
i.e. non-hydrochloride formof levofloxacin(LEV), dichloromethane
(DCM), acetone, ethyl acetate, Pluronic F-68, poly(vinyl alcohol)
(PVA, MW=13–23,000), polyethylene glycol (PEG, MW=4000),
tyloxapol, and chitosan, are purchased from Sigma–Aldrich (USA).
Drying adjuvants (i.e. lecithin and l-leucine) and biofilm growth
medium(i.e. Mueller HintonBrothor MHB) arealsopurchasedfrom
Sigma–Aldrich (USA). Phosphate buffer saline solution (PBS, pH
7.4) and dialysis membrane with a molecular weight cut-off size of
12,400g/mol used in the in-vitro drug release study are purchased
from 1st Base (Singapore) and Sigma–Aldrich (USA), respectively.
P. aeruginosa PAO1 strain is obtained from American Type Culture
Collection (ATCC, USA).
2.2. Methods
2.2.1. Nanoparticle preparations by standard methods
2.2.1.1. LEV-loaded PLGA nanoparticles by nanoprecipitation (NPC).
PLGA and LEV are dissolved in water-miscible acetone. Upon addi-
tion of the PLGA/LEV solution into water, the acetone rapidly
diffuses into the aqueous phase resulting in the formation of LEV-
loaded PLGA nanoparticles. Specifically, 8mg of LEV and 100mg of
PLGA are dissolved in 5mL of acetone after which the solution is
poured into 10mL aqueous solution of 0.1% (w/v) Pluronic F-68.
The resulting nanoparticle suspension is stirred overnight at room
temperature to evaporate off the acetone present in the aqueous
phase, which is followed by twice centrifugations at 14,000RPM
and two washing cycles to remove the non-encapsulated LEV and
the remaining acetone.
2.2.1.2. LEV-loaded PLGA nanoparticles by single emulsification-
solvent-evaporation (ESE). PLGA and LEV are dissolved in a
water-immiscible solvent (i.e. DCM). Upon addition of the
PLGA/LEV solution into water under ultrasonication, an oil-in-
water nano-emulsion is formed. The DCM is slowly evaporated
to transform the nano-emulsion into a nanoparticle suspension.
Specifically, 8mg of LEV and 80mg of PLGA are dissolved in 2mL
of DCM after which the solution is poured into 6mL aqueous solu-
tion of 1.0% (w/v) PVA. The resulting solution is emulsified for 60s
using a Vibra-Cell probe sonicator (VC 5040, Sonics and Materi-
als, USA). The nano-emulsion is next poured into 10mL aqueous
solution of 0.1% (w/v) PVA and the DCM is evaporated overnight at
room temperature. The resulting nanoparticle suspension is cen-
trifuged twice at 11,000RPM and two washing cycles to remove
the non-encapsulated LEV and DCM trace.
2.2.1.3. LEV-loaded PLGA nanoparticles by double-emulsification-
solvent-evaporation (DESE). PLGA and LEV are dissolved separately
in a water-immiscible solvent (i.e. DCM) and deionized water,
respectively. The two solutions are emulsified under ultrasonica-
tion to form w/o nano-emulsion, which is subsequently added to a
second aqueous phase to form w/o/w nano-emulsion. Specifically,
9mg of LEV is dissolved in 300␮L of deionized water and 90mg
of PLGA is dissolved in 3mL of DCM. The aqueous LEV solution
is emulsified in the PLGA organic solution by sonication for 60s.
The resulting nano-emulsion is poured into 12mL aqueous solu-
tion of 2.0% (w/v) PVA and is sonicated for 60s. The nano-emulsion
is stirred overnight at roomtemperature to evaporate off the DCM.
The resulting nanoparticle suspension is centrifuged and washed
following similar procedures described in the ESE method.
2.2.2. Nanoparticle physical characterizations
The drug encapsulation efficiency (i.e. % encapsulation) is deter-
mined from the mass ratio of the encapsulated drug to the drug
initially added. The encapsulated drug amount is determined by
subtracting the drug amount present in the supernatant after the
first centrifugation fromthe drug amount initially added. The drug
W.S. Cheow, K. Hadinoto / Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86 81
concentration in the supernatant is measured using UV–vis spec-
trophotometer (UV Mini-1240, Shimadzu, Japan) at absorbance
wavelength of 254nm. The drug loading is determined from the
mass ratio of the encapsulated drug to the total nanoparticles
produced. The nanoparticle production yield is determined from
the mass ratio of the nanoparticle produced to the PLGA initially
added. The nanoparticle size and zeta potential are measured by
photon correlation spectroscopy (PCS) using a Brookhaven 90Plus
Nanoparticle Size Analyzer (Brookhaven Instruments Corporation,
USA).
2.2.3. In-vitro drug release study
Thedrugreleaseratefromthenanoparticles is determinedusing
the dialysis-bagmethodunder a sinkcondition. The dissolutiontest
is conducted at 37
◦
C in a water bath under a continuous stirring.
The dialysis bag containing 2mL nanoparticle suspension is placed
in an opaque bottle to prevent photodegradation of LEV. 6mL PBS
is used as the release medium. At fixed intervals, 4mL of the release
medium is withdrawn and the amount of LEV released is mea-
sured by the UV–vis spectrophotometer. The withdrawn sample
is replaced with 4mL of fresh PBS to maintain the sink condition.
2.2.4. Spray-drying experiment
The LEV-loaded PLGA nanoparticle suspension is transformed
into dry-powder nanoparticle aggregates using a Büchi B-290
mini spray dryer (Büchi, Switzerland). Drying adjuvants are use
(i) to protect the structural integrity of the nanoparticles upon
drying and (ii) to ensure the nanoparticle aggregates can be recon-
stituted into primary nanoparticles upon their exposure to an
aqueous environment. The particle geometric size (d
G
) and shape
are characterized using a Scanning Electron Microscope (SEM)
model JSM-6700F (JEOL, USA), whereas the particle effective den-
sity (
e
) is determined using a tap densitometer (Quantachromme,
USA). The aerodynamic diameter (d
A
) is determined fromEq. (1) in
which
s
=1g/cm
3
.
d
A
= d
G


e

S
(1)
2.2.5. Antibacterial activity testing
P. aeruginosabiofilmcells arecultivatedaccordingtothemethod
of Harrison et al. [16]. Briefly, a bacterial cell suspension used as the
inoculum is adjusted to 1.0 McFarland standard and is diluted by
30-foldinMHBtoproduce1.0×10
7
colonyformingunits (CFU)/mL.
Next, 150␮L of the inoculum is transferred to each well of a 96-
well microplate and a 96-peg lid is fitted on top of the microplate
to provide a surface for the biofilm growth. After 24-h-incubation
in a shaking incubator at 150RPM and 37
◦
C, the peg lid is lifted
and the biofilm formed on the pegs is rinsed with PBS to remove
loosely attached planktonic cells.
Table 1
Physical characteristics of nanoparticles prepared by the standard methods.
Run Method Size (nm) % Encapsulation Drug loading (%)
A NPC 80 ± 30 15 0.7
B ESE 190 ± 50 16 1.1
C DESE 190 ± 80 10 1.0
To determine the antibacterial activity, biofilm cells formed on
the pegs are exposed to a 20␮L suspension of the dry-powder
nanoparticle aggregates and 180␮L of MHB at dosing equal to
7␮g/mL of LEV. Biofilmcells exposed to 20␮L of PBS and 180␮L of
MHB are used as the experimental control. After 6, 12, and 24h of
incubation at 37
◦
C, biofilm cells remaining on the pegs are rinsed
with PBS and transferred to a newmicroplate containing 200␮L of
PBS after a five-minute sonication. The recovered biofilm cells are
serially diluted 10-fold and are plated onto agar plates to be incu-
bated overnight at 37
◦
C to determine the viable CFU. The results
reported are based on two replicates. The antibacterial activity of
the nanoparticle suspension is also tested to examine, if any, the
effect of the dry-powder transformation.
3. Results and discussions
3.1. LEV-loaded PLGA nanoparticles prepared by standard
methods
3.1.1. Physical characteristics
A summary of the LEV-loaded PLGA nanoparticles produced
by using the three standard methods (i.e. NPC, ESE, and DESE in
Runs A–C, respectively) is presented in Table 1. Despite the similar
initial amount of PLGA used (80–100mg), nanoparticles prepared
by the NPC method exhibit a considerably smaller size (≈100nm)
compared to that of nanoparticles prepared by the ESE and DESE
methods (≈200nm). The sizes of the three nanoparticles produced
nevertheless remain within the range ideal for sputum penetra-
tions [3].
Particle size distribution of the nanoparticles produced by the
ESE method is presented in Fig. 1A, where a highly uniform distri-
bution is observed. Similarly, uniform size distributions are also
observed for the nanoparticles produced by the NPC and DESE
methods. SEMimage of the nanoparticles inFig. 1Bindicates spher-
ical nanoparticles are produced and verifies their highly uniform
size distribution. Zeta potentials of the three nanoparticles are in
the range of (−) 10–30mV indicating colloidally stable nanoparti-
cles.
Inadditionto the smaller size, the NPCmethodresults ina lower
production yield (≈65%) than the ESE and DESE methods (≈90%),
which can be attributed to greater particle losses during the cen-
trifugationandwashing steps as a result of the smaller particle size.
Fig. 1. (A) Particle size distribution of LEV-loaded PLGA nanoparticles indicating a highly monodisperse distribution and (B) their SEM image for verification.
82 W.S. Cheow, K. Hadinoto / Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86
Due to the high aqueous solubility of LEV, % encapsulations of LEV
in nanoparticles prepared by the standard methods are expectably
low(≤16%, w/w) as LEV diffuses out into the aqueous phase during
the nanoparticle preparation steps. Even for the DESE method, the
% encapsulation is also low, despite the double emulsification, due
to the high LEVsolubility in the oil phase, which leads to the escape
of LEV from the inner to the outer aqueous phase.
Consequently, drug loadings of the nanoparticles are also low
(≤1.1%, w/w), where the NPC method results in the lowest drug
loading at 0.70% (w/w). Experimental uncertainties in the % encap-
sulation and drug loading measurements are between 4–8% and
5%, respectively, based on three replicates. The extremely lowdrug
loading of the nanoparticles produced by the NPC method is caused
byLEVdiffuses out intotheaqueous phase, together withthewater-
miscible acetone, upon PLGA precipitations, which results in LEV
being predominantly adsorbed onto the nanoparticle surface. Con-
sequently, LEV is washed off more easily during the centrifugation
step, compared to when LEV is encapsulated within the polymer
matrix, resulting in a low drug loading. The dominant presence
of LEV on the surface of the nanoparticles produced by the NPC
method is manifested in its fast release profile as presented in the
next section.
The use of water-immiscible DCM in the ESE and DESE meth-
ods prevents LEV from being predominantly adsorbed onto the
nanoparticle surface as reflected in the higher drug loading. In
this regard, the NPC and ESE methods are commonly employed to
encapsulate hydrophobic drugs into nanoparticles, where the lack
of drug solubility in the aqueous phase ensures high encapsulation
efficiency. Thus, when highly water-soluble drugs, such as LEV, are
encapsulated using the NPC and ESE methods, % encapsulations
achieved are expectably low.
The DESE method, on the other hand, is typically employed
to encapsulate water-soluble, but not solvent-soluble, drugs, as
the inner aqueous phase acts as a reservoir in which the drug is
dissolved, whereas the oil phase serves as a dissolution barrier pre-
venting the drug leakage fromthe inner tothe outer aqueous phase.
For LEV, which is highly soluble in both the aqueous and oil/organic
phases, standard NPC, ESE, and DESE methods have been shown
to be incapable of producing high % encapsulations. Hence, mod-
ifications to the standard methods are needed to enhance the %
encapsulation of LEV.
3.1.2. In-vitro drug release
In-vitro release profiles of LEV fromthe nanoparticles produced
in Runs A–C are shown in Fig. 2. Experimental uncertainties in the %
drug released, based on three replicates, are approximately 5%. As
previously mentioned, nanoparticles prepared by the NPC method
0
20
40
60
80
100
0 3 6 9 12
%

d
r
u
g

r
e
l
e
a
s
e
d
Time (hours)
Run A
Run B
Run C
Fig. 2. In-vitro drug release profiles from nanoparticles prepared by the standard
methods.
exhibit a monophasic burst release profile as a result of LEV being
predominantly present on the nanoparticle surface. Specifically,
≈80% of the loaded LEV is released within the first hour and the
entire loaded drug is completely released within 3h.
In contrast, nanoparticles prepared by the ESE and DESE meth-
ods display biphasic release profiles, where a high burst release
in the first hour (≈40–50%) is observed due to the presence of
LEV on the nanoparticle surface. The burst release is followed by
a slower release of the remaining LEV contributed by LEV that is
encapsulated within the PLGA matrix. Not all of the loaded LEV,
however, only 80–90%, is released after 24h. Nevertheless, the
extended biphasic LEV release profiles produced by the ESE and
DESE methods are more desirable than the burst release profile as
they can ensure that LEV concentrations are maintained above the
level needed to inhibit the biofilmgrowth in-between dosings [17],
as well as reducing the risk of systemic toxicity.
Thus, nanoparticles prepared by the ESE and DESE methods are
deemed superior compared to those prepared by the NPC method,
because of (i) the higher production yield and drug loading and (ii)
the sustained release profiles. For this reason, subsequent modifi-
cation attempts to improve the % encapsulation and drug loading
are focused on the ESE and DESE methods. A summary of modifica-
tions that are attempted and their corresponding results (i.e. size,
% encapsulation, and drug loading) is presented in Tables 2–4. The
modifications reported in the present work are not aimed towards
Table 2
Physical characteristics of nanoparticles prepared by the modified ESE methods.
Run Modified phase ESE modification Size (nm) % Encapsulation Drug loading (%)
B1
Organic
50% (v/v) DCM:ethyl acetate
≈200 ≈15 ≈1.0
B2 50% (v/v) DCM:acetone
B3 Aqueous 50% (w/w) PVA:lecithin 240±50 23 2.3
Table 3
Physical characteristics of nanoparticles prepared by the modified single-phase DESE methods.
Run Modified phase DESE modification Size (nm) % Encapsulation Drug loading (%)
C1
Inner aqueous
PBS core 190 ± 70 6 0.6
C2 0.1% PEG (w/v) 200 ± 70 8 0.8
C3 0.1% PVA (w/v) 360 ± 120 22 2.1
C4
Oil
50% (v/v) DCM:acetone 110 ± 40 22 2.4
C5 50% (v/v) DCM:ethyl acetate 140 ± 50 20 2.1
W.S. Cheow, K. Hadinoto / Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86 83
Table 4
Physical characteristics of nanoparticles prepared by the modified multiple-phase DESE methods.
Run Inner aqueous phase Oil phase Outer aqueous phase Size (nm) % Encapsulation Drug loading (%)
C6
0.1% (w/v) PVA
DCM:acetone
Standard
170 ± 80 6 0.5
C7 DCM:ethyl acetate 180 ± 80 10 1.0
C8
Standard
50% (w/w) PEG 290 ± 110 11 1.0
C9 50% (w/w) chitosan 720 ± 290 8 0.8
C10 50% (w/w) tyloxapol 720 ± 340 4 0.4
the more usual parameters, suchas stirring speed, pHchange, start-
ing polymer and drug concentrations, or the volume ratio of the
aqueous to organic phase, because these parameters have been
found in our preliminary study to be non-influential towards the
LEV encapsulation process.
3.2. LEV-loaded PLGA nanoparticles prepared by the modified ESE
methods
Three modifications to the standard ESE method are performed,
which are (1, 2) using alternative solvents composed of 50% (v/v)
mixture of DCM and either acetone or ethyl acetate in Runs B1
and B2, and (3) addition of 50% (w/w) lecithin into the aqueous
PVA solution in Run B3. For the effect of using alternative solvents,
precipitations of PLGAchains toformsolidnanoparticles are slower
whencompletelywater-immisciblesolvents, suchas DCM, areused
as it takes longer for the polymer molecules to be exposed to the
aqueous phaseuponemulsifications. Theslower PLGAprecipitation
rate provides an opening for the highly water-soluble LEV to leak
out of the organic phase into the aqueous phase resulting inthe low
% encapsulation.
For this reason, varying the water-miscibility level of the solvent
is thought to potentially have an impact on the % encapsulation.
The solvent composition is therefore varied from 100% (v/v) DCM
in the standard ESE method to 50% (v/v) of DCM and ethyl acetate
in Run B1, and to 50% (v/v) of DCM and acetone in Run B2. The
water-miscibility levels of these mixtures increase in the following
order: 100% (v/v) DCM<50% (v/v) DCM/ethyl acetate <50% (v/v)
DCM/acetone, which is attributed to the partial-miscibility of ethyl
acetate and full-miscibility of acetone in water.
Completely replacing DCM with a highly water-miscible sol-
vent, such as acetone, however, is not pursued as it denotes a shift
from the ESE method to the NPC method, which has been found to
produce less-than-ideal nanoparticles. The results of Runs B1 and
B2 in Table 2 nonetheless suggest that the effects of the water-
miscibility level of the solvent on the % encapsulation and drug
loading are minimal, where their respective values remain in the
15% (w/w) and 1.0% (w/w) ranges. Hence, instead of attempting
to reduce the time-scale of the drug leak-out, the next modifica-
tion attempt aims to incorporate physicochemical barriers into the
nanoparticles, so that the drug leak-out can be minimized.
In this regard, adding phospholipid lecithin into the aqueous
phase during PLGA nanoparticle preparation has been shown to
result in adsorptions of the amphiphilic lecithin onto the nanopar-
ticle surface through hydrophobic interactions [18]. As LEV is
lipophilic [19], it interacts with the hydrophobic tails of the lecithin
similar to the PLGA–lecithin interactions, such that incorporating
lecithin onto the nanoparticle surface can potentially enhance the
% encapsulation by entrapping LEV in the phospholipid layer.
Indeed, two-foldincreases inthe %encapsulationanddrug load-
ing to 23% and 2.3% (w/w), respectively, are observed in Run B3,
where lecithin is added to the 1.0% (w/v) aqueous PVA solution at
50% (w/w) PVA to lecithin ratio. Moreover, in spite of the higher
drug loading, the lecithin inclusion slows down the LEV release
rate due to the hydrophobic interactions as shown in Fig. 3. As a
result, the PLGA–lecithin nanoparticles exhibit the ideal extended
biphasic release profile.
3.3. LEV-loaded PLGA nanoparticles prepared by the modified
DESE methods
For the DESE method, the modifications are first performed
one-at-a-time on either the inner aqueous phase or the oil phase
of the w/o/w nano-emulsion to identify formulation parameters
(e.g. solvent, surfactant) that have the most significant impacts
on the % encapsulation. Subsequently, simultaneous modifications
involving the significant parameters are carried out in both the
inner/outer aqueous and oil phases.
3.3.1. Single-phase modifications
First, the effects of modifying the inner aqueous phase on the
% encapsulation and drug loading are investigated by substituting
deionized water used in the standard DESE method with 100% (v/v)
PBS in Run C1, and 0.1% (w/v) aqueous PEG and PVA solutions in
Runs C2 and C3, respectively. The oil and outer aqueous phases
remain unchanged from the standard DESE method.
First, owed to the higher observed solubility of LEV in PBS
than that in deionized water, the use of PBS as the inner aque-
ous phase is thought to be able to increase the % encapsulation,
where LEV would remain longer in the aqueous core upon emul-
sifications due to the lower degree of supersaturation. However,
the results of Run C1 presented in Table 3 suggest otherwise,
where the % encapsulation and drug loading are lowered to 6% and
0.6% (w/w), respectively, from 10% and 1.0% (w/w) in the standard
DESE method. The nanoparticle size is nevertheless unaffected and
remains in the ≈200–300nm range.
Second, the inclusion of a surfactant in the inner aqueous phase
of the DESE method has been shown at the micro-scale to enhance
the primary w/o emulsion stability, which increases the aqueous
pore volume of the carrier particles, resulting in higher % encap-
sulation and drug loading [20]. For this reason, the inclusion of
surfactants in the aqueous core is nowattempted at the nano-scale
0
20
40
60
80
0 6 12 18 24
%

d
r
u
g

r
e
l
e
a
s
e
d
Time (hours)
Run B w/o lecithin
Run B3 w/ lecithin
Fig. 3. Effect of lecithin inclusion on the in-vitro drug release profile.
84 W.S. Cheow, K. Hadinoto / Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86
to examine whether a similar finding is observed. Using PEG as
the surfactant, the results of Run C2 indicate that % encapsulation
and drug loading are lowered to 8% and 0.8% (w/w), respectively,
without any significant impacts on the nanoparticle size.
Incontrast, usingPVAas thesurfactant inRunC3yields nanopar-
ticles having higher % encapsulation and drug loading (i.e. 22% and
2.1%, w/w, respectively), which is similar to the results obtained
with the lecithin inclusion in the ESE method (i.e. Run B3). How-
ever, the nanoparticle size in Run C3 is increased to ≈400–500nm
range, which makes themless suitable for sputumpenetration. The
significant difference in the effects of PEGand PVA inclusion on the
%encapsulationis likelydue tothe fact that adsorbedPEGis washed
off the PLGA nanoparticle surface more easily than PVA [21]. Sig-
nificantly, the results of Runs B3 and C3 indicate that the presence
of surfactant layer on the PLGA nanoparticle surface can enhance
the % encapsulation and drug loading provided that the surfactant
remains adsorbed upon washing.
Third, the modifications are performed in the oil phase by
replacing the 100% (v/v) DCM in the standard DESE method with
50%(v/v) mixture of DCM/acetone inRunC4andDCM/ethyl acetate
in Run C5. DCMcannot be completely replaced as the DESE method
requires the presence of a completely water-immiscible solvent.
In this regard, the type of solvent used as the oil phase has been
reported to impact the % drug encapsulation in nanoparticles due
to the different nanoparticle precipitation rates depending on the
water-miscibility level of the solvent used [22].
Specifically, in the second emulsification step, the polymer pre-
cipitates at a rate proportional to the water-miscibility level of the
solvent, where the more water-miscible the solvent is, the faster
the polymer precipitates to form w/o/w nano-emulsion as a result
of the faster exposure to the aqueous phase. The faster the poly-
mer precipitates andencapsulates the drug inthe process, less drug
leaks out fromtheinner totheouter aqueous phase, whichintheory
should improve the % encapsulation.
For this reason, the effect of varying the water-miscibility level
of the oil phase is investigatedinRuns C4 andC5 by adding miscible
acetone and partially miscible ethyl acetate into the DCM. In con-
trast to the results of a similar study in the ESE method (i.e. Runs
B1–B2), the results in Table 3 indicate that increasing the water-
miscibility level of the oil phase does lead to improvements in the
% encapsulation and drug loading to 20–22% and 2.1–2.4% (w/w),
respectively, which is more than two-fold higher than the values
obtained from the standard DESE method.
The slightly higher drug loading achieved in Run C4 is attributed
to the higher water-miscibility of the DCM/acetone mixture com-
pared to the DCM/ethyl acetate mixture. Importantly, the higher
drug loading does not jeopardize the nanoparticle sizes, which in
fact are slightly reduced to be below200nm. Despite the increased
drug loading, the in-vitro LEV release profiles of Runs C4 and C5
presented in Fig. 4 are similar to that of the standard DESE method
in Run C, where 90% of the drug is released within 6h.
The difference in the effects of varying the water-miscibility
level on the % encapsulation between the ESE and DESE methods
is postulated to be caused by the fact that in the DESE method, the
drug is dissolved in the inner aqueous phase, such that faster poly-
mer precipitations at higher water-miscibility levels can prevent
the drug leak-out from the inner aqueous core to the outer aque-
ous phase resulting in higher % encapsulation. In contrast, the drug
is dissolved in the oil phase (i.e. solvent phase) in the ESE method,
such that even though faster polymer precipitations increase the
% encapsulation, the diffusion of the water-miscible component of
the oil phase (i.e. either ethyl acetate or acetone) into the aqueous
phase uponemulsificationresults inthe dissolveddrugbeingtrans-
ported out to the aqueous phase together with the water-miscible
solvent. As a result, the increase in the % encapsulation owed to the
faster polymer precipitation is cancelled out by the drug loss due to
0
20
40
60
80
100
0 6 12 18 24
%

d
r
u
g

r
e
l
e
a
s
e
d
Time (hours)
Run C
Run C4
Run C5
Fig. 4. Effect of varying the water-miscibility level of the oil phase on the in-vitro
drug release profile.
the solvent outward diffusion resulting in insignificant net effects
on the % encapsulation.
3.3.2. Multiple-phase modifications
To examine whether further improvements in the % encapsula-
tion and drug loading can be achieved, simultaneous modifications
involving the inner and outer aqueous phases, as well as the oil
phase, are investigated in Runs C6–C10 presented in Table 4. 0.1%
(w/v) aqueous PVA solution is used as the inner aqueous phase in
Runs C6–C10as it has beenfoundinRunC3to significantly improve
the % encapsulation and drug loading, though it also results in a
larger nanoparticle size.
First, the effects of simultaneously modifying the inner aqueous
phase and the oil phase, while keeping the outer aqueous phase
unchanged (i.e. 2.0% (w/v) aqueous PVA solution), are examined
in Runs C6 and C7. DCM/acetone and DCM/ethyl acetate mixtures
are used as the oil phase in Runs C6 and C7, respectively, as they
have beenfoundtoincrease the %encapsulationinRuns C4–C5. The
results, however, indicate that a synergistic effect is not observed
whentheinner aqueous phaseandtheoil phasearemodifiedsimul-
taneously.
In fact, less favorable results in terms of the % encapsulation and
drug loading are observed in Runs C6–C7 compared to the single-
phase modification results. Specifically, the % encapsulation is back
to be ≤10% (w/w) similar to that obtained using the standard DESE
method. Similarly, the drug loading falls below 1.0% (w/w) and the
nanoparticle size is back in the ≈200–300nm range.
Second, the effects of simultaneously modifying the inner
and outer aqueous phases, while keeping the oil phase remains
unchanged (i.e. 100% (v/v) DCM), are investigated in Runs C8–C10.
The outer aqueous phase is modified by adding new surfactants or
polymers widely used in drug delivery formulation, such as PEG,
chitosan, and tyloxapol, into the 2.0% (w/v) aqueous PVA solution.
The presence of these materials at 50% (w/w) ratio in the outer
aqueous phase is thought to have the potential to influence the
drug leak-out rate.
The results nonetheless indicate that the % encapsulation, drug
loading, and nanoparticle size deteriorate significantly compared
to the single-phase modifications in Runs C3–C5. Specifically, the
% encapsulations are low between 4% and 11% (w/w) resulting
in low drug loadings between 0.1% and 1.0% (w/w). Furthermore,
the nanoparticle sizes in Runs C9 and C10 are excessively large at
≈700nm, hence signifying the importance of either PVA or PEG
W.S. Cheow, K. Hadinoto / Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86 85
in maintaining the colloidal stability of the w/o/w nano-emulsion
formed.
To summarize, for the DESE method, simultaneous mod-
ifications involving multiple phases negatively affect the %
encapsulation and drug loading. In contrast, single-phase modifi-
cations in Runs C3–C5 result in two-fold improvements in the %
encapsulation and drug loading. However, the large nanoparticle
size produced in Run C3 (i.e. ≥400nm) is undesirable due to the
lack of sputumpenetration ability. For the ESE method, the lecithin
inclusion in Run B3 similarly results in a two-fold improvement
in the % encapsulation and drug loading, while maintaining the
nanoparticle size in the 200–300nm range.
Importantly, the nanoparticles produced in Run B3 exhibit a
slower sustained LEV release profile, which make them more
attractive than those produced in Runs C4 and C5, despite the sim-
ilarities in terms of their size, % encapsulation, drug loading, and
production yield. In addition, the ESE method is simpler than the
DESE method owed to the single-emulsification step. For this rea-
son, the lecithin-modified nanoparticles produced in Run B3 are
selected as the optimal formulation for further spray-drying and
antibacterial activity testing experiments.
3.4. Spray-drying production of dry-powder LEV-loaded PLGA
nanoparticles
The following spray-drying conditions are employed, i.e.
inlet temperature =100
◦
C, feed rate =2.8mL/min, atomizing flow
rate =350L/h, and feed concentration=1.0% (w/w). Due to the high
drying temperature, which is above the glass transition tempera-
ture of PLGA(i.e. ≈35
◦
C), the inclusion of leucine and PVAas drying
adjuvants is necessarytoprotect thestructural integrityof thePLGA
nanoparticles. The ratio of the nanoparticles to the drying adju-
vants is maintained at 1:1. Dry-powder nanoparticle aggregates
produced at this spray drying condition possess d
G
between 6 and
9␮mand low
eff
(≈0.3g/cm
3
) owed to the hollowmorphology as
shown in Fig. 5A. Due to the hollow morphology, d
A
of these par-
ticles is between 3 and 5␮m rendering them suitable for inhaled
delivery.
Importantly, the drug loading of the dry-powder nanoparticles
is similar in magnitude to that of the suspension form indi-
cating a minimal drug loss in the drying process. Furthermore,
these nanoparticle aggregates readily reconstitute into individual
nanoparticles upon their exposure to an aqueous environment. The
nanoparticle size after the re-constitution in Fig. 5B, however, is
about twice larger than their size prior to the dry-powder transfor-
mation, which is due to the coalescence of the PLGA nanoparticles
upon exposure to a high temperature above their glass transition
temperature as can be seen in Fig. 5C.
3.5. Antibacterial activity of dry-powder LEV-loaded PLGA
nanoparticles
To ensure that the activity of the nanoparticles is not affected
by their transformation into the dry-powder form, antibacterial
activities of both the dry-powder and suspension forms of the
nanoparticles are tested against P. aeruginosa biofilm cells. The
antibacterial activities of the nanoparticles are evaluated in a time-
kill study after 6, 12, and 24h of antibiotic exposures using a LEV
dose of 7±0.3␮g/mL, which is above the minimuminhibitory con-
centration (MIC) of the planktonic cells (≈2␮g/mL) as biofilmcells
are well-known to exhibit a higher antibiotic tolerances than their
planktonic cell counterparts.
The experimental control (i.e. biofilm cells exposed to PBS)
shown in Fig. 6A remains at a density of ≈8log (CFU/peg) through-
out the duration of the study signifying a steady-state biofilm
density when no antibiotic is introduced. After 6h of exposure to
Fig. 5. (A) SEM image of dry-powder nanoparticle aggregates prepared by spray-
drying, (B) particle size distribution of the nanoparticles before the dry-powder
transformation and after re-constitution, and (C) SEM image of the re-dispersed
nanoparticles.
the nanoparticles, 99.999% of the biofilm cells are killed with only
3log (CFU/peg) of surviving biofilm cells, which is attributed to
the high antibiotic concentration produced by the burst release.
The number of biofilm cells, however, increases at the 12th and
24th hour to 5log (CFU/peg) as the surviving biofilm cells begin to
multiply due to the subsequent slower antibiotic release. Overall,
99.9%of thebiofilmcells areeradicatedafter 24h. Importantly, at all
time-points, the extents of biofilmcell eradications afforded by the
dry-powder form are nearly identical to that of the nanoparticle
suspension, which indicates that the dry-powder transformation
does not affect the antibacterial activity of the nanoparticles.
In addition, to verify that the biofilm eradications presented in
Fig. 6A are solely attributed to the LEV-loaded nanoparticles, and
not because of the presence of hazardous solvent trace from the
nanoparticle preparation steps, antibacterial activity of the blank
nanoparticles (i.e. no antibiotic) is examined. The results in Fig. 6B
indicate that the blank nanoparticles, in their suspension and dry-
86 W.S. Cheow, K. Hadinoto / Colloids and Surfaces A: Physicochem. Eng. Aspects 370 (2010) 79–86
Fig. 6. (A) Antibacterial activities of the LEV-loaded nanoparticles at three differ-
ent time-points (DP3=dry-powder nanoparticles, NP3=nanoparticle suspension)
against P. aeruginosa biofilm cells and (B) non-activity of blank nanoparticles.
powder forms, do not exhibit any antibacterial activities towards
the biofilm cells as the log (CFU/peg) of the biofilm cells exposed
to the blank nanoparticles are similar in magnitude to that of the
control experiment.
4. Conclusions
LEV-loaded PLGA nanoparticles are prepared by employing the
standard and modified NPC, ESE and DESE methods. In compari-
son to the standard methods, single-phase modifications involving
(i) the addition of lecithin into the aqueous phase (ESE) and (ii)
the use of alternative solvents (DESE) result in two-fold improve-
ments in the % encapsulation and drug loading (i.e. ≈22–24% and
2.1–2.4%, w/w, respectively). Significantly, theexperimental results
indicatethat simultaneous modifications involvingmultiplephases
do not lead to additive effects of the single-phase modifications. In
other words, an observed doubling of the % encapsulation when
two single-phase modifications are performed separately does not
translate to quadrupled % encapsulation when the two modifi-
cations are performed simultaneously. The optimal formulation,
which is evaluated in terms of the nanoparticle size, % encapsu-
lation, drug loading, in-vitro drug release profile, and production
yield, has been identified in the form of PLGA-lecithin nanopar-
ticles prepared by the ESE method. Lastly, the nanoparticles can
be straightforwardly transformed into the inhalable solid dosage
form by spray-drying with the inclusion of drying adjuvants. The
dry-powder nanoparticles exhibit similar antibacterial activities as
their aqueous suspension form against P. aeruginosa biofilm cells.
Acknowledgement
A financial support from Nanyang Technological University’s
Start-Up Grant (Grant No. SUG 8/07) is gratefully acknowledged.
References
[1] P.G.A. Rogueda, D. Traini, The nanoscale in pulmonary delivery. Part 2: Formu-
lation platforms, Expert Opin. Drug Deliv. 4 (2007) 607–620.
[2] P.R.A. Pauwels, K.F. Rabe, Burden and clinical features of chronic obstructive
pulmonary disease (COPD), Lancet 364 (2004) 613–620.
[3] J.S. Suk, S.K. Lai, Y.-Y. Wang, L.M. Ensign, P.L. Zeitlin, M.P. Boyle, J. Hanes, The
penetration of fresh undiluted sputum expectorated by cystic fibrosis patients
by non-adhesive polymer nanoparticles, Biomaterials 30 (2009) 2591–2597.
[4] P. Meers, M. Neville, V. Malinin, A.W. Scotto, G. Sardaryan, R. Kurumunda,
C. Mackinson, G. James, S. Fisher, W.R. Perkins, Biofilm penetration, trig-
gered release and in vivo activity of inhaled liposomal amikacin in chronic
Pseudomonas aeruginosa lung infections, J. Antimicrob. Chemother. 61 (2008)
859–868.
[5] M. Halwani, S. Hebert, Z.E. Suntres, R.M. Lafrenie, A.O. Azghani, A. Omri,
Bismuth-thiol incorporation enhances biological activities of liposomal
tobramycin against bacterial biofilm and quorum sensing molecules produc-
tion by Pseudomonas aeruginosa, Int. J. Pharm. 373 (2009) 141–146.
[6] P.G.A. Rogueda, D. Traini, The nanoscale in pulmonary delivery. Part 1:
deposition, fate, toxicology and effects, Expert Opin. Drug Deliv. 4 (2007)
595–606.
[7] K. Hadinoto, W.S. Cheow, Hollow spherical nanoparticulate aggregates as
potential ultrasoundcontrast agent: shell thickness characterization, DrugDev.
Ind. Pharm. 35 (2009) 1167–1179.
[8] K. Hadinoto, Mechanical stability of hollowspherical nano-aggregates as ultra-
sound contrast agent, Int. J. Pharm. 374 (2009) 153–161.
[9] H. Ishida, Y. Ishida, Y. Kurosaka, T. Otani, K. Sato, H. Kobayashi, In vitro and in
vivo activities of levofloxacin against biofilm-producing Pseudomonas aerugi-
nosa, Antimicrob. Agents Chemother. 42 (1998) 1641–1645.
[10] S. Schwank, Z. Rajacic, W. Zimmerli, J. Blaser, Impact of bacterial biofilm for-
mation on in vitro and in vivo activities of antibiotics, Antimicrob. Agents
Chemother. 42 (1998) 895–898.
[11] A. Frick, H. Möller, E. Wirbitzki, Biopharmaceutical characterization of oral
immediate release drug products. In vitro/in vivo comparison of phe-
noxymethylpenicillin potassium, glimepiride and levofloxacin, Eur. J. Pharm.
Biopharm. 46 (1998) 305–311.
[12] T. Govender, S. Stolnik, M.C. Garnett, L. Illum, S.S. Davis, PLGA nanoparticles
prepared by nanoprecipitation: drug loading and release studies of a water
soluble drug, J. Controlled Release 57 (1999) 171–185.
[13] J.M. Barichello, M. Morishita, K. Takayama, T. Nagai, Encapsulation of
hydrophilic and lipophilic drugs in PLGA nanoparticles by the nanoprecipi-
tation method, Drug Dev. Ind. Pharm. 25 (1999) 471–476.
[14] L. Peltonen, J. Aitta, S. Hyvönen, M. Karjalainen, J. Hirvonen, Improved entrap-
ment efficiency of hydrophilic drug substance during nanoprecipitation of
poly(l)lactidenanoparticles, AAPSPharmSciTech5(2004) [electronic resource].
[15] D. Worlitzsch, R. Tarran, M. Ulrich, U. Schwab, A. Cekici, K.C. Meyer, P. Bir-
rer, G. Bellon, J.r. Berger, T. Weiss, K. Botzenhart, J.R. Yankaskas, S. Randell,
R.C. Boucher, G. Döring, Effects of reduced mucus oxygen concentration in
airway Pseudomonas infections of cystic fibrosis patients, Journal of Clinical
Investigation 109 (2002) 317–325.
[16] J. Harrison, H. Ceri, J. Yerly, C. Stremick, Y. Hu, R. Martinuzzi, R. Turner, The use
of microscopy and three-dimensional visualization to evaluate the structure of
microbial biofilms cultivated in the calgary biofilm device, Biol. Proced. Online
8 (2006) 194–215.
[17] W. Cheow, M. Chang, K. Hadinoto, Antibacterial efficacy of inhalable
levofloxacin-loaded polymeric nanoparticles against E. coli biofilm cells: the
effect of antibiotic release profile, Pharm. Res. 27 (2010) 1597–1609.
[18] L. Zhang, J.M. Chan, F.X. Gu, J.-W. Rhee, A.Z. Wang, A.F. Radovic-Moreno,
F. Alexis, R. Langer, O.C. Farokhzad, Self-assembled lipid-polymer hybrid
nanoparticles: a robust drug delivery platform, ACS Nano 2 (2008) 1696–1702.
[19] A. Lambert, J.-B. Regnouf-de-Vains, M.F. Ruiz-López, Structure of levofloxacin
in hydrophilic and hydrophobic media: Relationship to its antibacterial prop-
erties, Chem. Phys. Lett. 442 (2007) 281–284.
[20] Y.-Y. Yang, T.-S. Chung, N. Ping Ng, Morphology, drug distribution, and in vitro
release profiles of biodegradable polymeric microspheres containing protein
fabricated by double-emulsion solvent extraction/evaporation method, Bio-
materials 22 (2001) 231–241.
[21] L. Mu, P.H. Seow, S.N. Ang, S.S. Feng, Study on surfactant coating of polymeric
nanoparticles for controlled delivery of anticancer drug, Colloid Polym. Sci. 283
(2004) 58–65.
[22] E. Cohen-Sela, M. Chorny, N. Koroukhov, H.D. Danenberg, G. Golomb, A new
double emulsion solvent diffusion technique for encapsulating hydrophilic
molecules in PLGA nanoparticles, J. Controlled Release 133 (2009) 90–95.