Ow

DMa

Db

c

a

ARR1AA

KRWSO

1

tfitptttt(a

m

0h

Colloids and Surfaces B: Biointerfaces 115 (2014) 109–117

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces

j o ur nal ho me pa ge: www.elsev ier .com/ locate /co lsur fb

ptimization of the production of solid Witepsol nanoparticles loadedith rosmarinic acid

ébora A. Camposa, Ana Raquel Madureiraa,∗, Ana Maria Gomesa, Bruno Sarmentob,c,aria Manuela Pintadoa,∗∗

CBQF – Centro de Biotecnologia e Química Fina – Laboratório Associado, Escola Superior de Biotecnologia, Universidade Católica Portuguesa/Porto, Ruar. António Bernardino Almeida, 4200-072 Porto, PortugalINEB – Instituto de Engenharia Biomédica, NewTherapies Group, Universidade do Porto, Rua do Campo Alegre, 823, 4150-180 Porto, PortugalCICS – Health Sciences Reseach Center, Instituto Superior de Ciências da Saúde – Norte, Gandra, Portugal

r t i c l e i n f o

rticle history:eceived 7 June 2013eceived in revised form2 September 2013ccepted 22 October 2013vailable online 21 November 2013

eywords:osmarinic aciditepsol

LNsptimization

a b s t r a c t

During the last decade there has been a growing interest in the formulation of new food and nutraceu-tical products containing compounds with antioxidant activity. Unfortunately, due to their structure,certain compounds such as polyphenols, in particular rosmarinic acid (RA) are not stable and may inter-act easily with matrices in which they are incorporated. To overcome such limitations, the formulationof loaded polyphenols nanoparticles can offer an efficient solution to protect such compounds. Based onthis rationale, the aim of this study was to prepare solid lipid nanoparticles (SLNs) loaded with RA usinga hot melt ultrasonication method, where Witepsol H15 was used as lipid and Polysorbate 80 (Tween80) as surfactant, following a 32 fractional factorial design, resulting in the use of 3 different percentagesof surfactant (viz. 1, 2 and 3%, v/v) and lipid (0.5, 1.0 and 1.5%, w/v). The stability of the nanoparticlessystems were tested during 28 d in aqueous solution stored at refrigeration temperature (ca. 5 ◦C), track-ing the mean particle size of different formulations by photon correlation spectroscopy. To confirm RAentrapment, thermal analyses of the nanoparticles by DSC and FTIR were performed. The association effi-

ciencies percentages (AE%) were determined using HPLC to quantitatively assess the RA in supernatants.Results showed that Witepsol H15 produced nanoparticles with initial mean diameters between 270and 1000 nm, yet over time, a slight increase occurred, but without occurrence of aggregation. The AE%showed a high percentage of encapsulation (ca. 99%), which reveals low polyphenol releases from SLNsthroughout storage time. In general, results showed a successful production of SLNs with properties that

cation

can be used to food appli

. Introduction

Solid lipid nanoparticles (SLNs) were introduced for the firstime in the early 1990s and are prepared from a lipid matrix, withnal particle size ranging between 50 and 1000 nm [1]. The reduc-ion of particle size and the use of non-toxic materials make thesearticles important colloidal carriers, because they combine advan-ages such as physical stability, controlled release and excellentolerability, being one of the most currently used systems [2–4]. Forheir formulation, lipid, emulsifier and water are required as essen-

ial components, and all need to hold a Generally Recognized as SafeGRAS) status, if their final destination use is for oral and topicaldministration [4]. The lipids used, normally have a melting point

∗ Corresponding author. Tel.: +351 225580044.∗∗ Corresponding author.

E-mail addresses: rmadureira@porto.ucp.pt (A.R. Madureira),pintado@porto.ucp.pt (M.M. Pintado).

927-7765/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.colsurfb.2013.10.035

s.© 2013 Elsevier B.V. All rights reserved.

above room and body temperature and can be triglycerides, mono,di and triglycerides mixtures, waxes, hard fats and other types oflipids [4]. A good example is, Witepsol wax types, usually used inthe pharmaceutical industry as excipients but also approved forhuman consumption, which comprise a mixture of triglycerides,diglycerides and monoglycerides in different proportions. Thesewaxes are normally presented in four different formulations thatchange according to the type and amount of glycerides (mono- anddiglycerides) used, commonly identified as H, W, S and E Witepsolgrades. These wax grades differ in their melting and solidifica-tion points, leading to different consistencies from soft to hardlipid structure as well as different post-hardening tendencies andthese properties are essential selection criteria during formulationdevelopments [5]. On the other hand, the emulsifiers used are com-mon types of poloxamer and polysorbate, but lecithin, tyloxapol,

sodium cholate and sodium glycocholate, taurodeoxychocolic acidsodium, butanol and butyric acid, among others can also be used[4,6]. Among, the methods and techniques used to produce SLNs, acommon strategy is to employ the hot homogenization technique,

1 faces B: Biointerfaces 115 (2014) 109–117

waa

sttptHdtsnsclaC

etaNcascclctchtftfb

bucft

2

2

G(m(

2r

RdmaP

Table 1Boundaries of experimental design and spacing of levels of processing parametersexpressed in coded and natural units.

Level (code unit) Processing parameters

Wax (%, w/v)(X1)

Tween 80 (%)(X2)

−1 0.5 1

10 D.A. Campos et al. / Colloids and Sur

hich involves homogenizing a compound loaded lipid and anqueous phase together in the presence of a hydrophilic emulsifiert a temperature above the melting point of the lipid phase [7].

One of the main advantage of these nanoparticles is the pos-ibility of improving the stability of the incorporated compounds,hermal stability, increased oral bioavailability of compounds (par-icularly lipophilic molecules), e.g. labile drug molecules can berotected from external environment (during storage) and pro-ection from the digestive process (following oral administration).ydrophilic drugs can also be incorporated in SLNs, but such proce-ure is considered to be challenge because of the affinity betweenhe drug and the lipid, and the tendency of partitioning the encap-ulated molecules in the water during the production process ofanoparticles [8]. Another challenge is the type of lipid used, and itsuccessful adsorption by the intestinal epithelium, since medium-hain triglycerides (MCT) lipids are more effectively adsorbed thanong-chain triglycerides (LCT) [9]. For example, Witepsol H seriesre mainly composed of a mixture of mono, di- and triglycerides10-C18 [5].

Rosmarinic acid (RA) is a natural polyphenol carboxylic acid, anster of caffeic acid with 3,4-dihydroxyphenyllactic acid. Normally,his acid appears in higher amounts in families such as Boraginaceaend Lamiaceae, but, in the latter, it is restricted to a subfamily, theepetoideae. Rosmarinic acid has been identified as one of the activeomponents of several medicinal plants (e.g. Salvia officinalis), with

number of potential biological properties associated therewith,uch as antioxidant, anti-mutagenic, anti-bacterial and anti-viralapabilities [10]. Additionally, it exhibits various pharmacologi-al properties, including prevention of oxidation of low densityipoproteins, inhibition of murine cell proliferative activity and ofyclooxygenase, anti-inflammatory and anti-allergic actions [11],he most important activities being the actual protection againstancer and the high antioxidant activity [12]. The technologicalandling of such type of compounds is sometimes limitative dueo their reactivity, so their incorporation in an oral nutraceuticalormulation became difficult if they are not duly protected. Addi-ionally, conditions prevailing during digestion can be negativeor the stability of these compounds, as well as compromise theirioavailability.

Thus, the objective of this research work was to develop sta-le solid lipid nanoparticles containing rosmarinic acid (RA-SLNs)sing a Witepsol wax as lipid matrix. Formulation parameters, lipidoncentration and emulsifier concentration, were optimized via aractional factorial experimental design, aiming at their incorpora-ion in a nutraceutical oral formulation.

. Materials and methods

.1. Materials

The lipid wax tested was Witepsol H15 (Sasol, Hamburg,ermany), composed by a blend of hydrogenated coco-glycerides

C12–C18). The surfactant Polysorbate 80 (Tween 80) and ros-arinic acid (RA) were purchased from Sigma–Aldrich Chemistry

St. Louis, MI, USA).

.2. Formulation of solid lipid nanoparticles loaded withosmarinic acid

The SLNs were prepared by hot melt ultrasonication, loadingA at a final concentration of 0.15 mg/mL. Based on a 32 factorial

esign, several SLNs formulations were produced using the lipidatrix at three different concentrations (0.5, 1 and 1.5%, w/v) and

lso were used three different percentages of aqueous solutions ofolysorbate 80 as surfactant (1, 2 and 3%, v/v) as shown in Table 1.

0 1.0 21 1.5 3

The lipid was warmed to a temperature 5 ◦C above the melting point(36 ◦C), then the RA in solution was added to the melted matrix andsubmitted to ultrasonication (VCX 130, Sonics & Materials, New-town, USA) during 1 min at 70% of intensity. Finally, the aqueoussolution of emulsifier was added and mixed for a few seconds, forcompletely homogenization of the O/W emulsion. The resultingfluid solutions were left to cool at room temperature (20 ◦C). TheSLN emulsions were stored at 5 ◦C throughout 28 d until further use.When dried SLNs were used, the final SLNs emulsions were freeze-dried upon production using a Vacuum Freeze Drier (Model FT33,Armefield, UK), under a vacuum pressure of 100 mTorr; the tem-perature in the freezing chamber was −46 ◦C and the temperaturein the sample chamber was 15 ◦C.

2.3. Physical and morphological characterization

For evaluation of the mean particle size (PS), polydispersityindex (PI) and zeta potential (ZP), liquid samples were analyzed viaphase analysis dynamic light scattering (DLS) using a ZetaPALS, ZetaPotential Analyzer (Holtsville, NY, USA). All analyses were carriedout with an angle of 90◦ at 25 ◦C. These properties were evaluatedin triplicate at 0 and 28 d of storage time.

2.4. Association efficiency

The association efficiency percentages (AE%) were calculated bythe difference between the total amount of incorporated RA andthe amount of RA present in the supernatant of SLNs formula-tions (Eq. (1)). The separation of free RA from aqueous solution ofSLNs was performed by centrifugation at 3000 rpm, during 20 minand at 4 ◦C using centrifugal filter units with a cut-off of 10 K(Amicon® Ultra-4, Millipore; Billerica, MA, USA). The resultingsupernatant was removed and analyzed for free RA concentration.These concentrations were obtained by High Performance LiquidChromatography (HPLC) analysis of the supernatants of SLNs, usinga diode-array detector (HPLC–DAD) (Waters Series 600, Mildford,MA, USA). Separation was done in a C18 reverse-phase column cou-pled with a guard column containing the same stationary phase(Symmetry® C18, Waters, Mildford, MA, USA). Chromatographicseparation of phenolic compounds was carried out with mobilephase A – water, methanol (Panreac, Barcelona, Spain) and formicacid (Merck, Darmstadt, Germany) (92.5:5:2.5) – and mobile phaseB–methanol, water and formic acid (92.5:5:2.5) – under the fol-lowing conditions: linear gradient elution starting at 0–60% mobilephase B in 60 min at 0.65 mL/min, 60–10% in 5 min at 0.5 mL/minand from 5% to 0% in 5 min. The injection volume was 20 �l. Detec-tion was achieved using a diode array detector (Waters, Mildford,MA, USA) at wavelengths ranging from 200 to 600 nm measured in2 nm intervals.

The calculations were performed according to the following for-mula:

AE% = Total amount of RA − Amount of RA in supernatantTotal amount of RA

× 100

(1)

D.A. Campos et al. / Colloids and Surfaces B: Biointerfaces 115 (2014) 109–117 111

Table 2Mean values ± SD of particles size (PS), polydispersity index (PI), zeta potential (ZP) and percentage of association efficiency (AE%) of the produced nanoparticles over 28 dof storage at refrigerated temperatures.

Properties Storage time (d) A B C D E F G H I

PS (nm) 0 411 ± 1a 1149 ± 117bcd 659 ± 289abc 646 ± 291ab 1303 ± 147cd 1400 ± 31d 650 ± 328abc 372 ± 80abc 423 ± 109a

28 821 ± 48a 956 ± 225a 872 ± 132a 663 ± 130a 648 ± 5a 694 ± 194a 588 ± 223a 583 ± 97a 769 ± 102a

PI 0 0.23 ± 0.03a 0.28 ± 0.11a 0.23 ± 0.03a 0.29 ± 0.09a 0.30 ± 0.10a 0.29 ± 0.09a 0.32 ± 0.05a 0.25 ± 0.09a 0.28 ± 0.04a

28 0.30 ± 0.05a 0.30 ± 0.01a 0.25 ± 0.03a 0.28 ± 0.09a 0.20 ± 0.09a 0.29 ± 0.03a 0.30 ± 0.03a 0.24 ± 0.10a 0.29 ± 0.06a

ZP (mV) 0 −38.4 ± 3.0a −39.4 ± 3.1a −39.1 ± 3.2a −38.2 ± 2.9a −38.7 ± 3.0a −38.2 ± 2.9a −38.0 ± 2.9a −38.7 ± 3.0a −38.6 ± 3.0a

28 −38.9 ± 3.0a −38.2 ± 2.9a −38.2 ± 2.9a −39.5 ± 3.1a −37.7 ± 2.9a −38.3 ± 2.9a −38.0 ± 2.9a −38.8 ± 3.0a −37.8 ± 2.9a

AE (%) 0 99.88 99.89 99.87 99.84 99.80 99.83 99.78 99.80 99.7628 99.75 99.82 99.81 99.89 99.89 99.89 99.93 99.93 99.85

D B) 0.5a persc

2

bCRvtapveClE

C)ratio

2

tJmuwat

2

RwZrWtitWS

74spra

esignations of the letters are as follows (Witepsol%, w/v: Tween%, v/v): (A) 0.5:1, (,b,c,d The differences between the means in the same row labelled with the same su

.5. Thermal properties characterization

The thermodynamic behavior of the SLNs was determinedy using Differential Scanning Calorimetry (DSC-60, Shimadzu,olumbia, USA). All SLNs formulations, both inert and loaded withA, as well as all the raw materials used in the formulations, indi-idually and in combination, were tested. Briefly, samples wereaken from freeze-dried SLNs and 3 mg of SLNs were placed in anluminum crucible. Thermal behavior was determined in the tem-erature range of 20–100 ◦C at a heating rate of 10 ◦C/min. Enthalpyalues and optimal melting temperatures were calculated by thequipment software (ta60 version 2.10, DSC software, Shimadzu,olumbia, USA). The crystallinity indexes (CI%) of SLNs were calcu-

ated according to Kheradmandnia et al. (2010), using the followingq. (2):

I% = Melting enthaply (SLN dispersion) (J/gMelting enthaply (bulk material without RA) (J/g) × Concent

.6. Morphological properties

Morphology of nanoparticles was evaluated by Scanning Elec-ron Microscopy (SEM) using a JEOL-5600 Lv microscope (Tokyo,apan). Briefly, a small amount of freeze-dried SLNs was placed on

etallic stubs with carbon tape and coated with gold/palladiumsing a Sputter Coater (Polaron, Bad Schwalbach, Germany). SEMas operated at the high vacuum mode, using a spot size of 36–37

nd a potential of 15–22 kV. All analyses were performed at roomemperature (20 ◦C).

.7. Fourier transform infrared (FT-IR) spectroscopy

The freeze-dried formulations of SLNs with and without RA, pureA and pure wax lipid Witepsol H15, as well as physical mixtures,ere evaluated using an ABB MB3000 FT-IR spectrometer (ABB,

ürich, Switzerland) equipped with a horizontal attenuated totaleflectance (ATR) sampling accessory (PIKE Technologies, Madison,

I, USA) with a diamond/ZnSe crystal, obtaining different spec-ra. All samples were run in triplicate. Several controls were runn parallel, a background run (to remove the background noise ofhe instrument) was carried out as a negative control, the matrix

itepsol, as well as RA, were carried out as positive controls; theLNs with and without RA were also analyzed.

The mid-infrared absorbance region was between 4000 and00 cm−1 and the spectra were measured at a spectral resolution of

cm−1 with 200 scans co-added, to minimize differences between

pectra due to baseline shifts. In order to perform the spectra com-arison, spectra were truncated at 1800 and 700 cm−1, since thisegion displays typical absorption bands for the used compounds. Inddition, baseline 4–5 point adjustment and spectra normalization

:2, (C) 0.5:3, (D) 1.0:1, (E) 1.0:2, (F) 1.0:3, (G) 1.5:1, (H) 1.5:2, (I) 1.5:3.ript are not statistically significant (P > 0.05) (n = 9).

n of lipid phase(%)× 100 (2)

was performed. Treatment of all spectra was carried out with theHorizon MBTM FTIR software (ABB, Zürich, Switzerland).

2.8. Statistical and mathematical analyses

A full factorial design was implemented to evaluate the effect oflipid and surfactant concentration, on PS, PI, ZP, AE% and CI% at thetime of production (0 d).

The boundaries of the chosen domain of experimentation werecalculated using:

Xi = xi − xxi

�xi(3)

where Xi is the coded value of the ith independent variable, xi isthe natural value of that variable, xx

iis the natural value of said

variable at the center point, and �xi is half the amplitude of theexperimental range of that variable. The central values (zero level)chosen for the experimental design were 1% (v/v) lipid and 2% (v/v)of Polysorbate 80.

The aforementioned coded values of all three variables werethen laid out according to a full factorial – which allows estima-tion of second order terms (in addition to linear terms), besides arepresentative estimate of variability at the center point. The totalnumber of experiments was thus given by:

n = 2k−p + 2k + cp (4)

where n is the number of experiments, k is the factor number,p is the fractionalization number, and cp is the number of cen-ter points required for curvature estimation; the planned designencompassed a total of 9 experimental runs as depicted in Table 2.

The underlying quadratic model (including linear and quadratic,as well as and first order interaction factors), reads:

Y = b0 +2∑

i=1

bixi +2∑

i=1

biix2i + e (5)

where Y is the measured response; b0 is the intercept; bi, bii andbij, are the coefficients associated with linear, quadratic and inter-action effects, respectively, of variables xi and xj, respectively ande is the (random) error. All fits, and associated statistical analy-

ses related with the experimental design were performed usingStatistical software (StatSoft, Tulsa OK, USA).

The statistical significance at a 5% level of differences betweenthe means values of the tested parameters, viz. PS, PI, ZP, AE% and

1 faces B

Ctw

3

3

tNwot8fc13PIsdetr(ieltttZipcpt

rTa(susbsastwelh

assmunpw

12 D.A. Campos et al. / Colloids and Sur

I% values obtained for time 0 and 28 d, were determined by Tukey’sest, by running one-way ANalysis Of VAriance (ANOVA) carried outith the aid of SPSS (v. 20, Chicago, IL, USA).

. Results and discussion

.1. Rosmarinic acid – SLNs physical properties

The characterization of lipid nanoparticles is critical due tohe complexity of the system and colloidal size of the particles.onetheless, it is of great importance in order to generate systemsith required properties. In the optimization of the formulation

f different SLNs, the use of an experimental design was useful toest different combinations of lipids and surfactant (Polysorbate0) and their effect on the physical properties of the SLNs. The 32

actorial plan included nine SLNs formulations resulting from theombination of three different concentrations of lipid (0.5, 1 and.5%, w/v) and three different percentages of emulsifier (1, 2 and%, v/v). The physical stability was evaluated by measuring the PS,I and ZP at time of production (0 d) and upon storage for 28 d.t is well established that composition of the formulation (such asurfactant, lipid type, concentration and compound), as well as pro-uction methods and conditions (time, temperature, equipment,tc.) may affect SLNs physical properties. The quantitative estima-ion of effects (and interactions) of the various variables upon theesponses can be obtained by pareto charts in terms of positive+) and negative (−) effects, together with their statistical signif-cance. When the effects are statistically significant (P < 0.05) thequation of the response surface model can be obtained. Neverthe-ess, even if these effects are not statistically significant (P > 0.05),he response surface model can be used to obtain a prediction ofhe range of values for the variables (lipid and surfactant concentra-ion) to obtain the desired value of the measured parameter (e.g. PS,P, PI, CI% among). This analysis was performed with data achievedmmediately upon production – at 0 days – in order to optimize thearameters to be used in production of the SLNs with mean parti-les sizes above 300 nm, and the highest AE% and homogeneity (PI)ossible. An equally distributed surface charge of the nanoparticleso avoid aggregation is also targeted.

Analysis of Table 2 reveals SLNs with mean diameters in theange of 372–1400 nm size range, at the time of production (0 d).here is no statistically significant relationship with the lipidmount used in production. This is confirmed by the pareto chartsFig. 1a), which shows an inverse relationship between the Witep-ol concentration (linear effect) and PS, i.e. higher PS are obtainedsing lower concentrations of the lipid, but this relationship is nottatistically significant (P > 0.05). The increase in lipid amount wille expected to increase the size of the particle, and the presence of aufficient surfactant amount helps reducing the interfacial tensionnd the lipid tend to be more homogenized and thus reduces theize of the particle [13]. In fact, using higher contents of surfactant,he average SLNs sizes tended to decrease, which is in agreementith the found specifically for SLNs [7]. In the present study the only

xception found was for formulations produced with 1.5% (w/v) ofipid (P > 0.05), but this result can be attributed with the use of aigh concentration of lipid, as stated before.

Until the present date and to the best of our knowledge therere no published research works concerning the use of Witep-ol H15 as lipid matrix for the production of nanoparticles. But,everal research studies have used other Witepsol waxes as lipidatrix and the nanoparticles produced with these waxes and

sing polysorbates (Tween 80) as surfactant, generally producesanoparticles with small sizes (<500 nm) than the overall valuesresented reported in this study [14]. Particle sizes tend to be largerhen the addition of surfactant is done before crystallization of

: Biointerfaces 115 (2014) 109–117

nanoparticles, i.e. before nanoparticles formation, and this featuremay eventually affect the stability of the emulsion. Nevertheless,after 28 d of storage, the particle sizes in all formulations tendedto become more uniform and no significant statistical differenceswere found between these values (P > 0.05), which shows that theseformulations were stable over storage time and at refrigerated tem-peratures.

In the present work, aiming at the use of oral route for admin-istration a desirable value of PS for SLNs is established as ≥300 nm,in order to not exceed the epithelium of the gut, avoiding passageto the bloodstream and toxicity of SLNs by accumulation in theorgans [4]. All formulations reported PS > 300 nm, yet if better con-trolled are sought the surface response chart for PS (Fig. 2a) canbe used to estimate the average range of lipid and surfactants to beused; hence, the most favorable conditions to obtain SLNs with sizesabove 300 nm should include concentrations of surfactant between1–1.4% (v/v) and of lipid between 0.4–0.6% (w/v) and 1.2–1.4% (w/v)in the formulation.

The PI is an important value since it shows the size distributionof the nanoparticles; the maximum value is arbitrarily limited to1.0 and indicates that the sample has a very broad size distributionand may contain large particles or aggregates that could slowlysediment. An optimum PI value is acceptable at <0.30 as describedby Das and Chaudhury [15], which means that 66.7% of SLNs havethe same size. All Witepsol formulations presented PI values below0.30 showing high homogeneity of SLNs. According to the paretocharts (Fig. 1b), the lipid and the surfactant concentrations affectedthe PI positively, i.e. by increasing their concentrations an increasein PI values was registered (P > 0.05). This trend is further confirmedby the response surface chart (Fig. 2b). Along storage, the values ofPI remained similar after 28 d of storage (P > 0.05), which showsthat the particles conserved their size during this storage period.

The ZP values (Table 2) indicate the overall charge that the par-ticles acquire in a specific medium. It also ensures informationabout the degree of repulsion between close and similarly chargedparticles in the dispersion, and consequently may be used as anindicator of SLNs stability. A high absolute value of ZP indicatesthat there is no risk of aggregation of the particles due to electricrepulsion of SLNs; for low values of ZP, attraction between the par-ticles will occur and there is the risk of coagulation or flocculation.The more equally charged the particles are the higher are the elec-trostatic repulsions between them and the higher is the physicalstability. Previous studies have shown that a ZP above −60 mV isrequired for excellent, and above −30 mV for good physical sta-bility of colloidal carriers [16]. In this context, all formulations ofRA-SLNs prepared in this study presented ZP values between −38.0and −39.5 mV; upon 28 d of storage, the same narrow interval wasmaintained, which indicates that the SLNs maintained a good elec-trostatic stability throughout time. Accordingly to the pareto chart(Fig. 1c), both main effects were of statistical importance (P < 0.05),an increase in lipid concentration will increase the ZP whereas anincrease in surfactant concentration (quadratic effect) will decreasethe ZP. These conclusions were confirmed by the response surfacemodel graph (Fig. 2c), where the lipid concentrations and surfac-tant range values used were ideal to obtain ZP values between thetargeted −30 and −60 mV interval. This negative charge obtainedfor ZP values was likely caused by the slightly ionized fatty acidsfrom the glycerides used (Witepsol H15).

The AE% of a delivery system such as SLNs is one of the mostimportant features to be controlled since will give us informationabout the percentage of RA that was associated to the lipid andproduced the SLNs. The percentages of AE were high for all formu-

lations (ca. 99.8 ± 0.06%) even after 28 d of storage, which meansthat the polyphenol entrapment did not change for the differentformulations tested and throughout storage time. With even noapparent differences in the values of AE%, and with no statistical

D.A. Campos et al. / Colloids and Surfaces B: Biointerfaces 115 (2014) 109–117 113

F ), (c) zi l desig

stttWtT

ig. 1. Pareto charts obtained for (a) particle size (PS), (b) polydispersity index (PIndex percentage (CI%) for the SLNs produced according the fractional experimenta

ignificance (P > 0.05), the pareto charts obtained (Fig. 1d) showshat the lipid and surfactant concentrations influences negativelyhe AE%, i.e. increasing both concentrations the AE% decreases. This

rend can be confirmed by the response surface graphic (Fig. 2d).

hen using lipids/waxes to produce nanoparticles, it is expectedhat a physical entrapment of the encapsulated compound occurs.he range of lipid concentrations used were not significantly

eta potential (ZP), (d) association efficiency percentage (AE%) and (e) crystalinityn.

different to observe an influence in the AE%. Also, the surfactantis also a factor that does not influence this efficiency since it is usedto stabilize the particle and at time of production had no influence

in the association of the lipid to the polyphenol.

Scanning electron microscopy (SEM) was used to confirm thesize of the particles as well as their shape and arrangement.Micrographs (Fig. 3) obtained from the dried particles revealed

114 D.A. Campos et al. / Colloids and Surfaces B: Biointerfaces 115 (2014) 109–117

F dex (Pc xperi

WafTsSpita

3

io(

ig. 2. Response surface models obtained for (a) particle size, (b) polydispersity inrystalinity index percentage (CI%) for the SLNs produced according the fractional e

itepsol SLNs spherical shape, a dense lipid matrix and no visualggregation. However, the sizes measured by SEM were not inull agreement with those obtained by DLS. As can be seen inable 2, the selected Witepsol RA-SLNs formulation (1:2%; Witep-ol:Polysorbate 80) presented average sizes of 648 nm, whereasEM results show SLNs with sizes ca. <500 nm. Yet, since thesearticles were previously dried by freeze drying process and as

t has been reported that solvent removal may cause modifica-ions affecting the particle shape and size [15], these differencesre comprehensively accepted.

.2. Thermal properties

The thermal properties and thermodynamic behavior was stud-ed by differential scanning calorimetry (DSC). This equipmentffers a close look at the crystallization and melting behaviorfusion of the crystal lattice or breakdown), by heating the samples

olid. Index), (c) zeta potential, (d) association efficiency percentage (AE%) and (e)mental design.

and giving information about polymorphism and crystal ordering[17].

The RA and the surfactant does not shown any peaks for thetested temperatures (Fig. 4), as expected, however these compo-nents influences the lipid matrix and crystal formation (particles).On the other hand, for the tested temperatures, the studied formu-lations and the pure lipid presented endothermic peaks (Fig. 5). Ahigh value of melting enthalpy suggests a high level of organiza-tion in the crystal lattice, because the fusion of an organized crystal(perfect crystal) requires more energy to overcome the forces ofcohesion in the crystal lattice [18].

The thermal behavior of the different formulations of SLNs(with and without RA) presented range melting temperatures val-

ues between 35.5 and 38.5 ◦C (Fig. 4, line D). The values wereslightly lower than the melting point of control, Witepsol H15(41.1 ◦C) (Fig. 4, line C). The reduction in the melting temperaturevalues can be related to the nanocrystalline size of lipids in SLN

D.A. Campos et al. / Colloids and Surfaces B

FR

smrm

Fw

F1

ig. 3. Micrograph (2000×) of Witepsol RA-SLNs by scanning electron microscopy.ow indicates an example of SLN.

ystems ([19,20]). Also, for the SLNs formulations (Fig. 5), endother-ic peaks with lower enthalpy (−40.3 to −17.5 J/g) and different

anging melting temperatures were visualized (Table 3). Thiseans different rearrangement on lipid re-crystallization, with

20 30 40 50 60

DS

C (m

W)

A B

41.72 ˚C

-21 .79 J/ g

41.15 ˚C

-70 .63 J/ g

ig. 4. Thermograms of the bulk materials, (A) rosmarinic acid, (B) Tween 80, (C) Witepsol

ith 1.0% Witepsol H15: 2% Tween 80.

20 30 40 50 60

DS

C (m

W)

A B C D

ig. 5. Thermograms of RA-SLNs formulations, with Witepsol H15% (w/v): Tween 80% (v/.5:3.

: Biointerfaces 115 (2014) 109–117 115

lower energy being necessary to melt the lipid [18]. The decreaseof the melting temperature and enthalpy can be attributed to thepresence of surfactant and also to the colloidal low dimensions ofthe particles, in particular to their high surface area to volume ratio[21]. For this purpose correlation values were calculated betweenmelting temperatures, enthalpy values and the particle sizes. Asignificant and positive correlation was obtained between PS andmelting temperatures (ca. 0.62). This means that as higher the PShigher the melting temperatures, i.e. larger size particles need moretemperature to melt than the smaller ones, which is according withthe described elsewhere [19,20].

The CI% allows the understanding of the thermal behavior ofmaterials, and is directly correlated with compound incorporationand release rate, where thermal behavior is different for pure lipidand the SLNs [22,23]. As can be seen in Eq. (2), the percentage is cal-culated involving the enthalpy value of pure lipid (normally high)and the value of the SLNs (value normally lower than the pure com-pound). The decrease of enthalpy of the melting transition can berelated to the crystallinity of the SLNs matrix. Hence, high values

of CI% lead to faster compound release, but it also means that moreenergy is required to melt the crystal lipids. The ideal value of CI%should be sufficiently high to make sure that most of the SLNs arenot unstable in the formation of new particles, but it also has to be

70 80 90 100

Temperature (°C)

C D

H15 and (D) standard formulation without RA and without cryoprotectant produced

70 80 90 100

Temper ature (°C)

E F G H I

v), (A) 0.5:1, (B) 0.5:2, (C) 0.5:3, (D) 1.0:1, (E) 1.0:2, (F) 1.0:3, (G) 1.5:1, (H) 1.5:2, (I)

116 D.A. Campos et al. / Colloids and Surfaces B: Biointerfaces 115 (2014) 109–117

Table 3Mean values ± SD of change in enthalpy (�H) and values of melting temperature and percentage of crystallinity (CI%) of nanoparticles after production.

Thermal properties A B C D E F G H I

�H (J/g) −38 ± 6 −18 ± 0 −20 ± 8 −24 ± 8 −21 ± 0 −40 ± 0 −35 ± 0 −24 ± 2 −32 ± 12Melting temperature (◦C) 35.74 38.50 36.55 35.53 37.03 37.12 38.11 35.52 36.04CI% 90.27 49.47 55.36 34.13 30.85 57.07 33.14 22.68 29.79

Designations of the letters are as follows (Witepsol H15%, w/v: Tween 80%, v/v): (A) 0.5:1, (B) 0.5:2, (C) 0.5:3, (D) 1.0:1, (E) 1.0:2, (F) 1.0:3, (G) 1.5:1, (H) 1.5:2, (I) 1.5:3.

7009001100130015001700

Wavenumber cm-1

A B C D E

F Polysa g (E)

svmwatattCdwnuo

3

pFSatagriiitbmwpw

ig. 6. FTIR spectra of (A) rosmarinic acid, (B) SLNs formulation (0.5% Witepsol: 1%cid, (D) SLNs formulation (0.5% Witepsol: 2% Polysorbate 80) and the correspondin

ufficiently low to ensure the release of RA. The most appropriatealues are those close to 50%. The calculated CI% for the SLNs for-ulations are shown in Table 3, and as can be seen the lower valuesere obtained for the SLNs produced with the higher lipid percent-

ge, which means that at this concentration the lipid interferes withhe chemical stability of SLNs. This is confirmed by the statisticalnalysis performed for CI% values at 0 d of SLNs. By inspection ofhe pareto charts (Fig. 1e), only the lipid concentration is of sta-istical significance and it affects negatively the values of CI%, i.e.I% decreases with an increase in lipid concentration. These ten-encies were confirmed by the response surface analysis (Fig. 2e)hich enabled calculation of the necessary conditions to obtain a CIear 50%, i.e. a concentration of lipid lower than 1.1% (w/v) must besed, while the surfactant concentration will not have any impactn this property.

.3. FTIR

The three bands at 1605, 1520 and 1445 cm−1 are due to theresence of aromatic ring stretching, as can be seen in the graphs forTIR spectra (Fig. 6) of RA (line A) and physical mixtures betweenLNs and RA (lines C and E) [24]. The spectra from SLNs (lines Bnd D) did not present these peaks, probably due to the connec-ions between the matrix and the reactive connections O H ofromatic rings of RA. Further evidences for presence of phenolicroups were delivered through the bands at 1360 and 1180 cm−1

esulting from O H and C O stretches; these peaks were presentedn all spectra, indicating the presence of phenolic groups and thencorporated RA. Therefore, an overlap of two bands in the regions very likely. Carboxylic acid groups show a characteristic band inhe range 1725–1700 cm−1 for RA and for both physical mixturesetween the SLNs and the RA (lines C and E); these results reflect the

aintenance of the RA structure. The lipid as well as the surfactantere directly involved in the loading of the RA. The lipid by entrap-ing the RA and the surfactant by stabilizing the particles. When RAas encapsulated, these characteristic peaks generally disappear,

orbate 80), the corresponding (C) physical mixture of SLNs (0.5:1) and rosmarinicphysical mixture of SLNs (0.5:2%) and rosmarinic acid.

mostly due to chemical interactions between the reactive groups ofRA and the lipid matrix; hence these typical bands probably weremasked by the matrix and proving definitively the physical andnot chemical interaction between the lipid and polyphenol. Spec-tra also revealed a band at 1750 cm−1, which is an adsorption bandcharacteristic of an ester formation at 1750–1725 cm−1; sometimesthis band may appear at lower wavelength numbers 1700 cm−1 inpresence of aromatic structures [24,25].

In addition, a correlation between the loading of the polyphenoland ZP could also support the idea of physical and not chem-ical interaction between lipid and RA. According to Shukat andRelkin [26], the incorporated compound (antioxidant) could bedirectly correlated to ZP values. The incorporation of such com-pounds leads to the increment of 10 mV to the absolute ZP value,increasing the stability of the system. But, in our work, there wereno differences between the non-loaded SLNs and RA-SLNs (datanot shown), showing no influence on the stability of the SLNs. Asalready mentioned, these findings contribute to the hypothesis thatthe interaction of the lipid and RA is physical and not chemical.

4. Conclusions

Based on all the results achieved and the analysis thereof it ispossible to conclude that, in general, the percentage of lipid andsurfactant used were important factors for the production of RA-SLNs influencing their physical properties; in fact, the combinationsof these parameters lead to a certain particle size and influencedthe stability of SLNs. FTIR analysis enabled, for the first time, con-firmation of the efficient encapsulation of RA in the SLNs, wherethe disappearance of the bands proves the physical associationof lipid with RA. The DSC analysis also confirmed the interactionbetween these compounds, via differences between the spectra of

SLNs and RA-SLNs: the values of enthalpy were minor for SLNs, yetwhen related with RA the values of enthalpy and melting temper-atures decreased due to the established connections and the filledspaces. Based on standard recommendations, it is required that all

faces B

tsootisaa

niNtm

omita

A

NOce2

[[[

[[

[[[[

[[[[[

D.A. Campos et al. / Colloids and Sur

ested RA-SLNs should have a PS ≥ 300 nm, because the RA-SLNshould not transpose the gut epithelium, as well as present a CI%f ca. 50% for an appropriate release. After performing the batteryf characterization tests response surface models were built withhe generated data and substitution of the above required outputsn the models allowed establishment of the most suitable lipid andurfactant concentrations. The optimum values to obtain the desir-ble features suggest formulations containing 0.5% (w/v) of lipidnd 1–2% (v/v) of surfactant.

When thinking in producing an ingredient to produce a oralutraceutical formulation with these systems, all tests concern-

ng the stability of them in model matrices have to be explored.evertheless, the stability tests during 28 d in refrigerated condi-

ions give us an approach that at least in solution, these particlesaintain their initial properties.The present work employed an approach toward the evaluation

f SLNs using a model compound commonly used by the phar-aceutical industry – Witepsol. There is a wide space for further

nvestigation in this area, since the studies on applications of nano-echnology in the food industry are still limited and quite recentnd need to be more explored.

cknowledgments

Partial funding for this research work was provided via projectANODAIRY (PTDC/AGR-ALI/117808/2010) and project PEst-

E/EQB/LA0016/2011, administrated by FCT (Fundac ão para a Ciên-ia e Tecnologia, Portugal). Author Ana Raquel Madureira acknowl-dges FCT for the post-doctoral scholarship SFRH/BPD/71391/010.

[

[

[

: Biointerfaces 115 (2014) 109–117 117

References

[1] R.H. Müller, K. Mäder, S. Gohla, Eur. J. Pharm. Biopharm. 50 (161) (2000).[2] R. Parhi, P. Suresh, J. Chem. Pharm. Res. 2 (2010) 211.[3] J. Weiss, P. Takhistov, J. McClements, J. Food Sci. 71 (107) (2006).[4] P. Severino, T. Andreani, A.S. Macedo, J.F. Fangueiro, M.H. Santana, A.M. Silva,

E.B. Souto, J. Drug Deliv. 2012 (1) (2011).[5] L.N. Prista, A.C. Alves, R.M.R. Morgado, Técnica Farmacêutica e Farmácia

Galénica, vol. 3, Fundac ão Calouste Gulbenkian, Lisbon, 1990.[6] R.H. Müller, D. Rühl, S.A. Runge, Int. J. Pharm. 144 (115) (1996).[7] D.J. McClements, E.A. Decker, Y. Park, Crit. Rev. Food Sci. Nutr. 49 (2009) 48.[8] S. Singh, A.K. Dobhal, A. Jain, J.K. Pandit, S. Chakraborty, Chem. Pharm. Bull. 58

(2010) 650.[9] L. Sek, C.J.H. Porter, A.M. Kaukonen, W.N. Charman, J. Pharm. Pharm. 54 (2002)

29.10] M.J. Parnham, K. Kesselring, Drugs Fut. 10 (1985) 756.11] E. Szabo, A. Thelen, M. Petersen, Plant Cell Rep. 18 (1999) 485.12] F.S. D’Amelio, Botanicals: A Phytocosmetic Desk Reference, CRC Press, London,

1999, pp. 361.13] M.D. Triplett, J.F. Rathman, Nanopart. Res. 11 (2009) 601.14] S. Martins, I. Tho, D.C. Ferreira, E.B. Souto, M. Brandl, Drug Dev. Ind. Pharm. 37

(815) (2011).15] H. Das, A. Chaudhury, Am. Assoc. Pharm. Sci. Technol. 12 (62) (2010).16] R.H. Müller, S. Heinemann, Clin. Nutr. 11 (1992) 223.17] V. Jenning, K. Mäder, S.H. Gohla, Int. J. Pharm. 205 (2000) 15.18] N.A.B. Rosa, Development of Nanoparticles Loaded with Bioactive Compounds

for Application as Nutraceuticals, Engineering Faculty of Oporto University,2011 (Integrated Master Dissertation).

19] K. Westesen, H. Bunjes, Int. J. Pharm. 115 (1995) 129.20] J. Jee, S. Lim, J. Park, C.J. Kim, Eur. J. Pharm. Biopharm. 63 (2006) 134.21] K. Vivek, H. Reddy, R.S. Murthy, AAPS PharmSciTech 8 (2007).22] A. Radomska-Soukhaver, Adv. Drug Deliv. Rev. 59 (2007) 411.23] A. Estella-Hermoso, M.A. Campanero, F. Mollinedo, M.J. Blanco-Príeto, J. Chro-

matogr. B 877 (2009) 4035.

24] J. Coates, in: R.A. Meyers (Ed.), A Practical Approach. Encyclopedia of Analytical

Chemistry, Wiley and Sons, Chichester, 2000, pp. 1–23.25] N.P. Roeges, A Guide to Complete Interpretation of Infrared Spectra of Organic

Structures, Wiley, Chichester, UK, 1994.26] R. Shukat, P. Relkin, Colloids Surf. B 86 (2011) 119.