- Email: [email protected]
hydroxypropyl-β-cyclodextrin system
Journal of Controlled Release 102 (2005) 71 – 83 www.elsevier.com/locate/jconrel
How cyclodextrin incorporation affects the properties of protein-loaded PLGA-based microspheres: the case of insulin/hydroxypropyl-h-cyclodextrin system Giuseppe De Rosaa, Domenico Larobinab, Maria Immacolata La Rotondaa,*, Pellegrino Mustoc, Fabiana Quagliaa, Francesca Ungaroa a
Dipartimento di Chimica Farmaceutica e Tossicologica, Universita` degli Studi di Napoli Federico II, Via D. Montesano 49, 80131 Napoli, Italy b Dipartimento di Ingegneria dei Materiali e delle Produzione, Universita` degli Studi di Napoli Federico II, P.le Tecchio 80, 80100 Napoli, Italy c Istituto di Chimica e Tecnologia dei Polimeri, Consiglio Nazionale delle Ricerche (CNR), Via Campi Flegrei 34, 80078 Pozzuoli (NA), Italy Received 16 April 2004; accepted 21 September 2004 Available online 11 November 2004
Abstract The aim of this work was to study the influence of cyclodextrin (CD) incorporation on the properties of protein-loaded poly(lactide-co-glycolide) (PLGA) microspheres, with particular regards to protein release kinetics. To this purpose, insulinloaded microspheres were prepared by spray-drying emulsion or solution formulations, with or without hydroxypropyl-hcyclodextrin (HPhCD), and fully characterized for encapsulation efficiency and release kinetics of both insulin and cyclodextrin. Homogeneous populations of spherical microparticles entrapping both insulin and HPhCD were obtained. In order to get an insight into insulin/HPhCD interactions occurring inside microspheres, Fourier transform infrared (FTIR) analysis in the Amide I region was performed. FTIR spectra of dried microspheres containing HPhCD showed a change in insulin secondary structure, attributed to the presence of insulin/HPhCD complexes within microspheres. Insulin release was affected by the presence of HPhCD depending on the initial formulation conditions. In the case of microspheres prepared from emulsion, cyclodextrin reduced only insulin burst, whereas in the case of microspheres obtained from solution, the overall insulin release rate was slowed down. Combining the release kinetics of HPhCD with the FTIR results on hydrated microspheres, it was concluded that the formation of insulin/HPhCD complexes inside microspheres is critical to decrease protein diffusivity in the polymer matrix and achieve an effective modulation of protein release rate. D 2004 Elsevier B.V. All rights reserved. Keywords: Cyclodextrins; Microspheres; FTIR; Release rate; Insulin
* Corresponding author. Tel.: +39 81 678634; fax: +39 81 678647. E-mail address: [email protected] (M.I. La Rotonda). 0168-3659/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2004.09.030
72
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
1. Introduction Poly(lactide-co-glycolide) (PLGA)-based microspheres are biodegradable particulate delivery systems providing both the protection of the drug, encapsulated inside a polymeric matrix, and its release at slow and continuous rate. Depending on microsphere preparation technique, drug molecules are either dispersed within the polymer or deposited inside spherical, or nearly spherical, occlusions (macropores) formed within the particle during processing. Upon immersing microspheres in an aqueous medium, water penetrates toward the centre of the particle (hydration phase) and activate drug diffusion through the innate micropores of PLGA (angstrom- or nanometerdimension) and the macroporous structure of the particle. In the case of macromolecules, the diffusion in the porous network is highly limited due to the cramped space available and so extremely slow until pores grow in size and/or coalesce because of polymer erosion. Thus, drug release rate from biodegradable PLGA microspheres is mainly controlled by polymer erosion, structure of the porous microenvironment and drug diffusion [1,2]. These factors can be generally regulated by selecting adequate formulation conditions, such as polymer type and preparation method [3]. Actually, the use of PLGAs differing for molecular weight and copolymer composition can change initial hydration and erosion rate of the matrix [4,5]. Much more difficult is to modify the release features of microspheres once a polymer type and a preparation technique have been selected. In this case, control over the release rate could be exerted by either modifying the internal morphology of the system (i.e., internal porosity) or adding a third component that alters drug effective diffusivity in the polymeric matrix. Cyclodextrins (CDs) have been successfully used to modify the release features of polymeric systems mainly due to their capability of forming noncovalent complexes with drugs, that is a species with a different solubility and/or diffusivity [6,7]. CDs can form complexes also with proteins by including hydrophobic side chains inside their cone-shaped cavity, thus affecting three-dimensional structure and chemical/biological properties of the macromolecule [8]. In the light of these considerations, complexation with CDs can offer an additional tool to modulate protein
release rate from PLGA microspheres. Since complex formation is critical to this purpose, a careful investigation of protein/cyclodextrin interactions— that is CD effect on protein conformation—occurring inside microspheres is necessary. Among different experimental and theoretical methods aimed to the recognition of protein secondary structure, Fourier transform infrared (FTIR) spectroscopy has been demonstrated as a useful technique to collect detailed information on protein conformation within microspheres, without altering the controlled release system [9,10]. The aim of this work was to study the influence of hydroxypropyl-h-cyclodextrin (HPhCD) incorporation on the properties of insulin-loaded PLGA microspheres, with particular regard to protein release kinetics. In order to affect the internal morphology of the system, insulin-loaded microspheres were produced by spray-drying different liquids, namely W/O emulsions and solutions. Different microsphere formulations with or without HPhCD were obtained and fully characterized for encapsulation efficiency and release kinetics of both insulin and HPhCD. For a better understanding of the mechanisms by which the coencapsulation of HPhCD affects protein release rate, protein/cyclodextrin interactions within dried and hydrated particles were assessed by FTIR spectroscopy.
2. Materials and methods 2.1. Materials Poly(d,l-lactide-co-glycolide) (50:50) (PLGA, Resomer RG 504 H; Mw 50 KDa; inherent viscosity 0.5 dl/g) was purchased from Boehringer Ingelheim (Germany). Insulin from bovine pancreas, insulin from bovine pancreas FITC-labeled, trizma base (TRIS), trifluoroacetic acid (TFA), sodium azide and 1,4-diazabicyclo-[2.2.2]octane (DABCO) were obtained from Sigma (USA). Hydroxypropyl-h-cyclodextrin (HPhCD, Mw 1380, molar substitution 0.6), phenolphthalein, polysorbate 80, and polyvinylalcohol (PVA, MowiolR 40–88) were purchased from Aldrich (USA). Analytical grade methylene chloride (MC), glacial acetic acid and methanol, HPLC grade acetonitrile were supplied by Carlo Erba (Italy).
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
2.2. Microsphere preparation Insulin-loaded microspheres were produced by spray-drying slightly modifying a reported procedure [11]. Microspheres were prepared at the insulin theoretical loading of 2.5% (2.5 mg of insulin per 100 mg of polymer) by atomization of a W/O emulsion or a solution in glacial acetic acid in a Mini Spray-Dryer Bqchi 190 (Flawil, Switzerland). For the preparation of the W/O emulsion, 10 ml of an insulin solution (2.5 mg/ml) in 30% (v/v) acetic acid were dispersed in 50 ml of MC containing 1 g of PLGA by a high-speed mixer (Diax 900, Heidolph, Germany) operating at 10,000 rpm for 3 min (tool 10G). For the preparation of the solution in glacial acetic acid, 25 mg of insulin and 1 g of PLGA were dissolved in 50 ml of glacial acetic acid. In the case of the emulsion, HPhCD was coencapsulated by addition into the aqueous phase, whereas for the solution, HPhCD was dissolved with insulin and PLGA in acetic acid (insulin/HPhCD molar ratio was about 1:4). Emulsions and solutions were individually spray-dried with the following process parameters: feed rate 2 ml/min; aspirator setting 20; spray-flow 600 Nl/h; inlet temperature 42 8C. A 0.5-mm nozzle was used throughout the experiments. Microspheres were collected and freeze-dried for 24 h (0.01 atm, 60 8C; Modulyo, Edwards, UK). 2.3. Microsphere morphology and size Microsphere shape and morphology were analysed by Scanning Electron Microscopy (SEM; Leica S440, Germany). The samples were stuck on a metal stub and coated with gold under vacuum. The mean diameter and size distribution were determined by laser light scattering (Coulter LS 100Q, USA) on a dispersion of freeze-dried microspheres in 0.2% w/v aqueous PVA. Particle size is expressed as volume mean diameterFS.D. of values collected from three different batches. 2.4. Insulin distribution inside microspheres Insulin distribution inside microspheres was observed by confocal laser scanning microscopy. A mixture of insulin and FITC-labeled insulin (10:1 w/ w) was encapsulated. Microspheres containing FITC-
73
labeled insulin were suspended in 10% w/v DABCO aqueous solution and mounted with ConfocalMatrixR (Micro-Tech-Lab, Germany). Insulin distribution inside microspheres was investigated using an LSM 510 Zeiss confocal inverted microscope (Axiovert 100 M) equipped with a Zeiss 63X/1.2 NA oil immersion objective lens (C-Apochromat). 2.5. Insulin analysis Insulin quantitative analysis was carried out by a previously reported Reverse Phase High-Performance Liquid Chromatography (RP-HPLC) method [12]. The chromatograph (Shimadzu, Kyoto, Japan) was equipped with an HPLC LC-10AD pump, a 7725i injection valve (Rheodyne), a SPV-10A UV-vis detector set at the wavelength of 220 nm and a CR6 integrator. A Jupiter 5-Am C18 column (2504.6 mm, 300 2; Phenomenex, Torrance, CA) was employed. The mobile phase was a mixture of water containing 0.1% v/v of TFA and acetonitrile containing 0.1% v/v of TFA (70:30 v/v). The flow rate was 1 ml/min. The chemical stability of insulin was assessed by HPLC. Desamido insulin was detected by the abovereported RP-HPLC method, whereas soluble covalent aggregates were detected by size-exclusion chromatography (SE-HPLC) [12]. Briefly, a Biosep SEC-S 2000 column (3007.8 mm; Phenomenex) was used for the analysis. The mobile phase was a mixture of acetonitrile and 2.5 M acetic acid containing 4 mM larginine (96:4 v/v). The flow rate was 1 ml/min and the detection was performed at 280 nm. 2.6. Hydroxypropyl-b-cyclodextrin analysis HPhCD was quantified by spectrophotometrically analysing the fading of a phenolphthalein alkaline solution in the presence of the complexing agent. The phenomenon is due to the formation of the colourless stable inclusion complex phenolphthalein/HPhCD (molar ratio 1:1) and is directly related to the amount of cyclodextrin added to the solution [13]. In this work, the experimental conditions were slightly changed to ensure reproducibility and accuracy of the quantitative analysis of HPhCD into both media used to characterize microspheres, that is water for actual loadings and 0.05 M TRIS buffer at pH 7.4
74
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
(containing 0.05% w/v sodium azide as preserving agent and 0.02% v/v polysorbate 80 as dispersing agent) for release kinetics (mixture indicated as brelease mediumQ in the following). A stock phenolphthalein solution 3 mM in methanol was diluted 1:100 in 0.05 M carbonate buffer at pH 10.5 just prior to use. Into a test tube, 2.6 ml of phenolphthalein working solution were added to 400 Al of a HPhCD sample. The absorbance at 553 nm (phenolphthalein k max) of the resulting solution was measured just after mixing by means of a single-beam UV-vis spectrophotometer (model 1204 Shimadzu) fitted out with 1-cm quartz cell. In the control assay, HPhCD sample was replaced by an equal volume of water or release medium and the reference absorbance at 553 nm recorded. All measurements were performed in triplicate at room temperature against a reagent blank. Standard solutions within the range 0.01–0.20 mg/ml were prepared either in water or release medium and calibration curves were obtained plotting the decrease in absorbance of the control versus HPhCD concentration. In both case, Beer’s law was verified over this concentration range (R 2N0.99). The limits of quantitation (QOD) were 0.01 and 0.02 mg/ml in water and release medium, respectively. To validate daily the performance of the analytical method, the absorbance values of HPhCD standards in both water and release medium were evaluated. 2.7. Actual loadings Insulin content within the microspheres was measured as previously reported [11]. Briefly, 5 mg of microspheres were dissolved into 0.8 ml of an acetonitrile/water solution (9:1 v/v) and insulin was extracted into 2 ml of 0.05 N hydrochloric acid. The suspension was centrifuged (at 5000 rpm, room temperature for 15 min) and the supernatant analysed for insulin content by RP-HPLC. In order to exclude the presence of insulin fibrils inside particles, the microspheres were washed three times with methylene chloride in order to dissolve PLGA. The solid residue was recovered and redissolved in 0.05 N HCl. Turbidity of the solution was measured by UVvis spectrophotometry at 350 nm as previously reported [11].
The loading of HPhCD inside microspheres was determined by solvent extraction. Five milligrams of microspheres were firstly placed in 1 ml of MC to dissolve the polymer and then extracted with 1 ml of water. Cyclodextrin, extracted in the aqueous phase, was quantified as described above. Control experiments were run to verify the ability of the method in extracting HPhCD from microspheres. A known amount of HPhCD (about 1 mg) was suspended in 1 ml of MC containing PLGA (5 mg/ml), extracted in 1 ml of water and quantified after complexation with phenolphthalein. The experiment was run in triplicate and the recovery was found to be 91F3%. Both insulin and HPhCD loadings are expressed as actual loading percentage (mg encapsulated per 100 mg of microspheres). 2.8. In vitro release studies Insulin release was evaluated by suspending 3 mg of dried microspheres in 2 ml of 0.05 M aqueous TRIS buffer at pH 7.4 (containing 0.05% w/v sodium azide as preserving agent and 0.02% v/v polysorbate 80 as dispersing agent) in a thermostatic bath at 37 8C. At scheduled time intervals, the release medium was withdrawn and microspheres analysed for their insulin content as described in Section 2.7. In order to assess aggregation state of insulin released from microspheres, the medium was withdrawn at 24 h, filtered and analysed by circular dichroism as previously reported [11]. Protein quality inside microspheres was assessed by RP-HPLC and SE-HPLC as described in Section 2.5. Experiments were run in triplicate for each point of release kinetics. HPhCD release was evaluated by suspending 10 mg of dried microspheres in 1.5 ml of release medium. Quadruplicate samples for each batch were placed in a thermostatic bath at 37 8C. At scheduled time intervals, 0.5 ml of the release medium were withdrawn and replaced with the same volume of fresh medium. Samples were analysed for HPhCD content as described above. 2.9. FTIR spectroscopy 2.9.1. IR spectroscopy FTIR spectra of insulin solutions with or without cyclodextrin were collected in the attenuated total
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
2.9.2. IR data analysis To enhance the apparent resolution of the highly overlapped Amide I region (1700–1590 cm 1), the Fourier self-deconvolution (FSD) method was applied to the experimental band profiles prior to curve-fitting analysis. The FSD approach, based on the original formulation of Kauppinen et al. [14,15], was performed using a Lorenzian deconvolution function; smoothing was achieved by applying a Bessel truncation function to the data in the Fourier
(i.e., interferogram) domain. The characteristic parameters of the two above functions were determined as follows: the chosen frequency range was truncated and, keeping fixed the degree of smoothing at a value of 50%, the width of the Lorentzian deconvolution function was increased stepwise from 4 cm 1 upward. The procedure was iterated up to the appearance of negative lobes in the spectrum, which indicate that the line-width being used exceeded that of the narrowest component. At this point, the smoothing degree was gradually reduced down to a point where the noise started to increase abruptly, which represents the optimum value of smoothing. The procedure followed for the analysis of spectra prior to curve fitting is illustrated in Fig. 1. After the application of the FSD algorithm, the individual peaks were eventually separated out by a least-squares curve-fitting program based on the Levenberg–Marquardt method [16]. To reduce the number of adjustable parameters and to insure the uniqueness of the result, the baseline, the band shape and the number of components were fixed. The minimum number of components and their initial position were estimated by taking the second derivative spectrum of the experimental profiles. The curve-fitting algorithm was then allowed to calculate
A Absorbance
reflection mode (ATR) with a System 2000 FTIR spectrometer (Perkin-Elmer, Norwalk, CT) equipped with a wide-band DTGS detector and a Germanium/ KBr beam splitter. FTIR analyses were carried out at a resolution of 4 cm 1 using a multiple reflection ATR accessory (Benchmark from Specac, UK) with horizontal geometry equipped with a KRS-5 crystal (angle of incidence=458, number of reflections=6). The solution samples contained insulin (0.2 mg/ml) and HPhCD (4 mg/ml) in 10 mM potassium phosphate buffer at pH 7.4. FTIR spectra of microsphere samples were obtained in transmission mode with a rapid scanning infrared microscope (Autoimage, Perkin-Elmer) connected to a System 2000 interferometer also from Perkin-Elmer. The system is equipped with a video camera for the collection and storage of visual images and uses a Cassegrain objective and a 6 permanently aligned duplex Cassegrain condenser to couple IR and visible images. Detection of the IR radiation is achieved by a narrow-band MCT detector, cooled with liquid nitrogen. Localized acquisition was performed with the use of a 5050 Am upper variable aperture and a lower variable aperture of the same dimensions. One hundred fifty spectra were signal averaged in the conventional manner to improve the signal-to-noise ratio. The spectra were collected on samples prepared by flattening powder particles with a compression press (Carver, Wabash, IN) at room temperature to obtain samples of around 100 Am in diameter. The analysis was performed on dried and hydrated microspheres. In the case of hydrated samples, microspheres were incubated in water for 6 h, the suspension filtered under vacuum on a 0.45-Am acetate cellulose filter (Millipore, Bedford, MA) and the solid residue compressed as described above.
75
B
C D 1700
1660
1620
1580
1540
Wavenumber (cm-1)
1500
Fig. 1. Analysis of FTIR spectra prior to curve fitting: (A) as collected spectrum; (B) spectrum of neat PLGA microspheres; (C) subtraction spectrum A–B; (D) Fourier self-deconvoluted spectrum. As an example, the case of Sol/CD dried microspheres is shown.
76
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Table 1 Composition and overall characteristics of microspheres Formulation
Feeding
Additive
Production yield (%)
Mean size (AmFS.D.)a
Insulin loading (%FS.D.)b
HPhCD loading (%FS.D.)b,c
Em Em/CD Sol Sol/CD
emulsion emulsion solution solution
– HPhCDd – HPhCDd
42 51 40 50
13.9F1.2 15.2F1.3 10.1F1.0 9.8F0.9
2.44F0.20 2.75F0.07 2.74F0.07 2.45F0.08
– 2.28F0.11 – 2.47F0.06
a b
c d
Volume mean diameterFS.D. have been calculated on three different batches. Insulin and HPhCD actual loadings are expressed as mg of encapsulated substance per 100 mg of microspheres. The theoretical loadings were 2.5% (2.5 mg per 100 mg of microspheres). Values are corrected for recovery (91F3%). Insulin/HPhCD theoretical mole ratio was 1:4.
the height, the full width at half height (FWHH) and the position of the individual components. The peak function was a mixed Gauss–Lorentz line shape.
3. Results and discussion The aim of this study was to investigate how HPhCD incorporation affects the properties of insulin-loaded PLGA microspheres, with particular regard to protein release kinetics. This effect was studied on
microspheres prepared by spray-drying different feeding liquids (i.e. W/O emulsion and solution) and therefore differing in the internal porosity. Composition and overall characteristics of the prepared microspheres are reported in Table 1. Monodispersed microspheres with mean size ranging between 9.8 and 15.2 Am were prepared. The incorporation of HPhCD into the formulation did not appreciably affect size. As can be seen in Fig. 2, microspheres were regularly shaped and no insulin crystal was observed, thus suggesting an effective
Fig. 2. SEM micrographs of insulin-loaded PLGA microspheres.
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
77
Fig. 3. Confocal microscopy analysis of microspheres containing FITC-labeled insulin: (A) emulsion formulations; (B) solution formulations.
peptide entrapment inside the polymeric matrix. Nevertheless, microspheres prepared from emulsion and containing HPhCD were unexpectedly porous (Fig. 2B). It is reasonable that the presence of HPhCD in the aqueous phase modified solvent evaporation rate, which is generally recognised as a factor controlling surface morphology of microspheres produced by spray-drying [17]. Insulin distribution inside microspheres was assessed by confocal microscopy (Fig. 3). In both emulsion and solution formulations, insulin was well distributed within the polymeric matrix, and low amounts of drug were located near the surface. As expected, the internal morphology of microspheres depended on the initial formulation. When microspheres were prepared from emulsions (Fig. 3A), insulin was mainly located inside macropores distributed in the matrix, and originated from the aqueous phase of the emulsion during microsphere formation. In the case of the atomization of a continuous phase (i.e., solution formulations), a homogeneous insulin dispersion in the polymeric matrix was achieved (Fig. 3B). For all formulations, both insulin and HPhCD were effectively loaded inside microspheres (Table 1), so that the initial insulin/cyclodextrin molar ratio was kept. In agreement with our previous results [11], insulin was encapsulated in its native state. The results of the RP-HPLC analysis showed that desamido insulin content of microspheres was always lower than 5%, that is a value commonly present as impurity in the commercial insulin (as stated by USP 26). No covalent aggregate soluble in the extraction medium was detected by SE-HPLC. Furthermore, the presence
of insulin fibrils inside microspheres was evaluated by washing out the polymer and dispersing the residue in a medium unable to dissolve fibrils [11]. The resulting solution gave a transmittance value of 99F4%, which was considered indicative that insulin did not fibrillate inside microspheres. FTIR analyses in the Amide I region of insulin solutions with or without cyclodextrin are reported in Fig. 4. As can be seen, the addition of HPhCD changed insulin spectrum profile, indicating the occurrence of insulin/HPhCD interactions. Actually, previous NMR spectroscopy studies [18] have suggested that an interaction between insulin and hcyclodextrin takes place in solution, involving at least four sites on the monomeric form of insulin. The FTIR
Fig. 4. Fourier self-deconvoluted spectra in the Amide I region of aqueous solutions containing: (i) insulin (dashed line); (ii) insulin and HPhCD (continuous line).
78
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
spectra of dried microspheres, along with the curvefitting results, are reported in Fig. 5. In the case of dried samples, irrespective of the initial formulation, no significant spectral difference could be observed between Sol and Em formulations containing only insulin (Fig. 5A–C), whereas differences in the spectrum profiles are apparent between microspheres with or without HPhCD (Fig. 5A–D). In fact, the lowintensity shoulder at 1630 cm 1 present in the formulations without cyclodextrin turned into a peak of considerable intensity at the same wave number when HPhCD was coencapsulated inside microspheres. The actual changes of insulin conformation inside microspheres upon addition of HPhCD could be monitored after analysis of FTIR spectra by the leastsquares curve-fitting program described in the experimental section. The spectral profiles of the dried
0.20
A
0.15
0.20 0.15
0.10
0.10
0.05
0.05
0.00
0.00 1700
0.40
B
Absorbance
Absorbance
0.25
samples were suitably simulated by the sum of six Lorentz/Gauss functions (Fig. 5) on the basis of second-derivative analysis. The spectral parameters resulting from the analysis—position, FWHH and area of the peaks—are reported in Table 2. According to the literature, each component identified by curve-fitting analysis was tentatively assigned to the stretching vibration of specific carbonyl groups located in different physical environments, depending on protein secondary structure. After peak assignment, assuming the invariance of the molar absorptivity of the various carbonyl components, absorbance ratios were directly converted into concentration ratios, so as to obtain the percentage of the various amide species (a-helix, h sheet/turn, random coil) with respect to the total amide concentration [19]. In the numerous studies on FTIR characterization of the secondary structure of insulin,
1680
1660
1640
1620
Wavenumber (cm-1)
1700
1600
1680
1660
1640
1620
1600
Wavenumber (cm-1)
C
D
0.30
Absorbance
Absorbance
0.30
0.20
0.20
0.10
0.10 0.00 1700
0.00 1680
1660
1640
1620
Wavenumber (cm-1)
1600
1700
1680
1660
1640
1620
1600
Wavenumber (cm-1)
Fig. 5. FTIR spectra of insulin within dried PLGA microspheres in the Amide I region and curve-fitting results: (A) Em; (B) Em/CD; (C) Sol; (D) Sol/CD. The individual Lorentz/Gauss components are shown as symmetrical peaks underneath the FTIR spectra.
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
79
Table 2 Curve-fitting data relative to FTIR analysis of insulin within dried microspheres Formulation
Position (cm 1)
FWHHa (cm 1)
Area (cm 1)
R2
Tentative assignmentb
C i/C tot (%)c
Em
1692 1677 1659 1643 1630 1616
10.0 17.0 19.8 16.2 13.9 14.1
0.63 2.32 5.36 3.24 1.45 0.74
0.9964
h-sheet h-turn a-helix h-sheet h-sheet Side chains
4.9 17.8 41.2 25.0 11.1 –
Em/CD
1692 1677 1660 1643 1631 1617
9.5 15.8 19.8 16.2 13.8 14.1
0.39 1.51 3.93 2.61 1.72 0.51
1.0000
h-sheet h-turn a-helix h-sheet h-sheet Side chains
3.9 14.9 38.6 25.7 16.9 –
Sol
1692 1678 1660 1642 1629 1616
9.7 15.8 19.8 16.2 13.8 14.1
0.92 3.08 8.30 4.89 2.61 0.51
0.9979
h-sheet h-turn a-helix h-sheet h-sheet Side chains
4.6 15.6 41.9 24.7 13.2 –
Sol/CD
1692 1677 1660 1643 1631 1618
8.2 15.8 19.8 16.2 13.8 14.1
0.52 2.34 6.90 5.05 3.42 0.96
1.0000
h-sheet h-turn a-helix h-sheet h-sheet Side chains
2.9 12.9 37.8 27.7 18.8 –
a b c
Full weight at half height. According to Refs. [20–23]. Percentage of total area with side chain subtracted.
there is a general consensus about the assignment of the peak at 1658–1659 cm 1 to the a-helix structure [20–22]. The h-structures give rise to a more complex pattern arising from h-sheet and h-turn secondary structures. The multiplicity of components observed in the vibrational spectrum of these structures reflects a wider distribution of force constants within the different local environments and/or a stronger dependence of the force field from geometrical factors. The antiparallel h-sheet produces two components, one at low frequency, around 1630 cm 1, and one at high frequency, that appears at around 1690 cm 1. Another component at around 1642–1643 cm 1 has been attributed to parallel h-sheet [21–23]. The various bands appearing between 1660 and 1700 cm 1 are commonly assigned to different kinds of h-turns in a nonspecific way [23]. Finally, several authors assigned the component at around 1615 cm 1 to side chains
[19,21]. The presence of this spectral feature usually complicates the analysis of the Amide I region; in the present case, the relative contribution of 1615 cm 1 was neglected in the quantitative analysis. Upon addition of HPhCD within PLGA microspheres, the component at around 1630 cm 1 increased by a factor of 1.5 (e.g., 16.9% versus 11.1% in the case of emulsion formulation; Table 2), suggesting a rearrangement of insulin h-structures. This effect is consistent with previous NMR spectroscopy studies indicating that cyclodextrin can interact with insulin residues adopting an extended h-strand conformation [18]. The release profiles of insulin and HPhCD from all formulations are reported in Figs. 6 and 7 as insulin and HPhCD released over time. The release of insulin was a two-stage process where a first more rapid release stage (burst) was followed by a slower
80
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Fig. 6. Release profiles of insulin from emulsion (A) and solution (B) formulations evaluated in buffer at pH 7.4 and 37 8C.
second phase (Fig. 6). Concerning the burst, Em formulation displayed the highest burst (33F2%), whereas Em/CD microspheres resulted in a lower amount of insulin released in the first stage (18F3%). Sol and Sol/CD formulations showed a burst significantly lower than the formulations prepared from emulsions (8F1% and 6F2%, respectively). Concerning the second stage of insulin release, all formulations displayed a controlled release of insulin for more than 1 month. At 35 days, the amount of insulin released from Em and Sol formulations was about 94%, whereas Em/CD and Sol/CD formulations released about 80% and 66% of their insulin content, respectively. However, Em, Em/CD and Sol formulations released insulin at a similar rate, whereas release from Sol/CD occurred at a significantly lower rate. In each case, insulin
was released in its native conformation as indicated by circular dichroism data relative to insulin released after 24 h (the ratio between molar ellipticities at 223 and 208 nm was 0.8 as found for native insulin). Insulin was chemically stable inside microspheres along the release study since neither increased amounts of desamido insulin nor soluble insulin aggregates were detected in microsphere extracts. Insulin fibrils were not found inside microspheres as indicated by turbidimetry results at 35 days. Concerning HPhCD release, Em/CD and Sol/CD formulations displayed an HPhCD burst (Fig. 7) higher than the respective insulin burst. This effect can be reasonably attributed to a faster dissolution of HPhCD in the hydrating medium and its higher mobility in the matrix as compared to insulin. In particular, a two-fold higher burst was observed for Em/CD as compared to Sol/CD formulation (49F8% versus 25F1%, respectively), causing a depletion of cyclodextrin molecules inside microspheres. During the following sustained release stage, no difference in the HPhCD release rate was evident between Em/ CD and Sol/CD. Release studies have clearly shown that the modulator effect of HPhCD depends on the initial formulation conditions. In the case of Em/CD, cyclodextrin is able to reduce insulin burst, whereas in the case of Sol/CD, it slows down the overall insulin release rate. In order to elucidate this aspect, supplementary FTIR experiments were performed on hydrated microspheres at burst time. As shown in
Fig. 7. Release profiles of HPhCD from Em/CD and Sol/CD formulations evaluated in buffer at pH 7.4 and 37 8C.
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Fig. 8, the incorporation of HPhCD affected only the FTIR spectrum of Sol/CD formulation, where a higher absorption at about 1630 cm 1 was recorded (Fig. 8D). For hydrated microspheres, the spectral profiles were suitably simulated by the sum of eight Lorentz/Gauss functions. On the basis of literature data, the additional components at 1649 and 1668 cm 1 were assigned to unordered structures (random coils) and h-turns, respectively [20–23]. As shown in Table 3, significant differences in the relative concentration of the h-sheet component at 1630 cm 1 were observed only between solution formulations with or without cyclodextrin (i.e., 15.0% versus 11.9%), indicating that insulin/HPhCD interactions are expected to take place during the release stage only inside Sol/CD microspheres. Combining release kinetics of HPhCD with FTIR results and assuming that protein diffusion in the
81
polymeric matrix is a factor controlling the release, a hypothesis on the mechanism by which cyclodextrins modulate release rate can be advanced. The formation of insulin/CD complexes inside microspheres, as demonstrated by FTIR study on both dried and hydrated microspheres, likely reduces insulin mobility in the polymer through decrease of its effective diffusion coefficient. In the case of Em/CD, an early depletion of cyclodextrin from microspheres (HPhCD/insulin mole ratio decreases from 3.4 in dried microspheres to 1.7 in microspheres at burst) shifts insulin/CD complexation equilibrium toward free protein, which has a diffusion coefficient lower than complexed form. As a consequence, cyclodextrins reduce insulin release rate only in the very first stages of release. In the case of Sol/CD, a lower HPhCD burst generates a higher fraction of cyclodextrin molecules inside the polymeric matrix avail-
Fig. 8. FTIR spectra of insulin within hydrated PLGA microspheres in the Amide I region and curve-fitting results: (A) Em; (B) Em/CD; (C) Sol; (D) Sol/CD. The individual Lorentz/Gauss components are shown as symmetrical peaks underneath the FTIR spectra.
82
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Table 3 Curve-fitting data relative to FTIR analysis of insulin within hydrated microspheres Formulation
Position (cm 1)
FWHHa (cm 1)
Area (cm 1)
R2
Tentative assignmentb
C i/C tot (%)c
Em
1691 1678 1668 1658 1649 1642 1630 1616
11.9 13.2 10.0 13.6 7.6 11.8 11.4 8.2
0.98 2.08 1.37 3.99 0.68 2.09 0.71 0.54
0.9993
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
7.5 17.5 11.5 33.6 5.7 17.6 5.9 –
Em/CD
1689 1677 1668 1658 1649 1642 1631 1617
15.0 13.2 10.0 13.6 8.5 11.8 12.6 17.4
1.18 1.51 1.03 3.15 0.73 1.60 0.65 0.45
1.0000
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
12.0 15.3 10.4 32.0 7.4 16.3 6.6 –
Sol
1691 1677 1667 1658 1649 1642 1630 1618
10.6 14.4 10.0 13.5 8.1 11.7 12.0 19.6
0.85 2.45 1.35 3.91 0.79 2.04 1.54 1.01
0.9972
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
6.6 18.9 10.4 30.2 6.1 15.8 11.9 –
Sol/CD
1691 1677 1667 1658 1649 1642 1630 1615
10.6 14.4 10.0 13.6 8.5 11.7 12.0 20.0
0.60 1.84 1.07 3.08 0.67 1.51 1.55 1.04
0.9976
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
5.8 17.9 10.4 29.8 6.5 14.6 15.0 –
a b c
Full weight at half height. According to Refs. [20–23]. Percentage of total area with side chain subtracted.
able for insulin complexation (HPhCD/insulin mole ratio is 3.4 in Sol/CD microspheres at burst), which contributes to reduce protein mobility and slow down the overall release rate. Furthermore, it cannot be neglected that the lack of macroporosity—which is typical of microspheres produced from solution formulations—can affect the stability constant of insulin/CD complex in the hydrated microspheres and play a role in decreasing protein mobility through the polymeric matrix.
4. Conclusions In this work, we have investigated how cyclodextrins can affect the properties of insulin-loaded PLGA microspheres prepared by spray-drying. Release studies have shown that the modulator effect of HPhCD depends on the initial formulation conditions. FTIR analysis coupled with in vitro release studies of both insulin and HPhCD have pointed out that the formation of insulin/cyclodextrin complexes
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
inside microspheres is critical to decrease protein mobility in the polymer matrix and achieve an effective modulation of protein release rate.
Acknowledgements
[11]
[12]
The financial support of MIUR (PRIN 2002) is gratefully acknowledged. [13]
References [1] R.P. Batycky, J. Hanes, R. Langer, D.A. Edwards, A theoretical model of erosion and macromolecular drug release from biodegradable microspheres, J. Pharm. Sci. 86 (1997) 1464 – 1477. [2] V. Lemaire, J. Be´lair, P. Hildgen, Structural modeling of drug release from biodegradable porous matrices based on a combined diffusion/erosion process, Int. J. Pharm. 258 (2003) 95 – 107. [3] C. Washington, Drug release from microparticulate systems, in: S. Benita (Ed.), Microencapsulation, Marcel Dekker, New York, 1996, pp. 155 – 182. [4] G. Crotts, T.G. Park, Protein delivery from poly(lactic-coglycolic acid) biodegradable microspheres: release kinetics and stability issues, J. Microencapsul. 15 (1998) 699 – 713. [5] M.A. Tracy, K.L. Ward, L. Firouzabadian, Y. Wang, N. Dong, R. Qian, Y. Zhang, Factors affecting the degradation rate of poly(lactide-co-glycolide) microspheres in vivo and in vitro, Biomaterials 20 (1999) 1057 – 1062. [6] D.C. Bibby, N.M. Davies, I.G. Tucker, Mechanisms by which cyclodextrins modify drug release from polymeric drug delivery systems, Int. J. Pharm. 197 (2000) 1 – 11. [7] D. Duchene, D. Wouessidjewe, G. Ponchel, Cyclodextrins and carrier systems, J. Control. Release 62 (1999) 263 – 268. [8] T. Irie, K. Uekama, Cyclodextrins in peptide and protein delivery, Adv. Drug Deliv. Rev. 36 (1999) 101 – 123. [9] K. Fu, K. Griebenow, L. Hsieh, A.M. Klibanov, R. Langer, FTIR characterization of the secondary structure of proteins encapsulated within PLGA microspheres, J. Control. Release 58 (1999) 357 – 366. [10] T.H. Yang, A. Dong, J. Meyer, O.L. Johnson, J.L. Cleland, J.F. Carpenter, Use of infrared spectroscopy to assess secondary
[14]
[15]
[16] [17]
[18]
[19]
[20]
[21]
[22]
[23]
83
structure of human growth hormone within biodegradable microspheres, J. Pharm. Sci. 88 (1999) 161 – 165. F. Quaglia, G. De Rosa, E. Granata, F. Ungaro, E. Fattal, M.I. La Rotonda, Feeding liquid, non-ionic surfactants and cyclodextrin affect the properties of insulin-loaded poly(lactide-coglycolide) microspheres prepared by spray-drying, J. Control. Release 86 (2003) 267 – 278. Oliva, J.B. Farin˜a, M. Labre´s, Development of two highperformance liquid chromatographic methods for the analysis and characterization of insulin and its degradation products in pharmaceutical preparations, J. Chromatogr., B, Biomed. Sci. Appl. 749 (2000) 25 – 34. P.K. Zarzycki, H. Lamparczyk, The equilibrium constant of hcyclodextrin-phenolphtalein complex: influence of temperature and tetrahydrofuran addition, J. Pharm. Biomed. Anal. 18 (1998) 165 – 170. J.K. Kauppinen, D.J. Moffat, H.H. Mantsh, D.G. Cameron, Fourier self-deconvolution: a method for resolving intrinsically overlapped bands, Appl. Spectrosc. 35 (1981) 271 – 276. J.K. Kauppinen, D.J. Moffat, D.G. Cameron, H.H. Mantsh, Fourier transforms in the computation of self-deconvoluted and first-order derivative spectra of overlapped band contours, Anal. Chem. 53 (1981) 1454 – 1457. W.F. Maddams, The scope and limitations of curve fitting, Appl. Spectrosc. 34 (1980) 245 – 267. F.J. Wang, C.H. Wang, Sustained release of etanidazole from spray dried microspheres prepared by non-halogenated solvents, J. Control. Release 90 (2002) 263 – 280. F.L.A. Aachmann, D.E. Otzen, K.L. Larsen, R. Wimmer, Structural background of cyclodextrin–protein interactions, Protein Eng. 16 (2003) 905 – 912. C. Jung, Insight into protein structure and protein–ligand recognition by Fourier transform infrared spectroscopy, J. Mol. Recognit. 13 (2000) 325 – 351. A. Dong, P. Huang, W.S. Caughey, Protein secondary structures in water from second-derivative amide I infrared spectra, Biochemistry 29 (1990) 3303 – 3308. M.J. Pikal, D.R. Rigsbee, The stability of insulin in crystalline and amorphous solids: observation of greater stability for the amorphous form, Pharm. Res. 14 (1997) 1379 – 1387. L. Nielsen, S. Frokjaer, J.F. Carpenter, J. Brange, Studies of the structure of insulin fibrils by Fourier transform infrared (FTIR) spectroscopy and electron microscopy, J. Pharm. Sci. 90 (2001) 29 – 37. J.L.R. Arrondo, F.M. Goni, Structures and dynamics of membrane proteins as studied by infrared spectroscopy, Prog. Biophys. Mol. Biol. 72 (1999) 367 – 405.
How cyclodextrin incorporation affects the properties of protein-loaded PLGA-based microspheres: the case of insulin/hydroxypropyl-h-cyclodextrin system Giuseppe De Rosaa, Domenico Larobinab, Maria Immacolata La Rotondaa,*, Pellegrino Mustoc, Fabiana Quagliaa, Francesca Ungaroa a
Dipartimento di Chimica Farmaceutica e Tossicologica, Universita` degli Studi di Napoli Federico II, Via D. Montesano 49, 80131 Napoli, Italy b Dipartimento di Ingegneria dei Materiali e delle Produzione, Universita` degli Studi di Napoli Federico II, P.le Tecchio 80, 80100 Napoli, Italy c Istituto di Chimica e Tecnologia dei Polimeri, Consiglio Nazionale delle Ricerche (CNR), Via Campi Flegrei 34, 80078 Pozzuoli (NA), Italy Received 16 April 2004; accepted 21 September 2004 Available online 11 November 2004
Abstract The aim of this work was to study the influence of cyclodextrin (CD) incorporation on the properties of protein-loaded poly(lactide-co-glycolide) (PLGA) microspheres, with particular regards to protein release kinetics. To this purpose, insulinloaded microspheres were prepared by spray-drying emulsion or solution formulations, with or without hydroxypropyl-hcyclodextrin (HPhCD), and fully characterized for encapsulation efficiency and release kinetics of both insulin and cyclodextrin. Homogeneous populations of spherical microparticles entrapping both insulin and HPhCD were obtained. In order to get an insight into insulin/HPhCD interactions occurring inside microspheres, Fourier transform infrared (FTIR) analysis in the Amide I region was performed. FTIR spectra of dried microspheres containing HPhCD showed a change in insulin secondary structure, attributed to the presence of insulin/HPhCD complexes within microspheres. Insulin release was affected by the presence of HPhCD depending on the initial formulation conditions. In the case of microspheres prepared from emulsion, cyclodextrin reduced only insulin burst, whereas in the case of microspheres obtained from solution, the overall insulin release rate was slowed down. Combining the release kinetics of HPhCD with the FTIR results on hydrated microspheres, it was concluded that the formation of insulin/HPhCD complexes inside microspheres is critical to decrease protein diffusivity in the polymer matrix and achieve an effective modulation of protein release rate. D 2004 Elsevier B.V. All rights reserved. Keywords: Cyclodextrins; Microspheres; FTIR; Release rate; Insulin
* Corresponding author. Tel.: +39 81 678634; fax: +39 81 678647. E-mail address: [email protected] (M.I. La Rotonda). 0168-3659/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2004.09.030
72
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
1. Introduction Poly(lactide-co-glycolide) (PLGA)-based microspheres are biodegradable particulate delivery systems providing both the protection of the drug, encapsulated inside a polymeric matrix, and its release at slow and continuous rate. Depending on microsphere preparation technique, drug molecules are either dispersed within the polymer or deposited inside spherical, or nearly spherical, occlusions (macropores) formed within the particle during processing. Upon immersing microspheres in an aqueous medium, water penetrates toward the centre of the particle (hydration phase) and activate drug diffusion through the innate micropores of PLGA (angstrom- or nanometerdimension) and the macroporous structure of the particle. In the case of macromolecules, the diffusion in the porous network is highly limited due to the cramped space available and so extremely slow until pores grow in size and/or coalesce because of polymer erosion. Thus, drug release rate from biodegradable PLGA microspheres is mainly controlled by polymer erosion, structure of the porous microenvironment and drug diffusion [1,2]. These factors can be generally regulated by selecting adequate formulation conditions, such as polymer type and preparation method [3]. Actually, the use of PLGAs differing for molecular weight and copolymer composition can change initial hydration and erosion rate of the matrix [4,5]. Much more difficult is to modify the release features of microspheres once a polymer type and a preparation technique have been selected. In this case, control over the release rate could be exerted by either modifying the internal morphology of the system (i.e., internal porosity) or adding a third component that alters drug effective diffusivity in the polymeric matrix. Cyclodextrins (CDs) have been successfully used to modify the release features of polymeric systems mainly due to their capability of forming noncovalent complexes with drugs, that is a species with a different solubility and/or diffusivity [6,7]. CDs can form complexes also with proteins by including hydrophobic side chains inside their cone-shaped cavity, thus affecting three-dimensional structure and chemical/biological properties of the macromolecule [8]. In the light of these considerations, complexation with CDs can offer an additional tool to modulate protein
release rate from PLGA microspheres. Since complex formation is critical to this purpose, a careful investigation of protein/cyclodextrin interactions— that is CD effect on protein conformation—occurring inside microspheres is necessary. Among different experimental and theoretical methods aimed to the recognition of protein secondary structure, Fourier transform infrared (FTIR) spectroscopy has been demonstrated as a useful technique to collect detailed information on protein conformation within microspheres, without altering the controlled release system [9,10]. The aim of this work was to study the influence of hydroxypropyl-h-cyclodextrin (HPhCD) incorporation on the properties of insulin-loaded PLGA microspheres, with particular regard to protein release kinetics. In order to affect the internal morphology of the system, insulin-loaded microspheres were produced by spray-drying different liquids, namely W/O emulsions and solutions. Different microsphere formulations with or without HPhCD were obtained and fully characterized for encapsulation efficiency and release kinetics of both insulin and HPhCD. For a better understanding of the mechanisms by which the coencapsulation of HPhCD affects protein release rate, protein/cyclodextrin interactions within dried and hydrated particles were assessed by FTIR spectroscopy.
2. Materials and methods 2.1. Materials Poly(d,l-lactide-co-glycolide) (50:50) (PLGA, Resomer RG 504 H; Mw 50 KDa; inherent viscosity 0.5 dl/g) was purchased from Boehringer Ingelheim (Germany). Insulin from bovine pancreas, insulin from bovine pancreas FITC-labeled, trizma base (TRIS), trifluoroacetic acid (TFA), sodium azide and 1,4-diazabicyclo-[2.2.2]octane (DABCO) were obtained from Sigma (USA). Hydroxypropyl-h-cyclodextrin (HPhCD, Mw 1380, molar substitution 0.6), phenolphthalein, polysorbate 80, and polyvinylalcohol (PVA, MowiolR 40–88) were purchased from Aldrich (USA). Analytical grade methylene chloride (MC), glacial acetic acid and methanol, HPLC grade acetonitrile were supplied by Carlo Erba (Italy).
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
2.2. Microsphere preparation Insulin-loaded microspheres were produced by spray-drying slightly modifying a reported procedure [11]. Microspheres were prepared at the insulin theoretical loading of 2.5% (2.5 mg of insulin per 100 mg of polymer) by atomization of a W/O emulsion or a solution in glacial acetic acid in a Mini Spray-Dryer Bqchi 190 (Flawil, Switzerland). For the preparation of the W/O emulsion, 10 ml of an insulin solution (2.5 mg/ml) in 30% (v/v) acetic acid were dispersed in 50 ml of MC containing 1 g of PLGA by a high-speed mixer (Diax 900, Heidolph, Germany) operating at 10,000 rpm for 3 min (tool 10G). For the preparation of the solution in glacial acetic acid, 25 mg of insulin and 1 g of PLGA were dissolved in 50 ml of glacial acetic acid. In the case of the emulsion, HPhCD was coencapsulated by addition into the aqueous phase, whereas for the solution, HPhCD was dissolved with insulin and PLGA in acetic acid (insulin/HPhCD molar ratio was about 1:4). Emulsions and solutions were individually spray-dried with the following process parameters: feed rate 2 ml/min; aspirator setting 20; spray-flow 600 Nl/h; inlet temperature 42 8C. A 0.5-mm nozzle was used throughout the experiments. Microspheres were collected and freeze-dried for 24 h (0.01 atm, 60 8C; Modulyo, Edwards, UK). 2.3. Microsphere morphology and size Microsphere shape and morphology were analysed by Scanning Electron Microscopy (SEM; Leica S440, Germany). The samples were stuck on a metal stub and coated with gold under vacuum. The mean diameter and size distribution were determined by laser light scattering (Coulter LS 100Q, USA) on a dispersion of freeze-dried microspheres in 0.2% w/v aqueous PVA. Particle size is expressed as volume mean diameterFS.D. of values collected from three different batches. 2.4. Insulin distribution inside microspheres Insulin distribution inside microspheres was observed by confocal laser scanning microscopy. A mixture of insulin and FITC-labeled insulin (10:1 w/ w) was encapsulated. Microspheres containing FITC-
73
labeled insulin were suspended in 10% w/v DABCO aqueous solution and mounted with ConfocalMatrixR (Micro-Tech-Lab, Germany). Insulin distribution inside microspheres was investigated using an LSM 510 Zeiss confocal inverted microscope (Axiovert 100 M) equipped with a Zeiss 63X/1.2 NA oil immersion objective lens (C-Apochromat). 2.5. Insulin analysis Insulin quantitative analysis was carried out by a previously reported Reverse Phase High-Performance Liquid Chromatography (RP-HPLC) method [12]. The chromatograph (Shimadzu, Kyoto, Japan) was equipped with an HPLC LC-10AD pump, a 7725i injection valve (Rheodyne), a SPV-10A UV-vis detector set at the wavelength of 220 nm and a CR6 integrator. A Jupiter 5-Am C18 column (2504.6 mm, 300 2; Phenomenex, Torrance, CA) was employed. The mobile phase was a mixture of water containing 0.1% v/v of TFA and acetonitrile containing 0.1% v/v of TFA (70:30 v/v). The flow rate was 1 ml/min. The chemical stability of insulin was assessed by HPLC. Desamido insulin was detected by the abovereported RP-HPLC method, whereas soluble covalent aggregates were detected by size-exclusion chromatography (SE-HPLC) [12]. Briefly, a Biosep SEC-S 2000 column (3007.8 mm; Phenomenex) was used for the analysis. The mobile phase was a mixture of acetonitrile and 2.5 M acetic acid containing 4 mM larginine (96:4 v/v). The flow rate was 1 ml/min and the detection was performed at 280 nm. 2.6. Hydroxypropyl-b-cyclodextrin analysis HPhCD was quantified by spectrophotometrically analysing the fading of a phenolphthalein alkaline solution in the presence of the complexing agent. The phenomenon is due to the formation of the colourless stable inclusion complex phenolphthalein/HPhCD (molar ratio 1:1) and is directly related to the amount of cyclodextrin added to the solution [13]. In this work, the experimental conditions were slightly changed to ensure reproducibility and accuracy of the quantitative analysis of HPhCD into both media used to characterize microspheres, that is water for actual loadings and 0.05 M TRIS buffer at pH 7.4
74
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
(containing 0.05% w/v sodium azide as preserving agent and 0.02% v/v polysorbate 80 as dispersing agent) for release kinetics (mixture indicated as brelease mediumQ in the following). A stock phenolphthalein solution 3 mM in methanol was diluted 1:100 in 0.05 M carbonate buffer at pH 10.5 just prior to use. Into a test tube, 2.6 ml of phenolphthalein working solution were added to 400 Al of a HPhCD sample. The absorbance at 553 nm (phenolphthalein k max) of the resulting solution was measured just after mixing by means of a single-beam UV-vis spectrophotometer (model 1204 Shimadzu) fitted out with 1-cm quartz cell. In the control assay, HPhCD sample was replaced by an equal volume of water or release medium and the reference absorbance at 553 nm recorded. All measurements were performed in triplicate at room temperature against a reagent blank. Standard solutions within the range 0.01–0.20 mg/ml were prepared either in water or release medium and calibration curves were obtained plotting the decrease in absorbance of the control versus HPhCD concentration. In both case, Beer’s law was verified over this concentration range (R 2N0.99). The limits of quantitation (QOD) were 0.01 and 0.02 mg/ml in water and release medium, respectively. To validate daily the performance of the analytical method, the absorbance values of HPhCD standards in both water and release medium were evaluated. 2.7. Actual loadings Insulin content within the microspheres was measured as previously reported [11]. Briefly, 5 mg of microspheres were dissolved into 0.8 ml of an acetonitrile/water solution (9:1 v/v) and insulin was extracted into 2 ml of 0.05 N hydrochloric acid. The suspension was centrifuged (at 5000 rpm, room temperature for 15 min) and the supernatant analysed for insulin content by RP-HPLC. In order to exclude the presence of insulin fibrils inside particles, the microspheres were washed three times with methylene chloride in order to dissolve PLGA. The solid residue was recovered and redissolved in 0.05 N HCl. Turbidity of the solution was measured by UVvis spectrophotometry at 350 nm as previously reported [11].
The loading of HPhCD inside microspheres was determined by solvent extraction. Five milligrams of microspheres were firstly placed in 1 ml of MC to dissolve the polymer and then extracted with 1 ml of water. Cyclodextrin, extracted in the aqueous phase, was quantified as described above. Control experiments were run to verify the ability of the method in extracting HPhCD from microspheres. A known amount of HPhCD (about 1 mg) was suspended in 1 ml of MC containing PLGA (5 mg/ml), extracted in 1 ml of water and quantified after complexation with phenolphthalein. The experiment was run in triplicate and the recovery was found to be 91F3%. Both insulin and HPhCD loadings are expressed as actual loading percentage (mg encapsulated per 100 mg of microspheres). 2.8. In vitro release studies Insulin release was evaluated by suspending 3 mg of dried microspheres in 2 ml of 0.05 M aqueous TRIS buffer at pH 7.4 (containing 0.05% w/v sodium azide as preserving agent and 0.02% v/v polysorbate 80 as dispersing agent) in a thermostatic bath at 37 8C. At scheduled time intervals, the release medium was withdrawn and microspheres analysed for their insulin content as described in Section 2.7. In order to assess aggregation state of insulin released from microspheres, the medium was withdrawn at 24 h, filtered and analysed by circular dichroism as previously reported [11]. Protein quality inside microspheres was assessed by RP-HPLC and SE-HPLC as described in Section 2.5. Experiments were run in triplicate for each point of release kinetics. HPhCD release was evaluated by suspending 10 mg of dried microspheres in 1.5 ml of release medium. Quadruplicate samples for each batch were placed in a thermostatic bath at 37 8C. At scheduled time intervals, 0.5 ml of the release medium were withdrawn and replaced with the same volume of fresh medium. Samples were analysed for HPhCD content as described above. 2.9. FTIR spectroscopy 2.9.1. IR spectroscopy FTIR spectra of insulin solutions with or without cyclodextrin were collected in the attenuated total
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
2.9.2. IR data analysis To enhance the apparent resolution of the highly overlapped Amide I region (1700–1590 cm 1), the Fourier self-deconvolution (FSD) method was applied to the experimental band profiles prior to curve-fitting analysis. The FSD approach, based on the original formulation of Kauppinen et al. [14,15], was performed using a Lorenzian deconvolution function; smoothing was achieved by applying a Bessel truncation function to the data in the Fourier
(i.e., interferogram) domain. The characteristic parameters of the two above functions were determined as follows: the chosen frequency range was truncated and, keeping fixed the degree of smoothing at a value of 50%, the width of the Lorentzian deconvolution function was increased stepwise from 4 cm 1 upward. The procedure was iterated up to the appearance of negative lobes in the spectrum, which indicate that the line-width being used exceeded that of the narrowest component. At this point, the smoothing degree was gradually reduced down to a point where the noise started to increase abruptly, which represents the optimum value of smoothing. The procedure followed for the analysis of spectra prior to curve fitting is illustrated in Fig. 1. After the application of the FSD algorithm, the individual peaks were eventually separated out by a least-squares curve-fitting program based on the Levenberg–Marquardt method [16]. To reduce the number of adjustable parameters and to insure the uniqueness of the result, the baseline, the band shape and the number of components were fixed. The minimum number of components and their initial position were estimated by taking the second derivative spectrum of the experimental profiles. The curve-fitting algorithm was then allowed to calculate
A Absorbance
reflection mode (ATR) with a System 2000 FTIR spectrometer (Perkin-Elmer, Norwalk, CT) equipped with a wide-band DTGS detector and a Germanium/ KBr beam splitter. FTIR analyses were carried out at a resolution of 4 cm 1 using a multiple reflection ATR accessory (Benchmark from Specac, UK) with horizontal geometry equipped with a KRS-5 crystal (angle of incidence=458, number of reflections=6). The solution samples contained insulin (0.2 mg/ml) and HPhCD (4 mg/ml) in 10 mM potassium phosphate buffer at pH 7.4. FTIR spectra of microsphere samples were obtained in transmission mode with a rapid scanning infrared microscope (Autoimage, Perkin-Elmer) connected to a System 2000 interferometer also from Perkin-Elmer. The system is equipped with a video camera for the collection and storage of visual images and uses a Cassegrain objective and a 6 permanently aligned duplex Cassegrain condenser to couple IR and visible images. Detection of the IR radiation is achieved by a narrow-band MCT detector, cooled with liquid nitrogen. Localized acquisition was performed with the use of a 5050 Am upper variable aperture and a lower variable aperture of the same dimensions. One hundred fifty spectra were signal averaged in the conventional manner to improve the signal-to-noise ratio. The spectra were collected on samples prepared by flattening powder particles with a compression press (Carver, Wabash, IN) at room temperature to obtain samples of around 100 Am in diameter. The analysis was performed on dried and hydrated microspheres. In the case of hydrated samples, microspheres were incubated in water for 6 h, the suspension filtered under vacuum on a 0.45-Am acetate cellulose filter (Millipore, Bedford, MA) and the solid residue compressed as described above.
75
B
C D 1700
1660
1620
1580
1540
Wavenumber (cm-1)
1500
Fig. 1. Analysis of FTIR spectra prior to curve fitting: (A) as collected spectrum; (B) spectrum of neat PLGA microspheres; (C) subtraction spectrum A–B; (D) Fourier self-deconvoluted spectrum. As an example, the case of Sol/CD dried microspheres is shown.
76
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Table 1 Composition and overall characteristics of microspheres Formulation
Feeding
Additive
Production yield (%)
Mean size (AmFS.D.)a
Insulin loading (%FS.D.)b
HPhCD loading (%FS.D.)b,c
Em Em/CD Sol Sol/CD
emulsion emulsion solution solution
– HPhCDd – HPhCDd
42 51 40 50
13.9F1.2 15.2F1.3 10.1F1.0 9.8F0.9
2.44F0.20 2.75F0.07 2.74F0.07 2.45F0.08
– 2.28F0.11 – 2.47F0.06
a b
c d
Volume mean diameterFS.D. have been calculated on three different batches. Insulin and HPhCD actual loadings are expressed as mg of encapsulated substance per 100 mg of microspheres. The theoretical loadings were 2.5% (2.5 mg per 100 mg of microspheres). Values are corrected for recovery (91F3%). Insulin/HPhCD theoretical mole ratio was 1:4.
the height, the full width at half height (FWHH) and the position of the individual components. The peak function was a mixed Gauss–Lorentz line shape.
3. Results and discussion The aim of this study was to investigate how HPhCD incorporation affects the properties of insulin-loaded PLGA microspheres, with particular regard to protein release kinetics. This effect was studied on
microspheres prepared by spray-drying different feeding liquids (i.e. W/O emulsion and solution) and therefore differing in the internal porosity. Composition and overall characteristics of the prepared microspheres are reported in Table 1. Monodispersed microspheres with mean size ranging between 9.8 and 15.2 Am were prepared. The incorporation of HPhCD into the formulation did not appreciably affect size. As can be seen in Fig. 2, microspheres were regularly shaped and no insulin crystal was observed, thus suggesting an effective
Fig. 2. SEM micrographs of insulin-loaded PLGA microspheres.
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
77
Fig. 3. Confocal microscopy analysis of microspheres containing FITC-labeled insulin: (A) emulsion formulations; (B) solution formulations.
peptide entrapment inside the polymeric matrix. Nevertheless, microspheres prepared from emulsion and containing HPhCD were unexpectedly porous (Fig. 2B). It is reasonable that the presence of HPhCD in the aqueous phase modified solvent evaporation rate, which is generally recognised as a factor controlling surface morphology of microspheres produced by spray-drying [17]. Insulin distribution inside microspheres was assessed by confocal microscopy (Fig. 3). In both emulsion and solution formulations, insulin was well distributed within the polymeric matrix, and low amounts of drug were located near the surface. As expected, the internal morphology of microspheres depended on the initial formulation. When microspheres were prepared from emulsions (Fig. 3A), insulin was mainly located inside macropores distributed in the matrix, and originated from the aqueous phase of the emulsion during microsphere formation. In the case of the atomization of a continuous phase (i.e., solution formulations), a homogeneous insulin dispersion in the polymeric matrix was achieved (Fig. 3B). For all formulations, both insulin and HPhCD were effectively loaded inside microspheres (Table 1), so that the initial insulin/cyclodextrin molar ratio was kept. In agreement with our previous results [11], insulin was encapsulated in its native state. The results of the RP-HPLC analysis showed that desamido insulin content of microspheres was always lower than 5%, that is a value commonly present as impurity in the commercial insulin (as stated by USP 26). No covalent aggregate soluble in the extraction medium was detected by SE-HPLC. Furthermore, the presence
of insulin fibrils inside microspheres was evaluated by washing out the polymer and dispersing the residue in a medium unable to dissolve fibrils [11]. The resulting solution gave a transmittance value of 99F4%, which was considered indicative that insulin did not fibrillate inside microspheres. FTIR analyses in the Amide I region of insulin solutions with or without cyclodextrin are reported in Fig. 4. As can be seen, the addition of HPhCD changed insulin spectrum profile, indicating the occurrence of insulin/HPhCD interactions. Actually, previous NMR spectroscopy studies [18] have suggested that an interaction between insulin and hcyclodextrin takes place in solution, involving at least four sites on the monomeric form of insulin. The FTIR
Fig. 4. Fourier self-deconvoluted spectra in the Amide I region of aqueous solutions containing: (i) insulin (dashed line); (ii) insulin and HPhCD (continuous line).
78
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
spectra of dried microspheres, along with the curvefitting results, are reported in Fig. 5. In the case of dried samples, irrespective of the initial formulation, no significant spectral difference could be observed between Sol and Em formulations containing only insulin (Fig. 5A–C), whereas differences in the spectrum profiles are apparent between microspheres with or without HPhCD (Fig. 5A–D). In fact, the lowintensity shoulder at 1630 cm 1 present in the formulations without cyclodextrin turned into a peak of considerable intensity at the same wave number when HPhCD was coencapsulated inside microspheres. The actual changes of insulin conformation inside microspheres upon addition of HPhCD could be monitored after analysis of FTIR spectra by the leastsquares curve-fitting program described in the experimental section. The spectral profiles of the dried
0.20
A
0.15
0.20 0.15
0.10
0.10
0.05
0.05
0.00
0.00 1700
0.40
B
Absorbance
Absorbance
0.25
samples were suitably simulated by the sum of six Lorentz/Gauss functions (Fig. 5) on the basis of second-derivative analysis. The spectral parameters resulting from the analysis—position, FWHH and area of the peaks—are reported in Table 2. According to the literature, each component identified by curve-fitting analysis was tentatively assigned to the stretching vibration of specific carbonyl groups located in different physical environments, depending on protein secondary structure. After peak assignment, assuming the invariance of the molar absorptivity of the various carbonyl components, absorbance ratios were directly converted into concentration ratios, so as to obtain the percentage of the various amide species (a-helix, h sheet/turn, random coil) with respect to the total amide concentration [19]. In the numerous studies on FTIR characterization of the secondary structure of insulin,
1680
1660
1640
1620
Wavenumber (cm-1)
1700
1600
1680
1660
1640
1620
1600
Wavenumber (cm-1)
C
D
0.30
Absorbance
Absorbance
0.30
0.20
0.20
0.10
0.10 0.00 1700
0.00 1680
1660
1640
1620
Wavenumber (cm-1)
1600
1700
1680
1660
1640
1620
1600
Wavenumber (cm-1)
Fig. 5. FTIR spectra of insulin within dried PLGA microspheres in the Amide I region and curve-fitting results: (A) Em; (B) Em/CD; (C) Sol; (D) Sol/CD. The individual Lorentz/Gauss components are shown as symmetrical peaks underneath the FTIR spectra.
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
79
Table 2 Curve-fitting data relative to FTIR analysis of insulin within dried microspheres Formulation
Position (cm 1)
FWHHa (cm 1)
Area (cm 1)
R2
Tentative assignmentb
C i/C tot (%)c
Em
1692 1677 1659 1643 1630 1616
10.0 17.0 19.8 16.2 13.9 14.1
0.63 2.32 5.36 3.24 1.45 0.74
0.9964
h-sheet h-turn a-helix h-sheet h-sheet Side chains
4.9 17.8 41.2 25.0 11.1 –
Em/CD
1692 1677 1660 1643 1631 1617
9.5 15.8 19.8 16.2 13.8 14.1
0.39 1.51 3.93 2.61 1.72 0.51
1.0000
h-sheet h-turn a-helix h-sheet h-sheet Side chains
3.9 14.9 38.6 25.7 16.9 –
Sol
1692 1678 1660 1642 1629 1616
9.7 15.8 19.8 16.2 13.8 14.1
0.92 3.08 8.30 4.89 2.61 0.51
0.9979
h-sheet h-turn a-helix h-sheet h-sheet Side chains
4.6 15.6 41.9 24.7 13.2 –
Sol/CD
1692 1677 1660 1643 1631 1618
8.2 15.8 19.8 16.2 13.8 14.1
0.52 2.34 6.90 5.05 3.42 0.96
1.0000
h-sheet h-turn a-helix h-sheet h-sheet Side chains
2.9 12.9 37.8 27.7 18.8 –
a b c
Full weight at half height. According to Refs. [20–23]. Percentage of total area with side chain subtracted.
there is a general consensus about the assignment of the peak at 1658–1659 cm 1 to the a-helix structure [20–22]. The h-structures give rise to a more complex pattern arising from h-sheet and h-turn secondary structures. The multiplicity of components observed in the vibrational spectrum of these structures reflects a wider distribution of force constants within the different local environments and/or a stronger dependence of the force field from geometrical factors. The antiparallel h-sheet produces two components, one at low frequency, around 1630 cm 1, and one at high frequency, that appears at around 1690 cm 1. Another component at around 1642–1643 cm 1 has been attributed to parallel h-sheet [21–23]. The various bands appearing between 1660 and 1700 cm 1 are commonly assigned to different kinds of h-turns in a nonspecific way [23]. Finally, several authors assigned the component at around 1615 cm 1 to side chains
[19,21]. The presence of this spectral feature usually complicates the analysis of the Amide I region; in the present case, the relative contribution of 1615 cm 1 was neglected in the quantitative analysis. Upon addition of HPhCD within PLGA microspheres, the component at around 1630 cm 1 increased by a factor of 1.5 (e.g., 16.9% versus 11.1% in the case of emulsion formulation; Table 2), suggesting a rearrangement of insulin h-structures. This effect is consistent with previous NMR spectroscopy studies indicating that cyclodextrin can interact with insulin residues adopting an extended h-strand conformation [18]. The release profiles of insulin and HPhCD from all formulations are reported in Figs. 6 and 7 as insulin and HPhCD released over time. The release of insulin was a two-stage process where a first more rapid release stage (burst) was followed by a slower
80
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Fig. 6. Release profiles of insulin from emulsion (A) and solution (B) formulations evaluated in buffer at pH 7.4 and 37 8C.
second phase (Fig. 6). Concerning the burst, Em formulation displayed the highest burst (33F2%), whereas Em/CD microspheres resulted in a lower amount of insulin released in the first stage (18F3%). Sol and Sol/CD formulations showed a burst significantly lower than the formulations prepared from emulsions (8F1% and 6F2%, respectively). Concerning the second stage of insulin release, all formulations displayed a controlled release of insulin for more than 1 month. At 35 days, the amount of insulin released from Em and Sol formulations was about 94%, whereas Em/CD and Sol/CD formulations released about 80% and 66% of their insulin content, respectively. However, Em, Em/CD and Sol formulations released insulin at a similar rate, whereas release from Sol/CD occurred at a significantly lower rate. In each case, insulin
was released in its native conformation as indicated by circular dichroism data relative to insulin released after 24 h (the ratio between molar ellipticities at 223 and 208 nm was 0.8 as found for native insulin). Insulin was chemically stable inside microspheres along the release study since neither increased amounts of desamido insulin nor soluble insulin aggregates were detected in microsphere extracts. Insulin fibrils were not found inside microspheres as indicated by turbidimetry results at 35 days. Concerning HPhCD release, Em/CD and Sol/CD formulations displayed an HPhCD burst (Fig. 7) higher than the respective insulin burst. This effect can be reasonably attributed to a faster dissolution of HPhCD in the hydrating medium and its higher mobility in the matrix as compared to insulin. In particular, a two-fold higher burst was observed for Em/CD as compared to Sol/CD formulation (49F8% versus 25F1%, respectively), causing a depletion of cyclodextrin molecules inside microspheres. During the following sustained release stage, no difference in the HPhCD release rate was evident between Em/ CD and Sol/CD. Release studies have clearly shown that the modulator effect of HPhCD depends on the initial formulation conditions. In the case of Em/CD, cyclodextrin is able to reduce insulin burst, whereas in the case of Sol/CD, it slows down the overall insulin release rate. In order to elucidate this aspect, supplementary FTIR experiments were performed on hydrated microspheres at burst time. As shown in
Fig. 7. Release profiles of HPhCD from Em/CD and Sol/CD formulations evaluated in buffer at pH 7.4 and 37 8C.
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Fig. 8, the incorporation of HPhCD affected only the FTIR spectrum of Sol/CD formulation, where a higher absorption at about 1630 cm 1 was recorded (Fig. 8D). For hydrated microspheres, the spectral profiles were suitably simulated by the sum of eight Lorentz/Gauss functions. On the basis of literature data, the additional components at 1649 and 1668 cm 1 were assigned to unordered structures (random coils) and h-turns, respectively [20–23]. As shown in Table 3, significant differences in the relative concentration of the h-sheet component at 1630 cm 1 were observed only between solution formulations with or without cyclodextrin (i.e., 15.0% versus 11.9%), indicating that insulin/HPhCD interactions are expected to take place during the release stage only inside Sol/CD microspheres. Combining release kinetics of HPhCD with FTIR results and assuming that protein diffusion in the
81
polymeric matrix is a factor controlling the release, a hypothesis on the mechanism by which cyclodextrins modulate release rate can be advanced. The formation of insulin/CD complexes inside microspheres, as demonstrated by FTIR study on both dried and hydrated microspheres, likely reduces insulin mobility in the polymer through decrease of its effective diffusion coefficient. In the case of Em/CD, an early depletion of cyclodextrin from microspheres (HPhCD/insulin mole ratio decreases from 3.4 in dried microspheres to 1.7 in microspheres at burst) shifts insulin/CD complexation equilibrium toward free protein, which has a diffusion coefficient lower than complexed form. As a consequence, cyclodextrins reduce insulin release rate only in the very first stages of release. In the case of Sol/CD, a lower HPhCD burst generates a higher fraction of cyclodextrin molecules inside the polymeric matrix avail-
Fig. 8. FTIR spectra of insulin within hydrated PLGA microspheres in the Amide I region and curve-fitting results: (A) Em; (B) Em/CD; (C) Sol; (D) Sol/CD. The individual Lorentz/Gauss components are shown as symmetrical peaks underneath the FTIR spectra.
82
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
Table 3 Curve-fitting data relative to FTIR analysis of insulin within hydrated microspheres Formulation
Position (cm 1)
FWHHa (cm 1)
Area (cm 1)
R2
Tentative assignmentb
C i/C tot (%)c
Em
1691 1678 1668 1658 1649 1642 1630 1616
11.9 13.2 10.0 13.6 7.6 11.8 11.4 8.2
0.98 2.08 1.37 3.99 0.68 2.09 0.71 0.54
0.9993
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
7.5 17.5 11.5 33.6 5.7 17.6 5.9 –
Em/CD
1689 1677 1668 1658 1649 1642 1631 1617
15.0 13.2 10.0 13.6 8.5 11.8 12.6 17.4
1.18 1.51 1.03 3.15 0.73 1.60 0.65 0.45
1.0000
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
12.0 15.3 10.4 32.0 7.4 16.3 6.6 –
Sol
1691 1677 1667 1658 1649 1642 1630 1618
10.6 14.4 10.0 13.5 8.1 11.7 12.0 19.6
0.85 2.45 1.35 3.91 0.79 2.04 1.54 1.01
0.9972
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
6.6 18.9 10.4 30.2 6.1 15.8 11.9 –
Sol/CD
1691 1677 1667 1658 1649 1642 1630 1615
10.6 14.4 10.0 13.6 8.5 11.7 12.0 20.0
0.60 1.84 1.07 3.08 0.67 1.51 1.55 1.04
0.9976
h-sheet h-turn h-turn a-helix Random coil h-sheet h-sheet Side chains
5.8 17.9 10.4 29.8 6.5 14.6 15.0 –
a b c
Full weight at half height. According to Refs. [20–23]. Percentage of total area with side chain subtracted.
able for insulin complexation (HPhCD/insulin mole ratio is 3.4 in Sol/CD microspheres at burst), which contributes to reduce protein mobility and slow down the overall release rate. Furthermore, it cannot be neglected that the lack of macroporosity—which is typical of microspheres produced from solution formulations—can affect the stability constant of insulin/CD complex in the hydrated microspheres and play a role in decreasing protein mobility through the polymeric matrix.
4. Conclusions In this work, we have investigated how cyclodextrins can affect the properties of insulin-loaded PLGA microspheres prepared by spray-drying. Release studies have shown that the modulator effect of HPhCD depends on the initial formulation conditions. FTIR analysis coupled with in vitro release studies of both insulin and HPhCD have pointed out that the formation of insulin/cyclodextrin complexes
G. De Rosa et al. / Journal of Controlled Release 102 (2005) 71–83
inside microspheres is critical to decrease protein mobility in the polymer matrix and achieve an effective modulation of protein release rate.
Acknowledgements
[11]
[12]
The financial support of MIUR (PRIN 2002) is gratefully acknowledged. [13]
References [1] R.P. Batycky, J. Hanes, R. Langer, D.A. Edwards, A theoretical model of erosion and macromolecular drug release from biodegradable microspheres, J. Pharm. Sci. 86 (1997) 1464 – 1477. [2] V. Lemaire, J. Be´lair, P. Hildgen, Structural modeling of drug release from biodegradable porous matrices based on a combined diffusion/erosion process, Int. J. Pharm. 258 (2003) 95 – 107. [3] C. Washington, Drug release from microparticulate systems, in: S. Benita (Ed.), Microencapsulation, Marcel Dekker, New York, 1996, pp. 155 – 182. [4] G. Crotts, T.G. Park, Protein delivery from poly(lactic-coglycolic acid) biodegradable microspheres: release kinetics and stability issues, J. Microencapsul. 15 (1998) 699 – 713. [5] M.A. Tracy, K.L. Ward, L. Firouzabadian, Y. Wang, N. Dong, R. Qian, Y. Zhang, Factors affecting the degradation rate of poly(lactide-co-glycolide) microspheres in vivo and in vitro, Biomaterials 20 (1999) 1057 – 1062. [6] D.C. Bibby, N.M. Davies, I.G. Tucker, Mechanisms by which cyclodextrins modify drug release from polymeric drug delivery systems, Int. J. Pharm. 197 (2000) 1 – 11. [7] D. Duchene, D. Wouessidjewe, G. Ponchel, Cyclodextrins and carrier systems, J. Control. Release 62 (1999) 263 – 268. [8] T. Irie, K. Uekama, Cyclodextrins in peptide and protein delivery, Adv. Drug Deliv. Rev. 36 (1999) 101 – 123. [9] K. Fu, K. Griebenow, L. Hsieh, A.M. Klibanov, R. Langer, FTIR characterization of the secondary structure of proteins encapsulated within PLGA microspheres, J. Control. Release 58 (1999) 357 – 366. [10] T.H. Yang, A. Dong, J. Meyer, O.L. Johnson, J.L. Cleland, J.F. Carpenter, Use of infrared spectroscopy to assess secondary
[14]
[15]
[16] [17]
[18]
[19]
[20]
[21]
[22]
[23]
83
structure of human growth hormone within biodegradable microspheres, J. Pharm. Sci. 88 (1999) 161 – 165. F. Quaglia, G. De Rosa, E. Granata, F. Ungaro, E. Fattal, M.I. La Rotonda, Feeding liquid, non-ionic surfactants and cyclodextrin affect the properties of insulin-loaded poly(lactide-coglycolide) microspheres prepared by spray-drying, J. Control. Release 86 (2003) 267 – 278. Oliva, J.B. Farin˜a, M. Labre´s, Development of two highperformance liquid chromatographic methods for the analysis and characterization of insulin and its degradation products in pharmaceutical preparations, J. Chromatogr., B, Biomed. Sci. Appl. 749 (2000) 25 – 34. P.K. Zarzycki, H. Lamparczyk, The equilibrium constant of hcyclodextrin-phenolphtalein complex: influence of temperature and tetrahydrofuran addition, J. Pharm. Biomed. Anal. 18 (1998) 165 – 170. J.K. Kauppinen, D.J. Moffat, H.H. Mantsh, D.G. Cameron, Fourier self-deconvolution: a method for resolving intrinsically overlapped bands, Appl. Spectrosc. 35 (1981) 271 – 276. J.K. Kauppinen, D.J. Moffat, D.G. Cameron, H.H. Mantsh, Fourier transforms in the computation of self-deconvoluted and first-order derivative spectra of overlapped band contours, Anal. Chem. 53 (1981) 1454 – 1457. W.F. Maddams, The scope and limitations of curve fitting, Appl. Spectrosc. 34 (1980) 245 – 267. F.J. Wang, C.H. Wang, Sustained release of etanidazole from spray dried microspheres prepared by non-halogenated solvents, J. Control. Release 90 (2002) 263 – 280. F.L.A. Aachmann, D.E. Otzen, K.L. Larsen, R. Wimmer, Structural background of cyclodextrin–protein interactions, Protein Eng. 16 (2003) 905 – 912. C. Jung, Insight into protein structure and protein–ligand recognition by Fourier transform infrared spectroscopy, J. Mol. Recognit. 13 (2000) 325 – 351. A. Dong, P. Huang, W.S. Caughey, Protein secondary structures in water from second-derivative amide I infrared spectra, Biochemistry 29 (1990) 3303 – 3308. M.J. Pikal, D.R. Rigsbee, The stability of insulin in crystalline and amorphous solids: observation of greater stability for the amorphous form, Pharm. Res. 14 (1997) 1379 – 1387. L. Nielsen, S. Frokjaer, J.F. Carpenter, J. Brange, Studies of the structure of insulin fibrils by Fourier transform infrared (FTIR) spectroscopy and electron microscopy, J. Pharm. Sci. 90 (2001) 29 – 37. J.L.R. Arrondo, F.M. Goni, Structures and dynamics of membrane proteins as studied by infrared spectroscopy, Prog. Biophys. Mol. Biol. 72 (1999) 367 – 405.