Transport Properties and Aggregation Phenomena of Polyoxyethylene

Langmuir 2007, 23, 10933-10939

10933

Articles Transport Properties and Aggregation Phenomena of Polyoxyethylene Sorbitane Monooleate (Polysorbate 80) in Pig Gastrointestinal Mucin and Mucus G. Lafitte,*,† K. Thuresson,†,‡ P. Jarwoll,§ and M. Nyde´n§ Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund UniVersity, P.O. Box 124, SE-221 00 Lund, Sweden, Camurus AB, Ideon Science Park, SE-223 70 Lund, Sweden, and Department of Materials and Surface Chemistry (Applied Surface Chemistry), Chalmers UniVersity, SE-Gothenburg, Sweden ReceiVed April 13, 2007. In Final Form: June 11, 2007 The aqueous environment in the gastrointestinal tract frequently requires solubilization of hydrophobic drug molecules in appropriate drug delivery vehicles. An effective uptake/absorption and systemic exposure of a drug molecule entails many processes, one being transport properties of the vehicles through the mucus layer. The mucus layer is a complex mixture of biological molecules. Among them, mucin is responsible of the gel properties of this layer. In this study, we have investigated the diffusion of polyoxyethylene sorbitane monooleate (polysorbate 80), a commonly used nonionic surfactant, in aqueous solution, in mucin solutions at 0.25 and 5 wt %, and in mucus. These measurements were done by using the pulsed field gradient spin echo nuclear magnetic resonance (PGSE-NMR) technique. We conclude that polysorbate 80 is a mixture of non-surface-active molecules that can diffuse freely through all the systems investigated and of surface-active molecules that form micellar structures with transport properties strongly dependent on the environment. Polysorbate 80 micelles do not interact with mucin even though their diffusion is hindered by obstruction of the large mucin molecules. On the other hand, the transport is slowed down in mucus due to interactions with other components such as lipids depots. In the last part of this study, a hydrophobic NMR probe molecule has been included in the systems to mimic a hydrophobic drug molecule. The measurements done in aqueous solution revealed that the probe molecules were transported in a closely similar way as the polysorbate 80 micelles, indicating that they were dissolved in the micellar core. The situation was more complex in mucus. The probe molecules seem to dissolve in the lipid depots at low concentrations of polysorbate 80, which slows down their transport. At increasing concentration of polysorbate 80, the diffusion of the probe molecules increases indicating a continuous dissolution of hexamethyldisilane in the core of polysorbate 80 micelles.

Introduction A successful oral drug formulation should give a sufficiently high uptake and provide a desired plasma profile of the active substance of interest.1 To meet these requirements, the drug formulation must be optimized for the challenging environment in the gastrointestinal tract (GIT), which may result in chemical degradation, binding and complexation, and metabolism. In particular, this means that dilution effects, pH variations, and exposure to GIT content (food, bile, etc.) should be accounted for if possible. Apart from these factors that have primary effects on physical and/or chemical properties of the drug-containing vehicle or the naked drug molecule, there are barriers to drug absorption that hinder and resist drug molecules to enter the systemic circulation.2 One obvious barrier, which has been extensively investigated over the years, is the cellular barrier. Here the drug molecule’s * To whom correspondence should be addressed. Fax: +46 46 222 44 13. E-mail: [email protected]. † Lund University. ‡ Camurus AB. § Chalmers University. (1) Ranade, V. V.; Hollinger, M. A. Drug DeliVery Systems; CC Press: Boca Raton, FL, 2004. (2) Washington, N.; Washington, C.; Wilson, C. G. Physiological Pharmaceutics. Barriers to Drug Adsorption; Taylor and Francis: London, 2001.

trans- and/or paracellular permeabilities are important factors, as well as the influence by active transporters or secretory efflux proteins. Another barrier is the mucus layer that covers the surface of the GIT and is responsible for a stagnant water layer outside the epithelial cells. This adherent gel layer on the gastrointestinal cell surface acts as a lubricant and a protective barrier against harmful agents, such as hydrogen ions and pepsins. Its barrier properties rely on the unstirred water layer through which a substance has to be transported by diffusion, and on its composition, since some compounds may hinder the transport because of physical binding. In a relatively recent study, the dry mass of pig intestinal mucus was determined to contain 37% lipids, 39% proteins, 6% DNA, and 5% mucin, while the aqueous content amounts to about 85%.3 The mucus layer can be seen as a hydrogel that obtains its gel properties from the polyelectrolyte mucin, which spans the aqueous domain. Mucin is a glycoprotein with a peptide backbone that is glycosylated to 70-80%,4 oligosaccharide side chains (typically, 2-15 sugars in length terminated with sialic acid or sulfate groups), and C- or N-terminal groups.4,5 Since the mucus layer is constantly degraded on the luminal side by (3) Larhed, A. W.; Artursson, P.; Bjork, E. Pharm. Res. 1998, 15, 66-71. (4) Deplancke, B.; Gaskins, H. R. Am. J. Clin. Nutr. 2001, 73, 1131s-1141s. (5) Khanvilkar, K.; Donovan, M. D.; Flanagan, D. R. AdV. Drug DeliVery ReV. 2001, 48, 173-193.

10.1021/la701081s CCC: $37.00 © 2007 American Chemical Society Published on Web 09/26/2007

10934 Langmuir, Vol. 23, No. 22, 2007

Lafitte et al.

Table 1. Composition, Apart from PS-80, in 5 wt % PGMuc Samples as Estimated from Larhed et al. (ref 3) component

wt %

mucin lipids protein DNA other

0.25 1.85 1.95 0.30 0.65

sum

5.00

proteolytic enzymes, such as pepsins, mucin is continuously synthesized and secreted by goblet cells.4 The mucus layer is best represented by being an inhomogeneous gel,6 with lipids likely to be assembled in hydrophobic domains. With this in mind, recent investigations have focused on transport properties of drug molecules in mucin and mucus gels,5 and it was found that the lipid (depots) were largely responsible for slowing down the movement of hydrophobic drugs, whereas the mucin polyelectrolyte had a negligible effect.7,8 Although hydrophilic drugs may be exposed to the GIT environment naked, it is likely that hydrophobic drugs are presented to the mucus layer complexed with and carried by other excipients. In the present investigation, we have therefore used pulsed field gradient spin echo nuclear magnetic resonance (PGSE-NMR)9 to follow transport properties of a commonly used nonionic surface-active agent (polyoxyethylene sorbitane monooleate, PS-80) in phosphate buffer, in native pig gastric mucus (PGMuc), and in pig gastric mucin (PGM). PGMuc was not purified and therefore contained all the original components of native mucus, whereas PGM contained exclusively mucin molecules. Materials and Methods Materials. The phosphate buffer contained 137 mM NaCl, 13.4 mM Na2HPO4‚2H2O, 1.40 mM KHPO4, all from Merck, and 6.15 mM NaN3 from BDH. The pH was adjusted to 7.1 by adding H3PO4. PEG-20-sorbitan monooleate (PS-80, average Mw ) 1310 g/mol, d ) 1.06-1.08 g/mL) was obtained from Uniqema, and mucin (PGM), type III, partially purified from porcine stomach, batch 61K7041, was obtained from Sigma. All chemicals were used as received. Pig gastric mucus (PGMuc) was prepared according to the scheme given by Wikman-Larhed et al. and stored at -20 °C before use.7 The dry weight of PGMuc was by lyophilization determined to be 6.9 wt %. All samples contained the phosphate buffer with 0.04 wt % sodium azide to prevent bacterial growth, and they were prepared by weight from stock solutions and thoroughly mixed a few hours before PGSE-NMR measurement. To check reproducibility repeated measurements were performed after approximately 2 months for many samples. In between the measuring occasions, the samples were kept frozen (-20 °C). As a result from the used route of preparation, PGMuc samples were diluted to a concentration of 5 wt % and thus, by using the results of Larhed et al.,3 had an anticipated composition, as specified in Table 1. Concentrations series of PS-80 (0.1, 0.2, 0.5, 1.0, 2.0, 5.0, and 10.0 wt %) were prepared in PB, PGMuc (5 wt %), and PGM (0.25 and 5 wt %), respectively. Hexamethyldisilane HMDS (Mw ) 146.4 g/mol) from Sigma (used as received) was added to a concentration 1.2 wt % based on PS-80 in the samples. Reported aggregation numbers of PS-80 micelles as determined by fluorescence techniques vary between (6) Wiedmann, T. S.; Herrington, H.; Deye, C.; Kallick, D. Chem. Phys. Lipids 2001, 112, 81-92. (7) Wikman-Larhed, A.; Artursson, P.; Gråsjo¨, J.; Bjo¨rk, E. J. Pharm. Sci. 1997, 86, 660-665. (8) Lafitte, G.; Thuresson, K.; Soderman, O. Langmuir 2005, 21, 7097-7104. (9) Stilbs, P. Prog. Nucl. Magn. Reson. Spectrosc. 1987, 19, 1-45.

Figure 1. Chemical structure of PS-80 for which the manufacturing process allows for a variation in the chemical composition. Nagg ) 2210 and Nagg ) 124.11,12 This corresponds to approximately 2.4-13 HMDS molecules per PS-80 micelle. Methods. The NMR self-diffusion measurements were carried out on a Varian Unity Inova 500 MHz spectrometer equipped with a dedicated diffusion probe supplied by DOTY Sci., U.S.A., and on a Bruker DMX-200 spectrometer, equipped with a Bruker diffusion probe having a maximum gradient strength of 9.6 T/m. All measurements utilized the stimulated echo sequence13 to minimize the effects from differences in transverse relaxation rates for molecules in different domains. The experimental diffusion time, ∆, was typically kept constant at 100 ms, and the length of the gradient pulses was either δ ) 4 or 8 ms depending on the speed of diffusion. The gradient strength was varied linearly between g ) 0.05 and 4.8 or 9.6 T/m. In order to test for restricted diffusion, time-dependent diffusion measurements were performed. ∆ was varied logarithmically between ∆ ) 20 and 2000 ms, but no time dependence in the functional form of the echo-decay was obtained, indicating the unrestricted motions at the length scales investigated here (around a few micrometers).

Results and Discussion Diffusion of PS-80 in Water. PS-80 is an ethoxylated sorbitane monooleate, Figure 1, which has been reported to have a critical micellar concentration (cmc) in the range of 0.01-0.02 mM.10,12,14,15 This surfactant has a hydrophilic-lipid balance (HLB) value of 1515 and forms rather small aggregated structures with an average aggregation number which has been reported to be in the range of 2210 to 124.11,12 The difficulty to give an exact value of cmc and aggregation number is due to a large chemical dispersity of the surfactant. Such fact is not surprising if one looks at the chemical structure and at the distribution of the ethoxylated groups. In the PGSE-NMR measurements, the chemical dispersity should result in a distribution of diffusion coefficients, meaning that aggregates formed have a wide size distribution. In order to detect such phenomenon and to get diffusion coefficients from PGSE-NMR, the attenuation of the peak from oxyethylene protons in PS-80 can be fitted with an expression to the Stejskal-Tanner relation, see eq 1.

I ) I0(p exp(-kD1) + q exp(-kD2) + s exp(-kD3) + ...) with p + q + s + ... ) 1 (1) D1, D2, D3, ... , are the diffusion coefficients, and k ) (γgδ)2(∆ - δ/3), where γ is the magnetogyric ratio for protons, ∆ is the effective diffusion time, and δ and g are the length and the strength of the field gradient pulse, respectively. The p-value obtained from a fit of eq 1 to the experimental data corresponds to the T2-weighted fraction of oxyethylene protons contributing to D1. (10) Glenn, K. M.; Moroze, S.; Palepu, R. M.; Bhattacharya, S. C. J. Dispersion Sci. Technol. 2005, 26, 79-86. (11) Acharya, K. R.; Bhattacharyya, S. C.; Moulik, S. P. J. Photochem. Photobiol., A 1999, 122, 47-52. (12) Haque, M. E.; Das, A. R.; Moulik, S. P. J. Colloid Interface Sci. 1999, 217, 1-7. (13) Callaghan, P. T. Principles of Nuclear Magnetic Resonance Microscopy; Oxford University Press Inc.: New York, 1991. (14) Patist, A.; Bhagwat, S. S.; Penfield, K. W.; Aikens, P.; Shah, D. O. J. Surfactants Deterg. 2000, 3, 53-58. (15) Prak, D. J. L.; Pritchard, P. H. Water Res. 2002, 36, 3463-3472.

Transport and Aggregation of Polysorbate 80

Langmuir, Vol. 23, No. 22, 2007 10935

about 20 EO groups. This value is also the number of EO groups provided by the supplier as presented in Figure 1. It is then reasonable to consider that the PS-80free is mainly constituted of polyoxyethylene groups.

(

M)6

Figure 2. PGSE-NMR attenuations for aqueous solutions of PS80. For the 5 and 10 wt % samples a biexponential description of the echo-decay has been fitted to the data. The inset shows residuals between the fit and the experimental data for the sample with 10 wt % PS-80.

In Figure 2, the echo-decay from the signal corresponding to the EO groups of PS-80 at 3.6 ppm is described by a biexponential function, i.e., eq 1 becomes I ) I0(p exp(-kD1) + (1 - p) exp(-kD2)). This fitting is highly accurate as can be seen in the inset of Figure 2. This means that a PS-80 sample can be considered as being composed of only two well-defined sizes of species. The results of the biexponential fitting for the 5 wt % PS-80 sample are p ) 0.35 and D1 ) 1.9 × 10-10 m2 s-1 for the fraction and diffusion constant for the fast component and (1 - p) ) 0.65 and D2 ) 3.5 × 10-11 m2 s-1 for the fraction and diffusion constant for the slow component, respectively. Similar results were obtained both from results presented in Figure 2 and from experiments performed with more points and stronger gradients (results not shown). Such biexponential fitting of the StejskalTanner plot have previously been observed for other polydisperse surfactants.16,17 In cited studies, the biexponential fitting was explained by the existence of two groups of molecules within a single sample. A less surface-active fraction, free in solution, diffuses quickly (∼1.7 × 10-10 m2 s-1), whereas micellized surfactant molecules have a much smaller diffusion coefficient (∼2 × 10-11 m2 s-1). In comparison, we can conclude that the two fractions obtained from echo-decay of EO peak of PS-80 correspond to the presence of one less surface-active group of molecules present as free molecules in solution (noted PS-80free in the text) and one surface-active species forming micelles (noted PS-80mic in the text). A representative value of the diffusion coefficient of PS-80free is about D1 ) 1.9 × 10-10 m2 s-1, and they have a hydrodynamic radius of about 11 Å (calculated from Stokes-Einstein equation, see eq 2).

RH )

kT 6πηD

(2)

where RH is the hydrodynamic radius, k is the Boltzmann constant, T is the temperature, η is the viscosity of the solvent, and D is the diffusion coefficient. Assuming that the radius of gyration, Rg, and the hydrodynamic radius, RH, are the same (see eq 3),18 these molecules have an average molecular weight of 900 g/mol, which corresponds to (16) Hakansson, B.; Hansson, P.; Regev, O.; Soderman, O. Langmuir 1998, 14, 5730-5739. (17) Momot, K. I.; Kuchel, P. W. Concepts Magn. Reson. 2003, 19A, 51-64.

Rg

0.89 × 10-10

)

2

(3)

PGSE-NMR data also provide information about the dynamics of the system. The two species of PS-80 having distinguishable diffusion coefficients implies that there is no fast exchange between them. However, a straight line is observed on the second part of the echo-decay of PS-80 in Figure 2. Therefore, the PS80mic molecules are either surface-active molecules free in solution or surface-active molecules forming micelles with a relatively narrow size distribution, and a fast exchange takes place between those two groups. Indeed, in the case of monodisperse surfactant samples above the cmc, i.e., when micelles are in equilibrium with monomers and fast exchange takes place between free surfactant molecules and micellized one, a straight line in the echo-decay is expected when presented as in Figure 2. The diffusion coefficient derived from such plot would then correspond to the time- and weighted-average of the fast and slow diffusion constant from the free and micellized surfactant molecules, respectively.19 Finally, we know from Figure 2 that the relative fractions (p) slightly change as the total concentration of PS-80 is increased. For PS-80 in water, the fraction of the slow diffusion species (PS-80mic) increases going from p ) 0.65 to p ) 0.70 when the concentration increases from 5 to 10 wt %. This indicates that the process of forming micelles is continuous and the fraction of micellized PS-80 increases as the total concentration is increased. Similar observations have been done with other techniques such as surface tension measurements14 and compression isotherms.20 As presented in Figure 1, the PS-80 molecules are rather large mainly due to the large head groups. In a similar way as done for C17E84,16 PS-80 molecules can be considered as “polymerlike” molecules. A scaling law for self-diffusion of such large surfactant and polymer molecules has been presented by Phillies.21-25

D(c) ) D0 exp(-Rcγ)

(4)

where D(c) is the diffusion coefficient at a concentration c, D0, is the concentration at infinite dilution, and R and γ are two constants. This law fits well our results for PS-80mic (Figure 3), and the value of D0 is calculated and equal to 4.86 × 10-11 m2 s-1. Consequently, a hydrodynamic radius of PS-80 micelles at infinite dilution of R0 ) 45 Å is calculated with Stokes-Einstein equation. With the use of the literature values for the micelle aggregation number Nagg ) 22 and Nagg ) 124, the area per head group becomes 1157 or 205 Å2, respectively. These numbers can be compared to the surface area obtained at the air-water interface,12 which is reported to be 248 Å2 under a surface pressure of 20 dyn/cm. Taking the reported values to be extreme and a (18) Gregory, P.; Huglin, M. B. Makromol. Chem. 1986, 187, 1745-1755. (19) Soderman, O.; Stilbs, P. Prog. Nucl. Magn. Reson. Spectrosc. 1994, 26, 445-482. (20) Lu, J. R.; Li, Z. X.; Thomas, R. K.; Staples, E. J.; Thompson, L.; Tucker, I.; Penfold, J. J. Phys. Chem. 1994, 98, 6559-6567. (21) Phillies, G. D. J. Macromolecules 1986, 19, 2367-2376. (22) Phillies, G. D. J. Macromolecules 1987, 20, 558-564. (23) Phillies, G. D. J. Macromolecules 1988, 21, 3101-3106. (24) Phillies, G. D. J. J. Phys. Chem. 1989, 93, 5029-5039. (25) Phillies, G. D. J.; Peczak, P. Macromolecules 1988, 21, 214-220.

10936 Langmuir, Vol. 23, No. 22, 2007

Lafitte et al.

Figure 3. Evolution of diffusion coefficients of PS-80mic with respect to concentrations. The full line is a fit of eq 2 to D as a function of PS-80 concentration. The following parameters were obtained: D0 ) 4.86 ((0.09) × 10-11 m2 s-1, R ) 0.065 ( 0.022, γ ) 0.97 ( 0.15. Table 2. Summary of the Values Obtained to Calculate the Effective Radius of PS-80 Micellesa concn of PS-80 wt %

Aobs

Dtheo (m2 s-1)

Agrowth

r (Å)

5 10

0.9 0.8

4.4 × 10-11 3.9 × 10-11

0.79 0.72

57 63

a

Figure 4. Echo-decays obtained for different concentration of PS80 in the presence of 0.25 wt % PGM.

where φsurf is the volume fraction of the surfactant micelles. If PS-80 micelles were only affected by obstruction from other micelles, the diffusion coefficients (Dtheo) would be 4.4 × 10-11 and 3.9 × 10-11 m2 s-1 for 5 and 10 wt %, respectively. However, the values presented in Figure 3 are lower. The stronger concentration dependency means that the micelles have also grown in size. Attenuation due to growth of the micelles (Agrowth) is equal to

Details of calculations can be found in the text.

representative value to be intermediate (i.e., about 680 Å2), it appears that the area obtained from the hydrodynamic radius is larger than that obtained at the air-water interface. This difference may be explained by the inherently different physical properties measured by the two different techniques (PGSE-NMR and surface tension). With the PGSE-NMR method, the area is calculated from the measured hydrodynamic radius. This corresponds to the area per surfactant at the outer most part of the spherical micelle. The area measured by surface tension measurements at the air-water interface reports on the head group area given a close-packed planar surfactant layer. As an illustrative example of this, an analogous comparison in a model system such as with the nonionic surfactant C12E8 is valuable. The hydrodynamic radius for C12E8 micelles is 32 Å as calculated from dynamic light scattering data.26 By using the known aggregation number of Nagg ) 90,26 a surfactant head group area of 143 Å2 for the C12E8 system is obtained. At the air-water interface, the same surfactant gives 60 Å2 by neutron reflection measurements.20 It thus appears that these two techniques measure different physical properties although both techniques are reporting on the surfactant head group area. It can be concluded that the numbers obtained for PS-80 are realistic seen in the perspective of literature data for C12E8. The decrease of diffusion coefficients of PS-80mic with increasing concentration may be the result of attenuation due to both obstruction of micelles (Aobs) and growth of the micelles (Agrowth). By estimating the different attenuation, it is then possible to work out the size of the micelles at 5 and 10 wt %. The attenuation due to the high concentration (Aobs) is calculated with eq 5 assuming spherical micelles.

Aobs ) 1 - 2φsurf

(5)

(26) Feitosa, E.; Brown, W.; Wang, K.; Barreleiro, P. C. A. Macromolecules 2002, 35, 201-207.

Agrowth )

D Dtheo

(6)

where D is the diffusion coefficient measured by PGSE-NMR and Dtheo is the diffusion coefficient expected if no micellar growth occurred. Then, by knowing Agrowth, one can calculate the effective radius r of the micelles with eq 7, since the diffusion coefficient is inversely proportional to the radius of the micelle according to Stokes-Einstein equation (eq 2).

r)

R0 Agrowth

(7)

where R0 is the radius of micelles at infinite dilution calculated above. The results of such calculations for 5 and 10 wt % are presented in Table 2. In conclusion, the decrease of diffusion coefficient of PS-80mic forming micelles is the result of obstruction and also from a growth in size of the micelles. Diffusion of PS-80 in 0.25 wt % PGM. As mentioned in the Introduction, PS-80 is one commonly used excipient when delivery of hydrophobic drugs is concerned. In the case of oral drug delivery, it is then important to determine and understand the possible interactions with the gastrointestinal layer and consequently with mucin molecules in order to be able to optimize the drug efficiency. In this study, 0.25 wt % PGM was chosen since it corresponds to the concentration in 5 wt % mucus (see Table 1). The echo-decays obtained when mixing different concentration of PS-80 with 0.25 wt % PGM are shown in Figure 4. The curvature of the echo-decays is very similar to the results obtained in the pure PS-80 system. For concentrations of PS-80 between 0.1 and 1 wt %, the diffusion of PS-80 is almost independent of the concentration of the surfactant. In Figure 4, the relative intensities for those concentrations actually overlap. On the other hand, at 5 and 10 wt %, the echo-decays can be accurately described by a biexponential fitting characteristic of

Transport and Aggregation of Polysorbate 80

Langmuir, Vol. 23, No. 22, 2007 10937 Table 3. Theoretical Values of Attenuation of Diffusion of PS-80 Micelles in the Presence of 0.25 wt % PGM Calculated from 2D-Lattice Model and Experimental Values Obtained by PGSE-NMRa

a

concn of PS-80 wt %

r (Å)

A⊥ (%)

A exptl (%)

5 10

57 63

16 19

14 ( 2 22 ( 2

Details of the calculation of theoretical values are given in the text.

from AFM and light scattering regarding the shape and the size of mucin molecules, a cylinder of radius (R) of 0.75 nm30 and of length of about 40 nm30-32 is a useful representation of a mucin molecule. As far as PS-80 micelles are concerned, they can be considered as spheres of radius (r) of 57 and 63 Å for 5 and 10 wt %, respectively (see Table 2). Notably, PS-80 micelles are small entities compared to mucin molecules at both concentrations. The mucin solution is then simulated by a 3D square lattice filled with obstructing cylinders (see Figure 5a). In such lattice, the diffusion of spheres can occur parallel (D|) and perpendicular (D⊥) to the cylinder, and it is equal to

1 2 D ) D| + D⊥ 3 3

Figure 5. 3D-lattice (a) and 2D-lattice (b) model for a system constituted of obstructing cylinders and spherical diffusing particles. Panel c is the cell model formed in the presence of both obstructing and diffusing particles.

the presence of two different sized components. At 5 wt %, the fraction and diffusion constant for the fast component is p ) 0.28 and D1 ) 2.0 × 10-10 m2 s-1, and for the slow component it is (1 - p) ) 0.72 and D2 ) 3.0 × 10-11 m2 s-1, respectively. At 10 wt %, the fraction and diffusion coefficient of the fast component is p ) 0.29 and D1 ) 1.8 × 10-10 m2 s-1, respectively, and for the slow component (1 - p) ) 0.71 and D2 ) 2.2 × 10-11 m2 s-1, respectively. The molecules of PS-80free diffuse similarly in aqueous solution as in the presence of mucin. One can then deduce that there is no specific interaction with mucin molecules. This conclusion is not surprising if one considers the chemical structure of PS-80free molecules. As mentioned earlier, these molecules are mainly constituted of EO groups and are expected to behave as PEG does. Several studies have already shown that PEG molecules and mucin molecules do not interact, letting the PEG molecules diffuse freely through mucin gel.27-29 On the other hand, the diffusion of the slow component decreases significantly in the presence of mucin compared to the pure aqueous solution: 14% ( 2% for 5 wt % PS-80 and 22% ( 2% for 10 wt % PS-80. Thus, mucin molecules affect the diffusion of the surfactant micelles. By assuming that there is no specific interaction between mucin and PS-80mic, the theoretical attenuation of diffusion of PS-80mic micelles by mucin molecules can be estimated by using a three-dimensional (3D) square lattice model. Mucin molecules and PS-80 micelles need to be schematized by simple geometrical forms. Considering the results (27) Lafitte, G.; Soderman, O.; Thuresson, K.; Davies, J. Biopolymers 2007, 86, 165-175. (28) Huang, Y.; Leobandung, W.; Foss, A.; Peppas, N. A. J. Controlled Release 2000, 65, 63-71. (29) Yoncheva, K.; Lizarraga, E.; Irache, J. M. Eur. J. Pharm. Sci. 2005, 24, 411-419.

(8)

When small spheres diffuse through a solution of long obstructing cylinders, the diffusion in the parallel plane is not affected. The obstruction occurs essentially in the transverse plane perpendicular to the cylinders.33 The 3D-lattice can be substituted by a 2D-lattice as illustrated in Figure 5b. Since the area fraction of cylinders is equal to 0.0025 (volume fraction of mucin), the mean field result for cylindrical obstruction at low concentration (i.e., area fraction φobs below 0.4) can be used to estimate the attenuation in the transverse plane (A⊥):33

A⊥ )

1 1 + φobs

(9)

When PS-80 micelles and mucin are in solution, the 2D-lattice model contains circles of radius R + r (Figure 5c). Consequently, φobs is calculated from eq 10:

φPS-80 )

π(R + r)2 (2R + a)2

(10)

with a obtained from

φmucin )

πR2 (2R + a)2

(11)

The results of this calculation are reported in Table 3. The theoretical and experimental results are in the same range for both concentrations. In conclusion, the attenuation of the diffusion coefficient seems to be essentially due to an obstruction of mucin on the PS-80 micelles, and no specific interactions such as electrostatic and/or hydrophobic appear to occur. This conclusion has also been drawn for mixed micelles formed by cationic surfactant and an excess of PS-80 in solution with mucin.8 (30) Hong, Z.; Chasan, B.; Bansil, R.; Turner, B. S.; Bhaskar, K. R.; Afdhal, N. H. Biomacromolecules 2005, 6, 3458-3466. (31) Round, A. N.; Berry, M.; McMaster, T. J.; Stoll, S.; Gowers, D.; Corfield, A. P.; Miles, M. J. Biophys. J. 2002, 83, 1661-1670. (32) Bansil, R.; Turner, B. S. Curr. Opin. Colloid Interface Sci. 2006, 11, 164-170. (33) Johannesson, H.; Halle, B. J. Chem. Phys. 1996, 104, 6807-6817.

10938 Langmuir, Vol. 23, No. 22, 2007

Lafitte et al.

Figure 6. Echo-decays obtained by PGSE-NMR for different concentrations of PS-80 in the presence of 5 wt % PGM.

Figure 7. Echo-decays obtained by PGSE-NMR for different concentrations of PS-80 in the presence of 5 wt % PGMuc.

Diffusion of PS-80 in 5 wt % PGM. The echo-decays obtained for PS-80 diffusing in a 5 wt % PGM matrix are shown in Figure 6. When increasing the mucin concentration to 5 wt %, we note that, at 10 wt % PS-80, the fraction and diffusion coefficient of PS-80free are p ) 0.16 and D1 ) 1.2 × 10-10 m2 s-1, respectively, and for PS-80mic (1 - p) ) 0.84 and D2 ) 0.8 × 10-11 m2 s-1, respectively. In comparison to the case of 0.25 wt % PGM, the diffusion constant of PS-80mic is significantly smaller. If one considers the same 2D-lattice model as for 0.25 wt %, a (Figure 5b) is too small to allow the diffusion of PS-80 micelles in the transverse plane. This means that transport of the micelles is severely hindered in this plane. One must then use the 3D-lattice (Figure 5a) and eq 8 becomes eq 12:

1 D ) D| 3

(12)

The attenuation is then expected to be 67%. From the results presented in Table 4, the experimental attenuation is about 70% for both concentrations of PS-80. Again we conclude that no interactions occur between mucin and surfactant molecules even at high concentrations of mucin. Diffusion of PS-80 in 5 wt % PGMuc. When mixing PS-80 with 5 wt % PGMuc (Figure 7), a significant change is seen in the echo-decay in comparison with the previous experiments. The signal from the PS-80mic is smaller, and as a consequence, the relative signal from PS-80free is more pronounced. This means

Figure 8. Diffusion coefficients for the hydrophobic probe molecule HMDS obtained in the presence of PS-80 in water, in PGM (5 wt %), and PGMuc (5 wt %).

that PS-80mic micelles interact with molecules constituting the PGMuc sample, however, excluding mucin molecules as shown in the previous section. The lipids being the most abundant species of mucus, they are also the most probable sites for interaction with PS-80 micelles. Lipids have a strong tendency to act as adsorption sites for nonionic amphiphile molecules via hydrophobic interactions between aliphatic chains. In NMR experiment, a small and broad signal from protons of the investigated molecule, due to fast relaxation effects, is often recorded in the case of aggregation with other molecules. Consequently, PS-80 molecules firmly bound to the lipid depots are less visible in the NMR experiment than the free PS-80 molecules, which bind less strongly. As seen in Table 4, reported diffusion coefficient of PS-80mic is slower in 5 wt % PGMuc than in 0.25 wt % PGM. Since a 5 wt % PGMuc sample contains 0.25 wt % mucin and other molecules, the other compounds obviously affect the diffusion of PS-80 micelles either by simple obstruction or by interaction. In view of the changed appearance of the NMR signal, a main reason of the substantially slower diffusion in mucus appears to be association. The overall diffusion of PS-80 is then subjected to an additional hindrance in mucus compared to PGM. At equivalent volume fraction of obstructing entities (5 wt % PGMuc and 5 wt % PGM), PS-80 micelles are expected to diffuse slower in PGMuc. However, the opposite is observed in Table 4. This observation may be due to the difficulties to detect the NMR signal for the slowest diffusing molecules as explained above. In addition, this observation may also imply that the mucus sample is heterogeneous. Thus, the measured diffusion coefficient is the average of fast diffusion coefficient in water-rich domains and slow diffusion in polymer-rich domains. The faster diffusion in 5 wt % PGMuc than in 5 wt % PGM is the result of a large contribution from fast diffusion PS-80 in water-rich domains. Diffusion PS-80 Micelles Doped with Hydrophobic Molecules in PGM and PGMuc. With the aim to understand more about transport properties of a hydrophobic drug carried in PS80 micelles in these complex mixtures, the water-insoluble hydrophobic molecule HMDS was added to the PS-80 solution before mixing it with PGMuc and PGM. First of all, as shown in Figure 8, the diffusion coefficients for HMDS in PS-80-water systems give a diffusion coefficient very close to that of PS-80mic as should be expected from a hydrophobic probe dissolved in the core of the PS-80 micelles (compare Figure 3). This indicates that the micelles are practically unaffected by the presence of the probe molecules.

Transport and Aggregation of Polysorbate 80

Langmuir, Vol. 23, No. 22, 2007 10939

Table 4. Summary of Parameters (p, D1, and D2) from Fitting Eq 1 to the PGSE-NMR Data from the Various Systemsa 5 wt % PS-80 p aqueous solution 0.25 wt % PGM 5 wt % PGM 5 wt % PGMUC a

0.35 ( 0.05 0.28 ( 0.07 b 0.73 ( 0.02

D1 × 10

-10

1.9 ( 0.2 2.0 ( 0.5 b 1.8 ( 0.1

(1 - p) 0.65 ( 0.05 0.72 ( 0.07 b 0.24 ( 0.02

10 wt % PS-80 D2 × 10

p

D1 × 10-10

(1 - p)

D2 × 10-11

0.30 ( 0.05 0.29 ( 0.01 0.16 ( 0.01 0.27 ( 0.07

2.1 ( 0.3 1.8 ( 0.1 1.2 ( 0.1 1.8 ( 0.5

0.70 ( 0.05 0.71 ( 0.01 0.84 ( 0.01 0.63 ( 0.07

2.8 ( 0.2 2.2 ( 0.1 0.8 ( 0.1 1.5 ( 0.2

-11

3.5 ( 0.3 3.0 ( 0.2 1.0 ( 0.1 1.3 ( 0.1

Errors calculated from Monte Carlo estimations during the fitting. b These values could not be determined precisely.

On the other hand, the diffusion of HMDS at low concentration of PS-80 (concentrations ranging from 0.5 to 5 wt %) in the PS-80/PGM and PS-80/PGMuc systems is strongly retarded. The observation may be due to association between HMDS and the slowly diffusing components in PGM and PGMuc, such as mucin molecules and lipids depots, instead of the inclusion of the hydrophobic molecules inside the PS-80 micelles. The slowing down is counteracted by an increasing PS-80 concentration. For example, at 5 wt % of PS-80, the diffusion of HMDS in PGM is close to the one in water. This increase of the HMDS diffusion with PS-80 concentration is possibly due to a continuously increasing proportion of HMDS dissolved in PS-80 micelles. PS-80 micelles act then as transporting vehicles. Similar behavior is observed both in PGMuc and in PGM, but the diffusion of HMDS is slower in PGMuc compared to PGM. This trend is expected since interactions with, and binding to, lipids depots should result in a slower diffusion as explained above. In conclusion, it appears that lipid depots affect transport of the hydrophobic HMDS more than they affect transport of PS-80. The reason may be that distribution to stagnant lipid depots is more pronounced for HMDS than for PS-80.

Conclusions Diffusion measurements of PS-80 in water shows the existence of two different sized components, one having a fast diffusion constant around 2 × 10-10 m2 s-1 and another having a slower diffusion constant around 3 × 10-11 m2 s-1. This is due to the presence of a less surface-active fraction of the PS-80 and one fraction with aggregated surfactant. The micelles formed by the surface-active fraction of PS-80 have tendency to increase in size by increasing concentration. Addition of the negatively charged, partially hydrophobic polyelectrolyte, mucin, to the solution of nonionic PS-80 leads

to a decrease of the diffusion of the surfactant micelles. Rationalizing the data in a cell model shows that the decrease of the diffusion is due to obstruction from the mucin molecules and not due to any specific (hydrophobic) interaction. The less surface-active fraction is essentially constituted of oligo oxyethylene. They are small molecules that do not form aggregated structures (micelles) and are therefore less affected by obstruction. Because they also do not interact with mucin, they diffuse freely through the samples independently of the mucin concentration. In the presence of pig intestine mucus, the diffusion of PS-80 micelles is not only affected by obstruction by mucin and other molecules but they also tend to interact strongly with lipid depots. In order to estimate the potential of PS-80 as a drug carrier for hydrophobic drugs, the micelles were doped with hydrophobic probe molecules (HMDS). The measurements revealed that in water the probe molecules were transported in the core of polysorbate 80 micelles. The situation was more complex in mucus and mucin where transport was slowed down, likely due to interactions of HMDS and hydrophobic groups of mucin molecules and dissolution of HMDS in lipid depots at low concentration of PS-80. At increasing concentration, the diffusion of the probe molecules increased indicating a continuous dissolution of HMDS in the more abundant polysorbate 80 micelles. Acknowledgment. O. So¨derman, Physical Chemistry I Lund University, is thanked for the fruitful discussion. The Swedish NMR center is acknowledged for spectrometer time on the NMR spectrometer. The Swedish Agency for Innovation Systems (VINNOVA) and the Swedish Research Council is acknowledged for financial support. LA701081S