Related publication: I. Arčon, A. Kodre, R.M. Abra, A. Huang, J. J. Vallner, D. D. Lasič Colloids and Surfaces B: Biointerfaces, Vol. 33/3-4 (2004) 199-204

Efficient and stable encapsulation of un-protonatable hydrophilic drugs is one of the most difficult tasks in liposomal drug delivery [1]. In many cases, using high lipid and drug concentrations may be the only way to achieve meaningful drug loading. However, formation of homogeneous unilamellar vesicles in such conditions is often very difficult or even impossible.

Fig. 1. Encapsulation of anticancer agent cisplatin into liposomes

The anticancer agent cisplatin [2] is characterized by a solubility of approximately 1-2 mg/ml at room temperature, which increases to 8-10 mg/ml at 60°C; this solubility dependence on the temperature can be exploited for efficient drug encapsulation. Liposomes can be prepared at high temperature, and on cooling the unencapsulated drug crystallizes and can be separated (filtered or decanted) from drug?containing liposomes.
This procedure yields liposomes that contain in their aqueous core cisplatin at approximately 8 mg/ml, i.e. about eight times its solubility at room temperature, as shown by analytical methods and indirectly by the anticancer activity of the formulation [3]. The drug-to-lipid ratio is ca. 0.01 mg/M of total lipid. Cisplatin does not leak out from liposomes on storage and dilution with the external medium or plasma, because of its insolubility in the bilayer and mechanically very strong and impermeable bilayers.
Because the drug concentration exceeds its solubility eightfold, this preparation raises the fundamental question of the physical state of the drug in liposomes. The drug molecules can be precipitated, adsorbed to the lipid surface, intercalated between polar heads, complexed with (or dissolved within) polyoxyethylene chains grafted to the liposome surface, or can simply form a supersaturated solution. Cryo electron microscopy, IR (infrared), and Raman spectroscopy have not provided any definitive answer to this question.


Pt L3-edge EXAFS spectra were measured on the liposome-encapsulated cisplatin to determine the local structure around Pt atoms in the sample. For comparison, Pt L3-edge EXAFS of free drug in solid and dissolved form was also measured, to give insight into the state of the encapsulated drug.
Platinum L3-edge EXAFS spectra of the samples were measured in a transmission mode at the X-ray beamline ROEMO2 (X1.1) in Hamburger Synchrotronstrahlungslabor HASYLAB at Deutschen Elektronen-Synchrotron DESY (Hamburg, Germany). A Si(311) fixed-exit double-crystal monochromator was used with 2-eV resolution at 12 keV. Harmonics were effectively eliminated by a slight detuning the monochromator crystals using a stabilization feedback control. Ionization cells filled with argon at 1 bar were used to detect the incident flux of the monochromatic x-ray beam and the transmitted flux through the sample. Standard stepping progression within a 1000-eV region above the edge was adopted with an integration time of 1 s/point
Liquid liposome-encapsulated cisplatin samples and aqueous solution of cisplatin were inserted in a variable-length liquid absorption cells with kapton windows. Aqueous cisplatin solution (1 mg/ml) was prepared in situ by dissolving crystalline cisplatin in distilled water. The optimal total absorption thickness (d) of about 2 was found with 5 mm thick sample layer in the cell in the case of liquid sample. Because of the very low concentration of Pt in the samples, the obtained Pt L3 edge jump was only 0.04 and 0.1 for liposome-encapsulated sample and the aqueous solution, respectively. Ten experimental runs were superimposed in case of liposome-encapsulated samples and three runs in case of aqueous solution to improve the signal-to-noise ratio. A reference spectrum for the liposome-encapsulated samples was taken on a 5-mm thick layer of an aqueous solution of placebo liposomes while reference spectrum for aqueous cisplatin solution was measured on a 5-mm thick layer of distilled water. A powdered crystalline sample was prepared on multiple layers of adhesive tape, with Pt L3 edge jump of about 1. A reference spectrum was measured on empty tapes.

EXAFS spectra were analyzed with the University of Washington analysis program UWXAFS [9]. Standard k^2-weighted EXAFS spectra (k) at the Pt L3 edge for different cisplatin samples are shown in Fig. 2. Due to low Pt concentration in the liposome-encapsulated sample, the obtained signal-to-noise ratio is inferior to that in the crystalline cisplatin and its aqueous solution.


Fig. 2. k^2-weighted Pt L3-edge EXAFS spectra of cisplatin in the crystal, in aqueous solution, and in liposome-encapsulated sample. Spectra are shifted vertically for clarity. Experiment - dots; best fit model function - solid line.

Fourier-transformed (FT) k^2-weighted EXAFS spectra calculated in the k interval 3 to 12 for individual samples are shown in Fig. 2. In the initial step, the signal of the crystalline cisplatin sample was analyzed. Although it was not directly relevant for the study of the encapsulation effects, this high-quality signal served as a benchmark of the achievable resolution. The model of the neighborhood was built ad hoc with some help from crystallographic data [11,12]. Ref. 11 provides data on the immediate neighborhood of the Pt atom, defined by the structure of the molecule itself. Position of further neighbours is defined by the particular stacking of the molecules in the crystal. Two crystal modifications of cisplatin are known. The sample used in this study was identified by x-ray powder diffraction analysis as a monoclinic "beta" form of Ref. 12.
Consequently, the R range of 1.0 .. 2.4 in the FT spectrum was perfectly explained by a first-neighbor shell of two N and two Cl neighbors at 2.05 and 2.33 A, respectively. The further range between 2.4 and 4.0 was resolved into contributions of higher-order scattering on the first shell neighbors and of the second shell of neighbors, comprising two Pt at 3.37 . The complete overview of best-fit parameters is given in Table 1. The quality of the fit is shown on Figures 1 and 2.

Fig. 3. The k^2-weighted Fourier transform (k = 3 .. 12 ) of the EXAFS spectrum of cisplatin in the crystal, in aqueous solution, and in liposome-encapsulated sample. Spectra are shifted vertically for clarity. Experiment - solid line; best fit model function - dashed line.

The crystalline sample with its well-defined nearest neighbors was also used to determine another parameter, the amplitude reduction factor of the EXAFS signal for Pt atom. This number is transferable between different samples with the central atom in a similar chemical (valence, coordination) state. The result (= 0.770.03) was in good agreement with theoretical estimates [13] and was used in the analysis of subsequent cisplatin samples.
The EXAFS spectrum of the aqueous solution of cisplatin is another, possibly closer template for identification of the Pt atom neighborhood in the encapsulated samples. The model of the Pt neighborhood was based on the closest shell of two N and two Cl atoms of the molecule, as in the crystalline sample. A very good fit (Table 1, Figs. 2 and 3) was obtained for the interval from 1.1 to 2.5 . The structure parameters are equal, within the error interval, to those of the crystalline sample. With higher-order scattering contributions from the same shell the validity of the model is extended up to 4 , providing an explanation of the small double peak within the region. The presence of Pt atom neighbors in a second coordination shell, such as confirmed in the crystalline sample, was excluded on the basis of statistical quality-of-fit measure. EXAFS models that include Pt atoms in the second shell do not fit the data in the R range from 2.5 to 4.0
This was to be expected: the opposite finding would point to aggregation of the cisplatin molecules in the solution, which, if true, should certainly have been supported by physicochemical data (osmotic pressure, freezing-point depression). Diffuse vestiges of the EXAFS signal beyond 4 can be attributed to the hydration shell of the molecule. Due to random positions and motion of solvent molecules the signal is largely averaged out. The important point, however, is that the hydration shell does not extend inward to the immediate vicinity of the central Pt atom.
The spectrum of the encapsulated cisplatin, although considerably noisier, agree perfectly with that of the cisplatin solution. The quantitative analysis in the R interval from 1.1 to 2.5 confirms the presence of two Cl and two N atoms in the first coordination shell at the same distances as found in the solution, within the experimental error (Table 1, Figs. 2 and 3). The presence of the second-shell Pt atom neighbors is again excluded, so that aggregation of encapsulated cisplatin molecules is not indicated.

Table 1. Parameters of the nearest coordination shells around Pt in cisplatin samples: type of the neighbour atom, average number N, distance R, and Debye-Waller factor s2. Uncertainty of the last digit is given in parentheses. The r-factor is a measure of the quality of EXAFS fit [9].

The determination of the local structure in the neighborhood of the Pt atom not only helps to determine the aggregation state but also can shed light on its chemical structure and consequently, its chemical stability. The first neighbors of the Pt atom, the directly bonded Cl and N atoms, were found at the same distance and coordination number in all samples, which shows that the cisplatin molecule in the liposome is not appreciably affected by the physical state of the system. This finding is supported by chemical and pharmacological studies.
The observation of two N and two Cl atoms in the first coordination shell demonstrates that the encapsulated cisplatin molecules are chemically stable and do not hydrolyze, i.e. exchange one or two Cl atoms with water. The absence of Pt neighbors at 3.37 clearly demonstrates that the cisplatin solution inside the liposomes does not crystallize. To be more specific, the cisplatin solution does not crystallize in the modification of the crystalline sample [11,12]. However, another crystal modification of cisplatin is known [14,15] - a hydrated cisplatin in which the closest Pt neighbors are removed to a distance beyond 4 , on behalf of interposed water molecules that contribute two O neighbors at the distance of 2.03 . Neither the encapsulated samples nor even the plain solution show O atoms so close to the central Pt. Therefore, the formation of hydrated crystal structures inside liposomes may be excluded.
The data shown above, as well as NMR observations [5], suggest that cisplatin in the liposome interior forms a supersaturated solution. From the liposome size, and lipid and drug concentrations we can estimate that each liposome contains on average approximately 2000 to 3000 cisplatin molecules. It may be that the number of compartmentalized molecules is simply too small to permit crossing the energy barrier from a crystal embryo to a real crystal. If this is the case, the encapsulated cisplatin molecules would constantly associate into crystallization nuclei and, because there are not enough molecules to overcome the barrier to start crystallization, constantly dissociate.
In conclusion, EXAFS data show that cisplatin forms a supersaturated solution in liposomes in which the drug is chemically stable and does not hydrolyze.

The study was supported by BMBF, Germany, and the Ministry of Sciences and Technology, Slovenia. L. Troeger from HASYLAB provided expert advice on beamline operation. We are thankful to A. Meden (University of Ljubljana, Faculty of Chemistry and Chemical Technology) for the powder x-ray diffraction analysis of the crystalline drug.



Last change: 02-Jun-2006