INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The underlying problem in cystic fibrosis (CF) is a mutation in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR) protein. This protein, which acts as a Cl channel, is localized in exocrine glands and secretory epithelia. A defective CFTR protein in the respiratory tract reduces the secretion of Cl in the overlying mucus layer and enhances Na⁺ absorption by epithelial cells (1). This alters the appearance and composition of respiratory mucus and leads to the respiratory complications of CF, which include airway obstruction, chronic lung infection, and inflammatory reactions. Current treatment of CF at the respiratory level is directed at reducing the seriousness of these complications (1). However, such treatment does not change the mutation in the CFTR gene that is the basic defect in CF.

In CF gene therapy, either viral or nonviral transfection vectors are used. Although viral vectors are clearly advantageous with respect to transfection efficiency, they may have some disadvantages, such as the induction of an immune response, cytopathic effects, and the possibility of recombination with wild-type viruses (2). Attention has therefore recently been given to nonviral vectors, and in particular to cationic liposomes that can spontaneously form complexes with nucleic acids (2). This association results in the formation of so-called lipoplexes (5).

A limited number of clinical trials of CF gene therapy have been reported (6). With regard to lipoplexes, only transfection of the nasal epitheluim has been clinically evaluated (6). However, the outcome of the first clinical trial of lipoplexes, in which these complexes were nebulized into the lungs of CF patients, was recently reported (4). All clinical studies of CF gene therapy have reported only a very limited correction of Cl transport. Moreover, in vivo transfection appears to be a thousand times less efficient than in vitro transfection (7). This might be attributable to differences in cell characteristics as well as to the presence of noncellular barriers. Because the target cells for gene therapy in CF are the submucosal gland cells and the epithelial cells lining the small conducting airways in the lungs (7, 8), gene transfection systems must permeate the tenacious overlying mucus layer in order to be effective. As compared with normal airway secretions, CF mucus has a higher consistency because of a high content of actin, serum proteins, DNA, alginate, and rigidifying lipids (9). These negatively charged biopolymers (mucin, DNA, and alginate) are connected to each other through physical entanglements of their chains and noncovalent interactions such as electrostatic, hydrogen, and hydrophobic bonding. The result is a viscoelastic network. Because it is well known that the mucus blanket in CF protects the epithelial cells by forming a diffusion barrier for pathogens and harmful particles (12), and because it has been observed that gastrointestinal mucus retards the diffusion of macromolecules (13), we wondered about the extent to which the large (100 to 500 nm), positively charged lipoplexes (14) and other systems used for transfection in CF gene therapy would be able to cross the CF sputum layer. No information about the diffusion of such large particles through pathologic respiratory sputum is available. However, it has recently been suggested that cationic liposomes and adenovirus-mediated transfection in vitro are significantly reduced in the presence of CF sputum (7, 15).

Because we were primarily interested in the influence of particle size and the viscoelasticity of CF sputum on the transport of nanosized particles through such sputum, we used inert fluorescence-labeled polystyrene nanospheres of various sizes as a model for gene particles. Our study was aimed at answering the following questions: (1) To what degree are nanospheres retarded by CF sputum and COPD sputum? (2) Is there a relation between the number of nanospheres that move through CF sputum and the viscoelasticity of the sputum? (3) What is the effect of the size of the nanospheres on their passage through CF sputum? and (4) Does the mucolytic agent recombinant human deoxyribonuclease (rhDNase) I enhance the transport of nanospheres through CF sputum?

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sputum

After approval by the Ethics Committee of the University Hospital of Ghent, sputum was obtained from CF patients and from a patient with COPD who attended the hospital. The sputum was produced spontaneously during respiratory therapy. After collection, sputum samples were immediately frozen at 20° C. Sputum viscoelasticity was measured with a controlled-stress rheometer (AR 1000-N; TA-Instruments, Ghent, Belgium) at 20° C, using a cone-plate geometry. The angle between the cone and the plate was 2 degrees, and the sample volume required was approximately 0.9 ml. Dynamic oscillatory measurements were made at a constant frequency of 1 Hz, with a stress ranging from 0.01 to 0.1 Pa. To avoid disruption of the weak biopolymer network in sputum, the elastic (G') and viscous (G) moduli of the sputum samples were determined in the linear viscoelastic region.

The concentration of DNA in the sputum samples was determined fluorometrically with a modified diaminobenzoic acid (DABA) assay (16). Sputum pretreated with 10 µg rhDNase I (Pulmozyme; Roche, Brussels, Belgium; 1 µg rhDNase I = 1 U) per milliliter of sputum was diluted 20-fold with buffer (0.01 M Na₂HPO₄, 0.04% NaN₃; pH 7.4) and placed in a sonification bath for 15 min for further dispersion. The dispersed sputum was incubated at 60° C for 1 h. Subsequently, 300 µl of diluted sputum was incubated with 300 µl of 20% DABA solution at 60° C. After 1 h, the reaction was stopped by adding 10 ml of 1.76 M HCl. The fluorescence of the reaction product was measured at an excitation wavelength of 390 nm and emission wavelength of 530 nm (SLM-Aminco Bowman; Spectronic Instruments Inc., Rochester, NY).

To determine its mucin concentration, sputum was also pretreated with 10 µg rhDNase I per milliliter of sputum. The sputum was then diluted 200-fold in buffer (0.01 M Na₂HPO₄, 0.04% NaN₃; pH 7.4) and placed in a sonication bath for 15 min for further dispersion. Subsequently, 200 µl of the diluted sputum was incubated with 240 µl of an alkaline solution of 0.6 M 2-cyanoacetamide at 100° C for 30 min. After incubation, 10 ml of 0.06 M borate buffer at pH 8.0 was added. A borate buffer was used because it was observed that the fluorescence of the reaction product was enhanced in the presence of borate ions (17). The fluorescence was measured at an excitation wavelength of 336 nm and emission wavelength of 383 nm.

Fluorescent Nanospheres

Fluorescent (yellow-green) polystyrene nanospheres of different sizes, bearing carboxyl groups on their surface, were purchased from Molecular Probes (Eugene, OR). The average particle size of the nanospheres was measured by photon correlation spectroscopy (PCS), using an Autosizer 4700 (Malvern, Worcestershire, UK). For this purpose, the commercial nanosphere suspension was diluted with phosphate buffer (0.01 M Na₂HPO₄, 0.04% NaN₃; pH 7.4). The weight-average hydrodynamic diameters of the nanospheres were 124 ± 2 nm (mean ± SD), 270 ± 6 nm, and 560 ± 11 nm. The zeta potentials of the nanospheres were determined with a Zetasizer 2000 (Malvern, Worcestershire, UK), and were 50 ± 2 mV, 47 ± 4 mV, and 57 ± 1 mV for the 124-nm, 270-nm, and 560-nm nanospheres, respectively.

Transport of Nanospheres through Sputum

Transport of the fluorescent nanospheres through CF and COPD sputum was studied at 20 ± 1° C with a vertical diffusion chamber system (Corning-Costar Corp., Cambridge, MA). As shown in Figure 1, this equipment was modified to a tricompartimental system as developed by Norris and colleagues (18). The modified system consisted of a donor side and an acceptor side, separated from each other by a layer of CF or COPD sputum (220 µm thick, 1.13 cm² area). The sputum was held in place with the aid of two track-etched porous polycarbonate membranes (Corning-Costar) with an average pore size of 3 µm. The donor compartment was filled with 5.0 ml of nanosphere dispersion (6.8 × 10⁹ nanospheres/ml in 0.01 M Na₂HPO₄, 0.04% NaN₃, pH 7.4). To prevent a difference in hydraulic pressure across the thin sputum layer upon filling of the donor compartment with nanosphere dispersion, the acceptor compartment was simultaneously filled with 5.0 ml buffer. Mixing was done by a flow of nitrogen through both compartments. The transport of nanospheres through the sputum was determined by measuring the nanosphere concentration at the acceptor side of the system after 150 min. This concentration was determined fluorometrically by taking 1.0 ml of sputum out of the acceptor side of the system. To again prevent a difference in hydraulic pressure across the thin sputum layer, 1.0 ml of sputum was simultaneously pipetted out of the donor side of the system. After measurement of their fluorescence, the 1.0-ml volumes of sputum were returned to the donor and acceptor compartments, respectively, of the system.

View larger version (24K): [in this window] [in a new window]   Figure 1.   Schematic drawing of the vertical diffusion chamber system (A) and the modification to a tricompartimental system (B through E ). The device that connects the donor side (I) to the acceptor side (II) of the system is made by modifying commercially available Snapwells (B), which are usually used to study drug transport through cell layers. The Snapwells contain a polycarbonate membrane (pore size: 3 µm) developed for culturing a cell monolayer between the donor and acceptor sides. A polystyrene ring (thickness: 220 µm) was glued to the Snapwell (C and D). This ring was filled with CF sputum and sealed with a second polycarbonate membrane (E ). The modified Snapwell was then placed between the donor and acceptor side.

To validate the transport experiments, we evaluated the binding of nanospheres to the acrylic walls of the diffusion chamber and the size of the nanospheres during the time course of the experiments. We also determined the leakage of mucin and DNA from the sputum layer, and the effect of rhDNase I on this leakage. The loss of mucin and DNA from the sputum layer was measured by simultaneously adding 5.0 ml of buffer to the donor and acceptor compartments. Mixing was again done by a flow of nitrogen through both compartments. After 150 min, the solutions in both compartments were collected. After evaporation of the solutions, the dry material was respectively solubilized in 300 µl of buffer when the DNA concentration was measured, and in 240 µl buffer when the mucin concentration was measured. We also determined the mucin and DNA concentrations of the sputum samples used in the transport experiments, in order to allow us to calculate the percentage of loss.

Influence of rhDNase I on Transport of Nanospheres through CF Sputum

To evaluate the effect of rhDNase I on the transport of the 270-nm nanospheres through CF sputum, we performed two types of experiments on sputum Samples 5 and 6 in Table 2. In a first series of experiments, we added rhDNase I to the nanosphere dispersion in the donor compartment of the diffusion chamber system to obtain final rhDNase I concentrations of 10, 30, and 60 µg/ml, respectively. In a second series of experiments, we mixed rhDNase I directly with the sputum before it was placed between the acceptor and donor compartments. The final rhDNase I concentration in the CF sputum was 6 µg/ml, a concentration observed in CF sputum after aerosolization of rhDNase I in CF patients (19). Upon mixing rhDNase I into the sputum, we followed mucolysis rheologically until no further mucolysis occurred. This revealed a reduction of G' and G of 50% and 75%, respectively, for sputum Samples 5 and 6. Subsequently, the degraded sputa were placed between the donor and acceptor sides of the diffusion chamber system. The nanosphere concentration in the donor compartment was again 6.8 × 10⁹ spheres/ml, and transport was followed for 150 min.

Confocal Microscopy

Confocal scanning laser microscopy (CSLM) (MRC1024; Bio-Rad, Hemel Hempstead, UK) was used to confirm the increased transport of nanospheres through the rhDNase I-pretreated CF sputa. After diffusion of the nanospheres, the 220-µm-thick sputum layers (Figure 1) were removed from the diffusion chamber system. The polycarbonate membranes were carefully washed with distilled water. Consequently, the Snapwell holding the sputum layer was placed on a cover slip. Confocal images (154 × 154 µm) of the nanospheres in the sputa were made at 10 µm below the surface of the sputum layer that had been in contact with the nanosphere dispersion in the donor compartment during the transport experiments. The number of fluorescent nanospheres in a microregion of each sputum sample was counted with appropriate image analysis software.

Electron Microscopy

CF sputum samples were processed for scanning electron microscopic examination according to a protocol used for the study of uterine cervical mucus (20). Samples were fixed for 1 h in 2% glutaraldehyde in phosphate-buffered saline (PBS) (2.5 mM KH₂PO₄, 7.0 mM K₂HPO₄, 0.12 M NaCl, pH 7.6), subsequently rinsed in buffer, and postfixed for 30 min in 1% OsO₄ in PBS. After three washes in PBS, the sputum samples were dehydrated through a series of ethanol solutions, critical-point dried with liquid CO₂, and sputter-coated with platinum. These prepared CF sputum samples were then examined with a scanning electron microscope (JSM-5000 LV; JEOL, Tokyo, Japan).

Statistical Analysis

The experimental results in this report are expressed as mean ± SD. Spearman's rank correlation test was used to study the relation between sputum viscoelasticity and sputum DNA and mucin content. In studying the effect of sputum viscoelasticity on nanosphere transport, we tested the equality of means through analysis of variance. Student's t test was used in evaluating the effect of rhDNase I on nanosphere transport. The t test was also used to evaluate the rheologic stability of CF sputum as a function of temperature, and to evaluate the effect of freezing on sputum viscoelasticity. Significance was set at p = 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Because the sputum samples were stored at 20° C, we evaluated the effect of freezing and thawing on their viscoelasticity. Table 1 shows the rheologic properties of eight sputum samples, all from different CF patients, before and after freezing at 20° C in airtight containers for 3 wk. No significant change in G' or G could be detected (p > 0.05).

Since we were interested in the influence of intact, nondegraded sputum on the transport of nanospheres, and because the transport experiments took about 150 min, we evaluated the rheologic stability of CF sputa and the COPD sputum sample. Sputum may undergo physical degradation after its collection, through the action of endogenous and bacterial proteases (21). This enzymatic degradation, which results in liquefaction of the sputum by cleavage of the protein backbone of mucin biopolymers, is often ignored in drug transport studies through mucus or sputum. To evaluate the rheologic stability of the CF sputum samples, we placed them between the cone and plate of the rheometer and made measurements of G' and G at 20° C and 37° C, respectively. To avoid sputum dehydratation from water evaporation, we used a solvent trap. The sputum viscoelasticity dropped by more than 30% after 2 h at 37° C, even in the presence of 0.21 mg/ml phenylsulfonylfluoride, a protease inhibitor, and 0.40 mg/ml NaN₃, an antibacterial agent. In contrast, no significant change in G' and G was observed after 3 h when CF sputum samples were kept at 20° C in the presence of phenylsulfonylfluoride (0.21 mg/ml) and NaN₃ (0.40 mg/ml) (p > 0.05). The COPD sputum sample also showed no change in viscoelasticity under these conditions. On the basis of these results, we conducted the transport experiments at 20° C for 150 min while carefully adding phenylsulfonylfluoride (0.21 mg/ml) and NaN₃ (0.40 mg/ml) to the sputum samples.

Table 2 shows the rheologic moduli of the six CF sputum samples, all from different patients, and of the COPD sputum sample used in the transport experiments. In the case of the CF sputum samples, each value is based on the rheologic analysis of three sputum fractions of each sample. Because of the limited amount of COPD sputum, only one fraction could be taken of this sample. Notwithstanding the visually observed heterogeneity of the CF sputum, a good reproducibility of G' and G was obtained (Table 2). This may be attributed to the large volume fractions (0.9 ml) taken from the CF sputum samples used in the determination of G' and G. In this way, the heterogeneity of the samples became more averaged. Because G' of each sputum sample is significantly larger than G, CF sputum can be considered as a dominantly elastic biomaterial.

The DNA and mucin contents of the CF sputa and the COPD sputum sample are shown in Table 2. A significant correlation was found between mucin content and the rheologic moduli G' and G (r_s = 0.83; p = 0.04). Additionally, a positive correlation, although not significant, was observed between the DNA concentration and the viscoelasticity of the CF sputa (r_s = 0.37; p = 0.47). Because a modified DABA assay was used, only the total DNA content of the sputum samples was measured. These data gave no information about the length of the DNA chains in CF sputum, which may play a major role in increasing its viscoelasticity. This might explain the lack of a significant correlation between the DNA content and viscoelasticity of the CF sputa.

During the transport study, no significant change in particle size was observed for any type of nanospheres used. The loss of mucin and DNA from the sputa into the donor and acceptor compartments of the diffusion chamber system depended on the addition of rhDNase I to the sputa. The loss of mucin from a first sputum sample (G' = 22 ± 6 Pa; G = 6 ± 2 Pa) increased from 7 ± 1% (before treatment with rhDNase I) to 15 ± 2% upon adding rhDNase I (6 µg/ml). For a second CF sputum sample (G' = 10 ± 3 Pa; G = 3 ± 1 Pa) the loss of mucin increased from 4 ± 2% to 13 ± 5%. To determine the loss of DNA, two other sputum samples were used. The leakage of DNA from the first sputum sample (G' = 17 ± 6 Pa; G = 4 ± 1 Pa) increased from 5 ± 1% (before treatment with rhDNase I) to 16 ± 5% upon adding rhDNase I (6 µg/ml). For the second CF sputum sample (G' = 13 ± 3 Pa and 3 ± 1 Pa) the leakage increased from 7 ± 1% to 14 ± 2%. Therefore, in the absence of rhDNase I, only a small amount of mucin and DNA leaked into the chambers. However, mixing rhDNase I with the sputa significantly increased these amounts of mucin and DNA. This was not surprising, knowing that rhDNase I cleaves DNA chains and thereby causes a weakening of the sputum viscoelastic network.

The binding experiments showed that the smallest nanospheres (124 nm) clung to the acrylic wall of the acceptor compartment of the diffusion chamber system, whereas the 270-nm and 560-nm nanospheres did not cling. To correct the concentration of the 124-nm nanospheres in the acceptor compartment for nonspecific adsorption, we performed additional experiments. The acceptor compartment was filled with 124-nm nanosphere dispersions, which, after adsorption, resulted in approximately the same nanosphere concentrations as measured in the transport experiments on the acceptor side of the diffusion chamber system after 150 min. From this decrease in nanosphere concentration we calculated, at each measuring time, a correction factor. To obtain the real quantity of 124-nm nanospheres that had moved through the sputum layer, we multiplied the experimentally determined 124-nm nanosphere concentrations in the acceptor compartment by the corresponding correction factors.

Transport experiments with the nanospheres were initially done on sputum Samples 1 through 4 in Table 2. Additional transport experiments were done with the 270-nm nanospheres on sputum Samples 5 and 6 and on the COPD sample. Each transport experiment was repeated four times. Figure 2 shows the percentages of nanospheres that were transported into the acceptor compartment after 150 min for the 124-nm, 270-nm, and 560-nm nanospheres, respectively. For each sputum sample, we observed a nonlinear decrease in the percentage of transported nanospheres as a function of the size of the nanospheres. The mean percents of nanospheres transported after 150 min through the four CF sputum samples were 0.24 ± 0.08% (124 nm), 0.022 ± 0.008% (270 nm), and 0.0017 ± 0.0009% (560 nm), respectively. On the basis of the size of the nanospheres, we calculated the percentage of nanospheres that could be expected in the acceptor compartment after 150 min with buffer present instead of a sputum layer between the donor and acceptor compartments (22). The calculated percents were 0.32%, 0.15%, and 0.071% for the 124-nm, 270-nm, and 560-nm nanospheres, respectively.

View larger version (19K): [in this window] [in a new window]   Figure 2.   Percentages of nanospheres that diffused through a 220-µm-thick layer of CF or COPD sputum into the acceptor compartment of the diffusion chamber system after 150 min as a function of the average sizes of the nanospheres. Transport experiments with the 124-nm and 560-nm nanospheres were performed only with sputum Samples 1 through 4 (n = 4 for each data point).

Figure 3 shows the percents of nanospheres that were transported into the acceptor compartment of the diffusion chamber system after 150 min as a function of G' of the CF sputa. Similar profiles were obtained when the G was plotted on the x axis. Surprisingly, for the 124-nm and 270-nm nanospheres, the percentages of transported nanospheres through the most tenacious sputum sample, Sample 4, was significantly greater than the percentages of nanospheres transported through the less viscoelastic sputum samples (p < 0.05). Additional transport studies (using the 270-nm nanospheres) with CF sputum Samples 5 and 6 (Table 2) confirmed that there was less nanosphere transport through less viscoelastic CF sputum samples (Figures 2 and 3). This phenomenon was also observed for the COPD sputum sample.

View larger version (17K): [in this window] [in a new window]   Figure 3.   Percentages of nanospheres that diffused through 220-µm- thick layers of CF and COPD (gray circle) sputum and into the acceptor compartment of the diffusion chamber system after 150 min as a function of the elastic moduli of the sputum samples (n = 4 for each data point).

The COPD sputum sample showed a much lower DNA concentration and viscoelasticity than did the CF sputa, whereas the mucin content of the two kinds of sputum samples was similar (Table 2). This confirms that a main reason for the high viscoelasticity of CF sputum is the presence of massive amounts of DNA. To reduce sputum viscoelasticity, rhDNase I is frequently used in treating CF. Because rhDNase I partly disrupts the biopolymer network in CF sputum, we wondered whether it would enhance the transport of nanospheres. Figure 4 shows the results. When rhDNase I was present on the donor side of the diffusion chamber system during the transport experiments, the greatest increase in transport was observed when a rhDNase I concentration of 60 µg/ml was used. When using lower rhDNase I concentrations (10 µg/ml and 30 µg/ml) on the donor side of the system, the transport of nanospheres was only moderately enhanced. The strongest increase in transport was seen when rhDNase I was mixed directly with CF sputa, even at a low concentration of 6 µg/ml. The increase in nanosphere transport upon mixing rhDNase I with the sputa was confirmed by confocal scanning laser microscopy of the CF sputum layers that were removed from the diffusion chamber system at the end of the transport experiments. Figures 5A and 5B show representative microregions of a CF sputum layer in which rhDNase I was mixed with the sputum before the transport experiment and when it was absent, respectively. The microregion of the sputum that had been mixed with rhDNase I contained 95 nanospheres, whereas only 14 nanospheres were present in an equal-size microregion of sputum with which no rhDNase I was mixed.

View larger version (10K): [in this window] [in a new window]   Figure 4.   Percentages of nanospheres (270 nm) that diffused into the acceptor compartment of the diffusion chamber system after 150 min in the absence and in the presence of rhDNase I. Results are mean values obtained after transport of nanospheres through sputum Samples 5 and 6. The transport experiment through each sputum sample was repeated four times. *p < 0.05; **p < 0.001; n = 8.

View larger version (164K): [in this window] [in a new window]   View larger version (160K): [in this window] [in a new window]   Figure 5.   Confocal scanning laser microscopic images (154 × 154 µm) of nanospheres in CF sputum at 10 µm below the surface of a sputum layer that was in contact with a nanosphere dispersion in the donor compartment of the diffusion chamber system used in the study during the transport experiment. The dark points are fluorescent nanospheres. The number of nanospheres in each microregion was counted with image analysis software. The microregion of the sputum into which rhDNase I was directly mixed contained 95 nanospheres (A). Only 14 nanospheres were present in an equal-size microregion of sputum with which no rhDNase I was mixed (B). The microregions are representative of findings from five separate microregions each. Bars = 10 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It was our goal to study the effect of particle size, sputum viscoelasticity, and rhDNase I on the transport of nanosized particles through CF and COPD sputum. In this way we hoped to be able to more fully elucidate the extent to which sputum represents a physical barrier to the transport of nanosized particles. Moreover, we hoped to provide an additional viewpoint to that of recently reported studies (7, 15) of the effect of CF sputum on in vitro gene transfection efficiency by lipoplexes and adenoviral carriers.

In our study, we used negatively charged polystyrene nanospheres because it is known that in vivo, cationic gene complexes are less efficient than anionic or neutral gene complexes (23). This lower efficacy of positively charged gene complexes is attributed to electrostatic interactions with anionic biopolymers in vivo (24).

CF sputa dramatically retarded the transport of large particles (Figures 2 and 3). We showed that the influence of surface hydrophobicity on the transport of nanospheres was much smaller than the effect of particle size (data not shown). We therefore concluded that the difference in transport of 124-nm, 270-nm, and 560-nm nanospheres was mainly caused by stronger steric obstruction with increasing particle size.

As compared with diffusion through buffer, the quantities of nanospheres that diffused through CF sputum were 1.3-fold, 6.8-fold and 42-fold smaller for the 124-nm, 270-nm and 560-nm nanospheres, respectively. Because the smallest nanospheres were retarded only by a factor of 1.3, it appears that the water channels in CF sputum were large enough to allow passage of these spheres. This indicates that the limited transport of the smallest nanospheres in CF sputum was mainly due to the distance they had to travel before they reached the acceptor compartment of the diffusion chamber system used in the study. On the other hand, the 560-nm nanospheres were almost completely blocked sterically by the CF sputum. This agreed with our electron microscopic data for CF sputa (Figure 6), which revealed pores with a diameter ranging from approximately 100 nm to 400 nm. However, it should be noted that distortions may have occurred in the network structure of the sputum during the fixation and dehydration steps of the sample preparation for electron microscopy.

View larger version (168K): [in this window] [in a new window]   Figure 6.   Scanning electron microscopic image of CF sputum, showing the pores in the biopolymer network. Bar = 0.5 µm.

Generally, diffusion slows when gels become more viscoelastic (25). This contrasts with the findings of the present study. As shown in Figure 3, a greater number of nanospheres moved through the more viscoelastic sputa. To explain these results, it is necessary to consider the structure of the three- dimensional network in sputum. For cervical mucus it has been proposed that the mucin chains, owing to their ordered arrangement, form a three-dimensional, macroporous network of channels whose large pores permit macromolecular drug transport (20). It has been mentioned that liposomes (~ 200 nm) can move through gastrointestinal mucus (13), whereas the transport of particles larger than 500 nm is almost completely blocked (18, 26). Additionally, Yudin and coworkers showed that in cervical mucus, a fine structure with interfiber spacings of ~ 100 nm is present within a more macroporous structure with interfiber spacings of ~ 500 nm (27). On the basis of this information, the following hypothesis may explain the greater transport of nanospheres through more viscoelastic sputa: An increase in the viscoelasticity of sputum originates from a high number of junctions between biopolymer chains per volume unit of sputum. An increase in the concentration of junctions may originate from an increase in the concentration of biopolymers in the sputum. Indeed, Table 2 shows a significant positive correlation between G' and mucin concentration (r_s = 0.83; p = 0.04) and a positive correlation between G' and DNA concentration (r_s = 0.37; p = 0.47). For synthetic gels it was observed that increasing the concentration of junctions may change the network structure of the gel from one of homogeneous microporous type to one of more heterogeneous macroporous type (28). A similar phenomenon may occur in biologic gels such as sputum. In weakly viscoelastic CF and COPD sputum, a more homogeneous microporous network may be present, with many free biopolymer chains in the aqueous channels of the network. By contrast, a more heterogeneous macroporous network may be established in highly viscoelastic sputum, with a limited number of free biopolymer chains in its aqueous channels. Both the macroporous channels and the low number of free biopolymer chains may facilitate the transport of nanospheres in the more viscoelastic sputum.

The experiments on the transport of 270-nm spheres through the COPD sputum sample revealed that the transport characteristics, as shown in Figures 2 and 3, are not unique to CF sputum. In particular, the numbers of nanospheres transported through the COPD sputum were fully in accord with the influence of viscoelasticity (Figures 2 and 3).

Figure 4 shows that when rhDNase I was added to the donor compartment of the diffusion chamber system used in the study, only a small increase occurred in nanosphere transport through CF sputum. The greatest, albeit only a limited, increase in nanosphere transport occurred when rhDNase I was directly mixed with CF sputum (6 µg/ml) before the sputum was placed between the donor and acceptor compartments of the diffusion chamber system. The limited enhancement of transport may be explained by the presence of mucins and other biopolymers as well as DNA. Consequently, rhDNase I does not completely break down the biopolymer network of sputum. Moreover, the smaller DNA fragments that result from DNA degradation may themselves hinder the transport of nanospheres through viscous effects. The loss of DNA and mucin from the sputum layer, which was significantly greater when rhDNase I was mixed with the sputum, may have enhanced the transport.

Assuming that the penetration of rhDNase I through CF sputum was rapid when a concentration of 10 µg/ml rhDNase I was added to the donor compartment, it can be expected that equilibrium of rhDNase I in such sputum will occur at a concentration of approximately 5 µg/ml sputum. This would suggest a similar effect of rhDNase I on nanosphere transport to the effect observed when 6 µg of rhDNase I was directly mixed with 1.0 ml sputum. Figure 4 shows that this was not the case. Because no loss of rhDNase I activity in the donor compartment of the diffusion chamber system was observed during the transport experiment, denaturation of rhDNase cannot explain the weak effect on nanosphere transport. Weak penetration of rhDNase I from the donor side of the system into the sputum layer may explain the weak effect on nanosphere transport. However, another explanation is that rhDNase I may have been inactive in the sputum because calcium and magnesium ions, which are necessary for optimal activity of rhDNase I (29), diffused from the sputum into the donor and acceptor compartments of the diffusion chamber system during the transport experiments. Because the addition of calcium and magnesium ions to the nanosphere dispersion might have destabilized the dispersion, we had to exclude these ions from our buffers.

Given that rhDNase I decreases sputum viscoelasticity, it can be asked, on the basis of the data shown in Figure 3, why a decrease in nanosphere transport was not observed in the presence of rhDNase I. This may be explained by the dualistic network structure of sputum as discussed earlier. In the macroporous network, the DNA chains may participate in ordered bundles of polymers that may be less accessible to degradation by rhDNase I than are DNA chains that are present in the homogeneous microporous network. Consequently, rhDNase I may first degrade the more accessible DNA chains in the microporous network, which may reduce the viscous drag on the diffusion of nanospheres, and may explain their increased transport through CF sputum.

In this study, we showed that CF sputum offers a size-dependent barrier to the transport of nanospheres of a size comparable to that of lipoplexes and other transfection systems. As compared with their diffusion through buffer, the smallest nanospheres (124 nm) used in our study were only moderately retarded by CF sputum. However, the larger, 560-nm nanospheres were almost completely blocked by CF sputum. This difference in transport was caused mainly by greater steric obstruction of the largest particles in the sputum. Our study also demonstrated that nanospheres moved more easily through sputum samples with a greater viscoelasticity. To explain this unexpected observation, we hypothesized that the network in CF sputum changes toward having a more macroporous structure when the sputum becomes more viscoelastic. COPD sputum retarded the transport of nanospheres to the same extent as did CF sputum. The strongest enhancement of the transport of nanospheres through CF sputum by rhDNase I, albeit by only a factor of 3, was observed when this mucolytic agent was mixed with the sputum. When rhDNase I was presented on the surface of CF sputum, as occurs in vivo, only a very small increase was observed in nanosphere transport through the sputum.