ANALYST, NOVEMBER 1991, VOL. 116 1113 Estimation of Effective Diffusion Coefficients of Model Solutes Through Gastric Mucus: Assessment of a Diffusion Chamber Technique Based on Spectrophotometric Analysis M. A. Desai and P. Vadgama Department of Medicine (Clinical Biochemistry), University of Manchester, Hope Hospital, Salford M68HD, UK A diffusion chamber technique based on spectrophotometric analysis t o determine effective diffusion coefficients for solutes with various relative molecular mass ( Mr) values and properties, passing through native mucus gel, is reported. For all solutes studied, a reduction in effective diffusion coefficients is observed with a retardation of solute flux by a factor of at least two. For the solutes investigated (with Mr values ranging from 126-14400 u), no consistent effect of solutes of low Mr was evident with regard t o the retarding effect of mucus; however, at high Mr values (>4000 u) the retardation was greatly enhanced.A possible relationship between charged solutes of low Mr and the degree of retardation was observed, which possibly suggests the presence of ionic interactions of the solutes with the largely negatively charged mucus gel. The results provide further evidence for the suggestion that mucus is acting more than simply as a gel support for an unstirred water layer and is perhaps a more potent diffusion barrier t o specific solutes. Keywords: Diffusion coefficient; gastric mucus; diffusion chamber; steady state Mucus forms a continuous viscoelastic gel layer that covers many epithelial surfaces, including those of the gastrointesti- nal tract, the upper respiratory tract and parts of the genitourinary tract.In particular, it constitutes an extracel- Mar barrier to the diffusion of nutrients and therapeutic agents to the absorptive epithelial surface. The precise thickness of the gel has been a matter of controversy depending on the techniques used.1.2 However, the thickness of the gel has been reported by some to be between 100 and 600 pm in the gut3.4 and its key gel forming constituent has been considered to be a 2 x 106 u glycoprotein subunit5 with a high (70% m/m) carbohydrate component,6 mainly present as oligosaccharides, confined to certain regions along the linear protein core. The glycoprotein is thought to be responsible for the physical and structural properties of mucus.' Macro- molecules such as secretory IgA, lysozyme and lactoferrin may also contribute to the protective function of mucus.3 Although rheological properties are important for the action of mucus as a lubricant, mixing barrier and particle trap, and have been the focus of much attention,%lO less information is available about mucus as a diffusion barrier. Again, some controversy exists as to whether mucus acts as a diffusion barrier by stabilizing a surface water layer,ll-12 or whether it could present an additional, more potent, and perhaps more selective barrier to the transfer of nutrients of both low and high relative molecular mass ( M , ) and drugs within the small intestine.13 Peppas et al.l4 have developed a theory for solute diffusion in intestinal mucus; this takes into account the concentration of the constituent glycoprotein, the size of the diffusing species and the density of the macromolecular cross-links.However, direct practical observation is necessary concerning the influence of mucus upon the diffusion of specific solutes. There are indications that the diffusion of certain species such as the ergo1 alkaloids,*5 aminoglycosides,I6 fatty acids17 and some antibioticsI*,lg is significantly retarded. Here, an estimation of the effective diffusion coefficients ( D e ) for a series of solutes of various M , and properties passing through a native mucus gel is reported, in an attempt to examine the precise retarding effect of mucus on the access of such species to the intestinal epithelium. Theory The diffusion coefficients of compounds with various M , values were determined using steady-state analysis.2° Mem- branes we1.e held between two well mixed chambers of concentrations c1 and c2 in the component whose diffusion coefficient was to be measured.Assuming that the mass transfer resistance of the film between the bulk fluids in the two chambers and the membrane was negligible, the transient diffusion process inside one volume element within the membrane is governed by the partial differential equation dc d2c dt dx2 -=D- where c is the concentration in the membranes, D is the diffusion coefficient, t is time, and x is the distance subject to the boundary conditions' c = c l at x = O c = c 2 at x = l (2) At steady state, eqn.(1) becomes d2c O=D- d x 2 (3) Applying the boundary conditions [eqn. (2)J to eqn. (3) gives the steady-state concentration profile Y A. c = c1 + (c2 - c1)- (4) 1 The flux at steady state is given by where A is the surface area of the membrane. The slope of the straight line in the graph of Q (total amount of solute transferred through the membrane) versus t is dQldt at steady state. Therefore, Experimental Reagents Phenolphthalein diphosphate tetrasodium salt, phenolph- thalein, 5-hydroxytryptamine, 1,3,5-trihydroxybenzene (phloroglucinol), 5-hydroxy-~-tryptophan, b-nicotinamide adenine dinucleotide (NAD) and lysozyme (muramidase E.C.1114 ANALYST, NOVEMBER 1991, VOL. 116 3.2.1.17) were purchased from Sigma (Poole, Dorset, UK). Ribonucleic acid from T.utilis (RNA), glycyrrhizic acid and cyanocobalamine (vitamin BI2) were purchased from Fluka (Glossop, Derbyshire, UK). Track-etched porous poly- carbonate membranes (0.1 pm pore size) were obtained from Poretics (Livermore, CA, USA). The remaining chemicals including the buffer components were purchased from Merck (Poole, Dorset, UK). An isotonic phosphate buffer was used consisting of 2.44 g of NaH2P04, 7.5 g of Na2HP04, 3 g of NaCl and 0.6 g of ethylenediaminetetraacetic acid in 11, pH 7.4. Porcine gastric mucus was obtained from an abattoir using animals immediately after slaughter. The stomach was split open, the luminal surface washed with water and the gel collected by scraping the intact mucosal surface with a glass slide. The mucus was frozen at -20°C until required for diffusion studies.Validation experiments on acid diffusion through fresh and frozen porcine gastric mucus have given similar responses over the entire pH spectrum;21 freezing and thawing of porcine mucus has also been demonstrated to have no apparent effect on permeability.17 Diffusion Experiment Diffusion experiments carried out at room temperature (21 k 2 "C) were performed in a Perspex diffusion chamber consist- ing of two half-cells. Each half-cell contained a round chamber of 170ml volume, held together with screw connections (Fig. 1). Two polycarbonate membranes (50 mm; 0.1 pm pore size) were placed on either side of a 200pm nylon netting spacer, and the laminate held between two stainless-steel discs. The chambers were secured with O-rings placed in position to provide a water-tight connection.The nylon spacer provided an unstirred aqueous layer for diffusion. For diffusion through mucus, thawed porcine mucus was layered on the nylon spacer before placing the upper polycarbonate membrane in position. Both chambers were filled with isotonic buffer and allowed to equilibrate for 1 h. The chambers were then emptied, one chamber was filled with buffer alone and the other with a 1 mg ml-* solution of the solute in buffer. Both chambers were filled simultaneously and rapid stirring was instituted; 1OOpl volumes of samples were taken from the buffer chamber at fixed time intervals. Initial and subsequent fixed time concentrations were deter- mined by measuring the absorbance of the solutions at 280 nm, except for phenolphthalein which was measured at 550 nm.Results and Discussion The graphs of Q (amount of solute transferred) versus time for two selected compounds 5-hydroxytryptamine and lysozyme are shown in Fig. 2(a) and (b), respectively, in the presence and absence of mucus. The plots show that the relationship Q versus t is linear and therefore the application of eqn. (6) is Motor drive A I) Fig. 1 Schematic representation of the diffusion chamber assembly valid for the operating conditions and duration of the experiment, allowing calculation of D, values. Furthermore, linear regression analysis gave regression coefficients of >0.99 [Fig. 2(a) and (b)] for both aqueous and mucus layers. The effect of mucus is to reduce Q substantially over a given time.Table 1 shows the effect of including mucus within the 200 pm diffusion gap created by the nylon spacer. For all solutes studied, a reduction in D, values is observed with a retardation of solute flux by a factor of at least two. Over the range of solutes investigated with M, values between 126 and 14400u, no consistent effect of solutes of low M, is evident with regard to the retarding effect of mucus. However, at very high M, values (>4000u) the retardation due to mucus is greatly increased as shown by a plot of log M, versus the ratio A diffusion chamber technique (IUPAC22) based on spectrophotometric analysis for the estimation of D, values of solutes that absorb in the ultraviolet region and which are therefore easier to monitor is described here.Gastrointestinal of De(aqueous) : De(mucus) (Fig. 3)- 30 (4 25 20 15 10 5 € 0 cn a 40 35 30 25 20 15 10 5 0 25 50 75 100 125 150 175 200 Time/min Fig. 2 Plots of Q (amount of solute transferred) versus time in aqueous and mucus layers. (a) 5-Hydroxytry tamine, A, aqueous (y = 0.37857 + 0.13167x, r = 0.998); B, mucus & = -9.2857 x 10-2x, r = 0.998); and (b) lysozyme, A, aqueous (y 1.3740 + 0.26857x, r = 0.996); B, mucus 0, = 24250 + 1.517 x lO-*x, r = 0.957) Table 1 Effective diffusion coefficients of selected solutes through aqueous and native porcine mucus layers Solute Phloroglucinol 5-Hydroxy-~- tryptophan Phenolphthalein 5-H ydroxytryptamine Phenolphthalein diphos- NAD Glycyrrhizic acid Cy anocobalamine RNA Lysoz yme phate OJlO-7 cm* s-1 De(aqueous): M , Aqueous Mucus De~,u,,s~ 126 78 24 3.3 220 68 14 4.9 318 83 18 4.6 387 63 14 4.5 566 49 33 1.5 663 11 1.7 6.5 840 67 27 2.5 1355 26 10 2.6 4000-8000 160 9 17.8 14400 120 4.5 26.7ANALYST, NOVEMBER 1991, VOL.116 1115 ,- 30 I i 30 I 2 Fig. 3 Plot of log M, versus ratio of De(aqueous):De(mucus) (values obtained from Table 1) mucus forms a tightly bound gel that is thought to restrict the diffusion of protein molecules.6 Estimates of D, values given in Table 1 clearly demonstrate that although species of low M , are retarded by a factor of 2-5, which is consistent with recently reported values,23 they nonetheless diffuse more freely than species of high M,. Diffusion of hydrogen ions through mucus has been demonstrated to be retarded approxi- mately 5-fold when compared with diffusion through an unstirred layer of similar thickness .24 This study has found anomalously high diffusion resistance for NAD (Fig.3) and to a lesser extent for 5-hydroxy-~- tryptophan, phenolphthalein and 5-hydroxytryptamine through mucus (Table l), raising the possibility of some specific binding of these solutes by mucus; some antibiotics bearing nitrogen groups have been shown to be selectively retarded, possibly on the basis of ionic interactions and charge differences.16 Mucus gel has also been shown to retard diffusion of ionic species such as K+ ions.lO This may be explained by the fact that mucus has a net surplus of negative charges due to the presence of charged amino, carboxyl, sulphate and neuraminic acid groups on mucus fibrils, thus providing a highly electrified surface.The charged glyco- protein is capable of retarding diffusion of other charged species by the Donnan exclusion phenomenon .25 The significant retardation of charged solutes such as NAD, 5-hydroxy-~-tryptophan and 5hydroxytryptamine in this study may therefore be attributed to the interaction of these species with charged mucus, whereas neutral species such as phloroglucinol , phenolphthalein diphosphate and cyanoco- balamine are less retarded (Table 1 and Fig. 3). The reduction of net charge by treatment with a species such as N-acetyl- neuraminidaselo or N-acetylcysteine, sodium taurodeoxy- cholate and acetylsalicylic acid24 has been reported to de- crease the effectiveness of the mucus as an ionic diffusion barrier.It can be seen from Table 1 that solutes as large as RNA (4-8000 u) and lysozyme (14 400 u), although significantly retarded, can still diffuse through the tightly formed mucus matrix and that defined diffusion coefficients can be measured as for small solutes. However, a network of channels due to the ordered arrangement of glycoprotein molecules in mucus has been proposed,26 and this could permit macromolecular transfer. Indeed, even multilayer liposomes (-200 nm) intro- duced into the lumen of the small intestine have been shown to be able to penetrate the mucus layer.27 It may be possible therefore that the existence of such channels in porcine gastric mucus can account for the finite diffusion of RNA and lysozyme. It is important to point out that lysozyme did not appear to have any proteolytic effect on mucus under the conditions used in this study, as confirmed by the straight line graph of Q versus t [Fig.2(b)], which does not show any abnormality that would have occurred in the presence of any decomposition of mucus. I .- t E 0 2 000 4 000 6 000 8000 10 000 Time/s Fig. 4 Plots of amount of 1*5I-lysozyme transferred through an aqueous layer versus time at stirring speeds of A, 200 rev min-l 0 = 105.60 + 0.29160x, r = 0.987); and B, 1400 rev min-' 0, = 536.73 + 2.7042x, r = 0.996) It should also be noted that the diffusion coefficient values for the solutes reported here are effective diffusion coeffi- cients obtained strictly under the experimental conditions described here and assumptions made in the calculation of D,.They may therefore show a deviation from literature values normally obtained under ideal diffusion conditions. However, the main purpose of this paper is to assess the technique and compare mass transfer of solutes of various M , values under identical experimental conditions. The high estimates of D, values obtained in this study may be due to the fact that very rapid stirring (1400 rev min-I) of solutions was instituted to ensure that the barrier to diffusion presented by the stagnant boundary layer was minimized, and, therefore, D, values reflected only the barrier presented by the membrane/gel/water barriers. As a consequence, the model here assumes that the barrier to diffusion presented by such a stagnant boundary layer is negligible. However, under such rapid stirring conditions, some turbulence in the non- stirring aqueous diffusion compartment in the control experi- ments was unavoidable, and may have given rise to higher D, values as a result of convection (Fig.4). Lucas28 has also reported diffusion coefficients of sodium and hydrogen ions to be of one order of magnitude higher than free solution values when similar stirred compartmental systems were used, a similar explanation being provided for this observation. This problem could have been avoided by use of smaller pore polycarbonate membranes, but then the diffusion barrier presented by the membranes themselves would have been significant, particularly for macromolecular diffusion. The highly retarded diffusion of RNA and lysozyme, however, emphasizes the role of mucus as a barrier to molecules such as final nutrient peptide or saccharide products which are required to diffuse to their hydrolase or transport sites on the epithelial membrane.The intestinal mucus coat has been shown by some workers to be an important diffusion barrier for such nutrients and other oligomers that require to be digested, transported or bound to receptor sites on the outer intestinal membrane. 13 The high diffusional resistance demonstrated for mucus here is therefore of relevance to the absorption of a range of therapeutic and other exogenous compounds to which the small intestine is exposed. The ratios of aqueous to mucus D, in this study clearly show a significant dependence on M,, and therefore provide further evidence for the suggestion that mucus is acting as more than simply a gel support for an unstirred water layer.It is not clear from this general study what the precise relationship of M , may be to mucus resistance as the solutes examined varied in charge, shape and class of compound. Future work will determine diffusional resistance using model diffusants of consistent shape and charge. Also, comparison of diffusion coefficients over an extended range of M , values remains to be carried out.1116 ANALYST, NOVEMBER 1991, VOL. 116 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 References Kerss, S., Allen, A., and Garner, A., Clin. Sci., 1982, 63. 187. Bickel, M., and Kauffman, G. L., Gastroenterology, 1981, 80, 770. Munster, D. J., Roberton, A. M., and Bagshaw, P. F., N . Z .Med. J . , 1989, 102,607. van Hoogdalem, E. J., De Boeboer, A. G., and Brimer, D. D., Pharmacol. Ther., 1989, 44, 407. Carlstedt. I., and Sheehan, J. K., Biochern. SOC. Trans., 1984. 12, 615. Allen, A., Br. Med. Bull., 1978, 34, 28. Neutra, M. R., and Forstner, J . F., in Physiology of the Gastrointestinal Tract, ed. Johnson, L. R., Raven Press, New York, 2nd edn., 1987, vol. 2, p. 975. Crowther, R. S., Marriott, C., and James, S. L., Biorheology, 1984, 21,253. Allen, A., in Physiology of the Gastrointestinal Tract, ed. Johnson, L. R., Raven Press, New York, 1981, vol. 1, p. 617. Lee, S. P., and Nicholls, J. F., Biorheology, 1987, 24, 565. Morris, G. P., Gastroenterol. Clin. Biol., 1985, 9, 106. De Simone, J. A., Science, 1982, 220, 221. Smithson, K. W., Millar, D. B., Jacobs, L. R., and Gray, G., Science, 1981, 214, 1241. Peppas, N. A., Hansen, P. J.. and Buri, P. A., Int. J . Pharrn., 1985, 20, 107. Nimmerfall, F. N., and Rosenthaler, J., Biochem. Biophys. Res. Cornmun., 1980, 94, 960. 16 17 18 19 20 21 22 23 24 25 26 27 28 Niibuchi, J . J . , Aramaki, Y., and Tsuchiya, S., Int. J . Pharm., 1986, 30, 181. Smith, G. W., Wiggins, P. M., Lee, S. P., andTasman-Jones, C.. Clin. Sci., 1986, 70, 271. Cheema, M. S., Rassing, J. E., and Marriott, C., J . Pharm. Pharmacol. Suppl., 1986, 38, 53. Kearney, P., and Marriott, C., Int. J. Pharm.. 1987, 38, 211. Hanoun, B. J. M., and Stephanopoulos, G., Biotechnol. Bioeng., 1986, 28, 829. Nicholas, C. V.. Desai, M., Vadgama, P., McDonnell, M. B., and Lucas, S., Analysr, 1991, 116,463. IUPAC, Pure Appl. Chem.. 1979, 51, 1575. Desai, M. A., Nicholas. C. V., and Vadgama, P., J . Pharrn. Pharmacol., 1991,43, 124. Turner, N. C., Martin, G. P., and Marriott, C.. J. Pharm. Pharrnacol.. 1985, 37, 776. Bokris, J. O’M., and Reddy, A. K. N., in Modern Electro- chemistry, eds. Bokris, J. O’M., and Reddy. A. K. N., Plenum Press, New York, 1973. vol. 1, p. 623. Lazarev, P. I., Dokl. Akad. Nauk SSSR, 1986, 286, 761. Brodskii, R. A., Gal’perin, Yu. M., Lazarev, P. I.. Nadkochii, V., and Popov, G. A., Dokl. Akad. Nauk SSSR, 1983,273,464. Lucas, M. L., Dig. Dis. Sci.. 1984, 29, 336. Paper 1 I02 71 8 B Received June 7th, 1991 Accepted June 20th, 1991