Membranes retention

Second, most membrane materials adsorb proteins. Worse, the adsorption is membrane-material specific and is dependent on concentration, pH, ionic strength, temperature, and so on. Adsorption has two consequences it changes the membrane pore size because solutes are adsorbed near and in membrane pores and it removes protein from the permeate by adsorption in addition to that removed by sieving. Porter (op. cit., p. 160) gives an illustrative table for adsorption of Cytochrome C on materials used for UF membranes, with values ranging from 1 to 25 percent. Because of the adsorption effects, membranes are characterized only when clean. Fouling has a dramatic effect on membrane retention, as is explained in its own section below. 

Sugano et al. [561,562] explored the lipid model containing several different phospholipids, closely resembling the mixture found in reconstituted brush border lipids [433,566] and demonstrated dramatically improved property predictions. The best-performing lipid composition consisted of a 3% wt/vol lipid solution in 1,7-octadiene (lipid consisting of 33% wt/wt cholesterol, 27% PC, 27% PE, 7% PS, 7% PI). The donor and acceptor compartments were adjusted in the pH interval between 5.0 and 7.4 [562]. With such a mixture, membrane retention is expected to be extensive when lipophilic drugs are assayed. The use of 1,7-octadiene in the assay was noted to require special safety precautions. 

The system reported by Avdeef and co-workers [25-28,556-560] is an extension of the Roche approach, with several novel features described, including a way to assess membrane retention [25-28,556,557] and a way to quantify the effects of iso-pH [558] and gradient pH [559] conditions applied to ionizable molecules. A highly pure synthetic phospholipid, dioleoylphosphatidylcholine (DOPC), was initially used to coat the filters (2% wt/vol DOPC in dodecane). Other lipid mixtures were subsequently developed, and are described in detail in this chapter. 

Yamashita et al. [82] also studied the effect of BSA on transport properties in Caco-2 assays. They observed that the permeability of highly lipophilic molecules could be rate limited by the process of desorption off the cell surface into the receiving solution, due to high membrane retention and very low water solubility. They recommended using serum proteins in the acceptor compartment when lipophilic molecules are assayed (which is a common circumstance in discovery settings). 

DERIVATION OF MEMBRANE-RETENTION PERMEABILITY EQUATIONS (ONE-POINT MEASUREMENTS, PHYSICAL SINKS, IONIZATION SINKS, BINDING SINKS, DOUBLE SINKS)... 

It is important to remember that Eqs. (7.10) and (7.11) are both based on assumptions that (1) sink conditions are maintained, (2) data are taken early in the transport process (to further assure sink condition), and (3) there is no membrane retention of solute. In discovery settings where Caco-2 assays are used, the validity of assumption 3 is often untested. 

The more general solutions (but still neglecting membrane retention, hence still apparent ) are given by two-way flux in Eqs. 7.12 (disappearance kinetics) and (7.13) (appearance kinetics). 

The popular permeability equations [(7.10) and (7.11)] derived in the preceding section presume that the solute does not distribute into the membrane to any appreciable extent. This assumption may not be valid in drug discovery research, since most of the compounds synthesized by combinatorial methods are very lipophilic and can substantially accumulate in the membrane. Neglecting this leads to underestimates of permeability coefficients. This section expands the equations to include membrane retention. 

When membrane retention of the solute needs to be considered, one can derive the appropriate permeability equations along the lines described in the preceding section Eqs. (7.1)—(7.3) apply (with P designated as the effective permeability, Pe). However, the mass balance would need to include the membrane compartment, in addition to the donor and acceptor compartments. At time t, the sample distributes (mol amounts) between three compartments ... 

At pH 3, ketoprofen is mostly in an uncharged state in solution. The dashed curve in Fig. 7.16 corresponding to pH 3 shows a rapid decline of the sample in the donor well in the first half-hour this corresponds to the membrane loading up with the drug, to the extent of 56%. The corresponding appearance of the sample in the acceptor well is shown by the solid line at pH 3. The solid curve remains at zero for t < xLAG. After the lag period, the acceptor curve starts to rise slowly, mirroring in shape the donor curve, which decreases slowly with time. The two curves nearly meet at 16 h, at a concentration ratio near 0.22, far below the value of 0.5, the expected value had the membrane retention not taken a portion of the material out of the aqueous solutions. 

Ordinarily it is not possible to determine the membrane retention of solute under the circumstances of a saturated solution, so no R terms appear in the special equation [Eq. (7.25)], nor is it important to do so, since the concentration gradient across the membrane is uniquely specified by S and CA (t). The permeability coefficient is effective in this case. 

When the pH is different on the two sides of the membrane, the transport of ioniz-able molecules can be dramatically altered. In effect, sink conditions can be created by pH gradients. Assay improvements can be achieved using such gradients between the donor and acceptor compartments of the permeation cell. A three-compartment diffusion differential equation can be derived that takes into account gradient pH conditions and membrane retention of the drug molecule (which clearly still exists—albeit lessened—in spite of the sink condition created). As before, one begins with two flux equations... 

The membrane retention fraction R may be defined as membrane-bound moles of sample, divided by the total moles of sample in the system ... 

Four neutral lipid models were explored at pH 7.4 (1) 2% wt/vol DOPC in dode-cane, (2) olive oil, (3) octanol, and (4) dodecane. Table 7.5 lists the effective permeabilities Pe, standard deviations (SDs), and membrane retentions of the 32 probe molecules (Table 7.4). The units of Pe and SD are 10 6 cm/s. Retentions are expressed as mole percentages. Figure 7.22a is a plot of log Pe versus log Kd (octanol-water apparent partition coefficients, pH 7.4) for filters loaded with 2% wt/vol DOPC in dodecane (model 1.0, hlled-circle symbols) and with phospholipid-free dodecane (model 4.0, open-circle symbols). The dashed line in the plot was calculated assuming a UWL permeability (see Section 7.7.6) Pu, 16 x 10-6 cm/s (a typical value in an unstirred 96-well microtiter plate assay), and Pe of 0.8 x 10-6 cm/s... 

Figure 7.22 Lipophilic nature of membrane retention, log(%R) versus octanol-water apparent partition coefficient, pH 7.4, neutral lipid models.

Kansy et al. [550] reported the permeability-lipophilicity relationship for about 120 molecules based on the 10% wt/vol egg lecithin plus 0.5% wt/vol cholesterol in dodecane membrane lipid (model 15.0 in Table 7.3), shown in Fig. 7.23. The vertical axis is proportional to apparent permeability [see Eq. (7.9)]. For log Kd > 1.5, Pa decreases with increasing log Kd. In terms of characteristic permeability-lipophilicity plots of Fig. 7.19, the Kansy result in Fig. 7.23 resembles the bilinear case in Fig. (7.19d). Some of the Pa values may be underestimated for the most lipophilic molecules because membrane retention was not considered in the analysis. 

It was thus interesting for us to examine the permeability and membrane retention properties of olive oil. As Table 7.5 shows, most of the Pe values for olive oil are less than or equal to those of 2% DOPC, with notable exceptions for instance, quinine is 4 times more permeable and progesterone is 16 times less permeable in olive oil than in DOPC. Both lipids show progesterone retention to be >80%, but quinine retention in olive oil is substantially greater than in DOPC. 

Membrane retention of lipophilic molecules is significantly increased in octanol, compared to 2% DOPC. Chlorpromazine and progesterone show R > 90% in octanol. Phenazopyridine, verapamil, promethazine, and imipramine show R > 70%. 

It is also quite interesting that lipid model 4.0 may be used to obtain alkane partition coefficients at high-throughput speeds, as suggested by Faller and Wohnsland [509,554], It is also interesting to note that since our Pe are corrected for membrane retention, the slope in Fig. 7.11 corresponding to the dashed line (our data) is 1.0, whereas the data not corrected for retention (solid line) show a lesser slope. This may not matter if the objective is to obtain alkane-water log Kp values at high speeds. 

Membrane Retention (under Iso-pH and in the Absence of Sink Condition)... 

Figure 7.24 Membrane retention in octanol-soaked filters versus octanol-water apparent partition coefficients.

Figure 7.25 Membrane retention in 2%DOPC/dodecane-soaked filters versus dodecane-soaked filters.

Figure 7.25 is a plot of %R (2%DOPC in dodecane) versus %R (100% dode-cane). It shows that even 2% DOPC in dodecane can influence membrane retention to a considerable extent, compared to retentions observed in the absence of DOPC. Many molecules show retentions exceeding 70% in DOPC, under conditions where the retentions in dodecane are below 10%. However, it cannot be assumed that retention is always very low in dodecane, since several points in Fig. 7.25 are below the diagonal line, with values as high as 90% (chlorpromazine). 

Since there would be increased overall lipid concentration in the dodecane solution, we decided to create a sink condition in the acceptor wells, to lower the membrane retention. We discovered that the pH 7.4 buffer saturated with sodium laurel sulfate serves as an excellent artificial sink-forming medium. Since the new PAMPA membranes would possess substantial negative charge, the negatively charged micellar system was not expected to act as an aggressive detergent to the two-component artificial membrane infused in the microfilter. 

Figure 7.27 Membrane retentions with and without sink, 2% DOPC model.

Furthermore, the membrane retentions of the lipophilic probe molecules are dramatically decreased in the presence of the sink condition in the acceptor wells, as shown in Fig. 7.27. All molecules show R < 35%, with progesterone and phenazo-pyridine showing the highest values, 34% and 26%, respectively. 

Figure 7.28b shows that membrane retention is very systematically increased for almost all of the weak bases. This is a general pattern for bases with any of the negatively charged membrane models, and is probably best explained by the increased electrostatic attractions between the drugs and the membranes. Still, all retentions are below 50%, due to the offsetting sink condition created in the acceptor wells. 

---