Substituent contributions to the transport of substituted p-toluic acids across lipid bilayer membranes

The fluxes of p-toluic acid and seven alpha-methylene-substituted analogs have been determined as a function of pH across planar egg lecithin/decane bilayers to construct a set of well-isolated polar functional group contributions to the free energy of transfer from water to the bilayer transport barrier domain. Nonlinear regression analyses of flux-pH profiles using a model which accounts for unstirred layer effects yielded membrane permeability coefficients (PRX) that varied from 1.1 cm/s for p-toluic acid to 4.1 x 10(-5) cm/s for the alpha-carbamoyl-p-toluic acid. Bulk organic solvent/water partition coefficients (KRX) were obtained for the same set of permeants using four solvent systems to identify a bulk solvent which closely resembles the chemical nature of the bilayer barrier microenvironment for these permeants. The slopes of plots of log PRX vs log KRX were 0.85, 0.91, 0.99, and 2.4, respectively, for hexadecane/water, hexadecene/water, 1,9-decadiene/water, and octanol/water with the best model solvent being that which yielded a slope closest to unity. A significant deviation in the slope from 1, as observed in the correlation with octanol/water partition coefficients, reveals that this relatively polar, hydrogen-bonding solvent is a poor model solvent for describing the barrier microenvironment for these permeants. Thus, the polar interfacial regions occupied by phospholipid head groups are not the barrier domain for the transport of the series examined in this study. The incremental group contributions to the free energy of transfer to the barrier domain (cal/mol) for the functional groups, CI, OCH3, CN, OH, COOH, and CONH2, were found to be 325, 687, 2170, 3860, 5170, and 6060, respectively.(ABSTRACT TRUNCATED AT 250 WORDS)

Solubilization of a tripeptide HIV protease inhibitor using a combination of ionization and complexation with chemically modified cyclodextrins

Kynostatin (KNI-272), an experimental HIV protease inhibitor, is currently undergoing preclinical testing for the treatment of AIDS. This transition state mimetic tripeptide exhibits extremely low aqueous solubility (4 micrograms/mL) making target concentrations (5-50 mg/mL) for parenteral solution formulations difficult to achieve. The presence of an ionizable (5-isoquinolinyloxy)acetyl moiety makes solubilization via pH adjustment possible, but a solubility > 5 mg/mL requires an adjustment in pH below 2.0, which would be physiologically unacceptable. This study examines and compares two approaches for solubilizing kynostatin: (1) inclusion complex formation at chemically distinct hydrophobic binding sites using (2-hydroxypropyl)-beta-cyclodextrin (HPCD) and a sulfobutyl ether derivative of beta-cyclodextrin (beta-CD-SBE) and (2) a combined strategy utilizing ionization of the isoquinoline moiety coupled with inclusion complex formation at the remaining binding site(s). Macroscopic binding constants determined from solubility profiles as a function of pH and HPCD concentration have been compared with the microscopic binding constant for formation of the isoquinoline-HPCD inclusion complex determined by UV difference spectroscopy to examine the independence of binding domains within KNI-272. As demonstrated in this report, combination strategies tailored to the properties of different domains within the molecule may be highly effective in solubilizing compounds such as poorly soluble peptides.

The role of intramolecular nucleophilic catalysis and the effects of self-association on the deamidation of human insulin at low pH

The influence of intramolecular catalysis and self-association on the kinetics of deamidation at the A-21 Asn residue of human insulin was explored at low pH and 35 degrees C. Observed rate constants of overall insulin degradation were determined as a function of pH over a pH range of 2.0-5.0 and as a function of total insulin concentration between pH 2.0-4.0. The pH-rate behavior of both monomeric and associated insulin degradation from pH 2.0 to 5.0 indicated intramolecular catalysis by the unionized carboxyl terminus of the A chain. Anhydride trapping with aniline at pH 3.0 provided evidence supporting the formation of a cyclic anhydride intermediate in the rate limiting step indicative of intramolecular nucleophilic catalysis. Insulin in the presence of aniline at low pH formed two anilide products, A-21 N delta 2-phenyl asparagine and N delta 2-phenyl aspartic acid human insulin, at the expense of desamido A-21 formation, consistent with the partitioning of a common intermediate. Self-associated insulin degraded at a rate approximately 2.5 times greater than that of the monomer at pH 2.0 and pH 3. However, self-association had a negligible or slight stabilizing effect on insulin decomposition at pH 4.0. An apparent downward shift in the pKa of the carboxyl terminus of approximately 0.75 units upon self-association and a catalytic rate constant which increases with -COOH acidity are proposed to account for these observations.

The relationship between permeant size and permeability in lipid bilayer membranes

Permeability coefficients (Pm) across planar egg lecithin/decane bilayers and bulk hydrocarbon/water partition coefficients (Kw-->hc) have been measured for 24 solutes with molecular volumes, V, varying by a factor of 22 and Pm values varying by a factor of 10(7) to explore the chemical nature of the bilayer barrier and the effects of permeant size on permeability. A proper bulk solvent which correctly mimics the microenvironment of the barrier domain was sought. Changes in Pm/Kw-->hc were then ascribed to size-dependent partitioning and/or size-dependent diffusivity. The diffusion coefficient-size dependency was described by Dbarrier = Do/Vn. When n-decane was used as a reference solvent, the correlation between log Pm/Kw-->hc and log V was poor (r = 0.56) with most of the lipophilic (hydrophilic) permeants lying below (above) the regression line. Correlations improved significantly (r = 0.87 and 0.90, respectively) with more polarizable solvents, 1-hexadecene and 1,9-decadiene. Values of the size selectivity parameter n were sensitive to the reference solvent (n = 0.8 +/- 0.3, 1.2 +/- 0.1 and 1.4 +/- 0.2, respectively, for decane, hexadecene, and decadiene). Decadiene was selected as the most suitable reference solvent. The value for n in bilayer transport is higher than that for bulk diffusion in decane (n = 0.74 +/- 0.10), confirming the steep dependence of bilayer permeability on molecular size. Statistical mechanical theory recently developed by the authors suggests that a component of this steep size dependence may reside in size-dependent solute partitioning into the ordered chain region of bilayers. This theory, combined with the above diffusion model, yielded the relationship, Pm/Kw-->hc = D(o)exp(-alpha V)Vn. A fit of the experimental data to this model gave the best fit (r = 0.93) with alpha = 0.0053 +/- 0.0021 and n = 0.8 +/- 0.3, suggesting that both diffusion and partitioning mechanisms may play a role in determining the size dependence of lipid bilayer permeabilities.

Protein binding of human immunodeficiency virus protease inhibitor KNI-272 and alteration of its in vitro antiretroviral activity in the presence of high concentrations of proteins

KNI-272 represents a peptide-based protease inhibitor having potent antiretroviral activity against human immunodeficiency virus (HIV) in vitro. The structure contains allophenylnorstatine [(2S,3S)-3-amino-2-hydroxy-4-phenylbutyric acid] with a hydroxymethylcarbonyl isostere. We asked whether this experimental anti-HIV agent could exert its activity in vitro in the presence of relatively high concentrations of fetal calf serum (FCS) and assessed its protein-binding properties by using fresh human plasma preparations. The 50 and 75% inhibitory concentrations of KNI-272 against HIV type 1 replication in vitro were 3- to 5-fold and 5-fold higher in the presence of 50% FCS and 15- to 25-fold and 25- to 100-fold higher in the presence of 80% ECS, respectively, than those with 15% FCS, whereas the antiviral activity of 2',3'-dideoxyinosine was not significantly affected by FCS concentrations in the culture. Detailed studies of the protein binding of KNI-272 suggest that in human plasma binding occurs predominantly to alpha 1-acid glycoprotein and that KNI-272 is probably extensively (approximately 98 to 99%) protein bound at concentrations likely to be achieved in the circulation. Thus, higher levels of KNI-272 in plasma may be required when this compound undergoes clinical trials relative to those inferred from in vitro data involving the use of 10 to 15% FCS-containing culture media. The current data may have a relevance to other antiretroviral drugs that are under development and that have a high protein-binding capacity.

Molecular distributions in interphases: statistical mechanical theory combined with molecular dynamics simulation of a model lipid bilayer

A mean-field statistical mechanical theory has been developed to describe molecular distributions in interphases. The excluded volume interaction has been modeled in terms of a reversible work that is required to create a cavity of the solute size against a pressure tensor exerted by the surrounding interphase molecules. The free energy change associated with this compression process includes the configuration entropy as well as the change in conformational energy of the surrounding chain molecules. The lateral pressure profile in a model lipid bilayer (30.5 A2/chain molecule) has been calculated as a function of depth in the bilayer interior by molecular dynamics simulation. The lateral pressure has a plateau value of 309 +/- 48 bar in the highly ordered region and decreases abruptly in the center of the bilayer. Model calculations have shown that for solute molecules with ellipsoidal symmetry, the orientational order increases with the ratio of the long to short molecular axes at a given solute volume and increases with solute volume at a given axial ratio, in accordance with recent experimental data. Increased lateral pressure (p perpendicular) results in higher local order and exclusion of solute from the interphase, in parallel with the effect of surface density on the partitioning and local order. The logarithm of the interphase/water partition coefficient for spherical solutes decreases linearly with solute volume. This is also an excellent approximation for elongated solutes because of the relatively weak dependence of solute partitioning on molecular shape. The slope is equal to (2p perpendicular - p parallel)/3KBT, where p parallel is the normal pressure component, and different from that predicted by the mean-field lattice theory. Finally, the lattice theory has been extended herein to incorporate an additional constraint on chain packing in the interphase and to account for the effect of solute size on partitioning.

Diffusion of ionizable solutes across planar lipid bilayer membranes: boundary-layer pH gradients and the effect of buffers

The diffusion of weak acids or bases across planar lipid bilayer membranes results in aqueous boundary layer pH gradients. If not properly taken into account, such pH gradients will lead to errors in estimated membrane permeability coefficients, Pm. The role of the permeant concentration, the buffer capacity, and the physicochemical properties of both permeant and buffer on the magnitude and impact of such pH gradients have been explored. A theoretical model has been developed to describe the diffusion of both permeant and buffer species. Significant pH gradients develop depending on solution pH and the pKa's, concentrations, and Pm values of both permeant and buffer. The relative error in experimentally determined Pm values was calculated as the ratio, r, between apparent Pm values (obtained from flux measurements using an equation which neglected boundary layer pH gradients) and its true value. Simulated r values ranged from 1 (0% error) to < 0.01 (> 100% error) for weak acids, decreasing with decreasing buffer capacity and increasing solute flux. The buffer capacity required for an r > 0.95 was calculated versus pH for permeants varying in pKa and Pm. Membrane-permeable buffers significantly reduce boundary layer pH gradients through a feedback effect due to buffer cotransport. Apparent Pm values of p-hydroxymethyl benzoic acid across lecithin bilayer membranes at 25 degrees C were obtained as a function of permeant concentration in various buffers [glycolic, 2-(N-morpholino)ethane-sulfonic, and formic acids]. Predictions agreed closely with experimental fluxes.

Salt and mesophase formation in aqueous suspensions of lauric acid

The solubility and solution behavior of lauric acid (LA) and its 1:1 acid soap (potassium hydrogen dilaurate) were investigated at 32 degrees C over a pH range of 2.5-8.5 and at varying KCl concentrations to examine the self-association of this long-chain carboxylic acid under these conditions. LA's solubility in water exhibited the classical pH dependence of a monocarboxylic acid with no evidence of self-association. In 0.1 M KCl between pH 6.3 and pH 7.3, filtered samples were turbid, suggesting the presence of high molecular weight aggregates (mesophase), which could be removed by ultrafiltration. The apparent LA solubility vs pH profile in ultrafiltered samples was consistent with a solid phase consisting of either the free acid (pH < 6.5) or potassium hydrogen dilaurate (pH > 6.5), again with no evidence of self-association to form low molecular weight species (dimers, etc.). Quasi-elastic light scattering (QLS) studies and mannitol trapping experiments indicated that vesicles were present in samples containing mesophase. The mesophase composition was characterized and a mass-action law for mesophase formation was developed to describe the apparent LA solubility versus pH in the mesophase region in terms of three parameters. The index of cooperativity, theta, indicated that the mesophase consists of approximately 25 molecules of LA with an acid:anion ratio, rho, of 1.7. The standard free energy of mesophase formation per mole of monomer was determined to be -6.3 kcal/mol. The aggregate size determined thermodynamically is several orders of magnitude less than that of the mesophase particle size determined by QLS measurements, suggesting that the LA monomer concentration in equilibrium with mesophase may be governed by a small unit domain of the vesicle. These observations may have a bearing on the thermodynamics of self-assembly of lipid bilayer membranes.

Solubilization of thiazolobenzimidazole using a combination of pH adjustment and complexation with 2-hydroxypropyl-beta-cyclodextrin

The thiazolobenzimidazole 1-(2,6-difluorophenyl)-1H,3H-thiazolo[3,4-a] benzimidazole, TBI, is an experimental drug for the treatment of AIDS which exhibits a low water solubility (11 micrograms/mL) and is therefore difficult to administer in an injectable solution dosage form at a target solution concentration of 10 mg/mL. The compound has a single ionizable functional group and exhibits an increase in solubility with decreasing pH consistent with a pKa of 3.55, but the maximum solubility attainable by pH adjustment has been shown to be only 0.4 mg/mL (at pH 2). TBI has been found to form inclusion complexes in either its neutral or its protonated form with 2-hydroxypropyl-beta-cyclodextrin (HPCD). The equilibrium constants for 1:1 complex formation were found to be 81 and 1033 M-1 for the protonated and neutral species, respectively. Although the formation of protonated complex is less favored in comparison to the neutral complex, the contribution of this species to the overall solubility of TBI predominates at low pH. Thus, using a combined approach of pH adjustment and complexation with HPCD, a solubility enhancement of 3 orders of magnitude is possible. NMR proton spectroscopy and molecular modeling studies, conducted to understand the orientation of TBI in the complex and the effect of protonation, are described.

Solubilization and stabilization of an anti-HIV thiocarbamate, NSC 629243, for parenteral delivery, using extemporaneous emulsions

The O-alkyl-N-aryl thiocarbamate, I, (2-chloro-5-[[(1-methyl-ethoxy)thioxomethyl]amino]benzoic acid, 1-methylethylester, NSC 629243, also known as Uniroyal Jr.) is an experimental anti-HIV drug with very low water solubility (1.5 micrograms/mL). Early clinical studies required an injectable solution at approximately 15 mg/mL, representing a solubility increase of approximately 10(4)-fold. Adequate solubilization of this hydrophobic drug was achieved in 20% lipid emulsions. Extemporaneous emulsions were prepared by adding a concentrated drug solution to a commercially available parenteral emulsion. Various methods of preparation to minimize drug precipitation during its addition and enhance redissolution of precipitated drug were evaluated. The stability and mechanism(s) of decomposition of NSC 629243 in both 20% lipid emulsions and in natural oil vehicles were examined. In lipid emulsions, the shelf life at 25 degrees C varied from 1 to > 10 weeks, depending on the extent to which air was excluded from the preparation. The shelf life of 50 mg/mL solutions in natural oils at 25 degrees C varied from < 1 to > 100 days depending on the oil and its supplier. A qualitative correlation was found between the initial rate of oxidation and the peroxide concentration in the oil. The primary degradation product in both systems was shown to be a disulfide dimer, II, formed via oxidation. Oxidation was inhibited by vacuum-sealing of emulsion formulations or incorporation of an oil-soluble thiol, thioglycolic acid (TGA), into oil formulations. TGA may inhibit oxidation by consuming free radicals or peroxide initiators or by reacting with the disulfide, II, to regenerate the starting drug.

A method for the early evaluation of the effects of storage and additives on the stability of parenteral fat emulsions

The combination of sedimentation field-flow fractionation (SedFFF) and photon correlation spectroscopy (PCS) is shown to provide a detailed record of the droplet sizes present in fat emulsions commonly used in parenteral nutrition. The technique presented has been used to record size distribution data for a particular emulsion (Liposyn-II), demonstrating its high stability and lot-to-lot uniformity. The technique is also able to demonstrate how additions of small amounts of electrolytes [0.45% (w/v) NaCl, 0.05% (w/v) CaCl2] tend to destabilize the emulsion, suggesting some caution in the use of total parenteral nutrition (TPN) mixtures. In contrast, a 1:1 mixture with human serum caused no sign of instability in the Liposyn-II. Using the emulsion as a carrier for lipophilic drugs necessitates adding solutions of the drug in nonaqueous solvents, such as DMSO (dimethyl sulfoxide). This solvent's destabilizing effect results in a droplet coalescence that becomes severe after 3-5 days following a 10% (v/v) addition, while a 5% (v/v) addition reaches the same level of coalescence in 10 days.

Changes in drug sensitivity of human immunodeficiency virus type 1 during therapy with azidothymidine, dideoxycytidine, and dideoxyinosine: an in vitro comparative study

Human immunodeficiency virus type 1 (HIV-1) strains were isolated from nine patients before and after prolonged therapy with either an alternating regimen of 3'-azido-3'-deoxythymidine (AZT) and 2',3'-dideoxycytidine (ddC) (AZT/ddC) or 2',3'-dideoxyinosine (ddI) alone. All strains obtained from four patients who received AZT/ddC for up to 41 mo were highly insensitive to AZT in vitro. Only one strain obtained after AZT/ddC therapy showed reduced susceptibility to ddC in addition to AZT and had previously unreported amino acid substitutions in the viral polymerase-encoding pol region, whereas three other strains had one or more of the five previously reported AZT-related mutations. In five HIV-1 strains from patients who received ddI for up to 29 mo, no appreciable decrease in sensitivity to ddI was detected. Two strains isolated after ddI therapy had no significant amino acid mutations, although three strains had a mutation reportedly associated with ddI administration. These data suggest that HIV-1 develops reduced susceptibility to AZT more readily than to ddC and ddI and/or that the reduced susceptibility to ddC and ddI is modest in degree. Moreover, the present data suggest that an alternating regimen of AZT and ddC does not block the emergence of AZT-insensitive variants. It should be noted, however, that the current results do not provide a basis for concluding that AZT/ddC or ddI is inferior, equivalent, or superior to AZT as therapy of AIDS.

Central nervous system targeting of 2',3'-dideoxyinosine via adenosine deaminase-activated 6-halo-dideoxypurine prodrugs

AIDS dementia complex is a neurologic disorder, characterized by increasingly severe cognitive, behavioral, and motor impairment, which is associated with human immunodeficiency virus (HIV) infection in the central nervous system (CNS). Many of the dideoxynucleosides effective systemically in the treatment of HIV infections, such as 2',3'-dideoxyinosine (ddI), exhibit limited penetration into the CNS and limited or variable effectiveness in reversing the symptoms of AIDS dementia. Thus, approaches for increasing the CNS uptake of ddI and other dideoxynucleosides are needed. The CNS uptake of a series of 6-halo-2',3'-dideoxypurine ribofuranosides (6-halo-ddPs) previously shown to be active against HIV because of their conversion to ddI through the action of adenosine deaminase was examined in rats. In vitro studies in rat blood and brain tissue homogenate suggested a favorable selectivity for bioconversion in brain tissue, but with bioconversion half-lives varying widely within the series. In vivo infusions of 6-chloro-ddP (6-Cl-ddP), 6-bromo-ddP (6-Br-ddP), and 6-iodo-ddP (6-I-ddP) resulted in significant increases (20- to 34-fold) in the ddI concentration ratios in brain parenchyma/plasma when compared with those after an infusion of ddI alone. Absolute concentrations of ddI in brain parenchyma were increased 10- and 4-fold, respectively, following 30-min infusions of 6-Cl-ddP or 6-Br-ddP, but were 2.4-fold lower after an infusion of 6-I-ddP relative to that after a control infusion of ddI. Detailed studies of the plasma pharmacokinetics, CNS uptake kinetics, and bioconversion of 6-Cl-ddP were conducted to compare in vivo transport and bioconversion parameters with those predicted from in vitro measurements and to rationalize the efficiency of CNS delivery of ddI from 6-Cl-ddP. The results show that increased lipophilicity alone does not ensure that a given prodrug will deliver higher levels of a parent compound to the CNS. Both the selectivity and absolute rate of bioconversion in the brain are important factors.

Transport methods for probing the barrier domain of lipid bilayer membranes

Two experimental techniques have been utilized to explore the barrier properties of lecithin/decane bilayer membranes with the aim of determining the contributions of various domains within the bilayer to the overall barrier. The thickness of lecithin/decane bilayers was systematically varied by modulating the chemical potential of decane in the annulus surrounding the bilayer using different mole fractions of squalene in decane. The dependence of permeability of a model permeant (acetamide) on the thickness of the solvent-filled region of the bilayer was assessed in these bilayers to determine the contribution of this region to the overall barrier. The flux of acetamide was found to vary linearly with bilayer area with Pm = (2.9 +/- 0.3) x 10(-4) cm s-1, after correcting for diffusion through unstirred water layers. The ratio between the overall membrane permeability coefficient and that calculated for diffusion through the hydrocarbon core in membranes having maximum thickness was 0.24, suggesting that the solvent domain contributes only slightly to the overall barrier properties. Consistent with these results, the permeability of acetamide was found to be independent of bilayer thickness. The relative contributions of the bilayer interface and ordered hydrocarbon regions to the transport barrier may be evaluated qualitatively by exploring the effective chemical nature of the barrier microenvironment. This may be probed by comparing functional group contributions to transport with those obtained for partitioning between water and various model bulk solvents ranging in polarity or hydrogen-bonding potential. A novel approach is described for obtaining group contributions to transport using ionizable permeants and pH adjustment. Using this approach, bilayer permeability coefficients of p-toluic acid and p-hydroxymethyl benzoic acid were determined to be 1.1 +/- 0.2 cm s-1 and (1.6 +/- 0.4) x 10(-3) cm s-1, respectively. From these values, the -OH group contribution to bilayer transport [delta(delta G0-OH)] was found to be 3.9 kcal/mol. This result suggests that the barrier region of the bilayer does not resemble the hydrogen-bonding environment found in octanol, but is somewhat less selective (more polar) than a hydrocarbon solvent.

Synthesis and anti-HIV activity of isonucleosides

A series of isomeric 2',3'-dideoxynucleosides which contains a modified carbohydrate moiety has been prepared. This class of compounds was designed to mimic the activity of known anti-HIV dideoxynucleosides, while imparting enhanced chemical and enzymatic stability. Isonucleosides containing the standard heterocyclic bases (A, C, G, T) were synthesized via nucleophilic addition of the base to an isomeric sugar unit. Modified derivatives were generated by manipulation of the intact isonucleoside. Two of the compound prepared, iso-ddA (1) and iso-ddG (6), exhibit significant and selective anti-HIV activity, as well as beneficial hydrolytic stability.

Dose dependence in the plasma pharmacokinetics and uptake kinetics of 2',3'-dideoxyinosine into brain and cerebrospinal fluid of rats

Dose dependence in the plasma pharmacokinetics of 2',3'-dideoxyinosine (ddI) was examined during and after 2-hr iv infusions in rats at infusion rates of 12.4, 32.7, and 125 mg/kg/hr. After termination of the infusions, the disappearance of ddI from plasma was distinctly biphasic, suggesting that the majority of ddI is eliminated before distribution equilibrium is achieved. The mean alpha t1/2 following the infusions was 2.7 min and was independent of dose. The mean terminal half-life (beta t1/2) was approximately 24 min and also independent of dose. Nonlinear pharmacokinetic behavior in plasma after infusions was manifested in a decreased clearance with increasing dose, as determined from steady state plasma concentrations of ddI during infusions. In parallel with the decreased clearance, the apparent volume of distribution of the central compartment, Vcapp, decreased with increasing dose. Nonlinearity in clearance with increasing dose could be accounted for using a model which includes rapid, saturable tissue binding. Dose dependence in the kinetics of uptake of ddI into brain tissue and cerebrospinal fluid (CSF) was also examined during and after iv infusions. Steady state concentrations of ddI in brain tissue and CSF varied linearly with steady state plasma concentrations over a plasma concentration range of greater than 30-fold. Mean tissue to plasma concentration ratios, expressed as percentages, were 2% in CSF, 5% in brain tissue, and 1-2% in brain parenchymal tissue (corrected for the contribution of the cerebral vascular space).