Study of the mechanisms of drug passage through biological barriers aimed to optimize bioavailability and/or blood-brain barrier permeation

Phospholipophilicity, i.e. the affinity of a drug for phospholipids, is one of the key features modulating pharmacokinetics. It would be desirable to assess it in the early stages of pharmaceutical drug development. The fastest method to achieve phospholipophilicity measures is liquid chromatography performed on stationary phases functionalized with phospholipid analogues, so called Immobilized Artificial Membrane (IAM). The logarithms of the chromatographic retention coefficients of drugs, measured or extrapolated to 100% aqueous phase on such stationary phases (log kwIAM), are often used to surrogate drug membrane passage data measured in vivo or in situ. Indeed, the research group I worked with proposed a novel method to unravel the total drug/phospholipid interaction forces in a lipophilic/hydrophobic component, mainly described by the n-octanol/water lipophilicity, possibly acting as a “driving force” in membrane barrier passage, and a polar/electrostatic one, possibly acting instead as a “trapping force”. The latter is parameterized by ∆log kwIAM, a novel physico-chemical parameter proposed by the research team I worked with. It is the difference between phospholipophilicity values determined by IAM-HPLC (log kwIAM) and the values expected on the basis of n-octanol/water lipophilicity of the neutral forms of the analytes (log PN). The rationale of this parameter arises from the fact that log kwIAM values of structurally non-related neutral compounds having polar surface area (PSA) equal to zero relate unambiguously with n-octanol lipophilicity values by highly significant direct linear relationships (r2 0.96) in a log P range 1.15-4.80; in contrast, such relationships are no longer observed for analytes having PSA greater than 0 and, in larger extent, for ionizable analytes. In fact, for instance, poorly and medially lipophilic bases, ionized at the experimental conditions, show an IAM chromatographic retention higher than that of neutral isolipophilic compounds. Such peculiar behavior was referred to the occurrence of an extra interaction component, mainly of electrostatic nature, actually taking place in drug-phospholipid interaction. ∆log kwIAM was found as a suitable descriptor of the drug/biomembrane polar/electrostatic interaction component. This component was demonstrated as related by highly significant inverse linear relationships to Blood-Brain Barrier (BBB) penetration data (log BB). For the BBB passage of acidic compounds, it was necessary to magnify the electrostatic interaction component by calculating it taking into account log D7.4 rather than log PN so yielding ∆’log kwIAM values. In this Ph.D. thesis, a study has been performed for (a) the elucidation of the molecular mechanisms actually involved in drug/biomembrane interactions (b) verifying the soundness of ∆log kwIAM in predicting different data of passive drug uptake either measured in vivo or in vitro (c) the improvement of the throughput of the technique so as to offer to the pharmaceutical companies an high-throughput screening method to evaluate new drug candidates. The relationships between data of BBB penetration (log BB) and ∆log kwIAM reported in recent works were based on a limited dataset (n= 21). As a consequence, the first part of the present research was focused on validating the proposed model on an enlarged dataset. Therefore, further 21 analytes, whose log BB values were known and taken from a single bibliographic source, were taken into account and their chromatographic retention coefficients were determined on two different stationary phases, i.e. IAM.PC.MG and IAM.PC.DD2. A good relationship between log BB and log PN values could be observed after the exclusion of two data points (chlorambucil and domperidone), but fairly good and highly significant inverse linear relationships were observed between log BB and ∆log kwIAM.MG (r2 = 0.681) and ∆log kwIAM.DD2 (r2 = 0.825), respectively, for the all the analytes. The results from our previous work were then assembled with the newly determined values in a single doubled dataset (n = 42) to verify the model proposed; again remarkably significant linear inverse relationships were achieved when plotting log BB values vs either ∆log kwIAM.MG (r2 = 0.738) or ∆log kwIAM.DD2 (r2 = 0.826), with only two points (haloperidol and chlorpromazine) behaving as outliers. Subsequently, starting from the observation according to which membrane passive diffusion of analytes has been suggested to be a universal process, regardless the different composition or function of the biological barrier involved, possible relationships between delta values and intestinal absorption data as measured by the in situ LOC-I-GUT perfusion technique were for the first time investigated. For this second part of this research activity, 15 structurally unrelated analytes, known to be passively absorbed at intestinal level, were taken into account and their IAM chromatographic retention coefficients were determined. Although moderate linear direct relationships between intestinal absorption data and log P values were found, again much more significant relationships were obtained by plotting intestinal absorption data vs ∆log kwIAM.MG (r2 = 0.803) and ∆log kwIAM.DD2 (r2 = 0.784). However, these results, albeit really encouraging, were based on a limited dataset. However, this model had to be validated by taking into account a larger number of biological data, and this opportunity was provided by permeation data achieved on cultured cells model, such as Caco-2 and MDCK cultured cells. Indeed, Caco-2 and MDCK permeation assays, for the morphological similarity of cellular monolayers employed to the intestinal epithelium, are claimed as mirroring rather closely drug intestinal absorption. However, in such assays the in vitro apparent permeability values, log Papp, i.e. the crude permeation data, can be separated into four contributions i) aqueous boundary layer (represented by the accessible intestinal surface area - PABL), ii) filter-determined permeability related to the polycarbonate porous support of the cultured cells (Pf), transcellular permeability (PC), and paracellular permeability (Ppara). The role played by these contributions markedly differs between in vitro and in vivo systems. Therefore, the investigation was carried out employing two datasets: the first one consisting of 38 compounds whose crude Caco-2 permeation data, log Papp, were reported in the literature; the second one consisting of 47 compounds whose Caco-2/MDCK permeation data were corrected to express the sole transcellular intrinsic permeability of the drugs, log P0Caco-2/MDCK. As to the the first dataset, log Papp values were found as related to the apparent lipophilicity of the analytes measured at pH 7.4 (log D7.4) by a parabolic trend, and reasonable relationships between log Papp and delta values were only visible selecting the analytes heavier than 300 Da. Indeed, for these compounds, the occurrence of paracellular passage mechanisms can be reasonably excluded. On the other hand, when taking into account the second dataset, highly significant inverse linear relationships between log P0Caco-2/MDCK and ∆log kwIAM.MG (r2 = 0.765) and ∆log kwIAM.DD2 (r2 = 0.806) were achieved. Indeed, log P0Caco-2/MDCK express the sole transcellular intrinsic permeability. The last part of this Ph.D. project was devoted to the improvement of throughput technique so as to appeal pharmaceutical companies, too. The first strategy applied was the development and validation of partial-least-squares (PLS) based statistic models, starting from molecular descriptors calculated in silico aimed at predicting phospholipophilicity data measured on IAM stationary phases. This led to the development of two mathematical models able to predict phospholipophilicity as measured on IAM.PC.MG and IAM.PC.DD2, with an accuracy of 75% and 79%, respectively. These results allow a rapid and reliable in silico prediction of delta log kwIAM values suitable for accurate estimates of the intestinal absorption/BBB entering potential of new leads or hypothetical molecules. In parallel, the conditions of the analytical methods were optimized to gain experimental values in a reasonably short time. The coupling of the LC system to an Electrospray Ionization Source (ESI) – Time of Flight (TOF) mass spectrometer detector allowed to analyze the compounds of interest simultaneously in mixtures of up to 10 compounds at the same time thanks to the higher selectivity of m/z ratio. This approach resulted in the development of an MS analytical method 100 times faster than the one traditionally employed in this sort of determinations.