Chromatographic Behaviour Predicts the Ability of Potential Nootropics to Permeate the Blood-Brain Barrier

The log BB parameter is the logarithm of the ratio of a compound’s equilibrium concentrations in the brain tissue versus the blood plasma. This parameter is a useful descriptor in assessing the ability of a compound to permeate the blood-brain barrier. The aim of this study was to develop a Hansch-type linear regression QSAR model that correlates the parameter log BB and the retention time of drugs and other organic compounds on a reversed-phase HPLC containing an embedded amide moiety. The retention time was expressed by the capacity factor log k′. The second aim was to estimate the brain’s absorption of 2-(azacycloalkyl)acetamidophenoxyacetic acids, which are analogues of piracetam, nefiracetam, and meclofenoxate. Notably, these acids may be novel nootropics. Two simple regression models that relate log BB and log k′ were developed from an assay performed using a reversed-phase HPLC that contained an embedded amide moiety. Both the quadratic and linear models yielded statistical parameters comparable to previously published models of log BB dependence on various structural characteristics. The models predict that four members of the substituted phenoxyacetic acid series have a strong chance of permeating the barrier and being absorbed in the brain. The results of this study show that a reversed-phase HPLC system containing an embedded amide moiety is a functional in vitro surrogate of the blood-brain barrier. These results suggest that racetam-type nootropic drugs containing a carboxylic moiety could be more poorly absorbed than analogues devoid of the carboxyl group, especially if the compounds penetrate the barrier by a simple diffusion mechanism.


Abstract
The log BB parameter is the logarithm of the ratio of a compound's equilibrium concentrations in the brain tissue versus the blood plasma. This parameter is a useful descriptor in assessing the ability of a compound to permeate the bloodbrain barrier. The aim of this study was to develop a Hansch-type linear regression QSAR model that correlates the parameter log BB and the retention time of drugs and other organic compounds on a reversed-phase HPLC containing an embedded amide moiety. The retention time was expressed by the capacity factor log k'. The second aim was to estimate the brain's absorption of 2-(azacycloalkyl)acetamidophenoxyacetic acids, which are analogues of piracetam, nefiracetam, and meclofenoxate. Notably, these acids may be novel nootropics. Two simple regression models that relate log BB and log k' were developed from an assay performed using a reversed-phase HPLC that contained an embedded amide moiety. Both the quadratic and linear models yielded statistical parameters comparable to previously published models of log BB dependence on various structural characteristics. The models predict that four members of the substituted phenoxyacetic acid series have a strong chance of permeating the barrier and being absorbed in the brain. The results of this study show that a reversed-phase HPLC system containing an embedded amide moiety is a functional in vitro surrogate of the blood-brain barrier. These results suggest that racetam-type nootropic drugs containing a carboxylic moiety could be more poorly absorbed than analogues devoid of the carboxyl group, especially if the compounds penetrate the barrier by a simple diffusion mechanism.

Introduction
The blood-brain barrier (BBB) is one of the most important barriers in the human body. It protects brain and spinal cord tissues against unwanted compounds. Endothelial cells and brain capillaries comprise the barrier. Unlike other capillaries in the body, spaces between capillaries in the BBB are filled and tightened by tight junctions, which are membrane formations that are produced by parenchymal cells from the brain. Although the BBB can be considered both a physical barrier and a complex biochemical interface with important physiological functions, such as specific uptake by the brain [1], several transport mechanisms are involved in allowing drugs and other xenobiotics to pass through the BBB. These mechanisms are absorptive-mediated transcytosis, receptor-mediated transcytosis, passive transport, and carrier-mediated transport [2]. Most drugs are transported into the brain by means of passive transport or simple diffusion. It is generally accepted that the transport velocity depends on the lipophilicity of the compound. When the lipophilicity is expressed as log P (octanol/water), the optimal values generally range between 1.5 and 2.7, and the maximum value of log P is typically around 2.1 [1]. However, an increasing body of evidence suggests that the velocity of passive transport also depends on hydrogen bonding. The log BB parameter, which is the logarithm of the equilibrium brain/blood concentrations ratio, is frequently used to assess the likelihood that a compound can permeate the BBB. The parameter is expressed as follows: log BB = c brain /c blood . Young and co-workers realised that log BB values of a small set of six centrally acting histamine H 2 antagonists correlated much more closely to the difference Δlog P, where Δlog P = log P (octanol/water) -log P (cyclohexane/water), than with log P (octanol/water) itself [3]. These results were in accordance with the results of Seiler, who had previously introduced the parameter Δlog P as a measure of the overall hydrogenbonding capacity of a compound [4]. The logarithm of the capacity factor log k' is expressed in equation 1 where t r is the retention time of the compound, and t 0 is the dead retention time, namely the retention time of a compound that is not retained on the column. Log k' is a measure of the position of a sample peak during column chromatography [either gas chromatography (GC) or high-performance liquid chromatography (HPLC)]. If log k' is calculated from the retention time of a compound in a reversed-phase high-performance liquid chromategraphy (RP-HPLC) system, it can be used as a substitute for a log P value. The partition between the hydrophobic stationary phase and the hydrophilic mobile phase is the primary means by which compounds separate on a column. Notably, log k' values depend on the environment in which the constituent parameters are measured. Octadecyl silica gel phases represent a purely lipophilic environment, whereas a reversed-phase system containing an "embedded" amide moiety ( Figure 1) may be able to account for the hydrogen bonding that may occur during the partitioning process.  Fig. 1.
The structure of a reversed-phase column system containing an "embedded" amide moiety (such as Supelcosil ABZ Plus+ used in this study).
The aim-and simultaneously the novelty-of this study was to mimic the behaviour of drugs and other organic compounds at the BBB by using RP-HPLC embedded with an amide moiety. Specifically, the aim was to find a function (equation 2)

Eq. 2.
log BB = f (log k') by means of regression analysis and to subsequently use it to predict whether some potential and experimental nootropic drugs that may function as novel CNS therapeutics can cross the BBB. Nootropics are drugs that enhance learning and memory by facilitating the flow of information between the cerebral hemispheres. Nootropics also fortify the CNS against chemical and physical injuries. Nootropics are also used during the initial treatment stages of neurodegenerative diseases. Although this therapeutic group is structurally very diverse, the racetames subgroup is relatively structurally uniform. The racetames are characterised by the presence of a pyrrolidine-2-on moiety [5,6]. Piracetam, 2-(2-oxopyrrolidine-1-yl)acetamide, is the most frequently prescribed drug of this group. Its more hydrophobic derivative nefiracetam, N- , is closely structurally related to the experimental drugs, the log BB values of which were estimated by means of regression models developed in this study ( Figure 2).

Development of regression models
The retention times of the initial set of 21 compounds (listed in Excluding the outliers improved the correlation. Two approaches to identify outliers were tested: (a) the stepwise exclusion of influential points, which were identified by statistical tests; (b) the stepwise exclusion of points with the greatest residual value. The residual value is the difference between a particular log BB value reported in the literature and the predicted value. The residual value is expressed as a percentage of the reported value. Approach (a) led to the exclusion of indomethacin, which was also identified as an outlier by both the Atkinson distance and its influence on the regression parameters. This model was not better than the previous one (equation 6). These results indicated that exclusion of compounds as outliers based on their residual value is an unreliable method.

Discussion
The two final regression models are, from a statistical viewpoint, equal to or better than previous models that have aimed to predict the structural properties that enable a molecule to permeate the BBB. The correlation of log BB and log k' in the quadratic model (equation 10) is comparable to the model of Lombardo and co-workers [10], where log BB is dependent on the free energies of solvation. The plain linear model (equation 15) exhibits better characteristics, although it is based on a training set containing fewer members. Both models presented here also have better statistical parameters than those of Young and co-workers [3]. The models developed in this study are also comparable to more complex models that use computed parameters, including log BB dependence and descriptors expressing the electrotopological state of a molecule [12]. In conclusion, equations 10 and 15 confirm that the behaviour of a compound in a reversed-phase HPLC containing an embedded amide moiety can mimic the behaviour of the BBB. The models presume permeation by a passive diffusion mechanism. Notably, previous models are also based on this assumption [3,10,12].

Tab. 2.
Compounds of the prediction set with predicted log BB values.

Cpd. code
Position of (2-oxoazacycloalkyl)acetamido moiety at the benzene ring Differences between the log k' and log BB values of 2-(2-oxoazacycloalkyl)acetamidophenoxyacetic acids are small. Notably, the values are predicted with both of the models in this study. Cruciani and co-workers [13] classified organic drugs and druglike compounds by their predicted ability to cross the blood-brain barrier; these predictions were based on the log BB values of the structures. If log BB > 0, the compound readily crosses the BBB and reaches a greater concentration in the brain than in the blood. If 0 > log BB > 0.3, the compound may or may not cross the BBB. However, if log BB < −0.3, the compound is highly predicted not to cross the BBB. If this classification is applied to the results shown in Table 2 [7], N-(4-methoxyphenyl)-2-(2-oxopyrrolidin-1yl)acetamide, has log BB values of 0.078 and −0.011 in the quadratic and plain linear models, respectively. Thus, compound No. 12 has a markedly stronger chance of permeating the BBB. Based on these results, the carboxy group seems to significantly limit a compound's blood-brain permeability. This is in agreement with results of Platts and co-workers [8], who introduced a carboxylic acid indicator variable to obtain a multi-linear regression model for the dependence of log BB on several calculated structural properties. Some authors even demonstrated carrier-mediated transport mechanisms for both short-and long-chain aliphatic carboxylic acids as well as aromatic carboxylic acids [14]. The absorption of these experimental nootropics could be improved if members in the FOK series were transported through the BBB via such mechanisms.

Conclusions
A reversed-phase HPLC that had been fitted with an embedded amide moiety was used to build two simple QSAR regression models between log BB and log k', namely, a quadratic model as well as a plain linear one. These models were derived by a repeated linear regression that excluded outliers in a stepwise manner. The outliers were influential points. Both models showed that such an HPLC system can mimic the behaviour of the bloodbrain barrier. In addition, estimating the log BB values of potentially novel racetam-type nootropics using these models, suggests that carboxy groups markedly inhibit a compound's ability to pass through the BBB by a simple diffusion mechanism.

Materials
All of the drug substances used in this study were of pharmacopoeial grade and were obtained and purchased from pharmaceutical suppliers. Potential nootropics, used for log BB estimation, were prepared in our laboratory as previously described [7,11]. Their identities were confirmed by 1 H-NMR, 13 C-NMR, and IR spectroscopies. Compounds were at least 95% pure (HPLC). Methanol and water were of HPLC grade and were purchased from Merck, Darmstadt, Germany. Potassium dihydrogen phosphate, toluene, xylenes, propan-2-one, butan-2-one, and 1,1,2-trichloroethylene were all of analytical grade. Sodium hydroxide, sodium phosphate, and sodium azide, purchased from Lach-Ner, Neratovice, Czech Republic, were all of analytical grade. The reversed-phase analytical HPLC column with an embedded amide moiety, namely, the hexadecanoylamidopropyl substitution on silica gel hydroxyls, was purchased from Sigma-Aldrich, Prague, Czech Republic (Supelcosil ABZ Plus+, 150 x 4.6 mm, 3 µm particle size).

HPLC equipment and measurement conditions and procedure
HPLC measurements were performed on a set consisting of an LCP 4020 isocratic pump (Ecom, Prague, Czech Republic), Rheodyne® (Rheodyne, Rohnert Park, USA). The system utilised a manual sample injector equipped with a 20-μL loop. The ABZ Plus+ column (above) was maintained at ambient temperature. An LCD 2083 UV-VIS detector was linked to Clarity integration software (Ecom, Prague, Czech Republic). A mobile phase consisting of methanol and a phosphate buffer (pH 7.4, 0.008 mol/L) was used. The flow rate was maintained at 0.6 mL/min. A wavelength of 254 nm was used for spectrophotometric detection. Solid compounds were dissolved in the mobile phase composition to a final concentration of 0.1 mg/mL. Aromatic liquids were diluted to 0.1 μL/mL; non-aromatic liquids were diluted to 1 μL/mL. Basic compounds, supplied as their salts, had been previously converted into their bases with sodium phosphate. Sodium azide at a final concentration of 0.1 mg/mL in the mobile phase composition, was used to determine the dead retention time; the determined value was 2.62 min.

Data processing tools
Linear regression analyses were performed using QCExpert 2.5 Statistical Software (Trilobyte, Pardubice, Czech Republic) run on a Pentium 4 2.66 GHz computer.