Kinetic solubility is measured by taking a DMSO stock solution (20 mg/mL) of a compound and adding it to a pH 7.4 buffer, filtering, and measuring the concentration of the resulting solution. The assay is run in a 96-well format. Results are reported in |g/mL and classified as high (>60 |g/mL), moderate (1060 |g/mL), or low (<10 |g/mL). These solubility ranges correlate with high, moderate, and low oral absorption (Lipinski et al., 1997). However, solubility is also important for the behavior of compounds in in vitro and in vivo assays in the discovery phase of drug research. The kinetic solubility assay closely resembles the conditions of an in vitro bioassay where a DMSO stock solution of a compound is added to a buffer.
Permeability is measured using the Parallel Artificial Membrane Permeability Assay (PAMPA) (Kansy et al., 1998). In this assay, an artificial membrane composed of phosphatidyl choline and dodecane is adhered to a filter plate, which is stacked on top of a receiving plate filled with pH 7.4 buffer. A DMSO stock solution of a compound is added to a pH 7.4 buffer solution that is added to the filter plate. Following an 18 hour incubation, the concentration of compound is measured in the receiving plate and the filter plate. The assay is run in a 96-well format. Results are reported as cm/sec and classified as high (>1 x 10-6 cm/sec), moderate (0.1 - 1 x 10-6 cm/sec), or low (<0.1 x 10-6 cm/sec). These ranges also correlate to high, moderate, and low oral absorption for compounds having average to high solubility. A second version of this assay for categorizing blood brain barrier permeability (Di et al., 2003) is also part of the standard profile, so two permeability data points are generated for each compound. Results are reported as cm/sec and classified as CNS+ (>4 x 10-6 cm/sec), CNS+/- (2 - 4 x 10-6 cm/sec), or CNS- (<2 x 10-6 cm/sec). Since these assays utilize artificial membranes, they only measure passive diffusion across membranes.
Metabolic stability is measured by incubation of compounds with rat liver microsomes. The stability is measured at a single compound concentration of 3 |M, and results are reported as percent of parent compound remaining after 15 minutes and half life (t1/2). Ranges are classified as high (>80% remaining), moderate (20 - 80% remaining), and low (<20% remaining).
Cytochrome P450 (CYP) inhibition is measured for three CYP isozymes: 3A4, 2C9, and 2D6. The inhibition is measured at a single concentration of 3 | M and is reported as percent inhibition at this concentration. The results are classified as high (>50% inhibition), moderate (15 - 50% inhibition), and low (<15% inhibition).
As mentioned above, all of these assays are run in a 96 well format at a single concentration and are intended to quickly indicate potential issues with compounds. As compounds advance in the discovery phase, any issues are followed up with more in depth work, such as determining IC50 values for CYP inhibition, looking at other measurements of permeability in Caco-2 cells, and running pharmacokinetic studies. However, the standard pharmaceutical profile is useful for looking at trends in series of compounds, particularly when there are large numbers of compounds in a series where more detailed information such as IC50s may not be available. Pharmaceutical profiling data can be used in a negative sense to rank series and to deprioritize those that have an issue across many members of the series. It can also be used in a more positive sense to highlight properties of a series that need to be optimized in parallel with biological properties such as potency and functional activity. A series of compounds undergoing optimization may range in size from tens of compounds initially to hundreds of compounds as the optimization process progresses. In the early stage of discovery there are generally several types of biological data generated, including binding, functional and or cellular activity, and selectivity. When these data are combined with the pharmaceutical profiling data for tens or hundreds of compounds, interpretation by simple inspection of spreadsheets is not possible, unless one focuses on only one variable. But by focusing on only one variable one may miss correlations that involve multiple variables. Table I illustrates part of a typical data matrix. This table includes twenty-six compounds from a larger set that were synthesized for a serotonin 5-HT6 receptor modulator project (Ellingboe et al., 2004). The twelve variables include eight calculated properties (molecular weight, plogD at pH 4.0, plogD at pH 7.4, plogD at pH 9.0, number of hydrogen bond donors, topological polar surface area, number of hydrogen bond acceptors, and number of rotatable bonds) and the results from four assays (solubility at pH 7.4, PAMPA, blood brain barrier PAMPA, and 5-HT6 binding).
While this data table is relatively small, it is clear that it is not easy to understand all correlations with only visual inspection. The effective utilization of pharmaceutical profiling data to optimize physical properties along with biological properties requires the use of multivariate data analysis tools.
Was this article helpful?