Discovery ADMET Profiling: Solubility Techniques

Throughput, accuracy, and consistency using a unified approach
Arnon Chait, Ph.D.

Figure1: The worktable of AsolW

ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) and other drug-like properties have recently taken center stage in drug discovery and development. ADMET screening is receiving the same press coverage that combinatorial chemistry and high throughput screening (HTS) received in the early 90's and for a good reason. According to the Kennedy study,⁽¹⁾ 39% of all drug candidate failures during development have poor pharmacokinetics to blame. The abundance of targets, libraries and the means to combine those using HTS resulted in a backlog of bioactive compounds, without delivering on the promise of more productive pipelines. Consistent candidate failure in pharmacokinetics necessitates using a second filter for hit selection that is not related to bioactivity. ADMET profiling is the most rational candidate.

Solubility evaluation, an important early parameter
Solubility is a key ADMET parameter involved in oral absorption. While bio-activity and metabolic stability are relatively easy to modify during development, solubility is much more difficult to improve.⁽²⁾ For this reason, solubility evaluation is an important early parameter for screening hits or even libraries prior to primary screening during the discovery process. Solubility is also an important parameter for development scientists. While solubility has a different meaning to these two crowds, the continuous blurring between discovery and development activities requires a single methodology for solubility assessment. This article will provide brief descriptions of the meaning and use of the solubility concept and the principal methods used for its measurement; it will also introduce a method to reconcile the controversy.

For discovery scientists, solubility is used to assist in rational selection of hits for further consideration. On the other hand, development scientists need to select amongst salt or crystalline forms, evaluate formulation excipients and determine minimum absorbable dose. Discovery scientists deal with DMSO-dissolved compounds, whereas only dry materials of particular crystalline form are considered in development. Thus, on a superficial level, the two disciplines are not compatible for a unified method.

Solubility is a conceptually easy parameter to measure:
1. Add an excess amount of material to water or your buffer of choice
2. Wait until the maximum possible amount is dissolved (typically under active mixing and/or at elevated temperature).
3. Measure the compound concentration in the solution (usually after filtering).

Protocols based on this definition the concentration of the dissolved compound in equilibrium with its solid are considered not useful for discovery profiling for the following key reasons and assumptions:
1. The dissolution step usually takes at least 24 hours and is thus incompatible with high throughput operations.
2. The compound is dissolved in DMSO, which has a potentially very significant influence on solubility. In other words, important information about the crystalline form of the material is already lost when the compound is pre-dissolved anyway.
3. All that is needed in discovery is an "indicator" of solubility to rationally weed out the losers anyway so why measure the real thing?

While many solubility protocols exist in practice, they typically fall into two distinct categories: methods that rely on detection of small aggregates in solution as indicators of solubility, and methods that faithfully follow the definition of solubility (measured at equilibrium). The former methods became popular in recent years due to their ease of adaptation for screening large numbers of compounds using automated liquid handlers. In most protocols, one uses titration to establish the onset of cloudiness in the solution, presumably indicating the limit of solubility. Both UV absorbance and laser light backwards scattering form the basis for commercially available systems. The methods are fast and could be completely automated, but suffer from fundamental limitations that must be explicitly recognized and should be considered when making use of data from such techniques.

Near thermodynamic saturation conditions, the dissolved solute exists as a complex population comprised of monomers, dimers, trimers, etc. Relying on detection of higher order aggregates, e.g., using an arbitrary shift in the time-averaged value of static light scattering, could be problematic due to the lack of generality of the assumption that cloudiness is related to solubility in a direct and quantifiable manner. Relaxing the definition of the parameter obtained using scattering techniques removes the pretense, allows for fast measurements of a large number of compounds, and delivers an early indicator of solubility to be measured upstream in discovery. But scattering or nephelometry methods⁽³⁾ suffer from additional serious limitations: They must use pre-dissolved compounds, and unfortunately, even small amounts of DMSO may have significant effects on the apparent solubility (referred to as "kinetic" solubility) especially for poorly soluble compounds, detection of which is after all the main reason for acquiring the data in the first place. In practice, setting an arbitrary scattering limit has been proven to be difficult to quantify due to variable background noise and other operational details.

Another method for measuring kinetic solubility is based on mixing a DMSO solution of a compound of interest with a given aqueous solvent, incubation of the mixture for a fixed period of time, removing the precipitant formed by filtration, and assaying the compound concentration by measuring the optical absorbance of the filtrate at the maximum wavelength specific for the compound.⁽⁴⁾ A parallel concentration assay (without incubation and filtration) is also performed in a separately prepared mixture of the same DMSO solution of the compound with the same aqueous solvent under higher DMSO solution/aqueous solvent ratio in the mixture. The higher DMSO/aqueous solvent ratio provides conditions under which the compound is not precipitated out of solution. The results of the compound concentration measurements in such a mixture are used as a reference point, and the concentration of the compound in the filtrate is calculated based on assuming linearity in the relation between the optical absorbance and concentration over the range of the DMSO/aqueous solvent ratio used. This procedure is based upon an additional assumption that the initial concentration of a compound under study in DMSO is known a highly questionable assumption for most parallel synthesized libraries. This method is also limited to compounds with good chromophoric groups, and since DMSO absorbs strongly in the ultraviolet region of the spectrum, it could contribute to the error inherent in optical methods. Finally, both the nephelometric and spectrophotometric methods usually necessitate measurements of multiple wells with different dilutions of the same compound, thus requiring a significant amount of sample and multiple liquid handling steps.

Measuring solubility using traditional equilibrium techniques requires mixing (shaking) the compound with the solvent of choice for at least 24 hours or until the solid reaches equilibrium (no further dissolution), then separating the saturated solution from un-dissolved residue by centrifugation or filtration, and determining the concentration of the dissolved compound by a suitable analytical assay. The analytical assay typically requires calibration, which includes preparation of several solutions of the compound in known varied concentrations and establishing a quantitative relationship between the measurable analytical signal and the compound concentration. This approach is inappropriate in a modern drug discovery environment due to the inherent low throughput of such measurements. Careful examination of the process points out to one principal hindering step: the concentration assay since mixing and incubation for an arbitrarily long time is not an issue in modern process engineered automation in which compounds are staged through the process. Utilization of equimolar detectors (elemental or light scattering) has recently demonstrated the ability to provide excellent dynamic range for solubility determination without requiring calibration or standards by using a single injection in a flow mode. The technique is useful for DMSO-dissolved and dry compounds alike, requiring very small sample sizes due to elimination of multiple dilutions and preparation of standards. It is fully automated and has a throughput range that is compatible with the vast majority of secondary screening operations today (presently ca. 250/day). Reverting to modern high throughput equilibrium techniques also offers a very important downstream benefit data consistency and integration with development.

A solubility workstation
The ASolW (Automated Solubility Workstation) from ANALIZA provides a simple and completely automated solution for true solubility testing for both development and discovery scientists. This system allows true thermodynamic or kinetic solubility testing starting from either DMSO-dissolved samples or dry samples and delivers data with integrated analysis software, all with minimal user intervention.

The ASolW includes:

Hamilton robotic liquid handling workstation

Antek Instruments equimolar nitrogen chemiluminescent detector

Custom fluidics and electronics, HPLC and vacuum pumps

Automated plate presser and temperature-controlled shaker for up to four plates

Innovative graphical user interface for assay setup, control and data analysis and reporting

ASolW can process up to 250 samples in 24 hours, typically with less than one hour of user interface. The ASolW eliminates many of the problems associated with solubility testing described above, while delivering the quality data of solubility testing for each sample. With the Antek equimolar detector, the ASolW system eliminates the necessity for good chromophores in the sample using absorbance methods, errors in detection due to background noise using turbidity techniques, and the requirement of standard curves for concentration determination and wasted sample for serial dilutions in both. ASolW can also be used to determine the concentration of the library with the addition of one extra sample.

The ASolW workstation represents a simple, turnkey method for solubility testing for both development and discovery scientists by eliminating the traditional problems with solubility measurements. As the boundaries between discovery and development continue to blur, a consistent methodology for solubility will become increasingly useful for optimizing drug candidate selection and development.

About the author
Arnon Chait, Ph.D., is the President of ANALIZA, Incorporated. Dr. Chait's training and experience covers physics, engineering and biosciences in numerous fields, concentrating on interdisciplinary research for the past two decades. ANALIZA is a Cleveland, Ohio company developing ADMET technologies for discovery and development a using a unified approach. ANALIZA offers a line of high throughput automation platforms for physicochemical parameters measurements. Using its proprietary Structural Signature Technology, ANALIZA has also pioneered applications for rapid characterization of high-order structure changes in biomolecules.

References1. Kennedy, T. Managing the drug discovery/development interface. Drug Discovery Today, 2:436-444 (1997).

2. Lipinski, C.A. Reducing the investment made in likely drug development failures. In Transforming the Pharmaceutical Industry: Adapting to Change in Technology and Markets. Cambridge Healthtech Institute, Newton, MA, 2001.

3. Bevan, C.D., Lloyd, R.S. A high throughput screening method for the determination of aqueous drug solubility using laser nephelometry in microtiter plates. Anal. Chem., 72:1781-1787 (2000).

4. Avdeef, A. High throughput measurements of solubility profiles. In: Test, B., van de Waterbeemd, H., Folkers, G., Guy, R., Eds., Pharmacokinetic optimization in drug research: Biological, physicochemical, and computational strategies. Verlag Helvetica Chimica Acta, Zurich, pp.305-326 (2001).