Is in silico approach to high-throughput drug screening viable? An example of aminosalicylates and folic acid

Determination of various parameters such as lipophilicity, plasma protein binding etc. through conventional means is time-consuming, therefore the effectiveness of drug discovery is diminished. In silico approaches offer faster insight into several characteristics of the molecule related to pharmacokinetics, which is especially important with two or more compounds in a single formulation. Seeing that folic acid (FA) is malabsorbed in inflammatory bowel disease (IBD) patients, a fixed dose combination of one of the aminosalicylates mesalazine (MSZ), olsalazine (OSZ), sulfasalazine (SSZ) or balsalazide (BSZ) and FA is proposed. The goal of this work was to evaluate the applicability of in silico approach to predict the physico- chemical and ADME parameters of aminosalicylates and FA obtained by two chromatographic techniques. In silico approximations of logP, logD, plasma protein binding and Caco-2 permeability were predicted by 14 on-line based platforms. TLC measurements were done using RP- C18 F254 silicagel plates. HPLC measurements were done using conventional octadecylsilyl column as well as biomimetic chromatographic columns: immobilized artificial membrane, human serum albumin and α-1 acid glycoprotein. Experimental measurements showed increasing lipophilicity in the following order: MSZ (RMwTLC = -0.36, logkwC18 = 0.37) < FA < BSZ ≈ OSZ < SSZ (RMwTLC = 2.98, logkwC18 = 3.71). SSZ and OSZ showed longest retention times on HSA and AGP columns, implying high percentage of plasma protein binding (higher than 90%). Of the 14 in silico approaches, 5 of them have shown significant correlations with at least one of physico- chemical and ADME parameter (R2 0.864 to 0.998): Chemexper, Chemicalize, Molsoft, PreADME and admetSAR. Although further improvements in theoretical approaches should be done, in silico predictions show great potential in high- throughput drug screening due to high processing capability of a large sample pool and acceptable correlations with the experimental data. This work has been supported in part by the Croatian Science Foundation under the project number [UIP-2017- 05-3949]. This work has been supported in part by the European Union from the European Social Fund.