Naturally occurring molecules are excellent sources of lead compounds. A series of oleanolic acid (OA) derivatives previously synthesized in our laboratory, which show promising antitumor activity, have been analyzed in terms of lipophilicity evaluation applying chromatographic and computational approaches.
Journal of Chromatography A 1656 (2021) 462552 Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma Multivariate assessment of anticancer oleanane triterpenoids lipophilicity Monika Pastewska a, Barbara Bednarczyk-Cwynar b, Strahinja Kovacˇ evic´ c, Natalia Buławska a, d Szymon Ulenberg d, Paweł Georgiev d, Hanna Kapica a, Piotr Kawczak d, Tomasz Baczek ˛ , a a,∗ Wiesław Sawicki , Krzesimir Ciura a ´ sk, Al Gen Hallera 107, 80-416 Gdan ´ sk, Poland Department of Physical Chemistry, Medical University of Gdan ´ , Poland Department of Organic Chemistry, Poznan University of Medical Science, Grunwaldzka 6, 60-780 Poznan University of Novi Sad, Faculty of Technology Novi Sad, Department of Applied and Engineering Chemistry, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia d ´ sk, Al Gen Hallera 107, 80-416 Gdan ´ sk, Poland Department of Pharmaceutical Chemistry, Medical University of Gdan b c a r t i c l e i n f o Article history: Received June 2021 Revised September 2021 Accepted 10 September 2021 Available online 15 September 2021 Keywords: Oleanane triterpenoids Lipophilicity Chemometrics RP-HPLC IAM a b s t r a c t Naturally occurring molecules are excellent sources of lead compounds A series of oleanolic acid (OA) derivatives previously synthesized in our laboratory, which show promising antitumor activity, have been analyzed in terms of lipophilicity evaluation applying chromatographic and computational approaches Retention data obtained on three reversed-phase liquid chromatography stationary phases (RP-HPLC) and immobilized artificial membrane chromatography (IAM-HPLC) were compared with computational methods using chemometric tools such as cluster analysis, principal component analysis and sum of ranking differences To investigate the molecular mechanism of retention quantitive structure retention relationship analysis was performed, based on the genetic algorithm coupled with multiple linear regression (GA-MLR) The obtained results suggested that the ionization potential of studied molecules significantly affects their retention in classical RP-HPLC In IAM-HPLC additionally, polarizability-related descriptors also play an essential role in that process The lipophilicity indices comparison shows significant differences between the computational lipophilicity and chromatographically determined ones © 2021 The Author(s) Published by Elsevier B.V This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Introduction Optimizing lipophilicity is an essential process in drug discovery since it noticeably affects the diffusion of molecules through a biological membrane Consequently, lipophilicity determines pharmacokinetic processes, including absorption, distribution, metabolism, excretion (ADME), as well as the toxicity of drug candidates [1–3] According to the International Union of Pure and Applied Chemistry (IUPAC) definition, lipophilicity represents the affinity of a molecule or a moiety towards a lipophilic environment Lipophilicity is typically determined by solute distribution in biphasic liquid-liquid or solid-liquid systems [4] The first protocol for lipophilicity assessment was initially proposed by Hansh and based on the shake-flask procedure, which enables the determination of the partition coefficient of the target compound between n-octanol and water (logP) [5] ∗ Corresponding author E-mail address: krzesimir.ciura@gumed.edu.pl (K Ciura) Nevertheless, this method has many disadvantages - it is effortful, time-consuming, and requires large amounts of organic solvent and pure substances For these reasons, methods based on solidliquid partitioning, such as reversed-phase liquid chromatography (RP-LC), are currently used for lipophilicity estimation Both IUPAC and organisation for Economic Co-operation and Development (OECD) consider RP-LC equivalent to the shake-flask approach It is worth emphasizing that other separation techniques, like micellar liquid chromatography (MLC), micellar electrokinetic chromatography (MEKC), or microemulsion electrokinetic chromatography (MEEKC), can also be used for lipophilicity value designation Separation methods owe their popularity to numerous advantages First of all, they are reproducible, easy to automate, rapid, and require small amounts of analytes that don’t need to be absolutely pure because their impurities are readily separated during the chromatographic process Consequently, the separation methods have become the primary approach to lipophilicity assessment and easily fit into a high-throughput technique As a property affecting the biological activity, chromatographically determined lipophilicity can then be processed using the Quantitative https://doi.org/10.1016/j.chroma.2021.462552 0021-9673/© 2021 The Author(s) Published by Elsevier B.V This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 Structure-Retention Relationship (QSRR) approach [6] QSRR is a valuable support in predicting the behavior of new molecular entities and can provide insight into molecular mechanisms Another way for lipophilicity assessment are the theoretical methods that provide quick information about the lipophilic properties of target molecules The calculation approach enables the estimation of lipophilicity during the design of drug candidates before their synthesis Several programs dedicated to logP calculation are plain, freely available online, and generate results nearly instantaneously Mentioned in silico approaches are significantly faster and cheaper when compared to experimental methods They require no need for laboratory experiments, specialized equipment, and any chemical reagents Nevertheless, it is worth keeping in mind that various studies noted significant differences between the calculated logP and experimental data for some chemicals [2] Biomimetic chromatography with an immobilized artificial membrane (IAM) can be used to assess affinity for phospholipids due to the presence of a stationary phase that contains the phosphatidylcholine head group As a consequence, IAM-HPLC can emulate the lipid membrane monolayer [4] This affinity is essential for the drug molecule to achieve its therapeutic target (e.g., receptor) and to produce therapeutic effects IAM-HPLC links the advantages of HPLC, such as rapid analysis and automatization, with more biosimilar nature of stationary phase Therefore, it is frequently used by both academia and pharmaceutical companies Terpenoids are a large and structurally diverse family of natural compounds [7] Their structure is based on the five-carbon 2methyl-1,3-butadiene group, also called isoprene units These compounds can be divided into several classes depending on the connection pattern of isoprene units, which is related to the quantity of those moieties [7] Common classes that we can distinguish are oleanes, lupanes, and ursanes In the oleane group the most important is oleanolic acid (OA), which shows various pharmacological effects [8], such as anti-inflammatory [9], spasmolytic [10], hepatoprotective [11], antiviral [12], antiallergic [13], hypoglycemic [14,15], cytoprotective and primarily antitumorigenic potential [7] On account of simply altering the chemical structure and tremendous potential in activity at various stages of tumor development, OA is a perfect hit molecule [7,16,17] A library of OA derivatives synthesized in our laboratory shows various anticancer activities [16–23] The target compounds present a higher level of anticancer activity towards breast cancer MCF-7, cervical cancer KB (HeLa) [19], Cellosaurus CCRF-CEM, CCRF-VCR10 0, CCRF-ADR50 0, and human leukemia HL-60, SR [18,24] cell lines [17] The main aim of this study was to assess the lipophilicity indices of previously synthesized OA derivatives from the retention behavior on three different RP-HPLC stationary phases and IAMHPLC and compare the experimental indices with those obtained through computational methods using a chemometric approach Additionally, our investigation has two secondary goals The first of them was the selection of the OA structures which should exhibit optimal lipophilicity, high antitumor activity, and the best pharmacokinetic properties for further studies In order to realize this purpose, investigated molecules were also characterized using SwissADME software The second goal focuses on exploring the molecular mechanism of retention in RP-HPLC using the QSRR approach dissolved in DMSO (1 mg/ml) except for compound 18, dissolved in THF (1 mg/ml) Each stock solution of analytes was stored at 2–8 °C Dilutions of the compounds (100 μg/mL) were made just before analysis 2.1.2 The analytical standards In order to carry out chromatographic hydrophobicity index of IAM (CHIIAM ) determination, the model substances were used The analytical standards of octanonophenone, butyrophenone, and acetanilide were provided by Alfa Aesar (Haverhill, MA, USA); acetaminophen, acetophenone, were purchased from SigmaAldrich (Steinheim, Germany); heptanophenone, hexanophenone, valerophenone, propiophenone, and acetophenone were bought from Acros Organic (Massachusetts, United States) 2.1.3 Reagents Ultrapure water (18.2 M × cm−1 ) used to prepare the mobile phase was purified and deionized in our laboratory via a Millipore Direct-Q UV Water Purification System (Millipore Corporation, Bedford, MA, USA) All the analytical reagents were used without prior purification Disodium phosphate (Na2 HPO4 ) and monosodium phosphate (NaH2 PO4 ) were supplied by POCH (Gliwice, Poland) Dimethyl sulfoxide (DMSO), used as a solvent, was from Merck (Darmstadt, Germany) Acetonitrile (LiChrosolv®) and tetrahydrofuran (gradient grade for liquid chromatography) were purchased from Sigma-Aldrich (Steinheim, Germany) Metabolic stability was performed using the following reagents: pooled human liver microsomes and sodium salt of NADPH were obtained from Sigma (Sigma-Aldrich, Saint-Louis, MO, USA), while monopotassium phosphate and dipotassium phosphate were from POCH (POCH, Gliwice, Poland) 2.2 Chromatographic analysis All HPLC analyses were performed on Prominence-1 LC-2030C 3D HPLC system (Shimadzu, Japan) equipped with DAD detector and controlled by LabSolution system (version 5.90 Shimadzu, Japan) In all cases, the concentrations of the investigated analytes were 100 μg/mL (in DMSO or THF), and the injected volume was 20 μL In each system phase A was 10 mM phosphate buffer, and phase B was acetonitrile (ACN) Hence, retention times (tR ) of investigated triterpenoids were collected, and their detection in all systems was performed at 200 nm The temperature, ACN concentration, pH of the buffer, columns suppliers, and dead times, together with the flow for each chromatographic system used in the study, are given below • • • Materials and methods 2.1 Materials • 2.1.1 Oleanolic acid derivatives The 2D structures and SMILES notation of the target OA derivatives are presented in Table S1, whereas their synthesis and characterization were described in the literature [18] All samples were CN chromatography: 10 mM phosphate buffer at pH 7.4; the flow rate 1.5 mL/min The chromatography was carried out at 40 °C on Agilent SB-CN column (4.6 mm x 150 mm x 3,5 μm; Zorbax; USA, dead time = 1.22 min) with a linear gradient phase B 50–100% C18 chromatography: 10 mM phosphate buffer at pH 7.4; the flow rate 1.5 mL/min The chromatography was carried out at 40 °C on Symmetry C18 column (3.9 × 150 mm x μm; Waters; USA, dead time = 0.847 min) with a linear gradient phase B 70–100% Phenyl (Ph) chromatography: 10 mM phosphate buffer at pH 7.4; the flow rate 0.2 mL/min The chromatography was carried out at 30 °C on Unison UK-Phenyl column (2 mm x 150 mm x μm; Imtakt; USA dead time = 2.461 min) with a linear gradient phase B 30–100% IAM chromatography: 10 mM phosphate buffer at pH 7.4; the flow rate 1.5 mL/min The chromatography was carried out at 30 °C on IAM.PC.DD2 column (10 × 4.6 mm x 10.0 μm; Regis Technologies; USA), additionally equipped with IAM guard column M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 In CN, C18 , and Ph chromatography, two gradient runs were applied during experiments differing in gradient time (tG equal to 20 and 40 min) According to the assumption proposed by Snyder and co-workers [25,26], appropriate logkw values (i.e., the retention factor logk extrapolated to 0% organic modifier, as an alternative to logP) were calculated DryLab 6.0 software (Molnar Institute, Berlin, Germany) was used to make this computation Dwell volume for these HPLC systems was measured at 0.780 mL The IAM-HPLC analyses were carried out in one gradient run and the analysis time was 6.5 The studied compounds’ CHIIAM indices were obtained using a calibration set of reference substances using the protocol proposed by Valko and co-workers [27] Each HPLC analysis was run in triplicate The obtained retention times for the target OA derivatives are summarized in Table S2, S3 and S4 2.4.2 Sum of ranking differences (SRD) analysis In the present study, the SRD approach was carried out in order to rank and select the best lipophilicity measures of oleanane triterpenoids obtained by both in silico and experimental (chromatographic) methods The benchmark was defined as row-average (consensus approach) calculated on the basis of all data in one row The variables used in the SRD analysis were normalized by minmax normalization method and scaled in the range between 0.01 and 0.99 The SRD results were validated by comparison of rank by random numbers (CRRN) and 7-fold cross-validation method [54] The cross-validation was carried out by omitting approximately 1/7 of objects and by performing the ranking on the remaining objects [55] The SRD values were normalized (SRD%) in order to easily compare the results of different SRD analyses 2.4.3 QSRR calculation QSRR analysis was done using QSARINS 2.2.4 version software developed by Gramatica et al [27,28] Descriptors selection was supported by a genetic algorithm (GA), whereas multiple regression was employed as a regression method The set of parameters applied to control GA was the size of the population—100 and the mutation rate—25% The analyzed OA derivatives were divided into two groups, the training group (n = 25, ≈ 76%) and the testing group (n = 8, ≈24%, compounds no: 12, 19, 20, 22, 23, 24, 27, 31) The split was random prior to QSRR analysis The model fitting, robustness, and predictive abilities were assessed by the coefficient of determination (R2 ), predictive squared correlation coefficient (Q2 ), and root-mean-squared error of cross-validation (RMSECV) coming from the leave-one-out cross-validation technique Furthermore, root-mean-square error in prediction (RMSEP) deriving from external validation was calculated 2.3 In-silico calculation 2.3.1 Theoretical physicochemical descriptors Models were prepared as mol files using ACD Chemsketch (Advanced Chemistry Development, Inc., Toronto, Canada), and afterward converted into Gamess [25] input files using Open Babel software (The Open Babel Package, version 2.3.1 http://openbabel.org; accessed Oct 2011) [26] Geometrical optimization of 2D models was performed on DFT level of theory, using B3LYP/6–311 G(d) parameters Optimized compounds were afterward subjected to Dragon 7.0 (Talete, Milano, Italy) software to calculate physicochemical descriptors used later to develop presented models 2.3.2 In-silico calculation of lipophilicity Several software packages were used for lipophilicity calculations, whereas each is based on different algorithms Four different logP values (ILOGP, WLOGP, Silicos-IT LogP, Consensus LogP) were calculated using a virtual logP calculator available online: http: //www.swissadme.ch (developed and maintained by the Molecular Modeling Group of the Swiss Institute of Bioinformatic) KOWWIN logP values were obtained using KOWWIN v 1.68 software (EPI Suite package v.4.2, U.S EPA) XLOGP3, MLOGP, AlogP were attained with Virtual Computational Chemistry Laboratory (VCCLAB, http:// www.vcclab.org/) ACD/ChemSketch logP was derived from ChemSketch software (version 12.1.0.31258; ACD/Labs, Toronto, Canada) Results and discussion 3.1 Lipophilicity estimation by computational methods Nowadays, computer-aided drug design (CADD) plays an essential role in the drug discovery and development process Among huge numbers of theoretical descriptors, lipophilicity remains the most important one As a consequence, many computer programs can estimate the logP values based on various algorithms Generally, the computational approaches for lipophilicity determination have several advantages over the experimental procedures First, the calculated logP indices may be obtained before synthesis and supported the design process of new derivatives with desired lipophilicity Another benefit is that computational methods can save time and chemical reagents, making it very attractive from the economic and environmental points of view Computational logP values studied compounds estimated by means of different software are summarized in Table 1, whereas the classification of the investigated software based on algorithm type is presented in Table The 2D structures and SMILES codes of target solutes are listed in Table S1 The logP values calculated by iLogP are significantly lower for each tested substance It is worth emphasizing that the computed value of logP is almost twice lower than that of other programs iLogP platform is based on calculating solvation free energy, which is a relatively new approach for lipophilicity calculation [29] On the contrary, the highest values are obtained for ACD/LogP This software implemented a classic algorithm that is based on the principle of isolating carbons It should be highlighted that the same chemical structure features completely different logP parameters For example, obtained value iLOGP for molecule is equal 5.09, whereas attained ACD/LogP quantity is 10.42, while the lipophilicity for this compound, calculated using programs based on hybrid and fragmentary algorithms, is about 6–7 Among 2.3.1 In-silico calculation of biological properties SwissADME data were obtained using a web-based application available online: http://www.swissadme.ch (developed and maintained by the Molecular Modeling Group of the Swiss Institute of Bioinformatic) Input structures for both calculations were generated on the basis of optimized structures as a set of mol files The calculated pharmacokinetic properties are summarized in Table S5 2.4 Data analysis 2.4.1 Cluster analysis and principal component analysis Cluster Analysis (CA) and Principal Component Analysis (PCA) were applied for databases which included the obtained chromatographic lipophilicity data and the calculated logP using STATISTICA 13.3 (StatSoft, Tulsa, Oklahoma, USA) Before analysis, data were standardized to eliminate the impact of different scales by using the Z-score scaling algorithm (V = mean of V/δ , where V is the value of variables and δ is the standard deviation) CA has been carried out using Ward’s amalgamation rule and the Euclidian distance measure For the purpose of clustering lipophilicity measures based on the sum of ranking differences, the double dendrograms in clustered heat maps were obtained using NCSS 2021 Statistical Software (NCSS, LLC Kaysville, Utah, USA, ncss.com/software/ncss.) applying Ward’s algorithm and Euclidean distances M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 Table The calculated logP values of the OA derivatives with respect to the computational model No iLogP XLogP3 WLogP MLogP Silicos-IT LogP Consensus LogP ALogP LogP KOWWIN ACD/ChemSketch logP 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 3.94 3.96 4.67 5.09 4.57 4.52 4.38 4.76 4.54 4.44 4.21 4.20 4.04 4.46 4.20 3.77 3.79 3.63 3.44 4.48 4.32 4.19 4.28 4.03 4.38 4.20 5.06 4.89 4.66 4.73 4.54 5.02 4.91 7.49 8.06 7.81 8.39 7.5 7.75 6.57 7.17 8.07 7.87 7.56 7.81 6.63 7.23 8.13 7.17 7.43 6.24 7.74 7.32 6.75 6.43 6.69 5.50 6.10 7.00 7.62 7.05 6.73 6.99 5.80 6.40 7.30 7.23 7.80 7.32 7.89 7.53 7.79 6.45 7.25 8.02 7.28 7.49 7.75 6.41 7.21 7.98 7.44 7.70 6.36 7.93 7.07 6.50 6.71 6.97 5.63 6.43 7.20 7.20 6.63 6.84 7.10 5.75 6.55 7.33 5.82 6.06 6.01 6.25 5.92 5.88 5.47 5.92 5.82 6.21 6.11 6.07 5.66 6.11 6.01 5.73 5.69 5.28 5.63 5.33 5.06 4.97 4.96 4.55 4.97 4.89 5.56 5.31 5.22 5.20 4.79 5.22 5.12 5.85 6.36 6.41 6.92 6.99 6.70 6.27 7.72 7.32 6.58 7.17 6.87 6.44 7.90 7.50 6.44 6.14 5.71 6.76 6.63 6.10 6.69 6.39 5.96 7.42 7.03 6.76 6.24 6.83 6.53 6.10 7.55 7.18 6.07 6.45 6.44 6.91 6.50 6.53 5.83 6.56 6.76 6.48 6.51 6.54 5.84 6.58 6.76 6.11 6.15 5.44 6.30 6.17 5.75 5.80 5.86 5.14 5.86 6.06 6.44 6.02 6.05 6.11 5.40 6.15 6.37 6.42 6.80 6.67 7.05 6.63 6.72 5.77 6.67 7.17 6.65 6.61 6.70 5.75 6.65 7.15 6.38 6.47 5.52 6.92 6.12 5.74 5.70 5.79 4.84 5.74 6.24 6.29 5.92 5.88 5.97 5.02 5.92 6.42 7.95 8.96 8.24 9.25 7.73 7.52 7.27 7.15 9.14 8.27 8.10 7.54 7.64 7.52 8.48 7.44 7.23 6.98 8.85 8.19 7.18 6.67 6.46 6.21 6.10 8.08 8.00 7.00 6.48 6.27 6.02 5.91 7.89 9.06 9.96 9.52 10.42 8.94 9.30 7.86 8.73 9.45 7.94 7.37 7.73 6.28 7.15 7.87 8.48 8.84 7.39 8.98 8.14 7.25 7.01 7.37 5.92 6.79 7.40 9.92 9.02 8.44 8.80 7.36 8.23 8.95 Table List of software used with information regarding algorithms and suppliers No softwere Algorhitms Supplier iLogP http://www.swissadme.ch XLogP3 WLogP MLogP Silicos-IT LogP Consensus LogP ALogP KOWWINlogP ACD/ChemSketch logP Physics-based method relying on free energies of solvation in n-octanol and water calculated by the Generalized-Born and solvent accessible surface area (GB/SA) model An atomistic method including corrective factors and knowledge-based library Purely atomistic method based on the fragmental system of Wildman and Crippen An archetype of topological method relying on a linear relationship with 13 molecular descriptors An hybrid method relying on 27 fragments and topological descriptors Arithmetic mean of the values predicted by the five proposed methods Ghose-Crippen octanol-water partition coeff (logP); based on molecular properties Atom-based approach and fragmental contribution Based on the principle of isolating carbons tested derivatives and 32 have the greatest theoretical lipophilicity, especially when compared to parent OA structures Generally, the newly synthesized structures demonstrate similar lipophilicity properties as OA Slightly more hydrophilic are molecules 16, 17, 18, 25, and 31 All tested substances represented relatively high lipophilicity Referring to the Lipinski’s rule of five [30], it should be noted that the given compounds not exceed the critical value of lipophilicity (logP 0.79), suggesting that similar intermolecular interactions govern retention in investigated RP-HPLC systems Considering the limits of this traditional RP-HPLC stationary phase to mimic the electrostatic interactions between molecule and phospholipid membranes, IAM chromatography experiments (which combines membrane simulation with rapid measurements) were performed The first thing that draws attention is the lowest correlation coefficient (range between 0.60–0.68) that we observe in the obtained correlation matrix, comparing all tested chromatographic systems This probably occurred because the ionization for some OA derivatives significantly affects its retention under IAM whereas neutral compounds showed similar retention on RP-HPLC and IAM [40–48] Generally, the IAM stationary phase surface is mainly zwitterionic at pH 7.4, and interact with the negatively and positively charged molecules [49–51] According to the theory proposed by Avdeef and co-workers, choline moieties (positively charged at pH 7.4) are located in the outer part of the IAM layer In contrast, the phosphate groups are negatively charged at the same pH and present in the phase’s inner part Consequntly, the bases are more retained than acids with the same log P value because they interact more in-depth [50-51] Whereas the C18 , Phe, and CN phases are neutral, excluding negative and undesirable influences of free-silanol groups These findings suggest that the membrane interaction of target OA derivatives has a significantly different nature than lipophilic/hydrophobic interactions Nevertheless, the less lipophilic structures (molecule no 18), also showed the weakest affinity to phospholipids membranes CHIIAM of investigated OA derivatives ranges from 38.51 to 72.22, which seems to be lower values than might be expected given for consideration the highly lipophilic nature of the compounds In the case of OA’s initial structure, it exhibits a moderate affinity to phospholipids from among the tested compounds, CHIIAM = 56.03 (whereas mean and median are 56.03 and 55.50, respectively), even though both calculated and chromatographically determined lipophilicity values were the lowest among tested molecules This suggests that although the introduced modifications to the structure of the OA usually increased lipophilicity, it did not translate unequivocally into a change in the affinity for phospholipids M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 Table The summarized values of logkw and CHIIAM indices for OA derivatives No logkw C18 logkw Ph logkw CN CHIIAM 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 2.16 3.75 4.21 5.64 4.69 4.37 3.35 4.19 4.84 3.56 3.94 3.69 2.98 3.62 4.18 2.93 3.07 3.41 3.16 4.23 3.33 3.53 3.32 3.06 3.43 3.96 4.26 3.26 3.60 3.45 3.25 3.72 3.27 3.66 3.67 3.98 4.38 4.05 4.07 3.66 4.04 4.33 3.87 3.57 3.92 3.59 3.81 4.07 3.34 3.69 3.36 3.52 3.84 3.65 3.53 3.51 3.56 3.79 3.83 4.06 3.77 3.77 3.85 3.89 4.04 3.70 4.22 4.87 4.87 5.39 4.99 5.03 4.47 4.95 5.36 4.67 4.86 5.03 4.36 4.77 5.11 4.21 4.28 4.09 4.43 5.08 4.67 4.71 4.66 4.25 4.56 5.00 5.10 4.63 4.72 4.72 4.68 4.97 4.21 56.03 58.12 72.22 69.40 64.52 71.58 58.55 60.76 60.90 61.69 56.40 61.49 49.83 53.49 55.01 49.46 56.98 38.51 47.27 55.39 54.15 51.58 54.66 45.86 49.19 51.83 60.14 60.06 55.50 60.37 48.02 52.05 53.34 mation of lipophilicity database The scree plot (Fig 4) confirms that the three principal components have the most crucial information Presented on figure PCA loadings indicated that PC2 differentiates chromatographic and computational lipophilicity indexes The situation with the CHIIAM parameter, which lies between the two approaches, is interesting Taking into account the significant differences between IAM-HPLC and other tested lipophilicity indices, the present observations suggest that in the case of OA derivatives, the IAM-HPLC provides new information about compounds that go beyond classical lipophilicity Weighing the advantages and limitations of unsupervised chemometric tools such as PCA and CA, it should be highlighted that both methods not allow for the selection of the best and the worst approaches for lipophilicity measurement What is more, they not include information about statistical figures of performed analysis For this reason, the SRD analysis, introduced by Héberger [53,54], was applied for the ranking and selection of lipophilicity indices [53,55,56] The sum of ranking differences (SRD) as a non-parametric and robust method in the last decade has become a widely used tool in ranking compounds, mathematical models, samples, objects, analytical techniques, etc [53–57] The application of the SRD approach in molecular science is reflected in the possibility to rank the molecules in terms of their molecular features, such as lipophilicity [55–57] The SRD method is based on the calculation of absolute differences of ranks between defined reference ranking (known as a benchmark) and each variable which describes every object [53,54] Those absolute differences are eventually summed into SRD values A smaller SRD value means that the particular variable is closer to the benchmark; in other words: the smaller SRD, the better variable The results of conducted SRD-CRRN analysis are presented in Fig As it can be seen from the graph the Consensus LogP lipophilicity descriptor is the closest to the reference ranking and can be considered to be the best lipophilicity measure of oleanane triterpenoids derivatives or “real” consensus logP descriptor It is followed by ALOGP and XLOGP3 as quite suitable lipophilicity descriptors of the analyzed series of compounds It must be emphasized that according to the Wilcoxon’s matched pairs test, these lipophilicity descriptors are statistically different in terms of their SRD% values obtained in 7-fold cross-validation procedure (Fig 6) The following measures, including logkw Ph, WLOGP, logkw C18, logkw CN, CHIIAM , MLOGP, KOWWINlogP, and ACD/ChemSketchlogP descriptors are grouped close to each other in terms of similar SRD values in Figs and in as well This is confirmed by Wilcoxon’s matched pairs test as well The descriptor Silicos-IT LogP is separated from this group significantly The iLOGP descriptor is placed furthest from the reference ranking and aforementioned group of descriptors, but it is placed quite close to random number distribution described by Gaussian curve Therefore, iLOGP can be considered to be the worst parameter for the description of lipophilicity of the analyzed group of oleanane triterpenoids Chromatographically determined lipophilicity parameters including logkw Ph, logkw C18, logkw CN and CHIIAM not overmatch the computationally estimated lipophilicity measures – Consensus LogP, ALOGP and XLOGP3 Nevertheless, the parameters logkw Ph, logkw C18, logkw CN and CHIIAM are placed relatively close to the reference ranking and far from the random number distribution CHIIAM , as an experimentally determined phospholipophilicity measure, is placed very close to MLOGP in silico lipophilicity descriptor This can be observed in the SRD graph in Fig and in the Box-Whisker plot in Fig Generally, the SRD-CRRN analysis indicated that all the experimentally determined lipophilicity indices are statistically very similar to WLOGP, MLOGP, KOWWINlogP and ACD/ChemSketchlogP in silico lipophilicity descriptors, as it can be noticed in Fig 3.4 Comparison of chromatographic and computational lipophilicity indexes In order to investigate similarities and dissimilarities between chromatographic and computational lipophilicity measurements of the investigated molecules, principal component analysis (PCA), cluster analysis (CA), and the sum of ranking differences (SRD) were performed The above chemometric methods are complementary to each other and provide insight into the data structure Among the agglomerative clustering methods, Ward’s method was selected due to its unique properties It is based on a classical sumof-squares criterion which produces groups and minimizes dispersion within-group at each binary fusion [52] The results of CA are presented in Fig Two main clusters can be separated The first cluster included logkw obtained from C18 , CN and Ph chromatography together with two calculated logP: iLOGP and SilicosIT logP These software calculations were based on solvation free energy and topological descriptors corrected by fragmental information, respectively In the second cluster, two subclusters can be distinguished In the first subcluster, IIa following descriptors are located: calculated logP, XLOGP3, WLOGP, ALOGP, Consensus logP, KOWWINlogP and MLOGP The second subcluster IIB comprises of CHIIAM indices and logP calculated by ACD/ChemSketch This CA analysis indicated significant differences between phospholipids affinity determined chromatographically and others lipophilicity indices PCA analysis confirmed the conclusions of the CA PCA is one of the basic multivariate techniques which provides an insight into a data structure, similarities and dissimilarities of variables, disposition of objects, tendencies for their grouping, and outlying effects Summary of PCA analysis is noticed in supplementary information as Table S7 The first three-component included 86% of the infor7 M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 Fig The results of CA The probabilities that lipophilicity measures are derived randomly are provided in Supplementary materials in Table S8 For Consensus LogP descriptor this probability is so small (less than 3.95E-10) so it can be neglected The highest likelihood of random character is assigned to iLOGP descriptor (between 1.51 and 1.73) The clustering in the form of double dendrogram in clustered heat maps of the lipophilicity measures of oleanane triterpenoids was carried out based on their SRD% values The double dendrogram is presented in Fig in which two main clusters of the lipophilicity measures can be observed Silicos-IT LogP and iLOGP descriptors are placed into the separate cluster having the highest SRD% values in all steps of 7-fold cross-validation All the other descriptors are collected in other main cluster consisting of two subclusters The dendrogram indicates the grouping of Consensus LogP, ALOGP and XLOGP3 having the lowest SRD% values These results can be excepted, since Consensus logP the average of in silico logP values calculated by SwissADME software (iLOGP, XLOGP3, WLOGP, MLOGP and Silicos-IT LogP) The rest of the descriptors are placed into the other subcluster according to similar SRD% values This is the confirmation of the grouping of the descriptors assumed in Figs and 3.4 QSRR analysis Fig The PCA scree plot Insights into the molecular mechanism of chromatographically determined lipophilicity indices were performed using the QSRR approach, accepting the assumption that all determined chromatographic indices are the expression of the lipophilic character of the investigated analytes Generally, QSRR methodology, introduced by Kaliszan, linked the relationship between retention and analyte structures mathematically [58] Obtained QSRR models allow for the prediction and explanation of the nature of interaction involved in the retention mechanism which takes place between the compounds and the employed stationary phases The selection of theoretical descriptors, which affected the logkw , was performed by using GA-MLR method As a result, four Despite the fact that CA and SRD methods have quite different computational basics, the comparison between the gruping of the lipophilicity measures obtained by CA (Fig 3) and SRD indicates some similarities For example, the Consensus LogP descriptor is placed in the same subcluster as ALOGP and XLOGP3 descriptors, which are quite close to each other in the SRD graph as well Also, iLOGP and Silicos-IT LogP descriptors are separated from other in silico descriptors as in SRD graph M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 Fig The ranking of normalized lipophilicity measures of oleanane triterpenoids by sum of ranking differences and comparison of ranks by random numbers with row average as a reference ranking The statistical characteristics of Gaussian fit are the following: first icosaile (5%), XX1 = 294; first quartile, Q1 = 334; median, Mediana (Med) = 360; last quartile, Q3 = 388; last icosaile (95%), XX19 = 428 Fig The Box-Whisker plot representing the normalized sum of ranking differences values (SRD%) obtained by 7-fold cross-validation of each lipopilicity parameter The vertical lines separate statistically different lipophilicity parameters by the means of Wilcoxon’s matched pairs test models, each describing retention in investigated chromatographic systems, have been calculated and listed in Table together with the statistical figures The values of theoretical descriptors and their description were listed in tables S9 and S10, respectively Considering the number of OA derivatives in the tested group, a maximum of five molecular descriptors were added into calculated QSRR equations since at most five analytes should be used for one independent variable [59] The descriptors in the equations were introduced from most important to less significant, taking the pvalue into account The statistical features of obtained models show an excellent fitting of the data and indicate a good prediction ability of the models The small RMSEEXT confirmed that the model is not only well-fitted for the training but also predicts correctly Each obtained model dedicated for classical RP-HPLC the GATS7i descriptor was included These results suggested that the ionization potential of molecules significantly affected their retention Additionally, the obtained models suggested that 3D-MoRSE descriptors can be very useful for predicting physicochemical properties of this class of compounds since they appear very often in calculated models On the other hand, the polarizability (H7p) and ionization (MATS8i) related descriptors are present and significantly affect the CHIIAM values logkPh = −7.853(±1.224)GATS7i – 65.9445(±27.345)VE2sign_G/D + 0.303(±0.115)Mor20v + 0.095(±0.0249)Mor29s + 0.0225(±0.0036)ALOGP2 + 11.389(±1.250) logkCN = 154.291(±22.817)VE2sign_D – 11.227(±2.177)GATS7i – 0.701(±0.179)Eig13_EA(ri) – 2.662(±0.896)E2e – 17.392(±1.989) logkC18 = −23.978(±7.382)GATS7i + 38.491(±7.664)SpMin1_Bh(v) + 0.829(±0.120)Mor05p + 0.684(±0.158)Mor26i – 31.912(±4.753)Du – 34.418(±16.549) CHIIAM = −145.279(±14.572)MATS8i −100.391(±11.688)SpMin4_Bh(m) + 12.463(±4.551)H7p + 11.636(±0.838)SssssC – 0.373(±0.138)G(N .S) + 233.858(±21.322) RMSEext Q2 RMSEcv R2 0.888 0.812 0.109 0.208 0.874 0.819 0.148 0.108 0.912 0.858 0.263 0.281 0.947 0.906 2.236 2.171 Journal of Chromatography A 1656 (2021) 462552 Fig The clustered heat map (double dendrogram) of the lipophilicity measures of oleanane triterpenoids based on SRD% values obtained by 7-fold cross-validation of each lipophilicity parameter Conclusions Nowadays, concern about the development of new anticancer drugs is still increasing on account of enlarged susceptibility for neoplastic disease among society A series of OA derivatives previously synthesized in our laboratory show auspicious various antitumor activity Among tested solutes promising anticancer-activity, IC50 2.8 and 1.6 μg/ml for HeLa and MCF-7 cell line and relatively low lipophilicity among tested substances showed molecule no 24 SwissADME calculation suggested that these molecules should be absorbed orally, which is highly desirable and should not exceed the BBB, limiting side effects related to CNS QSRR analysis indicated that the ionization potential of studied molecules significantly affects their retention in classical RP-HPLC, whereas on IAMHPLC, polarizability-related descriptors also play an essential role The chemometric analysis presented differences between calculated logP and lipophilicity obtained chromatographically This fact can explain the nature of computed lipophilicity parameters, which refer to octanol/water liquid/liquid static partition equilibrium At the same time, chromatographically measure parameters refer to the dynamic equilibrium constant on a large surface of the stationary phases According to SRD, the best lipophilicity indices of studied derivatives are consensus LogP descriptors followed by ALOGP and XLOGP3, while chromatographically determined lipophilicity parameters not overmatch them Equation Model Table The four models which describe retention in investigated chromatographic systems M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al 10 M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 Declaration of Competing Interest [15] X.-Y Zeng, Y.P Wang, J Cantley, T.J Iseli, J.C Molero, B.D Hegarty, E.W Kraegen, Y Ye, J.-M Ye, Oleanolic acid reduces hyperglycemia beyond treatment period with Akt/FoxO1- induced suppression of hepatic gluconeogenesis in type-2 diabetic mice, PLoS ONE (2012) [16] B Bednarczyk-Cwynar, P Ruszkowski, T Jarosz, K Krukiewicz, Enhancing anticancer activity through the combination of bioreducing agents and triterpenes, Future Med Chem 10 (2018) 511–525, doi:10.4155/fmc- 2017- 0154 [17] B Bednarczyk-Cwynar, P Ruszkowski, D Atamanyuk, R Lesyk, L Zaprutko, Hybrids of oleanolic acid with norbornene-2,3-dicarboximide-n-carboxylic acids as potential anticancer agents, Acta Pol Pharm - Drug Res 74 (2017) 827–835 [18] D Kaminskyy, B Bednarczyk-Cwynar, O Vasylenko, O Kazakova, B Zimenkovsky, L Zaprutko, R Lesyk, Synthesis of new potential anticancer agents based on 4-thiazolidinone and oleanane scaffolds, Med Chem Res 21 (2012) 3568–3580, doi:10.10 07/s0 044- 011- 9893- ´ Anticancer ef[19] B Bednarczyk-Cwynar, L Zaprutko, P Ruszkowski, B Hładon, fect of A-ring or/and C-ring modified oleanolic acid derivatives on KB, MCF7 and HeLa cell lines, Org Biomol Chem 10 (2012) 2201–2205, doi:10.1039/ c2ob06923g [20] A Paszel-Jaworska, B Rubis´ , B Bednarczyk-Cwynar, L Zaprutko, M Ry´ bczynska, Proapoptotic activity and ABCC1-related multidrug resistance reduction ability of semisynthetic oleanolic acid derivatives DIOXOL and HIMOXOL in human acute promyelocytic leukemia cells, Chem Biol Interact 242 (2015) 1–12, doi:10.1016/j.cbi.2015.07.011 ´ B Rubis´ , M Pakuła, B Bednarczyk[21] N Lisiak, A Paszel-Jaworska, E Toton, ´ Cwynar, L Zaprutko, M Rybczynska, Semisynthetic oleanane triterpenoids inhibit migration and invasion of human breast cancer cells through downregulated expression of the ITGB1/PTK2/PXN pathway, Chem Biol Interact 268 (2017) 136–147, doi:10.1016/j.cbi.2017.03.008 ´ ˙ [22] V Krajka-Kuzniak, B Bednarczyk-Cwynar, M Narozna, H Szaefer, W BaerDubowska, Morpholide derivative of the novel oleanolic oxime and succinic acid conjugate diminish the expression and activity of NF-κ B and STATs in human hepatocellular carcinoma cells, Chem Biol Interact (2019) 311, doi:10 1016/j.cbi.2019.108786 [23] K Krukiewicz, M Cichy, P Ruszkowski, R Turczyn, T Jarosz, J.K Zak, M Lapkowski, B Bednarczyk-Cwynar, Betulin-loaded PEDOT films for regional chemotherapy, Mater Sci Eng C 73 (2017) 611–615, doi:10.1016/j.msec.2016 12.115 [24] M.K.A Paszel, B Rubis´ , B Bednarczyk-Cwynar, L Zaprutko, M.R.J Hofmann, Oleanolic acid derivative metyl 3,11-dioxoolean-12-en28-olate targets multidrug resistance related to ABCB1, Pharm Rep (2011) 1500–1517 [25] G.M.J Barca, C Bertoni, L Carrington, D Datta, N De Silva, J.E Deustua, D.G Fedorov, J.R Gour, A.O Gunina, E Guidez, T Harville, S Irle, J Ivanic, K Kowalski, S.S Leang, H Li, W Li, J.J Lutz, I Magoulas, J Mato, V Mironov, H Nakata, B.Q Pham, P Piecuch, D Poole, S.R Pruitt, A.P Rendell, L.B Roskop, K Ruedenberg, T Sattasathuchana, M.W Schmidt, J Shen, L Slipchenko, M Sosonkina, V Sundriyal, A Tiwari, J.L Galvez Vallejo, B Westheimer, M Włoch, P Xu, F Zahariev, M.S Gordon, Recent developments in the general atomic and molecular electronic structure system, J Chem Phys (2020), doi:10.1063/5.0 05188 [26] N.M O’Boyle, M Banck, C.A James, C Morley, T Vandermeersch, G.R Hutchison, Open Babel: an open chemical toolbox - 1758-2946-3-33.pdf, J Cheminform (2011) 1–14 [27] P Gramatica, N Chirico, E Papa, S Cassani, S Kovarich, QSARINS: a new software for the development, analysis, and validation of QSAR MLR models, J Comput Chem 34 (2013) 2121–2132, doi:10.1002/jcc.23361 [28] P Gramatica, S Cassani, N Chirico, QSARINS-chem: insubria datasets and new QSAR/QSPR models for environmental pollutants in QSARINS, J Comput Chem 35 (2014) 1036–1044, doi:10.1002/jcc.23576 [29] A Daina, O Michielin, V Zoete, ILOGP: a simple, robust, and efficient description of n-octanol/water partition coefficient for drug design using the GB/SA approach, J Chem Inf Model (2014), doi:10.1021/ci500467k [30] C.A Lipinski, F Lombardo, B.W Dominy, P.J Feeney, Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings, Adv Drug Deliv Rev 46 (2001) 3–26, doi:10.1016/ s0169-409x(0 0)0 0129-0 [31] A Daina, O Michielin, V Zoete, SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules, Sci Rep (2017), doi:10.1038/srep42717 [32] A Daina, V Zoete, A BOILED-egg to predict gastrointestinal absorption and brain penetration of small molecules, ChemMedChem 11 (2016) 1117–1121, doi:10.10 02/cmdc.20160 0182 [33] E Choong, M Dobrinas, P.-.A Carrupt, C.B Eap, The permeability Pglycoprotein: a focus on enantioselectivity and brain distribution, Expert Opin Drug Metab Toxicol (2010) 953–965, doi:10.1517/17425251003789394 [34] P Berton, M.K Mishra, H Choudhary, A.S Myerson, R.D Rogers, Solubility studies of cyclosporine using ionic liquids, ACS Omega (2019) 7938–7943, doi:10.1021/acsomega.9b00603 ´ H Kapica, P Baranowski, W Saw[35] K Ciura, J Fedorowicz, P Žuvela, M Lovric, icki, M.W Wong, J Saczewski, ˛ Affinity of antifungal isoxazolo [3,4-b]pyridine3(1H)-Ones to phospholipids in immobilized artificial membrane (IAM) chromatography, Molecules 25 (2020) 4835, doi:10.3390/molecules25204835 [36] D Casoni, A Kot-Wasik, J Namies´ nik, C Sârbu, Lipophilicity data for some preservatives estimated by reversed-phase liquid chromatography and different computation methods, J Chromatogr A (2009), doi:10.1016/j.chroma.2009 01.029 None CRediT authorship contribution statement Monika Pastewska: Investigation, Writing – original draft, Conceptualization Barbara Bednarczyk-Cwynar: Resources, Conceptualization Strahinja Kovacˇ evic´ : Software, Visualization, Writing – original draft Natalia Buławska: Investigation, Software, Conceptualization, Formal analysis Szymon Ulenberg: Software, Writing – review & editing Paweł Georgiev: Investigation Hanna Kapica: Investigation Piotr Kawczak: Software Tomasz Baczek: Writing – ˛ review & editing Wiesław Sawicki: Funding acquisition Krzesimir Ciura: Conceptualization, Writing – original draft, Methodology, Supervision, Project administration, Software, Formal analysis, Investigation Acknowledgement This research was funded by the Ministry of Science and Higher Education by means of ST3 02-0 03/07/518 statutory funds The computing part was supported by the computational cluster of Copernicus Computing rendering farm, located in Włocławska 161, ´ Poland We also thank Prof Paola Gramatica for free 87-100 Torun, academic licenses for the use of QSARINS software Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.chroma.2021.462552 References [1] E.H Di, L Kerns, Drug-Like Properties, Acad Press New York, NY, USA, 2008 ´ P Žuvela, K.E Greber, P Baranowski, [2] K Ciura, J Fedorowicz, F Andric, P Kawczak, J Nowakowska, T Baỗzek, J Saỗzewski, Lipophilicity determination of antifungal isoxazolo [3,4-b]pyridin-3(1h)-ones and their n1-substituted derivatives with chromatographic and computational methods, Molecules (2019), doi:10.3390/molecules24234311 [3] M.J Waring, Lipophilicity in drug discovery, Expert Opin Drug Discov (2010) 235–248, doi:10.1517/17460441003605098 ´ [4] D Kempinska, T Chmiel, A Kot-Wasik, A Mróz, Z Mazerska, J Namies´ nik, State of the art and prospects of methods for determination of lipophilicity of chemical compounds, TrAC Trends Anal Chem 113 (2019) 54–73, doi:10 1016/j.trac.2019.01.011 ´ K.E Greber, A Gurgielewicz, W Sawicki, [5] K Ciura, J Fedorowicz, F Andric, J Saczewski, ˛ Lipophilicity determination of quaternary (fluoro)quinolones by chromatographic and theoretical approaches, Int J Mol Sci 20 (2019) 1–15, doi:10.3390/ijms20215288 ´ cˇ ic, ´ M.K Banjac, N Miloševic, ´ J Cur ´ D Marjanovic, ´ N Todor[6] S Kovacˇ evic, ´ J Krmar, S Podunavac-Kuzmanovic, ´ N Banjac, G Ušcumli ´ ´ Comparaovic, c, tive chemometric and quantitative structure-retention relationship analysis of anisotropic lipophilicity of 1-arylsuccinimide derivatives determined in highperformance thin-layer chromatography system with aprotic solvents, J Chromatogr A (2020), doi:10.1016/j.chroma.2020.461439 [7] A Loboda, E Rojczyk-Golebiewska, B Bednarczyk-Cwynar, L Zaprutko, A Jozkowicz, J Dulak, Targeting Nrf2-mediated gene transcription by triterpenoids and their derivatives, Biomol Ther 20 (2012) 499–505, doi:10.4062/ biomolther.2012.20.6.499 [9] S Singh, G.B Singh, S Bani, B.D Gupta, S.K Banerjee, Anti-inflammatory activity of oleanolic acid in rats and mice, J Pharm Pharmacol 44 (1992) 456–45 [10] S Begum, I Sultana, B.S Siddiqui, F Shaheen, A.H Gilani, Structure and spasmolytic activity of eucalyptanoic acid from Eucalyptus camaldulensis var obtusa and synthesis of its active derivative from oleanolic acid, J Nat Prod 65 (2002) 1939–1 [11] J Liu, Q Wu, Y.-F Lu, J Pi, New insights into generalized hepatoprotective effects of oleanolic acid: key roles of metallothionein and Nrf2 induction, Biochem Pharmacol 76 (2008) 922–992 [12] C.M Mengoni, F Lichtner, M Battinelli, L Marzi, M Mastroianni, G Vullo, V Mazzanti, In vitro anti-HIV activity of oleanolic acid on infected human mononuclear cells, Planta Med 68 (2002) 111–114 [13] J.-E Yuk, M.Y Lee, O.K Kwon, X.F Cai, H.Y Jang, S.R Oh, H.K Lee, K.S Ahn, Effects of astilbic acid on airway hyperresponsiveness and inflammation in a mouse model of allergic asthma, Int Immunopharmacol 11 (2011) 266–273 [14] C.D Liu, J Liu, Y Mao, Q Klaassen, The effects of 10 triterpenoid compounds on experimental liver injury in mice, Fundam Appl Toxicol 22 (1994) 34–40 11 M Pastewska, B Bednarczyk-Cwynar, S Kovacˇ evi´c et al Journal of Chromatography A 1656 (2021) 462552 [37] Giaginis Constantinos, Kakoulidou Anna Tsantili, Current State of the Art in HPLC Methodology for Lipophilicity Assessment of Basic Drugs A review, Journal of Liquid Chromatography & Related Technologies 31 (1) (2007) 79–96, doi:10.1080/10826070701665626 [38] S Teague, K Valko, How to identify and eliminate compounds with a risk of high clinical dose during the early phase of lead optimisation in drug discovery, Eur J Pharm Sci 110 (2017) 37–50, doi:10.1016/j.ejps.2017.02.017 [39] K.L Valko, Application of biomimetic HPLC to estimate in vivo behavior of early drug discovery compounds, Futur Drug Discov (2019) FDD11, doi:10 4155/fdd- 2019- 0 04 [40] A Taillardat-Bertschinger, P.-.A Carrupt, F Barbato, B Testa, Immobilized artificial membrane HPLC in drug research, J Med Chem 46 (2003) 655–665, doi:10.1021/jm020265j [41] A Taillardat-Ertschinger, A Galland, P.-.A Carrupt, B Testa, Immobilized artificial membrane liquid chromatography: proposed guidelines for technical optimization of retention measurements, J Chromatogr A 953 (2002) 39–53, doi:10.1016/s0 021-9673(02)0 0119-x [42] A Taillardat-Bertschinger, C.A.M Martinet, P.-.A Carrupt, M Reist, G Caron, R Fruttero, B Testa, Molecular factors influencing retention on immobilized artifical membranes (IAM) compared to partitioning in liposomes and n-octanol, Pharm Res 19 (2002) 729–737, doi:10.1023/a:1016156927420 [43] M Chrysanthakopoulos, F Tsopelas, A Tsantili-Kakoulidou, Biomimetic chromatography: a useful tool in the drug discovery process, Adv Chromatogr 51 (2014) 91–125 [44] F Tsopelas, T Vallianatou, A Tsantili-Kakoulidou, Advances in immobilized artificial membrane (IAM) chromatography for novel drug discovery, Expert Opin Drug Discov 11 (2016) 473–488, doi:10.1517/17460441.2016.1160886 [45] L Grumetto, C Carpentiero, P Di Vaio, F Frecentese, F Barbato, Lipophilic and polar interaction forces between acidic drugs and membrane phospholipids encoded in IAM-HPLC indexes: their role in membrane partition and relationships with BBB permeation data, J Pharm Biomed Anal 75 (2013) 165–172, doi:10.1016/j.jpba.2012.11.034 [46] L Grumetto, G Russo, F Barbato, Immobilized artificial membrane HPLC derived parameters vs PAMPA-BBB data in estimating in situ measured bloodbrain barrier permeation of drugs, Mol Pharm 13 (2016) 2808–2816, doi:10 1021/acs.molpharmaceut.6b00397 [47] G Russo, L Grumetto, F Barbato, G Vistoli, A Pedretti, Prediction and mechanism elucidation of analyte retention on phospholipid stationary phases (IAMHPLC) by in silico calculated physico-chemical descriptors, Eur J Pharm Sci 99 (2017) 173–184, doi:10.1016/j.ejps.2016.11.026 [48] G Russo, L Grumetto, R Szucs, F Barbato, F Lynen, Determination of in vitro and in silico indexes for the modeling of blood-brain barrier partitioning of drugs via micellar and immobilized artificial membrane liquid chromatography, J Med Chem 60 (2017) 3739–3754, doi:10.1021/acs.jmedchem.6b01811 [49] A Avdeef, K.J Box, J.E Comer, C Hibbert, K.Y Tam, pH-metric logP 10 Determination of liposomal membrane-water partition coefficients of ionizable drugs, Pharm Res 15 (1998) 209–215, doi:10.1023/a:1011954332221 [50] L Grumetto, C Carpentiero, F Barbato, Lipophilic and electrostatic forces encoded in IAM-HPLC indexes of basic drugs: their role in membrane partition and their relationships with BBB passage data, Eur J Pharm Sci 45 (2012) 685–692, doi:10.1016/j.ejps.2012.01.008 [51] L Grumetto, C Carpentiero, P Di Vaio, F Frecentese, F Barbato, Lipophilic and polar interaction forces between acidic drugs and membrane phospholipids encoded in IAM-HPLC indexes: their role in membrane partition and relationships with BBB permeation data, J Pharm Biomed Anal 75 (2013) 165–172, doi:10.1016/j.jpba.2012.11.034 [52] K Ciura, M Belka, P Kawczak, T Baczek, ˛ J Nowakowska, The comparative study of micellar TLC and RP-TLC as potential tools for lipophilicity assessment based on QSRR approach, J Pharm Biomed Anal 149 (2018) 70–79, doi:10.1016/j.jpba.2017.10.034 [53] K Héberger, Sum of ranking differences compares methods or models fairly, TrAC - Trends Anal Chem 29 (2010) 101–109, doi:10.1016/j.trac.20 09.09.0 09 [54] K Héberger, K Kollár-Hunek, Sum of ranking differences for method discrimination and its validation: comparison of ranks with random numbers, J Chemom 25 (2011) 151–158, doi:10.1002/cem.1320 ´ K Héberger, Chromatographic and computational assessment of [55] F Andric, lipophilicity using sum of ranking differences and generalized pair-correlation, J Chromatogr A 1380 (2015) 130–138, doi:10.1016/j.chroma.2014.12.073 ´ D Bajusz, A Rácz, S Šegan, K Héberger, Multivariate assessment [56] F Andric, of lipophilicity scales—computational and reversed phase thin-layer chromatographic indices, J Pharm Biomed Anal 127 (2016) 81–93, doi:10.1016/j.jpba 2016.04.001 ´ K Héberger, Towards better understanding of lipophilicity: assess[57] F Andric, ment of in silico and chromatographic logP measures for pharmaceutically important compounds by nonparametric rankings, J Pharm Biomed Anal 115 (2015) 183–191, doi:10.1016/j.jpba.2015.07.006 [58] R Kaliszan, H Foks, The relationship between the RM values and the connectivity indices for pyrazine carbothioamide derivatives, Chromatographia 10 (1977) 346–349, doi:10.1007/BF02274482 ´ M Natic, ´ Z Džambaski, R Markovic, ´ D Milojkovic-Opsenica, ´ [59] D Dabic, ´ Quantitative structure–retention relationship of new N-substituted 2Ž Tešic, alkylidene-4-oxothiazolidines, J Sep Sci 34 (2011) 2397–2404, doi:10.1002/ jssc.201100266 [8] Bednarczyk-Cwynar B, Ruszkowski P, Bobkiewicz-Kozlowska T, Zaprutko L Oleanolic Acid A-lactams Inhibit the Growth of HeLa, KB, MCF-7 and HepG2 Cancer Cell Lines at Micromolar Concentrations Anticancer Agents Med Chem 2016;16(5):579-92 doi: 10.2174/1871520615666150907095756 PMID: 26343139 12 ... et al Journal of Chromatography A 1656 (2021) 462552 Fig The ranking of normalized lipophilicity measures of oleanane triterpenoids by sum of ranking differences and comparison of ranks by random... likelihood of random character is assigned to iLOGP descriptor (between 1.51 and 1.73) The clustering in the form of double dendrogram in clustered heat maps of the lipophilicity measures of oleanane triterpenoids. .. the best lipophilicity measure of oleanane triterpenoids derivatives or “real” consensus logP descriptor It is followed by ALOGP and XLOGP3 as quite suitable lipophilicity descriptors of the analyzed