SUPPORT RELATED DIFFERENTIAL IMPACT OF SUBSTITUENTS ON PERFORMANCE OF (ALKOXY-PHENYL)BENZAMIDES IN NORMAL PHASE TLC

Abstract

The performance of 14 (alkoxy-phenyl)benzamide derivatives was studied on silica and alumina surfaces by normal phase TLC. The impact of substituent of benzanilide moiety on retention of analytes was determined by Free Wilson Analysis. It was found that the introduction of the methyl group in ortho position on benzyl submoiety, as well as alkoxy group in ortho position on anilide submoiety, altered the retention of analytes by a significantly different manner on two stationary phases; meanwhile, this influence in meta position depended on the size of the alkoxy group. The substitution in para position on anilide submoiety influenced the retention on silica and alumina surfaces by a similar manner. Weighting the impact of the substituent on retention of benzanilides, the canonical correlation analysis was successfully applied for reduction of the number of variables without losing relevant information.

The relationships between measured retention parameters of analytes and computed molecular descriptors were calculated by linear regression analysis. It was established that the computed molecular descriptors in 78 of 3346 of cases correlated with measured retention parameters (P = 0.1–10%). This finding indicates that further developments are requested for the use of in silico descriptors in Quantitative Structure Retention Relationship (QSRR) studies to predict expected physicochemical properties of this type of molecules.

Keywords:

• adsorption
• benzanilide
• canonic correlation
• QSSR
• reduce variables
• thin layer chromatography

ACKNOWLEDGMENT

This study was supported by the Hungarian Scientific Research Fund and the National Office for Research and Technology.

Notes

^a Descriptors computed with the following free programs, sets, and web engines ACD, Molinspiration, Actelion, and Chemaxon, D = Dragon did not significantly correlated with observed parameters.

^b Probablity of relationship between experimental values and “in silico” molecular descriptor.

^c Total number of descriptors offered by engine.

^d Number of descriptors that fits to the structure of analytes.

FG = 14, .

The abbreviations are as followes: Si = silicagel, A = aluminiumoxide, AO- = alkoxy group, Cn- = number of carbon atoms in alkoxy-groups. The partial regression coefficients (R) were calculated by Multiple Regression Analysis (r_0.05 = 0.497) using Statictica 5 module. The L_Si-R and L_Al-R are canonic variables of the first canonic equation calculated of two sets of variables (Table 1 and Figure 3), intercepts and slopes of the set of benzanilides determined on two supports. The coefficients b_Si and b_Al were used for prediction of expected values plotted in Figure 3/II applying the general formula: R_M0 = b*[R₁CH₃] + b*[R₂AO] + b*[R₂Cn] + b*[R₃AO] + b*[R₃Cn] + b*[R₄AO] + b*[R₄Cn].

In the table are shown those cases, where the correlation between observed parameters and in silico molecular descriptors proved to be significant at minimum p_0.1 level.

^a FG = 14, r_0.001 =  0.742 , r_0.01 =  0.623 , r_0.05 = 0.479, r_0.1 = 0.426, and their squared values are 0.551 , 0.388, 0.247, and 0.181, respectively. a_Si and a_Al are the intercepts, while b_Si and b_Al the slopes, respectively (see Table 3). The L1_Si, R1_Al and L2_Si, R2_Al are canonic variables of left and second sets of chromatographic parameters by first and second roots of canonic equation.

^b The explanation of descriptors is given in Table 1. Bold labeled names are programs that were used computing molecular descriptors for in silico QSAR calculations.

^c Number of descriptors computed by modules of programs.

^d Number of descriptors fits to structures of molecules investigated in this study.

^e The number of cases where the given observed of derived retention parameters correlate with molecular descriptors.