Using Topomer Comparative Molecular Field Analysis to Elucidate Activity Differences of Aminomethylenethiophene Derivatives as Lysyl Oxidase Inhibitors: Implications for Rational Design of Antimetastatic Agents for Cancer Therapy

Department of Electronic and Information Engineering, Changsha Social Work College, Changsha 410004, Hunan, China College of Mining Engineering, Taiyuan University of Technology, Taiyuan, Shanxi, China Western Australia School of Mines: Minerals, Energy and Chemical Engineering, Curtin University, Kalgoorlie, WA 6430, Australia College of Chemistry and Chemical Engineering, Hunan Institute of Engineering, Xiangtan 411104, Hunan, China

LOX and LOXL2 play important roles in stimulating tumor growth in various kinds of cancer [4][5][6]. Noticeably, LOX has been proved to be a significant mediator of cancer metastasis [7]. us, therapeutic agents inhibiting the activity of LOX are suggested as cancer treatments, particularly against metastasis where successful therapeutic methods are currently rare.
To date, few drug-like inhibitors of LOX have been reported [8,9]. It was published that trifluoromethyl-substituted aminomethylene-pyridine exhibited weak inhibition against LOX [8]. Until recently, aminomethylenethiophene (AMT) derivatives were proved to be potent inhibitors of LOX and, more importantly, can significantly reduce tumor growth as well as metastatic burden in a mouse model [9,10].       Journal of Chemistry e preliminary structure-activity relationship study of AMT derivatives has been performed, which was dictated by chemical intuition and serendipity [9,[11][12][13]. Considering the fact that we are entering the era of big data and artificial intelligence, empirical and labor-intensive approaches are being incrementally replaced by automatic and rational strategies [14][15][16][17]. Consistent with this emerging trend, sophisticated drug design method (i.e., topomer CoMFA) in this paper is introduced into the three-dimensional quantitative structure-activity relationship (3D-QSAR) study of aminomethylenethiophene derivatives, which holds great promise for rational design of LOX inhibitors with improved potency.

Substituted Aminomethylenethiophene.
e aminomethylenethiophene (AMT) derivatives and their anti-LOX potency are taken from the recently published literature [9]. e LOX inhibition values of AMT derivatives, that is, LOX IC 50 (μΜ), were obtained under identical experimental conditions [9]. e molecular structures of AMT derivatives and their anti-LOX potency are shown in Table 1. irty-six of the forty-five aminomethylenethiophene-based inhibitors were chosen as the training set, while the remaining nine molecules were used as the test set. During the selection of the training and test set, great efforts were made to ensure that the most potent, moderately active, and low active AMT inhibitors were included in the training set to spread the activity range. e test set compounds were selected in such a manner that at least one structural analog of the training set was chosen for the test set. Each AMT derivative was geometrically optimized by the Tripos force field under the convergence criterion of 0.005 kcal/mol [15,16]. e MMFF94 method was utilized to determine the partial atomic charges of AMT derivatives [15,18].

Topomer CoMFA.
e molecular modelling of 3D-QSAR was carried out using SYBYL 8.1.1 software running on a HP Z600 workstation. For the topomer CoMFA method, the aim is to build a quantitative relationship between molecular field information of topomers (i.e., steric and electrostatic fields) and LOX inhibition activity [19][20][21]. A topomer indicates an invariant three-dimensional representation of the molecular fragment with absolute orientation and conformation [21]. It is well known that comparative molecular field analysis (CoMFA) is very sensitive to the molecular alignment scheme by which the training set compounds are superimposed over the maximum common substructure [22]. To circumvent this problem, topomer CoMFA is invented as an alignment-free 3D-QSAR technique [19,20].
In this paper, topomer comparative molecular filed analysis (topomer CoMFA) is used to build the quantitative relationship between the molecular field information and the anti-LOX potency of aminomethylenethiophene derivatives.

Statistical Evaluation of the QSAR Model.
To avoid overfitted 3D-QSAR model, partial least squares (PLS) is utilized to correlate the molecular filed descriptors with the anti-LOX potency of AMT derivatives [23]. e predictive capability of the QSAR model is measured by the cross-validated coefficient (q 2 ), which is determined by the leave-one-out cross-validation procedure and can be formulated as where Y pred , Y actual , and Y mean are the theoretically predicted, experimentally measured, and mean LOX inhibition activities of the training set, respectively.

Molecular Fragmentation.
e structural variation of AMT derivatives lies in the R substitute, while aminomethylenethiophene is the common scaffold. us, all the substituted aminomethylenethiophenes are fragmented into two parts: (1) the common scaffold and (2) R substituent. e fragmentation pattern is shown in Figure 1.

Statistical Analysis of the QSAR Model.
e statistical results of the topomer CoMFA model are presented in Table 2, which shows the non-cross-validation correlation coefficient (r 2 ) of 0.912 and cross-validation correlation coefficient (q 2 ) of 0.540. It is recognized that a 3D-QSAR model would have decent predictive ability provided that q 2 is greater than 0.5 and r 2 is more than 0.6 [19,24,25]. As such, the resultant 3D-QSAR model in this paper should be statistically robust. In terms of the topomer CoMFA model, the predicted anti-LOX potency of the AMT derivatives is listed in Table 1. e experimentally measured LOX inhibition activities versus the theoretically predicted values are graphically shown in Figure 2, and the data points reside closely to the diagonal line. It is known that if the leave-one-out correlation coefficient of a 3D-QSAR model is above 0.3, the likelihood of chance correlation could be less than 5% [26]. Furthermore, the correlation coefficient (R 2 ) for the test set is 0.6637, which indicates a decent predictive power of the otpomer 3D-QSAR model. erefore, the built topomer CoMFA model in our paper should be statistically significant and has sound predictive ability, which could be used to predict and screen the untested AMT derivatives [19,20,25,27].

Topomer CoMFA Contour Analysis.
Given that the R substitute is the main factor governing the anti-LOX potency of AMT derivatives, the molecular field contour map of the R2 fragment is presented in Figure 3. Compound 22 with the highest LOX inhibition was employed as the reference molecule. Actually, Figure 3 provides hints about the possible regions for the structural modification of R2 fragments (i.e., the blue area favors the electron-donating substituent and the green area for steric groups, while the red area is Journal of Chemistry favourable for electron-withdrawing substituent and the yellow area for the less steric groups).
e steric contour map of the R2 fragment is shown in Figure 3(a). In Figure 3(a), large polyhedra covering the 4position of phenyl signify that steric hindrance at this location may be beneficial for improving the anti-LOX potency of AMT derivatives, which is convincingly confirmed by comparing the LOX inhibition activity of compound 15 with compound 22. Due to the sole introduction of methanesulfonyl into the 4-position of phenyl, the anti-LOX potency of compound 22 (IC 50 : 0.26) is significantly improved as compared to compound 15 (IC 50 : 1.7). e yellow polyhedra revolving around the 3-position of phenyl indicate that sterically bulky groups at this location may exert negative impact on the LOX inhibition of AMT derivatives, which is unambiguously validated by the comparison of compound 15 and compound 18; the LOX inhibition activity of the latter molecule (IC 50 : 6.4) is much lower than the former compound (IC 50 : 1.7) because of the addition of benzene at the 3-position of phenyl. e electrostatic contour plots of the R2 fragment are displayed in Figure 3(b). In Figure 3(b), a red contour encapsulating the end of methanesulfonyl moiety implies that electron-donating groups at this end disfavor the LOX inhibition, which is clearly verified through comparing the structures and anti-LOX potency of compound 28 (IC 50 : 1.8) and compounds 29-32 (IC 50 : 0.29∼0.72); compound 28 has a benzene substituent (i.e., electron-donating group) at the end of methanesulfonyl moiety, while compounds 29-32 have oxygen or nitrogencontaining groups (i.e., electron-withdrawing groups).

Newly Designed Compound and Predicted Anti-LOX Potency
Since the above 3D-QSAR model is statistically significant, it would be of utmost importance to theoretically design new molecule with enhanced LOX inhibition. e scaffold of aminomethylenethiophene was modified by various substituents, and the anti-LOX potency of the resulting AMT derivatives was predicted in terms of the already established 3D-QSAR model. Two of the newly designed AMT derivatives with the theoretically decent LOX inhibition are presented in Table 3.
ough the theoretically predicted anti-LOX potency of the two novel compounds is slightly lower as compared to compound 22, it should be noted that    great caution is necessary in using QSAR to make predictions that extrapolate beyond the area of knowledge of the training set [28,29]. In addition, these two newly proposed AMT derivatives should be synthetised to experimentally determine their LOX inhibition activity.

Conclusion
In this work, the topomer CoMFA method is applied to the quantitative structure-activity relationship study of aminomethylenethiophene (AMT) derivatives as novel lysyl oxidase (LOX) inhibitors. e resulting topomer CoMFA model is effective for estimating anti-LOX potency of the AMT derivatives. Molecular field information has illustrated the critical structural factors governing the LOX inhibition activity of substituted aminomethylenethiophene. In terms of the developed topomer CoMFA model, the theoretically predicted anti-LOX potency of AMT derivatives matched well with the experimentally measured values, indicating the reasonable predictive ability of the QSAR model. e newly designed AMT derivatives with theoretically decent anti-LOX potency have been proposed. is study may serve as the stepping stone for the rational design/selection of LOX inhibitors as antimetastatic agents for cancer therapy.

Data Availability
All the data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare no conflicts of interest. Yellow contours imply that less steric hindrance is favourable, while green polyhedra indicate that the sterically bulky group is beneficial. Blue contours signify that the electron-donating groups are favored, while red contours reveal that the electron-withdrawing substituents are preferred.