Skip to main content

QSAR, molecular docking studies, ligand-based design and pharmacokinetic analysis on Maternal Embryonic Leucine Zipper Kinase (MELK) inhibitors as potential anti-triple-negative breast cancer (MDA-MB-231 cell line) drug compounds

Abstract

Background

Cancer of the breast is known to be among the top spreading diseases on the globe. Triple-negative breast cancer is painstaking the most destructive type of mammary tumor because it spreads faster to other parts of the body, with high chances of early relapse and mortality. This research would aim at utilizing computational methods like quantitative structure–activity relationship (QSAR), performing molecular docking studies and again to further design new effective molecules using the QSAR model parameters and to analyze the pharmacokinetics “drug-likeliness” properties of the new compounds before they could proceed to pre-clinical trials.

Results

The QSAR model of the derivatives was highly robust as it also conforms to the least minimum requirement for QSAR model from the statistical assessments of (R2) = 0.6715, (R2adj) = 0.61920, (Q2) = 0.5460 and (R2pred) of 0.5304, and the model parameters (AATS6i and VR1_Dze) were used in designing new derivative compounds with higher potency. The molecular docking studies between the derivative compounds and Maternal Embryonic Leucine Zipper Kinase (MELK) protein target revealed that ligand 2, 9 and 17 had the highest binding affinities of − 9.3, − 9.3 and − 8.9 kcal/mol which was found to be higher than the standard drug adriamycin with − 7.8 kcal/mol. The pharmacokinetics analysis carried out on the newly designed compounds revealed that all the compounds passed the drug-likeness test and also the Lipinski rule of five.

Conclusions

The results obtained from the QSAR mathematical model of parthenolide derivatives were used in designing new derivatives compounds that were more effective and potent. The molecular docking result of parthenolide derivatives showed that compounds 2, 9 and 17 had higher docking scores than the standard drug adriamycin. The compounds would serve as the most promising inhibitors (MELK). Furthermore, the pharmacokinetics analysis carried out on the newly designed compounds revealed that all the compounds passed the drug-likeness test (ADME and other physicochemical properties) and they also adhered to the Lipinski rule of five. This gives a great breakthrough in medicine in finding the cure to triple-negative breast cancer (MBA-MD-231 cell line).

Background

Cancer has long been a challenging malady to the human race with great occurrence and death rates. Nearly 8 million humans pass away from tumors yearly, and 14.2 million different cancer patients are treated worldwide yearly (Xu et al. 2019). Cancer of the breast is the most reoccurring type of disease and the second leading cause of death among women globally. Yearly, approximately 1–1.3 million cases of breast cancer are detected globally (Ge et al. 2019). The high occurrence and death rate of cancer occur from the fact that there are over 200 kinds of cancer and it is indeed hard to diagnose at a premature stage (Lolak et al. 2019).

Triple-negative breast cancer (TNBC) is painstaking the most destructive type of mammary tumor because it spreads faster to other parts of the body, with high chances of early relapse and mortality (Hu et al. 2012). Yearly, an estimation of over a million female beings is detected with mammary tumor and TNBC is responsible for almost 15–20% of the overall breast cancer detected (Jo et al. 2019). TNBC does not express estrogen receptor (ER), progesterone receptor and human epidermal growth factor 2 (HER2) (Hu et al. 2012).

Early detection, understanding of the cause and pathway of this disease and advancing in treatment have played a key role in decreasing breast cancer death rates during the past few years. Chemotherapy remains the main key to complete therapy since it would extend and terminate tumor cells faster within the human system (Kaplan 2013). Computational methods of drug discovery have proven to be faster than traditional methods of drug discovery, and it saves more time, resources and also tends to have higher efficacy with less toxicity. Efficiency and safety of the drug to the system are the two major causes leading to drug failure. Therefore, it is compulsory to find potent molecules with better ADMET properties “drug-likeliness” (Guan et al. 2019).

Novel 61 series of parthenolide derivatives were obtained from the literature of (Ge et al. 2019) as inhibitors against the MDA-MB-231 cell line. Edupuganti et al. 2017 suggest that MELK with Protein Data Bank (PDB: 4BKY) has therapeutic value for targeting various cancers, especially the critical survival functions like TNBCs.

The purpose of this research is to utilize computational methods like quantitative structure–activity relationship (QSAR), in calculating the activities of parthenolide derivative compounds as inhibitors against breast cancer cell line MDA-MB-231 based on an established QSAR mathematical model, molecular docking studies in understanding how the ligands (parthenolide compounds) interrelate with a protein receptor Maternal Embryonic Leucine Zipper Kinase (MELK), again, to further design new effective molecules using the QSAR model parameters and to analyze the pharmacokinetics “drug-likeliness” properties of the new compounds.

Methods

QSAR studies

Data set

In total, 61 derivative compounds of parthenolide were obtained from (Ge et al. 2019) article.

Anti-proliferative activities

Their bioactivities were changed to (pIC50) using the formula below. It was measured in inhibitory concentration (IC50) values in the micromolar concentration of (µM) (Abdulrahman et al. 2020a, b, c). The parent compounds as shown in Fig. 1a–d are the structural derivatives of parthenolide. Tables 1, 2, 3, 4 and 5 show the various substituents measured in (IC50) and (pIC50) that were attached to the parent compound.

$${\text{pIC}}_{50} = - \log 10 \, \left( {{\text{IC}}_{50} \times 10^{ - 6} } \right).$$
Fig. 1
figure1

a Parthenolide template molecule for Table 1. b Parthenolide template molecule 2 for Table 2. c Parthenolide template molecule 3 for Table 3. d Parthenolide template molecule 4 for Table 4

Table 1 The inhibitory concentration of parthenolide derivatives for template molecule 1
Table 2 The inhibitory activities of parthenolide derivatives for template molecule 2
Table 3 The inhibitory activity of parthenolide derivatives for template molecule 3
Table 4 The inhibitory activity of parthenolide derivatives for template molecule 4
Table 5 The inhibitory activities of parthenolide derivatives
Table 6 Suggested values used in the assessment of QSAR model
Table 7 Model parameters used in validating the model externally
Table 8 Continuation of the external validation of the model
Table 9 The bioactivities (pIC50), predicted (pIC50) and residual of the model
Table 10 Definition of descriptors and their classes for the model
Table 11 Statistical analysis of the descriptor used in the QSAR model
Table 12 Binding affinity, interaction type, bond type and distances in some complexes
Table 13 Newly designed parthenolide derivative compounds with their new predicted activities (pIC50)
Table 14 Physicochemical and ADME properties (pharmacokinetics) of designed parthenolide compounds against the MDA-MB-231 cell line

Structural optimization

The data set was drawn on software (V12.0.2) of ChemDraw and was transferred to Spartan 14 software (V1.1.4) for structural optimization, setting up the parameters density functional theory (DFT) at B3LYP, 6-31G/ basis set (Abdulrahman et al. 2020a, b, c).

Molecular descriptor calculations

In total, 61 derivative compounds of parthenolide were converted to SDF format after optimization. Pharmaceutical Data Exploration Laboratory Software V (2.20) was used in calculating physicochemical descriptors (Yap et al. 2011). The descriptors were pretreated using Data Pre-treatment software GUI 1.2. (Arthur et al. 2020; Abdulrahman et al. 2020a, b, c) to remove irrelevant values.

Training set and test set division

For a valid QSAR model to be obtained, the pretreated set must be divided into train and test set. The division should satisfy the following conditions: (i) it is essential for all the train set descriptive points to be evenly shared within the entire area used by the data set (Noolvi and Patel 2013). (ii) All descriptor values should be near the train set descriptor values. (iii) All characteristic molecular elements in the test set should spread within the descriptor space and should also be close to the train set (Noolvi and Patel 2013). The algorithm of Kennard–Stone was employed in dividing the set into 30% test and 70% train set (Kennard and Stone 1969).

Model building

Multiple linear regression technique was used in constructing a model in Version 8 of Material studio software. It was used to show the correlation between the dependent variables (pIC50) and the independent variables as the 2D model descriptors (Abdulrahman et al. 2020a, b, c). The model is suited so that the total squared difference between the actual and predicted value of the set of anti-proliferative activity is diminished. In regression analysis, the reliant mean of the dependent variable, (pIC50), relies on the independent descriptors (Abdullahi et al. 2020).

Validating the model (internal)

The prediction of the model built must be verified on a data set that was never used in building the model at first (Tropsha et al. 2003). The models obtained from the internal validation of the train test (sixty-one compounds) were evaluated using Friedman formula (Friedman 1991).

$${\text{LOF}} = \frac{{{\text{SEE}}}}{{{\text{M}}\left[ {1 - {\upbeta }\left( {\frac{{{\text{c}} + {\text{d}} \times {\text{p}}}}{{\text{M}}}} \right)} \right]^{2} }}$$

d equals user-defined smoothing parameter, C equals the sum of model definitions, M equals the sum of derivatives on the train set, while p is the summation of model definition (Abdulrahman et al. 2020a, b, c). SEE is the standard estimated error. The smaller SEE is, the more enhanced the model would be.

$${\text{SEE}} = \sqrt {\frac{{\left( {{\text{Y}}_{{{\text{exp}}}} - {\text{Y}}_{{{\text{pred}}}} } \right)^{2} }}{{{\text{N}} - {\text{P}} - 1}}}$$

The model from the regression structure becomes.

$$Y = k_{1} x_{1} + k_{2} x_{2} + k_{3} x_{3} + n.$$

Y is the response variable, ‘k’ is the coefficient of regression corresponding to ‘x’s that are the model parameters which are the predictor variables, and n is the regression constant.

(R2) is the regression coefficient, and it is the degree of fitness of the equation of regression. It signifies the variation section in the data that is described by regression. The nearer R2 value is 0. 1, the better the regression fitness. R2 is given as:

$$R^{{2{ }}} {\text{pred}} = 1 - \left[ {\frac{{\sum \left( {Y_{{{\text{exp}}}} - Y_{{{\text{pred}}}} } \right)^{2} }}{{\sum \left( {Y_{{{\text{exp}}}} - Y_{{{\text{meantraining}}}} } \right)^{2} }}} \right]$$

where Yexp and Ypred are the biological and calculated activities of the train set. Ymintraining indicates the average pIC50 of the train set molecules. Table 2 shows the accepted least required values in assessing a QSAR model (Ibrahim et al. 2018). (R2pred) is the predicted validation parameter, and it is assessed by calculating the prediction power of the model. An (R2pred) above 0.5 shows a high prediction power and robustness of the model.

Applicability domain of QSAR

Model validation should be within the training domain, and the performance of the compounds needs to be assessed within the domain to determine the fitness of the model. The applicability domain is evaluated using William's plot. The warning leverage is used in sorting the compounds within a particular space on the graph known as the applicability domain, compounds that fall within the space on the plot are the approved predicted compounds (Veerasamy et al. 2011). It is formulated as;

$$h* = \frac{{3\left( {j + 1} \right)}}{m}$$

where j equals the total model parameters and m is the total molecules of train sets.

Molecular docking studies

Some compounds with high pIC50 were chosen for molecular docking studies with Maternal Embryonic Leucine Zipper Kinase (MELK) as protein target. The crystal structure was obtained from RCSB PDB (https://www.rcsb.org) with the ID, 4BKY. The crystal structure was prepared with Discovery Studio Software and converted to PBD format. The binding affinity of the ligand–protein complex was calculated with AutoDock Vina of PyRX software employed in calculating (Abdulrahman et al. 2020a, b, c). Visualizer of Discovery Studio was used to understand the ligand–protein target interactions.

Computational pharmacokinetics (drug-likeness)

The SwissADME, a free web tool used in evaluating the pharmacokinetics, drug-likeness (physicochemical and ADME properties) and medicinal chemistry friendliness of small molecules (Daina et al. 2017), would be used in testing the drug-likeness of the newly designed compounds. Furthermore, some physicochemical properties and positive controls of the designed compounds were checked using the on-line tool for their adaptability with Lipinski's rule of five (Hou et al. 2019). Lipinski and co-workers proposed the "Rule of Five" in 1997, which was the original and most known rule-based filter for drug-likeness of a molecule, distinguishing whether a molecule can be orally absorbed well or not, following the criteria: molecular weight (MW) ≤ 500, octanol/water partition coefficient (AlogP) ≤ 5, number of hydrogen bond donors (HBD's) ≤ 5 and number of hydrogen bond acceptors (HBAs) ≤ 10.6. According to the Rule of Five, a molecule would not be orally active if it violates two or more of the four rules (Guan et al. 2019).

Results

QSAR of parthenolide derivatives

The goal of QSAR is to establish a model from the obtained descriptor that has a higher performance than the experimental values. In this research, parthenolide derivatives went through a quantitative structure–activity relationship with its actual activities.

The Model

$$\begin{aligned} pIC_{50} & = 0.112976697*\left( {AATS6i} \right) + 0.010871443*\left( {ATSC8i} \right) + 1.482833475*\left( {GATS5s} \right) \\ & \quad + 0.000807496*\left( {VR1\_Dze} \right) - 0.303714684*\left( {C2SP3} \right) - 12.639784038. \end{aligned}$$

The model best predicts the biological activities of parthenolide derivatives against the MDA-MB-231 cell line. The model also conforms to the least proposed requirement in QSAR modeling as indicated in Table 2.

Tables 7 and 8 reveal how the model was validated externally by carrying out calculations using the values of the descriptors obtained from the test set. The calculated external validation (R2pred) was 0.5304, which agrees with the least requirements as shown in Table 6.

Table 9 shows the experimental, calculated and residual values of the derivative compounds. The low residual values were calculated from the predicted activities. The low residual values show the potency of the built model.

Table 10 defines the model parameters (descriptors) used in building the model, and the descriptors were used in verifying the model both internally and externally. The mean effect was calculated statistically, and it shows the collinearity of the model parameters and the activities of the derivative compounds in the built model.

Table 11 shows correlation between the descriptors generated from the model (VIF) and the P values of the descriptors. The VIF was calculated using the equation;

$${\text{VIF}} = \frac{1}{{\left( {1 - X^{2} } \right)}}$$

X2 is the coefficient of correlation (Myers 1990).

The P values estimate the statistical difference between the model parameters at 95% confidence level.

A straight line graph was obtained from plotting the activities (predicted activity against experimental activity) of both the test set and train set of the derivative compounds as shown in Fig. 2. The anti-proliferative activities showed a good connection as shown on the plot.

Fig. 2
figure2

Graph of predicted activities versus inhibition concentration (anti-proliferative activities)

A graph of standardized residual versus biological activity of the entire data set was plotted as shown in Fig. 3. The compounds of both the test and train set spread on both sides of the y-axis, defining the potency of the model.

Fig. 3
figure3

Graph of standardized residual versus biological activity

The applicability domain was evaluated using the William's plot. The William’s plot is a graph of standardized residual against leverages, as shown in Fig. 4. The warning leverage is used in sorting the compounds within a particular space, and the compounds that fell within the space on the plot were the actual predicted compounds. The warning leverage was found to be (h* = 0.47).

Fig. 4
figure4

Graph of standardized residual against leverages (William’s plot)

Molecular docking results

Molecular docking interaction on compounds of parthenolide derivatives with the protein target, Maternal Embryonic Leucine Zipper Kinase (MELK), was performed. The molecular docking results are summarized in Table 12. The binding affinity was calculated using PyRX software, while Discovery Studio Software was used in understanding and visualizing the various interaction formed between the ligand (derivative compounds) and the binding pose of the receptor (MELK).

Figure 5 shows the prepared receptor and ligand that was used in the docking studies, Figs. 6 and 7 show the 2D interaction of the complexes 9 and 2, while Fig. 8 shows the 3D interaction of the prepared ligand 9 and 2 to form complexes.

Fig. 5
figure5

Prepared receptor and ligand

Fig. 6.
figure6

2D representation of complex 9

Fig. 7
figure7

2D representation of complex 2

Fig. 8
figure8

The bioavailability RADAR of molecules 12, 5 & 8

Ligand base drug design

Fifteen (15) new compounds new parthenolide derivative compounds were designed using the ligand-based approach. The ligand-based design used the derived mathematical model obtained from the QSAR studies, and adjustment was made on the lead compounds (1 and 13) based on the definition of the molecular descriptors obtained from the model. Table 13 shows the newly designed compounds with their new calculated activities.

Physicochemical and ADME properties (pharmacokinetics) of designed parthenolide compounds

A compound with poor drug-likeness and poor ADMET properties will not be allowed to progress into pre-clinical research, regardless of the high anti-proliferative activities. ADME properties are one of the main pipelines in drug discovery, for drug-likeliness of a molecules to be assessed, and it must pass through the pipeline of drug discovery. All the newly designed compounds were assessed for their drug-likeliness (pharmacokinetics analysis) as shown in Table 14.

Discussion

QSAR of parthenolide derivatives

From the statistical parameters, the squared coefficient of correlation (R2) was 0.6715, correlation coefficient adjusted squared (R2adj) was 0.61920, cross-validation coefficient (Q2) was 0.5460, and external validation (R2pred) was 0.5304. How the external validation of the model was verified is shown in Tables 7 and 8, which also conforms to the least requirement in QSAR modeling as shown in Table 6. The actual, calculated and residual values of parthenolide derivative compounds are shown in Table 9. The low residual value is the outcome between the difference between actual and predicted activities showing the high performance of the model. The model certified both internal and external parameters, thereby confirming the model to have high performance, very stable and robust.

The model's parameters are defined in Table 10 (names, definition and class). The mean effect (Table 10) shows the contribution of each descriptor in the constructed model, and the positive coefficient and values of the descriptor from the mean effect showed AAts6i has more impact followed by GATS5s, ATSC8i and VR1_Dze, therefore increasing the positive effect of the descriptors would increase the biological activities of parthenolide derivatives, while C2SP3 carries a negative effect and gives the least contribution in the model, reducing its negative effect would also contribute in increasing biological activities of parthenolide derivatives as proven in Tables 7 and 8. There was no much inter-correlation between the model parameters from the statistical analysis of Variance Inflation Factor (VIF), making the model highly stable. The null hypothesis shows no significant connection amid the bioactivity and model parameters of the constructed model at p > 0.05. At a 95% confidence level, the P values of the model parameters were below 0.05. Therefore, the null hypothesis is rejected and the alternative hypothesis is accepted. This indicates that there is a co-linearity between the bioactivity and model parameters of the constructed model, as shown in Table 11.

Figure 2 shows a plot of actual activities against the predicted activities of both the test set and the train set of compounds. The plot showed that the predicted activity was in good agreement with its experimental values as shown in Table 6, conforming to the effectiveness and strength of the built model. Figure 3 shows how the entire molecules spread on both negative and positive sides of zero point on the y-axis of the plot, indicating no systematic errors between the standardized residual versus inhibition concentration (experimental activity). Figure 4 shows the standardized residuals against the leverage values also called William's plot. The better part of the derivatives fell within the calculated leverage of (h = 0.47); except 8 compounds we found to be outside the warning leverage which might be due to a slight difference in structure compared to other compounds in the data set.

Molecular docking studies

Molecular docking interaction on compounds of parthenolide derivatives with the protein target, Maternal Embryonic Leucine Zipper Kinase (MELK), was performed. Compounds 2, 9 and 17 were chosen because of their high binging score. Compounds 2, 9 and 17 had the highest docking score of − 9.3, − 9.3 and − 8.9 kcal/mol as shown in Table 12. The visual examination of docked complexes was done by assessing the hydrogen bond interaction, hydrogen bond length and the hydrophobic interaction as shown in Table 12.

Compound 9 showed backbone conventional hydrogen bonds (2.74925, 2.4619, 2.98056, 3.34925, 3.35524 and 3.59598 A0) with THR19, GLY21, ARG53, CYS89, GLY18 and ILE17 of the protein target site. Furthermore, it formed hydrophobic bonds (5.96743, 5.00307, 5.43404, 4.42983, 3.89836, 5.15308, 4.67658 and 4.44028 A0) with PHE22, ILE17, VAL25, LEU139, ILE17, PHE22, TYR88 and ALA23 of the target site. It formed a halogen and electrostatic bond (3.33001 and 4.20988 A0) with amino acids of ASP150. The carbonyl group of the compound acts as a hydrogen acceptor and formed one hydrogen bond with THR19, while two of the fluorine (Halogens) attached to the phenyl group of the compound, act as a hydrogen donor and formed two hydrogen bonds with GLY21 and ARG53 of the protein crystal.

Compound 2 also showed four hydrogen bonding (1.91438, 3.50504, 3.40604 and 3.35007 A0) with LYS134, GLY18, ASP150 and THR19 of the active site of the receptor. The derivative also formed hydrophobic bonds (3.72576, 3.10836, 4.3497, 5.0501, 4.61294, 4.86094, 4.17368, 4.84776, 5.45664 and 4.54951 A0) with amino acids of PHE22, ALA23, ALA38, ILE17, VAL25, LEU139, VAL25, ALA38, LEU86 and ILE149 of the protein target. The carbonyl of the compound attached to 3-methylenedihydrofuran-2(3H)-one acts as a hydrogen acceptor to form one hydrogen bond with LSY134.

Compound 17 also formed hydrogen and hydrophobic bonds with the protein receptor at different bond distances as shown in Table 12. The hydrogen and hydrophobic interactions show that the ligands have high potency against the MELK receptor. The prepared receptor and ligand are shown in Fig. 5. The 3D and 2D representations of the binding pose are shown in Figs. 6 and 7, and how the ligands bind firmly to the active sites of the receptor and also a detailed interaction to form complexes.

Ligand-based drug design

Compounds 13 and 1 were chosen as lead compounds in the design of fifteen (15) new parthenolide derivative compounds due to their high predicted activity and low residual values as shown in Table 7. From the mean effect of the descriptors, AATS6i had a greater positive impact, while VR1_Dze had the least positive impact on the model. According to AATS6i (Centered Broto-Moreau autocorrelation—lag 6/weighted by first ionization potential) and VR1_Dze (Randic-like eigenvector-based index from Barysz matrix/weighted by Sanderson electronegativities) descriptors, adding more ionization potential and electronegative elements would increase the potency of the chosen templates. The structural modification occurred by adding more ionization potential and electronegative elements to template molecules (compounds 1 and 13). The predicted activities of the newly designed compounds were higher than that of the template molecules as shown in Table 13.

Physicochemical and ADME properties (pharmacokinetics) of designed parthenolide compounds

All the newly designed compounds were assessed for their drug-likeliness (ADME and physicochemical properties). None of the designed compounds violated two rules out of the Lipinski rule of five, a prominent principle used in certifying the drug-likeness of a compound, and this shows that all the designed compounds passed the drug-likeness test as shown in Table 14, making the compounds a breakthrough in finding the cure to triple-negative breast cancer. Figure 9 shows the bioavailability RADAR of molecules 2, 5 and 8. The bioavailability RADAR enables a first glance at the drug-likeness of a molecule. The pink area signifies the ideal range for each properties (lipophilicity: XLOGP3 between − 0.7 and + 5.0, size: MW between 150 and 500 g/mol, polarity: TPSA between 20 and 130 Å2, solubility: log S not higher than 6, saturation: fraction of carbons in the sp3 hybridization not less than 0.25 and flexibility: no more than 9 rotatable bonds) (Daina et al. 2017). The GI—gastrointestinal absorption was also high for all the designed molecules.

Fig. 9
figure9

3D representation of complex 9 and 2

Conclusion

The results obtained from the QSAR mathematical model of parthenolide derivatives were used in designing new derivatives compounds that were more effective and potent. The molecular docking result of parthenolide derivatives showed that compounds 2, 9 and 17 had higher docking scores than the standard drug adriamycin. The compounds would serve as the most promising inhibitors (MELK). Furthermore, the pharmacokinetics analysis carried out on the newly designed compounds revealed that all the compounds can move on to the next step of pre-clinical trial because they passed drug-likeness test (ADME and other physicochemical properties) and they also adhered to the Lipinski rule of five: a major principle used in analyzing the drug-likeness of small compounds. This gives a great breakthrough in medicine in finding the cure to triple-negative breast cancer (MBA-MD-231 cell line).

Availability of data and materials

Not applicable.

Abbreviations

ADME properties:

Absorption, distribution, metabolism and elimination properties

MELK:

Maternal Embryonic Leucine Zipper Kinase

QSAR:

Quantitative structure–activity relationship

TNBCs:

Triple-negative breast cancers

References

  1. Abdullahi M, Uzairu A, Shallangwa GA, Mamza P, Arthur DE, Ibrahim MT (2020) In-silico modelling studies on some C14-urea-tetrandrine derivatives as potent anti-cancer agents against prostate (PC3) cell line. J King Saud Univ Sci 32(1):770–779

    Article  Google Scholar 

  2. Abdulrahman HL, Uzairu A, Uba S (2020a) QSAR, ligand based design and pharmacokinetic studies of parviflorons derivatives as anti-breast cancer drug compounds against MCF-7 cell line. Chem Africa 29:1–3

    Google Scholar 

  3. Abdulrahman HL, Uzairu A, Uba S (2020b) Computer modeling of some anti-breast cancer compounds. Struct Chem 15:1–9

    Google Scholar 

  4. Abdulrahman HL, Uzairu A, Uba S (2020c) In silico studies of some 2-anilinopyrimidine derivatives as anti-triple-negative breast cancer agents. Beni-Suef Univ J Basic Appl Sci 9:1–2

    Article  Google Scholar 

  5. Arthur DE, Uzairu A, Mamza P, Abechi SE, Shallangwa G (2020) Activity and toxicity modelling of some NCI selected compounds against leukemia P388ADR cell line using genetic algorithm-multiple linear regressions. Journal of King Saud University-Science 32(1):324–331

    Article  Google Scholar 

  6. Daina A, Michielin O, Zoete V (2017) SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Sci Rep 7:42717

    ADS  Article  Google Scholar 

  7. Edupuganti R, Taliaferro JM, Wang Q, Xie X, Cho EJ, Vidhu F, Ren P, Anslyn EV, Bartholomeusz C, Dalby KN (2017) Discovery of a potent inhibitor of MELK that inhibits expression of the anti-apoptotic protein Mcl-1 and TNBC cell growth. Bioorg Med Chem 25(9):2609–2616

    CAS  Article  Google Scholar 

  8. Friedman JH (1991) Multivariate adaptive regression splines. The annals of statistics. 19:1–67

    MathSciNet  MATH  Google Scholar 

  9. Ge W, Hao X, Han F, Liu Z, Wang T, Wang M, Chen N, Ding Y, Chen Y, Zhang Q (2019) Synthesis and structure-activity relationship studies of parthenolide derivatives as potential anti-triple negative breast cancer agents. Eur J Med Chem 166:445–69

    CAS  Article  Google Scholar 

  10. Guan L, Yang H, Cai Y, Sun L, Di P, Li W, Liu G, Tang Y (2019) ADMET-score—a comprehensive scoring function for evaluation of chemical drug-likeness. MedChemComm 10(1):148–157

    ADS  CAS  Article  Google Scholar 

  11. Hou Y, Zhu L, Li Z, Shen Q, Xu Q, Li W, Liu Y, Gong P (2019) Design, synthesis and biological evaluation of novel 7-amino-[1, 2, 4] triazolo [4, 3-f] pteridinone, and 7-aminotetrazolo [1, 5-f] pteridinone derivative as potent antitumor agents. Eur J Med Chem 163:690–709

    CAS  Article  Google Scholar 

  12. Hu K, Law JH, Fotovati A, Dunn SE (2012) Small interfering RNA library screen identified polo-like kinase-1 (PLK1) as a potential therapeutic target for breast cancer that uniquely eliminates tumor-initiating cells. Breast Cancer Res 14(1):1–5

    CAS  Article  Google Scholar 

  13. Iqbal J, Ejaz SA, Khan I, Ausekle E, Miliutina M, Langer P (2019) Exploration of quinolone and quinoline derivatives as potential anticancer agents. DARU J Pharm Sci 27(2):613–626

    CAS  Article  Google Scholar 

  14. Jalali-Heravi M, Kyani A (2004) Use of computer-assisted methods for the modeling of the retention time of a variety of volatile organic compounds: a PCA-MLR-ANN approach. J Chem Inf Comput Sci 44(4):1328–1335

    CAS  Article  Google Scholar 

  15. Jo J, Kim SH, Kim H, Jeong M, Kwak JH, Han YT, Jeong JY, Jung YS, Yun H (2019) Discovery and SAR studies of novel 2-anilinopyrimidine-based selective inhibitors against triple-negative breast cancer cell line MDA-MB-468. Bioorg Med Chem Lett 29(1):62–65

    CAS  Article  Google Scholar 

  16. Kaplan W. Background Paper 6.5 Cancer and Cancer Therapeutics. World Health Organization (ed) Priority medicines for Europe and the world: update. 2013:6–5.

  17. Kennard RW, Stone LA (1969) Computer aided design of experiments. Technometrics 11(1):137–148

    Article  Google Scholar 

  18. Lolak N, Akocak S, Bua S, Supuran CT (2019) Design, synthesis and biological evaluation of novel ureido benzenesulfonamides incorporating 1, 3, 5-triazine moieties as potent carbonic anhydrase IX inhibitors. Bioorg Chem 82:117–122

    CAS  Article  Google Scholar 

  19. Myers RH (1990) Classical and modern regression application. Duxbury Press, CA

    Google Scholar 

  20. Noolvi MN, Patel HM (2013) A comparative QSAR analysis and molecular docking studies of quinazoline derivatives as tyrosine kinase (EGFR) inhibitors: A rational approach to anticancer drug design. J Saudi Chem Soc 17(4):361–379

    CAS  Article  Google Scholar 

  21. Qu FZ, Xiao SN, Wang XD, Zhang Y, Su GY, Zhao YQ (2019) Semi-synthesis and anti-tumor activity of novel 25-OCH3-PPD derivatives incorporating aromatic moiety. Bioorg Med Chem Lett 29(2):189–193

    CAS  Article  Google Scholar 

  22. Tropsha A, Gramatica P, Gombar VK (2003) The importance of being earnest: validation is the absolute essential for successful application and interpretation of QSPR models. QSAR Comb Sci 22(1):69–77

    CAS  Article  Google Scholar 

  23. Veerasamy R, Rajak H, Jain A, Sivadasan S, Varghese CP, Agrawal RK (2011) Validation of QSAR models-strategies and importance. Int J Drug Des Discov 3:511–519

    Google Scholar 

  24. Xu P, Chu J, Li Y, Wang Y, He Y, Qi C, Chang J (2019) Novel promising 4-anilinoquinazoline-based derivatives as multi-target RTKs inhibitors: design, molecular docking, synthesis, and antitumor activities in vitro and vivo. Bioorg Med Chem 27(20):114938

    CAS  Article  Google Scholar 

  25. Yap CW (2011) PaDEL-descriptor: an open source software to calculate molecular descriptors and fingerprints. J Comput Chem 32(7):1466–1474

    MathSciNet  CAS  Article  Google Scholar 

Download references

Acknowledgements

Authors acknowledge the technical effort of the physical chemistry department, Ahmadu Bello University Zaria for their immense advice and various contributions toward the successful completion of the research.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

Author information

Affiliations

Authors

Contributions

All author’s participated accordingly, HAL carried out the computational analysis, AU arranged the manuscript according to the journal format, and SU re-confirmed the results and edited the manuscript to ensure that error was minimized. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Hadiza Abdulrahman Lawal.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lawal, H.A., Uzairu, A. & Uba, S. QSAR, molecular docking studies, ligand-based design and pharmacokinetic analysis on Maternal Embryonic Leucine Zipper Kinase (MELK) inhibitors as potential anti-triple-negative breast cancer (MDA-MB-231 cell line) drug compounds. Bull Natl Res Cent 45, 90 (2021). https://doi.org/10.1186/s42269-021-00541-x

Download citation

Keywords

  • QSAR
  • Molecular docking
  • Ligand-based design
  • Pharmacokinetic analysis