Statistical Analysis of the Impact of Molecular Descriptors on Cytotoxicity of Thiourea Derivatives Incorporating 2-Aminothiazole Scaffold

Abstract

Chemical reactivity descriptors and lipophilicyty (log P) were evaluated via semi-empirical method for the quantum calculation of molecular electronic structure (PM3) in order to clarify the structure–cytotoxic activity relationships of disubstutited thioureas. Analysed compounds were obtained by the linkage of 2-aminothiazole ring, thiourea and substituted phenyl ring. The detailed examination was carried out to establish correlation between descriptors and cytotoxic activity against the MT-4 cells for 11 compounds. For the most active compounds (6 compounds) cytotoxic activity against three cancer cell lines (CCRF-CEM, WIL-2NS, CCRF-SB) and normal human cell (HaCaT) was determined. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction and lactate dehydrogenase (LDH) release were assessed. Regression analysis revealed that electrophilicity index and chemical potential significantly contributed to expain the thioureas cytotoxic potential.

Empirical method for building predictive models of the relationship between molecular structure and useful properties are becoming increasingly important. A large amount of biological target information is becoming available through molecular biology. Quantitative structure–activity relationship (QSAR) methods have much to offer in these areas. For bioactive molecules sometimes a single atomic charge in a small organic molecule may profoundly affect the biological properties of these species.¹⁾

Physicochemical parameters such as molar weight (M), volume (V), surface area (SA), surface area grid (^gSA), logarithm of the partition coefficient (logP), the difference between highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels (HLG), the HOMO energy, the LUMO energy, hardness (η), Mullkien electronegativity (χ), total energy (E_T), binding energy (E_B), isolated atomic energy (E_IA), electronic energy (E_E), core–core interaction (I_C-C), heat of formation (HF), refractivity (Rf), polarizability (α) are known as global reactivity descriptors. These electronic indexes have been successfully applied in describing and predicting the toxicity of oleoylamides,²⁾ 3-styryl-2H-chromenes,³⁾ benzodithiazine derivatives.⁴⁾

The aim of the study is to optimize a cytotoxic effect of 11 thiourea derivatives (listed in Table 1) towards the MT-4 cells in terms of chemical descriptors that are derived from electronic structure calculation. The research was extented and for the most active compounds (6 compounds—CC₅₀<10 µM) the cytotoxic activity against three cancer cell lines (CCRF-CEM, WIL-2NS, CCRF-SB) was determined. Toxic effects of derivatives 1, 2, 5, 6, 7 and 11 on normal human cell line HaCaT were also investigated by determination of mitochondrial dehydrogenase (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay) and lactate dehydrogenase (LDH) activities according to previously described procedures.⁵⁾

Table 1. Strcture of Thiourea Derivatives Used in the Study

Compound	Structure	MT-4 CC50a)
1		9.0
2		9.0
3		38.5
4		10.8
5		8.5
6		7.8
7		7.8
8		50.0
9		>100
10		10.8
11		9.0

a) Compound concentration (µM) required to reduce the viability of mock-infected MT-4 cells by 50%, as determined by the MTT method.

Results and Discussion

A basic skeletal structure of thiourea derivatives consists of two aromatic rings (ring A, ring B) covalently linked by thiourea moiety (Fig. 1).

Fig. 1. Linkage of 2-Aminothiazole Ring, Thiourea and Substituted Phenyl Ring

The conformation of these thiourea derivatives was confirmed by X-ray analysis.⁶⁾ The structures of tested compounds and cytotoxic effect towards MT-4 cells are presented in Table 1.

All investigated thiourea derivatives were found to be highly cytotoxic for exponentially growing MT-4 cells, and many of them reduced cell viability at the low micromolar concentrations (CC₅₀<10 µM). This research was extented and for the most active compounds (6 compounds—CC₅₀<10 µM) the cytotoxic activity against three cancer cell lines (CCRF-CEM, WIL-2NS, CCRF-SB) was determined (Table 2). The cytotoxicity (CC₅₀) of compounds 1, 2, 5, 6, 7 and 11 against these cells was in the same range as for MT-4 cell (12.9–3.5 µM).

Table 2. Antiproliferative Activity of Selected Compounds against Human Leukaemia/Lymphoma, Solid Tumour and Normal Cell Lines

Compound	CCRF-CEMb)	WIL-2NSc)	CCRF-SBd)	HaCaTe)
	CC50a)	CC50a)	CC50a)	CC50a)
1	9	10.2	8.2	>18
2	5.7	9.1	5.4	>9
5	9	12.9	9.7	>25
6	8.6	10.9	8.3	>25
7	10.3	9.5	9.5	>25
11	5.3	10.4	3.5	>9
Doxorubicin	0.02	0.02	0.03	nt

Data represent mean values for three independent determinations. a) Compound concentration (µM) required to reduce cell proliferation by 50%, as determined by the MTT assay, under conditions allowing untreated controls to undergo at least three consecutive rounds of multiplication. b) CD4+ human acute T-lymphoblastic leukaemia cell line. c) Human splenic B-lymphoblastoid cells. d) Human acute B-lymphoblastic leukemia cell line. e) Human immortal keratinocyte cells.

Treatment of normal human HaCaT cells with tested compounds revieled that compounds 5, 6 and 7 did not effect cell viability (MTT assay, Fig. 2) at concentrations 3-times higer than CC₅₀ for MT-4 cell, while compounds 1, 2 and 11 at similarly high concentrations reduced HaCaT cell viability by 50%. Results of LDH assay (Fig. 3) confirmed that compounds 5, 6 and 7 did not express cytotoxic activity against HaCaT cells at concentrations as high as 25 µM. Compounds 1, 2 and 11 at concentrations from 9 to 27 µM caused LDH release of 30 to 60% as compared to controls indicating enhanced number of dead cells.

Fig. 2. Cell Viability Assessed by MTT Mitochondrial Reduction in HaCaT Cells Treated for 24 h with Tested Compound

Data are expressed as the mean±S.D. (presented as a bar) from three independent experiments performed in triplicate. Statistical significance: * p<0.05 and ** p<0.001 refer to the control (untreated) cells.

Fig. 3. LDH Release as a Marker of Cell Death in HaCaT Cells Treated for 24 h with Tested Compound

Data are expressed as the mean±S.D. (presented as a bar) from three independent experiments performed in triplicate. Statistical significance: ** p<0.001 refer to the control (untreated) cells.

The QSAR modeling was performed by utilizing well known methods: simple linear regression and multilinear regression. Calculations were performed using the STATISTICA (data analysis software system), version 12.⁷⁾ The relationship of cytotoxicity against 18 molecular descriptors calculated were examined using HyperChem⁸⁾ and described.⁶⁾

There is a statistically significant correlation between cytotoxicity and M, log P, the HOMO energy, the LUMO energy, χ obtained from the equation χ=−(HOMO+LUMO)/2, electrophilicity⁹⁾ (ω) obtained from the equation ω=χ²/(LUMO−HOMO), where M expressed in atomic mass unit (amu), HOMO, LUMO, ω and χ expressed in eV. Table 3 shows all examined compounds together with molecular descriptors, which significantly correlate with their cytotoxicity. This work also examined compound agglomeration using both: Ward’s method and k-means algorithm.

Table 3. Mollecular Descriptors of Thiourea Derivatives Used in the Study

Compound	M	Log P	HOMO	ω	LUMO	χ
1	314.22	4.54	−8.835	49.1547	−1.566	5.201
2	304.21	4.78	−8.779	48.3625	−1.611	5.195
3	253.31	3.89	−8.786	48.2436	−1.521	5.154
4	253.31	3.89	−8.855	49.5293	−1.587	5.221
5	283.79	4.73	−8.718	47.1423	−1.513	5.116
6	287.76	4.40	−8.870	49.9340	−1.645	5.258
7	303.32	4.63	−8.663	46.2624	−1.506	5.085
8	235.32	3.75	−8.697	46.6636	−1.458	5.078
9	265.35	3.49	−8.503	43.5852	−1.416	4.960
10	253.31	3.89	−8.808	48.7527	−1.579	5.194
11	314.22	4.54	−8.829	49.0854	−1.579	5.204

M in amu, HOMO, LUMO, ω and χ in eV.

Table 4. Formulas Describing the Relationship of Cytotoxicity against Selected Molecular Descriptors

Equation	R	R2	R2adj.	F	SEE	Q2
Log(1/MT-4)=0.69(±0.16)×log P−4.1(±0.7)	0.81	0.65	0.61	17.18	0.239	0.55
Log(1/MT-4)=−4.57(±1.13)×LUMO−8.23(±0.75)	0.80	0.64	0.60	16.26	0.244	0.55
Log(1/MT-4)=3.43(±0.98)×χ−18.85(±5.08)	0.75	0.57	0.53	12.10	0.267	0.47
Log(1/MT-4)=0.15(±0.05)×ω−8.29(±2.35)	0.71	0.50	0.44	9.15	0.287	0.35
Log(1/MT-4)=−2.48(±0.86)×HOMO−23.06(±7.54)	0.69	0.48	0.43	8.42	0.293	0.31
Log(1/MT-4)=0.009(±0.004)×M−3.57(±1.01)	0.62	0.38	0.32	5.67	0.320	0.31

R—correlation coefficient; R²—determination coefficient; R²_adj.—adjusted determination coefficient for calibration; F—Fisher test value; SEE—a standard error of estimate; Q²—determination coefficient of LOO validation.

Figures 4 and 5 show dependencies for which the R² value is greater than 0.6 and Q² value is greater than 0.5. This fits the assumption for which the QSAR model is acceptable.¹⁰⁾ Figure 4 shows a strong positive correlation between cytotoxicity and log P. Compounds 4 and 10 were outside the confidence interval. The third compound 3, one of those having the fluorine atom in phenyl ring, is in range of the confidence interval. Even though, compounds 3, 4, 10 characterized the same value of log P, they differ from one another in terms of cytotoxicity. Figure 5 shows a strong negative correlation between cytotoxicity and LUMO (compounds 3, 5, 7 were outside the confidence interval).

Fig. 4. Correlation of Cytotoxicity and Log P

Fig. 5. Correlation of Cytotoxicity and LUMO

The analyzed compounds were divided by their cytotoxicity into two groups using the k-means algorithm. The agglomeration results are shown in Fig. 6 and Table 5 which also present distances from centers of each group. The first group, containing compounds 3, 8, 9 is characterised by larger distances from its center, what indicates significant variation within the group. This group is also strongly heterogeneous according to its complex structure. It only conatins compounds showing low cytotoxicity. The second group conatins highly cytotoxic compounds only. The distances from the center of the group are significantly smaller, what indicates its homogeneity.

Fig. 6. Results of Compound Agglomeration Using the k-Means Method

Table 5. Results of Compound Agglomeration Using the k-Means Method

Compound	Group	Distance
1	2	0.0231
2	2	0.0231
3	1	0.4547
4	2	0.1815
5	2	0.0872
6	2	0.1836
7	2	0.0231
8	1	0.1615
9	1	0.6161
10	2	0.1815
11	2	0.0231

Ward’s method of agglomeration, based on squared Euclidean distance, was also performed. The results are shown in Fig. 7. This method clearly validates the choice of molecular descriptors proposed earlier in this work.

Fig. 7. Dendrogram Based on Standardized: Log P, M, LUMO, HOMO, χ, Cytotoxic Activity (Expressed as Log 1/IC₅₀ in µM) against MT-4

Ward’s method of agglomeration, based on squared Euclidean distance, was used.

There are three groups of compounds:

• —Non-toxic compound: 9
• —Medium-toxic compounds: 3, 4, 8, 10
• —Toxic compounds: 1, 2, 5, 6, 7, 11.

This division is consistent with the pharmaceutical classification of those compounds.

Using multiple regression, a test was performed to check the correlation of the cytotoxicity and molecular descriptors. Best matching results of R² value are shown in Figs. 8–11. Linear correlations of the multiple regression were determined for two variables/descriptors only, so that at least five compounds accounted for one parameter.^11,12) All equations satisfy the coincidence condition.

Fig. 8. Correlation of the Cytotoxicity and Log P, HOMO; R—Correlation Coefficient; R²—Determination Coefficient; R²_adj.—Adjusted Determination Coefficient for Calibration; SEE—A Standard Error of Estimate, Q²—Correlation Coefficient for Cross-Validation

Fig. 9. Correlation of the Cytotoxicity and Log P, ω

Fig. 10. Correlation of the Cytotoxicity and Log P, LUMO

Fig. 11. Correlation of the Cytotoxicity and Log P, χ

As is clear from the equations obtained, the cytotoxicity of analyzed compounds is influenced by the logarithm of the partition coefficient log P as well as electron descriptors such as: HOMO, LUMO, ω, χ, log P and χ descriptors show the positive impact, whereas HOMO and LUMO ones, the negative impact on the cytotoxicity of studied compounds. Performed statistical analysis have confirmed the impact of HOMO, LUMO, χ, ω electron descriptors on the cytotoxic activity of compounds.¹³⁾ For all obtained dependencies, following compounds 3, 4, 7 and 10 are always outside the confidece interval. Those compounds differ from others by chemical structure, since they contain fluorine atoms in different positions of phenyl ring.

Figure 12 (part a) shows the highest occupied HOMO and lowest unoccupied LUMO molecular orbitals of compound 6 featuring the largest value of cytotoxicity. Figure 12 (part b) shows the highest occupied HOMO and lowest unoccupied LUMO molecular orbitals of compound 9 featuring the lowest value of cytotoxicity. There should be noticed the virtually symetric reversal of the spatial position of HOMO and LUMO orbitals, which may be a clue for further search for toxic substances.

Fig. 12. Highest Occupied HOMO (Light Gray) and Lowest Unoccupied LUMO (Dark Gray) Molecular Orbitals of Compounds

a) Compound 6. b) Compound 9.

Summary and Conclusion

The results of performed analysis show a strong correlation of cytotoxicity and log P. Also the correlation between cytotoxicity and LUMO satisfies the condition of the QSAR model acceptability.¹⁴⁾ Regardless to the correlation between cytotoxicity and two parameters, best match in terms of R² was obtained for log P and χ as well as log P and LUMO. Only a little worse match was obtained for log P and ω.

Analyzing the structure of studied compounds, it can be noted, that the position of fluorine in the phenyl ring has a meaningful impact on the cytotoxicity: Fluorine in position 2 of the phenyl ring significantly lowers the level of compound cytotoxicity.

This analysis can be used to search for other thiourea derivatives incorporating a 2-aminothiazole scaffold featuring properties.

Experimental

Cytotoxicity Assays

Exponentially growing MT-4 cells were seeded at an initial density of 1×10⁵ cells/mL in 96-well plates in RPMI-1640 medium, supplemented with 10% fetal bovine serum (FBS), 100 units/mL penicillin G and 100 µg/mL streptomycin. Cell viability was determined after 96 h at 37°C by the MTT method.¹⁵⁾

As far as stationary monolayers (analogous to those which support the replication of the other RNA and DNA viruses) are concerned, MDBK and BHK cells were seeded in 24-well plates at an initial density of 6×10⁵ and 1×10⁶ cells/mL, respectively, in Minimum Essential Medium with Earle’s salts (MEM-E), L-glutamine, 1 mM sodium pyruvate and 25 mg/L kanamycin, supplemented with 10% horse serum (MDBK) or 10% FBS (BHK). Cell viability was determined after 48–96 h at 37°C by the MTT method.

Vero-76 cells were seeded in 24-well plates at an initial density of 4×10⁵ cells/mL, in Dulbecco’s modified Eagle’s medium (DMEM) with L-glutamine and 25 mg/L kanamycin, supplemented with 10% FBS. Cell viability was determined after 48–96 h at 37°C by the crystal violet staining method.

Cell lines derived from human haematological tumours [CD4+ human T-cells containing an integrated human T-cell leukemia virus type-1 (HTLV-1) genome (MT-4); CD4+ human acute T-lymphoblastic leukaemia (CCRF-CEM), human splenic B-lymphoblastoid cells (WIL-2NS), human acute B-lymphoblastic leukaemia (CCRF-SB)] were seeded at an initial density of 1×10⁵ cells/mL in 96 well plates in RPMI-1640 medium supplemented with 10% fetal calf serum (FCS), 100 units/mL penicillin G and 100 µg/mL streptomycin. Cell viability was determined after 96 h at 37°C by the MTT method.

All cell cultures were then incubated at 37°C in a humidified, 5% CO₂ atmosphere, in the absence or presence of serial dilutions of test compounds in culture medium. Before dilutions, compounds were dissolved in dimethyl sulfoxide (DMSO) at 100 mM.

Cell Culture: Conditions and Treatments

Human immortal keratinocyte cell line from adult human skin (HaCaT), cell line was bought from American Type Culture Collection (Rockville, U.S.A.), cultured in DMEM supplemented with antibiotics (penicillin and streptomycin), and 10% heat-inactivated FBS-fetal bovine serum (Gibco Life Technologies, U.S.A.), at 37°C and 5% CO₂ atmosphere. Cells were passaged using trypsin-ethylenediaminetetraacetic acid (EDTA) (Gibco Life Technologies, U.S.A.) and cultured in 24-well plates (2.5×10⁴ cells per well). Experiments were conducted in DMEM with 2% FBS.

Cell Viability Assessment (Mitochondrial Function Assessment)

The cell viability was assessed by determination of MTT salt (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, Sigma-Aldrich, Germany) conversion by mitochondrial dehydrogenase.⁵⁾ The cells were incubated for 24 h in 24-well plates with concentration of tested compounds, and subsequently for another 2 h with 0.5 mg/mL of MTT solution which is converted in live cells under the effect of mitochondrial dehydrogenase into insoluble formazan. The converted dye was then solubilized with the use of 0.04 M HCl in absolute isopropanol. Absorbance of solubilized formazan was measured spectrophotometrically at 570 nm (using Epoch microplate reader, BioTek Inc., U.S.A.) equipped with Gen5 software (BioTech Instruments, Inc., Biokom).

Cell viability was presented as a percent of MTT reduction in the treated cells versus the controls (cells incubated in serum-free DMEM without extracts). The relative MTT level (%) was calculated as [A]/[B]×100, where [A] is the absorbance of the test sample and [B] is the absorbance of control sample containing the untreated cells. Decreased relative MTT level indicates decreased cell viability.

Lactate Dehydrogenase Release Assay (Cellular Membrane Integrity Assessment)

Release of LDH from the cytosol to culture medium is a marker of cell death. The assay was performed after 24 h incubation of HaCaT cells in 24-well plates with investigated concentrations of each compounds. The activity of LDH released from cytosol of damaged cells to the supernatant was measured using the protocol of the cytotoxicity detection kit LDH test described by the manufacturer (Roche Diagnostics, Germany). Absorbance was measured at 490 nm using a microplate reader (using Epoch microplate reader, BioTek Inc.) equipped with Gen5 software (BioTech Instruments, Inc.).⁵⁾

Compounds mediated cytotoxicity expressed as the LDH release (%) was determined by the following equation: [(A test sample−A low control)/(A high control−A low control)]×100% (A-absorbance); where “low control” were cells in DMEM with 2% FBS without tested compounds, and “high control” were cells incubated in DMEM with 2% FBS with 10% Triton X-100 (100% LDH release).

Statistical Analysis

All results are presented as experimental mean values and they were compared using one-way ANOVA with the Tukey’s post-hoc test (Statistica ver. 8, StatSoft, Poland); asterisk (*) indicates a significant difference p<0.05 and double asterisk (**) indicates a significant difference p<0.001.

Methodology of Molecular Modeling and QSAR Models Development

Conformational search and physicochemical parameters were calculated using HyperChem ver. 8.0.⁸⁾ Extensive conformational search was performed at molecular mechanics level with Optimized Potentials for Liquid Simulations (OPLS) force field. The most stable structures obtained were subsequently optimized to the closest local minimum at the semiempirical level using PM3 parametrizations. Converegence criteria were set to 0.1 and 0.01 kcal·mol⁻¹·Å⁻¹ for OPLS and PM3 calculations, respectively. The selected descriptors were then used to develop a QSAR model.

The QSAR model was performed utilizing Stepwise Multiple Linear Regression Method. Calculations were performed using the STATISTICA (data analysis software system), version 12.⁶⁾ Log(1/MT-4) was taken as a dependent value for analysis, while molecular descriptors were independent values. For the Multiple Linear Regression Analysis (MLRA) technique, stepwise regression was chosen in the development of the QSAR model, in which a selection algorithm was used to select a subset of the input variables, X. The stepwise method combines two approaches, which are the forward and backward stepping.

In forward stepping, the partial F (statistical significance) values for all variables outside the model were calculated. This process is continued until no more variables qualified to enter the model. In backward stepping, the partial F values for all variables inside the model were calculated. The variable with the lowest partial F value was removed from the model. This process is continued until no more variables were qualified to be removed from the model.¹³⁾ Created models were validated using Leave-one-out cross-validation (LOO).^16,17) We consider a QSAR model predictive, if the following conditions are satisfied: R²>0.6 and Q²>0.5.¹⁰⁾

Compound agglomeration (Ward’s method, k-means method) was conducted using STATISTICA 12.0 software. Ward’s method of agglomeration, based on squared Euclidean distance, was used. Dendograms, based on standardized: cytotoxity activity against MT-4 (expressed as log(1/MT-4), log P, M, LUMO, HOMO, χ and ω, was created. Compound agglomeration using k-means was carried out on a standardized value of cytotoxicity against MT-4.

Acknowledgments

This work was partially supported by the Ministry of Science and Higher Education funding for statutory activities of young researchers of Faculty of Automatic Control, Electronics and Computer Science. This work was partially supported by the Medical University of Warsaw and carried out with the use of CePT infrastructure financed by the European Union—the European Regional Development Fund within the Operational Programme Innovative Economy for 2007–2013. The authors wish to thank the scientific group of Professor Paolo La Colla, Universita di Cagliari, Italy, for performing cytoxicity screenings.

Conflict of Interest

The authors declare no conflict of interest.

References

• 1) Winkler D. A., Brief. Bioinform., 3, 73–86 (2002).
• 2) Sakagami H., Uesawa Y., Ishihara M., Kagaya H., Kanamoto T., Terakubo S., Nakashima H., Takao K., Sugita Y., Anticancer Res., 35, 5341–5351 (2015).
• 3) Uesawa Y., Sakagami H., Ishihara M., Kagaya H., Kanamoto T., Terakubo S., Nakashima H., Yahagi H., Takao K., Sugita Y., Anticancer Res., 35, 5299–5307 (2015).
• 4) Pogorzelska A., Sławiński J., Brożewicz K., Ulenberg S., Bączek T., Molecules, 20, 21960–21970 (2015).
• 5) Stefanska J., Nowicka G., Struga M., Szulczyk D., Koziol A. E., Augustynowicz-Kopec E., Napiorkowska A., Bielenica A., Filipowski W., Filipowska A., Drzewiecka A., Giliberti G., Madeddu S., Boi S., La Colla P., Sanna G., Chem. Pharm. Bull., 63, 225–236 (2015).
• 6) Grudzinski I. P., Bystrzejewski M., Cywinska M. A., Kosmider A., Poplawska M., Cieszanowski A., Fijalek Z., Ostrowska A., Parzonko A., J. Appl. Toxicol., 34, 380–394 (2014).
• 7) StatSoft, Inc. (2014). STATISTICA (data analysis software system), version 12. 〈www.statsoft.com〉
• 8) Hyper Chem (TM), Professional, Hypercube, Inc., 1115 NW 4th Street, Gainesville, Florida 32601, U.S.A.
• 9) Parr R. G., Szentpály L., Liu S., J. Am. Chem. Soc., 121, 1922–1924 (1999).
• 10) Golbraikh A., Tropsha A., J. Comput. Aided Mol. Des., 5, 231–243 (2002).
• 11) Dehmer M., Varmuza K., Bonchev D., Emmert-Streib F., “Statistical Modeling of Molecular Descriptors in QSAR/QSPR,” Wiley-Blackwell, Hoboken, NJ, 2012.
• 12) Topliss J. G., Edwards R. P., J. Med. Chem., 22, 1238–1244 (1979).
• 13) Kupcewicz B., Jarzęcki A., Małecka M., Krajewska U., Rozalski M., Bioorg. Med. Chem. Lett., 24, 4260–4265 (2014).
• 14) Frimayanti N., Yam M. L., Lee H. B., Othman R., Zain S. M., Rahman N. A., Int. J. Mol. Sci., 12, 8626–8644 (2011).
• 15) Pauwels R., Balzarini J., Baba M., Snoeck R., Schols D., Herdewijn P., Desmyter J., De Clercq E., J. Virol. Methods, 20, 309–321 (1988).
• 16) Afantitis A., Melagraki G., Sarimveis H., Koutentis P. A., Markopoulos J., Igglessi-Markopoulou O., Bioorg. Med. Chem., 14, 6686–6694 (2006).
• 17) Khaledian S., Saaidpour S., Orient. J. Chem., 31, 1969–1976 (2015).