Antimicrobial Activity of Nitrogen-Containing 5-α-Androstane Derivatives: In Silico and Experimental Studies

We evaluated the antimicrobial activity of thirty-one nitrogen-containing 5-α-androstane derivatives in silico using computer program PASS (Prediction of Activity Spectra for Substances) and freely available PASS-based web applications (such as Way2Drug). Antibacterial activity was predicted for 27 out of 31 molecules; antifungal activity was predicted for 25 out of 31 compounds. The results of experiments, which we conducted to study the antimicrobial activity, are in agreement with the predictions. All compounds were found to be active with MIC (Minimum Inhibitory Concentration) and MBC (Minimum Bactericidal Concentration) values in the range of 0.0005–0.6 mg/mL. The activity of all studied 5-α-androstane derivatives exceeded or was equal to those of Streptomycin and, except for the 3β-hydroxy-17α-aza-d-homo-5α-androstane-17-one, all molecules were more active than Ampicillin. Activity against the resistant strains of E. coli, P. aeruginosa, and methicillin-resistant Staphylococcus aureus was also shown in experiments. Antifungal activity was determined with MIC and MFC (Minimum Fungicidal Concentration) values varying from 0.007 to 0.6 mg/mL. Most of the compounds were found to be more potent than the reference drugs Bifonazole and Ketoconazole. According to the results of docking studies, the putative targets for antibacterial and antifungal activity are UDP-N-acetylenolpyruvoylglucosamine reductase and 14-α-demethylase, respectively. In silico assessments of the acute rodent toxicity and cytotoxicity obtained using GUSAR (General Unrestricted Structure-Activity Relationships) and CLC-Pred (Cell Line Cytotoxicity Predictor) web-services were low for the majority of compounds under study, which contributes to the chances for those compounds to advance in the development.


Introduction
According to the World Health Organization (https://www.who.int/news-room/fact-sheets/detail/ the-top-10-causes-of-death), infectious diseases are among the top ten leading causes of death worldwide. This is mainly due to the emerging antimicrobial resistance, which is a threat to global health itself (https://www.who.int/news-room/feature-stories/ten-threats-to-global-health-in-2019). In particular, nosocomial infections caused by methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococcus faecium (VRE), and drug-resistant Streptococcus pneumoniae have been designated as severe public threats by the US Centers for Disease Control and Prevention [1]. Furthermore, the resistant bacteria capable of surviving in the presence of the almost all known antibiotics, such as multidrug-resistant Staphylococcus aureus (MRSA), are the major source of concerns worldwide [2][3][4][5][6][7]. It is essential to outline the main factors contributing to the antimicrobial resistance to find a way to deal with it. The main reasons why bacteria can acquire and demonstrate resistance in the clinic are as follows: (1) high rates of mutations (in some bacteria); (2) exchange of genetic information via mobile genetic elements (plasmids) in some bacteria; (3) violation of medical prescriptions for taking antibiotics; (4) a limited number of antimicrobial agents in clinical practice.
Therefore, new approaches are needed to fight antimicrobial resistance. Both modifications of known and discovery of novel antibacterial and antifungal molecules are applied to develop the antimicrobial agents active against the resistant pathogens [8][9][10][11][12][13].
One of the promising strategies is the chemical modification of the steroids. Two of the adopted ways of doing so are the introduction of the oxime group in the steroid scaffold and attachment of amino groups to steroids. Previously, it was shown that such modifications improve many biologically relevant properties of steroids: modified derivatives are often less toxic and possess the pleasant bioavailability. Moreover, many such compounds were shown to be active against the bacteria, including resistant ones [14][15][16][17][18][19][20]. Also, steroidal oximes [21,22], and azides [23][24][25] are considered as the suitable starting points for the development of more complex molecules having their advantages [26][27][28][29][30][31].
It is worth to notice that we found strong structure-activity relationships for antiarrhythmic and radioprotective activity (RPA) of epimeric 3-amino-5α-androstan-17-ol and 17-amino-5α-androstan-3-ole. 17β-Amino-5α-androstan-3 β-ole is characterized by the best antiarrhythmic activity and 3α-amino-5α-androstan-17α-ole with the best RPA [21]. 3α-Amino-5α-androstan-17α-ole was selected and evaluated for antibacterial and antifungal activity. Results proved the high antimicrobial activity of this epimer [22]. According to our previous studies on the N-containing derivatives of 5α-androstane series, the presence of 3α-amino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] is essential. Recently, we found that the antimicrobial action of N-containing 5α-androstane derivatives is probably due to the very selective interaction since even slight changes in the molecular structure may reduce or increase their activity significantly [22].
These data [23,24] prompted us to continue study in this field and investigate the antimicrobial activity of 17-amino-5α-androstan-3-oles and derivatives as well as intermediate N-containing compounds that we have synthesized earlier.

Chemistry
In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds.

N Structure N Structure
In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds.

Structure N Structure
In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds.

Structure N Structure
In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. In continuation of our studies of N-containing 5α-androstane derivatives we conducted further in silico and in vitro studies of their antimicrobial activity and its selectivity [22]. Most of the compounds synthesized earlier revealed different pharmacological effects. According to our previous studies on the N-containing derivatives of 5α-androstane series, the importance of 3αamino-and 17α-hydroxy functional groups for antimicrobial and radioprotective activity [21,22] was shown.
D-homoandrostane derivatives 20 and 21 were synthesized from oxime 22 using Beckmann molecular rearrangement procedure [42] and 3α-phthalimido-5α-androstane-17-one 23 from epiandrosterone by Mitsunobu reaction [43]. The data about any biological activity of steroidal oximes 15 and 22, phthalimido-23, and azido steroids 24 and 27 have not been found so far in the literature. The structure of compounds is presented in Table 1 and way of their preparation in Scheme 1. The results of biological testing provide a strong impetus for a more extensive study of the derivatives mentioned above and the continuation of the search for new, highly effective antimicrobial agents among N-containing 5α-steroidal compounds. Compounds 28-30 were synthesized according to the procedure described earlier [44,45,46]; compound 31 was purchased from Fluka.

Biological Activity and Toxicity Predictions
PASS prediction of antimicrobial activities was performed for thirty-one compounds selected for investigation. Antibacterial activity was predicted for 27 out of 31 compounds with probability "to be active" Pa values ranging from 0.298 to 0.458. Antifungal activity was predicted for 25 compounds with Pa values ranging from 0.171 to 0.427 (Table 2). Compounds 28-30 were synthesized according to the procedure described earlier [44,45,46]; compound 31 was purchased from Fluka.

Biological Activity and Toxicity Predictions
PASS prediction of antimicrobial activities was performed for thirty-one compounds selected for investigation. Antibacterial activity was predicted for 27 out of 31 compounds with probability "to be active" Pa values ranging from 0.298 to 0.458. Antifungal activity was predicted for 25 compounds with Pa values ranging from 0.171 to 0.427 (Table 2). Compounds 28-30 were synthesized according to the procedure described earlier [44][45][46]; compound 31 was purchased from Fluka.

Biological Activity and Toxicity Predictions
PASS prediction of antimicrobial activities was performed for thirty-one compounds selected for investigation. Antibacterial activity was predicted for 27 out of 31 compounds with probability "to be active" Pa values ranging from 0.298 to 0.458. Antifungal activity was predicted for 25 compounds with Pa values ranging from 0.171 to 0.427 (Table 2). Pa values below 0.5 indicate not only the probability for the chemical compound to be found active in the experiment, but also testify on its relative novelty to the training set or the presence of similar compounds among the ones having activities besides predicted, which is probably the case for steroids, known for their wide range of biological activities [47,48].
PASS predicts the antibacterial and antifungal effects for chemical compounds in general, furthermore, also activity against the limited number of bacteria and fungi. In addition, to rationally select the particular bacterial and/or fungal target for chemical compound, AntiBAC-pred [49,50] and AntiFun-Pred [51] may be used, since they are able to predict activity against many distinct bacterial and fungal species and strains. AntiBac-Pred and AntiFun-Pred differ from standard version of PASS in training sets, which consist only of the structures of chemical compounds evaluated experimentally against bacteria (AntiBac-Pred) or fungi (AntiFun-Pred).
Application of the AntiBac-Pred to chemical structures of the studied 5-α-androstane derivatives provided the following results: 23 out of 35 compounds were predicted as active ones against the L. plantarum and S. lugdunensis; 4 compounds were predicted as active against B. sphaericus, C. ramosum, P. gingivalis, resistant S. simulans, and S. mutans. Besides, at least one compound has been predicted as active against one or more of 25 other bacteria, including two resistant strains (resistant S. simulans and resistant M. ulcerans).
Therefore, the compounds under study may be tested experimentally against the vast and diverse set of microbial organisms. The results of prediction, including up to three best-rated chemical structures for the selected bacteria and fungi, are given in Supplementary Materials.
Predictions of rat acute toxicity for intraperitoneal and oral routes of administration obtained using computer program GUSAR [52][53][54] are given in the Supplementary Materials. As could be seen CLC-Pred [49,50], one more PASS-based web resource, was used to assess the potential cytotoxicity of the studied compounds against the 22 non-tumor cell lines. 22 out of 31 compounds were not predicted as cytotoxic at the cutoff Pa It is necessary to notice that the PASS-based approach estimates the probability of belonging to the classes of "actives". However, it does not determine the concentration/dose, which will induce the predicted action. Therefore, the dose-cytotoxic effect relationships should be studied for the compounds mentioned above, particularly against the predicted vulnerable cell lines.
Overall, PASS and PASS-based web applications are able not only to provide the computational assessment for chemical compounds to have general antimicrobial effect and activity against the particular microbial species and strains, but also to give some insights about cytotoxicity against the particular non-tumor cell lines.

Antimicrobial Activity
The antimicrobial activity of the synthesized compounds was evaluated using the microdilution method for determining the minimal inhibitory and minimal bactericidal/fungicidal concentrations.
Results of evaluation of antibacterial activity of compounds 1-31 are shown in     It should be noticed that bacteria, in general, showed different sensitivities to compounds tested. Nevertheless, three bacteria species, S. aureus, L. monocytogenes, and P. aeruginosa appeared to be very sensitive to compound 19. Completely different was the sensitivity of E. coli and S. typhimirium toward compounds tested. Thus, the antibacterial potency against S. aureus can be presented as: Completely different was the sensitivity of E. coli and S. typhimirium to compounds tested. As it is already mentioned, compound 19 exhibited excellent activity against all bacterial strains tested. On the other hand compounds 19, 1, 3, 4, and 2 exhibited good activity also against S. aureus and L. monocytogenes with MIC and MBC at 0.0005-0.02 mg/mL and 0.0007-0.04 mg/mL respectively, while some of them (19, 2, 4) with MIC at 0.0015-0.025 mg/mL and MBC 0.003-0.037 mg/mL against P. aeruginosa and 1, 3, 4, and 19, with MIC and MBC at 0.005-0.02 and 0.007-0.037 mg/mL respectively against S. typhimirium. Good activity against this bacterium was observed also for compound 5. Compounds 1, 4, 10, and 28 exhibited good activity against E. coli with MIC at 0.007-0.025 and MBC at 0.015-0.075 mg/mL.
In particular, for the Gram-positive bacteria, the range of MIC and MBC were at 0.0005-0.3 mg/mL and 0.0007-0.45 mg/mL, respectively, while for the Gram-negative bacteria, MIC and MBC ranged at 0.0015-0.3 and 0.003-0.6 mg/mL. It seems that Gram-positive bacteria are more sensitive to the tested compounds than Gram-negative bacteria.
At the same time, it was observed that many compounds exhibited the same potency among the same bacteria species. Thus, for example, compounds 1, 3, and 22, as well as 4 and 10, have the same potency against S. aureus. Compounds 2 and 4, as well as 28 and 29, showed the same activity against L. monocytogenes. The same was observed for other species as well. On the other hand, some compounds appeared to be inactive against some bacteria species. Thus, compounds 6, 7, 9, 12, 13, 15, 17, 21, 25, 26, and 27 did not display any activity against S. aureus being active against almost all other species . Compounds 6, 13, 25, 26, and 27 were also inactive against L. monocytogenes.
In general, compounds 6 and 27 were found to be the most inactive compounds. It should be mentioned that compounds 1-4, 10, 11, 28, and 29 showed better antibacterial potency than both antibiotics used as reference drugs.
For 5α-androstan derivatives the presence of 17α-amino-3α-hydroxy-(10), as well as 17β-amino-3β-hydroxy groups (1) is beneficial for antibacterial activity. In general, replacement of the 17β-amino with 17α-amino group (2) as well as of alkyl substitution of 17β-amino group led to compounds 4 and 11 with decreased, but still good activity. Introduction of 17β-formamido group led to compound 6 which was completely inactive against all bacteria tested. In general, he order of activity of these derivatives can be presented as 10 > 1 > 2 > 4 > 11 > 3 > 8 > 7 > 6.
It should be mentioned that in general, azido derivatives, together with hydroxyimino derivatives, were among the less active steroids.
All compounds were tested against three resistant bacterial strains (MRSA, P. aeruginosa and E. coli) ( Table 3) and their antibacterial potential can be presented as follows: Compound 19 again showed the best activity as in the case of ATCC bacteria with MIC and MBC at 0.000015-0.015 mg/mL and 0.0003-0.037 mg/mL, respectively. The lowest antibacterial activity was observed for compound 27 with MIC 0.30 mg/mL and MBC 0.60 mg/mL.
The resistant strains, as in case of the non-resistant strains, expressed different sensitivity-towards compounds tested as well. Nevertheless, all three resistant strains were susceptible to 19 and very resistant to 27.
In a group of 5α-androstan-3β-ol derivatives, the most beneficial for antibacterial activity against resistant strains was the presence of the 17β-amino group (1). Epimerization to 17α-amino (2) decreased a little activity. The replacement of the 3β-hydroxy group in compound 2 by 3α-hydroxy resulted in less active compound 10. In general, the substitution of the free 17-amino group was not beneficial for activity against resistant strains. Thus, the presence of 17β-N-methylamino-(4), as well as 17α-cyanomethylamino groups (8), appeared to be very negative for activity.
For 5α-androst-2-en derivatives, the most beneficial was the presence of the 17β-amino group (19). This compound, in general, was the most active among all 31 compounds tested. The positive influence also had 17β-formamido substituent (14), while acetoximino-(13) had a negative effect on antibacterial activity against resistant strains. In a group of hydroximino derivatives, the best result was observed with 3α-methoxy-16-hydroximino substitution of 5α-androstan-17-one core (22), which showed, in general, good activity among all compounds tested. On the contrary, the presence of a 3,17-hydroximino group and two double bonds 1,2 and 4,5 in A ring of steroid core (17) was very negative. Among azido derivatives, the most positive contribution to the activity was shown by the presence of 3α-azido-17β-hydroxy groups in 5α-androstan ring (24). The substitution of the 17β-hydroxy group by tosyloxy was detrimental for the activity (25). Finally, for 17a-aza-d-homoandrost-5-en-17-one 28 derivatives beneficial for activity was the presence of 3β-hydroxy group as well as the double bond in 5,6 positions (28). The reduction of this double bond had a negative effect leading to compound 29, which is among the less active compounds.
In conclusion, the structure-activity relationship studies against resistant strains revealed that substituents beneficial for antibacterial activity appeared to be the same as in case of non-resistant bacteria.

Evaluation of Antifungal Activity
The antifungal potential of tested compounds is shown in Table 4      Again, it was observed that many compounds exhibited the same sensitivity against the same fungi. For example, compounds 3, 4, 9-11, 15, 20-24, and 28 exhibited the same moderate activity against A. fumigatus.
Compounds 16, 18 and 19 showed very good activity against A. fumigatus with MIC at 0.007-0.075 mg/mL and MFC at 0.015-0.15 mg/mL. A good activity was observed for 19 against A. versicolor. Compound 16 exhibited promising activity against A. ochraceus, while compound 3 and 28 against P. funiculosum and P. ochrochloron with MIC 0.007mg/mL and MFC 0.015 mg/mL. Very potent appeared to be compounds 3 and 28 as well as 4 against T. viride with MIC and MFC at 0.003 mg/mL and 0.007 mg/mL. The good activity was shown by compound 1 against A. fumigatus and A. versicolor and compounds 10 and 11 against P. funiculosum with MIC and MFC at 0.015 mg/mL and 0.037 mg/mL, respectively.
The analysis of the structure-activity relationship revealed that the presence of 17α-aza-and 3β-hydroxy groups in d-homo-androst-5-en-17-one core (28) was the most beneficial for antifungal activity followed by the 17β-amino-on 5α-androst-2-en moiety (19). On the contrary with antibacterial activity substitution of the 17-amino group appeared to be responsible for good activity. Thus, the presence of 17β-(N,N-dimethylamino)-as well as 17β-aminoethylamino substitution resulted in compounds 3 and 11, which are among five the most active. The 17β-amino-(1), as well as 3-hydroximino substitution (15) of 5α-androstan-17β-ol core, also had a positive impact on antifungal activity.

Docking to Antibacterial Targets
To elucidate the probable mechanism of antibacterial activity of tested compounds, docking studies were performed on five bacterial targets including DNA Topo IV, DNA Gyrase, E. coli Primase, Thymidylate kinase, and E. coli MurB enzymes. The obtained results are given in Table 5.  Docking studies revealed that the scoring function associated with the free energy of binding to E. coli UDP-N-acetylenolpyruvoylglucosamine reductase (MurB) was lower than those obtained for the other enzymes. Hence, it may be concluded that E. coli MurB is the putative target responsible for the antibacterial activity of the tested compounds.
The binding mode of the most active compound 19 (Est. binding energy: 9.62kcal/mol) (Figure 1) showed one hydrogen bond formed between the hydrogen atom of the NH 2 group and the oxygen atom of the side chain of Ser228 (distance 2.53 A). The fused rings interact hydrophobically with the residues Arg213, Gly122, Arg158, Ala123, Ile109, Ile121, Pro110, Ser49, Arg326, Gln119, Asn50, Ala226, Glu324, and Leu217. Docking studies revealed that the scoring function associated with the free energy of binding to E. coli UDP-N-acetylenolpyruvoylglucosamine reductase (MurB) was lower than those obtained for the other enzymes. Hence, it may be concluded that E. coli MurB is the putative target responsible for the antibacterial activity of the tested compounds.

Docking to Antifungal Targets
All the synthesized compounds and reference drugs were docked to different antifungal targets (Squalene synthase, Dihydrofolate reductase, and of C. albicans). It was found that the enzyme lanosterol 14α-demethylase of C. albicans was the most suitable for antifungal activity (Table 6) since the free binding energy was the lowest.

Docking to Antifungal Targets
All the synthesized compounds and reference drugs were docked to different antifungal targets (Squalene synthase, Dihydrofolate reductase, and of C. albicans). It was found that the enzyme lanosterol 14α-demethylase of C. albicans was the most suitable for antifungal activity (Table 6) since the free binding energy was the lowest. Docking results showed that all the synthesized compounds may bind to CYP51Ca in a way that is similar to the binding of ketoconazole (Figure 2). The best docking score was calculated for compound 28, which appeared to be the most favorable inhibitor experimentally. The docking pose of this compound is represented in Figure 3. Based on the docking results, compound 28 takes place inside the enzyme alongside to heme group, forming a hydrogen bond interaction between the oxygen atom of -OH substituent and the hydrogen atom of the side chain of the residue Ser378 (distance 1.98 Å). Moreover, fused rings interact hydroponically with the residues Tyr118, Leu121, Thr122, Leu376, Thr311, Met508, as well as with the heme group ( Figure 4). In the case of compound 19, docking scores revealed that it forms plenty of hydrophobic interactions. Furthermore, 19 forms positive ionizable interactions between the heme group and the -NH2 substituent (Figure 4), which stabilized more the complex of the ligand with the enzyme. This interaction is probably responsible for the lower free energy of binding compare to other compounds and ketoconazole.

Antimicrobial and Cytotoxic Activity Prediction
Prediction of the general antimicrobial activity was carried out using PASS (Prediction of Activity Spectra for Substances) software [47,48]. PASS uses structure-activity relationships derived from the data on biological activity of more than one million molecules, including twenty thousand

Antimicrobial and Cytotoxic Activity Prediction
Prediction of the general antimicrobial activity was carried out using PASS (Prediction of Activity Spectra for Substances) software [47,48]. PASS uses structure-activity relationships derived from the data on biological activity of more than one million molecules, including twenty thousand with antibacterial and five thousand with antifungal activity; to classify previously unseen structures

Antimicrobial and Cytotoxic Activity Prediction
Prediction of the general antimicrobial activity was carried out using PASS (Prediction of Activity Spectra for Substances) software [47,48]. PASS uses structure-activity relationships derived from the data on biological activity of more than one million molecules, including twenty thousand with antibacterial and five thousand with antifungal activity; to classify previously unseen structures

Antimicrobial and Cytotoxic Activity Prediction
Prediction of the general antimicrobial activity was carried out using PASS (Prediction of Activity Spectra for Substances) software [47,48]. PASS uses structure-activity relationships derived from the data on biological activity of more than one million molecules, including twenty thousand with antibacterial and five thousand with antifungal activity; to classify previously unseen structures of chemical compounds as belonging or not belonging to one or more of the 5066 biological activity classes. PASS takes the chemical structure(s) of the molecule(s) under study as MDL MOL file or SDF (structure-data file) as input value and outputs the list of activities with corresponding assessments: Pa, assessment of probability for the structure to represent active molecule, and Pi, assessment of probability for the structure to represent inactive molecule.
The probable action of the studied compounds on the distinct microbial species and strains was estimated using web applications AntiBac-Pred [49,50] and AntiFun-Pred [51]. These tools are based on PASS and provide, in addition to its capabilities, the novel bioactivity data and web interface, also they are free to use. AntiBac-Pred allows to evaluate chemical compounds against 353 bacterial strains, and AntiFun-Pred, against 38 fungi. The results of the prediction are provided in a similar manner to that used in PASS. However, instead of the Pa and Pi values, only their difference is provided. The higher the value, the higher the confidence that compound will show activity.
CLC-Pred [56,57] is another PASS-based web application, which allows to predict cytotoxicity for chemical compounds against tumor and non-tumor cell lines. This tool was used to assess the potential cytotoxic effect of the chemical compounds under study. The antibacterial assay was carried out by the microdilution method [57] in order to determine the antibacterial activity of compounds tested against the the above strains of human pathogenic bacteria. Compounds were diluted in DMSO, which was used as negative control (5%).

Biological Evaluation
The bacterial suspensions were adjusted with sterile saline to a concentration of 1.0 × 10 −5 cfu/mL. The innocula were prepared daily and stored at +4 • C until use. Dilutions of the innocula were cultured on solid medium to verify the absence of contamination and to check the validity of the inoculum [58,59].

Microdilution Test
The minimum inhibitory and bactericidal concentrations (MICs and MBCs) were determined using 96-well microtiter plates. The bacterial suspension was adjusted with sterile saline to a concentration of 1.0 × 10 −5 cfu/mL. Compounds to be investigated were dissolved in broth LB medium (100 µL) with bacterial inocula (1.0 × 10 −4 cfu per well) to achieve the wanted concentrations (1 mg/mL). The microplates were incubated for 24 h at 48 • C. The lowest concentrations without visible growth (under the binocular microscope) were defined as concentrations that completely inhibited bacterial growth (MICs). The compounds investigated were dissolved in 5% DMSO (1 mg/mL) and added in the LB medium to the inoculum. The MBCs were determined by serial sub-cultivation of 2 µL into microtiter plates containing 100 µL of broth per well and then submitted to further incubation for 72 h. The lowest concentration with no visible growth was defined as the MBC, indicating 99.5% killing of the original inoculum. The optical density of each well was measured at 655 nm by a Bio-Rad Laboratories Microplate Manager 4.0 and compared with a blank and the positive control. Streptomycin and ampicillin were used as positive controls (1 mg/mL) [58,59]. All experiments were performed in duplicate and repeated three times.
The micromycetes were maintained on malt agar and the cultures stored at 4 • C and sub-cultured once a month. In order to investigate the antifungal activity of the extracts, a modified microdilution technique was used [51][52][53]. The fungal spores were washed from the surface of agar plates with sterile 0.85% saline containing 0.1% Tween 80 (v/v). The spore suspension was adjusted with sterile saline to a concentration of approximately 1.0 × 10 −5 in a final volume of 100 µL per well. The innocula were stored at 4 • C for further use. Dilutions of the innocula were cultured on solid malt agar to verify the absence of contamination and to check the validity of the inoculum.
Minimum inhibitory concentration (MIC) determinations were performed by a serial dilution technique using 96-well microtiter plates. The compounds investigated were dissolved in 5% DMSO (1 mg/mL) and added in broth malt medium to the inoculum. The microplates were incubated for 72 h at 28 • C, respectively. The lowest concentrations without visible growth (under the binocular microscope) were defined as MICs.
The fungicidal concentrations (MFCs) were determined by serial subcultivation of a 2 mL into microtiter plates containing 100 µL of broth per well and then submitted to further incubation for 72 h at 28 • C.
The lowest concentration with no visible growth was defined as MFC, indicating 99.5% killing of the original inoculum. DMSO was used as a negative control; commercial fungicides, bifonazole and ketoconazole were used as positive controls (1-3000 mg/mL). All experiments were performed in duplicate and repeated three times.

Docking Studies
The AutoDock 4.2 ® (version 4.2.6, San Diego, California, CA, U.S.A) software was used for the docking simulation. The free energy of binding (∆G) of DNA topoisomerase IV, E. coli primase, E. coli DNA GyrB, E. coli MurB, Thymidylate kinase, Squalene synthase, Dihydrofolate reductase and CYP51 of C. albicans in complex with the inhibitors were generated using this molecular docking program. The X-ray crystal structures data of all the enzymes used were obtained from the Protein Data Bank (PDB ID: 1S16, 1DDE, AKZN, AQGG, 2Q85, 1EZF, 4HOF, and 5V5Z, respectively). All procedures were performed according to our previous papers [60].

Conclusions
Thirty-one compounds were studied for antimicrobial activity in silico using PASS software as well as freely available web-services AntiBAC Pred, MICF Pred, and CLC-Pred. PASS predicted antibacterial activity for 27 of 31 molecules, and antifungal activity was predicted for 25 of 31 compounds with relatively low probability. Such a result leads us to the suggestion that the analyzed compounds are structurally different from well-known antimicrobial agents. Therefore, the studied compounds may be active against the resistant strains. Prediction of antibacterial and antifungal action on particular microbial strains with AntiBAC Pred and MICF Pred web-services demonstrated that the compounds may exhibit rather broad spectra of antimicrobial activities. GUSAR predicted rather low general toxicity for all compounds. CLC-Pred provided estimates that allow selecting the compounds with low probability of cytotoxicity for further studies. Therefore, testing of the antimicrobial activity against different microbial species for compounds with low chance of general toxicity and cytotoxicity looks reasonable.
The evaluation of the antibacterial activity of the tested compounds revealed that these molecules exhibit a significant pharmacological potential, having higher in vitro potency than the approved antibacterial drugs: Ampicillin and Streptomycin. In particular, studied compounds were more active against the resistant bacterial E. coli, and P. aeruginosa strains as well as methicillin-resistant Staphylococcus aureus. It should be mentioned that in general Gram-positive bacteria are more sensitive to the tested compounds than Gram-negative bacteria.
The presence of 17α-amino-3α-hydroxy, as well as 17β-amino-3β-hydroxy groups in 5α-androstan core was found to be beneficial for antibacterial activity whereas the presence of 17β-tosyloxy-as well as 3α-and 3β-azido groups was detrimental on activity.
Compounds' antifungal effect (MIC at 0.007-0.45 mg/mL and MFC at 0.075-0.60 mg/mL) appeared to be superior to Ketoconazole and Bifonazole, which are widely used in clinical practice. The most sensitive fungi appeared to be T. viride, while P. cyclpoium var verucosum was the most resistant.
Despite that, all compounds exhibited good activity against all bacteria and fungi tested, their sensitivity towards compounds, in general, was different.
The molecular docking analysis indicated that the putative mechanism of antibacterial activity is probably the inhibition of the E. coli MurB enzyme.
Docking analysis to 14α-lanosterol demethylase (CYP51) and tetrahydrofolate reductase of Candida albicans indicated a probable implication of CYP51 reductase in the anti-fungal activity of the compounds.