Introduction

β-Amino acids are promising intermediates in organic, bioorganic, medical, and peptide chemistry. Many compounds with structural fragments of β-amino acids—peptides, coenzymes, antibiotics, and alkaloids—are found in nature. A group of very important β-amino acid derivatives are the β-lactams, which are present in antibiotics, human leukocyte elastase inhibitors, and cholesterol uptake inhibitors. The anticancer compound taxol also has a structural component of β-alanine. An extensive description of the synthesis and application of β-alanines is presented in [1]. Likewise, N-substituted β-amino acids and their derivatives are important and have been widely studied for many different purposes, exhibiting biological activity, e.g., antibacterial [2] and antifungal [3, 4], as well as growth stimulation of agricultural crops [5, 6], antagonism of lysophosphatidic acid (EDG-2 receptor) [7], etc. N-Substituted β-amino acids are also excellent synthons for synthesis of azetidine [8–10], pyridinone [11–13], dihydrouracil [12, 14, 15], and quinolinone [16–18] heterocyclic systems. Modification of β-amino acids as potential substances for organic synthesis is presently a prevailing trend. One of the ways to modify N-substituted β-amino acids is to use carbohydrazides of N-substituted β-amino acids in synthesis of heterocyclic compounds [19–21]. On the other hand, many azole derivative compounds have shown a variety of different biological activities, including antibacterial and antifungal activity [22].

The aim of this work is to synthesis and evaluate in silico new potentially biologically active azoles from N-(4-iodophenyl)-β-alanine hydrazide, to establish their biological potential.

Nowadays, computer-aided methods are widely used in drug discovery, particularly for lead finding and optimization [23]. Target-based methods are mostly applied to find compounds inhibiting a particular endogenous macromolecule, whose three-dimensional structure is determined by X-ray analysis or nuclear magnetic resonance (NMR) [24]. Thus, such methods are focused on a particular target and are less applicable for evaluation of the biological potential of drug-like organic compounds. In contrast to this, ligand-based approaches provide information about the general biological activity profiles of drug-like compounds [25], either based on chemical similarity assessment or on machine learning methods. Thus, for the purpose of our study, we selected ligand-based methods, in particular ChemSpider, chemical structure lookup service (CSLS), Integrity, SuperPred, and PASS (see below for description of “Computational studies”).

Results and discussion

Chemistry

In continuation of our interest in the chemistry of N-substituted β-amino acids, N-(4-iodophenyl)-β-alanine hydrazide 1 was synthesized by reaction of N-(4-iodophenyl)-β-alanine and hydrazine hydrate in toluene under reflux. The hydrazones 2–4 were synthesized in good yields (89–97 %) by reacting N-(4-iodophenyl)-β-alanine hydrazide 1 with aromatic aldehydes in methanol at reflux (Scheme 1). The structure of the hydrazones 2–4 was established mainly on the basis of 1H and 13C NMR spectra. Theoretically, this compound, possessing amide and azomethine groups, can exist as an inseparable mixture of four isomers. The amide group determined the splitting of resonances in the 1H and 13C NMR spectra due to the restricted rotation around the amide bond. The 1H and 13C NMR spectral data allowed us to conclude that the geometrical isomers of the azomethine group were not observed and hydrazones 2–4c in dimethyl sulfoxide (DMSO)-d6 solution existed as a mixture of E/Z isomers, where the Z isomer predominated due to hindered rotation around the CO–NH bond.

Scheme 1
scheme 1

Synthesis of new N-(4-iodophenyl)-β-alanine derivatives 2–13

Full size image

Azole derivatives can be obtained by interaction of carbohydrazide and diketones. Dimethylpyrrole derivative 5 was prepared by refluxing a mixture of the corresponding carbohydrazide 1, 2,5-hexanedione, 2-propanol, and a catalytic amount of glacial acetic acid (Scheme 1). 3,5-Dimethylpyrazole derivative 6 was prepared in good yield by refluxing a mixture of carbohydrazide 1, 2,4-pentanedione, 2-propanol, and a catalytic amount of hydrochloric acid (Scheme 1). The resonances at 2.18 and 2.47 ppm (CH3 groups), and 6.18 ppm (C=CH=C group) in the 1H NMR spectra confirmed the formation of pyrazole rings. The resonances at 13.4 and 14.0 ppm of CH3 groups in the 13C NMR spectra of molecule 6 and the signals at 111.1 ppm, which were attributed to the CH, proved the presence of a five-membered heterocycle. The resonances observed at 143.1 and 148.1 ppm were ascribed to the carbons of the N–C and N=C groups of the heterocycle.

In this work, 1,3,4-oxadiazole 8 was prepared by refluxing carbohydrazide 1, carbon disulfide, and potassium hydroxide in 2-propanol, followed by dissolution of the intermediate potassium dithiocarbazate 7 in water and treatment of the obtained solution with acetic acid to pH 6. The presence of broad singlets at 14.20 ppm of NH proton in 1H NMR spectra and two resonances at 162.5 and 177.7 ppm, attributed to the N=C and C=S groups, respectively, in 13C NMR spectra confirmed the formation of a five-membered oxadiazole ring in compound 8 (Scheme 1).

The 4-amino-1,2,4-triazole derivative 9 was prepared by heating intermediate dithiocarbazate 7 with hydrazine in water solution under reflux, and then the reaction mixture was acidified with acetic acid to pH 6. Compound 9 was also prepared by heating 1,3,4-oxadiazole 8 with hydrazine in 2-propanol solution under reflux, and then the reaction mixture was acidified with acetic acid to pH 6, but the reaction yield was low (24 %). The characteristic resonances at 13.51 ppm (NH) in the 1H NMR spectrum and at 150.3 ppm (CH2 C=) and 165.8 ppm (C=S) proved the formation of a 1,2,4-triazole ring in compound 9. The presence of a NH2 group was confirmed by the resonance multiplet at about 5.00–5.25 ppm in the 1H NMR spectrum.

The starting compounds for the synthesis of phenyl-substituted triazoles (Scheme 1)—N-phenylhydrazinecarboxamide 10 and N-phenylhydrazinecarbothioamide 11—were synthesized from hydrazides 1 with phenyliso- and phenylisothiocyanates in methanol. Triazoles 12, 13 were obtained by refluxing compounds 10, 11 in aqueous 2 % NaOH for 3 h with subsequent acidification of the reaction mixture with acetic acid. The sharp singlet at 11.72 ppm, assigned to the proton of the NH group, in the 1H NMR spectrum, and resonances at 145.0, 154.3 ppm were assigned to C=N, C=O carbon atoms in the 13C NMR spectrum, proving the formation of a five-membered heterocycle in compound 12. The resonances at 150.3 ppm (C=N) and at 167.5 ppm (C=S) in the 13C NMR spectrum proved the formation of a thioanalogous five-membered heterocycle in compound 13.

Synthesis of compounds 15–23 (Scheme 2) was carried out in earlier work [26].

Scheme 2
scheme 2

Synthesis of new bis-N-(4-iodophenyl)-β-alanine derivatives 15–23

Full size image

Computational studies

To identify the most promising directions for experimental research into the synthesized compounds, we decided to use an in silico approach. There exist several tools that may provide estimation of probable biological activity for drug-like compounds based on assessment of structural similarity or machine learning methods. All these tools require the structural formulae as input information, which should be presented as MOL files (for a single compound) or as an SDF file (for the set including all 20 compounds). We prepared such files using the ISIS Draw and ISIS Base programs [27].

ChemSpider

First, we tried to determine whether the synthesized compounds corresponded to known chemicals available via ChemSpider [28]. ChemSpider is a freely available chemical database providing structure search access to over 29 million structures from hundreds of data sources. No identical structure was found for any of the 20 synthesized compounds (see, e.g., the example search for compound 5 in Fig. 1).

Fig. 1
figure 1

Example search for compound 5 using ChemSpider

Full size image

Chemical structure lookup service

Second, we tried to find the synthesized compounds among the known chemicals in different chemical databases using the CSLS [29]. The CSLS search engine provides the retrieval of 74 million indexed structures (46 million unique structures) from over 100 chemical databases (PubChem, NIST, EPA, etc.). As one may see from Fig. 2, no identical structure was found for any of the 20 synthesized compounds as well. Therefore, all synthesized compounds do not have any equivalent structure among the 46 million unique structures in the different databases available through CSLS.

Fig. 2
figure 2

Results of search for 20 structures using CSLS service

Full size image

Therefore, structure searches using both ChemSpider and CSLS did not find any identical structures among the dozens of millions of structures in over 100 different databases. This result provides additional evidence regarding the relative novelty of the structures synthesized in this work.

Integrity

Then, we tried to find similar structures within the Integrity database (Thomson Reuters) [30]. Integrity is a commercially available Internet portal providing information about drugs and pharmacological substances that are currently under different stages of research and development as potential human medicines. It contains information about ca. 330,000 chemical structures. In addition to the exact structure search, which gave no results for any of the 20 synthesized structures, in Integrity it is possible to perform a search based on structure similarity. If some structures that have known biological activity and that are similar to the synthesized compounds could be identified, it is reasonable to suggest that our 20 compounds may have the same biological activities.

Similarity searching finds molecules that are similar to a query structure. The calculation applies the Tanimoto coefficient (TC), which has two arguments: (1) the fingerprint (a bit string that contains structural information on the molecule) of the query structure, and (2) the fingerprint of the molecule in the Integrity database. The TC is calculated by the following formula:

$${\text{TC}} = N_{\text{AB}} /\left( {N_{\text{A}} + \, N_{\text{B}} - N_{\text{AB}} } \right),$$

where N A and N B are the number of bits set for molecule A and B, respectively, and N AB is the number of bits common to the two molecules.

The similarity threshold specifies a lower limit for the TC. If a TC is greater than this threshold, the query structure and the given structure from the database are considered to be similar. In our study, we used 70 % similarity as a lower threshold, which can still provide a reasonable basis for the inductive conclusion that such structures would have similar biological activities.

Figure 3 presents an example query for compound 12.

Fig. 3
figure 3

Example similarity search using the structure of compound 12 as a query (threshold TC > 70 %)

Full size image

The result of the similarity search for compound 12 is given in Fig. 4.

Fig. 4
figure 4

Result of similarity search using the structure of compound 12 as a query (threshold TC > 70 %)

Full size image

As one can see from Fig. 4, the only compound with greater than 70 % similarity to the structure of 12 is the compound with ID # 443457 from the Integrity database. This compound was studied by the Birla Institute of Technology and Science as an agent for treatment of neuropathic pain and as a potential antiepileptic drug.

For 16 of the 20 synthesized compounds, no similar structures were found using the threshold of TS > 70 %. The results for the other four compounds are given in Table 1.

Table 1 Results of similarity search for compounds 12, 13, 22, and 23 (threshold TC > 70 %)
Full size table

From the results presented in Table 1, it is clear that, for these 4 of the20 synthesized compounds, only five similar structures were found in the Integrity database when using the TC > 70 % threshold. One may suggest that biological activities known for these five structures, which are presented in Table 1, could be found in the query structures. However, it is necessary to bear in mind the result obtained by Yvonne Martin and coworkers [31] that there is only a 30 % chance that structures similar at the TS > 85 % threshold will exhibit similar biological activities.

SuperPred

SuperPred is a freely available web tool that provides estimation of biological activity based on chemical similarity [32, 33]. This tool calculates the TC values for the similarity between two compounds using their structural fingerprints as generated by the Chemistry Development Kit [34]. Biological activities are described here in terms of the Anatomical Therapeutic Chemical (ATC) classification and using a list of human proteins that are considered as pharmacological targets. Estimation of chemical similarity for each of the synthesized 20 compounds using SuperPred did not give TC > 70 % for any pairs of compounds (see data presented in Electronic Supplementary Material 1). The highest TC value, calculated for the similarity between compound 14 and 3-[acetyl(3-amino-2,4,6-triiodophenyl)amino]-2-methylpropanoic acid, was about 65 %; however, no biological activity was predicted in this case. Bearing in mind the above-mentioned poor accuracy of biological activity prediction based on chemical similarity [31], one may conclude that SuperPred does not provide any useful hints about the biological activity of the 20 synthesized compounds.

PASS

Since our attempts to identify the same or similar structures with known biological activities among dozens of millions of chemical substances from several Internet resources did not help in determination of the most promising directions for testing the biological activity of our 20 synthesized compounds, we decided to predict their biological activity using the PASS computer program [35, 36]. The prediction of activity spectra for substances (PASS) estimates the probabilities of belonging to the classes of “actives” (Pa) and “inactives” (Pi) for over 6,000 biological activities on the basis of analysis of structure–activity relationships for a training set including more than 300,000 biologically active compounds. The machine learning method implemented in PASS is based on a Bayesian algorithm [36] and multilevel neighborhoods of atoms (MNA) descriptors [37]. The average accuracy of PASS prediction estimated by a leave-one-out cross-validation procedure for the whole training set and all predictable biological activities is about 95 %; similar values were obtained in 20-fold cross-validation. The freely available web resource PASS Online [35] has been available since 2000 [38–40], and more than 50 papers with independent confirmation of the prediction results for various biological activities and different chemical classes have been published [35, 40].

We performed PASS Online predictions of the pharmacotherapeutic effects and mechanisms of action for the 20 synthesized compounds, and then analyzed the obtained results using the PharmaExpert computer program [41]. A report on the prediction results for all the compounds as generated by PharmaExpert is presented in Electronic Supplementary Material 2. Statistics on the number of compounds for which a particular activity is predicted with Pa > 50 % for the top 40 activities are given in Table 2.

Table 2 Statistics of number of compounds with activities predicted with Pa > 50 %
Full size table

As one can see from the data presented in Table 2, complement factor 1s inhibiting activity is predicted for all 20 compounds; DNA ligase 1 inhibiting and protein disulfide-isomerase inhibiting activities are predicted for 8 of the 20 compounds; Epstein–Barr nuclear antigen 1 inhibiting and NACHT, LRR, and PYD domains-containing protein 3 inhibiting activities are predicted for 6 of the 20 compounds; PfA-M1 aminopeptidase inhibiting and eukaryotic translation initiation factor 4H inhibiting activities are predicted for 5 of the 20 compounds; etc. These activities can be considered as typical for the chemical series under consideration because they are predicted for an essential fraction of the whole set of compounds [42]. In contrast, transcription factor STAT1 inhibiting, botulin neurotoxin A light chain inhibiting, manganese peroxidase inhibiting, trimethylamine dehydrogenase inhibiting, HIV rev inhibiting, etc. activities are predicted just for one or two derivatives, and they can be considered as minor for this chemical series [39]. Spermidine dehydrogenase inhibiting, neuropeptide Y2 antagonistic, and glucose-6-phosphate dehydrogenase-6-phosphogluconolactonase inhibiting activities are predicted with probabilities Pa < 50 %, and the probability to find the activity in experiments is rather small.

It is necessary to stress that, at the threshold Pa > 50 %, neither antibacterial nor antifungal activity is predicted by PASS, which corresponds to the negative results of experimental testing of our 20 compounds on Escherichia coli B-906, Mycobacterium luteum B-917, Candida tenuis VKM Y-70, and Aspergillus niger F-1119 (see below).

These PASS predictions provide hints about priorities on biological activities which should be tested for each synthesized derivative, to increase the chances of finding biological activity in experiments. Let us consider, for example, the prediction results obtained with Pa > 50 % for compound 20:

figure a

Activity that is predicted with the highest probability, so-called focal activity, should be tested first. For compound 20 the focal activity is as a complement factor 1s inhibitor. Then, one should test the activities that have the second, third, etc. ranks in the predicted activity spectra: polyadenylate-binding protein 1 inhibitor, transcription factor STAT1 inhibitor, analgesic, and protein disulfide-isomerase inhibitor.

There are only four different focal activities for all 20 compounds: complement factor 1s inhibitor (compounds 5–13, 18–23), DNA ligase 1 inhibitor (compounds 2–4, 15–17), beta-lysine 5,6-aminomutase inhibitor (compound 1), and taurine dehydrogenase inhibitor (compound 14). It is clear that the majority of the focal activities (e.g., complement factor 1s inhibitor) correspond to typical activities in this chemical series; however, some of the focal activities (e.g., beta-lysine 5,6-aminomutase inhibitor) correspond to minor activities in this chemical series.

Analysis of the statistics of activities predicted at Pa > 50 % shows that even 20 N-(4-iodophenyl)-β-alanine derivatives, N-(4-iodophenyl)-N-carboxyethyl-β-alanine derivatives, and their cyclization products demonstrate wide pharmacological potential. Let us consider this potential in more detail.

A complement, as an essential component of the immune system, is of substantial relevance for destruction of invading microorganisms and for maintaining tissue homeostasis, including protection against autoimmune diseases [43]. Complement activation is tightly controlled by its regulators to prevent complement-mediated self-tissue injury [44]. It has been well established that excessive complement activation contributes to the pathogenesis of many autoimmune diseases, including rheumatoid arthritis [45, 46], type I diabetes [47], myasthenia gravis [48], glomerulonephritis [49], as well as transplanted graft rejection [50]. Thus, to focus on complement inhibition appears to be a logical approach to stop the process of inflammatory disorders. Compounds with predictive complement factor 1s inhibiting activity include all 20 synthesized compounds.

A hallmark of many neurodegenerative diseases, including Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis, and Huntington’s disease, is accumulation of misfolded proteins within neurons, leading to cellular dysfunction and cell death. Comprehensive understanding of how the protein disulfide-isomerase cell death pathway is controlled may aid in discovering therapeutics to treat protein misfolding diseases [51].

DNA reparation (DNA ligase I) inhibitors have potential to concomitantly increase the killing of cancer cells and reduce damage to normal tissues and cells if either the damaging agent or the inhibitor can be selectively delivered to cancer cells. A problem with the screening of random chemical libraries for DNA ligase inhibitors is that many of the hits are likely to be nonspecific inhibitors that either bind to the DNA substrate or are nucleotide analogs that inhibit a large number of adenosine triphosphate (ATP)-dependent enzymes [52].

PfA-M1 aminopeptidase localizes differently in trophozoites and schizonts, suggesting two different roles in critical steps of the intraerythrocytic life of the parasite [53]. Proteases that are expressed during the erythocytic stage of Plasmodium falciparum are newly explored drug targets for treatment of malaria [54]. Some of our compounds are predicted to be potential inhibitors of PfA-M1 aminopeptidase.

Substances for which peptide agonist activity was predicted are probably resistant to proteolytic attack, and may have prolonged activity as peptide-based drugs used to treat corneal ulceration, emphysema, rheumatoid arthritis, peptic ulcers, hypertension, and multiple sclerosis.

Ubiquinol cytochrome C oxidoreductase (bc1 complex) is an essential component of the electron transport systems critical for respiration in prokaryotes and eukaryotes, and photosynthesis in some bacteria [55, 56]. In fact, in many species, ubiquinol-cytochrome C oxidoreductase activity was found to be sensitive to quinone analogs such as myxothiazol, stigmatellin, antimycin, hydroxyquinoline N-oxide (HQNO), and diuron [57]. A variety of inhibitors of the bc1 complex have been important in elucidating the Q cycle hypothesis and, together with recent structural and genetic experiments, have also been useful in mapping the features of the Qo and Qi domains and their relationship to the reaction mechanism [58]. The inhibitors fall into two main categories: those that bind at the Qi site (class II), such as the antibiotic antimycin, and those that bind to the Qo site, such as strobilurin A (class I) [58]. The obtained information [59] may be useful for understanding the function of the bc1 complex, and for design of novel antiinfective agents for use as human remedies and in agricultural applications.

Comparison of the obtained Pa data for our 20 compounds shows that the Pa values for monosubstituted derivatives exceed those for disubstituted derivatives; also, wider predicted biological activity spectra are observed in monosubstituted derivatives.

Neither antibacterial nor antifungal activities were predicted by PASS for compounds under study with Pa > 0.5. However, since such activities are known for compounds from this chemical class, we decided to study their antibacterial and antifungal activities in vitro. The experimental results should shed light on the antimicrobial action of our compounds, and also act as a negative control for computer-aided predictions.

Biology

Compounds 1–6, 8, 9, and 12–23 were selected to identify the influence of azole fragments in the case of mono- and disubstituted derivatives of N-(4-iodophenyl)-β-alanine (substances 10 and 11 were not chosen for antimicrobial study because they were used only as intermediate compounds in synthesis of triazoles 12, 13). These substances were evaluated for their antibacterial and antifungal activity against strains of E. coli B-906, Staphylococcus aureus 209-P, M. luteum B-917 (as a nonpathogenic test bacterial culture representative of the genus Mycobacterium), C. tenuis VKM Y-70, and A. niger F-1119 by the diffusion technique [60] and serial dilution technique [61] (determination of minimal inhibition concentration, MIC) to correlate the in silico investigation with experimental data for the new potentially biologically active compounds based on N-(4-iodophenyl)-β-alanine hydrazides. It was established that all tested compounds showed no antibacterial or fungicidal action at concentrations of 0.5 and 0.1 % (diffusion technique) against strains E. coli, S. aureus, M. luteum, C. tenuis, and A. niger. No influence of the studied compounds on the growth of microorganisms was observed at concentrations of 3.9–500 mkg/ml (serial dilution technique). However, in the same experiment, antimicrobial activity was found for the reference antibacterial and antifungal drugs (vancomycin and nystatin at concentration of 0.5 and 0.1 %, respectively).

Conclusions

In silico studies of synthesized new potentially biologically active N-(4-iodophenyl)-β-alanine derivatives, N-(4-iodophenyl)-N-carboxyethyl-β-alanine derivatives, and their cyclization products allowed their biological potential to be established. Application of several freely available and commercial tools (ChemSpider and CSLS) showed the relative novelty of the synthesized derivatives; however, similarity searches (Integrity and SuperPred) did not provide many fruitful ideas regarding which kinds of biological activity of the synthesized compounds should be tested with high priority. This finding corresponds to earlier conclusions that there is a rather small chance of revealing the same activity in pairs of compounds with TC exceeding 85 % [30].

In contrast to the similarity assessment, machine learning methods implemented in the PASS program predicted probability to be active Pa > 50 % for 37 different biological activities, including typical activities for this chemical series (e.g., complement factor 1s inhibitor) and minor activities for this chemical series (e.g., taurine dehydrogenase inhibitor). To achieve the maximum efficiency in discovery of new biologically active compounds, it is necessary to study the biological activity of each compound starting from those predicted with the highest probability (so-called focal activity), continuing in descending order of probability to be active (Pa).

However, the predicted activities fall into very different pharmacological categories and may have various mechanisms of action. Therefore, it is rather difficult for an academic team to synthesize new chemical compounds with great pharmacological potential according to computational predictions to check their diverse biological activities. Based on the current study, we can only conclude that antibacterial and antifungal activities were not predicted by PASS with Pa > 50 %, and that these activities were not found in in vitro assays.

The essential role of academic studies in drug design and discovery is well established [62–65]. Being more flexible in their research topics, people working in academia often generate new ideas that may be translated into practice through further research and development. However, the past experience of many organic chemists shows that the main bottleneck in this process is the limited facilities for assaying biological activities. Creating new academic drug R & D centers, as observed during recent years [62], may significantly improve this situation. Collaboration through multidisciplinary teams and wide application of computational methods represent the current mainstream of new drug discovery projects [66].

Experimental

Chemistry

1H and 13C NMR spectra were recorded on a Varian Unity Inova (300, 75 MHz) spectrometer operating in Fourier-transform mode, using DMSO-d6 as solvent and tetramethylsilane (TMS) as internal reference (chemical shifts in δ, ppm). Infrared (IR) spectra (ν, cm−1) were recorded on a PerkinElmer Spectrum BX FT–IR spectrometer using KBr tablets. Elemental analyses were performed with a CE-440 elemental analyzer. Melting points were determined with an automatic APA1 melting point apparatus and are uncorrected. Thin-layer chromatography (TLC) was performed with Merck silica gel 60 F254 (Kieselgel 60 F254) silica gel plates.

General procedure for synthesis of hydrazones 2–4

A mixture of hydrazide 1 (0.92 g, 3 mmol) and 4 mmol of the corresponding aromatic aldehyde in 30 ml 1,4-dioxane was heated under reflux for 3 h. The reaction mixture was cooled down, the precipitate filtered, then washed with 2-propanol and crystallized from methanol to give hydrazones 2–4.

N′-[(Z/E)-(4-Fluorophenyl)methylidene]-3-[(4-iodophenyl)amino]propanehydrazide (2)

White solid (1.16 g, 94 %), m.p. 169–170 °C (methanol). IR, ν: 1,231 (ArNH), 1,599 (N=C), 1,660 (CO), 3,173 (CONH), 3,354 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.43–2.50 (0.4 (2H), m, CH2CO), 2.88 (0.6 (2H), t, J = 7.0 Hz, CH2CO), 3.26–3.36 (2H, m, CH 2NH), 5.92 (1H, t, J = 5.8 Hz, CH2 NH), 6.42–7.78 (8H, m, Har), 7.98 (0.6 (1H), s, N=CH), 8.15 (0.4 (1H), s, N=CH), 11.36, 11.41 (1H, 2s, CONH). 13C NMR (75.4 MHz, DMSO-d6): δ = 31.7, 33.8 (CH 2CO), 38.2, 38.8 (CH2NH), 75.8, 75.9, 114.5, 114.6, 115.6, 115.9, 128.6, 128.7, 129.0, 129.1, 130.7, 130.7, 130.9, 137.0, 141.6, 144.8, 148.1, 148.2, 161.1, 164.4 (C-ar), 167.0, 172.8 (CONH). Found, %: C 46.53, H 3.54, N 10.38. Calculated, %: C 46.73, H 3.68, N 10.22. C16H15FIN3O.

N′-[(Z/E)-(4-Chlorophenyl)methylidene]-3-[(4-iodophenyl)amino]propanehydrazide (3)

White solid (1.24 g, 97 %), m.p. 182–183 °C (methanol). IR, ν: 1,264 (ArNH), 1,596 (N=C), 1,664 (CO), 3,168 (CONH), 3,348 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.43–2.50 (0.4 (2H), m, CH2CO), 2.89 (0.6(2H), t, J = 7.0 Hz, CH2CO), 3.31 (2H, dd, 1 J = 6.7 Hz, 2 J = 12.7 Hz, CH 2NH), 5.92 (1H, t, J = 5.4 Hz, CH2 NH), 6.41–7.73 (8H, m, Har), 7.98 (0.6 (1H), s, N=CH), 8.14 (0.4 (1H), s, N=CH), 11.41, 11.48 (1H, 2s, CONH). 13C NMR (75.4 MHz, DMSO-d6): δ = 31.6, 33.8 (CH 2CO), 38.2, 38.7 (CH2NH), 75.8, 114.6, 114.6, 128.2, 128.5, 128.8, 133.1, 133.2, 134.0, 134.3, 137.1, 141.4, 144.6, 148.1, 148.2 (C-ar), 167.1, 172.9 (CONH). Found, %: C 44.81, H 3.83, N 9.98. Calculated, %: C 44.93, H 3.54, N 9.83. C16H15ClIN3O.

N′-[(Z/E)-(4-Bromophenyl)methylidene]-3-[(4-iodophenyl)amino]propanehydrazide (4)

White solid (1.36 g, 96 %), m.p. 181–182 °C (methanol). IR, ν: 1,288 (ArNH), 1,598 (N=C), 1,659 (CO), 3,167 (CONH), 3,334 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.42–2.50 (0.4 (2H), m, CH2CO), 2.88 (0.6 (2H), t, J = 7.0 Hz, CH2CO), 3.31 (2H, q, J = 6.8 Hz, CH 2NH), 5.92 (1H, t, J = 5.8 Hz, CH2 NH), 6.41–7.65 (8H, m, Har), 7.95 (0.6 (1H), s, N=CH), 8.12 (0.4 (1H), s, N=CH), 11.41, 11.47 (1H, 2s, CONH). 13C NMR (75.4 MHz, DMSO-d6): 31.6, 33.8 (CH 2CO), 38.2, 38.7 (CH2NH), 75.8, 114.6, 114.6, 122.8, 128.4, 128.7, 131.7, 133.4, 133.5, 137.1, 141.5, 144.6, 148.1, 148.2 (C-ar), 167.1, 172.1 (CONH). Found, %: C 40.64, H 3.42, N 9.05. Calculated, %: C 40.70, H 3.20, N 8.90. C16H15BrIN3O.

N-(2,5-Dimethyl-1H-pyrrol-1-yl)-3-[(4-iodophenyl)amino]propanamide (5)

A mixture of hydrazide 1 (0.92 g, 3 mmol), 2,5-hexanedione (0.57 g, 5 mmol), 2-propanol (10 ml), and acetic acid (0.5 ml) was heated under reflux for 6 h, then cooled, the precipitate was filtered off, washed with 2-propanol and crystallized from 2-propanol to give dimethylpyrrole derivative 5.

White solid (0.80 g, 70 %), m.p. 136–137 °C (2-propanol). IR, ν: 1,262 (ArNH), 1,660 (CO), 3,292 (CONH), 3,345 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 1.95, 2.00 (6H, 2s, CH3), 2.52 (2H, t, J = 6.6 Hz, CH2CO), 3.30–3.39 (2H, m, CH 2NH), 5.61 (0.9 (4H), s, CH–CH), 5.70 (0.1 (4H), s, CH–CH), 5.95 (1H, t, J = 5.9 Hz, CH2 NH), 6.34 (0.2 (2H), d, J = 8.9 Hz, Har-2.6), 6.47 (0.8 (2H), d, J = 8.9 Hz, Har-2.6), 7.29 (0.2 (2H), d, J = 8.8 Hz, Har-3.5), 7.35 (0.8 (2H), d, J = 8.8 Hz, Har-3.5), 10.17 (0.1 (2H), 2s, NH), 10.60 (0.9 (2H), 2s, NH). 13C NMR (75.4 MHz, DMSO-d6): δ = 10.9 (2CH3), 32.8 (CH 2 CO), 38.8 (CH2NH), 76.1 (C-4), 102.8 (CH–CH), 114.7 (C-2.6), 126.7 (N–CCH3), 137.1 (C-3.5), 148.0 (C-1), 170.1 (CO). Found, %: C 47.15, H 4.72, N 10.80. Calculated, %: C 47.01, H 4.73, N 10.96. C15H18IN3O.

1-(3,5-Dimethyl-1H-pyrazol-1-yl)-3-[(4-iodophenyl)amino]propan-1-one (6)

A mixture of hydrazide 1 (0.92 g, 3 mmol), 2,4-pentanedione (0.5 g, 5 mmol), 2-propanol (10 ml), and conc. hydrochloric acid (0.5 ml) was heated under reflux for 7 h, then cooled, the precipitate was filtered off and washed with water. Compound 6 was crystallized from a mixture of 2-propanol and water (2:1).

White solid (0.84 g, 76 %), m.p. 93–94 °C (2-propanol). IR, ν: 1,257 (ArNH), 1,311, 1,324, 1,361, 1,378 (CH3), 1,594 (N=C), 1,728 (CO), 3,323 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.18 (3H, s, CH3), 2.47 (3H, s, CH3), 3.15–3.45 (4H, m, NHCH 2 CH 2), 5.98 (1H, t, J = 4.6 Hz, CH2 NH), 6.18 (1H, s, CH), 6.46 (2H, d, J = 8.5 Hz, Har-2.6), 7.35 (2H, d, J = 8.5 Hz, Har-3.5). 13C NMR (75.4 MHz, DMSO-d6): δ = 13.4 (CH=CCH 3 ), 14.0 (CH-CCH 3 ), 34.5 (CH 2 CO), 37.9 (CH2NH), 76.0 (C-4), 111.1 (CH=C), 114.6 (C-2.6), 137.1 (C-3.5), 143.1 (CH=CCH3), 148.1 (C-1), 151.3 (CH-CCH3), 171.7 (CO). Found, %: C 45.38, H 4.39, N 11.23. Calculated, %: C 45.54, H 4.37, N 11.38. C14H16IN3O.

5-{2-[(4-Iodophenyl)amino]ethyl}-1,3,4-oxadiazole-2(3H)-thione (8)

A solution of hydrazide 1 (1.53 g, 5 mmol), potassium hydroxide (0.84 g, 15 mmol), carbon disulfide (0.76 g, 10 mmol), and methanol (20 ml) was heated under reflux for 24 h, then volatile fractions were evaporated under reduced pressure. The residue was dissolved in water (30 ml), and the solution was acidified with acetic acid to pH 6. The formed residue was filtered off, washed with water, and dried. Compound 8 was crystallized from a mixture of 2-propanol and water (2:1).

Light-yellow solid (1.44 g, 83 %), m.p. 111–112 °C (2-propanol). IR, ν: 1,172 (C=S), 1,261 (ArNH), 1,592 (N=C), 3,101 (NNH), 3,415 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.93 (2H, t, J = 6.6 Hz; CH2C=), 3.30–3.40 (2H, m, CH 2NH), 6.02 (1H, br. s. CH2 NH), 6.44 (2H, d, J = 8.8 Hz; Har-2.6), 7.34 (2H, d, J = 8.8 Hz; Har-3.5), 14.20 (1H, br. s, NNH). 13C NMR (75.4 MHz, DMSO-d6): δ = 25.0 (CH 2C=), 38.8 (CH2NH), 76.5 (C-4), 114.7 (C-2.6), 137.1 (C-3.5), 147.7 (C-1), 162.5 (OC=N), 177.7 (C=S). Found, %: C 34.23, H 3.99, N 12.16. Calculated, %: C 34.60, H 2.90, N 12.10. C10H10IN3OS.

4-Amino-5-{2-[(4-iodophenyl)amino]ethyl}-2,4-dihydro-3H-1,2,4-triazole-3-thione (9)

A mixture of hydrazide 1 (1.53 g, 5 mmol), potassium hydroxide (0.85 g, 15 mmol), carbon disulfide (0.76 g, 10 mol), and methanol (20 ml) was heated under reflux for 24 h, then volatile fractions were evaporated under reduced pressure; the residue was dissolved in water (5 ml), and hydrazine hydrate (1.00 g, 20 mmol) was added. The mixture was heated under reflux for 30 h, diluted with 10 ml of water, cooled, and acidified with acetic acid to pH 6. The formed residue was filtered off, washed with water, and compound 9 was crystallized from 2-propanol.

White solid (1.07 g, 59 %), m.p. 137–138 °C (2-propanol). IR, ν: 1,182 (C=S), 1,238 (ArNH), 1,590 (N=C), 3,171 (NNH), 3,271 (NNH2), 3,395 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.86 (2H, t, J = 7.1 Hz, CH2C=), 3.36 (2H, t, J = 7.1 Hz, CH 2NH), 5.00–6.25 (3H, br. s, NH2, CH2 NH), 6.46 (2H, d, J = 8.5 Hz, Har-2.6), 7.34 (2H, d, J = 8.4 Hz, Har-3.5), 13.51 (1H, s, NNH) ppm. 13C NMR (75.4 MHz, [D6]DMSO): δ = 24.3 (CH 2C=), 39.3 (CH2NH), 76.3 (C-4), 114.7 (C-2.6), 137.1 (C-3.5), 147.8 (C-1), 150.3 (CH2 C=), 165.8 (C=S). Found, %: C 33.45, H 3.48, N 19.18. Calculated, %: C 33.25, H 3.35, N 19.39. C10H12IN5S.

2-{3-[(4-Iodophenyl)amino]propanoyl}-N-phenylhydrazinecarboxamide (10)

A mixture of hydrazide 1 (1.53 g, 5 mmol), phenyl isocyanate (0.71 g, 6 mmol), and 30 ml methanol was heated under reflux for 2 h. The reaction mixture was cooled, the precipitate was filtered off, then washed with methanol, and compound 10 was crystallized from methanol.

White solid (0.92 g, 57 %), m.p. 157–158 °C (methanol). IR, ν: 1,230 (ArNH), 1,604–1,641 (C=O), 3,313–3,374 (NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.42 (2H, t, J = 6.9 Hz, CH2CO), 3.25 (2H, dd, 1 J = 6.8 Hz, 2 J = 13.0 Hz, CH 2NH), 5.86 (1H, t, J = 5.6 Hz, CH2 NH), 6.40–7.50 (9H, m, Har), 8.04, 8.37, 8.71, 8.90, 9.01, 9.74 (3H, 6s, CONH). 13C NMR (75.4 MHz, DMSO-d6): δ = 32.9 (CH 2CO), 39.7 (CH2N), 76.1, 114.7, 118.4, 121.8, 137.1, 139.5, 148.1 (C-ar), 155.3 (NHCONH), 170.6 (CH2 CONH). Found, %: C 45.19, H 3.92, N 13.28. Calculated, %: C 45.30, H 4.04, N 13.21. C16H17IN4O2.

2-{3-[(4-Iodophenyl)amino]propanoyl}-N-phenylhydrazinecarbothioamide (11)

A mixture of hydrazide 1 (1.53 g, 5 mmol), phenyl isothiocyanate (0.81 g, 6 mmol), and 50 ml methanol was heated under reflux for 3 h. The reaction mixture was cooled, the precipitate filtered off, then washed with methanol, and compound 11 was crystallized from methanol.

White solid (2.05 g, 93 %), m.p. 155–156 °C (methanol). IR, ν: 1,172 (C=S), 1,229 (ArNH), 1,682 (C=O), 3,064, 3,133, 3,272 (NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.43–2.50 (2H, m, CH2CO), 3.27 (2H, dd, 1 J = 6.7 Hz, 2 J = 12.6 Hz, CH2NH), 5.85 (1H, t, J = 5.4 Hz, CH2 NH), 6.40–7.55 (9H, m, Har), 9.57 (2H, s, NHCSNH), 9.98 (1H, s, CONH). 13C NMR (75.4 MHz, DMSO-d6): δ = 33.0 (CH 2 CO), 38.6 (CH2NH), 76.2, 114.7, 125.0, 125.8, 128.0, 137.1, 139.0, 148.1 (C-ar), 170.4 (C=O), 180.8 (C=S). Found, %: C 43.49, H 4.03, N 12.58. Calculated, %: C 43.65, H 3.89, N 12.72. C16H17IN4OS.

5-{2-[(4-Iodophenyl)amino]ethyl}-4-phenyl-2,4-dihydro-3H-1,2,4-triazole-3-one (12)

The semicarbazide 10 (0.85 g, 2 mmol) and 20 ml of aqueous 2 % sodium hydroxide solution were heated under reflux for 4 h, cooled, and acidified with acetic acid to pH 6. The formed residue was filtered off, washed with water, and dried. Compound 12 was crystallized from 2-propanol.

Light-blue solid (0.71 g, 87 %), m.p. 182–183 °C (2-propanol). IR, ν: 1,591 (C=N), 1,706 (C=O), 3,175 (NH), 3,345 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.60 (2H, t, J = 7.3 Hz, CH2C=), 3.15 (2H, dd, 1 J = 7.1 Hz, 2 J = 13.6 Hz, CH 2NH), 5.88 (1H, t, J = 6.0 Hz; CH2 NH), 6.18–7.57 (9H, m, Har), 11.72 (1H, s, NNH). 13C NMR (75,4 MHz, [D6]DMSO): δ = 25.5 (CH 2C=), 39.0 (CH2NH), 76.1, 114.4, 127.5, 128.6, 129.4, 132.8, 137.0, 147.7 (C-ar), 145.0 (C=N), 154.3 (C=O). Found, %: C 47.49, H 3.84, N 13.95. Calculated, %: C 47.31, H 3.72, N 13.79. C16H15IN4O.

5-{2-[(4-Iodophenyl)amino]ethyl}-4-phenyl-2,4-dihydro-3H-1,2,4-triazole-3-thione (13)

The thiosemicarbazide 11 (0.88 g, 2 mmol) and 20 ml of aqueous 2 % sodium hydroxide solution were heated under reflux for 3 h, cooled down, and acidified with acetic acid to pH 6. The formed residue was filtered off, washed with water, and dried. Compound 13 was crystallized from 2-propanol.

White solid (0.80 g, 95 %), m.p. 156–157 °C (2-propanol). IR, ν: 1,339 (C=S), 1,572 (C=N), 3,102 (NH), 3,393 (CH2 NH). 1H NMR (300 MHz, DMSO-d6): δ = 2.63 (2H, t, J = 7.2 Hz, CH2C=), 3.18 (2H, t, J = 7.0 Hz, CH 2NH), 5.50–6.0 (1H, br. s, CH2 NH), 6.15–7.60 (9H, m, Har), 13.77 (1H, s, NNH). 13C NMR (75.4 MHz, DMSO-d6): δ = 25.0 (CH 2C=), 39.2 (CH2NH), 76.2, 114.4, 128.3, 129.4, 129.4, 133.6, 137.0, 147.5 (C-ar), 150.3 (C=N), 167.5 (C=S). Found, %: C 45.65, H 3.69, N 13.26. Calculated, %: C 45.51, H 3.58, N 13.27. C16H15IN4S.

Antibacterial and antifungal activity testing in vitro

Diffusion technique

The antimicrobial activity of the compounds was evaluated by diffusion in agar on solid nutrient medium (nutrient agar—for bacteria, wort agar—for fungi). Disks (5 mm diameter) were soaked in 0.02 mg/ml solutions of compounds in DMSO. Disks were put on an exponentially growing plated culture. The microbial loading was 109 cells/1 ml. The plates were then incubated for bacteria for 24 h at 35 °C and for fungi for 48–72 h at 28–30 °C. The results were recorded by measuring the zones surrounding the disk. Control disk contained vancomycin (for bacteria) or nystatin (for fungi) as standard.

Serial dilution technique

Testing was performed in flat-bottomed 96-well tissue culture plates. The tested compounds were dissolved in DMSO; the concentration range was 500–1.9 μg/ml. Inoculum of bacteria and fungi was inoculated in nutrient medium (nutrient meat extract—for bacteria, wort—for fungi). The duration of incubation was 24–72 h at 37 °C for bacteria and 30 °C for fungi. The results were estimated according to the presence or absence of growth of microorganisms.

Computational studies using PASS

Prediction of the biological activity spectra for the synthesized compounds was performed using the PASS computer program [35, 36]. The latest PASS version (2012.11.22) predicts 6,400 kinds of biological activity based on analysis of a training set including information about ~330,000 drugs, drug candidates, lead compounds, etc. The average accuracy of prediction estimated in the leave-one-out cross-validation procedure for the whole training set is about 95 %. The PASS Online version is freely available for the scientific community via a website. Based on the PASS predictions, new pharmaceutical agents from various chemical classes with different kinds of biological activity have been discovered [35, 36]. PASS allows simultaneous prediction of the interaction of chemicals with many biological targets [67]. Based on the prediction results using the PharmaExpert computer program, a substance can be selected from the pleiotropic effects. In this way, antihypertensive [68] and antiinflammatory [69] pharmacological agents with dual mechanisms of action have been identified, and nootropic properties of antihypertensive drugs unrelated to their antihypertensive activity were discovered [70]. It has also been shown to predict the biological activity of the components of herbal medicines [71].

PASS input information is presented as MOL or SDF files with the structural formulae of the compounds under study; the PASS output is presented as a list of probable activities with two estimated probabilities: Pa—the probability to be “active,” and Pi—the probability to be “inactive.”

The chemical descriptors used in the PASS analysis are called MNA [37]. They are automatically generated on the basis of the MOL file of a molecule. The list of unique MNA descriptors currently consists of more than 75,800 different items. New ones are added to this list if found in a novel compound, refreshing the training set. MNA descriptors are effectively applied in structure–activity relationship (SAR), quantitative SAR (QSAR), and similarity analyses. They can also be used as keys or fingerprints to cluster libraries of chemical compounds, select representative subsets from chemical databases, etc.

The list of predicted activities for a certain chemical compound is called the biological activity spectrum. It includes both the main and side pharmacological effects (e.g., antihypertensive, hepatoprotective, sedative, etc.), molecular mechanisms of action (5-hydroxytryptamine agonist, acetylcholinesterase inhibitor, adenosine uptake inhibitor, etc.), specific toxicities (carcinogenic, hallucinogenic, hepatotoxic, etc.), antitargets (ATPase inhibitor, CYP3A4 inhibitor, HERG channel blocker, etc.), terms associated with drug metabolism (CYP1A substrate, CYP1A1 human substrate, CYP3A4 substrate, etc.), terms associated with the transport of drugs (P-glycoprotein substrate, P-glycoprotein inhibitor, P-glycoprotein inductor, etc.), and terms associated with gene expression (APOA1 expression enhancer, ErbB-2 expression inhibitor, etc.).

The result of the prediction is presented as a list of activities with the appropriate Pa, which estimates the probability for the compound to be active or inactive for each type of biological activity. Its values vary from 0.000 to 1.000. The higher the Pa value, the lower the predicted probability of obtaining false positives in biological testing; For example, if one selects for testing only compounds for which a particular activity is predicted with Pa > 0.9, the expected probability of finding inactive compounds in the selected set is very low; however, about 90 % of active compounds are missed. If one lowers the Pa threshold to 0.8, the probability of finding inactive compounds is still low, but (only) about 70 % of active compounds are missed; etc. PASS uses the criterion Pa = Pi as the default threshold; i.e., all compounds with Pa > Pi are declared as being active.

Another aspect of the predictions is the compounds’ novelty. If one limits oneself to high Pa values, one may find close analogs of known biologically active substances among the tested compounds; For example, for Pa > 0.7, the chance of finding the activity in an experiment is high, but some of the compounds may be close analogs of known pharmaceutical agents. If one chooses 0.5 < Pa < 0.7, the chances of obtaining activity in the experiment are lower, but the compounds may be less similar to known pharmaceutical agents. For Pa < 0.5, the chances of obtaining activity in the experiment are even lower, but if it is found, the compound might happen to be a new chemical entity.

The prediction results were obtained for all compounds using the PASS version which predicts 4,169 types of biological activity including 84 pharmacological effects, 3,965 molecular mechanisms, 110 metabolism-related actions, 2 gene expression regulation, 4 transporter-related actions, 4 side effects, and toxicity.

If PASS predictions are analyzed for a set of compounds from a certain chemical series, the predicted activities may be defined as typical (predicted for the majority of derivatives from this series) or minor (predicted only for some compounds) [42].

Therefore, the computer program PASS can be effectively used for the following purposes [66]: finding compounds with required properties and without undesirable side effects, revealing new effects and mechanisms of action of known substances from corporate and personal databases, selecting the most propitious compounds from a set of available samples for high-throughput screening, and determining the more relevant screens for a particular compound.