Synthesis, Spectroscopic and Theoretical Studies of New Quaternary N,N-Dimethyl-3-phthalimidopropylammonium Conjugates of Sterols and Bile Acids

New quaternary 3-phthalimidopropylammonium conjugates of steroids were obtained by reaction of sterols (ergosterol, cholesterol, cholestanol) and bile acids (lithocholic, deoxycholic, cholic) with bromoacetic acid bromide to give sterol 3β-bromoacetates and bile acid 3α-bromoacetates, respectively. These intermediates were subjected to nuclephilic substitution with N,N-dimethyl-3-phthalimidopropylamine to give the final quaternary ammonium salts. The structures of products were confirmed by spectral (1H-NMR, 13C-NMR, and FT-IR) analysis, mass spectrometry (ESI-MS, MALDI) as well as PM5 semiempirical methods and B3LYP ab initio methods. Estimation of the pharmacotherapeutic potential has been accomplished for synthesized compounds on the basis of Prediction of Activity Spectra for Substances (PASS).


Introduction
Steroids are a large class of organic compounds. They play a very important role in animals, plants and microorganisms. The best known steroid is certainly cholesterol. Cholesterol was isolated for the first time from gall stones nearly two centuries ago by Chevreul [1,2]. This sterol is an important OPEN ACCESS component of mammalian cell membranes; it is also present in significant concentrations in the brain and nervous tissue [3][4][5][6]. Cholesterol is the biosynthetic precursor of steroid hormones, bile acids, vitamin D and lipoproteins [7][8][9]. Like the functions of cholesterol in mammals, ergosterol is necessary to support the life of fungi. It serves two main purposes: a bulk membrane function and a sparking function. Ergosterol is a biological precursor to vitamin D 2 (ergocalciferol) [10][11][12].
All sterols are crystalline compounds with a secondary hydroxyl group in the position C(3) of the steroid skeleton, one or two double bonds and differently modified side chains. Rings A/B of the steroid skeleton may have trans geometry (the allo series) or cis (the normal series). Sterols have the hydroxy group on the C(3) position forming a number of β-sterols with respect to the average plane of the ring. By contrast bile acids have hydroxy groups on the C(3) position which prefer the α orientation [13][14][15][16][17].
Bile acids are major metabolites of cholesterol, being end products of its metabolism in the liver [18,19]. They are isolated from the bile of higher animals, where they are found as sodium salts of peptide conjugates with glycine and taurine. The most important are the primary bile acids, e.g., chenodeoxycholic acid and cholic acid, which are successively transformed into secondary bile acids such as ursodeoxycholic, deoxycholic and lithocholic acids [20][21][22]. Bile acids (e.g., lithocholic, deoxycholic and cholic) are very interesting because they display a large, rigid, and curved skeleton. Moreover, they possess chemically different polar hydroxy groups (3α, 3α,7α and 3α,7α,12α) and amphiphilic properties. Modifications of the functional groups of sterol molecules allow one to obtain systems with high pharmacological activity [23].
Quaternary alkylammonium salts play an important role in the living organisms and many functions of prokaryotic and eukaryotic cells have been shown to be alkylammonium salts dependent [24,25]. These compounds also exhibit excellent antimicrobial activity, and therefore they are used as antiseptics, bactericides and fungicides, as well as therapeutic agents. In general, quaternary alkylammonium salts with good antimicrobial activities contain one or two alkyl chains with lengths in the C 8 -C 14 range. For the applications as softeners and hair conditioning agents hydrocarbon chain lengths between C 16 and C 18 are used [26][27][28]. Phthalimides, and N-substituted phthalimides are also an important class of compounds because of their biological activities as antimicrobial agents [29,30]. It has recently been shown that tetrachlorophthalimide derivatives are good α-glucosidase inhibitors [31]. The use of microbiocides of the same type for a long time may cause an increase of resistance of microorganisms to the chemicals used, which is a very serious and dangerous problem. Antimicrobial resistance of bacteria comprises a wide variety of biochemical mechanisms and processes that allow microorganisms to grow in the presence of microbiocides [32][33][34]. There are many ways to overcome the risk of an increasing resistance of microorganisms, however the best one is a periodically application of new microbiocides with modified structures [35,36]. Therefore, connections of sterols and bile acids with various amines or polyamines appears to be an unusually interesting potential approach to such new structures [37][38][39][40].
New quaternary 3-phthalimidopropylammonium conjugates of steroids were obtained by reaction of ergosterol (1), cholesterol (2), cholestanol (5α-cholestan-3β-ol, 3), and bile acids 10-12 with bromoacetic acid bromide to give 4-6 and 13-15. The 3β-bromoacetates of sterols and 3α-bromoacetates of bile acids, as well as N,N-dimethyl-3-phthalimidopropylamine were prepared according to the literature procedures [41,42]. The structure of products was confirmed by 1 H-NMR, 13 C-NMR, and FT-IR analysis, as well as ESI-MS and MALDI. The syntheses of conjugates 7-9 and 16-18 are shown in Scheme 1. Potential pharmacological activities of the synthesized compounds have been determined on the basis of computer-aided drug discovery approach with in silico Prediction of Activity Spectra for Substances (PASSs) program. It is based on a robust analysis of the structure-activity relationship in a heterogeneous training set currently including about 60,000 biologically active compounds from HO HO (2) (1)

10'
BrCH 2 COBr, TEBA, H different chemical series with about 4,500 types of biological activities. Since only the structural formula of the chemical compound is necessary to obtain a PASS prediction, this approach can be used at the earliest stages of investigation. There are many examples of the successful use of the PASS approach leading to new pharmacological agents [43][44][45][46][47]. The PASS software is useful for the study of biological activity of secondary metabolites. We have selected the types of activities that were predicted for a potential compound with the highest probability (focal activities). If predicted activity is higher than 0.7 (PA > 0.7), the substance is very likely to exhibit the activity in experiment and the chance of the substance being the analogue of a known pharmaceutical agent is also high. If predicted activity is between 0.5 and 0.7 (0.5 < PA < 0.7), the substance is unlikely to exhibit the activity in experiment and the similarity to known pharmaceutical substance is very limited.
The structures of all synthesized compounds were determined from their 1 H-and 13 C-NMR, FT-IR and ESI-MS spectra. Moreover, PM5 calculations and B3LYP ab initio methods were performed for all compounds. Additionally, analyses of the biological prediction activity spectra for the new esters prepared herein are good examples of in silico studies of chemical compounds. We also selected the types of activity that were predicted for a potential compound with the highest probability (focal activities) ( Table 1). According to these data the most frequently predicted types of biological activity are: inhibitors glyceryl-ether monooxygenase, acylcarnitine hydrolase, alkylacetylglycerophosphatase, plasmanylethanolamine desaturase, N-(acyl)ethanolamine deacylase and protein-disulfide reductase. The In the 1 H-NMR spectra of sterol conjugates 7-9 and bile acid conjugates 16-18 a signal of protons of the COCH 2 N + group in the range 5.35-4.55 ppm is observed. The peaks of six methyl protons of the N + (CH 3 ) 2 and two methylene protons of the N + CH 2 appear as singlets and multiplets or broad singlets in the range 3.69-3.61 ppm and 4.96-3.87 ppm, respectively. Two methylene protons of attached to the phthalimide ring-N-CH 2 group are seen as a broad singlets in the 3.83-3.82 ppm range. The 13 C-NMR spectra of conjugates 7-9 and 16-18 in CDCl 3 show characteristic signals at 15.80 ppm (7) The proton chemical shift assignments of N, Table 2) are based on 2D COSY experiments, in which the proton-proton connectivity is observed through the off-diagonal peaks in the counter plot.
The relations between the experimental 1 H-and 13 C-NMR chemical shifts (δ exp ) and the Gauge-Independent Atomic Orbitals (GIAO) isotropic magnetic shielding tensors (σ calc ) for 8 are shown in Figure 2. Both correlations are linear, described by the equation: δ exp. = a + bσ calc . The a and b parameters are given in Table 2  The correlation between the experimental chemical shifts and calculated isotropic screening constants are better for carbon atoms than for protons. The protons are located on the periphery of the molecule and thus they are exposed to stronger interactions with solvent than carbon atoms, which are more hidden inside of structure. The differences between the exact values of the calculated and experimental shifts for protons are probably due to the fact that the shifts are calculated for single molecules in gas phase, whereas experimental values are due to the condensed phase. For this reason the agreement between the experimental and the calculated data for protons are worse than for carbons. PM5 semiempirical calculations were performed using the WinMopac 2003 program and B3LYP calculation are performed using the GAUSSIAN 03 program package with the 6-31G(d,p) basis set. The final heat of formation (HOF) and energies for the sterols 1-3, bile acids 10-12 as well as their conjugates 7-9 and 16-18 is presented in Table 3. Representative conjugates of sterol 7 and bile acid 18 are shown in Figure 3.
The lowest values of HOF for sterols are observed for cholestanol 3 and its conjugate 9, where there are no double bonds which stabilize the molecule and hinder its reactivity, in contrast to conjugates of ergosterol 7 and cholesterol 8 where the double bonds increase the reactivity of the molecule, thereby increasing values of HOF. The HOF relationship for methyl esters of bile acids 10-12 and their corresponding conjugates 16-18 can be explained in a similar manner. In this case, the number of hydroxyl groups in the steroid skeleton lowers the value of the determinant of HOF. This spatial arrangement of bile acids can facilitate the formation of stable host-guest complexes. These complexes may be stabilized by hydrogen bonding or electrostatic interactions that arise from the number of hydroxyl groups in the bile acid molecule. Similar correlations have been observed using the B3LYP method.
The spatial arrangement and interaction of the conjugates 7 and 18 are shown in Figure 4. The final heat of formation is -1249.429 kcal/mol for 7 and -1358.893 kcal/mol for 18 and the distances between the quaternary nitrogen and the anion bromide are 4.34 Å and 4.33 Å, respectively.
Compensation charges occurs only through intermolecular electrostatic interaction. This is a very good confirmation of the conclusion that interactions reduce HOF. The dipole moments and selected geometry parameters were calculated at the PM5 and B3LYP /6-31G(d,p) level of theory are presented in Table 4.     The calculated bond lengths and bond angles for 7-9 and 16-18 optimized by the PM5 and B3LYP methods are quite similar, however the bond lengths N(2)-C(2') and N(2)-C(4') are different. Also the torsion angles calculated by the PM5 and B3LYP methods are slightly different, especially the C(5')-C(4')-N(2)-C(2') angle for fatty acid conjugates 16-18. This shows a crucial role of electrostatic interaction between oppositely charged groups in the structure of the investigated compounds (Table 4). Hydrogen bonds and short contact lengths distances are shorter for compounds calculated by B3LYP method. These data prove that in the gas phase the type of quantum chemical methods used play an important role in the molecular structure of ionic compounds. The solid-state IR spectra of sterols and bile acids conjugates are shown in Figures 5 and 6, respectively.  The FT-IR spectra of conjugates show characteristic bands at 1,775-1,772 cm −1 which are due to the asymmetric carbonyl group ν as C=O stretching vibrations in a phthalimide moiety (Figures 5-8) [48,49]. The symmetric ν s C=O stretching vibration appears in the FT-IR spectrum as an intense and broad nonsymmetrical band at 1,716-1,699 cm −1 suggesting the small nonequivalence of carbonyl groups in the phthalimide moiety. Moreover strong characteristic bands in the region 1,251-1,240 cm −1 are present, which are assigned to the ν(C-O).   The FT-IR spectra of bile acid conjugates (Figure 7, blue line) shows characteristic vibration bands of ν as C=O and ν s C=O in a phthalimide moiety at 1,773 cm −1 and at 1,712 and 1,699 cm −1 , respectively. In this case the ν as C=O and ν s C=O bands are more nonsymmetrical in comparison to the carbonyl bands of 17 and 18. The nonequivalence of ν as C=O in the FT-IR spectrum is observed for N,N-bis-(phthalimidopropyl)-N-propylamine [50]. In contrast to the above examples, in N,N-dimethyl-3phthalimidopropylammonium hydrochloride monohydrate and N-n-butyltetrachlorophthalimide no split of the carbonyl bands in FT-IR spectra were observed, in spite of the different interactions of each carbonyl group in the supramolecular structure [51,52]. The ν(COO) stretching vibrations of carboxy groups are observed at 1,747-1,736 cm −1 (Figure 7).
The room-temperature solid-state FT-IR and the calculated spectrum of 9 are shown in Figure 8. The band frequencies, relative intensities and their assignments in the 4,000-400 cm −1 range are listed in Table 5. For convenience of comparison, the band intensities for the calculated spectrum are scaled.  The DFT harmonic vibrational wavenumbers are usually higher than the experimental values. However, in this case the overall agreement between the experimental and calculated frequencies for (9) is very good (Figure 9). Any discrepancy noted between the observed and calculated frequency may be due to the fact that the calculations have been done for a single molecule in the gaseous state contrary to the experimental spectrum recorded in the presence of intermolecular interactions. The scaling procedure, as recommended by Palafox was used [53,54]. The scaled IR spectrum is shown in Figure 8b  Calculated frequencies by B3LYP method Observed frequencies, cm -1 predicted frequencies are listed in Table 3 as ν scaleq . Scaling of the harmonic vibrational frequencies reproduce the experimental solid-state FT-IR frequencies with the r.m.s. error of 38.1 cm −1 . The vibrational band assignments of 9 were made using Gauss-View molecular visualization program [55].

General
The NMR spectra were measured with a Varian Mercury 300 MHz NMR spectrometer (Oxford, UK), operating at 300.07 and 75.4614 for 1 H and 13 C, respectively. Typical conditions for the proton spectra were: pulse width 32°, acquisition time 5 s, FT size 32 K and digital resolution 0.3 Hz per point, and for the carbon spectra pulse width 60°, FT size 60 K and digital resolution 0.6 Hz per point, the number of scans varied from 1200 to 10,000 per spectrum. The 13 C and 1 H chemical shifts were measured in CDCl 3 relative to an internal standard of TMS. Infrared spectra were recorded in the KBr pellets using a FT-IR Bruker IFS 66 spectrometer (Karlsruhe, Germany). The ESI (electron spray ionization) mass spectra were recorded on a Waters/Micromass (Manchester, UK) ZQ mass spectrometer equipped with a Harvard Apparatus (Saint Laurent, QC, Canada), syringe pump. The sample solutions were prepared in methanol at the concentration of approximately 10 −5 M. The standard ESI-MS mass spectra were recorded at the cone voltage 110 V. The MALDI (matrix-assisted laser desorption/ionization) mass spectra were recorded on a Waters Maldi Q-Tof Premiere. The sample solutions were prepared in methanol at the concentration of approximately 10 −5 M. The matrix was 2,5-dihydroxybenzoic acid (gentisic acid) and the standard was β-cyclodextrin (m/z 1157.3218). PM5 semiempirical calculations were performed using the WinMopac 2003 program [56][57][58]. The calculations were performed using the GAUSSIAN 03 program package [59] at the B3LYP [60][61][62] levels of theory with the 6-31G(d,p) basis set [63]. The NMR isotropic shielding constants were calculated using the standard Gauge-Independent Atomic Orbital (GIAO) approach of Gaussian 03 [64,65].