Structural Requirements for the Antifungal Activities of Natural Drimane Sesquiterpenes and Analogues, Supported by Conformational and Electronic Studies

Seventeen drimanes including polygodial (1), isopolygodial (2), drimenol (3) and confertifolin (4) obtained from natural sources and the semi-synthetic derivatives 5–17 obtained from 1–3, were evaluated in vitro for antifungal properties against a unique panel of fungi with standardized procedures by using two end-points, MIC100 and MIC50. A SAR analysis of the whole series, supported by conformational and electronic studies, allowed us to show that the Δ7,8 -double bond would be one of the key structural features related to the antifungal activity. The MEPs obtained for active compounds exhibit a clear negative minimum value (deep red zone) in the vicinity of the Δ7,8 -double bond, which is not present in the inactive ones. Apart of this negative zone, a positive region (deep blue) appears in 1, which is not observed either in its epimer 2 nor in the rest of the active compounds. The LogP of active compounds varies between 2.33 and 3.84, but differences in MICs are not correlated with concomitant variations in LogP values.


Introduction
In the course of our ongoing search for antifungal compounds from natural sources, we recently reported that the aerial parts of Polygonum acuminatum Kunth., used in the Argentinean traditional medicine for ailments related to fungal infections [1], possessed antifungal properties against yeasts and dermatophytes, these results supporting their ethnopharmacological use [2]. Polygodial (1, Figure 1) was isolated from the most active extract as the main compound responsible for the activity, although other drimane derivatives such as isopolygodial (2) drimenol (3) and confertifolin (4) (Figure 1) showed antifungal properties too [2]. Although the claims that 1 possesses antifungal activity were consistent in all previous reports [3][4][5][6], contradictory reports were found in the literature regarding the antifungal activities of its natural analogue 2, which was variously reported to be either active or inactive, and based on this information, conflicting opinions about the main structural features required for drimanes to display antifungal activities were put forth. They can be summarized as follows: Fujita and Kubo stated that the orientation of the C-9 carbonyl group was indispensable for activity, since 1 (with an aldehyde in the C-9 orientation) but not 2 (with a C-9 aldehyde) possessed antifungal activity [7]. Conversely, Anke and Sterner [8] found that 2 did possess antifungal activity, although lower than that of 1, which was corroborated by us in the paper cited above [2] and in a subsequent one [9]. Another structural feature which was assigned a role in the antifungal activity was the ,-unsaturated C-8 aldehyde moiety which, according to Taniguchi et al., was essential for activity [10]. However, this appears to be a contradiction with their own previous report in which 2 was devoid of antifungal activity, although it possess the ,-unsaturated C-8 aldehyde moiety [7,10]. Fujita and Kubo also reported that 1 lost its antifungal activity against Saccharomyces cerevisiae in certain media, because of formation of a pyrrole between its aldehydes and the amines present in the medium [7]. Considering that, with the formation of a pyrrole, both aldehydes disappear, this report opens a new question: if one or both aldehydes are necessary for the antifungal activity.
In order to clarify the main features required for drimanes to display antifungal activities, we tested herein a series of 17 compounds 1-17 (most of them not previously assayed for antifungal activities) for antifungal properties against a unique panel of nine fungal strains, with a standardized methodology, in the same laboratory. To guarantee confident and reproducible results, the whole series was tested following the Clinical and Laboratory Standards Institute (CLSI) [11,12] standardized procedures. CLSI has established consensus procedures to facilitate the agreement among laboratories in measuring the susceptibility of yeasts (updated in 2008 as M-27 A3) and of filamentous fungi (updated in 2008 as M-38 A2) to antifungal agents, with broth dilution methods. The standardized parameters detailed in both documents include protocols for the preparation of antifungal stock solutions, dilutions for testing, inocula, culture medium, temperature and incubation time, endpoint definitions and reference MIC ranges for microdilution testing of both, the established and newly introduced antifungal agents.
Among the tested analogues of 1, compounds 2-4 were isolated from P. acuminatum leaves and the remaining compounds 5-17 were prepared by isomerization, reduction, oxidation and methylenation reactions using the natural compounds 1, 2 and 3 as starting materials. They were tested against three yeasts: Candida albicans, S. cerevisiae and Cryptococcus neoformans; three Aspergillus spp. (A. flavus, A. fumigatus and A. niger) and three dermatophytes: Microsporum gypseum, Trichophyton rubrum and Trichophyton mentagrophytes by using two end-points MIC 100 and MIC 50 . A SAR analysis of the whole series, supported by conformational and electronic studies on some selected compounds is presented herein, with the aim of determining the stereo-electronic characteristics related to the antifungal behavior and thus contribute to the understanding of the structural properties of antifungal drimanes.

Chemistry
Sesquiterpenes 1-4 were obtained from P. acuminatum leaves as described previously [2]. Compounds 1-3 were used as starting materials for the preparation of compounds 5-17.

Antifungal Activity
To carry out the antifungal evaluation, concentrations of naturally occurring drimanes 1-4 and semi-synthetic derivatives 5-17 up to 250 µg/mL were incorporated to growth media according to CLSI guidelines [11]. Table 1 summarizes the concentration of drimanes that completely inhibited (MIC 100 ) the growth of nine opportunistic pathogenic fungi including yeasts (C. albicans, C. neoformans, and S. cerevisiae), hialohyphomycetes (Aspergillus spp.) as well as dermatophytes (Microsporum and Trichophyton spp). In addition to MIC determinations, the Minimum Fungicidal Concentration (MFC) of each active compound against this panel was accomplished by sub-culturing a sample of media from MIC tubes showing no growth, onto drug free agar plates (Table 1).  I  I  I  I  I  I  I  I  I   MFC  ---------12   MIC 100  I  I  I  I  I  I  I  I  I   MFC  ------ Regarding the sensitivity of the different fungi to drimanes, it can be stated that: (i) all compounds, but not 11, 12 and 17, exhibited moderate activity (MICs  250 µg/mL) against dermatophytes and yeasts; (ii) no member of the series was active against Aspergillus spp; (iii) all active compounds showed fungicidal rather than fungistatic properties against most fungi, since they not only inhibit but also kill them at Minimum Fungicidal Concentrations (MFC)  250 µg/mL.

Structure-Activity Relationships (SAR)
Since no differences in MICs among the members were observed against dermatophytes (one-dilution-step MIC changes are within the range of experimental error) [25], the SAR studies were performed on the activity of the series against yeasts (C. albicans, S. cerevisiae and C. neoformans) which showed striking differences.
For a better analysis of the differential behavior of these compounds against yeasts, we selected the clinically important fungi C. neoformans and C. albicans as targets. C. neoformans is a fungus that remains an important life-threatening complication for immunocompromised hosts, particularly for patients who have undergone transplantation of solid organs [26]. Therefore, the activity of compounds against this fungus has a true clinical relevance. In turn, C. albicans is one of the leading causes of nosocomial blood stream infections worldwide [27].
Thus, compounds 1-10, 14-16 that showed activity against at least one yeast were re-tested against the standardized strains C. albicans ATCC 10231 and C. neoformans ATCC 32264 determining, this time, MIC 50 (the minimum concentration of compounds that inhibit 50% of growth), a less stringent end-point highly used [27] and recommended by CLSI [10]. MIC 50 consistently represent the in vitro activity of compounds and many times provide a better correlation with other measurements of antifungal activity [27]. Results are shown in Table 2. From the analysis of the structures and the activities displayed by all compounds against C. albicans and C. neoformans, some structure-activity relationships can be extracted. They were grouped according to the influence of each structural feature such as  7,8 -double bond in the drimane skeleton, configuration of C-9, presence of an aldehyde at C-9 (or at C-8) and presence of an α,β-unsaturated aldehyde moiety at C-8. In addition, it can be observed that the  7,8 -double bond in active structures can be found either alone as in compounds 3, 5-10 and 14, or as being part of the C-8 α,β-unsaturated aldehyde moiety (constituting a Michael acceptor group) as in compounds 1 and 2. The fact that both drimane structural types do display antifungal activity suggests that antifungal drimanes could not act via a Michael addition. This finding differs from the antifungal structural requirements of some other sesquiterpenes such as the sesquiterpenelactones (+)-costunolide and (−)-dehydrocostunolide isolated from Centaurea plants [28] and the synthetic tuberiferin and dehydrobrachylaenolide [29], whose antifungal activities were reported to be enhanced by a Michael acceptor group. Nevertheless, our findings are not in any way strange since many other diverse structural compounds not containing a Michael acceptor group, were found to display strong antifungal properties [30,31].

Role of the Configuration of C-9
The comparison of activities of 1 and 2 shows that, in contrast to Fujita and Kubo's previous report [7], both compounds do possess antifungal activity. However, the 9-derivative 2 displays 16 and eight times lower activity than its epimer 1, against C. albicans and C. neoformans, respectively. This result is consistent with Anke and Sterner's [8] and Derita et al.'s [2] previous studies, who both found that 2 did display antifungal activities, but lower than 1. The different antifungal activities displayed by compounds 1 and 2 are supported by their different electronic distributions which are presented in Section 2.4.

2.3.3.
Role of an Aldehyde at C-9 and of a CH 2 OH at C-8 The presence of an aldehyde group at C-9 appears not to be indispensable for activity, as can be observed in compounds 3-7, 9, 10, 15 and 16. Otherwise, the comparison of diol 9 with the monoalcohol 10, suggests that a CH 2 OH at C-8 does not contribute to the activity either.

Conformational and Electronic Studies
To gain a better understanding of the above experimental results, computer-assisted conformational and electronic studies were performed on all compounds of the series. The purpose was to obtain precise information about the antifungal drimanes' stereoelectronic characteristics that give support to the relationships between structure and activity.

Conformational Analysis
For the conformational analysis, each ring of the drimane scheleton was named A and B (Figure 2). The essential conformational problem of drimane sesquiterpenes, involves two aspects: (a) The conformational behavior of torsional angles  1 - 5 which determines the spatial ordering of ring A. (b) The orientation of torsional angles  6 - 9 which determines the overall shape of ring B. In this ring, they are also important torsional angles  10 and  11 , which are related to CH 2 OH substituents as R 2 or R 1 .
Our B3LYP/6311G(d.p) calculations indicated that a chair, and not a boat form, is the energetically preferred spatial ordering for ring A in compound 1. The energy gap obtained between both conformers is 14.7 Kcal/mol, indicating that this compound is fairly rigid and possesses a restricted conformational flexibility. Similar results were obtained for all compounds possessing this structural feature like compounds 1-3, 5, 8-10 and 14. Considering that, in these compounds, ring B is a cyclohexene, each of the three possible conformations: sofa (also called half-boat), half-chair and boat, were evaluated. OH OH Our theoretical results indicated that the so-called "half-boat" conformation is the highly preferred low-energy form Figure 3a which is in agreement with experimental data obtained for structurally related compounds [32][33][34]. Considering that all these compounds adopt similar low flexible conformations, it is clear that this would not be the cause of the different activities displayed against fungi. The same can be said for compounds 4, 6 and 7, for which our theoretical calculations indicate that the conformational flexibility is even more restricted than the observed for the above compounds as a consequence of the presence of a third fused ring. The energetically preferred conformation obtained for compound 6 is shown in Figure 3b. Closely similar conformations were obtained for compounds 4 and 7. Thus, the knowledge of the stereo-electronic properties of these drimanes will surely help to elucidate the structural requirements for the antifungal activity. For this analysis, Molecular Electrostatic Potentials (MEPs) are particularly valuable, because they allow the visualization and assessment of the capacity of a molecule to interact electrostatically with a putative-binding site [35][36][37]. Figure 4 gives the MEPs obtained for compounds 1, 2 and 5, which showed high, moderate and low activities respectively. The MEP obtained for 1 (Figure 4a) exhibits three characteristic regions: a clear negative minimum value (deep red zone with V(r) min ≈ −0.1139 el/au 3 ) in the vicinity of Δ 7,8 double bond; a positive region (deep blue zone with V(r) max ≈ 0.1136 el/au 3 ) located near the CH 3 group at C 10 ; and an extended hydrophobic zone (deep and light green zone with an almost neutral potential V(r) med ≈ −0.0011 el/au 3 ) in the rest of the molecule. Considering that 1 is the most active of the series, this MEP should account for the characteristics of the electronic general pattern for all the molecules possessing antifungal effect. Interestingly, the MEP obtained for its epimer 2 (Figure 4b), showed a marked electronegative zone near the double bond Δ 7,8 similar to the showed by 1 but a little smaller. In turn, the positive zone is not observed in 2 which is a clear difference with 1. Regarding compound 5, concomitantly with its lower activity, it showed a red negative zone but considerably smaller than that of 2 (Figure 4c). For the inactive compounds 13 and 17, the electronegative zone is totally missing ( Figure 5). This is an expected result considering that these compounds do not possess a double bond in such portion of their structures.
Finally, we calculated logP (the logarithm of the partition coefficient in a biphasic system, e.g., n-octanol/water) for all drimanes which are shown in Table 2. It is known that LogP describes the macroscopic hydrophobicity of a molecule, which is a factor related to its ability to penetrate membranes of fungal cells and to reach the interacting sites, thus influencing the antifungal activity of compounds [38]. LogP values for the different active drimanes of the series varied from 2.33 to 3.84, which fall within the range of values observed for other antifungal compounds [31,39,40]. However, although these values indicate that they all are lipophilic compounds [39,41], LogP have similar values for drimanes with clear different activity, suggesting that this parameter would not have a direct correlation with the antifungal activity of the compounds of Table 2 when acting against C. albicans and C. neoformans. Thus, the most active drimane 1 and its epimer 2, both possess the same LogP value = 2.33 yet their activity is quite different. The same can be observed for compound 14 which has a LogP = 3.53 and displays lower activities. This behavior of drimanes was dissimilar to that of sesquiterpene lactones, for which a clear inverse relationship between polarity and antifungal activity was observed [28,29]. However, we must take into account that the sesquiterpene lactone skeletons are quite different from those of the drimane sesquiterpenes. In addition, different fungal species were used as the test species in each work; here, the yeasts C. albicans and C. neoformans were used, while in the previous reports the filamentous fungus Cunninghamella echinulata [28] and Phycomyces blakesleeanus were used [29]. It has been often observed completely different behaviors for certain compounds towards different type of fungi [42]. As the authors stated in [28], there is not much information about the chemical structureantifungal activity relationships for sesquiterpenoids and deeper studies should be necessary. In this work, we add information that could be useful for a better knowledge of drimane sesquiterpenes behavior against human opportunistic and pathogenic fungi.

Chemistry
Melting points were determined in a Büchi melting point apparatus and are uncorrected. The 1 Hand 13 C-NMR spectra were run on a Bruker DPX 300 spectrometer operating at 300 and 75 MHz, respectively, using chloroform-d as solvent and tetramethylsilane as internal standard. The mass spectra were scanned on a Hewlett Packard HP Engine-5989 spectrometer (equipped with a direct inlet probe) operating at 70 eV.

General Procedure for Synthesis of 5 and 6
To a magnetically well-stirred solution of 1 (2.1 mmol) in MeOH (25 mL) acidified with 5 drops of 1 M HCl, NaBH 4 (8.4 mmol) was added carefully in small portions. The mixture was refluxed at 50 °C for 4 h and most of the solvent was evaporated in vacuo. Water (20 mL) was added cautiously and extractions with EtOAc were performed. The organic phase was dried with MgSO 4 , evaporated and submitted to column chromatography (CC) to give compounds 5 (1.3 mmol) and 6 (0.7 mmol) in yields of 62% and 33%, respectively. All these data are coincident with the literature [13,16].
3.1.3. Synthesis of (−)-(1R,5aS,9aS,9bR)-6,6,9a-Trimethyl-1,3,5,5a,6,7,8,9,9a,9b-decahydro naphto [2,1c]furan-1-ol (isodrimeninol, 7) Compound 1 (0.55 mmol) was dissolved in MeOH (10 mL) acidified with 3 drops of 1 M HCl and NaBH 4 (0.27 mmol) was added carefully in small portions. The mixture was stirred during 30 min at 0 °C and then for 3 h at rt. Water (10 mL) was added cautiously and extractions with EtOAc were performed. The organic phase was dried with MgSO 4 , evaporated and submitted to CC to give compound 7 (0. 39   Compound 1 (2.14 mmol) was dissolved in anhydrous THF (5 mL), the solution was stirred at 0 °C under nitrogen, and Tebbe´s reagent [(C 5 H 5 ) 2 TiCH 2 ClAl(CH 3 ) 2 ] (4 mL of a 0.5M solution in toluene, 2.0 mmol) was added dropwise. The dark brown solution was allowed to warm to room temperature for 10 min, 0.1 M NaOH solution was added (10 drops) and the slurry was filtered through a short pad of silica gel. The filtrate was evaporated and purified by CC to give compound 8 as a colorless oil, yield 25%;  [13,16] A magnetically stirred and cold (0 °C) solution of the alcohol substrate 3 (10 mmol) in acetone (50 mL) was titrated with Jones reagent, until the orange-brown color of the reagent persisted for 30 seconds. The reaction mixture was stirred for an additional 1 h at rt, and then isopropanol (1 mL) was added dropwise to destroy any excess of reagent. Water (100 mL) was added, and the suspension was extracted with AcOEt, dried with MgSO 4 , concentrated, and purified by column chromatography to give compounds 14 (0.68 mmol) and 15 (0.85 mmol) with yields of 30 and 38%, respectively.  To a magnetically stirred solution of the alcohol substrate 3 (10 mmol) in DCM (50 mL), pyridinium chlorochromate (10 mmol) was added. The brown suspension was stirred overnight, until thin layer chromatography analysis showed the disappearance of the starting material. Celite (500 mg) and EtOAc (50 mL) were added to the reaction mixture, and the slurry was filtered through a short pad of silica gel, washing copiously with EtOAc. The filtrate was dried, evaporated, and purified by column chromatography to give compound 16 (  Strains were grown on Sabouraud-chloramphenicol agar slants for 48 h at 30 °C, maintained on slopes of Sabouraud-dextrose agar (SDA, Oxoid), and sub-cultured every 15 days to prevent pleomorphic transformations. Inocula of cells or spore suspensions were obtained according to reported procedures and adjusted to 1-5 × 10 3 cells/spores with colony forming units (CFU)/mL [11,12].

Antifungal Susceptibility Testing
Minimum inhibitory concentration (MIC) of each compound was determined by using broth microdilution techniques according to the guidelines of the CLSI for yeasts (M27-A3) and for filamentous fungi (M 38 A2) [11,12]. MIC values were determined in RPMI-1640 (Sigma, St. Louis, MO, USA) buffered to pH 7.0 with MOPS. Microtiter trays were incubated at 35 °C for yeasts and hialohyphomycetes and at 28-30 °C for dermatophytes strains in a moist, dark chamber, and MICs were visually recorded at 48 h for yeasts, and at a time according to the control fungus growth, for the rest of fungi.
For the assay, stock solutions of pure compounds were two-fold diluted with RPMI from 250 to 0.98 µg/mL (final volume = 100 µL) and a final DMSO concentration ≤1%. A volume of 100 µL of inoculum suspension was added to each well with the exception of the sterility control where sterile water was added to the well instead. Ketoconazole, terbinafine and amphotericin B were used as positive controls.
Endpoints were defined as the lowest concentration of drug resulting in total inhibition (MIC 100 ) of visual growth compared to the growth in the control wells containing no antifungal. MIC 50 was defined as the lowest concentration of a compound that inhibited 50% of the growth control, respectively (culture media with the microorganism but without the addition of any compound), and was determined spectrophotometrically with the aid of a VERSA Max microplate reader (Molecular Devices, Sunnyvale, CA, USA).
The Minimum Fungicidal Concentration (MFC) of each compound against each strain was also determined as follows: After determining the MIC 100 , an aliquot of 5 µL sample was withdrawn from each clear well of the microtiter tray and plated onto a 150-mm RPMI-1640 agar plate buffered with MOPS (Remel, Lenexa, KS, USA.). Inoculated plates were incubated at 30 °C, and MFCs were recorded after 48 h. The MFC was defined as the lowest concentration of each compound that resulted in total inhibition of visible growth.

Calculations Methods
The DFT calculations, full geometry optimization and calculation of the harmonic vibrational frequencies were performed using the GAUSSIAN 03 program [53]. Density Functional Theory (DFT) calculations were employed in order to properly account for the electron correlation effects. The widely employed hybrid method denoted by B3LYP was used, along with the double-zeta split valence basis set 6-311G(d,p). This method includes a mixture of HF and DFT exchange terms and the gradient corrected correlation functional of Lee, Yang and Parr [54,55], as proposed and parameterized by Becke [56,57].
A preliminary conformational search was carried out using the GASCOS algorithm [58][59][60] combined with PM6 an RHF/3-21G calculations. All the conformations obtained using this calculations were confirmed from B3LYP/6311G(d, p) computations. With any conformational search, it is very important to examine the structures obtained to make sure that they are true minima and not transition structures or other structures with very low or zero forces on the atoms (stationary points). Thus, minima were characterized through harmonic frequency analysis employing B3LYP/6311G(d,p) calculations.