Synthesis, Biological Evaluation, and Structure–Activity Relationships of 4-Aminopiperidines as Novel Antifungal Agents Targeting Ergosterol Biosynthesis

The aliphatic heterocycles piperidine and morpholine are core structures of well-known antifungals such as fenpropidin and fenpropimorph, commonly used as agrofungicides, and the related morpholine amorolfine is approved for the treatment of dermal mycoses in humans. Inspired by these lead structures, we describe here the synthesis and biological evaluation of 4-aminopiperidines as a novel chemotype of antifungals with remarkable antifungal activity. A library of more than 30 4-aminopiperidines was synthesized, starting from N-substituted 4-piperidone derivatives by reductive amination with appropriate amines using sodium triacetoxyborohydride. Antifungal activity was determined on the model strain Yarrowia lipolytica, and some compounds showed interesting growth-inhibiting activity. These compounds were tested on 20 clinically relevant fungal isolates (Aspergillus spp., Candida spp., Mucormycetes) by standardized microbroth dilution assays. Two of the six compounds, 1-benzyl-N-dodecylpiperidin-4-amine and N-dodecyl-1-phenethylpiperidin-4-amine, were identified as promising candidates for further development based on their in vitro antifungal activity against Candida spp. and Aspergillus spp. Antifungal activity was determined for 18 Aspergillus spp. and 19 Candida spp., and their impact on ergosterol and cholesterol biosynthesis was determined. Toxicity was determined on HL-60, HUVEC, and MCF10A cells, and in the alternative in vivo model Galleria mellonella. Analysis of sterol patterns after incubation gave valuable insights into the putative molecular mechanism of action, indicating inhibition of the enzymes sterol C14-reductase and sterol C8-isomerase in fungal ergosterol biosynthesis.


Introduction
Fenpropidin and fenpropimorph [1,2] (Figure 1) are well-established antifungals with a broad application in agrochemistry since the 1980s. The structurally related morpholine derivative amorolfine (Loceryl ® ) is used in human dermatology, especially for the treatment of onychomycosis and various local dermal mycoses. The mechanism of action of these antifungals is inhibition of ergosterol biosynthesis. These and related morpholines and piperidines inhibit, to various extents, the enzymes sterol C14-reductase and sterol C8isomerase of the post-squalene part of ergosterol biosynthesis [3,4]. At physiological pH In the last decades, an increase in fungal resistance against commonly used antifungals was seen in clinical fungal isolates such as Aspergillus fumigatus and Candida species; furthermore, an increase in emerging fungal pathogens, such as Mucorales, was observed [11][12][13][14]. Consequently, the antifungal therapy of immunocompromised patients is becoming even more difficult and the need for development of new antifungals is an urgent demand. In parallel to the clinical developments, the demand for novel antifungal agrochemicals is similarly high, as resistance is emerging in this field, too, and a number of old antifungals used in crop protection are meanwhile hampered by safety concerns [15].
One promising strategy in the development of new antifungal compounds with advantages such as higher potency, a broader spectrum, fewer side effects, and a better ecological balance sheet and to break resistance is to start from well-established antifungals as lead structures by means of design and further optimization of hybrids of different active chemotypes [16].
In continuation of our research on the development of antifungals starting from simple aliphatic amines such as benzylamines, (partly) hydrogenated quinolines, and isoquinolines [9,10,17,18] (Figure 1C), we merged essential fragments from these chemotypes and evaluated the 4-aminopiperidine motif as a core structure for novel antifungals. The introduction of a second protonable nitrogen into a sterol biosynthesis inhibitor designed to imitate a carbocationic HEI had shown great benefit in our recent investigations on oxidosqualene cyclase inhibitors [19]. The nature of the residues at both nitrogen atoms of the 4-aminopiperidine core structure was inspired by arylalkylamines (e.g., fenpropidin, Figure 1A), as well as allylamine-type drugs (squalene epoxidase inhibitors) In the last decades, an increase in fungal resistance against commonly used antifungals was seen in clinical fungal isolates such as Aspergillus fumigatus and Candida species; furthermore, an increase in emerging fungal pathogens, such as Mucorales, was observed [11][12][13][14]. Consequently, the antifungal therapy of immunocompromised patients is becoming even more difficult and the need for development of new antifungals is an urgent demand. In parallel to the clinical developments, the demand for novel antifungal agrochemicals is similarly high, as resistance is emerging in this field, too, and a number of old antifungals used in crop protection are meanwhile hampered by safety concerns [15].
One promising strategy in the development of new antifungal compounds with advantages such as higher potency, a broader spectrum, fewer side effects, and a better ecological balance sheet and to break resistance is to start from well-established antifungals as lead structures by means of design and further optimization of hybrids of different active chemotypes [16].
In continuation of our research on the development of antifungals starting from simple aliphatic amines such as benzylamines, (partly) hydrogenated quinolines, and isoquinolines [9,10,17,18] (Figure 1C), we merged essential fragments from these chemotypes and evaluated the 4-aminopiperidine motif as a core structure for novel antifungals. The introduction of a second protonable nitrogen into a sterol biosynthesis inhibitor designed to imitate a carbocationic HEI had shown great benefit in our recent investigations on oxidosqualene cyclase inhibitors [19]. The nature of the residues at both nitrogen atoms of the 4-aminopiperidine core structure was inspired by arylalkylamines (e.g., fenpropidin, Figure 1A), as well as allylamine-type drugs (squalene epoxidase inhibitors) such as naftifine and terbinafine ( Figure 1B) on the one side, and medium to long, linear or branched alkyl chains ( Figure 1C; see evidence described in ref. [9,10,17]) on the other side.
such as naftifine and terbinafine ( Figure 1B) on the one side, and medium to long, linear or branched alkyl chains ( Figure 1C; see evidence described in ref. [9,10,17]) on the other side.
We expected that fine-tuning of the antifungal activity of this new class of compounds should be possible by systematic modification of both N-substituents.

Chemistry
Commercially available N-substituted 4-piperidone derivatives 1a-c were subjected to reductive amination with diverse aliphatic amines, using sodium triacetoxyborohydride as the reducing agent [17,20], to give the secondary amines 2a-j, 3a-g, 4a-f and the tertiary amines 7a,b in moderate to virtually quantitative yields (Scheme 1). In order to ensure stability and sufficient water solubility, all resulting amines could be converted into their corresponding dihydrochlorides (monohydrochlorides in case of the N-Boc compounds 4a-f) by precipitation with hydrogen chloride in diethyl ether. Scheme 1. Synthesis of compounds 2a-j, 3a-g, 4a-f and 7a,b.
The N-Boc protecting group of 4a-c, e, f could be removed by trifluoroacetic acid treatment [21] to give the corresponding unprotected piperidines 5a-c, e, f in good yields (Scheme 2). Attempted deprotection with hydrogen chloride in diethyl ether [22] resulted only in incomplete conversions in same cases. The N-Boc protecting group of 4a-c, e, f could be removed by trifluoroacetic acid treatment [21] to give the corresponding unprotected piperidines 5a-c, e, f in good yields (Scheme 2). Attempted deprotection with hydrogen chloride in diethyl ether [22] resulted only in incomplete conversions in same cases. In order to investigate the importance of having two protonable nitrogen atoms in the molecules, the secondary amines 2b and 3b were reacted with diverse carboxylic acid chlorides (acetyl chloride, 10-undecenoyl chloride, butanoyl chloride, propanoyl chloride, cinnamoyl chloride, and 2-phenylbenzoyl chloride) to give the amides 6a-f. Notably, the amides 6d and 6e were reduced with LiAlH4 in THF to give the corresponding bulky tertiary amines 8d and 8e. The alkene group of 6d was hydrogenated in the same reaction.

Screening for In Vitro Antifungal Activity
First, the antifungal activity (minimum inhibitory concentration, MIC) of the resulting compounds was evaluated in an in-house microdilution assay against the non-pathogenic yeast strain Yarrowia lipolytica (Supporting Information, Table S1). Furthermore, the compounds causing the highest growth inhibition in Y. lipolytica (and 6a, the acetamide Scheme 2. Synthesis of compounds 5a-f, 6a-f, and 8d,e. In order to investigate the importance of having two protonable nitrogen atoms in the molecules, the secondary amines 2b and 3b were reacted with diverse carboxylic acid chlorides (acetyl chloride, 10-undecenoyl chloride, butanoyl chloride, propanoyl chloride, cinnamoyl chloride, and 2-phenylbenzoyl chloride) to give the amides 6a-f. Notably, the amides 6d and 6e were reduced with LiAlH 4 in THF to give the corresponding bulky tertiary amines 8d and 8e. The alkene group of 6d was hydrogenated in the same reaction.

Screening for In Vitro Antifungal Activity
First, the antifungal activity (minimum inhibitory concentration, MIC) of the resulting compounds was evaluated in an in-house microdilution assay against the non-pathogenic yeast strain Yarrowia lipolytica (Supporting Information, Table S1). Furthermore, the compounds causing the highest growth inhibition in Y. lipolytica (and 6a, the acetamide derivative of 3b) were subjected to an extended antifungal activity screening according to the standardized method of the European Committee of Antifungal Susceptibility Testing [23]. Commonly used antifungal agents (amorolfine, voriconazole, Figure 1) were used for comparison (Table 1). The compounds 2b, 3b, 4b, and 5b showed a complete growth inhibition (MIC 100 ) in the same range as the reference antifungals amorolfine hydrochloride and voriconazole against the model strain Yarrowia lipolytica. For clinically relevant species, complete growth inhibition (80% or 90%) was observed in a species-and strain-dependent manner. The lowest minimal inhibitory concentration (MIC) values were determined for compound 3b against Candida spp. and Aspergillus spp. (MIC range 1-4 µg/mL for yeasts and 1-8 µg/mL for Aspergilli). In both groups, MIC values were significantly lower (min. 2 dilution steps) than for amorolfine hydrochloride. Similar reduction in MICs was observed for 2b. Except for A. terreus, confrontation of spores with amide 6a did not result in a complete growth inhibition for Aspergilli or Candida spp. at the concentrations tested. Similarly, no complete growth inhibition was detected for A. flavus and A. fumigatus for compound 2c. All other candidates efficiently inhibited growth in these groups. For the group of Mucormycetes, MICs were considerable higher than for the yeasts, in many strains not reaching 90% growth inhibition at the concentrations tested. Again, the lowest MICs were obtained for 3b against this group of fungi (MIC range 4->16 µg/mL). As Mucormycetes exhibit high resistance to commonly used antifungals, having a compound at hand causing complete growth inhibition when applied on spores is a promising result [24].
A number of structure-activity relationships can be deduced from these screening results: Both a benzyl and a phenylethyl substituent at the piperidine nitrogen can lead to high antifungal activity (see compounds 2b and 3b), as long as they are combined with N-alkyl substituents with more than seven carbon atoms at the 4-amino group. Shorter, branched, or cyclic alkyl residues at the 4-amino group are detrimental to activity; the same holds for most of the arylalkyl residues (except 4-tert-butylbenzyl in 2a). Outstanding antifungal activity was found for the N-dodecyl (C 12 ) residue (see 2b and 3b). In case this residue was attached to the 4-amino group, even compounds that are unsubstituted at the piperidine nitrogen (5b) or substituted with a Boc group there (4b) showed noteworthy activity. These findings for the N-dodecyl residue are in good accordance with SAR detected for antifungal N-alkyl perhydroisoquinolines [9] and N-alkyl perhydroquinolines [10], ergosterol biosynthesis inhibitors with amorolfine-like mode of action (inhibition of the enzymes sterol C14-reductase and/or sterol C8-isomerase).
Acylation of the secondary exocyclic amino group (6a-f, Supporting Information, Table S1) led to a virtually complete loss of antifungal activity; the same holds for the introduction of small (7a,b, Supporting Information, Table S1) or large residues there (8d/e, Supporting Information, Table S1).

Evaluation of the Antifungal Activity of 2b and 3b on Clinical Isolates
The compounds 2b and 3b showed the most promising growth-inhibiting activity on Aspergillus spp. and Candida spp. In the next step, we determined the antifungal activity on a greater collection of clinical isolates of Aspergillus spp. (n = 18) and Candida spp. (n = 19), to rule out strain specific differences in antifungal susceptibility ( Table 2). In addition to MIC values, the minimal fungicidal concentrations (MFCs) were determined for selected Candida and Aspergillus isolates. MFCs are used to characterize the antifungal activity either into fungistatic or fungicidal. As seen before in the small set of strains, the MIC values for 2b and 3b were lower compared to those for amorolfine hydrochloride. In the case of C. krusei, the MIC values for amorolfine hydrochloride were highly variable between the individual isolates while they were consistent for the new substances. Compound 3b showed lower MICs than 2b. For the Aspergillus spp., MICs could be defined for all strains ranging from 4 µg/mL to 16 µg/mL, while no MIC 90 was detected for amorolfine hydrochloride at the concentrations tested. Interestingly, while no MFC could be observed for any of the tested species for amorolfine hydrochloride, the new compounds led to a clearly defined MFC (Supporting Information, Figure S1), resulting in no growth/no colony-forming units (CFUs) at concentrations that resembled the MIC or were one dilution step higher than the MIC. These data indicate that contrary to amorolfine hydrochloride, the novel substances do exhibit fungicidal activity on selected Candida and Aspergillus isolates at concentrations that resemble the MIC, or only one dilution step higher. These data point to a higher antifungal activity of the novel compounds compared to the approved ergosterol biosynthesis inhibitor amorolfine hydrochloride. Rex et al. [25] have already pointed out in a review that MFCs might be the values more relevant to predict clinical outcome compared to solely MIC values. Table 2. Antifungal activity of amorolfine hydrochloride (A), 2b, and 3b against clinical isolates (Aspergillus and Candida species). MIC ranges represent the MIC 90 values obtained for 6 Aspergillus isolates and the MIC 80 values obtained for 6-7 isolates of each Candida species in all experiments. MFC ranges represent the MFCs determined for selected Aspergillus (1 per species) and Candida (1 isolate of C. albicans and C. tropicalis and 2 isolates of C. glabrata and C. krusei) isolates; () = tested fungal isolates in total. All experiments were carried out in duplicates.

Evaluation of Toxicity in Human Cell Lines and an Alternative In Vivo Model
In order to evaluate potential development of the compounds as antifungals applied in mammalian systems, toxicity tests were carried out with 2b and 3b on 3 different human cell lines in a standard proliferation assay [26]. For this purpose, we used HL-60 cells, HUVEC cells, and MCF10A. HUVEC cells are primary endothelial cells that are commonly used in research to assess cytotoxicity for the blood vessel system [27], MCF10A cells represent a healthy epithelial cell line that is frequently used in cytotoxicity studies [28], and HL-60 cells are used as an alternative to primary neutrophils and used to assess cytotoxicity for immune cells [29]. By combining cytotoxicity studies in these cell lines, we gained data on general toxicity of the compounds in human cells. The antifungals amorolfine hydrochloride (used as topical formulation), posaconazole, and voriconazole (both used for the treatment of invasive fungal infections) were used as reference drugs. The results are shown in Table 3. Table 3. Cytotoxic activity of selected compounds against human cell lines. Cell viability was determined by MTT assay (HL-60) and CTB assay (HUVEC, MCF10A); mean IC 50 values (n = 3) for amorolfine hydrochloride (A), posaconazole (P), and voriconazole (V), and for the promising compounds 2b and 3b are shown.

Cell Line
IC 50 (µM (µg/mL)) Compared to the reference drugs amorolfine hydrochloride and voriconazole, compounds 2b and 3b showed enhanced cytotoxicity. In HL-60 cells, the IC 50 value for posaconazole was in the same range as compound 2b and 3b.
Galleria mellonella larvae have been widely used as an alternative infection in vivo model for fungal diseases [30] and have also been utilized to study the in vivo activity and toxicity of commonly used or novel antimicrobial agents [31]. Therefore, the larval system was used to determine a potential impact on survival of larvae injected with three different doses of either compound 2b or 3b; amorolfine hydrochloride was used for comparison. All dilutions were made in PBS, which served as negative control. The Kaplan Meyer curves ( Figure 2) show that none of the compounds tested significantly reduced survival compared to PBS control (log rank test; p < 0.05), indicating no toxic activity in this model system. These results are promising for the further testing of the in vivo activity of the new compounds in animal models. Nevertheless, the survival data presented do not give insight into the potential impact these compounds might have on the larval immune system, the stability of the compounds in the larval hemolymph, and their tissue availability [32]. Further assays are necessary to investigate in detail (a) whether the larval model is suitable to test the in vivo efficiency of these novel compounds, and (b) whether treatment leads to an increase in the survival of fungus-infected larvae.

Identification of Target Enzymes in Ergosterol and Cholesterol Biosynthesis
The orienting tests for antifungal activity shown in Section 2.2.1 clearly indicated SAR related to those identified by us for antifungal N-alkyl perhydroisoquinolines [9] and N-alkyl perhydroquinolines [10] before. In the isoquinoline series, we found that even almost equipotent compounds can inhibit different target enzymes. In contrast to amorolfine and related N-alkylmorpholines/-piperidines, which inhibit both sterol C8-isomerase and sterol C14-reductase, these compounds were found to be inhibitors of either one or the other of these two enzymes. The N-alkyl perhydroquinolines, however, exclusively inhibit the enzyme sterol C8-isomerase. This prompted us to investigate the effect of the top compounds 2b and 3b from this investigation on the post-lanosterol part of ergosterol biosynthesis using a cellular assay [3].
Since ergosterol biosynthesis in fungi and cholesterol biosynthesis in humans are very similar and share numerous closely related enzymes, we investigated the effects of these compounds on cholesterol biosynthesis in a cellular assay [35] as well.
In the cholesterol biosynthesis assay, we investigated, in addition to the most active compounds 2b/3b (Table 5), a number of additional compounds (2c, 2f, 2g, 3c, 4b, 4c, 5b, and 6b; see Supporting Information, Table S2) from this library in order to get a first insight into SAR on this alternative target as well. Not unexpectedly, most of the compounds showed impact on enzymes of the post-lanosterol part of this biosynthesis, with multi-enzyme inhibition dominating at higher concentrations and inhibition of both sterol C8-isomerase and sterol C14-reductase at lower concentrations. The analysis of the sterol pattern of the reference inhibitor amorolfine hydrochloride (Table S2) showed only an accumulation of cholesta-8,14-dien-3β-ol, which indicates inhibition of sterol C14-reductase and sterol C8-isomerase at every test level (0.1 µM, 1 µM, 10 µM). Antifungal compound 2b (Table 5) as well as 2f (Supporting Information, Table S2) showed inhibition of sterol C8-isomerase and sterol C14-reductase at 0.1 µM, and a multi-enzyme inhibition at higher concentrations. The other strong antifungal compound, 3b (Table 5), was identified as a multi-enzyme inhibitor at 1 µM (inactive at 0.1 µM, toxic at 10 µM). Only compound 4c (Supporting Information, Table S2) was selective for sterol C14-reductase and sterol C8-isomerase at 1 µM and 10 µM (inactive at 0.1 µM). Notably, compound 2g did not show any effect on cholesterol biosynthesis.

Conclusions
A library of more than 30 novel 4-aminopiperidines was prepared by reductive amination of 4-piperidone derivatives with a broad variety of aliphatic amines. A screening on the model yeast Yarrowia lipolytica disclosed that compounds 2b and 3b are almost equipotent to established antifungals. In tests on clinically relevant species (Candida spp., Aspergillus spp., Mucormycetes), compounds 2b and 3b favourably compared with the approved antifungals amorolfine and voriconazole. Analysis of SAR revealed that the combination of a benzyl or phenylethyl residue at the piperidine nitrogen with an n-dodecyl residue at the 4-amino group is most beneficial to enhance antifungal activity. The two top compounds 2b and 3b are, as expected, inhibitors of the fungal ergosterol biosynthesis enzymes sterol C14-reductase and sterol C8-isomerase. However, additional molecular targets cannot be excluded. Determination of MIC values and minimal fungicidal concentrations (MFCs) for selected Candida and Aspergillus isolates revealed that, contrary to amorolfine, the novel substances do exhibit fungicidal activity. Antifungal activity was further determined on a greater collection of clinical isolates of Aspergillus spp. and Candida spp., and here 3b was clearly superior to amorolfine. The top compounds 2b and 3b exhibit cytotoxicity on human cell lines, but not on the Galleria mellonella larvae in an alternative (more complex) test system. The moderate cytotoxicity on human cell lines can in part be explained by inhibition of cholesterol biosynthesis. This aspect needs deeper investigation in the future if these or related compounds are considered for further development.
The 4-aminopiperidine core has been identified as an interesting lead structure for development of novel antifungals. It might be the basis for the development of antifungals targeting (at least among others) ergosterol biosynthesis. Due to the easy availability starting from cheap building blocks, application as agrofungicides might be considered as well. In total, 1.0 mmol of the ketone and 1.5 mmol of the amine were dissolved in 20 mL dry THF, and 2.0 mmol sodium triacetoxyborohydride was added. The suspension was stirred at room temperature for 12 h. Then, 20 mL of a saturated aqueous NaHCO 3 solution was added and the mixture was extracted with ethyl acetate (3 × 20 mL). The combined organic layers were dried over Na 2 SO 4 and the solvent was evaporated. The residue was purified by flash column chromatography (ethyl acetate:triethylamine 10:1).

General Procedure 2 (Cleavage of Boc-Protecting Group)
The Boc-protected amine (1.0 mmol) was dissolved in 20 mL dichloromethane, and 10 mL trifluoroacetic acid was added. The solution was stirred at room temperature for 8 h and then 20 mL 10% aqueous sodium hydroxide solution was added. The mixture was extracted with dichloromethane (3 × 20 mL) and the combined organic layers were dried over Na 2 SO 4 . The solvent was evaporated and the residue was purified by flash column chromatography (ethyl acetate:triethylamine 10:1) [10].

General Procedure 3 (Synthesis of Amides)
In total, 1.0 mmol of the amine was dissolved in 20 mL toluene and 1.2 mmol of the acid chloride was added. After addition of 3.0 mL triethylamine, the mixture was stirred at room temperature for 6 h. The solvent was evaporated and the residue was dissolved in 20 mL 2 M aqueous sodium hydroxide solution and extracted with ethyl acetate (3 × 20 mL). The combined organic layers were dried over Na 2 SO 4 and the solvent was evaporated. The residue was purified by flash column chromatography (ethyl acetate:triethylamine 10:1).
As compounds 2b and 3b exhibited promising growth-inhibiting activity at low concentrations, minimal fungicidial concentrations (MFCs) were determined according to literature [36,37] against selected Candida and Aspergillus isolates. In brief, 100 µL of samples containing the MIC concentration, one dilution step lower than the MIC, and 1 and 2 dilution steps higher than the MIC were plated on Sabouraud agar plates and the number of CFUs was compared to the growth control (no antifungal drug). Amorolfin hydrochloride was used for comparison. Fungicidial and fungistatic activity were defined according to Warn et al. [36].
Larvae were incubated at 37 • C and survival was monitored every 24 h over 4 days. The average survival rate of two independent experiments (40 larvae in total) was plotted in Kaplan Meier curves and statistical difference determined by log rank test (utilizing GraphPad Prism). The concentrations to be tested were selected based on the average MIC values determined for each drug. Applying 20 µL of 100 µM represents the MIC (4 µg/mL) of 2b and 3b per g larvae, 500 µM represents the MIC obtained for amorolfine hydrochloride (16 µg/mL), and additionally 10-fold the MIC of the novel substances was chosen.
Supplementary Materials: The following are available online: Supporting Information Table S1: Results of the antifungal activity screening for all compounds on Yarrowia lipolytica; Supporting Information Table S2: Results of the identification of the target enzyme(s) in cholesterol biosynthesis for compounds 2b, 2c, 2f, 2g, 3b, 4b, 4c, 5b, and 6b on HL-60 cells; Supporting Information Table S3: Fungal strains used in this study; Supporting Information Figure S1: Exemplary antifungal activity of 2b, 3b and amorolfine (A) against the two major fungal pathogens, A. fumigatus and C. albicans.