Mixture-Based Heterocyclic Combinatorial Positional Scanning Libraries: Discovery of Bicyclic Guanidines Having Potent Antifungal Activities against Candida albicans and Cryptococcus neoformans

Abstract

A mixture-based synthetic combinatorial library of more than 100,000 bicyclic guanidines was generated in a positional scanning format and assayed for activity against Candida albicans. Potent individual bicyclic guanidines were directly identified following the screening of the library. Time-kill curve studies indicated bactericidal activities for the individual bicyclic guanidines. These compounds also showed potent activity against Cryptococcus neoformans. These studies demonstrate the value of using mixture-based combinatorial positional scanning libraries made up of heterocyclic compounds for the rapid identification of novel classes of antifungal compounds.

Candida albicans and Cryptococcus neoformans are two of the most common opportunistic fungi responsible for infections (15). In particular, C. albicans infections may become problematic in persons with suppressed immune systems and may induce oral candidiasis, esophageal candidiasis, and vaginal candidiasis (10). Candidiasis is the fungal infection most frequently associated with human immunodeficiency virus (HIV)-positive patients. C. neoformans is the causative agent of cryptococcosis, which is the leading cause of morbidity and mortality due to fungi in patients with AIDS (24). Amphotericin B is the most commonly used drug despite its toxic effect on humans. Newer classes of antifungals include azole derivatives, allylamines, thiocarbamates, and fluoropyrimidines (14). While these new classes of compounds are now frequently used in the treatment of fungal infections, resistance to these new drugs is rising (1), which clearly indicates an urgent need for new antifungal agents. To overcome rapid development of drug resistance, new agents should preferably have chemical characteristics that clearly differ from those of existing agents.

Synthetic combinatorial libraries (SCLs) have brought a new dimension to drug discovery by allowing the rapid screening of large numbers of novel chemical entities (reviewed in references 13 and 20). SCL approaches therefore enable hundreds to thousands of times more compounds to be synthesized and screened than do traditional approaches. Furthermore, the chemical diversity and the large number of compounds for each class of structures available in SCL format increase the probability of identifying biologically active compounds with chemical characteristics that are different from those of existing biologically active compounds. By such approaches, new antifungal and/or antimicrobial agents, ranging from peptides of different lengths (5, 6, 25) to peptidomimetics (21), polyamines (2, 18), and, more recently, heterocyclic molecules (3, 4), have been identified from SCLs.

SCLs can be generated by solid-phase or solution-phase methods, as single compound arrays or as mixtures of individual related compounds. The solid-phase method offers the advantage of having the capability of driving reactions on polymer supports to completion (often >99.8%), the ability to readily remove excess reagents or starting materials, and the ease of automation. The expanding development of standard organic reactions to solid-phase chemistry greatly facilitates the generation of small-organic-molecule SCLs (reviewed in reference 16). Another approach for the generation of small-organic-molecule SCLs involves the chemical transformation of existing peptide SCLs (this has been termed the “libraries from libraries” approach [9, 12, 21]), which relies on the straightforward synthesis of peptide SCLs and the use of simple chemical modifications. The use of mixture-based SCLs is comparable to the screening of natural product extracts and bacterial broths in that these sources of diversity are composed of complex mixtures of compounds. However, SCLs have a number of clear advantages. For example, the structural nature of the compounds making up the SCLs is known, the concentration of every individual compound within the libraries is approximately equal, and no synthetic hurdles exist once an individual active compound has been identified.

Heterocyclic compounds offer a high degree of structural diversity and have proven broadly and economically useful as therapeutic agents. For example, nitrocarbon heterocyclic structures are found in various therapeutic agents, including antibiotic, antihistamine, antiseptic, antiarrhythmic, and antirheumatic compounds (26). In a continuing effort to generate and screen various classes of antifungal compounds, potent antifungal bicyclic guanidines were identified from a mixture-based SCL composed of more than 100,000 compounds. This SCL was generated from acylated resin-bound dipeptides by using the libraries-from-libraries approach (9, 12, 21) and was prepared in a positional scanning (PS) format (23). As described below, the PS format enables the identification of active compounds by screening the entire SCL, without successive iterative deconvolution steps.

MATERIALS AND METHODS

Bicyclic guanidine synthesis.

The bicyclic guanidines were prepared by using simultaneous multiple synthesis technology (17) as described elsewhere (22). In brief, resin-bound dipeptides (having side chains R₂ and R₃ [Fig. 1]) were acylated with a range of carboxylic acids (having side chain R₁ [Fig. 1]) in the presence of hydroxybenzotriazole. The N-acylated dipeptides were then reduced with 1.0 M diborane in tetrahydrofuran at 65°C under nitrogen (9, 11). Cyclization occurred following treatment of the reduced acylated dipeptides with thiocarbonyldiimidazole in anhydrous dichloromethane. Following cleavage from the resin with anhydrous HF in the presence of anisole (19), the desired compounds were extracted and lyophilized. The identity and purity of each individual compound were analyzed by mass spectral analysis interfaced with a liquid-chromatography system (Finnigan LCQ) and/or analytical reverse-phase high-performance liquid chromatography (RP-HPLC) using a Vydac C₁₈ column and a Beckman System Gold HPLC. The compounds were purified by using a Waters Milliprep 300 preparative HPLC with a preparative Foxy fraction collector.

FIG. 1.

Representation of the bicylic guanidine PS-SCL. O is defined as a single building block; X is a mixture of building blocks.

Fungal strains and growth conditions.

C. albicans ATCC 10231 and C. neoformans ATCC 32045 were used in the bioassays (American Type Culture Collection, Manassas, Va.). The cultures were maintained on yeast medium (YM; Difco Laboratories, Detroit, Mich.) agar plates at 4°C. Prior to the assay, the cultures were grown on agar plates and incubated for 48 h at 30°C or 72 h at 26°C for C. albicans or C. neoformans, respectively. Two colonies of these newly grown fungal cultures were then inoculated in 5 ml of 2× YM broth, vortexed, and diluted 10-fold in 2× YM broth, for an approximate final assay concentration of 1 × 10⁵ to 5 × 10⁵ CFU/ml.

Susceptibility testing.

The MICs and 50% inhibitory concentrations (IC₅₀) were determined by a broth microdilution method according to the guidelines of the National Committee for Clinical Laboratory Standards. In 96-well tissue cultures plates, fungal suspensions in 2× YM broth were added to the mixtures or individual compounds dispensed at concentrations ranging from 1,000 to 1 μg/ml derived from serial twofold dilutions in sterile water. The plates were then incubated for 48 h at 30°C or 72 h at 26°C for C. albicans or C. neoformans, respectively. The relative percent growth of the fungi found for each test sample was determined by the optical density at 620 nm (OD₆₂₀) by using a Titertek Multiskan Plus apparatus. The MIC was defined as the lowest concentration of the test sample that resulted in ≤2% growth, and the IC₅₀ was defined as the test sample concentration that resulted in 50% growth inhibition (27). The IC₅₀ were calculated by using a sigmoidal curve-fitting software program (Graphpad Prism; ISI Software, San Diego, Calif.).

Time-kill procedure.

The bacteriostatic or bactericidal effects of the compounds were evaluated by carrying out time-kill curve studies. Aliquots (200 μl) of C. albicans (1 × 10⁵ to 5 × 10⁵ CFU/ml) in suspension in 2× YM broth were allowed to bind to an Eppendorf tube for 15 min at room temperature. An equal volume of test compound solution in H₂O was then added, and the tubes were incubated at 30°C for 1 to 48 h. Controls lacking test compounds were treated in an identical manner. Following incubation, 100 μl of each solution was diluted in YM broth (10⁻¹ to 10⁻⁶ dilution) and plated onto YM agar plates.

Hemolytic assay.

The compounds’ hemolytic activities were determined by using human erythrocytes (RBCs) (7). In 96-well tissue culture plates, 0.25% RBC suspensions were added to individual compounds dispensed at concentrations varying from 32 to 4 μg/ml derived from serial twofold dilutions in water. Following a 1-h incubation at 37°C, the plates were centrifuged at 2,800 rpm (1,420 × g) for 5 min. The supernatant was separated from the pellet and its OD₄₁₄ was measured. The hemolytic dose required to lyse 50% of the RBCs (HD₅₀) was calculated by using a sigmoidal curve-fitting software program (Graphpad Prism).

Statistical analyses.

One-way analysis of variance (ANOVA), combined with the Tukey posttest, was used to determine the relative significance of the activity of each mixture within a sublibrary (28). A P value of less than 0.05 was taken as significant.

RESULTS

Library description.

The bicyclic guanidine SCL was generated in a PS format by using 41 carboxylic acids at position R₁, 51 amino acids at position R₂, and 49 amino acids at position R₃ for a total of 102,459 (41 × 51 × 49) individual compounds (Fig. 1; Table 1). The complete PS-SCL was composed of three sublibraries, each of which had a single defined building block at one position and a mixture of building blocks at each of the other two positions. Each sublibrary contained the same total of 102,459 individual compounds. The pooling of each sublibrary varied based on the number of building blocks included at the defined position of that sublibrary. In total, the PS-SCL was composed of 141 separate mixtures (i.e., 41 + 51 + 49 = 141 samples to be assayed), each mixture containing 2,009 (41 × 49) to 2,499 (49 × 51) individual compounds, depending on the location of the defined position. The structure(s) of the active individual compound(s) present in the library can be directly determined from the screening of these 141 mixtures, since each individual compound is present in a single mixture in each of the three sublibraries.

TABLE 1.

Building blocks used in the bicyclic guanidine PS-SCL^a

No.	Building block
	R1	R2	R3
1	3-Phenylbutyric acid	l-Alanine	l-Alanine
2	m-Tolylacetic acid	l-Phenylalanine	l-Phenylalanine
3	3-Fluorophenylacetic acid	Glycine	l-Isoleucine
4	p-Tolylacetic acid	l-Isoleucine	l-Lysine
5	4-Fluorophenylacetic acid	l-Lysine	l-Leucine
6	3-Metoxyphenylacetic acid	l-Leucine	l-Methionine
7	4-Metoxyphenylacetic acid	l-Methionine	l-Arginine
8	4-Ethoxyphenylacetic acid	l-Arginine	l-Valine
9	3-(3,4-Dimethoxyphenyl)-propionic acid	l-Valine	l-Tyrosine
10	4-Biphenylacetic acid	l-Tyrosine	d-Alanine
11	3,4-Dimethoxyphenylacetic acid	d-Alanine	d-Phenylalanine
12	Phenylacetic acid	d-Phenylalanine	d-Isoleucine
13	Hydrocinnamic acid	d-Isoleucine	d-Lysine
14	4-Phenylbutyric acid	d-Lysine	d-Leucine
15	Butyric acid	d-Leucine	d-Arginine
16	Heptanoic acid	d-Arginine	d-Valine
17	Isobutyric acid	d-Valine	d-Tyrosine
18	(+/−)-2-Methylbutyric acid	d-Tyrosine	l-α-Aminobutyric acid
19	Isovaleric acid	l-α-Aminobutyric acid	α-Aminoisobutyric acid
20	3-Methylvaleric acid	l-Norvaline	l-Norvaline
21	4-Methylvaleric acid	d-Norvaline	d-Norvaline
22	t-Butylacetic acid	l-Norleucine	l-Norleucine
23	Cyclohexanecarboxylic acid	d-Norleucine	d-Norleucine
24	Cyclohexylacetic acid	l-Ornithine	l-Ornithine
25	Cyclohexylbutyric acid	l-3-(2-Naphthyl)-alanine	l-3-(2-Naphthyl)-alanine
26	Cycloheptanecarboxylic acid	d-3-(2-Naphthyl)-alanine	d-3-(2-Naphthyl)-alanine
27	Lactic acid	l-Cyclohexylalanine	l-Cyclohexylalanine
28	Acetic acid	d-Cyclohexylalanine	d-Cyclohexylalanine
29	Cyclobutanecarboxylic acid	l-Methionine sulfone	l-Methionine sulfone
30	Cyclopentanecarboxylic acid	l-p-Nitrophenylalanine	l-p-Nitrophenylalanine
31	3-Cyclopentylpropionic acid	d-p-Nitrophenylalanine	l-p-Nitrophenylalanine
32	3-Cyclohexylpropionic acid	l-p-Chlorophenylalanine	l-p-Chlorophenylalanine
33	4-Methyl-cyclohexanecarboxylic acid	d-p-Chlorophenylalanine	d-p-Chlorophenylalanine
34	4-t-Butyl-cyclohexanecarboxylic acid	l-p-Fluorophenylalanine	l-p-Fluorophenylalanine
35	2-Norbornaneacetic acid	d-p-Fluorophenylalanine	d-p-Fluorophenylalanine
36	1-Adamantaneacetic acid	l-ɛ-Acetyl-lysine	l-ɛ-Acetyl-lysine
37	2-Ethylbutyric acid	l-3-Pyridyl-alanine	l-3-Pyridyl-alanine
38	3,3-Diphenylpropionic acid	d-3-Pyridyl-alanine	d-3-Pyridyl-alanine
39	2-Methyl-4-nitroimidazolepropionic acid	l-Cyclohexylglycine	l-Cyclohexylglycine
40	Cyclopentylacetic acid	d-Cyclohexylglycine	d-Cyclohexylglycine
41	Indole-3-acetic acid	l-t-Glycine	l-t-Butyl-glycine
42		pFmoc-amino-l-phenylalanineb	p-Fmoc-amino-l-phenylalanine
43		p-Fmoc-amino-d-phenylalanine	p-Fmoc-amino-d-phenylalanine
44		O-Ethyl-l-tyrosine	O-Ethyl-l-tyrosine
45		O-Ethyl-d-tyrosine	O-Ethyl-d-tyrosine
46		l-Aspartic acid-β-fluorenylmethyl ester	p-Iodo-l-phenylalanine
47		d-Aspartic acid-β-fluorenylmethyl ester	p-Iodo-d-phenylalanine
48		p-Iodo-l-phenylalanine	O-Methyl-l-tyrosine
49		p-Iodo-d-phenylalanine	O-Methyl-d-tyrosine
50		O-Methyl-l-tyrosine
51		O-Methyl-d-tyrosine

^aSee reference 22. 

^bFmoc, 9-fluorenylmethoxycarbonyl. 

Screening of the bicyclic guanidine PS-SCL against C. albicans.

Each mixture was assayed in duplicate in three separate assays at four or eight concentrations derived from serial twofold dilutions starting at 250 μg/ml. The IC₅₀ varied from 18 to >250 μg/ml, and the MICs varied from 20 to >250 μg/ml (Fig. 2). The most active mixtures of each sublibrary are listed in Table 2. Larger differences were seen for the IC₅₀ relative to the MICs when mixtures in a given sublibrary were compared. These results confirm the value of determining the IC₅₀ when a large number of samples are compared for deconvolution purposes.

FIG. 2.

Activity of the bicyclic guanidine PS-SCL against C. albicans. Each graph represents the activity of a sublibrary, and each bar within a graph represents the inverse of the IC₅₀ of a separate mixture. The defined functionalities are numbered on the x axis as described in Table 1. The line represents the average activity for all of the mixtures making up the sublibrary.

TABLE 2.

Activities of the most active mixtures of the bicyclic guanidine PS-SCL against C. albicans

Building block	IC50 (μg/ml)	MIC (μg/ml)
R1 defined
4-t-Butyl-cyclohexanecarboxylic acid	19 ± 1.3	32–62
1-Adamantaneacetic acid	22 ± 7.8	32–62
Cyclohexylbutyric acid	40 ± 0.8	50–62
2-Norbornaneacetic acid	54 ± 10	125–250
Heptanoic acid	80 ± 3.3	125–250
3-Cyclopentylpropionic acid	80 ± 6.5	125–250
4-Methyl-cyclohexanecarboxylic acid	82 ± 2.4	125–250
Cyclohexylacetic acid	85 ± 17	125–250
3-Methylvaleric acid	119 ± 38	>250
R2 defined
l-Cyclohexylalanine	20 ± 1.1	32–62
d-Cyclohexylalanine	23 ± 5.9	32–62
l-Cyclohexylglycine	48 ± 8.5	70–125
d-Cyclohexylglycine	78 ± 1.8	125–250
d-Isoleucine	80 ± 1.1	125–250
d-p-Chlorophenylalanine	88 ± 23	>250
Glycine	138 ± 4.7	150–250
d-Leucine	169 ± 0.6	>250
R3 defined
l-Cyclohexylalanine	18 ± 1.2	20–32
d-Cyclohexylalanine	19 ± 0.6	25–32
d-Cyclohexylglycine	34 ± 1.1	40–62
l-Cyclohexylglycine	37 ± 4.8	45–62
l-p-Fluorophenylalanine	94 ± 15	>250
d-p-Fluorophenylalanine	97 ± 16	125–250
d-Norleucine	105 ± 23	>250
l-ɛ-Acetyl-lysine	145 ± 22	>250

To confirm the significance of the differences in activity observed between the mixtures in each sublibrary, statistical analyses were carried out. ANOVA was performed with the IC₅₀ derived from three different assays to calculate the probability values of significance (P values), combined with a multiple-comparison Tukey posttest using a 5% overall risk of false-positive results (generated with Graphpad Prism software). The results are represented in Fig. 3. Only data that are not in the same box are statistically significantly different from each other. Thus, while the two most active mixtures of the sublibraries having position R₂ or R₃ defined are significantly more active than the other mixtures, the relative significances of the activities of the mixtures having position R₁ defined were not as obvious. The statistical analysis indicated no significant difference in activity between the three most active mixtures of the sublibrary having position R₁ defined, but it indicated that the most active mixture (R₁ derived from 4-t-butyl-cyclohexanecarboxylic acid) is significantly more active than the fourth most active (R₁ derived from 2-norbornaneacetic acid). Based on these results, a set of active mixtures was selected for carrying out the deconvolution process.

FIG. 3.

Results of the Tukey test of SCL activity against C. albicans. Each graph represents the IC₅₀ of all mixtures within a sublibrary, sorted in increasing order. Values that are not statistically significantly different from one another (P ≥ 0.05) are grouped in a box.

Antifungal activities of the individual bicyclic guanidines derived from the PS-SCL.

A set of 32 individual bicyclic guanidines was then generated to confirm the connectivity between the selected building blocks (i.e., if the activities of the selected mixtures are due to the same individual compounds), as well as to determine the relative activities of the individual compounds. These bicyclic guanidines were derived from all possible combinations of two carboxylic acids at position R₁ (4-t-butyl-cycohexanecarboxylic acid and 1-adamantaneacetic acid), four amino acids at position R₂ (l-cyclohexylalanine, d-cyclohexylalanine, l-cyclohexylglycine, and d-cyclohexyglycine), and four amino acids at position R₃ (l-cyclohexylalanine, d-cyclohexylalanine, d-cyclohexylglycine, and l-cyclohexylglycine). Eighteen additional bicyclic guanidines were used as control compounds to confirm that the activity was specific to the combination of all selected building blocks rather than to the guanidinium backbone and/or only one of the selected building blocks. In 12 of the control compounds, one position was occupied by one of the selected building blocks while the other two positions were occupied by negative-control building blocks: acetic acid for position R₁, d-leucine or l-phenylalanine for position R₂, and l-phenylalanine for position R₃. Thus, two compounds had 1-adamantaneacetic acid at position R₁, d-leucine at position R₂, and d-cyclohexylglycine or l-cyclohexylglycine at position R₃ (Table 3, control compounds 1 and 2). Two other control compounds each had l-phenylalanine at positions R₂ and R₃ and one of the selected carboxylic acids at position R₁ (Table 3, control compounds 3 and 4); four compounds each had acetic acid at position R₁, l-phenylalanine at position R₃, and one of the four selected amino acids at position R₂ (Table 3, controls 5 to 8); and four compounds each had acetic acid at position R₁, l-phenylalanine at position R₂, and one of the four selected amino acids at position R₃ (Table 3, controls 9 to 12). Building blocks defining mixtures that did not exhibit activity at the screening level were selected to generate the last six control compounds (Table 3, controls 13 to 18).

TABLE 3.

Antifungal activities of bicyclic guanidines derived from the PS-SCL

Type of compound	Building blocka			C. albicans		C. neoformans		Hemolysis
	R1	R2	R3	IC50 (μg/ml)	MIC (μg/ml)	IC50 (μg/ml)	MIC (μg/ml)	HD50 (μg/ml)	Indexb
									C. albicans	C. neoformans
Selected	4-tBu-cyclohexanecarboxylic acid	l-chAla	d-chAla	2.3 ± 0.1	3–4	5.5 ± 2.9	8–16	9.6 ± 2.2	4.1	1.7
	1-Adamantaneacetic acid	l-chAla	d-chGly	2.4 ± 0.1	3–4	2.6 ± 0.8	4–8	5.5 ± 1.4	3.5	2.1
	1-Adamantaneacetic acid	d-chAla	l-chGly	2.5 ± 0.3	3–4	2.1 ± 0.6	3–4	4.5 ± 0.2	1.8	2.1
	1-Adamantaneacetic acid	l-chAla	d-chAla	2.9 ± 0.7	5–8	2.7 ± 1.4	4–8	6.8 ± 0.7	2.3	2.5
	1-Adamantaneacetic acid	d-chAla	l-chAla	3.0 ± 0.9	4–8	2.3 ± 0.7	4–8	4.8 ± 0.5	1.6	2.1
	4-tBu-cyclohexanecarboxylic acid	l-chAla	d-chGly	3.5 ± 1.2	5–8	2.8 ± 1.4	8–16	8.0 ± 2.5	2.3	2.9
	4-tBu-cyclohexanecarboxylic acid	l-chAla	l-chAla	4.2 ± 0.2	5–8	2.2 ± 0.6	8–16	6.0 ± 1.6	1.4	2.7
	4-tBu-cyclohexanecarboxylic acid	d-chAla	l-chAla	4.4 ± 0.1	5–8	3.4 ± 1.7	8–16	6.5 ± 1.0	1.5	1.9
	1-Adamantaneacetic acid	l-chAla	l-chAla	4.4 ± 0.2	5–8	5.7 ± 3.8	8–16	10 ± 2.5	2.4	1.8
	4-tBu-cyclohexanecarboxylic acid	d-chAla	d-chAla	4.5 ± 0.1	5–8	3.7 ± 1.7	4–8	6.7 ± 1.2	1.5	1.8
	4-tBu-cyclohexanecarboxylic acid	d-chAla	l-chGly	4.6 ± 0.2	5–8	1.7 ± 0.7	2–4	5.5 ± 0.8	1.2	3.2
	4-tBu-cyclohexanecarboxylic acid	d-chAla	d-chGly	6.6 ± 0.5	8–16	4.7 ± 1.7	8–16	14 ± 2.7	2.1	3.0
	1-Adamantaneacetic acid	d-chAla	d-chAla	7.3 ± 1.7	8–16	4.1 ± 1.9	16–32	11 ± 1.2	1.5	2.7
	4-tBu-cyclohexanecarboxylic acid	l-chGly	l-chAla	7.3 ± 1.9	8–16	2.3 ± 1.2	4–8	7.3 ± 0.2	1.0	3.2
	1-Adamantaneacetic acid	l-chGly	d-chAla	8.3 ± 0.7	10–16	2.1 ± 1.3	4–8	6.6 ± 0.5	0.8	3.1
	4-tBu-cyclohexanecarboxylic acid	l-chAla	l-chGly	8.7 ± 0.3	10–16	4.1 ± 2.2	8–16	16 ± 2.5	1.8	3.9
	4-tBu-cyclohexanecarboxylic acid	d-chGly	d-chAla	8.7 ± 0.9	10–16	2.0 ± 0.7	4–8	5.4 ± 0.5	0.6	2.7
	1-Adamantaneacetic acid	l-chGly	d-chGly	9.0 ± 0.1	10–16	2.4 ± 1.5	4–8	17 ± 2.4	1.8	7.1
	1-Adamantaneacetic acid	d-chGly	d-chAla	9.1 ± 0.1	10–16	3.7 ± 1.4	8–16	8.9 ± 0.3	1.0	2.4
	4-tBu-cyclohexanecarboxylic acid	d-chGly	l-chAla	9.2 ± 0.5	10–16	3.5 ± 1.4	8–16	18 ± 2.8	2.0	5.1
	1-Adamantaneacetic acid	l-chGly	l-chAla	9.3 ± 0.1	10–16	3.2 ± 2.0	16–32	9.7 ± 0.6	1.1	3.0
	1-Adamantaneacetic acid	d-chGly	l-chAla	9.4 ± 0.7	10–16	3.6 ± 1.7	8–16	10 ± 1.0	1.1	2.8
	4-tBu-cyclohexanecarboxylic acid	l-chGly	d-chAla	9.7 ± 0.1	16–32	2.1 ± 0.7	8–16	23 ± 2.9	2.4	11
	1-Adamantaneacetic acid	d-chGly	l-chGly	9.9 ± 0.1	11–16	2.5 ± 0.3	4–8	14 ± 1.9	1.4	5.6
	4-tBu-cyclohexanecarboxylic acid	l-chGly	d-chGly	16.8 ± 1.8	18–32	3.7 ± 0.9	4–8	28 ± 1.8	1.7	7.6
	1-Adamantaneacetic acid	l-chGly	l-chGly	18.6 ± 1.2	20–32	2.2 ± 0.5	8–16	21 ± 1.3	1.1	9.5
	1-Adamantaneacetic acid	d-chGly	d-chGly	23.1 ± 2.2	32–64	3.8 ± 0.8	8–16	45 ± 1.6	2.0	12
	1-Adamantaneacetic acid	d-chAla	d-chGly	23.8 ± 1.1	32–64	9.9 ± 3.9	16–32	52 ± 2.3	2.2	5.2
	4-tBu-cyclohexanecarboxylic acid	l-chGly	l-chGly	24.1 ± 7.4	32–64	3.3 ± 0.8	4–8	39 ± 1.6	1.6	12
	4-tBu-cyclohexanecarboxylic acid	d-chGly	l-chGly	29.3 ± 7.8	>64	4.5 ± 2.2	8–16	42 ± 2.4	1.4	9.3
	4-tBu-cyclohexanecarboxylic acid	d-chGly	d-chGly	34.2 ± 9.3	>64	3.6 ± 1.4	4–8	50 ± 2.4	1.4	14
	1-Adamantaneacetic acid	l-chAla	l-chGly	38.8 ± 2.7	>64	15.0 ± 5.7	16–32	46 ± 1.3	1.2	3.1
Control
1	1-Adamantaneacetic acid	d-Leu	l-chGly	26.6 ± 11	>64	4.5 ± 0.2	5–8	21 ± 0.2	0.8	4.7
2	1-Adamantaneacetic acid	d-Leu	d-chGly	48.6 ± 3.7	>64	9.1 ± 0.1	10–16	>32	ND	>3.5
3	1-Adamantaneacetic acid	l-Phe	l-Phe	24.9 ± 3.0	32–64	6.3 ± 4.2	16–32	20 ± 1.8	0.8	3.2
4	4-tBu-cyclohexanecarboxylic acid	l-Phe	l-Phe	50.6 ± 1.8	>64	2.5 ± 0.3	4–8	18 ± 2.8	0.4	7.2
5	Acetic acid	l-chAla	l-Phe	>64	>64	41.0 ± 6.4	62–125	>32	ND	>13
6	Acetic acid	d-chAla	l-Phe	>64	>64	37.9 ± 4.2	62–125	>32	ND	ND
7	Acetic acid	l-chGly	l-Phe	>64	>64	64.6 ± 11	>25	>32	ND	ND
8	Acetic acid	d-chGly	l-Phe	>64	>64	38.6 ± 1.7	62–125	>32	ND	ND
9	Acetic acid	l-Phe	l-chAla	>64	>64	45.2 ± 0.8	62–125	>32	ND	ND
10	Acetic acid	l-Phe	d-chAla	>64	>64	37.7 ± 4.1	62–125	>32	ND	ND
11	Acetic acid	l-Phe	l-chGly	>64	>64	72.2 ± 24	>125	>32	ND	ND
12	Acetic acid	l-Phe	d-chGly	>64	>64	34.5 ± 2.2	62–125	>32	ND	ND
13	Butyric acid	l-Phe	l-Phe	>250	>250	ND	ND	>32	ND	ND
14	3,4-Dimethoxyphenylacetic acid	l-Phe	l-Phe	>250	>250	ND	ND	>32	ND	ND
15	Acetic acid	d-Ala	l-Phe	>250	>250	ND	ND	>32	ND	ND
16	Acetic acid	d-Arg	l-Phe	>250	>250	ND	ND	>32	ND	ND
17	Acetic acid	l-Phe	Aib	>250	>250	ND	ND	>32	ND	ND
18	Acetic acid	l-Phe	l-tBu-Gly	>250	>250	ND	ND	>32	ND	ND

^at-Bu, t-butyl; l-chAla, l-cyclohexylalanine; d-chAla, d-cyclohexylalanine; l-chGly, l-cyclohexylglycine; d-chGly, d-cyclohexylglycine; d-Leu, d-leucine; l-Phe, l-phenylalanine; Aib, α-aminoisobutyric acid; ND, not determined. 

^bDefined as HD₅₀/IC₅₀. 

The activities of the 44 bicyclic guanidines against C. albicans were evaluated. As shown in Table 3, most of them exhibited potent antifungal activity, with MICs of 3 to 4 μg/ml for the most active compounds. To evaluate the effect of each selected building block on the activity of the individual compounds and to analyze the connectivity between the selected building blocks, the overall averages of the IC₅₀ of individual compounds having one or two common building blocks were calculated. The average IC₅₀ of all of the individual compounds that had 1-adamantaneacetic acid at position R₁ was similar to the average IC₅₀ of all of the individual compounds that had 4-t-butyl-cyclohexanecarboxylic acid at this position (average activity for 16 compounds each; Fig. 4). This is in agreement with the similar activities found for the relevant bicyclic guanidine mixtures (Table 2). Also in agreement with the relative activities found for the relevant mixtures are the higher average IC₅₀ observed for all of the individual compounds deriving from l- or d-cyclohexylalanine at position R₂ or R₃ (average activities of eight compounds each; Fig. 4). Upon evaluation of the connectivity between positions R₁ and R₂, higher average activity was found for those compounds with 4-t-butyl-cyclohexanecarboxylic acid at position R₁ and l- or d-cyclohexylalanine at position R₂ (average activities of four compounds each; Fig. 4). In contrast, the presence of l- or d-cyclohexylalanine at position R₃ resulted in a higher average activity independently of the selected carboxylic acids at position R₁ (average activities of four compounds each; Fig. 4). Upon evaluation of the connectivity between positions R₂ and R₃, a higher average activity was found when an l-amino acid was connected to a d-amino acid, especially with l- or d-cyclohexylalanine at position R₂ (average activities of two compounds each; Fig. 4).

FIG. 4.

Evaluation of the effects of the selected building blocks on the anticandidal activity of the individual bicyclic guanidines. Each bar represents the inverse of the average IC₅₀ of individual bicyclic guanidines having one or two common building blocks at a given position or positions. The common position(s) is described above each group of bars (i.e., R₁, R₂, and R₃ for a single common position, and R₁/R₂, R₁/R₃, and R₂/R₃ for two common positions). For a single common position, each bar pattern represents the common building block. For two common positions, one of the common building blocks is identified above each group of bars (R₁ for R₁/R₂ and R₁/R₃, and R₂ for R₂/R₃), and each bar represents the common building block at the other position.

As anticipated from the screening results of the PS-SCL, the control compounds exhibited lower activity or no activity relative to the selected compounds. For example, when a mixture having lower activity was selected at position R₂ (defined with d-leucine), the derived individual compounds had lower activities than their analogs with l-cyclohexylalanine or l-cyclohexylglycine at this position. Interestingly, although overall activities were lower, a d/l configuration at positions R₂/R₃ in the d-leucine-derived bicyclic guanidines also led to higher activity than a d/d configuration. Finally, the poor activities of the l-phenylalanine-derived control compounds confirm that the activities observed for the identified bicyclic guanidines are due to the combination of the selected building blocks at the three positions rather than the overall backbone structure of the molecule and/or a single selected building block.

Viable counts of C. albicans colonies in the presence of three of the bicyclic guanidines at a concentration equal to 4 times the MIC (the two most active compounds and one of the least active compounds) were then carried out to determine if these representative bicyclic guanidines were bacteriostatic or bactericidal. As shown in Fig. 5, the three bicyclic guanidines have bactericidal activity against C. albicans.

FIG. 5.

Time-kill curve of C. albicans by three bicyclic guanidines. The mean log₁₀ CFU per milliliter are plotted versus incubation time for C. albicans tested against the bicyclic guanidines. ■, R₁ = 1-adamantaneacetic acid, R₂ = l-cyclohexylalanine, R₃ = d-cyclohexylglycine; ⧫, R₁ = 4-t-butyl-cyclohexanecarboxylic acid, R₂ = l-cyclohexylalanine, R₃ = d-cyclohexylalanine; ▴, R₁ = 1-adamantaneacetic acid, R₂ = d-cyclohexylalanine, R₃ = d-cyclohexylglycine; ✶, control without test compounds.

Specificity of bicyclic guanidine activity.

To gain insight into the specificity of the newly identified bicyclic guanidines, their antifungal activities were also evaluated against the opportunistic fungus C. neoformans (Table 3). Higher overall activity and little variation between the 32 individual bicyclic guanidines was found against C. neoformans. In a manner similar to the anticandidal activity, the control compounds showed lower activities against C. neoformans relative to the 32 identified guanidines, except for four control compounds. These results suggest that the overall activity profiles of the PS-SCL against the two fungi are likely to differ. In an initial step, a selected number of mixtures were assayed against C. neoformans. These mixtures were defined by the building blocks used for the synthesis of the 32 individual bicyclic guanidines as well as the control compounds. As shown in Table 4, higher activities were also found against C. neoformans for the selected mixtures relative to their activities against C. albicans. In particular, the mixtures with l-phenylalanine or d-leucine at position R₂ or R₃ showed significant activities against C. neoformans. These results agree with the higher activities observed for the control compounds against C. neoformans relative to their anticandidal activity (Table 3). In further agreement with the relative antifungal activities found for the individual bicyclic guanidines, the presence of an acetic acid group at position R₁ resulted in lower activity against C. neoformans, as observed for the relevant mixture and individual bicyclic guanidines (Tables 3 and 4). These results again indicate that the selected building blocks chosen are responsible for the antifungal activities of the bicyclic guanidines.

TABLE 4.

Activities of selected mixtures of the bicyclic guanidine PS-SCL against C. neoformans

Building block	IC50 (μg/ml)	MIC (μg/ml)
R1 defined
4-t-Butyl-cyclohexanecarboxylic acid	2.3 ± 0.03	4–8
1-Adamantaneacetic acid	2.0 ± 0.52	4–8
Acetic acid	37 ± 9.4	50–62
R2 defined
l-Cyclohexylalanine	6.4 ± 0.26	16–32
d-Cyclohexylalanine	6.3 ± 0.46	8–16
l-Cyclohexylglycine	6.2 ± 0.35	16–32
d-Cyclohexylglycine	8.7 ± 0.21	16–32
d-Leucine	17 ± 1.19	32–62
l-Phenylalanine	15 ± 0.23	16–32
R3 defined
l-Cyclohexylalanine	5.4 ± 0.14	8–16
d-Cyclohexylalanine	4.1 ± 2.07	8–16
d-Cyclohexylglycine	5.7 ± 1.35	8–16
l-Cyclohexylglycine	6.7 ± 1.05	16–32
l-Phenylalanine	10 ± 3.70	16–32

The specificities of the bicyclic guanidines were also evaluated by determining their activities against mammalian eucaryotic cells by using a hemolytic assay. Thus, each of the 44 individual bicyclic guanidines was assayed against human RBCs at concentrations varying from 32 to 4 μg/ml. The HD₅₀ of each compound was calculated and compared to its IC₅₀ against the two fungi (Table 3). The resulting specificity indexes (HD₅₀/IC₅₀) were found to be structure dependent, and, as anticipated due to a higher overall level of antifungal activity, higher indexes were observed against C. neoformans. As shown in Table 3, the indexes varied from 0.4 to 4.1 and from 1.7 to 14 relative to activities against C. albicans and C. neoformans, respectively.

DISCUSSION

The number of fungal infections has dramatically increased with the emergence of AIDS. For example, by 1997 90% of AIDS patients had been infected by C. albicans, and C. neoformans is the most common life-threatening opportunistic infection in these patients (1). An important strategy to overcome antifungal drug resistance is the development of new drugs having a broad spectrum of activity. The present study shows the value of generating and screening small-molecule, mixture-based SCL in a PS format for the rapid identification of novel antifungal compounds. In particular, although the activities of the mixtures decrease gradually, these small differences in activity (from 2- to 10-fold) were found to be statistically significant and to reflect the antifungal activities of the derived individual compounds. Most of the reported antifungal or antimicrobial compounds that have been identified from combinatorial libraries are peptides or peptidomimetics (5). The development of SCLs of small organic molecules enhances the likelihood of discovering antifungal agents that are therapeutically relevant (i.e., that have higher stability and bioavailability relative to peptides and peptidomimetics).

The compounds identified in the present study from the bicyclic guanidine SCL have a common characteristic in that they derive from hydrophobic residues (aliphatic or aromatic) at positions R₂ and R₃, and contain a bulky aliphatic moiety at position R₁ (Fig. 6). This suggests a high affinity with the lipid and/or ergosterol component of the fungal cell wall, and the basic character of the guanidine backbone facilitates the initial electrostatic interactions with negatively charged phospholipids. The higher overall activity found for the compounds derived from l- or d-cyclohexylalanine compared with those derived from l- or d-cyclohexylglycine may be due to an expected higher flexibility of the cyclohexyl moiety resulting from an additional methylene group. Such an additional bond results in the cyclohexyl moiety being more distant from the guanidine moiety and, in the case of position R₂, from the aliphatic moiety at position R₁, which is emphasized in the case of R₁ derived from 4-t-butyl-cyclohexanecarboxylic acid. Higher flexibility is then anticipated to assist the cyclohexyl moieties in associating with the lipid groups of the fungal cell wall. As has been reported for polyene antifungal agents (8), this is expected to cause membrane disruption, increased permeability, leakage of cytoplasmic contents, and cell death.

FIG. 6.

Structures of four representative bicyclic guanidines identified from the SCL.