Synthesis and Bioactivities of Kanamycin B-Derived Cationic Amphiphiles

Abstract

Cationic amphiphiles derived from aminoglycosides (AGs) have been shown to exhibit enhanced antimicrobial activity. Through the attachment of hydrophobic residues such as linear alkyl chains on the AG backbone, interesting antibacterial and antifungal agents with a novel mechanism of action have been developed. Herein, we report the design and synthesis of seven kanamycin B (KANB) derivatives. Their antibacterial and antifungal activities, along with resistance/enzymatic, hemolytic, and cytotoxicity assays were also studied. Two of these compounds, with a C₁₂ and C₁₄ aliphatic chain attached at the 6″-position of KANB through a thioether linkage, exhibited good antibacterial and antifungal activity, were poorer substrates than KANB for several AG-modifying enzymes, and could delay the development of resistance in bacteria and fungi. Also, they were both relatively less hemolytic than the known membrane targeting antibiotic gramicidin and the known antifungal agent amphotericin B and were not toxic at their antifungal MIC values. Their oxidation to sulfones was also demonstrated to have no effect on their activities. Moreover, they both acted synergistically with posaconazole, an azole currently used in the treatment of human fungal infections.

Graphical abstract

INTRODUCTION

Aminoglycosides (AGs) are a valuable class of antibiotics that exhibit versatility in their biological activities.^1,2 Widely recognized for their antibacterial properties, naturally occurring AGs have also recently been investigated as antifungal agents.^3,4 One study evaluated the in vitro and in vivo activities of several AG antibiotics against six crop pathogenic oomycetes (Phytophthora and Pythium species) and 10 common fungi,³ while the other examined the in vitro activity of AGs against Pythium insidiosum, an aquatic oomycete that affects humans, mammals, and birds. These studies showed that, while gentamicin (GEN), neomycin B (NEO), paromomycin (PAR), and streptomycin (STR) might display antifungal activity at a concentration of 64 μg/mL, they would be unsuitable for therapeutic use because their antifungal MIC is higher than their safe serum concentration of 30–40 μg/mL.⁴

The economic and social burden associated with fungal infections is considerable.^5–7 Currently, amphotericin B (AmB) and azoles, such as fluconazole (FLC), itraconazole (ITC), and posaconazole (POS), are the drugs of choice in the treatment of fungal infections.⁸ However, the rapid increase in the occurrence of opportunistic fungal infections in immunocompromised or severely ill patients, the difficulty to treat systemic mycoses, and the recrudescence of fungal strains resistant to available drugs,⁹ along with the toxicity associated with azoles,¹⁰ call for the development of new and effective antifungal agents.

Our group and others have recently engaged in the development of semisynthetic AGs with improved antifungal activity. These were notably cationic amphiphilic AGs derived from neamine (NEA) (FG03 and FG08),^11,12 kanamycin A (KANA) (K20),^13,14 and tobramycin (TOB) (C₁₄).¹⁵ FG03 and FG08, obtained by glycodiversification, bear a C₈ linear alkyl chain at the 4″-hydroxyl, whereas K20 and C₁₄, obtained by direct modification of KANA and TOB, respectively, have an octanesulfonyl chain and a thiooctadecyl chain, respectively, at the 6″-position. In all cases, it was demonstrated that the hydrophobic moiety, that is, the linear alkyl chain, conferred some amphiphilic properties to the parental AGs and increased their affinity toward fungal cell membranes.^14–16 This eventually led to the disruption of the membranes in fungi. With K20 displaying unprecedented inhibitory capabilities toward fungi, and despite the fact that the clinically used¹⁷ kanamycin B (KANB) is a better antimicrobial agent than KANA, there has not been any report to our knowledge of the direct modification of KANB toward the development of antifungal cationic amphiphiles.

Herein, we present the synthesis and antibacterial activity of seven KANB derivatives. Also, the antifungal activities of these compounds alone and in combination with azoles are studied. In addition, we investigated the ability of the most active compounds to delay the development of bacterial and fungal resistance and determined spontaneous resistance frequency against a fungal strain. Finally, we report on their toxicity toward erythrocytes and mammalian cells, along with their time-kill study and mechanism of action against fungi. Two of these compounds, KANB derivatives with a C₁₂ and C₁₄ alkyl chain attached in a thioether bond at the 6″-position, prove to be very promising in all respects.

RESULTS AND DISCUSSION

Design and Chemical Synthesis

In designing the KANB derivatives for this study, our primary focus was to confer cationic amphiphilic properties to them. This could be achieved by attaching hydrophobic residues, such as linear alkyl chains, to the highly polar KANB. Similar techniques have been previously employed in the development of amphiphilic AGs and have led to the discovery of NEO, PAR, KANA, NEA, and amikacin (AMK) derivatives with enhanced antibacterial activity.^18–28 We envisaged that the length of these alkyl chains could range from C₈ to C₁₆, as these were previously shown to exhibit better potency.^11–13,26 Finally, the ease of synthesis of the KANB derivatives was another important criterion. Because K20 and C₁₄ were obtained by direct modification at the 6″-position of KANA and TOB, respectively, and hundreds of grams of K20 could be prepared, we reasoned that similar modification of KANB could be accomplished. By incorporating various linear alkyl chains (C₈–C₁₆) at the 6″-position of KANB through a thioether linkage, we could thus generate our target molecules.

In light of this, our KANB derivatives 3a–e were conveniently synthesized from the known precursor 1²⁹ via a facile two-step procedure that we recently utilized to generate TOB variants for their evaluation as ribosome-targeting agents³⁰ (Scheme 1). Nucleophilic displacement of the TIPBS group at the 6″-position in 1 by various alkanethiols resulted in the formation of intermediates 2a–e in good yields (67–78%). Acid treatment of 2a–e then afforded the expected products 3a–e as TFA salts in yields ranging from 82% to quantitative.

Scheme 1. Structure of Kanamycin B (KANB) and Its Derivatives Utilized in This Study^a.

^aReagents and conditions for the synthesis: (i) RSH, Cs₂CO₃, DMF, 67–78%; (ii) neat TFA, 82%–quantitative; (iii) m-CPBA/CHCl₃ then neat TFA, 88–98%.

Antibacterial Activity

The antibacterial activity of KANB and its derivatives 3a–e was evaluated against an array of Gram-positive and Gram-negative bacterial strains, including strains that are susceptible as well as some that are resistant to the parent AG KANB (Table 1). Among the 11 Gram-positive pathogens were the KANB-resistant and pathogenic methicillin-resistant Staphylococcus aureus (MRSA, strain III; S. aureus USA 100, strain VI; S. aureus USA 200, strain VII; S. aureus USA 300, strain VIII; and S. aureus USA 600, strain IX) and vancomycin-resistant enterococci (VRE, strain XI). The set of six Gram-negative bacteria also included a strain of Klebsiella pneumoniae (strain XV).

Table 1.

MIC Values (μg/mL)^a of KANA, KANB, 3a–e, and 4c, d against Various Gram-Positive and Gram-Negative Bacterial Strains

bacterial strains	KANA	KANB	3a	3b	3c	3d	3e	4c	4d
Gram-Positive
B. subtilis 168 (I)	2–4	4	>128	128	>128	>128	>128	>128	>128
L. monocytogenes ATCC 19115 (II)	>128	≤0.25	8	2	1	0.5–1	16	4–8	2–4
MRSAb (III)	>128	>128	128	32	32	32–64	>128	64	128
M. smegmatis MC2-155 (IV)	≤0.25	≤0.25	64	8	2	>128	8	4	2
S. aureus ATCC 29213 (V)	>128	≤0.25	64	16–32	8	16	>128	128	128
S. aureus NRS382 (USA 100) (VI)	>128	>128	128	32	8	32	128	64	64
S. aureus NRS383 (USA 200) (VII)	>128	128	64	16	4	8–16	>128	32	128
S. aureus NRS385 (USA 300) (VIII)	>128	32	128	32–64	16	16–32	>128	64	128
S. aureus NRS22 (USA 600) (IX)	>128	32	64	16	16	16	128	32	64
S. epidermidis ATCC 12228 (X)	1	0.5	>128	>128	>128	>128	>128	>128	>128
VREc (XI)	>128	>128	128	32	16	16	128	32	32
Gram-Negative
A. baumannii ATCC 19606 (XII)	4	4	>128	>128	>128	>128	>128	128	>128
E. cloacae ATCC 13047 (XIII)	2	0.5	>128	>128	>128	128	>128	>128	>128
E. coli MC1061 (XIV)	8–16	2	>128	>128	>128	>128	>128	>128	>128
K. pneumoniae ATCC 27736 (XV)	16–32	4	>128	>128	>128	>128	>128	128	>128
P. aeruginosa ATCC 27853 (XVI)	>128	16	64	32	32	64	>128	16	32
S. enterica ATCC 14028 (XVII)	16	2	>128	>128	>128	32	>128	64	128

^aAll experiments were performed at least in duplicate.

^bMRSA = methicillin-resistant Staphylococcus aureus.

^cVRE = vancomycin-resistant enterococci.

We also assessed the antibacterial activity of KANA. Except against B. subtilis 168 (strain I), M. smegmatis MC2-155 (strain IV), and S. epidermidis ATCC 12228 (strain X), KANA was ineffective at inhibiting the growth of our tested Gram-positive bacteria in vitro, with MIC values >128 μg/mL. Moreover, KANB was, in general, more potent than KANA, suggesting that the derivatives of the former AG might be better antibacterial agents.

KANB derivatives 3a–e exhibited a wide range of antibacterial activity against the tested panel of Gram-positive bacteria. All but 3e inhibited the growth of MRSA (strain III), S. aureus USA 100 (strain VI), S. aureus USA 200 (strain VII), and VRE (strain XI) at concentrations lower than that required by the parent AG. Furthermore, 3b–d showed an activity similar to that of KANB against the other two MRSA strains (VIII and IX). On the other hand, compounds 3a–e displayed a reduced antibacterial activity against L. monocytogenes ATCC 19115 (strain II), M. smegmatis MC2-155 (strain IV), and S. aureus ATCC 29213 (strain V) as compared to KANB. However, they still exhibited moderate to good antibacterial activity. Finally, while B. subtilis 168 (strain I) and S. epidermidis ATCC 12228 (strain X) were both susceptible to KANB (MIC = 4 and 0.5 μg/mL, respectively), none of the derivatives showed activity against these pathogens, displaying MICs ≥ 128 μg/mL. It is worthy to mention that as the length of the thioalkyl group increased from 8 to 12 carbons, the MIC value generally decreased. Once the chain was extended to 14 carbons, an increase in MIC value, sometimes not to a significant extent, was observed. Further elongation of the alkyl chain to 16 carbons resulted in a complete loss of antibacterial activity. Therefore, 3c, which carries a thiododecyl group (C₁₂) at the 6″-position, exhibited the optimal antibacterial activity, closely followed by 3d (C₁₄), then 3b (C₁₀), 3a (C₈), and 3e (C₁₆) successively. A similar trend was observed for TOB derivatives that also bear thioalkyl chains at the 6″-position, with their inhibitory efficacy improving from the compound with C₈ chain to C₁₄ chain.^15,26

Substitution of the 6″-hydroxyl group of KANB by a thioalkyl group appears to significantly decrease its inhibitory activity against Gram-negative bacteria, regardless of the chain length of the substituent. Indeed, while the Gram-negative bacterial strains tested were all susceptible to KANB (MIC ≤ 16 μg/mL), they regained some resistance in the presence of 3a–e (MIC >64 μg/mL). Only P. aeruginosa ATCC 27853 (strain XVI) showed intermediate sensitivity to all five KANB derivatives but 3e (MIC = 32–64 μg/mL), as well as S. enterica ATCC 14028 (strain XVII) to 3d (MIC = 32 μg/mL). Therefore, 3a–e exhibited higher potency against Gram-positive bacteria than Gram-negative bacteria, as observed with other AG cationic amphiphiles.^26,31

To evaluate the effect of metabolic S-oxidation on our thioethers, we oxidized our most potent KANB derivatives (3c and 3d) to their corresponding sulfones (4c and 4d; Scheme 1) and assessed their antibacterial activity (Table 1). An increase in MIC value of the sulfone analogues as compared to the parent thioethers was observed across the array of Gram-positive bacteria tested. Only 4d showed a 64-fold improvement in antibacterial activity against M. smegmatis MC2-155 (strain IV) with respect to its corresponding parent thioether 3d. On the other hand, both sulfones 4c and 4d exhibited comparable to slightly better antibacterial activity than their corresponding thioethers 3c and 3d against Gram-negative bacterial strains. Therefore, it appears that S-oxidation of our KANB derivatives to the corresponding sulfones may decrease their potency against Gram-positive bacteria, while improving their inhibitory efficacy against Gram-negative bacteria.

Resistance Studies

The recurring emergence of bacterial resistance to AGs remains one of the major threats associated with their long-term clinical use. To evaluate the ability of bacteria to develop resistance to our newly synthesized compounds, we decided to subject the most susceptible bacterial strain in our study (L. monocytogenes; strain II) to multiple rounds of MIC testing,³² taking subinhibitory concentrations (1/2 MIC) for the next round of testing for the parent KANB and the two most active derivatives 3c and 3d over 15 passages (Figure S25). Although the relative MIC value of 3c appears to increase by 4-fold as early as at the third passage, it took up to 6 passages to observe a 2-fold change in the MIC of 3d and a 4-fold change in the MIC of KANB. Surprisingly, compound 3c developed resistance faster than KANB, which may indicate a different method of resistance development. Furthermore, L. monocytogenes seems to be developing resistance to 3d slower than it does to KANB. This encouraging finding thus suggests that 3d will be able to delay the emergence of resistance as compared to KANB.

With more than 100 AG-modifying enzymes (AMEs) identified to date,³³ this family of enzymes represents the major causal agent of resistance in bacteria. The development of AG derivatives capable of evading the action of these enzymes thus seems a plausible avenue to overcome resistance issues. We decided to evaluate the ability of AMEs to modify our synthesized KANB derivatives 3a–e and 4c, d by measuring the relative activities of six enzymes with these KANB derivatives and comparing them to that of the parent AG, KANB (Figure 1). Among the six AMEs were APH(2″)-Ia,³⁴ AAC(3)-IV,³⁵ AAC(6′)-Ie/APH(2″)-Ia (used for its AAC(6′)-Ie activity only),³⁵ AAC(6′)-IId,³⁶ AAC(2′)-Ic,³⁷ and the multiacetylating enzyme Eis,^37,38 with the latter two both being specific to Mycobacterium tuberculosis. Overall, there was a noticeable decrease in activity of these AMEs with all seven derivatives 3a–e and 4c, d. This suggests that these enzymes have reduced effect on modifying our newly synthesized KANB derivatives. In addition, out of all six AMEs tested, AAC(6′)-IId and APH(2″)-Ia were the least active against 3a–e and 4c, d. Meanwhile, the multiacetylating enzyme Eis was the most active of them all. As the length of the alkyl chain attached to KANB increases, the effect of the AMEs seems to decrease from 3a to 3d, before increasing back on 3e. As a result, 3c (C₁₂) and 3d (C₁₄), which exhibited the most potent inhibitory activity against bacterial growth, also appeared to be the poorest substrates of AMEs, as was the case with the TOB derivatives.^15,26,30 Furthermore, their sulfone analogues 4c and 4d did not seem to be much more susceptible to the action of the AMEs tested.

Figure 1.

Bar graph displaying the relative initial rates of reactions of the various AMEs with KANB and its derivatives 3a–e and 4c, d. Rates are normalized to KANB.

AG acetyltransferases (AACs) represent a class of AMEs that regiospecifically acetylate a single amino group of an AG. They include AAC(3), AAC(2′), and AAC(6′), which modify the amino group at the 3, 2′, and 6′-positions of AGs, respectively. As AAC(3) and AAC(6′) can sometimes both be found in the same bacterial strain, and because our newly synthesized KANB derivatives could decrease the activity of AAC(2′), we wanted to see the simultaneous effect of multiple AACs on KANB and our best derivative 3d. By using silica gel thin layer chromatography (TLC),^37,39,40 we were able to assess the modifications of AAC(3), AAC(2′), and AAC(6′), alone and together, on KANB and 3d (Figure S26). When individually subjected to AAC(3), AAC(2′), or AAC(6′) overnight, KANB was completely converted to its corresponding monoacetylated product (lanes 2, 3, and 4). Meanwhile, in the presence of two enzymes, some of the KANB remained unmodified (lanes 5, 6, and 7). This suggests that the use of two AACs might cause one to inhibit the other. However, this was not the case for 3d. In both compounds, we observed that whenever AAC(3) was used in combination with AAC(2′) (lanes 5 and 13) or AAC(6′) (lanes 7 and 15), only the 3-acetylated product was formed. Similarly, only the 6′-monoacetylated adduct was obtained when AAC(2′) and AAC(6′) were used together (lanes 6 and 14). This is in agreement with the fact that AAC(3) is more catalytically efficient than AAC(6′), which in turn is more catalytically efficient than AAC(2′).³⁵ Furthermore, modification at the 3-position appears to prevent further modification of the 6′-position, which is consistent with what we previously observed in the study of sequential modifications of other AGs.^34,35 When all three enzymes were used simultaneously, KANB was converted to a diacetylated adduct (lane 8), while only the 3-monoacetylated product was obtained for 3d (lane 16). All together, these results showed that, although our KANB derivative 3d could not completely stop the action of AMEs, especially when several of them were present, it would still be able to slow their action and, as such, the emergence of bacterial resistance.

Antifungal Activity

Certain amphiphilic KANA and TOB derivatives were shown to have potent antifungal and/or antibacterial activities.^11,14,15 These reports have prompted us to explore the in vitro antifungal activities of our newly synthesized amphiphilic KANB derivatives 3a–e. The MIC values of KANB, its derivatives 3a–e, and three azoles (POS, FLC, and ITC) against various fungi are summarized in Table 2. The fungal strains tested included Candida and Aspergillus species, which are the most common causes of fungal infections.⁴¹ With C. albicans species being among the most prevalent fungal pathogens⁴² and accounting for a large number of invasive fungal infections observed in intensive care units,⁴³ our collection of fungal strains included C. albicans clinical isolates from immunocompromised patients that are either susceptible (C. albicans ATCC MYA-2876 (strain C) and C. albicans ATCC MYA-2310 (strain E)) or resistant (C. albicans ATCC 10231 (strain A), C. albicans ATCC 64124 (strain B), C. albicans ATCC 90819 (strain D), C. albicans ATCC MYA-1237 (strain F), and C. albicans ATCC MYA-1003 (strain G)) to azoles. Like amphiphilic TOB analogues, our KANB derivatives also exhibited chain length-dependent antifungal activities against fungi in the order 3a (C₈) < 3b (C₁₀) < 3c (C₁₂) < 3d (C₁₄) < 3e (C₁₆). Compounds 3c, 3d, and 3e showed moderate to excellent activity against the fungal strains tested, with MIC values ranging from ≤1.95 to 31.2 μg/mL, ≤1.95 to 15.6 μg/mL, and 1.95 to 3.9 μg/mL, respectively. However, the parent AG KANB and its derivatives 3a and 3b were inactive against all yeast strains tested, with MIC values ≥125 μg/mL. Compound 3a only displayed moderate activity against the filamentous fungi A. nidulans ATCC 38163 (strain H) (MIC = 15.6 μg/mL), while 3b was very active (MIC ≤ 1.95 μg/mL). It is noteworthy mentioning that 3c, 3d, and 3e showed either comparable or, in most cases, enhanced antifungal activity against all yeast strains when compared to MIC-2 (50% inhibition) values of all three commercial antifungal azoles. More intriguingly, 3b–e also exhibited higher inhibitory effect to A. nidulans ATCC 38163 (strain H), with MIC values ≤1.95 μg/mL, which were superior to FLC, but comparable to POS and ITC. These results demonstrate that our KANB derivatives 3a–d are more active against filamentous fungi than their TOB counterparts previously reported.¹⁵ Indeed, while TOB analogues with C₈, C₁₀, C₁₂, and C₁₄ alkyl chains displayed MIC values of 125, 62.5, 7.8, and 7.8 μg/mL against A. nidulans ATCC 38163, we observed an 8-, 32-, 4-, and 4-fold decrease in MIC values of their corresponding KANB derivatives 3a, 3b, 3c, and 3d, respectively. Aspergillus species are one of the most common and emerging causes of invasive aspergillosis in immunocompromised individuals, causing a high rate of morbidity and mortality worldwide.⁴⁴ Moreover, compounds 3c, 3d, and 3e effectively inhibited several azole-resistant fungi (C. albicans ATCC 10231 (strain A), C. albicans ATCC 64124 (strain B) that has mutation in ERG11 sequences, C. albicans ATCC 90819 (strain D), C. albicans ATCC MYA-1237 (strain F), and C. albicans ATCC MYA-1003 (strain G)), potentially suggesting that these compounds will remain unaffected by the action of azole-resistant fungi. This is in agreement with our results showing that, as with bacteria, 3d may also be able to delay the development of resistance by fungi (Figure S27).

Table 2.

MIC Values (μg/mL) Determined for KANB, Its Derivatives 3a–e and 4c, d, and Three Control Antifungal Agents (POS, ITC, and FLC) against Various Yeast Strains and Filamentous Fungi^a

yeast strains	KANB	3a	3b	3c	3d	3e	4c	4d	POS	ITC	FLC
C. albicans ATCC 10231 (A)b	>125	>125	125	31.2	7.8	3.9	31.2	3.9	0.5	0.5	62.5
C. albicans ATCC 64124 (B)b	>125	>125	125	31.2	7.8	3.9	31.2	3.9	>62.5	>62.5	>125
C. albicans ATCC MYA-2876 (C)c	>125	>125	125	31.2	7.8	3.9	31.2	3.9	7.8	7.8	15.6
C. albicans ATCC 90819 (D)b	>125	>125	125	31.2	15.6	3.9	62.5	7.8	31.2	31.2	>125
C. albicans ATCC MYA-2310 (E)c	>125	>125	62.5	7.8	7.8	3.9	62.5	7.8	31.2	31.2	>125
C. albicans ATCC MYA-1237 (F)b	>125	>125	125	31.2	7.8	3.9	62.5	7.8	15.6	31.2	62.5
C. albicans ATCC MYA-1003 (G)b	>125	>125	125	31.2	7.8	3.9	62.5	7.8	15.6	31.2	62.5
Filamentous Fungi
Aspergillus nidulans ATCC 38163 (H)	>125	15.6	≤1.95	≤1.95	≤1.95	1.95	3.9	1.95	≤1.95	≤1.95	>62.5

^aAll experiments were performed in duplicate. For yeast strains, MIC-0 values (complete inhibition) are reported for KANB and 3a–e, whereas MIC-2 values (50% inhibition) are reported for azoles. For filamentous fungi, MIC-0 values are reported for all compounds.

^bStrains resistant to azoles.

^cStrains susceptible to azoles.

We also determined the MICs of the sulfone analogues 4c and 4d against our collection of fungal strains and found that these values did not change much and were generally within 2-fold dilution of the MICs of their corresponding thioethers 3c and 3d (Table 2). This suggests that metabolic S-oxidation of our KANB derivatives will not affect their antifungal activity.

Azoles, such as POS, are considered first-line drugs in the treatment of candidiasis and aspergillosis in humans.⁸ POS kills fungi by inhibiting cytochrome P450-dependent enzyme sterol 14-α-demethylase involved in the synthesis of ergosterol, a key structural component of fungal membrane. Like other azoles, POS also faces many therapeutic challenges such as host side effects, drug–drug interaction, and poor absorption of the drug, resulting in limited therapeutic use.⁴⁵ The frequent report of emergence of fungal resistance against POS has further complicated the treatment of invasive fungal infections in humans, and higher drug dosage is often required. In this study, we have evaluated the in vitro interactions of POS with our best antifungal candidates 3c and 3d against various yeast strains (Table 3). These antifungal interactions were quantified by calculating the fractional inhibitory concentration index (FICI).⁴⁶ Both 3c and 3d exhibited remarkable synergistic interaction with POS against all tested fungal strains with FICI values ranging from 0.07 to 0.31, except for 3d against C. albicans ATCC 10231 (strain A) (FICI = 0.56). It is noteworthy to mention that the MIC values of POS were reduced by 16–64-fold in the presence of 3c and by 16–32-fold in the presence of 3d. Also, POS lowered the MIC values of 3c and 3d by 4–16-fold and 2–8-fold, respectively. So far, there has only been one report of a less potent antifungal AG-derived cationic amphiphile (MIC = 15.6 μg/mL) capable of such synergy with azoles.⁴⁷ Our results thus suggest that the use of 3c or 3d in conjunction with POS may provide better alternatives to treat refractory fungal infections and more importantly in the clinical cases, where low doses of POS are desirable to reduce toxicity or drug–drug interactions.

Table 3.

In Vitro Susceptibility of C. albicans Species to 3c (C₁₂) and 3d (C₁₄) Alone and in Combination with POS^a

yeast strains	MIC values (μg/mL) for 3c						MIC values (μg/mL) for 3d
	drug alone		combination		FICIs	interpretation	drug alone		combination		FICIs	interpretation
	POS	3c	POS	3c			POS	3d	POS	3d
C. albicans ATCC 10231 (A)b	1	32	0.06	8	0.31	SYN	1	8	0.06	4	0.56	IND
C. albicans ATCC 64124 (B)b	>25	32	0.78	2	0.09	SYN	>25	8	1.56	1	0.18	SYN
C. albicans ATCC MYA-2876 (C)c	25	16	0.78	2	0.15	SYN	25	8	1.56	1	0.18	SYN
C. albicans ATCC 90819 (D)b	>25	32	0.78	4	0.15	SYN	>25	8	0.78	1	0.15	SYN
C. albicans ATCC MYA-2310 (E)c	>25	16	0.39	1	0.07	SYN	>25	4	1.56	1	0.31	SYN
C. albicans ATCC MYA-1237 (F)b	>25	32	0.78	2	0.09	SYN	>25	8	0.78	1	0.155	SYN
C. albicans ATCC MYA-1003 (G)b	>25	32	0.78	2	0.09	SYN	>25	8	1.56	2	0.31	SYN

^aAll experiments were performed in duplicate. FICI (fractional inhibitory concentration index) was determined to assess the in vitro antifungal interactions. It is calculated using the formula (MIC of drug A combination/MIC of drug A alone) + (MIC of drug B combination/MIC of drug B alone). When FICI ≤ 0.5, there is synergy (SYN). If 0.5 < FICI < 4, there is no interaction between the drugs (IND). However, when FICI ≥ 4, there is antagonism (ANT).⁴⁶

^bStrains resistant to azoles.

^cStrains susceptible to azoles.

Spontaneous Frequency of Resistance Studies

To establish the spontaneous frequency of resistance of C. albicans ATCC 10231 (strain A) to the combination of compound 3c and POS, three concentrations of POS (0.06, 0.125, or 0.25 μg/mL) were tested in combination with 3c (8 μg/mL). Only the combination of 3c (8 μg/mL) and POS (0.25 μg/mL) yielded few mutants (24 colonies) after 96 h incubation. This result suggests that the spontaneous frequency rate of resistance to the combination of 3c and POS by this strain A is very low (2.4 × 10⁻⁷).

Time-Kill Studies

To evaluate the antifungal activity kinetics of our KANB derivatives, we performed time-kill assays of 3c and 3d on C. albicans ATCC 64124 (strain B) over a period of 24 h (Figure 2). At subinhibitory concentrations (1/2 MICs), 3c and 3d showed fungistatic activity similarly to the known antifungal agent AmB. However, at concentrations equivalent to their MICs, 3c and 3d both rapidly reduced the CFU of C. albicans by ≥2 log 10 after 6 h treatment. In addition, complete fungal cell death was observed after 24 h for 3c and after 12 h for 3d. These results show that 3c and 3d exhibited a fungicidal effect on the azole-resistant strain C. albicans ATCC 64124, which is in agreement with our previous observation with amphiphilic TOB analogues.¹⁵

Figure 2.

Representative time-kill studies of KANB derivatives 3c and 3d against azole-resistant C. albicans ATCC 64124 (strain B). (A) Cultures were exposed to 3c at 8 μg/mL (○), 16 μg/mL (▼), and 32 μg/mL (△). (B) Cultures were exposed to 3d at 2 μg/mL (○), 4 μg/mL (▼), and 8 μg/mL (△). In both panels, cultures were exposed to AmB at 1 μg/mL (■) or to a no drug control (●).

Antifungal Mechanism of Action

Cell membrane disruption has been shown to be a mode of action peculiar to antifungal AG-derived cationic amphiphiles.¹⁵ Using propidium iodide (PI), a dye that fluoresces upon penetrating membrane-compromised cells and binding to nucleic acids, we were able to investigate the effects of our best antifungal candidates on the fungal membrane integrity of the azole-resistant yeast C. albicans ATCC 64124 (strain B). After exposing these yeast cells to various concentrations of 3c and 3d (1× MIC or 2× MIC), we observed 44% or 90%, and 49% or 84% of cell staining with PI, respectively. However, cells treated with KANB and untreated showed minimal staining (5% and <2%, respectively) (Figure 3). This shows that 3c and 3d affect membrane permeability and are both more effective than KANB in causing PI permeation in the azole-resistant C. albicans ATCC 64124 cells. Furthermore, 3c and 3d showed equal rate of permeability of fungal membrane.

Figure 3.

(A) Representative dose-dependent membrane permeabilization effects of KANB and its derivatives 3c and 3d on azole-resistant C. albicans ATCC 64124 (B). From top to bottom: Propidium iodine (PI) dye uptake by yeast cells without drug, with KANB (62.5 μg/mL), with 3c (1× and 2× MIC), and with 3d (1× and 2× MIC). (B) Quantitative representation of the images shown in panel A.

Hemolysis

AG-derived cationic amphiphiles have previously been shown to trigger membrane disruption. To evaluate the selectivity of our KANB derivatives 3a–e toward bacterial or fungal membranes, hemolytic assay was performed using murine red blood cells (mRBCs) (Figure 4). As observed with their antifungal activity, KANB derivatives also showed chain-length-dependent effect on RBCs in the order 3a (C₈) < 3b (C₁₀) < 3c (C₁₂) < 3d (C₁₄) < 3e (C₁₆). Compound 3c lysed 40% of mRBCs at 125 μg/mL, which is a 4-fold higher concentration that its antifungal MIC. Compound 3d, on the other hand, lysed 15% of mRBCs at 7.8 μg/mL (3d’s antifungal MIC) and 73% of mRBCs at 500 μg/mL (32-fold higher than 3d’s antifungal MIC). Compound 3e also lysed 16% of mRBCs at 7.8 μg/mL (twice its antifungal MIC) and 81% of mRBCs at 62.5 μg/mL (16-fold higher than its antifungal MIC). However, KANB and compounds 3a and 3b caused little to no hemolysis at higher concentrations. We also observed that 3d and 3e reach a plateau a lot quicker than 3c, and do not attain 100% (Figure 4). This is probably due to aggregation as the length of the lipophilic alkyl chain increases. In comparison to the known antifungal AmB, KANB, 3a, 3b, 3c, and 3d all showed relatively lower hemolytic activities. 3e, on the contrary, displayed hemolytic activity similar to that of AmB. Similarly, KANB, 3a, 3b, 3c, and 3d displayed lower hemolytic activities than the known membrane targeting antibiotic gramicidin, while 3e was more hemolytic than gramicidin. Therefore, even though 3e was the most active compound against fungi, it appears to rapidly turn hemolytic, leaving us with 3c and 3d as our best antifungal agent candidates. Importantly, when used in conjunction with POS, as less of 3c and 3d was required to display antifungal activity, the hemolytic activity resulting from using this lesser amount would greatly decrease the hemolytic activity of these KANB derivatives. With the MIC value of 3c dropping from 31.2 to 1–8 μg/mL when used alone versus in combination with POS, 50× MIC would be needed to lyse 60% mRBCs, rendering 3c ~30% less hemolytic than AmB and ~20% less hemolytic than gramicidin. Therefore, it is predicted that our compounds, especially when used in combination with POS, may not display problematic hemolysis of red blood cells.

Figure 4.

Hemolytic activity of KANB, gramicidin, amphotericin B (AmB), and 3a–e on mouse red blood cells.

Cytotoxicity

An important parameter in the development of drugs is their ability to selectively target microbial cells without affecting mammalian cells. This is especially the case for antifungal agents because fungi and humans are both eukaryotes and their cells would thus be similar on a biological level. To evaluate the toxicity of our KANB derivatives 3a–d, cytotoxicity assays were conducted against two nucleated mammalian cell lines, A549 and BEAS-2B (Figure 5). While compounds 3a and 3b displayed little to no toxicity against these cell lines like KANB itself, 3c had an IC₅₀ value >62.5 μg/mL against A549 (Figure 5A) and an IC₅₀ value >125 μg/mL against BEAS-2B (Figure 5B). These values were 4–64-fold higher than the antifungal MIC values. Likewise, the IC₅₀ values of 3d against both cell lines were >62.5 μg/mL, which correspond to 8–32 times its antifungal MICs. These results indicate that our KANB derivatives display a low toxicity profile against mammalian cell lines. The abundance of lipid-like components in the fungal membranes (phosphatidylcholine, L-α-phosphatidylethanolamine, L-α-phosphatidylinositol, ergosterol) ^16,48 as compared to the mammalian membranes (phosphatidylcholine, cholesterol)^16,48 could account for this selectivity. Indeed, our KANB derivatives being cationic and rich in amino and hydroxyl groups are capable of hydrogen and ionic interactions with the polar and negatively charged head groups present on the glycerophospholipid and sterols constituents of fungal membranes, and as such should have a higher preference for fungal membranes than mammalian membranes. Compounds 3c and 3d thus appear attractive as potential candidates for the development as antifungal agents.

Figure 5.

Mammalian cell cytotoxicity of KANB and its derivatives 3a–d against (A) A549 cell line and (B) BEAS-2B cell line.

CONCLUSIONS

We have synthesized five cationic amphiphilic KANB derivatives, 3a–e, by attaching linear alkyl chains of various length (C₈, C₁₀, C₁₂, C₁₄, and C₁₆) at the 6″-position of parental KANB through a thioether linkage. Biological evaluation revealed that 3a (C₈) and 3b (C₁₀) exhibited little to no antimicrobial activities, whereas 3c (C₁₂) and 3d (C₁₄) showed promising antibacterial and antifungal potency in vitro, and 3e (C₁₆) was the most active against fungi and with no potency on bacteria. Compound 3c was the most effective at inhibiting the growth of Gram-positive bacteria, closely followed by 3d. Also, both of these compounds were the poorest substrates of AMEs and could evade the action of these bacterial resistance enzymes better than KANB. Compound 3d in particular was the most effective at delaying the development of bacterial resistance. Our KANB derivatives also exhibited chain length-dependent antifungal activities against fungi in the order 3a (C₈) < 3b (C₁₀) < 3c (C₁₂) < 3d (C₁₄) < 3e (C₁₆). Compounds 3c, 3d, and 3e were 4, 16, and 32 times more effective than KANB, respectively, against several of the tested yeast strains. Interestingly, these KANB derivatives showed comparable or, in most cases, enhanced antifungal activity against all yeast strains when compared to the commercial antifungal drugs FLC, ITC, and POS. Against the filamentous fungus Aspergillus nidulans (strain H), our KANB derivatives 3a–d were more active than their TOB counterparts previously reported.¹⁵ Compounds 3c, 3d, and 3e were 64 times better than KANB and 32 times better than FLC. In vitro antifungal combination studies of POS with 3c or 3d also revealed remarkable synergistic interactions, with the MIC values of POS dropping by 16–64-fold in the presence of 3c and by 16–32-fold in the presence of 3d. POS also lowered the MIC values of 3c and 3d by 4–16-fold and 2–8-fold, respectively. Finally, although 3e was the most active against yeasts, it appeared to be quite hemolytic to murine red blood cells. Compounds 3c and 3d, on the contrary, showed low hemolytic activities and mammalian cell toxicities. By synthesizing their sulfone analogues 4c and 4d, we were also able to show that their biological activities will not change much under the effect of metabolic S-oxidation. As compared to our previous reports of TOB variants that investigated their ribosome-targeting capabilities³⁰ or fungicidal activity,¹⁵ this present report further expands the study to include the development of resistance in bacteria and fungi, along with the exciting synergistic effects of two of our KANB derivatives to azoles. It thus appears that 3c and 3d could be useful lead compounds for the development of antifungal agents. Furthermore, extensive in vivo study of these KANB derivatives with azole antifungal drugs will be an encouraging research avenue to be pursued, as this may offer interesting alternatives in the treatment of human fungal infections.