Differential Effects of Linkers on the Activity of Amphiphilic Tobramycin Antifungals

As the threat associated with fungal infections continues to rise and the availability of antifungal drugs remains a concern, it becomes obvious that the need to bolster the antifungal armamentarium is urgent. Building from our previous findings of tobramycin (TOB) derivatives with antifungal activity, we further investigate the effects of various linkers on the biological activity of these aminoglycosides. Herein, we analyze how thioether, sulfone, triazole, amide, and ether functionalities affect the antifungal activity of alkylated TOB derivatives against 22 Candida, Cryptococcus, and Aspergillus species. We also evaluate their impact on the hemolysis of murine erythrocytes and the cytotoxicity against mammalian cell lines. While the triazole linker appears to confer optimal activity overall, all of the linkers incorporated into the TOB derivatives resulted in compounds that are very effective against the Cryptococcus neoformans species, with MIC values ranging from 0.48 to 3.9 μg/mL.

A growing interest in the identification of new targets of AGs [24] has recently led researchers to the investigation of AGs' action against fungi [23,[25][26][27][28][29][30]. It is worth mentioning that the social and economic burden associated with fungal infections is considerable, both in medicine and agriculture. While the frequency of invasive fungal infections continues to rise, the number of antifungal agents available remains limited, calling for the development of additional effective antifungal drugs. Indeed, only three classes of antifungals are currently in clinical use [31]. These include the polyenes (for example, amphotericin B (AmB)), the azoles (for example, fluconazole, voriconazole (VOR), itraconazole, posaconazole), and the echinocandins (for example, caspofungin (CAS), micafungin, anidulafungin). These drugs exert their antifungal activity by either extracting ergosterol from the fungal plasma membrane [32], blocking the production of ergosterol through the inhibition of lanosterol 14α-demethylase [33], or inhibiting the biosynthesis of the fungal cell membrane component (1,3)-β-D-glucan [34], respectively. Unfortunately, resistance to these antifungal drugs has already been observed [35]. However, new strategies to combat fungal infections are being investigated, notably the development of derivatives of currently approved antifungals with improved activity and biological safety profiles [36][37][38][39][40], the combination therapy of current antifungal drugs working synergistically with other drugs [41], and the development of new classes of antifungal agents [42].
We have recently demonstrated that incorporating a linear alkyl chain (for example, the dodecyl group (C 12 ) or tetradecyl group (C 14 )) at the 6 -position of TOB or KANB through a thioether linkage results in AGs with antifungal activity against C. albicans [23,43]. Although we found the TOB (C 14 ) derivative to display stronger antifungal activity than TOB (C 12 ), we elected to generate TOB (C 12 ) analogues in this study because, as we previously demonstrated by investigating the hemolytic activity of KANB (C 12 ) and (C 14 ), KANB (C 12 ) displayed lower hemolytic activity than KANB (C 14 ). Herein, with the goal of improving on the activity of TOB (C 12 ), we investigated the effects of different groups (for example, thioether, sulfone, triazole, amide, and ether) as linkers for the alkyl chain to the AG scaffold on the efficacy of TOB (C 12 ) derivatives against 22 fungal strains. We investigated the hemolytic activity and cytotoxicity of the five most promising antifungals. We also performed time-kill studies and membrane permeabilization assays for the most active compound generated.

Chemistry
The synthesis started from the commercially available AG TOB. Modification at the 6 -position of TOB requires the selective conversion of the 6 -hydroxyl group into a good leaving group, which could be accomplished by Boc protection of the amino groups of TOB, followed by a reaction with 2,4,6-triisopropylbenzenesulfonyl chloride (TIPBSCl) in pyridine (Scheme 1). This gave the central intermediate 1 [44], which upon treatment with 1-dodecanethiol and Cs 2 CO 3 , afforded another intermediate compound 2 [8,10]. Treatment of compound 2 with trifluoroacetic acid (TFA) efficiently removed all the Boc protecting groups to give the target compound 3 [10], representing the TOB derivative with the thioether linkage. Furthermore, S-oxidation of compound 2 with m-CPBA followed by TFA treatment yielded the target sulfone derivative 4 [8].
Compound 1 was also subjected to a S N 2 nucleophilic displacement reaction in the presence of tetrabutylammonium azide (TBAA) to give the azido intermediate 5 [11]. Using copper sulfate and sodium L-ascorbate in DMF under microwave conditions, a click reaction between compound 5 and commercially available alkynes (1-dodecyne and 1-tetradecyne) afforded the intermediates 6a and 6b [11], respectively, whose TFA treatment gave the target triazole derivatives 7a and 7b [11], respectively. The design for compound 7a (C 10 -triazole), which is two-carbon shorter than compound 7b, stems from the need to investigate whether or not the nitrogen atoms N2 and N3 of the triazole ring contribute to the overall length of the alkyl side chain. Compound 5 was also subjected to Staudinger reduction that converted the 6 -azido group into an amine, followed by amide coupling with lauric acid to yield compound 8, which was then converted to the target amide derivative 9.
Finally, based on the very promising and encouraging results obtained against the C. neoformans clinical isolates, we decided to assess the antifungal activity of the derivatives generated against other types of fungal strains. We opted for the filamentous fungi Aspergillus flavus ATCC MYA-3631 (strain K), Aspergillus nidulans ATCC 38163 (strain L), and Aspergillus terreus ATCC MYA-3633 (strain M) ( Table 3). In these cases, all the TOB derivatives, with the exception of compound 16, were better than both TOB and CAS, with a trend similar to that observed with the C. albicans and non-albicans Candida strains. Furthermore, while A. terreus ATCC MYA-3633 (strain M) was not sensitive to the TOB derivatives, compounds 3, 4, 7a, 7b, and 9 showed moderate activity against A. flavus ATCC MYA-3631 (strain K) (MIC values ranging from 7.8 to 31.3 µg/mL) and good activity against A. nidulans ATCC 38163 (strain L) (MIC values ranging from 1.95 to 3.9 µg/mL). Compound 16 only showed good activity against A. nidulans ATCC 38163 (strain L) with an MIC value of 3.9 µg/mL. In light of these results, it appears that the linkers play a major role in the activity of alkylated TOB derivatives, with triazole > "better than" thioether/sulfone > amide > ether. This may stem from the distinct ability of these linkers to form interactions with the polar lipid head groups of fungi. Indeed, with its three nitrogen atoms, the triazole ring may form more hydrogen bonds than the sulfone which has two oxygens, the amide with a nitrogen and an oxygen, and the ether with only an oxygen atom. It is worth mentioning that, while compound 7b (C 12 -triazole) was the most effective at inhibiting the growth of yeasts and filamentous fungi, compound 7a (C 10 -triazole) was the least effective of all six TOB derivatives synthesized (except compound 16). This is no surprise as it is in accordance with the chain length-dependent antifungal activity observed with other amphiphilic AGs [23,43]. Indeed, amphiphilic AGs bear a hydrophilic core structure, which is rich in polar hydroxyl and amino groups that are positively charged under physiological conditions, and a lipophilic alkyl chain capable of interacting with the lipid-rich fungal cell membrane. As the length of the alkyl chain increases, the ability of amphiphilic AGs to puncture the lipid bilayers also increases, which may result in an enhanced cell membrane perturbation, a mechanism well-known for this type of molecules [8,27]. This also shows that the additional atoms N2 and N3 of the triazole ring do not contribute to the overall length of the alkyl chain.

Hemolysis
As potential antifungal agents, it is necessary to assess the ability of the synthesized compounds to selectively target fungal membranes. Since compound 16 was in general as inactive as the parent AG TOB, we decided to focus our efforts on the remaining TOB derivatives. We performed a hemolytic assay of the active TOB derivatives 3, 4, 7a, 7b, and 9 using murine red blood cells (mRBCs) (Figure 1 and Table S1). Although these amphiphilic TOB derivatives appear to affect mRBCs more than the parent TOB, they all displayed relatively lower hemolytic activity compared to AmB, which is an FDA-approved antifungal prescription medicine. Overall, the following trend was observed: 7a (C 10 -triazole) < "less hemolytic than" 9 (C 11 -amide) < 4 (C 12 -sulfone) < 3 (C 12 -thioether) < 7b (C 12 -triazole). Once again, we noticed a chain length-dependent effect on hemolysis, as previously observed with KANB-derived cationic amphiphiles [23]. It also appeared that the sulfone and amide linkers may impart less hemolysis than the thioether and triazole ones. Oxidation of the thioether derivative 3 to its corresponding sulfone 4 also seemed to lessen the hemolytic activity, suggesting that metabolic S-oxidation may not be detrimental in this case. While the sulfone derivative 4 and the amide 9 lysed~15-25% and~18-32% of mRBCs, respectively, at their MIC values against all seven C. albicans strains, the thioether derivative 3 and the triazole 7b lysed~19-46% and~13-41% of mRBCs, respectively. At their MIC values against non-albicans Candida strains, the sulfone derivative 4 and the amide 9 lysed~6-11% and~18-26% of mRBCs, respectively, while the thioether derivative 3 and the triazole 7b lysed~19-46% and~14-25% of mRBCs, respectively. While more hemolysis was observed at their MIC values against the Aspergillus strains (~4-32% for compounds 4 and 9, and~12-74% for compounds 3 and 7b), the TOB derivatives 3, 4, 7a, 7b, and 9 caused little to no hemolysis when tested at their MIC values against C. neoformans, only lysing up to 13% of mRBCs at 3.9 µg/mL, which represents 2-to 8-fold their MIC values against the three clinical isolates tested. The generally low hemolytic activity observed strengthens the potential application of the newly synthesized TOB derivatives as antifungal agents against the C. neoformans species.  Table S1.

Cytotoxicity
To further evaluate the potential safety of the TOB derivatives 3, 4, 7a, 7b, and 9, we assessed the cytotoxicity of these compounds against two mammalian cell lines, BEAS-2B and A549 ( Figure 2). Like the parent AG TOB, these derivatives displayed little to no toxicity against both cell lines. Indeed, compounds 4 (sulfone), 7a (C 10 -triazole), 7b (C 12 -triazole), and 9 (amide) all had IC 50 values >62.5 µg/mL, which were 1-to 16-fold higher than their antifungal MIC values against the C. albicans species. Only compound 3 (thioether) had an IC 50 value in the range of 31.3-62.5 µg/mL in BEAS-2B cells. This observed selectivity of the TOB derivatives in targeting fungal cells in the presence of mammalian cells may stem from the difference in lipid composition of the cell membranes in fungi and humans [50].

Time-Kill Studies
In light of the results presented so far, we selected the most promising compound 7b for further investigations. We evaluated the antifungal potency of 7b (C 12 -triazole) by determining its time-kill course against a representative C. albicans strain, ATCC 10231 (strain A); a representative non-albicans Candida strain, C. parapsilosis ATCC 22019 (strain J); and a representative non-Candida yeast, C. neoformans (strain CN1) over a 24-h period. We also included the parent AG TOB and the triazole antifungal agent VOR as controls in our study ( Figure 3). While TOB did not affect the growth of any of the tested fungal strains, its triazole derivative 7b was able to reduce the C. albicans ATCC 10231 (strain A) CFU by ≥2 log 10 after 9 h of treatment at 3.9 µg/mL (1× MIC) and after 6 h of treatment at 15.6 µg/mL (4× MIC) (Figure 3a). This suggests a dose-dependent effect as previously observed with other antifungal amphiphilic AGs [23,43]. Furthermore, compound 7b showed a similar growth inhibitory effect to that of VOR, since the latter also reduced the CFU of C. albicans ATCC 10231 (strain A) by ≥2 log 10 after 9 h of treatment at its 1× MIC value of 1.0 µg/mL and maintained a fungistatic activity up to the 24-h period. Against C. parapsilosis ATCC 22019 (strain J), compound 7b displayed a fungistatic effect at 1.95 µg/mL (1× MIC) only for the first 3 h before the fungal cells followed a trend similar to the growth control ( Figure 3b). A decrease in CFU by ≥2 log 10 was only observed after 24 h of treatment. Meanwhile, at 7.8 µg/mL (4× MIC), compound 7b completely killed C. parapsilosis ATCC 22019 (strain J) after 24 h. At 0.49 µg/mL (1× MIC), compound 7b rapidly reduced the CFU of C. neoformans (strain CN1) by ≥2 log 10 after 3 h of treatment ( Figure 3c). In addition, complete fungal cell death was observed as early as 6 h after the treatment at 0.49 µg/mL (1× MIC) and 3 h after the treatment at 1.95 µg/mL (4× MIC), while TOB and VOR showed little to no growth inhibitory effect against C. neoformans (strain CN1). These results confirm that the TOB derivative 7b (C 12 -triazole) exhibits better fungal growth inhibition than the parent AG TOB. Furthermore, compound 7b displayed a similar fungistatic effect to that of the FDA-approved triazole antifungal agent VOR against C. albicans ATCC 10231 (strain A) and C. parapsilosis ATCC 22019 (strain J), and an enhanced fungicidal effect against C. neoformans (strain CN1). The promise of 7b as a fungicidal agent against Cryptococcus sp. is highly encouraging. In all panels, the cultures were exposed to a no drug control (black circle), TOB at 15.6 µg/mL (white circle), VOR at 1.0 or 0.25 µg/mL (black inverted triangle), and 7b at 1× MIC (white triangle) and 4× MIC (black square). Each test tube represents the corresponding sample treated with resazurin, which was added for the visualization of fungal growth. Note: Blue = no fungal growth; orange-pink = fungal growth.

Antifungal Mechanism of Action
We have previously shown that amphiphilic TOB derivatives, notably compound 3, exert their antifungal activity by perturbing the fungal cell membrane [43]. We then decided to evaluate the ability of the most potent compound 7b to affect the fungal membrane integrity. Using propidium iodide (PI), a dye that fluoresces upon binding nucleic acids in membrane-compromised cells [51], we observed that C. albicans ATCC 10231 (strain A) cells that were treated with compound 7b at concentrations equivalent to 2× MIC or 4× MIC allowed a larger influx of PI dye compared to the untreated cells or those treated with the parent TOB, which only showed negligible staining (Figure 4). Yeast killed by heat were used as the positive control. These results show that compound 7b also affects the fungal membrane permeability and is more effective than TOB in causing PI permeation into the azole-resistant C. albicans ATCC 10231 (strain A) cells.

Conclusions
In summary, we have synthesized six TOB derivatives with various linkers, including thioether, sulfone, triazole, amide, and ether. While the C 12 -triazole derivative 7b was the most potent, the C 12 -ether derivative 16 was the least effective at inhibiting the growth of several fungal strains. We also noticed a chain length-dependent effect on the antifungal activity. Furthermore, the active TOB derivatives displayed low hemolytic activity and low mammalian cell toxicity. Finally, the compounds generated were very active against C. neoformans clinical isolates, leaving the door open for a future (outside of the scope of this study) investigative avenue for the development of novel therapies for cryptococcosis.

Antifungal Susceptibility Testing
The MIC values of compounds 3, 4, 7a, 7b, 9, and 16, as well as the parent compound TOB and the antifungal CAS against yeasts (strains A-G (Table 1), and strains H-J, CG1-3, CP1-3, and CN1-3 (Table 2)) were evaluated in 96-well microtiter plates as described in the CLSI document M27-A3 [52] with minor modifications. The final concentrations of compounds 3, 4, 7a, 7b, 9, and 16, as well as that of TOB tested in this study ranged from 0.24-62.5 µg/mL. CAS was used as a positive control and the final concentrations tested for CAS ranged from 0.03-31.3 µg/mL. Briefly, overnight yeast cultures were grown in yeast peptone dextrose (YPD) broth and the cell density was adjusted to an OD 600 of 0.12 (~1 × 10 6 CFU/mL) by using a spectrophotometer. Yeast cell suspensions were further diluted to achieve 1-5 × 10 3 CFU/mL in RPMI 1640 medium, and 100 µL of these yeast cells was added to 96-well microtiter plates containing RPMI 1640 medium and titrated compounds. Each test was performed in triplicate. The plates were incubated at 35 • C for 48 h. The MIC values for compounds 3, 4, 7a, 7b, 9, and 16, CAS, and TOB were defined as the lowest drug concentration that prevented visible growth (also known as MIC-0) when compared to the growth control. These data are presented in Table 1 (for strains A-G) and Table 2 (for strains H-J, CG1-3, CP1-3, and CN1-3).
Similarly, the MIC values of compounds 3, 4, 7a, 7b, 9, and 16, as well as that of all control drugs against filamentous fungi (strains K-M (Table 3)) were determined as previously described in CLSI document M38-A2 [53]. The spores were harvested from sporulating cultures growing on potato dextrose agar (PDA) by filtration through sterile glass wool and enumerated by using a hemocytometer (Hausser Scientific, Horsham, PA, USA) to obtain the desired inoculum size. Two-fold serial dilutions of compounds 3, 4, 7a, 7b, 9, and 16, as well as CAS and TOB were made in sterile 96-well microtiter plates to obtain the final concentration range of 0.24-62.5 µg/mL for TOB and TOB derivatives as well as 0.03-31.3 µg/mL for CAS in the RPMI 1640 medium. The spore suspensions were added to the wells to afford a final concentration of 5 × 10 5 spores/mL. The plates were incubated at 35 • C for 72 h.
The MIC values of all compounds, including compounds 3, 4, 7a, 7b, 9, and 16 as well as CAS and TOB against filamentous fungi were based on the complete inhibition of growth (optically clear well) when compared to the growth control (MIC-0). Each test was performed in triplicate. These data are also presented in Table 3 (strains K-M).

Hemolytic Activity Assays
The hemolytic activity of compounds 3, 4, 7a, 7b, and 9 as well as the parent compound TOB was determined by using previously described methods with minor modifications [23]. Murine whole blood (1 mL) was suspended in 4 mL of PBS and centrifuged at 1000 rpm for 10 min at room temperature to obtain murine red blood cells (mRBCs). The mRBCs were washed four times in PBS and resuspended in the same buffer to a final concentration of 10 7 erythrocytes/mL. Two-fold serial dilutions of compounds 3, 4, 7a, 7b, and 9 were prepared using 100 µL of PBS buffer in Eppendorf tubes followed by the addition of 100 µL of mRBC suspension that made the final concentration of compounds and mRBCs to be 0.48-62.5 µg/mL and 5 × 10 6 erythrocytes/mL, respectively. The tubes were incubated at 37 • C for 1 h. The tubes with PBS buffer (200 µL) and Triton X-100 ® (1% v/v, 2 µL) served as the negative (blank) and positive controls, respectively. The percentage of hemolysis was calculated using the following equation: % hemolysis = [(absorbance of sample) − (absorbance of blank)] × 100/(absorbance of positive control). These data are presented in Figure 1 and Table S1.

In Vitro Cytotoxicity Assays
Mammalian cytotoxicity assays were performed as previously described with minor modifications [29]. The normal human bronchial epithelial cells BEAS-2B and the human lung carcinoma epithelial cells A549 were grown in DMEM containing 10% fetal bovine serum (FBS) and 1% antibiotics. The confluent cells were then trypsinized with 0.05% trypsin and 0.53 mM EDTA, and resuspended in fresh DMEM medium. The cells were transferred into 96-well microtiter plates at a density of 3000 cells/well and were grown for 24 h. The following day, the medium was replaced by a fresh culture medium containing serially diluted compounds 3, 4, 7a, 7b, and 9, as well as the parent drug TOB at final concentrations of 0.48-62.5 µg/mL or sterile ddH 2 O (negative control). The cells were incubated for an additional 24 h at 37 • C with 5% CO 2 in a humidified incubator. To evaluate cell survival, each well was treated with 10 µL (25 µg/mL) of resazurin sodium salt (Sigma-Aldrich) for 4-6 h. Metabolically active cells can convert resazurin (blue) to the highly fluorescent dye resorufin (pink), which was detected at λ 560 excitation and λ 590 emission wavelengths by using a SpectraMax M5 plate reader. Triton X-100 ® (1% v/v) gave the complete loss of cell viability and was used as the positive control. The percentage survival rates were calculated by using the following formula:

Time-Kill Assays
Time-kill assays were used to assess the inhibitory efficiency of 7b against three yeast strains, C. albicans ATCC 10231 (strain A), C. parapsilosis ATCC 22019 (strain J), and C. neoformans clinical isolate CN1. The protocol for time-kill assays followed methods previously described [29,54] with minor modifications. Yeast cultures were grown overnight in YPD broth at 35 • C with shaking (200 rpm). A working stock of fungal cells was made by diluting cultures in RPMI 1640 medium to an OD 600 of 0.125, approximately 1 × 10 6 CFU/mL. From the working stock, 100 µL of cells was added to 4.9 mL of RPMI 1640 medium in sterile culture tubes, making the starting fungal cell concentration approximately 10 5 CFU/mL. Compounds were then added to the fungal cells. The treatment conditions included sterile control, growth control, TOB (parent; negative control), voriconazole (VOR) at 1× MIC (positive control), 7b at 1× and 4× MIC. The treated fungal cultures were incubated at 35 • C with shaking (200 rpm) for 24 h. The samples were taken from the different treatments at regular time points (0, 3, 6, 9, 12, and 24 h.) and plated in duplicates. For each time point, cultures were vortexed, 100 µL of culture was aspirated, and 10-fold serial dilutions were made in sterile ddH 2 O. From the appropriate dilutions, 100 µL of fungal suspension was spread on PDA plates and incubated at 35 • C for 48 h before the colony counts were determined. At 24 h, 50 µL of 1 mM resazurin was added to the treatments and incubated at 35 • C with rotation for 2 h in the dark for visual inspection. Experiments were performed in duplicate. These data are presented in Figure 3.

Membrane Permeabilization Assay Using Propidium Iodide Staining
A fresh culture of C. albicans ATCC 10231 (strain A) was prepared in 5 mL of YPD broth in a Falcon tube and was grown overnight at 35 • C at 200 rpm. An overnight culture (40 µL) was transferred to RPMI 1640 medium (1 mL) containing no drug (negative control) or compound 7b at concentrations of 3.9 µg/mL (2× MIC) and 7.8 µg/mL (4× MIC). TOB (31.3 µg/mL) was used as a negative control. The cell suspensions were then treated for 2 h at 35 • C with continuous agitation (200 rpm). To prepare a positive control sample, C. albicans ATCC 10231 (strain A) cells were killed by heat shock at 95 • C for 5 min as described by Ocampo and Barrientos [55]. The cells were then centrifuged and resuspended in 500 µL of PBS buffer (pH 7.2 adjusted at room temperature). Subsequently, the cells were treated with propidium iodide (9 µM, final concentration) and incubated for 20 min at room temperature in the dark. Glass slides prepared with 10 µL of each mixture were observed in bright field and fluorescence modes (using Texas red filter set with λ 535 excitation and λ 617 emission wavelengths) using a Zeiss Axiovert 200 M fluorescence microscope. The data were obtained from at least two independent experiments. The images were also post-processed, utilizing automatic contrast and brightness setting in Microsoft PowerPoint 2013 to eliminate background noise. These images are presented in Figure 4.
Supplementary Materials: The supplementary materials include 1 H and 13 C-NMR as well as mass spectra (Figures S1-S36) for the molecules synthesized. All compounds tested for activity are ≥95% pure according to NMR spectra. Table S1 with the % hemolysis ± SDEV displayed in Figure 1 is also provided. These materials are available free of charge via the Internet.

Conflicts of Interest:
The authors declare no conflict of interest.