The Hydrophobicity and Antifungal Potentiation of Burkholdine Analogues

The burkholdines are a family of cyclic lipopeptides reported to exhibit antifungal activity. We synthesized a series of 18 burkholdine analogues in good yield by conventional Fmoc-SPPS followed by cyclization with DIPCI/HOBt in the solution phase. Although none of the synthesized peptides exhibited antifungal activity, several did potentiate the antibiotic effect of the antibiotic G418, including the Thr-bearing Bk analogue (4b) and the tartaramide-bearing Bk analogue (5b). This work exemplifies the potential of burkholdine analogues as potentiating agents.


Introduction
The burkholdines (Bks) are a family of 25-membered cyclic lipopeptides composed of eight amino acids arranged in a lactam ring bearing a lipophilic side chain, which were first isolated from the proteobacteria Burkholderia ambifaria 2.2N by Schmidt in 2010 [1][2][3][4]. All naturally occurring Bks isolated so far have demonstrated potent antifungal activity, especially Bk-1119, whose activity is up to 25 times higher than that of amphotericin B [5][6][7]. The mechanism of this activity is thought to entail the inhibition of fungal β-glucan synthase, as is the case for the candine antibiotics micafungin [8] (MIC 90 = 0.015 µg/mL (Candida albicans)) and caspofungin [9] (MIC 90 = 0.25 µg/mL (C. albicans)) [10,11]. Bk-1119 is a promising lead compound for the development of safer antifungal drugs because it targets a β-glucan synthase not found in humans. However, neither the active site nor target enzyme of the Bks has been definitively identified, nor has the total synthesis of any Bk ever been reported. The role of the unusual residues β-OHTyr and β-OHAsn present in the Bk lactam ring is also unknown, but they are highly consequential, as their instability in strongly acidic solutions precludes the synthesis of peptides that incorporate them by conventional SPPS, which relies on TFA [12][13][14]. Accordingly, the scalable synthesis of Bks is needed to clarify their mechanism of action, establish the structural moieties essential for their good activity, and to generate enough material for mechanistic studies.
In our laboratory, we have been conducting structure-activity relationship studies of the Bks using Bk-1097 (1) as a lead compound [15,16]. Our studies so far have yielded two main findings. The first is that the 25-membered octapeptide structure is important for antifungal activity, although 27-membered octapeptides are also weakly active. The second is that the unusual amino acids present in the Bks are not essential for antifungal activity; analogue (2), which incorporates L-Tyr, L-Asn, and N-lauryl-3-amino-4-carbamoylpropanoic acid (LAP) in place of the unusual amino acids β-hydroxy Tyr, β-hydroxy Asn, and 3amino-5,6,7-trihydroxyoctadecanoic acid (ATHOD) present in Bk-1097 (2), respectively, showed MIC = 25 µg/mL (S. cerevisiae). Herein, we report the synthesis and evaluation of Ser/Thr-type Bk analogues (3, 4, 6, and 7) and hydrophobic, tartaramide-type Bk analogues (5 and 8).

Results and Discussions
Our previous work has established the importance of 25-and 27-membered rings for antifungal activity, showing that Bk analogues incorporating β,γ-diaminobutanoic acid (Dab) residues (the Xaa8 position) of the S configuration had better activity than their epimers-even though the natural Bk products are composed exclusively of amino acids of the R configuration [16].
For this work, we aimed to synthesize two series of 25-and 27-membered analogues incorporating either enantiomer of Dab at the Xaa8 position, as well as a third series incorporating L-Orn to investigate the efficiency of the ring size. To assess the influence of the length of the fatty side chain on antifungal activity, Bk analogues bearing undecane-, dodecane-, and octane-based side chains were all designed within each series. Thus, we designed the target peptides (3a-8) (Figure 1, Scheme 1, Table 1). and evaluation of Ser/Thr-type Bk analogues (3, 4, 6, and 7) and hydrophobic, tartaramidetype Bk analogues (5 and 8).

Results and Discussions
Our previous work has established the importance of 25-and 27-membered rings for antifungal activity, showing that Bk analogues incorporating β,γ-diaminobutanoic acid (Dab) residues (the Xaa8 position) of the S configuration had better activity than their epimers-even though the natural Bk products are composed exclusively of amino acids of the R configuration [16].
For this work, we aimed to synthesize two series of 25-and 27-membered analogues incorporating either enantiomer of Dab at the Xaa8 position, as well as a third series incorporating L-Orn to investigate the efficiency of the ring size. To assess the influence of the length of the fatty side chain on antifungal activity, Bk analogues bearing undecane-, dodecane-, and octane-based side chains were all designed within each series. Thus, we designed the target peptides (3a-8) (Figure 1, Scheme 1, Table 1).   First, we prepared (S) and (R)-g-Fmoc-b-Alloc-diaminobutanoic acids (Dab) (9), Fmoc-Orn(Alloc)-OH, and tartaramide derivative (Tat) (10). (S)-and (R)-g-Fmoc-b-Allocdiaminobutanoic acids (Dab) (9) were synthesized as previously described by our group [16]. Fmoc-Orn(Alloc)-OH was derived from Fmoc-Orn(Boc)-OH in two steps. Tat derivative (10) was prepared in five steps from L-(+)-tartaric acid (see Supplementary Materials Figure S1).
Next, the designed Bk analogues were synthesized according to our previous synthetic strategy [12,13]. Macrolactamization between the Xaa8 amine and the C-terminus carboxylic acid of the linear peptides was accomplished over 14-62 h in the presence of DIPC/HOBt. Global deprotection was performed with TFA/TIPS/H 2 O (95:2.5:2.5). The corresponding linear peptides were prepared by Fmoc-SPPS on 2-chlorotrityl resin (2-CT resin) starting from the loading of Fmoc-Asn(Trt)-OH onto the C-terminus of the corresponding linear peptides. The Alloc groups protecting the branched amino acids on Xaa8 were removed using catalytic Pd(0) and the corresponding linear peptides bearing protected side chains achieved cleavage from resin with 20% HFIP/CH 2 Cl 2 (Scheme 1).
The synthesis of 5b is depicted in Scheme 2 and is representative of the procedure used for all the analogues. Fmoc-SPPS was performed to give the requisite linear peptide resin 11, which was coupled with Tat derivative (10) in the presence of DIPCI/HOBt/DIPEA. Pd(0)mediated Alloc deprotection was carried out to give the linear peptide 12 in excellent yield after cleavage from resin using 20% HFIP/CH 2 Cl 2 . Cyclization of the linear peptide 12 with DIPCI/HOBt for 24 h proceeded to give the cyclic peptide 13 in 74% yield. Finally, global deprotection of the cyclic peptide 13 by TFA/TIPS/H 2 O (95:2.5:2.5) gave Bk analogue 5b in 33% yield after the purification by RP-HPLC. The overall yield of 5b from resin loading was 25%, confirming the efficiency of the cyclization and the stability of 5b under acidic conditions (Scheme 2). Reactions were monitored by RP-HPLC and measured by ESI-MS (Table 2 and Supplementary Materials Figure S1). . The corresponding linear peptides were prepared by Fmoc-SPPS on 2-chlorotrityl resin (2-CT resin) starting from the loading of Fmoc-Asn(Trt)-OH onto the C-terminus of the corresponding linear peptides. The Alloc groups protecting the branched amino acids on Xaa8 were removed using catalytic Pd(0) and the corresponding linear peptides bearing protected side chains achieved cleavage from resin with 20% HFIP/CH2Cl2 (Scheme 1). The synthesis of 5b is depicted in Scheme 2 and is representative of the procedure used for all the analogues. Fmoc-SPPS was performed to give the requisite linear peptide resin 11, which was coupled with Tat derivative (10) in the presence of DIPCI/HOBt/ DIPEA. Pd(0)-mediated Alloc deprotection was carried out to give the linear peptide 12 in excellent yield after cleavage from resin using 20% HFIP/CH2Cl2. Cyclization of the linear peptide 12 with DIPCI/HOBt for 24 h proceeded to give the cyclic peptide 13 in 74% yield. Finally, global deprotection of the cyclic peptide 13 by TFA/TIPS/H2O (95:2.5:2.5) gave Bk analogue 5b in 33% yield after the purification by RP-HPLC. The overall yield of 5b from resin loading was 25%, confirming the efficiency of the cyclization and the stability of 5b under acidic conditions (Scheme 2). Reactions were monitored by RP-HPLC and measured by ESI-MS (Table 2 and Supplementary Materials Figure S1).  Analogues (3a)- (8) were prepared based on the synthesis of 5b; the chemical yields of linear, cyclic, and designed peptides (A, B, and C%) and overall yields (D%) are shown in Table 2. Yields of the requisite linear peptides incorporating the Dab residue by Fmoc-SPPS were satisfactory (A%); those of the Orn-containing peptides (6a-d, 7, and 8) were moderated. Macrolactamization of the linear peptides was accomplished according to our optimized conditions (B%). Purification of crude protected cyclic peptides was performed by silica gel column chromatography. The efficiency of the cyclization step was noted to be influenced by both the sequence of the amino acids of the linear precursor and their stereochemistry. The final deprotection of the cyclic peptides also gave various yields. However, the cyclic peptides incorporating a (R)-Dab residue (3e-5b) were obtained in moderate yields, confirming the stability of peptides (3e-5b) under acidic conditions. Yields for steps A-D for each peptide are presented in Table 2. HPLC profiles and ESI-MS spectra data are included in the Supplementary Materials Figure S1. (Table 3). The hydrophilicity of all of the cyclic peptides synthesized (3a)-(8) was investigated using RP-HPLC under the same eluting conditions (30-60% MeCN/H 2 O for 30 min); the corresponding retention times are shown in Table 3. In all cases, the retention times of the (S)-Dab-containing peptides were shorter than those of the (R)-Dab-containing peptides, which were similar to those bearing L-Orn. The retention times of the cyclic peptides were shorter for Tat > D-Ser > L-Thr > L-Ser, while peptides bearing the undecanoic acid moiety eluted faster than those bearing the decanoic acid by about 2 min.
In addition, the polar surface areas (PSA) and logP values of these peptides were calculated using SPARTAN'18 (Wavefunction), using their stable conformers as calculated using molecular mechanics. The PSA of the Tat-bearing cyclic peptides (5a) was found to be similar to that of Bk-1097 (1). The logP values of these peptides also showed a similar tendency to the PSA values. The calculated PSA and logP values of synthesized peptides (3a)-(8) were all negative and unlikely to exhibit cell penetration. Accordingly, any biological activities they exhibited almost certainly arose from their interactions with the cell surface.
The antifungal activities of Bk analogues (3a)- (8) were also evaluated at a concentration of 200 mg/mL. Against the yeast S. cerevisiae (ATCC204504), only 3b and 3f showed inhibitory effects, with MIC values of 100 and 200 mg/mL, respectively-far higher than that of Bk analogue (2) (MIC = 25 µg/mL) [16]. None of the peptides inhibited A. oryzae (NBRC100959) or Candida viswanathii (NBRC10321) at 200 mg/mL concentration. This lack of activity may have been due to the high hydrophobicities of the analogues precluding their entry into the cells; the PSA and logP values for the analogues were all higher than those of Bk analogues (2). Future work will include a structure activity-relationship study of natural Bk products and their analogues to better understand the mechanism of the antifungal inhibition (Table 3).
Two antibiotics with different mechanisms of action can have a synergistic effect, together exerting an antifungal activity higher than the sum of their individual activities. Accordingly, the potentiation [17,18] of the antifungal effect of G418 (Geneticin) [19,20], an aminoglycosyl-type antibiotic that inhibits protein biosynthesis by binding to the ribosome 70S and 80S subunits, and zeocin (phleomucin D1) [21,22], a glycopeptide antibiotic and DNA intercalator, by cyclic peptides (3a)-(8) was evaluated. Although the relatively high PSA and negative logP values of Bk analogues (3a)- (8) are not conductive to cell penetration, there was the possibility that their fatty side chains have an affinity for cell surfaces. A dose of each 100 mg/mL of the cyclic peptides (3a)-(8) did indeed potentiate the effect of G418 against the pathogenic fungus S. cerevisiae; the use of 3g, 4b, 5b, 6c, 7, and 8 dropped the MIC value of G418 from 25 mg/mL to 12.5 mg/mL, a two-fold increase in potency. The mechanism of action is not clear, but it is possible that the interaction of the cyclic peptides (3a)-(8) with the cell wall increases its susceptibility to penetration by G418. However, no potentiation of zeocin by the cyclic peptides (3a)-(8) was observed, perhaps because it proved a highly potent inhibitor on its own (MIC = 3.13 mg/mL) ( Table 4). Note: a 3g was used at a rate of 100 mg/mL; b these assays were attempted three times.

General
All solvents were reagent grade (Nacalai tesque, Kyoto, Japan and Kishida Chemical, Osaka, Japan). All commercial reagents were of the highest purity available (Watanabe Chemical, Hiroshima, Japan, Fujifilm-Wako, Tokyo, Japan and TCI, Tokyo, Japan). Optical rotations were determined with a JASCO P-2200 polarimeter at the sodium D line (Tokyo, Japan). IR spectra were recorded on a Spectrum Two instrument (PerkinElmer, Waltham, MA, USA). 1