Perpendicular orientation of amphotericin B methyl ester in hydrated

Subscriber access provided by Bibliothèque de l'Université Paris-Sud

Article

Perpendicular orientation of amphotericin B methyl ester in hydrated lipid bilayers supports the barrel-stave model Tomoya Yamamoto, Yuichi Umegawa, Masaki Yamagami, Taiga Suzuki, Hiroshi Tsuchikawa, Shinya Hanashima, Nobuaki Matsumori, and Michio Murata Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.9b00180 • Publication Date (Web): 11 Apr 2019 Downloaded from http://pubs.acs.org on April 14, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Perpendicular orientation of amphotericin B methyl ester in hydrated lipid bilayers supports the barrelstave model Tomoya Yamamoto,1,2 Yuichi Umegawa,2,1,3,* Masaki Yamagami,1 Taiga Suzuki,1 Hiroshi Tsuchikawa,1 Shinya Hanashima,1 Nobuaki Matsumori,1,4 and Michio Murata1,2,3,* 1Department

of Chemistry, Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan

2JST-ERATO

Lipid Active Structure Project, Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan 3Fundamental

Science Research Center, Graduate School of Science, Osaka

University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan 4Department

of Chemistry, Graduate School of Sciences, Kyushu University, Fukuoka 819-0395, Japan

Corresponding

Authors:

Phone,

(+81)-6-6850-5774;

Fax,

(+81)-6-6850-5774;

[email protected] (M.M.), [email protected] (Y.U.)

1 ACS Paragon Plus Environment

E-mail,

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

This study was partly supported by JSPS KAKENHI Grant Numbers 16H06315 and 25750383 and by the Japan Science and Technology Agency as part of the Lipid Active Structure Project (.

ABBREVIATIONS AmB, Amphotericin B; AME, Amphotericin B methyl ester; B3LYP, Becke’s three-parameter hybrid functional and Lee, Yang, and Parr correlation functional; CSA, chemical shift anisotropy; DMF, N,N-dimethylformamide; DMPC, dimyristoyl phosphatidylcholine; DMSO, dimethyl sulfoxide; DPPC, dipalmitoyl phosphatidylcholine; Erg, ergosterol; FID, free induction decay; GIAO, gauge including atomic orbital; HPLC, high-performance liquid chromatography; MAS, magic angle spinning; MD, molecular dynamics; MLV, multi-lamellar vesicles; PC, phosphatidylcholine; MP2, second-order Møller-Plleset; REDOR, rotational-echo double resonance; THP, tetrahydropyran; TMS, trimethylsilyl.


Page 2 of 39

Page 3 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

ABSTRACT

A clinically important antibiotic, amphotericin B (AmB), is a membrane-active natural product that targets membrane sterol. The antimicrobial activity of AmB is generally attributed to its membrane permeabilization, which occurs when a pore is formed across a lipid bilayer. In this study, the molecular orientation of AmB was investigated using solid-state NMR to better understand the mechanism of antifungal activity. The methyl ester of AmB (AME) labeled with NMR isotopes, d3-AME, and its fluorinated and/or 13C-labeled derivatives were prepared. All the AmB derivatives showed similar membrane-disrupting activities and UV spectra in phospholipid liposomes, suggesting that their molecular assemblies in membranes closely mimic those of AmB. Solid-state 2H NMR measurements of d3-AME in a hydrated membrane showed that the mobility of AME molecules depends on concentration and temperature. At an AME/Erg/dimyristoyl phosphatidylcholine (DMPC) ratio of 1:5:45, AME became sufficiently mobilized to observe the motional averaging of quadrupole coupling. Based on the rotational averaging effect of 19F CSA, 2H-quadrupolar

splitting, and

13C-19F

dipolar coupling of 14β-F-AMEs, we deduced that the

molecular axis of AME is predominantly parallel to the normal of a lipid bilayer. This result supports the barrel-stave model as a molecular assembly of AmB in membrane.


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

INTRODUCTION A biological membrane acts as a barrier separating inner and outer cells, and it controls the selective permeability of specific ions and other small molecules. This function is essential for living cells, but membrane-active compounds such as drugs or toxins can directly target the membrane lipids in a bilayer structure.1-7 There are various ways membrane-active drugs can interact with membranes, such as pore formation,8 abstraction of specific lipid molecules in membranes,9 and phase modulation of lipid bilayers.10 These interactions disrupt the membrane’s integrity, frequently leading to cell death. However, it is generally difficult to elucidate the mechanism of drug-induced membrane disruption chiefly because of the heterogeneous nature of biological membranes and the rapid lateral diffusion of lipids. Additionally, membrane-active drugs often form self-assemblies, which even complicate the problem but sometimes provide a challenging research target in the fields of structural biology and molecular pharmacology.11 To elucidate the mechanism of membrane-active compounds, it is essential to understand their molecular orientation in membranes.12,13 Solid-state NMR is a powerful tool for performing orientation analysis of a hydrated membrane system. In fact, the membrane-bound structure and orientation of α-helical peptides are some of the most suitable research targets for solid-state NMR12–20 because of their well-defined secondary structures and established chemical/biological methodologies, such as isotope labeling techniques. However, experimental strategies for determining the orientation of non-peptidic agents, such as natural products, in membranes are not well established because of the difficulties associated with isotope labeling of diverse chemical structures. Amphotericin B (AmB, Figure 1), an important drug in clinical settings, is a typical example of a membrane-active natural product. AmB is a polyene macrolide antibiotic produced by


Page 4 of 39

Page 5 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Streptomyces nodosus, and it is widely used to treat fungal infection because of its efficacy and reduced development of resistance.21,22 AmB directly interacts with cell membranes and increases the membrane permeability of ergosterol (Erg)-containing fungal cells, but not cholesterolcontaining mammalian cells.21,23,24 This unique feature makes AmB particularly important for treating deep-seated fungal infections compared to other sterol-dependent anti-fungal agents.1-5, 25 However, its selective toxicity against fungi is higher than other sterol-dependent agents26,27 but not infallible, which sometimes causes severe adverse effects.21,22 Therefore, it is important to understand the mechanisms of the antifungal activity and sterol selectivity of AmB more in detail. So far, discussions of the membrane assembly of AmB responsible for pharmacological activity have been primarily based on the barrel-stave model. In this model, AmB forms self-aggregates with Erg with ion channel activity, thus changing the membrane potential and, ultimately, inducing cell death.28 Another molecular mechanism of AmB, in which AmB extracts Erg from a cell membrane and causes lethal destabilization of the bilayer structure, was proposed in the sterol sponge model.29–31 To obtain detailed information about the topology and structure of the complex formed by AmB and Erg, it is crucial to gain information about the orientation of the drug in the membrane. Several studies have shown that AmB is oriented almost perpendicular to the bilayer plane,32– 35

and a recent computational study suggests that AmB is vertically inserted in the bilayer in either

a monomeric or oligomeric manner.36 In contrast, some linear dichroism and electron spin resonance (ESR) studies have indicated that AmB lies on the surface of the membrane.37,38 A solidstate NMR study investigating the orientation of AmB-d3-methylamide using the 2H NMR of mechanically aligned bilayers found that the rotational axis of AmB is largely perpendicular to the membrane plane.39 Yet, the authors did not report the angle between the molecular axis of the drug


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 39

and the rotational axis. To discover whether AmB preferentially forms a barrel stave or sterol sponge, the angle between the AmB molecular axis and membrane normal has to be determined. Thus, further investigation of the orientation of the AmB molecule with respect to the membrane is necessary. For solid-state NMR measurements, we established the isotope-labeling methods of AmB and Erg by performing organic synthesis and culturing the producing microorganism.40–43 Using the labeled compounds, we have attempted to elucidate the structure of an AmB molecular assembly. We successfully observed the interaction between AmB and Erg by 2H NMR44 and elucidated the structure of their bimolecular complex using rotational-echo double-resonance (REDOR) experiments. Similarly, a previous study observed the dipolar coupling between 13C-labeled AmB and phospholipid in 13C{31P} REDOR experiments.45 Another 13C{19F} REDOR study of 13C- and 19F-labeled

AmBs showed a change in intermolecular distance in the presence or absence of Erg.46

Additionally, a recent REDOR study of 13C-labeled Erg and 19F-labeled AmB demonstrated the presence of head-to-head and head-to-tail types of the AmB-Erg complex.47,48 However, the molecular orientation of AmB has not been precisely elucidated due to the poor lateral mobility of the AmB complex in the membrane and the difficulty in preparing oriented lamellar bilayers in the presence of AmB. In this study, we prepared 2H-,

13C,

and

19F-labeled

AmB methyl esters (AMEs, Figure 1) to

elucidate the molecular orientation of AmB using solid-state NMR. AME derivatives were used instead of AmB because the AmB-Erg complex tends to create larger aggregates in the membrane44 compared with AME-Erg assemblies. This aggregate formation hampers the diffusional motion of AmB molecules, which is a prerequisite for NMR-based orientation analysis of a non-oriented bilayer. More importantly, AME exhibits very similar biological activity to AmB,49 indicating a


Page 7 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

close resemblance between AmB-Erg and AME-Erg complexes. By measuring 19F chemical shift anisotropy (CSA), 2H quadrupole splitting, and

13C-19F

dipolar coupling averaged by fast axial

rotation, we can determine the orientation of AME in membranes. This information will help us better understand AmB’s mechanism of ion channel formation, which is probably responsible for its potent antimicrobial activity.

OH O HO

O

OH OH

HO

OH OH O

R2 14 16

COOR1

17

O

OH

41

O

HO

O OH

NH2

R1, R2 =H: Amphotericin B (AmB 1) R1 =CH3, R2 =H: AME (2) R1=CD3, R2 =H: d3-AME (2') R1 =CH3, R2 =F: 14-F-AME (3) R1 =CD3, R2 =F: 14-F-d3-AME (3') R1 =13CH3, R2 =F: 14-F-13CH3-AME (3")

HO

O OH NHAc

R1=H, R2=H: N-Ac-AmB (4)

Figure 1. Chemical structures of AmB 1, AME 2, d3-AME 2’, 14β-F-AMEs 3, 3’, 3”, and N-AcAmB 4.

MATERIALS AND METHODS Materials AmB was purchased from Nacalai Tesque (Kyoto, Japan). Erg was obtained from Tokyo Kasei (Tokyo, Japan), and DMPC was obtained from NOF Corp (Tokyo, Japan). All other analytic-grade chemicals were purchased from standard sources. Column chromatography was performed using silica gel 60 (Merck; particle size, 0.063–0.200 mm, 60–230 mesh). HPLC purification was


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

performed using a COSMOSIL packed column, 5C18-MS-II (Nacalai Tesque, 10 mm I.D. × 250 mm). Thin-layer chromatography was performed on a glass plate pre-coated with silica gel (Merck Kieselgel 60 F254).

Preparation of labeled AmB derivatives We prepared 2H-labeled AmBs (d3-AME) and 19F-labeled AmBs (14β-F-AMEs, 14β-F-d3-AME and 14β-F-13CH3-AME) via chemical reactions with AmB or via the cultivation method (see the supporting information for details, Scheme S1, S2, and Figure S1-S5).

UV spectral measurements AmB derivatives, sterols (Erg or cholesterol), and DMPC were dissolved in a mixed solvent of CHCl3-MeOH (2:1), which was evaporated to form a membrane film. The film was dried under a vacuum for 6 h or longer and then suspended with 8% sucrose aqueous solution (2 mL) by a vortex. The suspension was homogenized with a freeze and thaw cycle, which was performed several times, and then diluted with 8% sucrose solution to obtain a suspension with a concentration of an AmB derivative of 22.4 µM. The UV spectra were recorded on a V630BIO spectrophotometer by a quartz cuvette with a path length of 1.0 cm at 23 °C and 0.1 nm intervals in a wavelength range of 300–450 nm.

Preparation of multilamellar vesicles for solid-state NMR Two different methods of sample preparation were used for d3-AME and 19F-labeled AMEs. For 2H

NMR measurements of d3-AME, Erg and DMPC were dissolved in a mixed solvent of

CHCl3/MeOH (5:3 v/v), which was evaporated to produce a thin film. After being left in a vacuum


Page 8 of 39

Page 9 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

for 8 h, the film was hydrated with H2O and dispersed by vortexing and sonication. Then, the lipid suspension was freeze-thawed five times to form multi-lamellar vesicles (MLVs). The MLV dispersion was lyophilized and rehydrated with deuterium-depleted water (1.0 mg/µL) by vortexing and then packed into a glass tube (5 mm φ), which was sealed with epoxy glue. To suppress the isotropic signal of

19F-labeled

AMEs due to the micelles in the aqueous

dispersion, centrifugation was used to prepare a sample pellet. The other methods were quite similar to those described above for 2H NMR. A lipid film of 14β-F-d3-AME, Erg, and DMPC was prepared and hydrated with deuterium-depleted H2O (30.3 mg/mL) at 50 °C and then dispersed by vortexing. Next, the lipid suspension was freeze-thawed five times to form MLV. The suspension was centrifuged (15,000 rpm, 10 min) and the supernatant was removed by decantation. The obtained precipitate was dispersed again in deuterium-depleted H2O (30.3 mg/mL) and centrifuged (15,000 rpm, 10 min) to create a lipid pellet. Then, the supernatant was removed. For 19F NMR measurements, an approximate 15 mg of the liposome pellet was loaded into a Bruker highresolution magic angle spinning (MAS) insert. Then, the insert was sealed with epoxy glue and placed into an MAS rotor of φ = 4 mm. For 2H NMR measurements, ca. 100 mg of the same liposome pellet was loaded into a 5-mm glass tube and sealed with epoxy glue.

Sample preparation for the powder form of 14β-F-AME We used 14β-F-AME and 14β-F-13CH3-AME for the

19F

NMR measurements and

13C{19F}

REDOR experiments with powder samples, respectively. First, 2.3 mg of 19F-labeled AME was dissolved in 1 mL of CHCl3/MeOH (5:3 v/v) and 22 mg of Celite was added. Then, the solvent was evaporated, resulting in 19F-labeled AME powder supported on Celite. The Celite was packed


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

into a MAS rotor of φ = 4 mm. To keep the 19F-labeled AME powder in the middle of the MAS rotor, 15 mg of Celite was used as a spacer at the top and bottom of the rotor.

Solid-state NMR measurements The 2H NMR spectra were recorded by an AVANCE 400 (2H resonance: 61.441 MHz, Bruker, Karlsruhe, Germany) or ECA 400 spectrometer (2H resonance: 61.369 MHz, Jeol, Tokyo, Japan). For d3-AME, the 2H NMR spectra were recorded at 20 °C and 40 °C using a 5-mm 2H static probe and a solid echo sequence (90°x-t1-90°y-t2-acq). The 2H 90° pulse width was 3.38 µs, and the prepulse and post-pulse echo delays were 7.0 and 5.0 µs, respectively. The relaxation decay was 0.5 s and the spectral width was 400 kHz. For 14β-F-d3-AME, the 2H NMR spectra were recorded at 40 °C with a 5-mm 2H static probe using a magic quadrupolar-echo pulse sequence.50 The 90° pulse width was 4.6 µs, the echo delay was 12.65 µs, the relaxation decay was 0.5 s, and the spectral width was 250 kHz. The 19F NMR spectra in a membrane were recorded using an AVANCE 600 MHz spectrometer (19F resonance: 564.61 MHz; Bruker, Karlsruhe, Germany) at 40 °C using a 4-mm quadrupleresonance (1H/19F/X/Y) MAS probe. The spectra of the powder sample were recorded with 5-kHz MAS, and the spectra of the liposome sample were recorded without MAS. To eliminate the background signal from a Kel-F HR-MAS insert (Bruker, Karlsruhe, Germany), a spin-echo pulse sequence was used after direct excitation of 19F. Decoupling of 1H was performed during both the spin-echo and free induction decay (FID) acquisition periods. The 19F 90° pulse width was 4.5 µs, the echo delay was 20 µs, the spectral width was 113.6 kHz, the 1H decoupling field strength was 60 kHz, and the relaxation delay was 5 s. The 19F chemical shifts were externally referenced to 0.5% trifluoroacetic acid in D2O (−76 ppm).


Page 10 of 39

Page 11 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

The REDOR experiments were performed with an AVANCE 600 MHz spectrometer (19F resonance: 564.61 MHz, 13C resonance: 150.93 MHz, Bruker, Karlsruhe, Germany) for liposomes and an AVANCE 400 MHz spectrometer (19F resonance, 376.61 MHz, and 13C resonance, 100.64 MHz, Bruker, Karlsruhe, Germany) for powder samples using a 4-mm quadruple-resonance (1H/19F/X/Y) MAS probe. The experiments for liposome samples were performed at 40 °C with 7 kHz MAS. For the reference experiment to determine the magnitude of the

13C-19F

dipolar

interaction without fast axial rotation, we used powdered samples rather than the same liposomes at low temperature because the signals of the 42-13CH3 group in liposome samples heavily overlapped with the signal due to the N(CH3)3 groups of DMPC at —10 °C (Figure S8b). The experiments for liposome preparation were performed at 40 °C with 7 kHz MAS. Spectra were recorded under TPPM 1H decoupling with a field strength of 59.5 kHz, and xy-8 phase cycling was used for 19F. The π pulse width of 13C and 19F was 14.0 µs and 11.2 µs for liposome preparation and 8.13 µs and 9.2 µs for the powder, respectively. In the REDOR pulse sequences for the powder samples, 90°-180°-90° composite pulses were used for broadband inversion of 19F nucleus. The spectral width was 59.5 kHz with a relaxation delay of 3 s, and the contact time for crosspolarization transfer was 1.2 ms. More than 12,288 scans were performed for each experiment in liposome samples and 8192 in powder samples.

19F

CSA tensor calculation

The 19F CSA tensor was calculated using Gaussian 09W (Gaussian Inc.) on a personal computer. The simplified structure shown in Figure 4a was adopted as an input file, and the structure was optimized using a second-order Møller-Plleset (MP2) method with the 6-311G (d) basis set. 51 The CSA tensor of this structure was calculated with use of Becke’s three-parameter hybrid


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 39

functional52 and the Lee, Yang, and Parr correlation function (B3LYP)53 with a uniform 6-311++G (2d, 2p) basis set. The input structure shown in Figure 4a was generated by GaussView 5.0.9 software (Gaussian Inc.), and the output structure was visualized by Chem3D (Perkin-Elmer).

RESULTS Preparation of 14β-F- AMEs and d3-AME We attempted to prepare a mechanically oriented membrane for orientation analysis of AmB. However, even under diluted conditions, we failed to observe the 2H signals of aligned AmB in membranes (data not shown). Thus, we focused on preparation of non-aligned liposomes, which required a fast diffusional motion to produce orientation information. However, it has been reported that, at higher concentrations, AmB forms a large aggregate and the 2H NMR signals of AmB and bound Erg disappeared.44 This is the reason that we had to use more mobile derivatives, such as AME, for orientation analysis. A deuterated derivative of AME, d3-AME 2’, was easily prepared by simple esterification with deuterated methyl iodide. More information is necessary to precisely determine its orientation. We derivatized

19F-labeled

AME from AmB using a

regiospecific fluorination reaction at the C14 position, as previously reported. To prepare d3-AME and 19F-labeled AMEs (Scheme 1), the amino group was protected by an Fmoc group and a CD3 group was introduced using 2H-labeled methyl iodide (derivative 5). After removal of the Fmoc group, d3-AME 2’ was obtained. For fluorine labeling, derivative 5 was treated under dehydration and electrophilic fluorination conditions, producing a mixture of NFmoc-14-F-d3-AME and 14α-F/14β-F at a 1:4 ratio. Then, the Fmoc group of the α/β mixture was removed, and 14β-F-d3-AME 3’ was purified using HPLC (see supporting information). Moreover,


Page 13 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

single-labeled 14β-F-AME 3 and 14β-F-13CH3-AME 3” were prepared using non-labeled methyl iodide and 13C-methyl iodide, respectively, instead of 2H-labeled methyl iodide.

OH O HO

OH OH

O

OH

OH

OH

OH OH O

O

Amphotericin B (AmB 1)

COOH

O a, b

HO

O

O

OH OH

c

OH NH2

HO

OH OH

OH OH O

O 6

HO

COOCD3

HO

O

OH OH

O OH

R1 OH R2

OH OH O

O

O 14-F-d3-AME (7), R1=H, R2=F 14-F-d3-AME (3'), R1=F, R2=H

OH NHFmoc

d, e

NHR

OH f, g

COOCD3

HO

O

OH O

O

d3-AME 2', R=H

OH O

OH

OH OH O

5 R=Fmoc

HO

OH

OH COOCD3

O

HO

OH NH2

Scheme 1. Synthesis of labeled AmBs Reagents and conditions: (a) FmocOSu, pyridine, N,N-dimethylformamide (DMF), room temperature, 7 h 89%; (b) methyl iodide-d3, Na2CO3, DMF, room temperature, 23 h, 54%; (c) piperidine, DMF/iPrOH = 3/1, room temperature, 1 h, 68%; (d) TMSOTf, 2,6-lutidine, CH2Cl2, room temperature, 2 h; (e) HF/ pyridine THF, room temperature, 14 h, 43% (two steps); (f) Selectfluor®, DMF/pH 6.8 phosphate buffer=3/1, 0 °C, 10 min, 19%; (g) piperidine, DMSO/iPrOH =3/1, room temperature, 2 h, 31%.

Similarity between AmB and 14β-F-AME in terms of biological activity, molecular assembly, and membrane orientation The biological activities of the fluorinated derivatives were evaluated by using erythrocytes and artificial liposomes. The derivatives induced hemolysis at 3.25 µM against 1% human erythrocytes, producing results comparable to those of AmB (2.18 µM, Table S1). We also evaluated Erg


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

selectivity using K+ flux assays with liposomes.54 The 19F-labeled AME derivative, 14β-F-d3-AME 3’, induced K+-flux in an Erg-selective manner, very similar to AmB, while an inactive derivative, N-Ac-AmB 4, showed neither significant activity nor Erg selectivity (Figure S6). According to previous reports,55–57 the UV spectra of AmB in oligomeric forms differ significantly from that in a monomeric form, depending on the average distance of the heptaene moiety between two neighboring AmB molecules. In addition, the distribution of AmB derivatives into membranes can be estimated based on the wavelength of a monomeric form. To examine the distribution of AME into membrane, and AME-AME interactions in a membrane, we recorded the UV spectrum in Erg-containing DMPC liposomes and compared it with those of AmB and N-AcAmB (Figure 2 and Figure S7).49 The spectrum of 14β-F-d3-AME 3’, which was chiefly used for NMR measurements, was quite similar to that of AmB, but there were slight increases in the intensities of the longer wavelength peaks and a small decrease in the shoulder peak around 330 nm (Figure 2). These differences probably reflect AME’s lower propensity to form large aggregates.57 A rough estimation of interpolyene distances based on previous reports55-57 confirmed their similarity in the assembly structure; the distance between 14β-F-d3-AME heptaene groups (around 5 Å) was estimated to be only 1% greater than the distance between AmB-AmB groups. Additionally, we also confirmed from UV spectra that 14β-F-d3-AME retains sterol selectivity. In the Erg-containing membrane, the absorption peak derived from a monomeric form appeared at around 413 nm, indicating its distribution into the membrane. On the other hand, in the cholesterol-containing membrane, the additional peak appeared at around 406 nm, which implied the micelle formation of AME and/or the partition of AME to aqueous phase (Figure S7b). Furthermore, the absorption peak at around 327 nm, which corresponds to the inactive aggregate, was observed particularly in the cholesterol-


Page 14 of 39

Page 15 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

containing membranes (Figure S7b). The spectrum of the inactive analog, N-Ac-AmB, was significantly different from that of AmB or AME particularly in the longer wavelength region, indicating that the analog was primarily in a monomeric form in the membrane. These results indicate that the AME analog forms a membrane-bound assembly with a similar orientation to that of AmB in Erg-containing membranes. From the ion permeability assay and the UV spectra, we found that introduction of a fluorine atom makes AME more closely resemble AmB in membrane behavior as compared to non-fluorinated AME.58 Thus, we adopted 14β-F-d3AME and Erg-containing systems for further orientation analysis, where the analog was expected to self-assemble into an AmB-mimicking ion channel without forming heterogeneous aggregates that hampered the REDOR analysis.

Figure 2. UV spectra of AmB 1 (blue), 14β-F-d3-AME 3’ (green), and N-Ac-AmB 4 (orange) in 5 mol% Erg/DMPC liposome. The concentration of AmB and its derivatives was 22.4 µM.


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Mobility of AME in membranes Previously, we reported that AmB molecules form large aggregates in membranes at higher concentrations44 and Erg contents are important to avoid aggregate formation.46 Formation of a small assembly with fast axial rotation in the membrane is a prerequisite for orientation analysis of AME in liposomes. Saturated DMPC, which is known to have a higher affinity to AmB than palmitoyloleoylphosphatidylcholine (POPC),59,60 was expected to better disperse the aggregates of AME. Regarding an Erg concentration, it is known that higher sterol contents often produce liquidordered domains (Lo phase) that restrict the mobility of membrane-bound entities. To allow the AME complex to undergo rapid lateral motion, we reduced the Erg content to 10 mol%, which is less than the biological content of Erg in fungal cell membranes. We used to evaluate the mobility of d3-AME by changing the AME/lipid ratio in 10 mol% Erg-containing DMPC bilayers. The 2H NMR spectra of d3-AME in DMPC-Erg vesicles are shown in Figure 3. At 20 °C, the spectral shapes representing a typical Pake doublet and quadrupole splitting value (35.6 kHz) were the same for all AME/lipid ratios and were very similar to the powder pattern, which indicated that the AME molecules were virtually immobilized in the membrane. In contrast, at 40 ºC, a new set of a doublet signals exhibiting a smaller splitting width (21.6 kHz) appeared in addition to the doublet due to immobilized AME. As the AME concentration decreased, the intensity of these signals increased. At 40 °C, with a membrane containing 2 mol% of d3-AME in Erg/DMPC (10:90 molar ratio), the Pake doublet from the immobilized AME disappeared, indicating that the AME molecules primarily underwent fast axial rotation. This AME/lipid ratio and temperature was used for further orientation analysis.


Page 16 of 39

Page 17 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Figure 3. 2H NMR spectra of d3-AME at different d3-AME/lipid ratios and temperatures.

19F

CSA tensor calculation of 14β-F-AME Before performing orientation analysis based on

19F

NMR, it is necessary to determine the

direction and values of the CSA tensor for the 19F nucleus at the labeled position. Although the 19F CSA tensors of a fluorine atom bound to sp2 carbon, such as the F-phenyl group, have been well studied,61,62 the tensors of a fluorine atom bound to sp3 carbon have not been well defined. Thus, we calculated the CSA principal axis system with respect to the molecular frame using the gauge including an atomic orbital (GIAO) method. The partial structure of the tetrahydropyran (THP) ring of 14β-F-AME (Figure 4a) was used as the initial position for calculations. Then, the conformation and CSA principal axis system shown in Figure 4b were calculated as described. To compare the calculated 19F CSA principal values with the experimental values, we measured the 19F

NMR spectrum of powdered 14β-F-AME (Figure 4c) under MAS (5 kHz) and 1H decoupling

conditions and then simulated the spinning side-band patterns using SIMPSON 3.0 software. Table


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 39

1 shows the 19F CSA principal values obtained from the in silico calculation and experiment. The principal values were highly consistent according to CSA; all were within a 6% error range.

Figure 4. (a) Structural motif of the THP ring used for quantum calculation of

19F

CSA. (b)

Conformation of truncated THP ring optimized by MP2 calculation and simulated principal axis system. (c) Experimental (black) and simulated (red)

19F

19F

CSA

NMR spectra for the

powdered sample of 14β-F-AME. Spectra were recorded with a 600-MHz NMR instrument (19F resonance: 564.61 MHz) under 5-kHz MAS conditions. The simulated spectrum was generated by SIMPSON software using the principal values shown as experimental values in Table 1.

Table 1. Principal values (ppm) of

19F

CSA tensor for powdered 14β-F-AME from the

experimental spectrum and computational calculation with the GIAO method.

δ11

δ22

δ33

δiso

experimental

−178.2

−198.3

−224.1

−200.2

calculated

−177.5

−196.3

−226.8

−200.2


Page 19 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

NMR measurements of 14β-F-d3-AME in DMPC-Erg vesicles We mixed 14β-F-d3-AME with a 10 mol% Erg-containing DMPC membrane and performed solid-state 19F and 2H NMR measurements (Figure 5). The signals in the 2H NMR spectrum of 14β-F-d3-AME in DMPC-Erg vesicles (Figure 5a) were relatively broadened. Comparison of the experimental and simulated spectra resulted in an estimated quadrupole splitting value of +/– 24.0 kHz, which is similar to the value of d3-AME (Figure 3), indicating that the fluorination at the  position of C14 hardly disturbed the molecular orientation of AME. We determined the sign of the splitting value to be positive (+24.0 kHz). When the quadrupole coupling value in a high-motion state takes a negative value, its absolute value needs to be smaller than half of the splitting width in a powder state (Δν0), which was approximately 36 kHz in this study (Figure 3). However, the observed splitting width, 24 kHz, was obviously larger than 18 kHz. This quadrupole splitting width was used for further orientation analysis. The 19F NMR spectra at 40 °C (Figure 5b) exhibited a narrower signal than the powder sample (Figure 4c), confirming that 14β-F-d3-AME underwent fast axial rotation under the experimental conditions. The center peak observed in the spectrum (Figure 5b) was determined to be the result of AME-derivative-forming micelles in the bulk water phase, as a similar center peak was observed in the 2H NMR spectrum (Figure 5a). Based on the 19F spectrum, two tensor values, δ ⊥ and δ ∥ , were obtained, and the averaged CSA, Δδ = δ∥−δ⊥, was determined to be 17.0 ppm by fitting the observed signal with the simulated spectra. For further confirmation, we carried out REDOR experiments with 14β-F-13CH3-AME to determine the intramolecular

13C{19F}

13C-19F

dipolar

coupling. The REDOR dephasing magnitude of labeled AME in a powder form was larger than the possible value of intramolecular 13C-19F distance obtained from DFT calculations (Figures 5d and S9), possibly because the close proximity of molecules in the powder form gave the additional


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

intermolecular

13C-19F

dipolar coupling, which can be negligible in the liposome preparation.

When the labeled AME was mixed with DMPC/Erg (Figure 5c), a signal due to the 13CH3 group was observed at around 54 ppm as a very broad peak. The signal intensity was greatly enhanced in the cross polarization spectrum, indicating that this component has strong 1H-13C dipolar interactions (Figure S8a). We also deduced that a sharp signal at 55 ppm, which overlapped with a signal due to the choline group of DMPC, probably contained the signals from the labeled AME in a micellar form (Figure S8a). In the membrane preparation compose of DMPC/Erg, the REDOR dephasings became very weak at dephasing times of 3.3 and 7.4 ms, indicating that fast axial rotation efficiently averaged out the intramolecular dipolar interaction (Figure 5c). Based on this result, 13C-19F intermolecular dipolar coupling was estimated to be between —24.5 Hz and +24.5 Hz under the membrane conditions (Figure 5d).


Page 20 of 39

Page 21 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Figure 5. (a) Experimental (black) and simulated (red) 2H NMR spectra of 14β-F-d3-AME (2 mol% to lipids) in a 10 mol% Erg-containing DMPC membrane. The simulated spectrum was composed of the sum of two quadrupole splitting components (24.0 kHz, light orange; 0 kHz, light blue). (b) Experimental (black) and simulated (red) 19F NMR spectra of 2 mol% 14β-F-d3-AME in 10 mol% Erg-containing DMPC membrane. The simulated spectrum was composed of two CSA components (17.0 ppm, light orange; 0 ppm, light blue). *Isotropic peaks are likely the result of AmB derivatives forming micelles. (c) REDOR spectra of 14β-F-13CH3-AME with (red) and without (black) π pulses on 19F nuclei at an AmB/Erg/DMPC ratio of 1:5:45 and a temperature of 40 °C. The dephasing time was 3.3 ms. The arrow denotes the methyl 13C signal, which shows no significant dephasing. (d) REDOR dephasing values for a powder sample (black open circle) and a liposome preparation (red square). The dephasing curve for the intramolecular dipolar coupling between 13C and 19F (94.5 Hz) based on the conformation optimized with MP2 calculations was shown with the solid black line. One of the best fitted dephasing curve considering the multi-spin system (intermolecular dipolar coupling: 94.5 Hz, intermolecular dipolar coupling: 90.0 Hz, the angle between them: 110°) was plotted with the dashed black line. The dephasing curve for 24.5 Hz (solid red) of dipolar coupling was also plotted as the largest possible dipolar interaction derived from the experimental values within the experimental errors.

Orientation of AME in membrane Based on the 2H-quadrupole splitting, 19F CSA values, and 13C-19F dipolar coupling obtained from the spectra in Figure 5, the orientation of an AME molecule in the membrane were examined. Before beginning orientation analysis, we calculated the rotational potential energy with respect to the rotatable bonds between the macrolactone ring and the CD3 group in the ester moiety. Free


Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

energy curves were obtained for the dihedral angles around the C16-C41 and C41-O42 bonds via DFT free energy calculations (Figure S10). The results showed that the C41-O42 bond is exclusively in the anti-conformation since its free energy is lower than all others by 15 kcal/mol (Figure S10b). The free energy curve for the C16-C41 bond shows several local minima within the 2 kcal/mol energy difference (Figure S10b). Nevertheless, the O-C42 bond can be assumed to have the same orientation as the C16-C41 bond since rotation around the C16-C41 bond, which is oriented in parallel to the O-C42 bond, does not affect the direction of the CD3 group with respect to the molecular axis. To calculate the orientation of the O-C42, C14-14β-F, and C42-14β-F vectors, we assumed that one global molecular order parameter, Smol, could be used as a scaling factor for 2H quadrupole splitting, 19F CSA, and dipolar coupling. We also expected the 13C-19F dipolar interaction to have a different Smol since it encompasses a highly rotatable bond. However, the deviation in Smol had no significant influence on the orientation of the molecule in the membrane (Figure S11). The tilt angle, τ, was defined as the angle between the bilayer normal and the O17-C16 vector (Figure 6a). The azimuthal angle, ρ, was defined as shown in Figure 6. To simulate the Δδ and Δν values, the THP ring with a structure optimized by MP2 calculations was placed in an orthogonal coordinate system in which the z-axis was parallel to the bilayer normal. When (τ, ρ) was (0°, 0°), O17 was located at the origin, C16 was on the z-axis, and C41 was on the y-z plane (Figure 6b). These angles were successively changed by 1°, and a total of 180×180 orientations were generated for the THP ring via usual coordinate transformation. As the experimental values for an immobilized state, we used the quadrupole splitting value of d3-AME at 20 °C (35.6 kHz), 19F CSA of 14β-F-d3-AME in a powder state, and theoretical value of intramolecular dipolar coupling between 19F and 13C of 14β-F-13CH3-AME 3’’. The same set of experimental values obtained with liposome


Page 22 of 39

Page 23 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

preparations at 40 °C were used for an axial rotation system. Note that the orientation of the overall macrolactone ring can be roughly estimated based only on the orientation of the THP ring, since the conformation of the macrolactone ring is rigid and can be assumed to be essentially same as the crystal structure of N-iodoacetyl-AmB.63–65

Figure 6. (a) Molecular coordinates and angles defining the orientation of AME with respect to the membrane normal, n. The tilt angle, τ, was defined as the angle between the bilayer normal, n, and the O17-C16 vector, z, and the azimuthal angle, ρ, was defined as the dihedral angle between the plane formed by vectors n and z and the plane formed by the three atoms of C16, O17, and C41, respectively. (b) Molecular coordinates of the THP ring. The 19F CSA principal axes, 41OC42 vector, and 14F-C42 vectors are shown in red, blue, and gray, respectively.

The orientations that fulfill the experimental values of 19F CSA, 2H quadrupolar coupling and 13C19F

dipolar coupling are plotted in black in Figures 7a, 7b, and S11. When the Smol value fell 23 ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

between 1.00 and 0.94, the possible range was separated into two regions, one is around τ = 23° and another around τ = 5°. As Smol decreased, these two regions became closer. When Smol was between 0.93 and 0.88, these two regions merged at around τ = 13°. The overall range of these three experimental values was 4°

Perpendicular orientation of amphotericin B methyl ester in hydrated

Recommend Documents