A new azobenzene-based design strategy for detergents in membrane protein research

Mass spectrometry enables the in-depth structural elucidation of membrane protein complexes, which is of great interest in structural biology and drug discovery. Recent breakthroughs in this field revealed the need for design rules that allow fine-tuning the properties of detergents in solution and gas phase. Desirable features include protein charge reduction, because it helps to preserve native features of protein complexes during transfer from solution into the vacuum of a mass spectrometer. Addressing this challenge, we here present the first systematic gas-phase study of azobenzene detergents. The utility of gas-phase techniques for monitoring light-driven changes of isomer ratios and molecular properties are investigated in detail. This leads to the first azobenzene detergent that enables the native mass spectrometry analysis of membrane proteins and whose charge-reducing properties can be tuned by irradiation with light. More broadly, the presented work outlines new avenues for the high-throughput characterization of supramolecular systems and opens a new design strategy for detergents in membrane protein research.


Synthesis and Characterization
General Remarks to Synthesis. OGBAs and OGDs were synthesized by means of previously published procedures. 1-3 Synthesis protocols that led to the obtainment of OGBAs 1 -5 have been published before. 2,3 The general synthesis route that was used to obtain OGDs has also been published before. 1 In particular, the detergent batches 6 -10 are new and related synthesis protocols have not been reported before.
Chemicals were purchased from Sigma-Aldrich (Germany), Acros Organics (Germany), Alfa Aesar (Germany), Fluka (Germany), Fisher Scientific (Germany), Merck (Germany), TCI (Germany). Chemicals were used as supplied. Ethyl acetate (EtOAc) and n-hexane were distilled before they were used. Other solvents, such as methanol (MeOH), dimethylformamide (DMF), dichloromethane (DCM), and acetonitrile (ACN), were used as supplied. Dry solvents were purchased in bottles sealed with a septum or tapped from a solvent purification system (MS-SPS-800) that was purchased from M. Braun (Germany). Deionized water (H 2 O) used for synthesis was provided by a deionization system installed in the Freie Universität's Institute of Chemistry and Biochemistry. Argon was purchased from Linde (Germany). For working under dry and oxygen-free reaction conditions, chemicals and solvents were handled under argon atmosphere. To support dry conditions the glassware was evacuated, heated up to 300 °C with a heat gun, and filled with argon. Dendritic compounds, such as mesylated and acetal-protected [G2] dendron ([pG2]-OMes), and 4,4´-dihydroxyazobenzene were synthesized in analogy to previously published procedures. 1, [3][4][5] Monitoring of reactions and purification procedures was achieved by normal phase (NP) thin-layer chromatography (TLC) analysis.
NP TLC plates (DC-Fertigfolien ALUGRAM® Xtra SIL G/UV254) based on silica (SiO 2 ) were purchased from Macherey-Nagel (Germany). Silica gel (60 M) for preparative normal phase column chromatography was purchased from Macherey-Nagel. For NP TLC analysis and manual NP column purification mixtures of organic solvents (v:v) were prepared. If necessary, MeOH was added in percent per volume to the prepared mixtures (v:v + v%). TLC plates were either analyzed under UV irradiation (254 nm) using a lamp from CAMAG (Germany) or by staining the TLC plates with cerium reagent (940 mL H 2 O, 60 mL H 2 SO 4 , 25 g molybdic acid, 10 g cerium(IV) sulfate). For the staining process, the TLC plates were fully submerged into the cerium reagent, excess of staining reagent was wiped off with cellulose, and the plate was heated up to 300 °C with a heat gun until staining was completed.
For high-pressure liquid chromatography (HPLC) purifications a setup from Knauer was used, which consisted of a Smartline Manager 5000 (+ interface-module), two Smartline Pumps 1000, a 6-port-3-channel-injection valve, a sample loop (10 mL), UV Detektor 2500 und a high pressure gradient mixer. Spectra were recorded with a x-y-plotter (Knauer). A reversed-phase (RP) Luna™ C18(2) (10 μm, 250 x 21.20 mm, Phenomenex®) column was used as stationary phase, which was further equipped with a Security Guard™ PREP Cartridge Holder Kit (21.20 mm, ID, Phenomenex®). The setup was constructed by Dr. Carlo Fasting. Components of the mobile phase, such as H 2 O and ACN, were supplemented with 0.1% trifluoroacetic acid and degassed prior usage. The flow rate was adjusted to 20 mL/min and the detection wavelength was 360 nm.
Mass spectra were acquired on an Agilent 6210 ESI-TOF (ESI-ToF) from Agilent Technologies (Santa Clara, CA, USA). The solvent flow rate was adjusted to 4 μL/min and the spray voltage was set to 4 kV. Drying gas flow rate was set to 15 psi (1 bar). All other parameters were adjusted for a maximum abundance of the relative [M+H] + . The instrument was operated by the Core Facility BioSupraMol of the Freie Universität Berlin. 1 H NMR, 13  General Procedure for the Synthesis of C n -OMes (n = 8 -12) The desired alcohol (5 g) was dissolved in toluene (150 mL). Triethylamine was added (2 eq.) and the reaction flask was cooled with an mixture of water and ice (w:w, 1:1). Methanesulfonyl chloride (1.2 eq.) was added and the mixture was allowed to warm up to room temperature over a time period of 16 hours. Massive precipitation of triethylamine hydrochloride sometimes interrupted the stirring process. In this case, more toluene was added and the precipitate was suspended manually using a spatula. After 16 hours, the precipitate was filtered off with a filter paper and solvent was removed under reduced pressure. Column chromatography (SiO 2 , DCM) gave the desired product as colorless oil (n = 8, 91%; n = 9, 97%; n = 10, 98%; n = 11, 99%; n = 12, 69%).  concentrated hydrochloric acid (1 mL, 37%). The aqueous layer was extracted with DCM. Organic layers were combined and dried over magnesium sulfate, and magnesium sulfate was filtered off using a filter paper. The raw product was immobilized on silica gel by adding SiO 2 (25 -50 g) to the filtrate and removing the solvent under reduced pressure. Subsequent column chromatography (SiO 2 , DCM) gave the desired product (n = 8, 33%; n = 9, 55%; n = 10, 25%; n = 11, 33%; n = 12, 27%).

Quantification of Isomers in Solution
The cis/trans equilibrium of 3, 6, 8, and 10 was monitored in solution after full equilibration under irradiation at 366 nm either by UV/VIS spectroscopy and/or RP HPLC analysis. The applied light source is a UV/VIS lamp from CAMAG and was described previously in more detail. 2 The time required to achieve full saturation of cis under irradiation at 366 nm was determined as shown exemplarily in Figure S1. Isomer ratios are summarized in Table S1 - Fig S5). 7 Both instruments were equipped with a nano electrospray ionization (nESI) source and mass-calibrated using a solution of caesium iodide (c = 100 mg/mL).

Gas-Phase Infrared Spectroscopy
To obtain IR spectra in the gas phase, an infrared laser of high intensity was applied to allow for the absorption of multiple photons at IR-active sites of ions. This resulted in increased internal energies and dissociation of covalent bonds within ions. This way of obtaining an IR spectrum is called: IR multiple photon dissociation (IRMPD) experiment. 7 Here, the Fritz Haber Institute free electron laser (FEL) was coupled to the DT IM-MS instrument, which enabled spectroscopic investigation of drift-time selected species. Subsequent to the drift tubes, ions within a narrow drift-time window were selected using the quadrupoles and irradiated by a pulse of intense IR light. Subsequent fragment analysis occurred by means of TOF MS. An action spectrum was recorded by monitoring the intensity of the precursor ion signal and the corresponding fragment signals as a function of the laser wavelength.

Membrane Protein Solubilisation
Purification and preparation of AmtB for native MS. AmtB was overexpressed as an AmtB-MBP construct in E. coli and purified as described before. 13 The protein-containing membranes were solubilized at 4 °C in resuspension buffer (100 mM NaCl, 20 mM Tris,  5 V). Sample cone, extraction cone, trap DC bias, and source pressure were optimized for maximum ion intensity of the AmtB trimer. The protein sample was analysed before and after irradiation at 366 nm using similar instrumental parameters. The mass spectra were acquired and analysed using MassLynx and plotted using Origin V9.1.
CD Spectroscopy. AmtB solubilized in DDM (concentration of the AmtB trimer = 4.6 µM) was transferred into ammonium bicarbonate buffer (NH 5 CO 3 , 100 mM, pH = 8) containing DDM, trans 6, or a cis/trans mixture of 6 obtained after full conversion of the isomer mixture under irradiation at 366 nm. The protein solutions were loaded into cuvettes (Quartz Suprasil, volume = 300 μL, layer thickness = 1 mm) and analyzed using a CD spectrometer (Chirasan, USA). The setup was flushed with nitrogen gas overnight and the lamp as well as the thermostat was turned on 30 min before the first measurement. The experimental parameters were as follows: final concentration of the AmtB trimer (0.14 µM), temperature (22.5 °C), wavelength range (190 -260 nm), step size (1 nm), scan speed (0.5s/nm), bandwidth (1 nm), and repeats per sample (4). Detergent-containing NH 5 CO 3 buffers were used as blanks. Data were acquired with Pro-Data Chirascan V4.5 and analyzed with Origin V9.1 (see Fig.S8). The average intensity of four scans was plotted against the wavelength (Fig. S8). Dynamic Light Scattering. The cac of OGD 6 was determined by dynamic light scattering (DLS). A dilution series was prepared in MilliQ water with concentrations of 6 between 10 -8 and 10 -2 mol·L -1 . The samples were filtered (0.22 µm, RC) and equilibrated in the dark for at least five days at room temperature (approximately 22 °C) to ensure a maximum saturation of the trans form. The samples were analysed in cuvettes (Quartz Suprasil, width x length: 2 mm x 10 mm) using a Zetasizer Nano-ZS ZEN3600 (Malvern, UK). We found no evidence that the laser irradiation of this Zetasizer instrument (red laser, 632.8 nm) is affecting the cis/trans ratio of our amphiphiles in solution. 3 The instrumental parameters were as follows: material (polystyrene latex), dispersant (water), sample viscosity parameters (use dispersant viscosity as sample viscosity), temperature (22.5 °C), equilibration time (120 s), cell type (quartz cuvettes), measurement angle (173° backscatter), measurement duration (manual), number of runs (11), run duration (10 s), number of measurements (3), delay between the measurements (0 s), data processing (general purpose, normal resolution). The derived count rate values obtained from three measurements per concentration were averaged. The unit of the derived count rate is kilo counts per second (kcps). The logarithm of the derived count rate was plotted against the logarithm of the concentration. The double logarithmic plots showed two characteristic regions: (1) a flat region with low count rates at lower concentrations of 6 and (2) a linear growth of the count rate at higher concentrations of 6. Both regions were fitted to linear functions and the intersection was taken as the cac value (Fig. S9). 14 A monomodal average size distribution of particles was obtained above the cac, before and after irradiation at 366 nm (volume%). The diffusion coefficients related to the obtained intensity maxima (65.4 µm·s -2 ) and the calculated hydrodynamic diameters (5.61 nm) indicate that both trans 6 and its isomer mixture after irradiation at 366 nm are forming micelles above a concentration of 250 µM (Fig. S9). The diffusion coefficients of the aggregates formed (65.4 µm·s -2 ) as well as the calculated hydrodynamic diameters (5.61 nm) match with those that have been previously obtained from micelle-forming OGDs. 15 The DLS data therefore indicate that both trans 6 as well as the isomer mixture obtained after irradiation at 366 nm are forming micelles.

Remarks to the Design of [G2]etherC18, [G2]amideC18, and [G2]triazoleC18.
For the formation PDCs it is important that both protein and detergent are sufficiently hydrophobic. A substitution of azobenzene in OGD 6 with alternative linker groups, such as ether, amide, or triazole, would lead to a detergent with a [G2] head group and a comparatively short hydrophobic tail, e.g., C8. Such a detergent would be very hydrophilic. In order to increase the hydrophobicity of the detergent, we added ten more carbon atoms to the hydrophobic chain. We assume that the length of the hydrophobic chain is important for the formation of PDCs in solution and that it has a negligible impact on the basicity of the detergent compared to that of the alternative linker groups, such as ether, amide, and triazole. Estimating the Gas-Phase Basicity of OGD 6. Mass spectra obtained upon dissociation of PDCs show fractions of non-chargereduced BGL charge-reduced BLG ( ), and singly-charged detergent ions (see Figure 5 in manuscript). To estimate the gas-phase basicity of OGD 6 we assumed that the fractions of charge-reduced and non-charge-reduced BLG ions is proportional to the gas-phase basicity of the detergents' linker between head and tail ( ): The fractions of charge-reduced BLG (%) obtained from OGDs in which azobenzene was substituted for alternative linker groups, such as ether, amide, or triazole, have recently been reported elsewhere. 16 Here, the fractions of charge-reduced BLG (%) were plotted against the gas-phase basicity of the detergents' linker groups of [G2]etherC18, [G2]amideC18, and [G2]triazoleC18. The average gas-phase basicity values of OGD 6, before and after irradiation at 366 nm, were extrapolated by assuming a linear correlation between the fraction of charge-reduced BLG (%) and gas-phase basicity of the detergents' linker groups under the experimental conditions employed (Fig.12).