High-Resolution Studies of Proteins in Natural Membranes by Solid-State NMR

Summary

This work details robust basic routines on how to prepare isotope-labeled membrane protein samples and analyze them at high-resolution with modern solid-state NMR spectroscopy methods.

Abstract

Membrane proteins are vital for cell function and thus represent important drug targets. Solid-state Nuclear Magnetic Resonance (ssNMR) spectroscopy offers a unique access to probe the structure and dynamics of such proteins in biological membranes of increasing complexity. Here, we present modern solid-state NMR spectroscopy as a tool to study structure and dynamics of proteins in natural lipid membranes and at atomic scale. Such spectroscopic studies profit from the use of high-sensitivity ssNMR methods, i.e., proton-(¹H)-detected ssNMR and DNP (Dynamic Nuclear Polarization) supported ssNMR. Using bacterial outer membrane beta-barrel protein BamA and the ion channel KcsA, we present methods to prepare isotope-labeled membrane proteins and to derive structural and motional information by ssNMR.

Introduction

Structural and motional studies of membrane proteins in physiologically relevant environments pose a challenge to traditional structural biology techniques¹. Modern solid-state nuclear magnetic resonance spectroscopy (ssNMR) methods offer a unique approach for the characterization of membrane proteins^2,3,4,5,6,7 and has long been used to study membrane proteins, including membrane embedded protein pumps⁸, channels^9,10,11, or receptors^12,13,14,15. Technical advances such as ultra-high magnetic fields >1,000 MHz, fast magic angle spinning frequencies >100 kHz, and hyperpolarization techniques¹⁶ have established ssNMR as a powerful method for the study of membrane proteins in environments of ever-increasing complexity from liposomes to cell membranes and even whole cells. For example, DNP has become a powerful tool for such experiments (see reference^{17,18,19,20,21,22,23,24,25}). More recently, ¹H-detected ssNMR offers increasing possibilities to study membrane proteins at high spectral resolution and sensitivity^{25,26,27,28,29}. This work highlights two bacterial membrane proteins that are involved in essential functions, i.e., protein insertion and ion transport. The corresponding proteins, BamA^{25,30,31,32,33} and KcsA^{23,27,28,34,35,36,37,38,39} (or chimeric variants thereof^10,40) have been examined by ssNMR methods for more than a decade.

A representative protocol for the preparation and ssNMR characterization of bacterially originating membrane proteins is presented here. The different steps of the protocol are shown in Figure 1. First, the expression, isotope-labeling, purification, and membrane-reconstitution of BamA is explained. Then, a general workflow for the characterization of the membrane protein by ssNMR is presented; specifically, the assignment of membrane protein backbones using ¹H-detected ssNMR at fast magic angle spinning. Finally, basic setup and acquisition of dynamic nuclear polarization-(DNP)-supported experiments, which significantly boost ssNMR signal sensitivity, are detailed.

Protocol

1. Production of uniformly labeled ²H, ¹³C, ¹⁵N-labeled BamA-P4P5

NOTE: While this protocol requires working with non-pathogenic Gram-negative bacteria, adherence to basic biological safety procedures is a must, namely, wearing safety glasses, lab coats, gloves, and following institutional standard operating procedures for work with microorganisms.

1. Use a single colony of E. coli BL 21 Star (DE3) containing the pET11aΔssYaeT plasmid encoding for E. coli BamA-P4P5 to inoculate 50 mL of Lysogeny broth supplemented with 50 µg/L of ampicillin.
2. Grow the culture at 37 °C at 200 rpm until an OD₆₀₀ of 0.6 is reached. Spin at 2,000 x g for 10 min. Resuspend the pellet in 50 mL of M9 (see Table 1) to a maximum OD₆₀₀ of 0.1.
   NOTE: All future growing steps take place at the conditions stated above unless otherwise noted.
3. Grow the culture until OD₆₀₀ of 0.5. Spin at 2,000 x g for 10 min at room temperature. Resuspend the pellet in 50 mL of M9 containing 90% D₂O (see Table 1 for recipe) to a maximum OD₆₀₀ of 0.1. Grow the culture overnight at 30 °C at 200 rpm.
4. Spin down the culture at 2,000 x g for 10 min at room temperature. Resuspend the pellet in the prewarmed 50 mL 90% D₂O M9 to a maximum OD₆₀₀ of 0.1. Grow until an OD₆₀₀ of 1.0 is reached.
5. Spin down the culture at 2,000 x g for 10 min at room temperature. Prepare 100 mL of 100% D₂O M9 medium with isotope-enriched amounts of non-enriched glucose and ammonium chloride. Resuspend in 100% D₂O M9 medium to a maximum OD₆₀₀ of 0.1. Grow until an OD₆₀₀ of 0.7 is reached.
6. Spin down the culture at 2,000 x g for 10 min at room temperature. Discard the supernatant and resuspend the pellet in 500 mL of 100% D₂O M9 medium with isotopes (see Table 1) to a maximum OD₆₀₀ of 0.1.
7. Allow the culture to grow until an OD₆₀₀ of 0.6-0.9 is reached. Induce with 1 mM of isopropyl β-D-1-thiogalactopyranoside (IPTG). Express for 4 h and harvest cells at 4,000 x g for 15 min at room temperature.

2. Purification, refolding, and BamA-P4P5 proteo-liposome formation

NOTE: All the steps of this section should be conducted in a fume hood. Special care must be taken when opening tubes post-centrifugation limits harmful aerosols.

1. Thaw the pellet on ice. Resuspend the pellet in 20 mL of cold Buffer 1. See Table 2 for the Buffers.
2. Spin down the solution at 4,000 x g for 20 min at 4 °C. Resuspend the pellet in 20 mL of cold distilled H₂O. Allow the suspension to incubate on ice for 10 min.
3. Spin down the suspension at 4,000 x g for 20 min at 4 °C. Resuspend in 10 mL cold Buffer 2 and allow to incubate on ice for 30 min.
4. Supplement the mixture with an additional 10 mL of cold buffer 2 and incubate for a further 30 min on ice. Add 0.5% n-Dodecyl-B-D-Maltoside (DDM) and swirl gently.
5. Sonicate the sample on ice at 13 kHz until vicious-use 10 s long on/off pulses. Spin down at 25,000 x g for 20 min at 4 °C. Wash three times with 20 mL of Buffer 3. Centrifuge at 25,000 x g for 20 min at 4 ˚C.
6. Resuspend the pellet in 20 mL of Buffer 4. Incubate at 37 °C for 30 min. Sonicate the sample on ice at 13 kHz, until clear-use 10 s long on/off pulses.
7. Spin down the sample at 25,000 x g for 20 min at 4 °C. Repeat steps 2.4 to 2.6 omitting addition of protease inhibitor and incubation at 37 °C.
8. Wash the pellet with 20 mL of water and twice with Buffer 3. Spin down at 25,000 x g for 20 min at 4 °C. Aliquot the suspension into 1 mL microcentrifuge tubes and spin the microcentrifuge tubes at maximum speed on a bench-top centrifuge for 20 min. Inclusion body purity is assessed by 10% SDS-Page gel, see Figure 2A. Expected yield for purified BamA-P4P5 triply (²H,¹³C,¹⁵N) labeled is 30 mg/L.
9. Solubilize inclusion bodies in 200 µL of Buffer 5 and 300 µL of H₂O. Add 6M guanidium chloride (GdnCl). Adjust the volume of microcentrifuge tube to 1 mL with H₂O. Vortex the mixture and incubate for 4 h at room temperature. Vortex the microcentrifuge tubes periodically (every 30 min).
10. Spin at 100,000 x g for 1 h at 4 °C. Use Beer-Lambert's Law to determine protein concentration by 280 nm. The molar extinction coefficient is 117,120 M^-1cm^-1. If concentration is >100 µM dilute with 1x Buffer 5 with 6 M GdnCl. Rapidly dilute the protein 10x in Buffer 6. Incubate the mixture overnight at room temperature.
    ​NOTE: For ultracentrifugation, use ultracentrifugation grade 1.5 mL tubes.
11. Determine refolding efficiency using a semi-native SDS-Page gel. Take two 10 µL aliquots from step 2.10 and add 10 µL of Lameili buffer (buffer 8) to each aliquot. Boil one sample for 10 min at 95 °C; leave the other aliquot on ice. Afterwards, run both samples at 14 mA on a 10% SDS-Page gel. The stacking and running portions of the gels lack reducing agent and SDS. Protein staining is achieved using Coomassie dye. The expected results are shown in Figure 2B.
12. Spin down the sample at 4,000 x g for 20 min at 4 °C. Concentrate the sample to a working volume of 8 mL using a 30 kDa centrifugal filter. Spin the sample at 4,000 x g for 20 min at 4 °C. Measure protein absorbance at 280 nm. As in step 2.10, determine the protein concentration.
13. Calculate the amount of lipid required for a 10:1 lipid to protein ratio (LPR-mol/mol). From a chloroform stock, add the required amount of lipids, as calculated in step 2.12, into a 100 mL round bottom flask (RBF). Evacuate the chloroform from the RBF under a gentle stream of nitrogen. Place RBF on high vacuum for 3 h. Hydrate the lipid film with 1 mL of Buffer 7. Incubate RBF for 10 min at 37 °C.
14. Add the protein mixture (from step 2.12) to the RBF.
15. Dilute the protein/lipid mixture (from step 2.14) to the final volume of 50 mL in the RBF. Use Buffer 7 supplemented with 1x protease inhibitor dissolved in DMSO and incubate for 30 min at 37 °C.
16. Dialyze against 100 volumes of Buffer 7 for 2 weeks (use 12-14 kDa molecular weight cut-off dialysis tubing). For the first 24 h, perform the dialysis at room temperature with addition of fresh Buffer 7 every 8 hrs. Afterwards, change the buffer once daily, and perform dialysis at 4 °C.

3. Filling of the ssNMR rotor

1. After step 2.16, centrifuge for 30-60 min at 10,000 x g at 4 °C so that a pellet is formed. Remove the supernatant with a pipette.
   NOTE: The technique discussed below has been optimized for a 3.2 mm rotor but is applicable for almost any rotor diameter.
2. Pipette the sample into the rotor and gently spin the same down with a table centrifuge.
   NOTE: Depending on sample consistency, one may either use a densification tool to compact the sample, or a spatula instead of a pipette to put the sample into the rotor.
3. Repeat this process 2-3x until the rotor is filled to the correct height and check if there is enough space for the cap left, as indicated by the marker line on the densification tool. Close the rotor with the cap positioning tool on an even surface. Mark half the bottom edge of the rotor.
   NOTE: New ssNMR rotor generations come with a durable laser mark at the bottom edge for the optical detection of the rotor spinning frequency. Otherwise, the use of a sharpie pen is suggested; as the MAS detector has been optimized specifically for sharpie ink.
4. Use a magnifying glass to check whether the cap is well placed.

4. Sample characterization by 2D ¹³C- ¹³C ssNMR spectroscopy

1. Insert the rotor into the magnet, using the automatic spinning routine of the MAS unit, spin the sample up to the desired MAS frequency between 10 and 20 kHz MAS while cooling the sample to 270 K set temperature. Tune and match the ¹H and ¹³C channels.
   NOTE: For the choice of the MAS frequency, avoid accidentally matching the chemical shift distance between the spectral Cα region and the Carbonyl carbon region to the spinning frequency, i.e., avoid the first order rotational resonance condition⁴¹.
2. Optimize a ¹H-¹³C cross-polarization⁴² (CP) experiment by varying the rf power on the proton channel from 25-80 kHz. Typical parameters are 45 kHz rf power on ¹³C, 80-100 kHz heteronuclear ¹H-decoupling⁴³ during acquisition, 1.8-2.0 s recycle delay, and 64 scans.
   1. Optimize the CP contact time for maximal signal sensitivity on the aliphatic carbons.
   2. Determine ¹H 90° pulse length in the ¹³C-CP experiment by multiplying the starting pulse by four, i.e., by optimizing a 360° pulse, and vary the length until the signal has disappeared. Include a rectangular pulse directly after the CP transfer and optimize the ¹³C 90° pulse length analogously.
3. Run a 2D ¹³C-¹³C proton-driven spin diffusion (PDSD)-type experiment such as PARIS with 30 ms magnetization transfer time to detect intra-residual correlations and gauge the sample quality. Transfer the power-levels, the contact time, and 90° pulse lengths from the previously optimized ¹³C-CP experiment. Use an acquisition time of 4-6 ms in the indirect dimension. Process the spectrum with squared sin bell functions (QSIN) or, for insensitive samples, exponential line broadening in both dimensions.
4. Extract slices of isolated cross-peaks and determine the linewidth at half heights. A linewidth between ~0.3-1.0 ¹³C ppm is indicative of a well-ordered sample, ideal for ssNMR characterization, whereas a linewidth above 1.0 ¹³C ppm is indicative of conformational heterogeneity that may compromise ssNMR characterization. The 2D CC PARIS spectrum of BamA shown in Figure 3A is an example of a well-ordered protein that gives high-quality spectra.

5. Backbone assignment by ¹H-detected 3D ssNMR spectroscopy

1. Acquisition of 2D NH spectral fingerprints
   1. Fill the sample in a 1.3 mm rotor, as described above. Insert the rotor into the magnet, spin the sample to 10-20 kHz MAS and cool the sample to 240-250 K set temperature or lower, depending on the specifications of the ssNMR probehead. Use the MAS unit interface to increase the MAS frequency gradually in steps of 5 kHz to 60 kHz MAS.
   2. Optimize the CP amplitudes, contact times, and 90° pulse lengths in ¹³C- and ¹⁵N CP experiments. Use amplitudes around 30-50 or 70-100 kHz on the heteronuclear channels and vary the amplitude on the ¹H-channel from 80-180 kHz. Use a recycling delay of 0.7-1.2 s, along with 15 kHz low-power heteronuclear proton decoupling⁴⁴. Also, see reference⁴⁵ for further practical details on how to set up CP ssNMR experiments.
   3. Acquire a 2D NH experiment to gauge the ¹H-resolution. Transfer the ¹⁵N-CP parameters and use approximately 15 and 25 ms acquisition time in the direct and indirect dimensions. For the first ¹H to ¹⁵N CP transfer, use the previously optimized contact time. For the ¹⁵N to ¹H backtransfer, use 0.7-1.0 ms contact time to minimize interresidual magnetization transfer.
   4. Use 50-250 ms water MISSISSIPPI suppression⁴⁶, dependent on sample water content. Process the spectrum with squared sin bell functions (QSIN) in both dimensions. Extract slices of isolated cross-peaks and determine the linewidth at half heights. An example for a high-quality 2D NH spectrum²⁵ is shown for membrane-embedded BamA-P4P5 in Figure 3B.
2. Backbone assignments
   1. Optimize a 1D ¹⁵N to ¹³Cα specific CP experiment⁴⁷. Place the ¹³C carrier in the middle of the Cα region around 55 ppm. Use a ¹³C rf amplitude of 20 kHz and vary the ¹⁵N amplitude from 35 to 45 kHz for maximal transfer to the Cα signals and minimal transfer to the carbonyl signals. Optimize the contact time from 3 to 10 ms in increments of 1 ms.
   2. Run a 3D CαNH experiment. Transfer all CP amplitudes, contact times, and pulse lengths from the previously optimized steps. Use approximately 10 ms and 30 ms in the indirect and direct acquisition dimensions. Process the spectrum with squared sin bell functions (QSIN) in all three dimensions.
   3. Optimize a 1D Cα to CO DREAM magnetization transfer⁴⁸. Start with a ¹⁵N to ¹³Cα specific CP transfer. Then, transfer the magnetization from Cα to CO using a spin lock on the ¹³C channel. Vary the rf amplitude from 20 to 35 kHz for optimal transfer. Optimize the transfer time from 4 to 10 ms.
   4. Run a 3D Cα(CO)NH experiment. Transfer all rf amplitudes, contact times, and pulse lengths from the previously optimized steps. Use approximately 10 ms and 30 ms in the indirect and direct acquisition dimensions. Process the spectrum with squared sin bell functions (QSIN) in all three dimensions.
   5. Repeat and adapt steps 1-4 for the analogous 3D CONH and CO(Cα)NH experiments. Use this quartet of 3D experiments for sequential Cα and CO walks to assign the protein backbone, as shown on the example of KcsA in Figure 4²⁸.

6. Protein dynamics by ¹H-detected ssNMR spectroscopy

1. For ¹⁵N T₁ studies, use the previously optimized 2D NH experiment at 60 kHz MAS and include a delay after the first ¹H to ¹⁵N cross-polarization step. Acquire a series of 2D NH experiments and vary the delay from 0 to 16 s using steps of, e.g., 0, 2, 4, 8, and 16 s.
2. Plot the normalized intensities for resolved signals as a function of the increasing delay time. Fit the T₁ decay to single exponentials according to y = exp^(-x/T1), with y being the signal intensity at decay time x.
3. Analogously, for ¹⁵N T_1rho studies, include a spin-lock pulse in the pulse program on the ¹⁵N channel with an rf amplitude between 15-20 kHz after the first ¹H to ¹⁵N cross-polarization step^26,49. Acquire a series of 2D NH experiments and vary the delay from 0 to 150 ms and fit the T_1rho decay of resolved signals to single exponentials. Illustrative ¹⁵N T_1rho data is shown for membrane-embedded K⁺ channels in Figure 5^27,28.

7. Dynamic nuclear polarization

NOTE: The following preparative steps relate to the use of a commercial DNP setups using 3.2 mm sapphire MAS rotors (Figure 6)²⁰. Use of the zirconia rotors or other DNP equipment may lead to lower DNP signal enhancements.

1. Resuspend membrane protein pellet (from step 3.1) in 50 µL of deuterated DNP buffer (60% (v/v) deuterated d8- and glycerol, and 15 mM AMUPol⁵⁰. For experiments at 800 MHz, we recommend using 10 mM NATriPOL-3⁵¹. Also see reference⁵² for further practical aspects to DNP ssNMR experiments.
2. Centrifuge for 10-20 min at 100,000 x g at 4 °C so that a pellet is formed. Remove the supernatant with a pipette.
3. Precool the centrifugal adaptor and the 3.2 mm sapphire MAS rotor components to <4 °C. The following steps (7.4-7.7) need to be done as quickly as possible to prevent the DNP agents from getting reduced.
4. Place the 3.2 mm sapphire rotor in the centrifuge adapter and fill it with the suspension of proteo-liposomes from step 7.2.
5. Pipette the resuspended membrane protein sample against the inner wall of the rotor carefully using a 200 µL pipette tip. Allow the solution to slide down to the bottom of the rotor. Ensure that air bubbles are not formed.
6. Spin down the sample in the rotor for 5-10 mins at 4,500 x g at 4 °C. Remove the supernatant containing excess DNP juice by pipetting it out carefully without disrupting the cell pellet. Repeat steps 6.4-6.6 until the rotor is full.
7. Carefully position the polytetrafluoroethylene (PTFE) spacer using the spacer screw on top of the rotor.
8. Place the rotor, cap side down in a rotor plunger⁵³ and plunge the rotor into liquid nitrogen. Allow at least 30-60 s for the rotor and the sample to freeze.
9. Start up the heat exchanger unit and cool down the DNP probe to ~90K.
10. Set the MAS unit to Manual mode and set the variable temperature (VT) to 135 l/h and bearing and drive gas flows to ~6 l/h.
11. Engage Eject on the MAS console.
12. Transfer the frozen sample from Step 7.8 from the liquid N2 into the sample catcher and place it in the probe (also see reference⁵³).
13. Manually increase the probe bearing pressure (Bp) to ~500 mbar before engaging the "Insert". This is advisable for stable and safe spinning of the rotor.
14. Increase MAS rate to the desired value using the procedure described in reference⁵³.
15. Check for DNP enhancements with and without microwaves using a standard 1D CP experiments (also see, e.g., reference^51,54) (Figure 7) before proceeding with multidimensional experiments, e.g., 2D CC experiments described in section 3 of this article.

Results

Figure 2 shows representative gels for inclusion body purity (Panel A) and refolding of inclusion bodies (Panel B3). Figure 2 confirms the successful purification of ¹³C,¹⁵N-labeled BamA-P4P5.

Figure 3A shows a typical 2D ¹³C-¹³C spectrum of a well-ordered membrane protein, and Figure 3B shows a typical, high-quality 2D ^15...

Discussion

Membrane proteins are key players in the regulation of vital cellular functions both in prokaryotic and eukaryotic organisms; thus, understanding their action mechanisms at atomic levels of resolution is of vital importance. The existing structural biology techniques have pushed scientific understanding of membrane proteins quite far but have heavily relied on experimental data gathered from in vitro systems devoid of membranes. In this article, an experimental approach is presented that allows to obtain atomistic i...

Materials

Name	Company	Catalog Number	Comments
Ammonium molibdate	Merck	277908
Ammonium-15N Chloride	Cortecnet	CN80P50
Ampicillin	Sigma Aldrich	A9518
AMUpol	Cortecnet	C010P005
Benzonase	EMD Millipore Corp	70746-3
Boric acid	Merck	B6768
bromophenol blue	Sigma	B0126
calcium dichloride	Merck	499609
Choline chloride	Sigma	C-1879
Cobalt chloride	Merck	449776
Copper sulphate	Merck	C1297
D-Biotin	Merck	8512090025
Deuterium Oxide	Cortecnet	CD5251P1000
Dimethyl sulfoxide	Merck	D9170
Ethylenediaminetetraacetic acid	Sigma Aldrich	L6876
Folic acid	Sigma	F-7876
Glucose 13C + 2H	Cortecnet	CCD860P50
Glycerol	Honeywell	G7757
Glycerol (12C3, 99.95% D8, 98%)	Eurisotope	CDLM-8660-PK
glycerol (non-enriched)	Honeywell	G7757-1L
Glycine	Sigma Aldrich	50046
Guanidine hydrochloride	Roth Carl	NR.0037.1
Iron sulphate	Merck	307718
isopropyl β-D-1-thiogalactopyranoside	Thermofisher	R0392
Lysogeny Broth	Merck	L3022
Lysozyme	Sigma Aldrich	L6876
Magnesium chloride - hexahydrate	Fluka	63064
magnesium sulphate	Merck	M5921
monopotassium phosphate	Merck	1051080050
Myoinositol	Sigma	I-5125
n-Dodecyl-B-D-maltoside	Acros Organics	3293702509
N,N-Dimethyldodecylamine N-oxide	Merck	40236
Nicatinamide	Sigma	N-3376
Panthotenic acid	Sigma	21210-25G-F
protease inhibitor	Sigma	P8849
Pyridoxal-HCl	Sigma Aldrich	P9130
Riboflavin	Aldrich	R170-6
Sodium Chloride	Merck	K51107104914
Sodium dihydrogen phospahte - monohydrate	Sigma Aldrich	1,06,34,61,000
Sodium dodecyl sulfate	Thermo-scientific	28365
Sodium hydroxide	Merck	1,06,49,81,000
Sucrose	Sigma Life Science	S9378
Thiamine-HCl	Merck	5871
Tris-HCl	Sigma Aldrich	10,70,89,76,001
Zinc chloride	Merck	208086
E.coli BL21 DE3*	New England Biolabs	C2527
1.5 mL Ultra-tubes	Beckman Coulter	357448
30 kDa centrifugal filter	Amicon	UFC903024
3.2 mm sapphire DNP rotor with caps	Cortecnet	H13861
3.2 mm teflon insert	Cortecnet	B6628
3.2 mm sample packer/unpacker	Cortecnet	B6988
3.2 mm Regular Wall MAS Rotor	Cortecnet	HZ16913
3.2 mm Regular Wall MAS rotor	Cortecnet	HZ09244
Tool Kit for 3.2 mm Thin Wall rotor	Cortecnet	B136904
1.3 mm MAS rotor + caps	Cortecnet	HZ14752
1.3 mm filling tool	Cortecnet	HZ14714
1.3 mm sample packer	Cortecnet	HZ14716
1.3 mm cap remover	Cortecnet	HZ14706
1.3 mm cap set tool	Cortecnet	HZ14744
Dialysis tubing 12-14 kDa	Spectra/Por	132703
Sharpie - Black	Merck	HS15094