Detergent-free isolation, characterization, andfunctional reconstitution of a tetrameric K+ channel:The power of native nanodiscsJonas M. Dörra,1, Martijn C. Koorengevela, Marre Schäfera, Alexander V. Prokofyevb,2, Stefan Scheidelaara,Elwin A. W. van der Cruijsenb, Timothy R. Daffornc, Marc Baldusb, and J. Antoinette Killiana

aMembrane Biochemistry and Biophysics, Bijvoet Center for Biomolecular Research, Utrecht University, 3584 CH Utrecht, The Netherlands; bNMRSpectroscopy, Bijvoet Center for Biomolecular Research, Utrecht University, 3584 CH Utrecht, The Netherlands; and cSchool of Bio Sciences, University ofBirmingham, Edgbaston Birmingham B15 2TT, United Kingdom

Edited by Ramon Latorre, Centro Interdisciplinario de Neurociencias, Universidad de Valparaíso, Valparaíso, Chile, and approved November 21, 2014 (receivedfor review August 22, 2014)

A major obstacle in the study of membrane proteins is their sol-ubilization in a stable and active conformation when using de-tergents. Here, we explored a detergent-free approach to isolatingthe tetrameric potassium channel KcsA directly from the mem-brane of Escherichia coli, using a styrene-maleic acid copolymer.This polymer self-inserts into membranes and is capable of extract-ing membrane patches in the form of nanosize discoidal proteo-lipid particles or “native nanodiscs.” Using circular dichroism andtryptophan fluorescence spectroscopy, we show that the confor-mation of KcsA in native nanodiscs is very similar to that in de-tergent micelles, but that the thermal stability of the protein ishigher in the nanodiscs. Furthermore, as a promising new applica-tion, we show that quantitative analysis of the co-isolated lipids inpurified KcsA-containing nanodiscs allows determination of pref-erential lipid–protein interactions. Thin-layer chromatography ex-periments revealed an enrichment of the anionic lipids cardiolipinand phosphatidylglycerol, indicating their close proximity to thechannel in biological membranes and supporting their functionalrelevance. Finally, we demonstrate that KcsA can be reconstitutedinto planar lipid bilayers directly from native nanodiscs, whichenables functional characterization of the channel by electrophys-iology without first depriving the protein of its native environ-ment. Together, these findings highlight the potential of the useof native nanodiscs as a tool in the study of ion channels, and ofmembrane proteins in general.

membrane–protein solubilization | styrene-maleic acid copolymer |lipid–protein interactions | nanodisc | ion channels

Integral membrane proteins (MPs) are an abundant class ofproteins that play key roles in a wide range of essential cellular

processes (1). To facilitate their study in vitro, detergent mole-cules are commonly used to extract MPs out of their native lipid–bilayer environment (2). However, the use of detergents hassome inherent disadvantages. Most importantly, even though thereare promising developments to improve their properties (3, 4), theinsufficient mimicking of a lipid bilayer by detergent micelles oftenleads to destabilization and rapid loss of function of the in-corporated protein (5). For many functional and structural studies,it is thus necessary to reconstitute the MP into a more stabilizingenvironment; for example, by replacing the detergent with am-phipathic polymers (amphipols) (6) or incorporating the MP intolipid nanodiscs with a surrounding protein scaffold (7). Bothapproaches have proven to be valuable tools for the study ofstructural and functional properties of MPs (8, 9); however,a limitation remains, as transfer of MPs into any of these systemsrequires initial solubilization by detergent.Recently, a detergent-free approach has been described using

amphipathic styrene-maleic acid copolymers (SMAs) as an al-ternative to solubilize MPs directly from biological membranesin the form of nanodiscs, referred to as “Lipodisq” or SMA lipid

particles (10–13) (Fig. 1). The mechanism of action of SMAdiffers fundamentally from that of detergents: instead of dis-rupting the lipid bilayer completely, SMA spontaneously self-inserts and extracts intact membrane patches in the form ofdiscoidal particles that are stabilized by a SMA annulus (14, 15).Because these nanodiscs conserve a spatially delimited nativebiomembrane including MPs, we term them “native nanodiscs.”One of the main advantages of this system is the straightforwardextraction protocol without the need for detergent. It has beenshown that the SMA polymer is capable of directly extractingnative nanodiscs containing large functional protein complexesfrom yeast (12), bacterial proteins involved in cell division (16)and photosynthesis (17), and several members of the ABC trans-porter family (13). The isolation of these proteins from a varietyof different organisms suggests a general applicability of SMAsolubilization for all MPs, irrespective of their expression host ornative organism.To further explore the potential of native nanodiscs, we used the

SMA polymer to isolate an oligomeric bacterial membrane protein:the tetrameric potassium channel from Streptomyces lividans (KcsA)(18), expressed in Escherichia coli. KcsA is an ideal model proteinfor such studies because it is well-characterized and because

Significance

The study of membrane proteins is often hampered by theirtendency to misfold when extracted by detergent. Here, weexplore a detergent-free approach to isolating membrane pro-teins while retaining their native lipid environment, making useof an amphipathic polymer that solubilizes intact membranepatches in the form of nanodiscs. Using a potassium channel asa model protein, we show that these “native nanodiscs” arehighly thermostable particles that are suitable for spectroscopicstudies, allowing structural characterization of the protein in itsnative environment and direct analysis of the lipids in its im-mediate surroundings. We also demonstrate that the channelcan be reconstituted from nanodiscs into planar lipid bilayers forfunctional characterization, thus making native nanodiscs anexcellent alternative to detergent solubilization.

Author contributions: J.M.D., M.C.K., and J.A.K. designed research; J.M.D., M.C.K., M.S.,S.S., and E.A.W.v.d.C. performed research; T.R.D. and M.B. contributed new reagents/analytic tools; J.M.D., M.C.K., M.S., A.V.P., and J.A.K. analyzed data; and J.M.D. and J.A.K.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: j.m.dorr@uu.nl.2Present address: The BIOS Lab on a Chip Group, MESA+ Institute of Nanotechnology,University of Twente, 7500 AE Enschede, The Netherlands.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416205112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1416205112 PNAS | December 30, 2014 | vol. 111 | no. 52 | 18607–18612

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

June

3, 2

020

reconstitution studies have shown that both its function andstability are strongly affected by lipid composition (19–21). Inthis work, we apply SMA to prepare and purify native nanodiscswith KcsA to compare the conformational properties and sta-bility of the protein with those in detergent micelles. In addition,we use native nanodiscs to investigate preferential lipid–proteininteractions by analyzing the composition of small patches of na-tive membrane that are copurified with the protein. Finally, westudy the functional properties of KcsA on reconstitution fromnative nanodiscs into a planar lipid bilayer system. Our resultsunderscore the huge potential of SMA as a membrane-solubilizingagent, as well as the use of native nanodiscs as a membrane-mimetic system for biophysical studies on ion channels, and MPsin general.

ResultsKcsA Can Be Solubilized and Purified in Native Nanodiscs. Prepara-tions of KcsA in native nanodiscs were obtained by the additionof SMA polymer directly to lysed E. coli whole cells with over-produced His-tagged KcsA, followed by isolation of the KcsA-containing nanodiscs by Ni-affinity chromatography. SDS/PAGEanalysis of purified KcsA in native nanodiscs shows a pronouncedband at a molecular weight of ∼60 kDa (Fig. 2A, lane 2), similar tothat observed with tetrameric KcsA purified in a standard pro-tocol, using the nonionic detergent n-dodecyl-β-D-maltoside(DDM) (lane 4) (22). Exposing the samples to 95 °C in the pres-ence of SDS results in a virtually complete transition to the mo-nomeric form of KcsA, as evident from the appearance of a bandat ∼17 kDa (Fig. 2A, lanes 3 and 5). Thus, KcsA can be isolated innative nanodiscs as a stable tetramer that is resistant to SDS atroom temperature, similar to what is described for KcsA in de-tergent (23). In the isolations of KcsA in nanodiscs, we observed anadditional pronounced double band of varying intensity at ∼25 kDathat was heat-stable. Mass spectrometric analysis on a tryptic-digested sample of this band identified it as SlyD, a stress-up-regulated, soluble 20.9-kDa protein endogenous to E. coli thatcontains multiple histidine residues in its C terminus (24). Thiscommon contaminant of Ni-affinity purifications could be re-moved completely by size-exclusion chromatography (Fig. 2B)(25). The SMA polymer itself migrates at a low apparent mo-lecular weight because of its high intrinsic negative charge and isobserved as a blue-stained front on the depicted gels (Fig. 2A,lanes 2 and 3).The size and shape of purified KcsA-containing nanodiscs were

further investigated, using negative-stain transmission electronmicroscopy (Fig. 2C). The particles exhibit a round shape with afairly homogeneous size distribution, with diameters of 10± 2 nm,which is in agreement with previous studies on native nanodiscs(10, 12) and on protein-free nanodiscs obtained by SMA solubi-lization of synthetic liposomes (14, 15).

Native Nanodiscs Surpass Detergent Micelles in Conserving StructuralStability of KcsA. The purity of KcsA nanodiscs after size-exclu-sion chromatography renders them directly suitable for studieswith circular dichroism (CD) and tryptophan fluorescence spec-troscopy. Far-UV CD spectra of KcsA in native nanodiscs andDDM micelles at different temperatures are depicted in Fig. 3A.The shapes of the spectra acquired at room temperature arevirtually indistinguishable and show the features of a predom-inantly α-helical protein with minima at around 208 and 222 nm,in good agreement with previous studies (20, 26) and availableX-ray structures (27). Upon heat exposure, differences betweenthe spectra of KcsA in the different systems become visible. At95 °C, KcsA in DDM shows a loss of∼50% in secondary structure,with a well-defined transition at 72 °C, as measured by the ellip-ticity at 222 nm (Fig. 3B). In contrast, KcsA in nanodiscs lacksa genuine transition as a function of temperature, and the amountof helicity decreases only slightly with increasing temperature,with a loss of ∼20% at 95 °C. In addition, the characteristic fea-tures of α-helical proteins are qualitatively conserved over thewhole temperature range for KcsA in nanodiscs, whereas they arelost in DDM micelles above the transition.Similar observations were made when the intrinsic fluores-

cence of KcsA in different environments was monitored. In-creasing the temperature resulted in a strong decline in intensityof the recorded emission spectra of KcsA that was more pro-nounced in detergent than in nanodiscs (Fig. 3C). At 20 °C, themaximum emission wavelengths on excitation at 295 nm were ∼329and ∼332 nm for KcsA in nanodiscs and micelles, respectively. Theblue-shifted emission maximum in nanodiscs can be explained bythe decreased solvent exposure of some tryptophans resulting frombetter shielding by conserved lipid molecules than is seen in de-tergent micelles. The thermal stability of the tertiary structure ofKcsA was investigated by following the shift in the emission max-imum as a function of temperature. Consistent with the CD

A B

Styrene–Maleic Acid

C C C C

O OO O

H –mn

H HHH

H

Fig. 1. (A) Chemical structure of SMA polymers at neutral pH. For this study,a polymer with an average SMA ratio of n:m = 2:1 was used. (B) Schematicrepresentation of a native nanodisc containing a KcsA tetramer (blue) andnative lipids (green). The outer hydrophobic surface of the lipids is shieldedby SMA (orange).

0

10

20

30

40

50

5 10 15 20

A 280 (m

AU)

Vel (mL)

*

**

* **

KcsA

SlyD

B

50 nm

C

0

10

20

30

40

5 10 15

A 280

(mAU

)

V (mL)

*

**

KcsA

SlyD

B

25015010075

50

37

25

20

NN DDM – + – + heat

T

M

SMA

A

Fig. 2. Purification of KcsA in native nanodiscs. (A) SDS/PAGE analysis ofpurified KcsA in native nanodiscs (NN) and DDM micelles at room temper-ature (−) and after incubation at 95 °C (+). Complete transitions from tet-rameric (T) to monomeric (M) state are visible. (B) Size exclusion chromatogramshowing effective separation of KcsA (*) from the soluble contaminant SlyD(**). (C) Negative stain transmission electron micrograph of native nanodiscswith KcsA after size exclusion chromatography. Round particles of an aver-age size of 10 ± 2 nm are visible. (Scale bar in the enlarged images, 10 nm.)

18608 | www.pnas.org/cgi/doi/10.1073/pnas.1416205112 Dörr et al.

Dow

nloa

ded

by g

uest

on

June

3, 2

020

analysis, KcsA in detergent micelles shows a genuine transition at∼70 °C, whereas in nanodiscs, changes are less pronounced,without a clear transition for KcsA (Fig. 3D). The emission max-imum of micellar KcsA at 95 °C is ∼343 nm, and is thus consid-erably more red-shifted than in nanodiscs (∼337 nm). Virtually thesame results were obtained when the intensity of the fluorescenceat 330 nm was monitored as function of temperature. Thus, thetryptophan residues of detergent-solubilized KcsA show largerconformational changes than those in nanodiscs, where the tertiarystructure is more stable even at high temperatures.The higher stability of KcsA in nanodiscs compared with

DDM micelles is further supported by the observation of a faintmonomer band of KcsA purified in DDM on SDS/PAGE atroom temperature that is absent for nanodiscs (Fig. 2A, lanes 2and 4). Upon storage at 4 °C, we observed an increase of thisdifference over time. Tetrameric KcsA in nanodiscs was stableover months, with a constant tetramer fraction above 95%,whereas the tetramer fraction of DDM-solubilized protein de-creased to ∼60% after 6 weeks. In line with this, we found thatneither freeze thawing nor lyophilization of KcsA in nanodiscsaffects the integrity of the quaternary structure, whereas DDM-solubilized tetramers dissociate to some extent (Fig. S1). Thus,using a complementary set of techniques, it is shown that nativenanodiscs serve as a good tool to conserve the structural stabilityof the oligomeric KcsA.A possible explanation for the lower stability of detergent-

solubilized KcsA is the dynamic equilibrium between monomericand micellar DDM (2), which may promote the dissociation ofthe KcsA tetramer by segregation of monomers into separatemicelles, thus favoring further unfolding. In contrast, nativenanodiscs are stable particles, and no free SMA is needed insolution to maintain their stability. Thus, the more rigid structureof the particles could convey the higher stability of KcsA. How-ever, it should also be noted that the mere presence of lipids canincrease the stability of KcsA, as suggested by studies on recon-stituted protein (20, 22). Conserving and stabilizing a lipid en-vironment around MPs during their extraction by SMA polymers

thus directly yields particles that maintain a higher stability of thetrapped protein than detergent micelles.

Biochemical Analysis of the Local Lipid Environment of KcsA RevealsEnrichment of Anionic Lipids. The isolation of intact nanopatchesof biological membranes in native nanodiscs provides a uniqueopportunity to investigate the immediate local lipid environmentof KcsA, and thus characterize preferential lipid–protein inter-actions. Using TLC, the isolated lipids from native nanodiscs canbe separated according to headgroup. Fig. 4A shows a represen-tative chromatogram with lipid samples from the total cell lysate,the complete SMA-solubilized fraction, and purified KcsA nano-discs. In all samples, three pronounced bands are visible that canbe identified by comparison with synthetic reference lipids asthe zwitterionic phosphatidylethanolamine and the anionic lip-ids phosphatidylglycerol and cardiolipin. The lipid compositionof independent bacterial cultures was found to be very sensitiveto the growth conditions and time, resulting in a variation ofabsolute values in the range of 5–10 mol%. For comparison,data from each independent nanodisc preparation were there-fore normalized to the corresponding total cell lysate. Quanti-tative analysis of the intensity of the chromatogram bands (Fig.4B) shows that the lipid compositions of the total cell lysate andthe soluble fraction are very similar and in good agreement withother studies on lipids of E. coli K 12 wild-type strains (28, 29).Thus, SMA does not preferentially solubilize any specific lipidsin the E. coli membrane. Strikingly, KcsA nanodiscs show higheramounts of anionic lipids, with increases of 36% in phosphati-dylglycerol and 61% in cardiolipin content compared with thetotal extract. This isolation of enriched lipid species can hencebe attributed to the preferential interaction of anionic lipidswith KcsA.

-15

-10

-5

0

5

200 210 220 230 240 250(nm)

θ(m

deg)

λ

Nanodisc 20 °C DDM 20 °CNanodisc 95 °C DDM 95 °C

0

0.2

0.4

0.6

0.8

1

300 325 350 375 400

I nor

m(a

. u.)

(nm)emλ

Nanodisc 20°CNanodisc 95°CDDM 20°CDDM 95°C

-15

-10

-5

20 45 70 95

θ 222

nm(m

deg)

DDMNanodisc

D

C

B

A

327

333

339

345

20 45 70 95

em, m

ax(n

m)

λ

T (°C)

DDMNanodisc

T (°C)

Fig. 3. Comparison of the thermal stability of KcsA in different environ-ments. (A) Circular dichroism spectra of KcsA in native nanodiscs and DDMmicelles at 20 °C and 95 °C. Data are offset corrected averages of 8–10 scans.(B) Corresponding thermal unfolding traces monitored by the ellipticity at222 nm. Data are averages of two experiments with errors too small to bedepicted. Solid lines are depicted to guide the eye. (C) Normalized fluores-cence intensity of the intrinsic tryptophans of KcsA at 20 °C and 95 °C. Dataare averages of three scans normalized to the intensity at 330 nm of therespective spectra at 20 °C. (D) Corresponding thermal unfolding traces mon-itored by the wavelength of maximum fluorescence emission. Solid lines aredepicted to guide the eye.

CL

PEPG

Origin

1 2 3A B

C

40

0

20

60

80

CL PE PG

Rela

�ve

amou

nt (m

ol%

)

KcsA

TotalSoluble

15

0

30

45

16:112:0 14:0 16:0 17:c 18:0 18:1 19:c

Rela

�ve

amou

nt (w

eigh

t%)

Fa�y acid

KcsA

TotalSoluble

Fig. 4. Analysis of coisolated lipids in native nanodiscs. (A) TLC on extractedlipids from lysed whole bacterial cells after addition of SMA (1), completesoluble fraction separated by ultracentrifugation (2), and native nanodiscswith KcsA purified by Ni-affinity chromatography (3). (B) Molar compositionof the extracted lipid species according to headgroup of total lysate (or-ange), soluble fraction (blue), and KcsA nanodiscs (green). Data are shown asaverage mol percentages with SDs (red error bars) of three independentnanodisc isolations. Data for the soluble fraction and KcsA nanodiscs werenormalized to the corresponding total lysate. Black error bars denote SDs ofthe change compared with total lysate. (C) Gas chromatography analysis ofthe fatty acid composition of all isolated lipids. Depicted data are averagesof two independent experiments. Errors are given as in B.

Dörr et al. PNAS | December 30, 2014 | vol. 111 | no. 52 | 18609

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

June

3, 2

020

Preferential lipid–protein interactions were further analyzedby investigating the composition of the acyl chains of all isolatedlipids (Fig. 4C). Lipids from all samples predominantly containedpalmitic (16:0) and cis-vaccenic acid (18:1) chains with a combinedweight percentage of ∼70%, in agreement with other studies (28).In our analysis, no strong enrichment of specific acyl chains couldbe detected when comparing the lipids that were copurified withKcsA with the total extract or with the SMA-solubilized fraction.Only the amount of short lauric (12:0) and myristic acid (14:0)chains appeared to be slightly decreased in purified KcsA nano-discs in favor of a slightly higher cis-vaccenic acid content. Theseresults suggest a minor preference of KcsA for lipids with longerchains, possibly to promote hydrophobic matching (30, 31).

Reconstitution of KcsA into Planar Lipid Bilayers Enables FunctionalCharacterization. A final challenge remains in assessing thefunctionality of KcsA in native nanodiscs. Because analysis ofion transport is not possible in nanodiscs, we attempted to re-constitute the channel into a compartment-forming bilayer sys-tem that enables investigations of protein-facilitated potassiumconductivity across membranes. To this end, we used a planarlipid bilayer setup that has been used successfully for functionalstudies on KcsA delivered in protein-stabilized nanodiscs (32).Seconds to minutes after the addition of native nanodiscs withKcsA, single-channel conductivity was observed as transientdiscrete current changes with an amplitude of 10–15 pA (Fig.5), implying spontaneous fusion events of single nanodiscs withthe bilayer. The recorded traces show both high and low openprobability states that are characteristic for KcsA monitoredunder similar conditions by single-channel recordings of KcsAfused from liposomes (33) or incorporated via a cell-free ex-pression protocol (34). The results are also comparable withthose obtained by patch clamp studies of giant unilamellarvesicles with KcsA reconstituted from detergent (35). Together,this suggests identical functional properties of KcsA recon-stituted from native nanodiscs compared with standard recon-stitution protocols. Upon addition of the K+-channel blockertetraethylammonium, we observed a reversible complete de-pletion of all opening events (Fig. S2), further confirming thepresence of reconstituted KcsA in the bilayer. Native nanodiscsthus may serve as a powerful alternative to conventional ap-proaches for the functional reconstitution of ion channels.

DiscussionWe have shown that native nanodiscs with embedded KcsA areformed spontaneously from bacterial cells upon addition of SMApolymer and that, using Ni-affinity and size-exclusion chro-matography, sufficient protein can be purified in nanodiscs forextensive biophysical characterization. The isolation of KcsA inthis manner is less cumbersome than standard detergent proto-cols and has the additional advantage of conserving the nativeconformation of the protein in a stabilizing environment com-prising native lipids. Our data suggest that native nanodiscs ingeneral convey a higher stability of the incorporated protein thandetergent micelles, in agreement with previous findings on thethermostability of an ABC transporter (13) and photosyntheticreaction centers (17). In addition, as shown in this study, thecoisolation of MPs with a patch of biological membrane allowsfor the analysis of preferential interactions and further facilitatesstructural and functional studies on purified protein in a (near)native state, as discussed next.Insights into preferential lipid–protein interactions are im-

portant to understand structural and functional properties ofMPs in a membrane environment. However, detailed infor-mation is not easily obtained with standard methods. Here, weexploited the extraction of intact nanopatches of biologicalmembranes by the SMA polymer to directly identify preferentiallipid–protein interactions that allow approximations of the situ-ation in vivo. Analysis of the lipid composition revealed an en-richment of anionic lipids in close proximity to KcsA. Thisfinding is unlikely to be an artifact of the extraction protocolbecause we found that none of the three main E. coli lipid spe-cies is preferentially incorporated into nanodiscs. This is in-triguing, as the SMA polymer has a high negative charge densityat pH 8.0 because of its many carboxylic acid moieties, whichcould be expected to lead to electrostatic repulsion of anioniclipids. The absence of a measurable effect of this repulsion henceemphasizes the promiscuity of the polymer with respect to lipidspecies in biological membranes. Similarly, SMA did not pref-erentially solubilize lipids containing specific fatty acids. Thisindicates that lipid solubilization by SMA is applicable to allphospholipids in any membrane, irrespective of headgroup oracyl chains, as further supported by the successful extraction ofall major phospholipids of the mitochondrial membrane inyeast (12), as well as the membrane of the photosyntheticbacterium Rhodobacter sphaeroides (17).The importance of anionic lipids for KcsA functionality and

their close association with the channel has been well establishedby a wealth of studies using fluorescence quenching by bromi-nated lipids (33), mass spectrometry (36), electrophysiology (33,37, 38), and molecular dynamic simulations (39, 40). In addition,the diacylglycerol fragment found in crystal structures of KcsAwas attributed to phosphatidylglycerol (19), and it has beenshown that anionic lipids strongly stabilize the KcsA tetrameragainst dissociation by both heat (20, 22) and small fluorinatedalcohols (21). In these studies, KcsA was purified in detergent,and preferential interactions were either deduced from the co-purification of single tightly bound lipids or were observed in-directly after reconstitution into synthetic lipid bilayer systems.In contrast, the method described in this work facilitates directbiochemical analysis of preferential lipid–protein interactionsthat involve the native annular lipid environment. This paves theway for a new application to study specific lipid–protein inter-actions by using the SMA polymer to isolate nanodiscs withprotein that has been conventionally reconstituted into bilayersof well-defined composition. By systematically varying the com-position of these bilayers, one can then obtain very detailed in-formation on the specificity of KcsA or other proteins for bothlipid headgroups and acyl chains. Aside from assessing lipid–protein interactions, native nanodiscs offer the important

A B

Even

ts x

103

Current (pA)

2

0

0.5

1.5

1

100-5 5 2015

Even

ts x

103

Current (pA)15

0.5

0

1

1.5

5

2

-5 0 10 20

10 pA1 s

10 pA1 s

5 pA0.2 s

5 pA0.2 s

Fig. 5. Functional characterization of KcsA reconstituted from nativenanodiscs into a planar lipid bilayer of E. coli polar lipid extract. (A) Typicalsingle-channel current traces on addition of KcsA nanodiscs to a setup witha symmetric 150-mM KCl solution at +100 mV. The top part representsa channel in the high open probability mode that is characterized by longopening dwell times, as shown in the enlarged section indicated by a box ofdashed lines. Channels in the low open probability mode exhibit a distinctlydifferent pattern, with short opening times resulting in sequences of bursts(Bottom). (B) All-point histograms for open/closed distribution of channelswith high (Top) and low (Bottom) open probability calculated from theenlarged 2-s single-channel current traces from A.

18610 | www.pnas.org/cgi/doi/10.1073/pnas.1416205112 Dörr et al.

Dow

nloa

ded

by g

uest

on

June

3, 2

020

additional advantage of copurifying proteins or other moleculesthat weakly interact with the target MP, providing unique in-sights into the composition of its immediate membrane envi-ronment. This feature has been used very recently for the isolation ofa detergent-labile complex of penicillin-binding proteins in Staphy-lococcus aureus (16). Native nanodiscs thus are a powerful tool tostudy preferential interactions of MPs both in vitro and in vivo.In addition to purification of MPs and analysis of their pref-

erential interactions, native nanodiscs also allow structural andfunctional characterization. Here, we used CD and fluorescencespectroscopy to study the thermostability of KcsA in a straight-forward way. Moreover, their small size makes native nanodiscs,in principle, suitable for characterization by the full range ofbiophysical techniques, including NMR spectroscopy, that havebeen successfully applied to MPs incorporated into similar lipidnanodiscs that are stabilized by scaffold proteins instead of SMA(9). Native nanodiscs are also promising tools for studies on MPsin the rapidly developing field of single-particle cryo-electronmicroscopy. The use of amphipols to stabilize MPs has recentlyled to a high-resolution de novo structure determination of acation channel at a resolution of 3.4 Å (41), breaking the side-chain resolution barrier. Together with promising studies usingscaffold-protein nanodiscs (42, 43), as well as initial studies withnative nanodiscs (44), these new developments suggest thatnative nanodiscs may become a powerful alternative to enablehigh-resolution structural characterization of MPs in their na-tive environment.Several proteins have so far been functionally characterized in

native nanodiscs by investigating their ligand-binding (13) andoptical (10–12, 17) properties, as well as their enzymatic activity(10, 12). Our data demonstrate that native nanodiscs can be alsoused to directly reconstitute KcsA into planar lipid bilayers,allowing electrophysiologic characterization of single channels inmembranes with well-defined composition. To the best of ourknowledge, this represents the first evidence of purification andtransfer of an MP from the membrane of living cells to syntheticlipid bilayers without being deprived of its native lipid environ-ment during any stage in the procedure. The ability of nativenanodiscs to fuse with bilayers can be considered a promisingfirst step toward enabling crystallization trials of MPs in a near-native environment. Other prospects of successful reconstitutioninclude a plethora of assays available for MPs in liposomes andsupported bilayers in applications that require a well-definedlipid environment. With the inherent advantage of conservingthe native lipid environment of MPs, native nanodiscs consti-tute a highly promising and convenient alternative to detergentsolubilization and may lead the way toward an in situ approachfor structural and functional studies on MPs, including phar-macologic assays.

Materials and MethodsMaterials and SMA Preparation. All chemicals and enzymes were purchasedfrom Sigma-Aldrich unless otherwise indicated. DDM was from Affymetrix,and all reference lipids used were from Avanti Polar Lipids. The used polymerwas SMA2000, a styrene-maleic anhydride copolymer with a molar styrenemaleic anhydride ratio of 2:1, a weight average molecular weight of 7.5 kDa,and a molar mass dispersity (polydispersity index) of 2.5 (Cray Valley). Con-version into SMA was achieved by hydrolysis in 1 M KOH and reflux for 2 hwhile heating the suspension to 100 °C. Subsequently, the polymer wasprecipitated by the addition of HCl and washed 6 times with 100 mM HCl toremove K+ ions. The sample was lyophilized, and hydrolysis was confirmedby Fourier transform infrared spectroscopy (17). SMA stock solutions wereprepared by dissolving 6% (wt/vol) SMA powder in 50 mM unadjusted Trisbuffer with gradual addition of NaOH solution until the pH reached theneutral range. The solution was then stored at −20 °C and adjusted to pH 8.0and to the desired volume after thawing.

Gene Expression and Protein Purification. KcsA was produced as describedearlier (22). Cell pellets were resuspended in 50 mM Tris buffer at pH 8.0

containing 600 mM NaCl and 30 mM KCl (10 mL per 2 g wet cell weight) andincubated with 10 μg/mL lysozyme and 5 μg/mL DNase for 45 min on ice. Thesuspension was aliquoted and mixed 1:1 with 6% (wt/vol) SMA in 50 mM Trisat pH 8 for the preparation of native nanodiscs or 1:1 with 50 mM Tris bufferat pH 8 for detergent solubilization, respectively. The cell suspension withSMA was pushed through a disruptor at a pressure of 215 MPa and in-cubated overnight with gentle agitation at room temperature, followed by45 min of centrifugation at 100,000 × g. SDS/PAGE analysis of supernatantand pellet after centrifugation revealed that 70–80% of the total amount oftetrameric KcsA was solubilized by SMA. The supernatant was then in-cubated for at least 4 h at 4 °C with 2.5 mL HisPur Ni-NTA agarose beads(Thermo Scientific) per liter of culture medium. The beads were transferredto a gravity-flow column and washed three times each with buffers con-taining 10 and 50 mM imidazole. Protein in native nanodiscs was then elutedwith buffer containing 300 mM imidazole, yielding 1–2 mg protein per literof culture medium on average, estimated by SDS/PAGE analysis and densi-tometric comparison of the protein bands after Coomassie staining witha gradient of BSA of known concentration, using the Quantity One software(Biorad). For biophysical studies, the pooled eluted fractions were loaded ona Superdex 200 10/300 GL size exclusion column (GE Healthcare) connectedto an Äkta Prime Plus chromatography system (GE Healthcare) to separatethe protein-containing nanodiscs from contaminating soluble proteins and freepolymer. Thereby, the buffer was exchanged to 10 mM Tris at pH 8, 100 mMNaCl, and 5 mM KCl, which was used in all subsequent experiments.

Detergent purification was performed on aliquots of lysed cells, as de-scribed earlier (22), with the difference of using Tris buffer at pH 8.0 insteadof Hepes. Eluted protein was extensively dialyzed against 1 mM DDM in thesame buffer that was used for nanodiscs, using a dialysis membrane with a50-kDa molecular weight cutoff. Protein concentration was determined bythe absorption at 280 nm, using a calculated extinction coefficient of 34,950M−1 · cm−1 (45). For protein in native nanodiscs, this can only be consideredan estimate, as the phenyl groups of SMA contribute to the absorption inthis wavelength range.

Spectroscopy. Far-UV CD experiments were performed on a J-810 spec-tropolarimeter (Jasco)with a Peltier thermo control element (Jasco), using Teflon-sealed, polarimetrically checked quartz glass cuvettes with an optical pathlengthof 1 mm and a volume of 350 μL (Hellma Analytics). Experimental parametersincluded a wavelength increment of 1 nm, a scan speed of 20 nm/min, a re-sponse time of 4 s, and a KcsA concentration of 0.06 mg/mL, as determined forDDM-solubilized protein. KcsA in native nanodiscs was diluted such that sampleshad the same signal intensity as protein in DDM, correcting for inaccuracies inconcentration determination. Samples were allowed to equilibrate for 30 minat the desired temperature before measurement. All resulting spectra arebuffer- and offset-corrected averages of 8–10 scans in the range of 200–250 nm.

Tryptophan fluorescence measurements were performed on a Cary Eclipsespectrofluorometer (Varian) equipped with a thermo control unit (Varian) insealed quartz glass cuvettes with optical path lengths of 4 × 10 mm and aninner volume of 1.4 mL (Hellma). The excitation light beam had a wave-length of 295 nm at a slit size of 5 nm, and the fluorescence emission wasrecorded in the range of 300–400 nm, using a slit size of 5 nm, a wavelengthincrement of 1 nm, a scan speed of 60 nm/min, and an integration time of1 s. The KcsA concentration was 0.01 mg/mL. All samples were allowed toequilibrate for 30 min at the desired temperature. Spectra were recorded intriplicate and corrected for buffer contribution before analysis by nonlinearleast squares fitting to a bimodal log-normal distribution, using the Exceladd-in Solver (Frontline Systems) (46).

Lipid Analysis. Before lipid isolation, KcsA-containing native nanodiscs thatwere eluted from Ni-NTA beads were washed with buffer on spin columnswith a molecular weight cutoff of 30 kDa to remove imidazole. Lipids werethen extracted according to a modified version of the method of Bligh andDyer (47) and analyzed by quantitative TLC, as well as gas chromatography(for details, see SI Materials and Methods).

Electrophysiology. Single-channel recordings of KcsA were performed on aCompact setup for planar lipid bilayer electrophysiology (Ionovation) con-nected to an EPC 10 amplifier (HEKA). Lipid bilayer formationwas achieved bypainting E. coli polar lipid extract dissolved in n-decane (50 mg/mL) overa 200-μm hole in a Teflon-septum separating two compartments. Bothcompartments contained 150 mM KCl solution that was buffered with10 mM Hepes to pH 7.0 in the cis compartment and 10 mM succinic acid topH 4.0 in the trans compartment, respectively. After channel insertion, theconductivity at a constant voltage of +100 mV was recorded for several

Dörr et al. PNAS | December 30, 2014 | vol. 111 | no. 52 | 18611

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

June

3, 2

020

minutes, using the PatchMaster software (HEKA). Data were sampled at 10 kHzand digitally filtered at 1 kHz. All measurements were performed at 22 °C.

For a more detailed description of the method used for incorporation ofKcsA channels into planar bilayers as well as electron microscopy, see SIMaterials and Methods.

ACKNOWLEDGMENTS. We thank Mohammed Jamshad and RosemaryParslow for help with the initial trials of solubilization of KcsA in nativenanodiscs. We are indebted to Mirjam Damen for performing the massspectrometric analysis on the SlyD protein and Hans Meeldijk for help with

the transmission electron microscopy measurements. We further thank RuudCox for help with the gas chromatography analysis, Hugo van den Hoek forassistance with size-exclusion chromatography, and Cray Valley for theirkind gift of SMA2000 polymer. Financial support received from the seventhframework program of the European Union (Initial Training Network “Mani-Fold,” Grant 317371 to J.M.D.) and from the Netherlands Organisation forScientific Research (Grants 700.11.334 and 700.58.102 to E.A.W.v.d.C. andM.B.), as well as via the research program of the Foundation for Funda-mental Research on Matter (to S.S.), is gratefully acknowledged. T.R.D.acknowledges support from the Biotechnology and Biological Sciences Re-search Council (Grants BB/J017310/1, BB/I020349/1 and BB/G010412/1).

1. von Heijne G (2007) The membrane protein universe: What’s out there and whybother? J Intern Med 261(6):543–557.

2. Garavito RM, Ferguson-Miller S (2001) Detergents as tools in membrane biochemistry.J Biol Chem 276(35):32403–32406.

3. McQuade DT, et al. (2000) Rigid amphiphiles for membrane protein manipulation.Angew Chem Int Ed Engl 39(4):758–761.

4. Chae PS, et al. (2010) Maltose-neopentyl glycol (MNG) amphiphiles for solubilization,stabilization and crystallization of membrane proteins. Nat Methods 7(12):1003–1008.

5. Bowie JU (2001) Stabilizing membrane proteins. Curr Opin Struct Biol 11(4):397–402.6. Tribet C, Audebert R, Popot JL (1996) Amphipols: Polymers that keep membrane

proteins soluble in aqueous solutions. Proc Natl Acad Sci USA 93(26):15047–15050.7. Civjan NR, Bayburt TH, Schuler MA, Sligar SG (2003) Direct solubilization of heterol-

ogously expressed membrane proteins by incorporation into nanoscale lipid bilayers.Biotechniques 35(3):556–560, 562–563.

8. Popot JL (2010) Amphipols, nanodiscs, and fluorinated surfactants: Three non-conventional approaches to studying membrane proteins in aqueous solutions. AnnuRev Biochem 79:737–775.

9. Bayburt TH, Sligar SG (2010) Membrane protein assembly into Nanodiscs. FEBS Lett584(9):1721–1727.

10. Knowles TJ, et al. (2009) Membrane proteins solubilized intact in lipid containingnanoparticles bounded by styrene maleic acid copolymer. J Am Chem Soc 131(22):7484–7485.

11. Orwick-Rydmark M, et al. (2012) Detergent-free incorporation of a seven-trans-membrane receptor protein into nanosized bilayer Lipodisq particles for functionaland biophysical studies. Nano Lett 12(9):4687–4692.

12. Long AR, et al. (2013) A detergent-free strategy for the reconstitution of active en-zyme complexes from native biological membranes into nanoscale discs. BMC Bio-technol 13:41.

13. Gulati S, et al. (2014) Detergent-free purification of ABC (ATP-binding-cassette)transporters. Biochem J 461(2):269–278.

14. Orwick MC, et al. (2012) Detergent-free formation and physicochemical character-ization of nanosized lipid-polymer complexes: Lipodisq. Angew Chem Int Ed Engl51(19):4653–4657.

15. Jamshad M, et al. (2014) Structural analysis of a nanoparticle containing a lipid bilayerused for detergent-free extraction of membrane proteins. NanoResearch, in press.

16. Paulin S, et al. (2014) Surfactant-free purification of membrane protein complexesfrom bacteria: Application to the staphylococcal penicillin-binding protein complexPBP2/PBP2a. Nanotechnology 25(28):285101.

17. Swainsbury DJK, Scheidelaar S, van Grondelle R, Killian JA, Jones MR (2014) Bacterialreaction centers purified with styrene maleic Acid copolymer retain native membranefunctional properties and display enhanced stability. Angew Chem Int Ed Engl 53(44):11803–11807.

18. Schrempf H, et al. (1995) A prokaryotic potassium ion channel with two predictedtransmembrane segments from Streptomyces lividans. EMBO J 14(21):5170–5178.

19. Valiyaveetil FI, Zhou Y, MacKinnon R (2002) Lipids in the structure, folding, andfunction of the KcsA K+ channel. Biochemistry 41(35):10771–10777.

20. Triano I, et al. (2010) Occupancy of nonannular lipid binding sites on KcsA greatlyincreases the stability of the tetrameric protein. Biochemistry 49(25):5397–5404.

21. Raja M, Spelbrink REJ, de Kruijff B, Killian JA (2007) Phosphatidic acid plays a specialrole in stabilizing and folding of the tetrameric potassium channel KcsA. FEBS Lett581(29):5715–5722.

22. van Dalen A, Hegger S, Killian JA, de Kruijff B (2002) Influence of lipids on membraneassembly and stability of the potassium channel KcsA. FEBS Lett 525(1-3):33–38.

23. Cortes DM, Perozo E (1997) Structural dynamics of the Streptomyces lividans K+ channel(SKC1): Oligomeric stoichiometry and stability. Biochemistry 36(33):10343–10352.

24. Wülfing C, Lombardero J, Plückthun A (1994) An Escherichia coli protein consisting ofa domain homologous to FK506-binding proteins (FKBP) and a new metal bindingmotif. J Biol Chem 269(4):2895–2901.

25. Parsy CB, Chapman CJ, Barnes AC, Robertson JF, Murray A (2007) Two-step method toisolate target recombinant protein from co-purified bacterial contaminant SlyD afterimmobilised metal affinity chromatography. J Chromatogr B Analyt Technol BiomedLife Sci 853(1-2):314–319.

26. van den Brink-van der Laan E, Chupin V, Killian JA, de Kruijff B (2004) Stability of KcsAtetramer depends on membrane lateral pressure. Biochemistry 43(14):4240–4250.

27. Uysal S, et al. (2009) Crystal structure of full-length KcsA in its closed conformation.Proc Natl Acad Sci USA 106(16):6644–6649.

28. Raetz CR (1978) Enzymology, genetics, and regulation of membrane phospholipidsynthesis in Escherichia coli. Microbiol Rev 42(3):614–659.

29. Morein S, Henricson D, Rilfors L (1994) Separation of inner and outer membrane vesiclesfrom Escherichia coli in self-generating Percoll gradients. Anal Biochem 216(1):47–51.

30. Williamson IM, Alvis SJ, East JM, Lee AG (2002) Interactions of phospholipids with thepotassium channel KcsA. Biophys J 83(4):2026–2038.

31. Killian JA (1998) Hydrophobic mismatch between proteins and lipids in membranes.Biochim Biophys Acta 1376(3):401–415.

32. Banerjee S, Nimigean CM (2011) Non-vesicular transfer of membrane proteins fromnanoparticles to lipid bilayers. J Gen Physiol 137(2):217–223.

33. Marius P, et al. (2008) Binding of anionic lipids to at least three nonannular sites on thepotassium channel KcsA is required for channel opening. Biophys J 94(5):1689–1698.

34. Friddin MS, et al. (2013) Single-channel electrophysiology of cell-free expressed ionchannels by direct incorporation in lipid bilayers. Analyst (Lond) 138(24):7294–7298.

35. Chakrapani S, Cordero-Morales JF, Perozo E (2007) A quantitative description of KcsAgating II: Single-channel currents. J Gen Physiol 130(5):479–496.

36. Demmers JAA, van Dalen A, de Kruijff B, Heck AJR, Killian JA (2003) Interaction of theK+ channel KcsA with membrane phospholipids as studied by ESI mass spectrometry.FEBS Lett 541(1-3):28–32.

37. Cuello LG, Romero JG, Cortes DM, Perozo E (1998) pH-dependent gating in theStreptomyces lividans K+ channel. Biochemistry 37(10):3229–3236.

38. Iwamoto M, Oiki S (2013) Amphipathic antenna of an inward rectifier K+ channelresponds to changes in the inner membrane leaflet. Proc Natl Acad Sci USA 110(2):749–754.

39. Deol SS, Domene C, Bond PJ, Sansom MSP (2006) Anionic phospholipid interactionswith the potassium channel KcsA: Simulation studies. Biophys J 90(3):822–830.

40. Weingarth M, et al. (2013) Structural determinants of specific lipid binding to po-tassium channels. J Am Chem Soc 135(10):3983–3988.

41. Liao M, Cao E, Julius D, Cheng Y (2013) Structure of the TRPV1 ion channel de-termined by electron cryo-microscopy. Nature 504(7478):107–112.

42. Gogol EP, et al. (2013) Three dimensional structure of the anthrax toxin translocon-lethal factor complex by cryo-electron microscopy. Protein Sci 22(5):586–594.

43. McCoy JG, et al. (2014) A KcsA/MloK1 chimeric ion channel has lipid-dependentligand-binding energetics. J Biol Chem 289(14):9535–9546.

44. Postis V, et al. (2014) The use of smalps as a novel membrane protein scaffold forstructure study by negative stain electron microscopy. Biochim Biophys Acta – Bio-membranes, in press.

45. Gill SC, von Hippel PH (1989) Calculation of protein extinction coefficients from aminoacid sequence data. Anal Biochem 182(2):319–326.

46. Kemmer G, Keller S (2010) Nonlinear least-squares data fitting in Excel spreadsheets.Nat Protoc 5(2):267–281.

47. Bligh EG, Dyer WJ (1959) A rapid method of total lipid extraction and purification.Can J Biochem Physiol 37(8):911–917.

18612 | www.pnas.org/cgi/doi/10.1073/pnas.1416205112 Dörr et al.

Dow

nloa

ded

by g

uest

on

June

3, 2

020