Multilayered lipid membrane stacks for biocatalysis using membrane enzymes [original]

full paper

wileyonlinelibrary.com

1606265 (1 of 9)

and chloroplasts. In addition to greater

control over compartmentalization, the

extra layering of membranes gives the

organisms increased area for membrane

proteins. The inner membranes of mito-

chondria, for example, are folded into

cristae to greatly increase the surface area

available for the electron transport chain

proteins, packing high concentrations into

a small volume.[1] Similarly, chloroplasts

employ the stacking of interconnected

thylakoid membranes to greatly enhance

the concentrations of photosynthetic pro-

tein complexes required to efficiently

harness solar energy.[2] These structures

and many others show how the lamellar

stacking of membranes and their proteins

hold substantial technological potential

for biomimicry, where the stacking of

membranes may allow the design of novel

protein arrangements with possible appli-

cations in catalysis, photonics, sensing,

and the 3D crystallization of membrane

proteins.[3,4]

Supported lipid bilayers (SLBs) have been widely advocated

for biotechnological applications and frequently used as model

cell membranes in fundamental studies. SLBs provide a pow-

erful means to study the function of membrane proteins and

protein–protein interaction in membranes.[5–7] The planar

nature of SLBs allows investigations by an array of powerful

surface analytical techniques such as atomic force microscopy

(AFM), quartz crystal microbalance, surface plasmon resonance,

and total internal reflection fluorescence microscopy. The poten-

tial of SLBs to mimic complex multilayer membrane assemblies

has been reported,[8] yet the development of consistent meth-

odologies has been limited, and only few model systems are

described that investigate protein behavior in multiple mem-

branes. For instance, multilamellar membranes are routinely

used in studies of the structure of lipid membranes and mem-

brane-associated polypeptides by X-ray or neutron diffraction, or

by solid-state nuclear magnetic resonance spectroscopy.[9] Meth-

odologies used to form these multilayered membranes require

the drying of membranes in organic solvents, which is suitable

for lipid and peptide studies; however, during drying, mem-

brane proteins are prone to lose their biological activity.[10]

We have previously reported on a simple layer-by-layer

(LBL) methodology to form lipid multilayers via vesicle rup-

ture onto existing SLBs using poly-l-lysine (PLL) as an elec-

trostatic polymer “glue.”[11] Not only does this technique allow

Multilayered Lipid Membrane Stacks for Biocatalysis Using

Membrane Enzymes

George R. Heath, Mengqiu Li, Honling Rong, Valentin Radu, Stefan Frielingsdorf,

Oliver Lenz, Julea N. Butt, and Lars J. C. Jeuken*

Multilayered or stacked lipid membranes are a common principle in biology and

have various functional advantages compared to single-lipid membranes, such

as their ability to spatially organize processes, compartmentalize mole cules, and

greatly increase surface area and hence membrane protein concentration. Here,

a supramolecular assembly of a multilayered lipid membrane system is reported

in which poly-l-lysine electrostatically links negatively charged lipid membranes.

When suitable membrane enzymes are incorporated, either an ubiquinol oxi-

dase (cytochrome bo3 from Escherichia coli) or an oxygen tolerant hydrogenase

(the membrane-bound hydrogenase from Ralstonia eutropha), cyclic voltam-

metry (CV) reveals a linear increase in bio catalytic activity with each additional

membrane layer. Electron transfer between the enzymes and the electrode is

mediated by the quinone pool that is present in the lipid phase. Using atomic

force microscopy, CV, and fluorescence micro scopy it is deduced that quinones

are able to diffuse between the stacked lipid membrane layers via defect sites

where the lipid membranes are inter connected. This assembly is akin to that

of interconnected thylakoid membranes or the folded lamella of mitochondria

and has significant potential for mimicry in biotechnology applications such as

energy production or biosensing.

Dr. G. R. Heath, Dr. M. Li, H. Rong, Dr. V. Radu,

Dr. L. J. C. Jeuken

School of Biomedical Sciences

University of Leeds

Leeds LS2 9JT, UK

E-mail: [email protected]

Dr. S. Frielingsdorf, Dr. O. Lenz

Institut für Chemie, Sekretariat PC14

Technische Universität Berlin

Straße des 17. Juni 135, 10623 Berlin, Germany

Prof. J. N. Butt

Centre for Molecular and Structural Biochemistry

School of Biological Sciences and School of Chemistry

University of East Anglia

Norwich NR4 7TJ, UK

DOI: 10.1002/adfm.201606265

1. Introduction

Double or multilayered membranes, along with their asso-

ciated membrane proteins, are an integral part of the energy

producing pathways of eukaryotic cells, gram-negative bacteria,

The copyright line of this paper was changed 5 April 2017 after initial

publication.

This is an open access article under the terms of the Creative Commons

Attribution License, which permits use, distribution and reproduction in

any medium, provided the original work is properly cited.

www.afm-journal.de

Adv. Funct. Mater. 2017, 27, 1606265

www.advancedsciencenews.com

full paper

incorporation of membrane proteins, but it can also create

membrane stacks with different membrane proteins in each

layer. Alternating membrane protein stacks are found in, for

example, the electrocyte cells of electric eels in which the trans-

membrane proteins are asymmetrically distributed across two

primary membranes that, when added in series, can generate

potentials of ≈600 V.[12,13]

Here, we engineered a lipid multilayer matrix in which

native electron mediators can freely diffuse for bio-electro-

catalytic applications. Mimicking the function of cristae and

thylakoid stacks in mitochondria and chloroplasts, respec-

tively, the assembly increases the concentration of membrane

enzymes per electrode surface area in a stepwise manner. We

created multilayers of membrane enzymes in a native-like

lipid environment using the LBL assembly of bacterial mem-

brane extracts at gold electrodes. Assemblies with two different

membrane proteins are demonstrated: a membrane-bound

hydrogenase (MBH) from Ralstonia eutropha, which oxidizes

H2, and a ubiquinol oxidase, cytochrome bo3, from Escherichia

coli that reduces oxygen. With each deposition of a membrane

layer, we see the catalytic activity increasing as the total amount

of enzyme on the surface increases. This biomimetic system

demonstrates how the stacking of membranes can proportion-

ally increase the concentration of active membrane proteins at

surfaces.

2. Results and Discussion

2.1. Interconnected Lipid Multilayers

Initial structural characterization of the membrane stacks was

performed on glass and mica using fluorescence microscopy

and AFM, respectively. Quinone diffusion between the mem-

brane layers and biocatalytic activity of membrane enzymes

incorporated into the stack was then characterized using

electrochemistry, for which the stacks were formed on gold

surfaces. The formation of membrane stacks is schematically

shown in Figure 1.

Figure 2A displays fluorescence images of a single solid

supported bilayer membrane (doped with a fluorescent lipid

analog) on which a layer of PLL was adsorbed. Before PLL, a

homogenous membrane is present as evidenced by a uniform

fluorescence intensity across the field of view. After addition

of PLL, discrete regions with twofold fluorescent intensity are

evident (Figure 2A,B). These patches are also seen by AFM

(Figure 2D), and the height and fluorescence intensity are con-

sistent with the formation of two lipid bilayer membranes on

top of each other. These patches appear before adding any addi-

tional vesicles. Furthermore, before addition of PLL, no vesicles

were detected on top of the base lipid bilayer (Figure 2A). This

suggests that the lipid required to form the double membrane

patches is extracted from the base bilayer. To test if the second

“top” membrane patches are continuous with the first mem-

brane beneath, we performed fluorescence recovery after photo-

bleaching (FRAP) experiments (Figure 2A). Importantly, after

photobleaching both the base lipid membrane and the double

membrane patches show full fluorescent recovery with nearly

identical rates (Figure 2B,C). Analysis of the intensities of

the bleached double patches relative to the nonbleached base

bilayer (Figure 2C) shows recovery to 197% after 20 min. This

indicates that fluorescently labeled lipids rapidly diffuse into

both membranes in the double membrane patches. Recovery

to the double membrane patches can be analyzed in isolation

by subtracting fluorescence intensity from the base membrane.

Performing this subtraction (Figure 2C, black line) results in

a fluorescent recovery that is almost indistinguishable from

that of the base bilayer (Figure 2C, blue line). The lateral dif-

fusion constants of 0.7 ± 0.2 µm2 s−1 and 0.6 ± 0.1 µm2 s−1 for

the double patches and base bilayer, respectively, indicate that

diffusion into double membrane patches is indistinguishable

from the lateral diffusion throughout the base bilayer. Such

www.afm-journal.dewww.advancedsciencenews.com

Adv. Funct. Mater. 2017, 27, 1606265

Figure 1. Schematic representation of the method used to create lipid–protein multilayers on gold electrodes. (1) Negatively charged vesicles are added

to the gold electrode that has a 6-mercaptohexanol and mercapto-(ethylene-oxy)3-cholesteryl (at an ≈3:2 ratio) self-assembled monolayer aiding the

formation of the first planar lipid bilayer. (2) Poly-l-lysine is bound to the surface of the lipid bilayer creating a net positive surface charge. (3) More

negatively charged vesicles are added, which rupture on the poly-l-lysine-coated membrane to form a double membrane. (4) Adsorption of poly-l-lysine

to the double membrane. (5) Steps (3) and (4) are repeated n times to create n additional stacked bilayers.

full paper

wileyonlinelibrary.com

1606265 (3 of 9)

similar rates of diffusion imply that the two membranes in

the double membrane patches are interconnected via a lipid

phase. Similar patches of membrane stacks that are connected

to an underlying bilayer have previously been observed after the

addition of lipopolysaccharide[14] and, very recently, long chain

polyamines.[15]

As illustrated by the cartoons in Figure 2, there are a number

of possible configurations that allow rapid perpendicular diffu-

sion between the base and secondary membranes. Sliding or

folding structures may be induced by PLL in regions where

the base bilayer had a defect. Alternatively, rapid lipid transfer

between vesicles or from vesicles to a supported lipid bilayer

has previously been reported when the lipids adopt a so-called

stalk structure.[16,17] Typically, these stalks, which may be stabi-

lized by PLL, require two proximal leaflets to form a local neck-

like connection, resulting in a rapid lipid exchange between the

opposing membranes through lipid diffusion in the connected

leaflets.[18]

The formation of double membrane patches was monitored

in situ by AFM, which showed, upon addition of PLL, the

immediate formation (within tens of seconds; see Figure S1

in the Supporting Information) of double membrane patches

with clearly distinct sizes (Figure 2D): those with large diam-

eters of 500–5000 nm (3% by number) and those with diame-

ters averaging 90 ± 40 nm (the remaining 97%). The size of the

smaller patches remained stable over time, but larger patches

www.afm-journal.de www.advancedsciencenews.com

Adv. Funct. Mater. 2017, 27, 1606265

Figure 2. A) Series of fluorescence microscopy images before and after addition of PLL. After addition of PLL, fluorescence recovery is shown after

photo bleaching a central circular spot of a POPC/POPG 1:1 bilayer (0.5% TxRed DHPE). Note that the patches with double fluorescent intensity are

formed only after incubation with PLL. B) Intensity profiles across dashed lines in (A) at different time points. C) Plots of normalized fluorescence inten-

sity change where

and

denote intensity of single and double-bilayer patches, respectively, and prime (

′ and

″) the intensity within the bleached

area. The intensities of the bleached areas are normalized against either the base bilayer (

) or the double membrane patch (

), as indicated. The

black line represents the difference between “

′relative to

” and “

′relative to

”, as further explained in the text. D) AFM image of POPC/POPG

(1:1) bilayer after incubation with PLL (z-scale = 16.0 nm). The cartoon illustrations show the possible configuration that can allow diffusion between

multiple membranes: (1) sliding, (2) folding, or (3) stalk structures. Experiments were performed in 5 × 10−3 m MOPS, pH 7.0, 22 °C.

full paper

were observed to grow across the bilayer asymmetrically (i.e.,

in one direction only; see Figure S1 in the Supporting Informa-

tion). Although the resolution of the AFM is not sufficient to

confirm this, we speculate that the larger patches are created by

sliding or folding of the membrane, consistent with the growth

in one direction only, and the smaller patches are due to the

stalk formation.

The further addition of negatively charged vesicles to

bilayers coated with PLL produces an almost complete second

bilayer. AFM indicates that the new second lipid bilayer mem-

brane seamlessly connects to the previous double membrane

patches (Figure S2, Supporting Information). It is thus pos-

sible that the structures shown schematically in Figure 2 are

maintained when multilayers are formed. Further addition of

PLL to this double membrane produces an even higher den-

sity of third bilayer patches, while closing up the second bilayer.

This process continues with each additional bilayer, potentially

providing diffusion routes between the whole multilayer.

2.2. Quinone Diffusion in Lipid Multilayers

Quinones are typical coenzymes in electron transport chains

and shuttle electrons between many membrane proteins that

are associated with energy conservation/transduction. The

FRAP experiments described above suggest that lipids rapidly

diffuse between lipid membranes in the multilayer membrane

patches. It is therefore likely that lipophilic quinones such as

ubiquinone-10 (UQ10) and menaquinone-7 (MQ7), which stay

in the hydrophobic region of the bilayer, should also transfer

between membranes. To test this hypothesis and to confirm

the presence of intermembrane connections, quinone-medi-

ated electron transfer across stacked lipid membranes was

investigated by cyclic voltammetry of lipid multilayers con-

taining MQ7 or UQ10. The multilayers were built up layer-

by-layer on mixed self-assembled monolayers (SAMs) on the

surface of gold electrodes. To adhere the first “base” lipid layer,

the membrane was tethered to the surface with previously

established methods.[19,20] In short, a mixed SAM was formed

on a gold electrode composed of mercapto-(ethylene-oxy)3-

cholesteryl (membrane tether, EO3C) and 6-mercaptohexanol

(spacer). The “base” lipid membrane was then formed via

self-assembly from lipid vesicles, where the cholesterol tether

induces rupture of the vesicle and allows the formation of a

continuous planar lipid membrane (the “base” lipid mem-

brane). Consecutive membrane layers were then formed by

alternatively adsorbing PLL and vesicles (Figure 1). Figure 3A

shows a series of cyclic voltammograms (CVs) of the SAM

only, a 1:1 (w/w) POPC/POPG base membrane containing

1.5% (w/w) MQ7 and then each response after four cycles of

incubation with PLL followed by 3:1 POPC/POPG 1.5% (w/w)

MQ7 vesicles. These lipid compositions were chosen since we

have previously shown that this protocol creates additional

membranes with each PLL/vesicle incubation cycle on mica,

glass, and SiO2.[11] The CV of the first bilayer shows the MQ7

reduction (≈−0.45 V vs standard hydrogen electrode, SHE)

followed by oxidation (≈0.1 V vs SHE) peaks. The large peak

separation is partly caused by the thickness of the 6-mercap-

tohexanol layer in the SAM, reducing the electronic coupling

between electrode and MQ7. More importantly, however, it

is caused by the coupling of the electron transfer with pro-

tonation/deprotonation steps, which are slow due to the

lipid bilayer environment.[21] With each additional bilayer we

observe a linear increase in the current being transferred by

the MQ7 (i.e., increase in peak area), implying an increase in

the number of quinones that electrochemically interact with

the electrode.

The double-layer capacitance of the multilayer stacks

was estimated by electrochemical impedance spectroscopy

(Figure 3B). The formation of the first bilayer produced a

double-layer capacitance of ≈1.2 ± 0.1 µF cm−2 as estimated

from the diameter in the Cole–Cole plot. The capacitance is in

agreement with previous similar lipid membranes on the EO3C

system.[22] However, with each additional bilayer we observe

only small reductions in the double-layer capacitance, indi-

cating that the additional lipid bilayers have low resistance to

ions and thus likely contain defects.

CV analysis of the total charge transferred by MQ7

(Figure 3C) shows a linear increase for each additional lipid

membrane in the multilayer stack. An identically formed multi-

layer stack with UQ10 showed the same behavior (Figure S3,

Supporting Information). To determine if electron transfer was

limited by diffusion, CVs were performed on an eight-layered

system with scan rates varying between 10 and 200 mV s−1

(Figure 3D). The redox peak area shows no dependency on scan

rate, indicating that up to 200 mV s−1, UQ10 diffusion is not

limiting. There are no analytical solutions available to calcu-

late the diffusion in complex multilayered film systems such

as these, but it is clear that even in an eight-layered membrane

system, UQ10 in all layers is oxidized or reduced in less than a

second.

Although synthetic lipids offer a stable environment for

membrane proteins, activity of membrane proteins can be

dependent on the presence of particular annular lipids.[23] For

bacterial membrane enzymes, it can thus be beneficial to pro-

vide a more native lipid environment via the use of bacterial

lipid extracts. To study the layer-by-layer formation of multi-

layers composed of lipid extracts, we used E. coli “polar” lipid

extracts, which according to the supplier contain ≈23% phos-

phoglycerol (PG), 10% cardiolipin, and 67% phosphoethanola-

mine (w/w). The PG and cardiolipins give these membranes an

overall negative charge, as required for our LBL system with the

positively charged PLL. Figure 3E shows CVs at each stage after

incubating five cycles of E. coli polar extract vesicles containing

1% (w/w) UQ10. As with the POPC/POPG system, we observe

an increase in the quinone redox peaks with each addition of

vesicles/PLL. This suggests that, as long as the lipid mixture

contains sufficient negative charge, the LBL system can be

adapted to any lipid composition. It should be noted that E. coli

polar extract lipid also contains small amounts of ubiquinone-8,

native to E. coli.

From our results, the transfer of charge across multiple

membranes to the electrode via the reduction and oxidation

of the quinones is clear. What remains unclear is the mecha-

nism of how this process occurs. While AFM and fluores-

cence results suggest there are connections between adjacent

bilayers, allowing diffusion perpendicular to the plane of the

lipid membranes, there may be a number of other possible

www.afm-journal.dewww.advancedsciencenews.com

Adv. Funct. Mater. 2017, 27, 1606265

full paper

wileyonlinelibrary.com

1606265 (5 of 9)

transfer routes such as quinone–quinone electron transfer,

where a quinone in one bilayer passes an electron to a quinone

in the adjacent bilayer, or quinone hopping across membranes.

To further test the mechanism of electron transport, a double

membrane system was prepared in which the quinones were

omitted in the base bilayer. In the latter system, the reduc-

tion and oxidation of the quinones is still observed (Figure S4,

Supporting Information) confirming that the quinones are

able to diffuse between the membranes layers. Spontaneous

lipid transfer between membranes is very slow (timescale of

hours).[24,25] To test if the same applies to UQ10 transfer, a

single-membrane layer was formed without UQ10 and incu-

bated with a liposome solution containing UQ10. As expected,

no UQ10 was observed to transfer from the liposomes in solu-

tion to the membrane on the surface (Figure S5, Supporting

Information). Finally, to exclude direct quinol to quinone elec-

tron transfer across membranes, single bilayers containing

UQ10 were formed and incubated with liposomes also con-

taining UQ10. The addition of liposomes did not alter the

UQ10 signals, confirming that electrons are not passed on

from the membrane on the surface to liposomes in solution

(Figure S5, Supporting Information). Taken together, these

results strongly suggest that the primary mechanism is diffu-

sion of quinols through the lipid interconnections between the

membrane stacks.

For a given molecule to transfer between two adjacent

membranes, it needs to diffuse a certain distance to the

nearest membrane connection. As determined from AFM

images of double membrane patches on the base bilayer, the

average nearest neighbor distance between patches is 470 nm

(SD ± 250 nm). Assuming a quinone or lipid molecule would

need to travel on average half this distance to reach an inter-

membrane connection, we can roughly estimate the time

taken (t) to diffuse this distance (x) using the 2D diffusion

equation: t = x2/4D, where D is the diffusion constant. Given

the lateral diffusion constant of 2 µm2 s−1 for UQ10[26] it would

take ≈0.01 s to diffuse to the next bilayer in the multilayer.

This timescale corresponds to the fact that it takes less than a

second to oxidize or reduce all UQ10 in an eight-layered mem-

brane system (see above).

www.afm-journal.de www.advancedsciencenews.com

Adv. Funct. Mater. 2017, 27, 1606265

Figure 3. A) Cyclic voltammograms, B) Cole–Cole plots, and C) total charge transferred during MQ7 reduction of POPC-POPG (1.5% MQ7)/PLL

multilayers. D) Charge transferred (reduction of UQ10) in CVs as a function of scan rate for a POPC-POPG (0.5% UQ10)/PLL multilayer with eight

layers. E) Cyclic voltammograms of E. coli polar lipid (1% UQ10)/PLL multilayers (color coding as in (A)). Experiments were performed in the absence

of O2 and at 10 mV s−1 scan rates, 20 × 10−3 m MOPS, 30 × 10−3 m Na2SO4, pH 7.4, 20 °C.

Loading more pages...