Biomimetic membrane for desalination and water reuse 2

As detergent removal is critical for the membrane protein incorporation as well as maintaining the membrane stability and integrity, the first goal of this work aims to prove that the de

Chapter 2 AquaporinZ Incorporation via Binary-Lipid

Langmuir Monolayers

2.1 Introduction

LB technology is known to be a powerful method to study lipid-protein interaction at

the gas-liquid interface and form planar lipid bilayers Membrane insertion behavior

of water soluble proteins is commonly investigated on the LB surface to mimic

biological membranes However, hydrophobic membrane proteins like AqpZ which

have to be solubilized in a detergent solution are poorly studied with the LB

technique This is mainly due to the fact that the detergent can easily disrupt the

surface monolayer by either adsorbing onto the monolayer or solubilizing the

monolayer [92, 93] As a result, the process of membrane protein insertion is hindered

by the detergent-monolayer interaction Moreover, the incorporated protein has the

tendency to denature and lose its activity when it reaches the air-liquid interface [60]

All of these reasons make the incorporation of membrane proteins with the LB

method a challenging and intriguing work

In this study, we attempt for the first time to reconstitute AqpZ into the lipid bilayer

with the LB technique As detergent removal is critical for the membrane protein

incorporation as well as maintaining the membrane stability and integrity, the first

goal of this work aims to prove that the detergent adsorption on the lipid monolayer

can be suppressed through the addition of BioBeads in the subphase, and the

detergent removal rate is correlated to the amount of BioBeads and circulation in the

subphase For the second part of this work, we demonstrated a new AqpZ

incorporation approach with a binary-lipid Langmuir monolayer and propose a

three-step mechanism for protein incorporation As atomic force microscopy (AFM) is the unique method to characterize LB film in the nano-scale [94, 95], it was adopted for the membrane morphology study in this work

2.2 Materials and methods

2.2.1 Materials

Nickel-chelating lipids,

di-(9Z-octadecenoyl)-sn-glycero-3-(N-(5-amino-1-carboxypentyl)iminodiacetic acid)succinyl with Nickle (DOGSNTA) and dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) were purchased from Avanti Polar Lipids, Birmingham, United States n-dodecyl beta-D-maltoside (DDM) with purity > 99.5% was a product of Acros Organics, Geel, Belgium 10x Phosphate buffered saline (PBS) in ultra high purity (1st BASE, Singapore) was diluted 10 times with Milli-Q water (Millipore) before use (pH=7.4) The lipids solution was prepared in chloroform (Tedia, HPLC grade) at a concentration of 1mg/ml Bio-Beads SM-2 Absorbents from BIO-RAD was used to remove DDM in the subphase Grade V1 mica was purchased from SPI Supplies

1,2-2.2.2 Surface pressure measurement

The surface pressure measurement was carried out on a customized LB trough NIMA, Finland) with a trough area of 36x5 cm2 Before each experiment, the trough was wiped thoroughly with chloroform twice and subsequently rinsed with Milli-Q water The subphase in the trough was either Milli-Q water or PBS buffer solution, and all the experiments were performed at controlled room temperature (23±1°C) Surface pressure was measured by a pressure sensor based on the Wilhelmy plate method The effective surface area was controlled by a pair of Delrin barriers Lipid

(KSV-solutions were spread on the aqueous subphase when the barriers were opened to the

maximum and 60min was allowed for chloroform evaporation from the surface The

total volume of the subphase is 65±2 ml

For injection of DDM or AqpZ solution into the subphase, barriers were held

statically after the surface pressure reached 30mN m-1 An injection port located at the

center of the trough was utilized to reduce the disturbance on surface monolayer

during injection After injection, the subphase was circulated at 3ml/min to promote

equilibration for 2 hours and the change in surface pressure was recorded In the case

of detergent removal from the subphase, BioBeads were added into the dipping well

before spreading of the monolayer and stirred gently with a mini Teflon stir bar

Subphase circulation was performed throughout the entire detergent removal process

The air above the subphase was maintained at saturation level to limit evaporation of

water To obtain the pressure-area isotherms of lipids monolayer, the surface was

compressed or expanded at the rate of 10cm2/min All the experiments were repeated

for at least three times to ensure the reproducibility of the results

2.2.3 Surface tension measurement

A surface pressure sensor was used to determine the surface tension of

detergent/protein solutions A platinum Whilhelmy plate connected to the pressure

sensor was partially immersed in a 20ml PBS solution The surface pressure change

(π) of PBS buffer with changing DDM concentration in the buffer was measured by

the pressure sensor

The surface tensions of mixtures were calculated and plotted against DDM concentration Three times repetition were done for each data point and the error for all points are within 6%

2.2.4 LB film deposition and AFM scanning

LB film deposition was performed to transfer the monolayer from the air-liquid interface to the mica surface The surface pressure for deposition was always at 35mN m-1 To study the morphology of the DPPC layer with DDM disruption and BioBeads intervention, the DPPC monolayer was deposited onto a fresh cleaved mica surface first by the vertical LB method and followed with the horizontal LS method The transfer speed for the LB method is 2 mm min-1 The samples were immersed in Milli-Q water before and during AFM scanning In the case of protein-associated monolayer deposition, a first monolayer of pure DPPC-DOGSNTA (5:1) was prepared by LB deposition on the mica surface using another trough (KSV-NIMA, Inverted Microscopy Trough) The protein-associated DPPC-DOGSNTA monolayer was subsequently deposited onto the first layer by the LS method The samples were incubated in PBS buffer for 1 hour, and then imaged by AFM in PBS buffer environment

The films were imaged by a PicoSPM atomic force microscope (Agilent) in the Acoustic alternating current mode The samples were scanned with ULTRASHARP NSC15/AIBS cantilevers (resonance frequency between 265 to 400 kHz and typical force constant of 46N m-1) in aqueous solutions (Milli-Q water or PBS buffer) at room temperature (23 ±1°C) The scanning speed is less than 0.5 ln s-1

2.3 Results and discussion

2.3.1 DDM effects on DPPC monolayer

A well characterized lipid, DPPC, was selected in this experiment to demonstrate that

detergent penetration into the lipid monolayer could be eliminated with the addition

of BioBeads BioBeads are macroporous polystyrene beads that could remove

detergent efficiently by hydrophobic adsorption At room temperature (23°C±1),

DPPC at the air-water interface exhibits a liquid-condensed phase upon compression

to above 7 mN m-1; therefore, detergent has the poor accessibility to such a compact

monolayer [93] To avoid a heavy loss of surface lipids by detergent solubilization,

the DDM concentration in the subphase was controlled at 4 µM (2 mg L-1), which

was much lower than its critical micelle concentration (CMC), 0.18 mM [96] As

shown in the dashed lines of Figure 2.1, the isotherm cycle of the monolayer after the

addition of DDM displays a significant counterclockwise hysteresis, while it is

known that the pure DPPC monolayer does not show any hysteresis[97] The

hysteresis is caused by the DDM adsorption onto the monolayer at low surface

pressures and incomplete desorption from the monolayer at high surface pressures

Different amounts of BioBeads (50mg, 150mg and 300mg, in wet weight) were

introduced into the trough to study the influence of BioBeads addition on the

detergent removal rate Since BioBeads is only confined in the dipping well (1.5cm in

diameter) located at the center of the trough, effective detergent removal cannot be

achieved without subphase circulation

Figure 2.1 Pressure-area isotherm cycles of DPPC at 23°C after 6 hours

detergent removal with (a) 50mg, (b) 150mg, and (c) 300 mg BioBeads in the trough (dotted line) and subphase volume of 65±2ml The initial DDM concentration in the subphase is 2 mg L-1 The isotherm cycles after detergent removal are compared with pure DPPC isotherm (solid line) and isotherm cycle

without using BioBeads (dashed line)

The isotherm cycle of DPPC monolayer was obtained after 6 hours of detergent removal The relative amount of remaining DDM in the subphase can be reflected by the intensity of isotherm-cycle hysteresis as well as the lowest surface pressure in the cycle When 50mg BioBeads was used (Figure 2.1(a)), the isotherm cycle shows similar intensity of hysteresis with the one that has no BioBeads, but the lowest surface pressure has been reduced to 0mN m-1 from 5mN m-1, indicating that only a small fraction of the detergent in the subphase has been removed A significant

improvement can be observed from the isotherm cycle as the BioBeads amount

increased to 150mg (Figure 2.1(b)), with milder hysteresis and better proximity to the

pure DPPC isotherm With 300mg BioBeads (Figure 2.1(c)), though the monolayer

displays a gentle hysteresis, the shape of the compression isotherm has reverted to

that of a pure DPPC isotherm Therefore, DDM can be eliminated from the interface

at a faster rate when there is a higher BioBeads to DDM ratio, which is reflected by a

better recovery of DPPC isotherms

The effect of DDM on the DPPC monolayer was also demonstrated by AFM images

in Figure 2.2 Without the addition of BioBeads, the DPPC bilayer loses the original

compact structure (Figure 2.2(a)) and forms a “fluid-like” defective structure (Figure

2.2(b)), while with detergent removal by BioBeads, the compact bilayer structure

(Figure 2.2(c)) was restored Therefore, it can be concluded that the detergent content

at the air-liquid interface can be substantially reduced with a decrease in its subphase

concentration BioBeads reduce the adsorbed DDM in the DPPC monolayer by

removing the DDM in the bulk either before or after the DDM association with the

monolayer In the former case, DDM adsorbs to BioBeads faster than to the

monolayer; thus less detergent is available to disrupt the surface monolayer In the

latter case, DDM quickly reaches the surface monolayer, but the elimination of DDM

in the subphase leads to DDM desorption from the monolayer Both hypotheses may

be reasonable but require further verification

Figure 2.2 AFM topograph of (a) pure DPPC bilayer, (b) disrupted DPPC

bilayer by 2mg L-1 DDM in the subphase, and (c) DPPC bilayer with removal of subphase DDM All the images were scanned in ultrapure water environment The dark area corresponds to the mica surface The typical DPPC thickness should be around 4-5 nm as shown in (a) and (c) However with DDM disruption, the DPPC thickness is reduced below 4nm and the “fluid-like” defective structure is formed in the bilayer The cross-section profiles at the dashed lines are shown at the bottom of respective images The scale bars (in solid line) are 1µm in all the three images

Figure 2.3 Pressure-area isotherms of DPPC, DOGSNTA and their mixtures of different molar ratios (5:1 and 10:1) The temperature of experiments is 23±1°C

0 5 10

2.3.2 AqpZ incorporation via DPPC-DOGSNTA monolayers

Monolayers of DPPC and DOGSNTA mixture were produced by spreading the

mixed lipid solution with certain ratio on the PBS buffer surface Figure 2.3 shows

the pressure-area isotherms of pure DPPC, pure DOGSNTA and their mixtures of

different molar ratios (5:1 and 10:1) The pure DOGSNTA monolayer has no

liquid-expanded to liquid-condensed transition plateau in the isotherm showing that

DOGSNTA exist in the fluid-phase at the room temperature In contrast, DPPC has a

transition temperature of 40°C thus is in the gel-phase at the room temperature

Therefore, their mixtures display characteristics of both To preserve the compact

structure of the monolayer at high surface pressures, DPPC and DOGSNTA mixed at

a ratio of 5 to 1 was adopted for protein insertion As mentioned previously, DPPC is

able to form a closely packed monolayer at high surface pressures and resist the

invasion of detergents In the meantime, DOGSNTA provides binding sites for

His-tagged proteins [98], which ensures the monolayer to have a good affinity to AqpZ

Upon addition of DOGSNTA in the DPPC monolayer, the surface pressure at liquid

expanded-liquid condensed (LE-LC) phase transition increases and the transition

plateau becomes less prominent compared with the pure DPPC monolayer These

suggest that the monolayer formed with the DPPC-DOGSNTA mixture possesses

more fluidity than that formed with the pure DPPC

The AqpZ-DDM solution was injected at a surface pressure of 30mN m-1 This

surface pressure is selected for two reasons The first is to minimize the detergent

penetration into the monolayer As described above, DDM can be partially desorbed

(i.e squeezed out) from the tightly packed LC phase monolayer, which causes the

hysteretic feature in the isotherm cycle of DPPC/DDM Thus, at a high surface pressure of 30mN/m, less disruption of the monolayer by detergent is expected [99] Secondly, the transmembrane protein AqpZ has a great tendency to aggregate and be denatured when exposed to air Therefore, the densely packed monolayer would limit the exposure of proteins to air at the air-liquid interface

As the protein-detergent mixture is injected into the DPPC-DOGSNTA subphase, one critical issue is that the lipid monolayer may be dissolved when the detergent concentration is above CMC [93, 100] This is especially true for DOGSNTA because it is in the fluid phase at room temperature and therefore has no liquid condensed phase on the air-liquid interface However, it is known that AqpZ needs to

be stored at a detergent concentration higher than CMC to prevent aggregation The lowest association concentration, the critical aggregation concentration (CAC), of DDM in the presence of AqpZ in the system, was therefore studied by the surface tension method [101] With the presence of proteins or polyelectrolyte in the solution, complexation of the detergent with protein/polymer molecules is thermodynamically favored at the CAC rather than the formation of regular micelles [102, 103] The hydrophobic segment of proteins associates with hydrocarbon chains of detergents and forms a protein-surfactant complex As shown in Figure 2.4, at an AqpZ concentration of 0.5 mg L-1, the CAC of the mixed system is at 2 mg L-1 DDM, which

is well below the CMC value of pure DDM solution 85 mg L-1 [96] The existence of CAC proves that nonspecific hydrophobic interactions between DDM and AqpZ are strong enough to induce complex association of the two at a DDM concentration lower than CMC The protein-detergent complex formed at CAC has been named as a

“necklace” model with detergent micelles as beads on the protein chain [103-105]

Figure 2.4 Surface tension versus DDM concentration in the AqpZ-free solution

and AqpZ solution The base solution is PBS buffer with pH=7.4 For pure DDM

solutions (▲), the transition point of the trend indicates the CMC value of DDM

For DDM-AqpZ solutions (○), the first transition point (from left to right) is the

CAC value of the DDM-AqpZ complex system, and the second transition point

is the same with the CMC of DDM At an AqpZ concentration of 0.5mg/L

(0.02µM), the CAC of the mixed system is at about 2mg/L DDM, which is well

below the CMC value of pure DDM solution, 85mg/L

The CAC value was selected as the DDM concentration in the protein association

experiments According to previous research [98], the His-tagged membrane protein

with strong hydrophobicity could bind to the DOGSNTA monolayer within one hour

after injection into the subphase Due to the hydrophobic nature of transmembrane

proteins, AqpZ stays preferentially either in the detergent micelles or at the air-water

interface; therefore, a reduction in detergent concentration could promote association

of proteins with the surface monolayer In order to achieve a better detergent

removing rate, 300mg BioBeads was added for detergent removal for 3, 6 or 12

hours With the LS technique, the final protein-associated monolayer was deposited

onto a second DPPC-DOGSNTA monolayer prepared on the Mica surface The

bilayer was imaged by AFM after incubation in the PBS buffer for one hour Only the

images from 6 hours detergent removal condition show hints of incorporated AqpZ

25 30 35 40 45 50 55 60 65 70

CAC

CMC

(See Figure 2.5) AFM imaging shows small protrusions on top of the bilayer film with an average height of 4 Å above the lipid bilayer A similar protein protrusion height was reported by Scheuring et al who used AFM to study AqpZ 2D crystal structures in 1999 [106] Therefore, Figure 2.5 might give an indirect indication of incorporated AqpZ However, more research should be done in the future to validate the hypothesis Detergent removal for a longer time span may cause the protein aggregation in the low detergent environment, while for a shorter time span, DDM would destabilize lipid monolayer which can be rapidly collapsed after deposition Furthermore, as the DDM adsorption and removal from the surface monolayer is a complicated process, it is very unlikely to interpret AqpZ-monolayer association from the surface pressure evolution profile

Figure 2.5 AFM topograph of AqpZ associated lipid bilayer after 6 hours of

detergent removal (recorded in PBS buffer solution, pH=7.4) The dark area corresponds to the mica surface while the lighter area corresponds to the bilayer surface The thickness of the bilayer is about 4-5 nm (a) Lower magnification topograph of the bilayer shows hints of incorporated AqpZ (arrows) Scale bar is 1µm (b) Higher magnification topograph shows the inserted proteins are

2

1

distributed in the lipid bilayer Scale bar is 200nm (c) and (d) are the

cross-sectional images of protrusions pointed by arrow 1 and arrow 2, respectively

Figure 2.6 Schematic presentation of the proposed AqpZ incorporation

mechanism: (a) protein attachment on the DPPC-DOGSNTA monolayer, (b)

partial insertion of AqpZ into the DPPC-DOGSNTA monolayer as DDM is

slowly removed, (c) protein reconstitution into the lipid bilayer after LS

deposition onto the second DPPC-DOGSNTA monolayer

A possible mechanism of protein incorporation made up of three phases has been

proposed as shown in Figure 2.6 In the first phase, His-tagged AqpZ attaches to the

DOGSNTA domain in a short time after injection into the subphase and the

monolayer is meanwhile destabilized by DDM Proteins are maintained at the CAC

and form a “necklace” complex with detergent molecules In the second phase,

BioBeads further decreases the detergent concentration, which accounts for the partial

insertion of AqpZ into the DOGSNTA monolayer The transmembrane protein may

be reversibly unfolded in the first two phases due to the low detergent concentration; presumably the protein-protein aggregation process is slow in the diluted solution After combining the protein associated monolayer with a second lipid monolayer, AqpZ could possibly refold to its native state within the bilayer film Many previous studies have reported that the unfolded membrane protein could be refolded to its native state when it is inserted into the lipid bilayer as long as its secondary structure

is still preserved [107, 108] The amphiphilic environment provided by the lipid bilayer could spontaneously adsorb, host and refold the membrane proteins, while the detergent is not necessarily involved in the whole process This reconstitution of AqpZ in the lipid bilayer is the third phase However, it is known that membrane protein can only insert into fluid-phase lipid bilayers In this experiment, although the gel-phased DPPC contributes to maintain the monolayer integrity, it becomes an obstacle during the protein incorporation process In Figure 2.5(a), other than the small round protrusion features (pointed by arrows), there is a large corrugated domain (upper left corner) which is suspected to be the incompletely constituted protein clusters

2.4 Conclusion

In summary, as the first attempt to reconstitute hydrophobic membrane proteins with the aid of LB technique, the removal of detergent from the air-liquid interface was investigated and a protein incorporation model was predicted Since detergents bring many complications in LB studies, we selected the lowest detergent-protein association concentration to prevent dissolution of surface monolayer and to minimize the membrane defects Both the subphase circulation and the quantity of

Biobeads added could determine the detergent removal rate Upon removal of

subphase detergent with BioBeads, the incorporation was achieved by transfering the

AqpZ-associated binary-lipid monolayer onto a preformed pure binary-lipid

monolayer using the Langmuir-Schaefer deposition method The existence of AqpZ

in the lipid film was observed by AFM, and AqpZ refolding in the lipid bilayer was

largely expected This newly developed approach of AqpZ reconstitution may

potentially be applied in biomimetic membrane formation for water purification or

biosensor applications if continuous lipid bilayer could be deposited onto a porous

support However, we are not able to fully remove the defective areas using current

technique Therefore, better solutions have to be developed

Chapter 3 Stabilization and Immobilization of Aquaporin Reconstituted Lipid Vesicles for Water Purification

3.1 Introduction

Inspired from biological membranes where AQPs provide extraordinary water permeability and selectivity, the use of aquaporin proteins in membrane for water desalination has recently attracted worldwide attention Although many efforts have been made to realize this target, most of these efforts have focused on AQP incorporation in planar lipid membranes, either supported lipid bilayers [88] or black lipid membranes [82] The stability of pore-suspending planar lipid membranes formed is questionable under high hydraulic pressure and turbulence because the lipid bilayer thickness is only few nanometers Therefore, the challenge in this current work is to design a stable aquaporin embedded water selective layer that can maintain its structure and perform under high pressure and surface agitation In this study, we aim to develop a type of stable liposomes with aquaporin incorporation, which can be cross-linked and immobilized on a porous polymeric support to produce a biomimetic nanofiltration membrane

Liposomes are typically unstable and tend to rupture on hydrophilic surfaces such as mica and silica, which result in the development of different approaches for their stabilization Cross-linkable lipids like diyne lipids can be polymerized easily in the bilayer but the lateral mobility of the lipids that are required for the protein functioning are largely diminished after cross-linking [109] Hence, we choose to use methacrylate monomers to cross-link the liposomes, which has also been suggested

by earlier studies [79, 110, 111] With the aid of UV-induced free radical