Membrane Transport
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport Taishi Tonooka, Koji Sato, Toshihisa Osaki, Ryuji Kawano, and Shoji Takeuchi*
This
paper describes picoliter-sized lipid bilayer chambers and their theoretical model for the rapid detection of membrane transport. To prepare the chambers, semispherical aqueous droplets are patterned on a hydrophilic/hydrophobic substrate and then brought into contact with another aqueous droplet in lipid-dispersed organic solvent, resulting in the formation of the lipid bilayers on the semispherical droplets. The proposed method implements the lipid bilayer chambers with 25-fold higher ratio of lipid membrane area (S) to chamber volume (V) compared to the previous spherical droplet chambers. Using these chambers, we are able to trace the time-course of Ca2+ influx through α-hemolysin pores by a fluorescent indicator. Moreover, we confirm that the detection time of the substrate transport is inversely proportional to the S/V ratio of the developed chambers, which is consistent with the simulation results based on the developed model. Our chambers and model might be useful for rapid functional analyses of membrane transport phenomena.
1. Introduction Fluorescence detection of membrane transport by artificial lipid bilayers has been used for functional analyses of membrane proteins such as transporters that have low transport capacity or the function of weakly-charged molecule transport.[1–4] The detection system used for this purpose generally utilizes fluorescence molecules and a small chamber sealed with a lipid bilayer (lipid bilayer chamber), and traces the change in fluorescence intensity resulting from transportation through the bilayer. For rapid fluorescence detection targeting high-throughput analyses of membrane transport phenomena, the ratio of lipid bilayer area and
T. Tonooka, K. Sato, T. Osaki, S. Takeuchi Institute of Industrial Science The University of Tokyo 4–6–1 Komaba Meguro-ku, Tokyo 153–8505, Japan E-mail: [email protected] T. Osaki, R. Kawano, S. Takeuchi Kanagawa Academy of Science and Technology 3–2–1 Sakado Takatsu-ku, Kawasaki Kanagawa, 213–0012, Japan DOI: 10.1002/smll.201303332 small 2014, 10, No. 16, 3275–3282
chamber volume (S/V) should be as high as possible; it is considered that a larger bilayer area increases the number of transporters incorporated while a smaller chamber volume condenses the transported substrate.[5] Although liposomes are generally used as the lipid bilayer chambers,[2–4,6,7] it is difficult to control the chamber size of these systems, owing to which they are unsuitable for high-throughput quantitative analyses. Two systems have been proposed for use as volumecontrollable chambers: micromachined chambers sealed with lipid bilayers[5,8–11] and droplet chambers in which monodisperse microdroplets come into contact with other droplets in a lipid-dispersed organic solvent.[12,13] Micromachined chambers provide controlled volume reduction, resulting in a variable S/V ratio;[5,8–11] however, it is difficult to achieve reproducible sealing of the chamber.[13–15] On the other hand, droplet chambers provide high reproducibility of the sealing;[13] nevertheless, the S/V ratio is relatively low (S/V < 4 µm2/pL) owing to the large droplet volume (several nanoliters).[12,13] Although scaling down of the lipid bilayer chamber is a straightforward approach for increasing the S/V ratio, handling of floating microdroplets, especially for volumes on the picoliter scale, is difficult that in turn makes the formation of reproducible lipid bilayers difficult. Consequently, droplet chambers with a high S/V ratio have not been realized yet.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
3275
full papers
T. Tonooka et al.
Herein, we propose a picoliter droplet chamber array with improved S/V ratio of up to 100 µm2/pL. To fabricate the droplet chambers, we implement an array of picoliter aqueous droplets in lipid-dispersed organic solvent using a hydrophilic/hydrophobic patterned substrate. The droplets become lipid bilayer chambers by bringing another droplet into contact on top of them. Since the droplets are arrayed and stabilized on the substrate, this method allows reproducible formation of the lipid bilayer chambers on the droplets of picoliter volumes. Additionally, the semispherical shape of this droplet chamber provides a higher S/V ratio than that achievable using previously reported spherical droplet chambers of the same volume. In this paper, we first discuss a theoretical model for fluorescence detection using the lipid bilayer chamber to investigate how the detection time of membrane transport changes according to the S/V ratio of the chamber. We then conduct formation of the picoliter droplet chambers and evaluate the S/V ratio of the chambers. Finally, we examine fluorescence detection of membrane transport by the droplet chambers with different S/V ratios.
2. Results and Discussion 2.1. A theoretical Model for Fluorescence Detection at Various S/V ratios For fluorescence observation of the membrane transport phenomena, the area (S [m2]) of a lipid bilayer membrane is an important factor in addition to the volume (V [m3]) of the chamber that contains the transported substrates.[5] Thus, we evaluated the influence of the S/V ratio on the fluorescence detection of substrate transportation by the following model. The relative fluorescence intensity (ΔF/F0) in the chamber, i.e., the intensity increment (ΔF) relative to the intensity at time zero (F0), was used as the index for fluorescence detection. The intensity increase was due to the binding of fluorescence indicators in the chamber with transported substrates with a dissociation constant (Kd [mol/m3]). We assumed that the number of transporters in the lipid bilayer chamber sigmoidally increased with time as the incorporation of transporters would start after the bilayer formation at time zero and saturate at the maximum surface density.[16] Under this assumption, the number of transporters is given by N (t ) =
N max 1 + exp[ −Gn (t − t s )]
t
∫0 N (t )ST dt V
(2)
where T [mol/s] is the transport rate of a single membrane protein per unit time. Therefore, the time-course of the
3276 www.small-journal.com
( )
is the maximum relative fluorescence intensity where ⎛⎜⎝ ΔFF0 ⎞⎟⎠ max of a fluorescent indicator and Gi is the gain of the sigmoidal curve of relative fluorescence intensity. Figure 1a shows sigmoidal curves plotted based on Equation(1), expressing four different examples of transporter incorporation (from steep (no. 1) to gradual (no. 4)). Figure 1b shows the time courses of the fluorescence intensity normalized by (ΔF/F0)max as 100 [%] with different S/V ratio values. We defined the detection time td as the time that corresponds to the fluorescence detection threshold. Note that by fluorescence observation, substrate transportation will be detected only after a certain intensity threshold is exceeded. Although the threshold for the detection of fluorescence signals is defined based on the background noise, we here simply set the threshold to 15% owing to the difficulty in reliable calculation of the noise level. As shown in Figure 1b, the curve shifts toward the left with an increase in the S/V ratio. These results reconfirm that a larger S/V ratio will enable faster detection of substrate transportation. Figure 1c and 1d show the relationship between td and S/V ratio. The results show that td becomes shorter with increasing S/V ratio and the maximum slope of the curves fits in inverse manner. Note that there is a range where the slope is gradually changing, deviating from the inverse proportion. This range is due to the time-dependent incorporation of transporters. When td exists at the transition state of the incorporation in Figure 1a, increasing S/V ratio to shorten td is not effective because the total number of transporters incorporated drastically decreases with decreasing td; smaller number of transporters causes longer detection time even at the larger S/V ratio. Moreover, we found that the curves mainly shift to the right direction when NmaxT, i.e. transport capacity per unit area of the lipid bilayers, becomes smaller (from 10×NmaxT to 0.01×NmaxT shown in Figure 1c), while the curves mainly shift to the upper direction when the sigmoidal curve of incorporation becomes more gradual (Figure 1d). These characteristics may provide a guideline to design lipid bilayer chambers with various S/V ratios.
(1)
where t [s] is time, Nmax [proteins/m2] is the maximum number of membrane proteins per unit area of a lipid bilayer, ts [s] and Gn are parameters that determine the sigmoidal curve of the membrane protein incorporation per unit area of the lipid bilayer. Using N(t), the concentration of substrates (transported molecules) in a lipid bilayer chamber is given by C (t ) =
relative fluorescence intensity is expressed by a sigmoidal curve as ΔF F0 max ΔF (t ) = (3) F0 1 + exp[ −Gi ln[C (t )/K d ]]
2.2. Formation and Characterization of Lipid Bilayer Chambers on Semispherical Microdroplets Lipid bilayer chambers were formed on a hydrophilic/hydrophobic patterned glass substrate fabricated by standard photolithography (Figure 2a). The fabricated substrate was patterned such that the diameters of hydrophilic circles were 40 and 200 µm, and the spaces between the patterns were 40 and 200 µm, respectively (Figure 2b). First, a phospholipid dispersed in n-hexadecane was introduced on the hydrophilic/ hydrophobic patterned glass substrate (Figure 2c1). Second, a semispherical droplet array was formed on the circular hydrophilic areas by spotting a buffer solution (Figure 2c2).
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 16, 3275–3282
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport
(a)
(b) 100
N(t)/Nmax (%)
80 60 no. 1
(ts:50,Gn:0.12)
40
no. 2
(ts:100,Gn:0.06)
no. 3
20
(ts:200,Gn:0.03)
no. 4
(ΔF/F0)/(ΔF/F0)max (%)
100
(ts:400,Gn:0.015)
0 0
300 600 Time (s)
(c) 106
(d)
105
60 40
td (s)
103
S/V=4
600 1200 Time (s) td
1800
no. 1 no. 2 no. 3 no. 4
105 104 103 102
10
10 1
Detection threshold
106
102
1
S/V=10 (μm2/pL)
20
td (s)
104
S/V=30
0 0
900
0.01×NmaxT 0.1×NmaxT 1×NmaxT 10×NmaxT
S/V=100
80
10
102 103 104 105 106
S/V (μm2/pL)
1
1
10
102 103 104 105 106
S/V (μm2/pL)
Figure 1. Simulation results of fluorescence detection of molecular transport through transporters by using a lipid bilayer chamber. (a) Time courses of transporter incorporation into a lipid bilayer; the curves represent examples of steep (no.1) to gradual (no.4) incorporation. The parameters are shown in the legend. (b) Time course of the relative fluorescence intensities calculated based on the curve no.1 in Figure 1a with different S/V ratios. Detection threshold was set at 15%. (c) Detection time vs. S/V ratio calculated based on the curve no.1 in Figure 1a with different NmaxT values. (d) Detection time vs. S/V ratio with different types of transporter incorporation corresponding to no.1–4 in Figure 1a. The parameters used in the solid curves were NmaxT [mol/s·m2] = 5.0 × 10−8, Gi = 2.34, Kd [mol/m3] = 0.662. Gi and Kd were experimentally estimated (Supporting information 1).
Lipid monolayers autonomously covered the droplet surfaces because of the amphiphilic characteristics of lipid molecules. Third, an upper aqueous droplet was infused on and brought into contact with the droplet array on the substrate (Figure 2c3). Since a lipid monolayer covered the upper droplet, a thick lipid/hexadecane/lipid membrane was formed when the droplets came into contact. The thick membrane became a lipid bilayer within several minutes, forming a lipid bilayer chamber. The lipid bilayer formation started when a part of the two lipid monolayers[18–20] on the patterned and upper droplets came into contact with each other; the area of the lipid bilayer gradually increased and finally reached a steady state (Figure S2).[21] Figure 3a shows the lipid bilayer chamber array: Thirty-six lipid bilayer chambers were observed in the field of view. The boundary of a lipid bilayer on a semispherical droplet is visible in a closeup image (Figure 3b). It should be noted that the patterned semispherical droplets were always smaller than the hydrophilic areas on the substrate, probably because of the balance of interfacial tension between the patterned aqueous droplet, small 2014, 10, No. 16, 3275–3282
lipid solution, and substrate surface. For precise control of the droplet diameter, this balance of the tension should be adjusted by surface modification of the substrate or contents of the solutions. Lipid bilayer chambers were then characterized by the volume of the chamber (V), the area of the lipid bilayer (S), and the S/V ratio (Table 1). Figure 3c shows cross-sectional images of a lipid bilayer chamber formed on a substrate with 40-µm-diameter hydrophilic patterns (named as 40-µm chamber). Interestingly, the formed lipid bilayer membranes had slight curvature along the surface of the patterned semispherical droplets, as shown in the longitudinal image. We considered that the curvature was due to the difference between the internal pressure in the upper and patterned droplets. The average volume of the 40-µm chambers was hundred-fold smaller than that of the 200-µm chambers, while the membrane area of the 40-µm chambers became 35 times smaller than that of the 200-µm chambers. The variations in both chamber volume and area were relatively small, as described in Table 1. Accordingly, the S/V ratio obtained
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3277
full papers
T. Tonooka et al.
(a)
(b)
(i) Cytop spincoat, Al deposition, S1818 spincoat
Hydrophilic area
(iv) Cytop etching
Hydrophobic area
(ii) S1818 patterning (v) S1818, Al removal
(iii) Al etching Glass Cytop
(c1)
S1818 Al
(c2)
Capillary Semispherical Lipid solution aqueous (hexadecane) Hydrophobic droplet
Hydrophilic
Lipid
Lipid monolayer
Capillary
Upper aqueous droplet
Lipid bilayer
Semispherical microdroplet
Hexadecane
(c3)
Lipid bilayer chamber
Figure 2. Formation of lipid bilayer chambers on a hydrophilic/hydrophobic patterned substrate. (a) Fabrication process for the hydrophilic/ hydrophobic patterned substrate. (b) Microscopic image of the substrate. The scale bar represents 100 µm. (c) Formation procedure for lipid bilayer chambers: (c1) filling with lipid-dispersed organic solvent, (c2) patterning of an aqueous solution on the substrate provides a picolitersized semispherical droplet array, and (c3) contact of an upper aqueous droplet on the semispherical droplets, resulting in lipid bilayer formation on the semispherical droplets.
for the 40-µm chambers was 100 µm2/pL, and three times larger than that of the 200-µm chambers. Note that the droplet chambers commonly used in previous works showed lower S/V ratio of up to 4 µm2/pL; these droplet chambers consist of droplets with volumes as low as 10 nL and diameters of a few hundred micrometers.[12,13] Our droplet patterning method using a hydrophilic/hydrophobic pattern is simple, and the implementation of a reproducible methodology scales down the chamber volume to a few picoliters. Additionally, the easily observable bilayer-membrane area is another advantage of the semispherical chamber design, since it potentially allows microscopic observations of membrane proteins on the lipid bilayer membrane directly.[22–24] 2.3. Fluorescence Detection of Ca2+ Transport Through Membrane Proteins The Ca2+ influx through an α-hemolysin nanopore [25–27] was monitored by the developed lipid bilayer chambers to validate our method for fluorescence detection of membrane transport. The lipid bilayer chambers were formed under the following conditions. The upper aqueous droplet contained Ca2+ with/without α-hemolysin, while the patterned semispherical droplets contained a Ca2+ fluorescent indicator, Fluo-4. Figure 4a is representative time-lapse images of a lipid
3278 www.small-journal.com
bilayer chamber array showing the fluorescence intensity increase due to Ca2+ influx through α-hemolysin pores. In this experiment, 30 lipid bilayer chambers out of 36 spots (83%) remained after the upper aqueous droplet came into contact, while the rest were fused with the upper droplet via membrane rupture. The fluorescence intensities of all 30 chambers increased within 15 min. Figure 4b shows the time courses of the fluorescence intensities in the individual lipid bilayer chambers under the initial conditions of including/excluding α-hemolysin. We confirmed that the relative fluorescence intensities (ΔF/F0) increased in all lipid bilayer chambers only with the addition of α-hemolysin; the convex increase followed by plateau is consistent with the model derived in section 2.1. Without α-hemolysin, the change in the relative fluorescence intensity was negligible throughout the experiment. These results demonstrated that the developed lipid bilayer chambers were useful for introducing α-hemolysin nanopores and for observing ion transports through the pores.
2.4. Comparison Between 40- and 200-μm Chambers for the Fluorescence Detection We compared the detection time at two S/V ratios of the 40and 200-µm chambers using Ca2+ transport experiments with
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 16, 3275–3282
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport
to increase steeply with Ca2+ influx and reached the detection threshold within a shorter period than in the case of the 200-µm chambers (Figure 5b). As the S/V ratio becomes 3-fold higher from the 200-µm chamber to the 40-µm chamber, the detection time reduced by about 3-fold. This result also agreed with the simulation results that showed the detection time was inversely proportional to the S/V ratios (Figure 1c). We further performed Ca2+ imaging with a lower number of incorporated nanopores using the two chambers (Figure 5c). As a result, we confirmed that 59% of the 40-µm chambers exceeded the detection threshold within 15 min; on the other hand, no 200-µm chamber reached the detection threshold throughout the experiment. A higher S/V ratio was advantageous for detecting minimal membrane transport phenomena within a given period. These results confirmed that increasing the S/V ratio for a lipid bilayer chamber system not only allows rapid detection but also promotes high sensitivity detection of membrane transport activities.
3. Conclusion
Figure 3. Lipid bilayer chamber array formed on a substrate with the 40-µm-diameter pattern. (a) Array of lipid bilayer chambers. (b) Close-up image of a lipid bilayer chamber. (c) Confocal fluorescence image of (b). Green fluorescence stained by calcein shows a semispherical aqueous droplet; magenta fluorescence stained by DiI shows the lipid-dispersed organic solvent. (d) Longitudinal image at the white dotted line in (c). The white dotted line in (d) indicates the height of the transverse image in (c). The black and white scale bars are 100 and 10 µm, respectively.
α-hemolysin pores. A representative result of Ca2+ imaging for each chamber is presented in Figure 5a. The relative fluorescence intensity increased in a sigmoidal manner, which agrees with the model shown in Figure 1b. We determined the detection threshold at F0+10σF0; the average (F0) and standard deviation (σF0) of background intensity were estimated from the initial 30 s of recording in each chamber. The fluorescence intensities of the 40-µm chambers tended
In this study, we developed lipid bilayer chambers with increased S/V ratios of up to 100 µm2/pL by using picolitersized semispherical droplets. The developed chambers successfully detected Ca2+ influx through the α-hemolysin pores by a fluorescent indicator. The increase of the fluorescence in the chambers agreed with the theoretical model. By comparing two chambers having different S/V ratios, we confirmed that the larger S/V ratio provides rapid detections. The model we derived would allow us to understand how the S/V ratio affects the detection time with various properties of membrane proteins including transport capacity and incorporation manner of proteins. We believe that our picoliter-sized droplet chambers are useful for rapid screening of membrane transport such as drug resistance transporters on tumor cells.
4. Experimental Section Materials and Chemicals: The substrate for obtaining a lipid bilayer chamber array was fabricated from a glass slip (thickness of 150–200 µm; Matsunami Glass Ind., Osaka, Japan) and CytopTM (perfluoro polymer; CTL-809M; Asahi Glass, Tokyo, Japan). S1818 (positive photoresist; Shipley, MA, USA), aluminum etchant (Wako Pure Chemical Industries, Tokyo, Japan), and NMD (S1818 developer; Tokyo Ohka Kogyo, Kawasaki, Japan) were used in the fabrication process. Diphytanoylphosphatidylcholine (DPhPC; Avanti Polar Lipids, AL, USA), asolectin (Sigma-Aldrich, MO, USA), and n-hexadecane (Kanto Chemical, Tokyo, Japan) were used for pre-
Table 1. Properties of 40- and 200-µm lipid bilayer chambers. The S/V ratios were significantly different (p < 0.01, Mann-Whitney U test). n
Lipid bilayer area (S)
Chamber volume (V)
S/V ratio
40-µm chamber
9
136 ± 24 µm2
1.4 ± 0.1 pL
100 ± 10 µm2/pL
200-µm chamber
4
4760 ± 1257 µm2
158 ± 25 pL
30 ± 3 µm2/pL
small 2014, 10, No. 16, 3275–3282
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3279
full papers
T. Tonooka et al.
Device Fabrication: The hydrophilic/ hydrophobic pattern was fabricated on a 0 min 2 min glass substrate by a common photolithography process.[17] A cleaned glass substrate was used as the hydrophilic surface while the Cytop pattern was used as the hydrophobic part on the glass. The detailed process is as follows. First, a glass substrate was rinsed with acetone and 2-propanol. Cytop (9 wt%) was then spin-coated on the glass Lipid bilayer chamber substrate at 3000 rpm (MS-A100; Mikasa, Tokyo, Japan). Coated Cytop was then baked at 50 °C for 30 min and then at 180 °C for 1 h to evaporate the solvent. A thin Al layer was deposited on the substrate by using a high-vacuum evaporator (SVC-700 TM; Sanyu 3 min 5 min Electron, Tokyo, Japan). On the Al layer, a positive photoresist was spin-coated. An array of 40-µm or 200-µm circles was patterned on the photoresist by UV photolithography (PEM-6M; Union Optical, Tokyo, Japan). The pattern on the resist and Al was developed using NMD and Al etchant. Finally, Cytop was etched by applying O2 plasma (FA-1; Samco, Kyoto, Japan) and a circular pattern was obtained on the Cytop layer. The remaining S1818 and Al were removed using acetone and etchant. The substrate was heated at 150 °C for 1 h to reactivate the hydropho(b) 35 bicity of Cytop, and immersed in a Piranha α-hemolysin (+) solution at 60 °C for 1 min to enhance the 30 α-hemolysin (-) hydrophilicity of the glass surface. The 25 obtained hydrophilic/hydrophobic pattern is shown in Figure 2b. The thickness of Cytop 20 was 1.0 µm, measured by a stylus profilom15 eter (Dektak 6M; Veeco, NY, USA). Formation of Lipid Bilayer Chamber: The 10 formation process for lipid bilayer cham5 bers consists of the following three steps. In the first step (Figure 2c1), about 100 µL 0 of n-hexadecane was poured on the hydro-5 philic/hydrophobic patterned substrate. In 0 1 2 3 4 5 6 7 8 the second step (Figure 2c2), droplet patTime (min) terning was conducted using a glass capilFigure 4. Ca2+ imaging of Ca2+ influx through α-hemolysin pores using the 40-µm lipid bilayer lary (inner diameter ranged from 200 µm to chamber array. (a) Fluorescence time-lapse images of lipid bilayer chambers with 100 nM α-hemolysin. Scale bar is 100 µm. (b) The relative fluorescence intensity obtained from Ca2+ 1 mm; Narishige, Tokyo, Japan) and a microimaging in each lipid bilayer chamber over time. Each line corresponds to the conditions with pump (PicoPipet; Altair, Kanagawa, Japan); the capillary was filled with a buffer solution the presence (black lines, n = 30) or absence (gray lines, n = 27) of 100 nM α-hemolysin. (10 mM HEPES, pH 7.4 NaOH) and mounted on a micromanipulator (Eppendorf, NY, paring the lipid bilayer chambers. Calcein (Sigma-Aldrich) and USA). The buffer solution was ejected from the capillary by using 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlo- the pump while the capillary tip traced the pattern on the subrate (DiI; Invitrogen, CA, USA), α-hemolysin from Stapylococcus strate. This procedure automatically patterned the buffer on the aureus (Sigma-Aldrich), and Fluo-4 (Pentapotassium salt; Ca2+ flu- hydrophilic circle area of the substrate, and produced an array orescent indicator; Dojindo Laboratories, Kumamoto, Japan) were of semispherical microdroplets. In the third step (Figure 2c3), a used for the characterization of the lipid bilayer chamber array. buffer solution for the upper aqueous droplet (10 mM HEPES, pH Other chemicals were purchased from Kanto Chemical or Wako 7.4 NaOH) was ejected into n-hexadecane in the same manner, Pure Chemical Ind. (Tokyo, Japan). Milli-Q water (Millipore, MA, and brought into contact with the patterned semispherical droplets on the substrate by controlling the manipulator. Owing to USA) was used as deionized water (DI water).
ΔF/F0
(a)
3280 www.small-journal.com
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 16, 3275–3282
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport
(a)
(b) 25
8
40-μm chamber
20 Time (min)
6
ΔF/F0
15 10 Threshold (40-μm)
5 0
200-μm chamber
3
6 9 12 Time (min)
5 4 3 2 1
Threshold (200-μm)
0
*
7
15
0
S/V: 100 μm2/pL (40-μm chamber)
S/V: 30 μm2/pL (200-μm chamber)
(c) 1.8 1.5 40-μm chamber
ΔF/F0
1.2 0.9 0.6 0.3
Threshold (40-μm) Threshold (200-μm)
0
Supporting Information
200-μm chamber
0
used a 10 mM HEPES buffer solution (pH 7.4 NaOH) containing 100 µM Fluo-4 and 500 µM EDTA for patterned semispherical aqueous droplets, and the same 10 mM HEPES buffer solution (pH 7.4) containing 100 mM CaCl2 for an upper aqueous droplet. To make the osmotic pressure equal, 150 mM of KCl was added to the semispherical droplets. EDTA was used as a chelator to suppress background Ca2+ in the semispherical droplets. The fluorescence images were taken by an inverted microscope (IX71; Olympus, Tokyo, Japan). The time course of fluorescence intensity in the droplets was monitored using Aquacosmos software and an EM-CCD camera (C9100; Hamamatsu Photonics, Shizuoka, Japan). The average fluorescence intensities inside the droplets were evaluated using Image J. The fluorescence intensities (F) were normalized by the intensities right after lipid bilayer formation (F0) as (F – F0)/ F0 (denoted as ΔF/F0).
3
6 9 12 Time (min)
15
Supporting Information is available from the Wiley Online Library or from the author.
Figure 5. Comparison between 40- and 200-µm lipid bilayer chambers for fluorescence detection of Ca2+ influx through the α-hemolysin pores. (a) Representative time-courses of fluorescence intensity in the 40- and 200-µm chambers with 100 nM α-hemolysin. The detection thresholds, determined as F0+10σF0, are presented in the graph. (b) Detection times obtained in the 40-µm chamber (n = 30) and 200-µm chamber (n = 4). Error bars: standard deviation; *: p < 0.05 (Student’s t-test). (c) Representative time-courses of fluorescence intensity in the 40- and 200-µm lipid bilayer chambers with 10 nM of α-hemolysin.
the pressure from the upper aqueous droplet, the membrane at the interface between the droplets was transformed into a lipid bilayer (Figure 2c3).[19] Estimation of Volume and Membrane Area of Lipid Bilayer Chambers: The volume of the lipid bilayer chambers was determined as follows. A series of cross-sectional images of the lipid bilayer chambers was obtained using an inverted confocal microscope (LSM780, Carl Zeiss Microscopy, Jena, Germany) together with a bright field image. Here, the chambers and n-hexadecane were visualized with calcein (green) and DiI (magenta), respectively, as shown in Figure 3c. The transverse confocal images of the chambers were taken in increments of 0.5 µm step in perpendicular direction. The images of calcein were binarized by the particular threshold using an image processing software (Image J; NIH, USA), and areas with intensity higher than the threshold were integrated along the perpendicular direction, defining the volume of the lipid bilayer chamber. The lipid bilayer areas were measured from the bright filed images by using Image J. Ca2+Imaging of the Lipid Bilayer Chambers: Ca2+ imaging of the lipid bilayer chambers was conducted as follows. We
small 2014, 10, No. 16, 3275–3282
Acknowledgements
We thank Prof. H. Noji, Drs. K. Tabata and H. Onoe at the University of Tokyo for their helpful discussion. T. Tonooka was supported by Grant-in-Aid for JSPS Fellowship, JSPS, Japan. This work was partly supported by Platform for Dynamic Approaches to Living System, MEXT, Japan.
[1] R. Peters, E. Coutavas, J. P. Siebrasse, J. Struct. Biol. 2002, 140, 268–278. [2] K. Venema, R. Gibrat, J. P. Grouzis, C. Grignon, Biochim. Biophys. Acta 1993, 1146, 87–96. [3] K. Akashi, H. Miyata, H. Itoh, K. Kinosita, Biophys. J. 1996, 71, 3242–3250. [4] D. Basting, M. Lorch, I. Lehner, C. Glaubitz, Faseb J. 2008, 22, 365–373. [5] K. Sumitomo, A. McAllister, Y. Tamba, Y. Kashimura, A. Tanaka, Y. Shinozaki, K. Torimitsu, Biosens. Bioelectron. 2012, 31, 445–450. [6] T. Robinson, P. Kuhn, K. Eyer, P. S. Dittrich, Biomicrofluidics 2013, 7, 044105. [7] H. Sasaki, R. Kawano, T. Osaki, K. Kamiya, S. Takeuchi, Lab Chip 2012, 12, 702–704. [8] A. Kleefen, D. Pedone, C. Grunwald, R. Wei, M. Firnkes, G. Abstreiter, U. Rant, R. Tampe, Nano Lett. 2010, 10, 5080–5087. [9] S. Ota, H. Suzuki, S. Takeuchi, Lab Chip 2011, 11, 2485–2487.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3281
full papers
T. Tonooka et al.
[10] P. V. Ganesan, S. G. Boxer, Proc. Nati. Acad. Sci. USA 2009, 106, 5627–5632. [11] T. Osaki, Y. Watanabe, R. Kawano, H. Sasaki, S. Takeuchi, J. Microelectromech. Syst. 2011, 20, 797–799. [12] O. K. Castell, J. Berridge, M. I. Wallace, Angew. Chem. Int. Ed. 2012, 51, 3134–3138. [13] S. Leptihn, O. K. Castell, B. Cronin, E. H. Lee, L. C. M. Gross, D. P. Marshall, J. R. Thompson, M. Holden, M. I. Wallace, Nat. Protoc. 2013, 8, 1048–1057. [14] Y. L. Cheng, R. J. Bushby, S. D. Evans, P. F. Knowles, R. E. Miles, S. D. Ogier, Langmuir 2001, 17, 1240–1242. [15] H. Bayley, B. Cronin, A. Heron, M. A. Holden, W. L. Hwang, R. Syeda, J. Thompson, M. Wallace, Mol. Biosyst. 2008, 4, 1191–1208. [16] E. Gallucci, S. Micelli, G. Monticelli, Biophys. J. 1996, 71, 824–831. [17] S. Sakakihara, S. Araki, R. Iino, H. Noji, Lab Chip 2010, 10, 3355–3362. [18] K. Funakoshi, H. Suzuki, S. Takeuchi, Anal. Chem. 2006, 78, 8169–8174. [19] A. M. El-Arabi, C. S. Salazar, J. J. Schmidt, Lab Chip 2012, 12, 2409–2413.
3282 www.small-journal.com
[20] R. Kawano, Y. Tsuji, K. Sato, T. Osaki, K. Kamiya, M. Hirano, T. Ide, N. Miki, S. Takeuchi, Sci. Rep. 2013, 3, 1995. [21] H. Suzuki, K. V. Tabata, H. Noji, S. Takeuchi, Biosens. Bioelectron. 2007, 22, 1111–1115. [22] T. Ide, Y. Takeuchi, T. Yanagida, Single Mol. 2002, 3, 33–42. [23] T. Ide, Y. Takeuchi, H. Noji, K. V. Tabata, Langmuir 2010, 26, 8540–8543. [24] A. J. Heron, J. R. Thompson, B. Cronin, H. Bayley, M. I. Wallace, J. Am. Chem. Soc. 2009, 131, 1652–1653. [25] E. N. Ervin, R. J. White, H. S. White, Anal. Chem. 2009, 81, 533–537. [26] L. Z. Song, M. R. Hobaugh, C. Shustak, S. Cheley, H. Bayley, J. E. Gouaux, Science 1996, 274, 1859–1866. [27] D. Branton, D. W. Deamer, A. Marziali, H. Bayley, S. A. Benner, T. Butler, M. Di Ventra, S. Garaj, A. Hibbs, X. H. Huang, S. B. Jovanovich, P. S. Krstic, S. Lindsay, X. S. S. Ling, C. H. Mastrangelo, A. Meller, J. S. Oliver, Y. V. Pershin, J. M. Ramsey, R. Riehn, G. V. Soni, V. Tabard-Cossa, M. Wanunu, M. Wiggin, J. A. Schloss, Nat. Biotechnol. 2008, 26, 1146–1153.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Received: October 24, 2013 Published online: March 10, 2014
small 2014, 10, No. 16, 3275–3282
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport Taishi Tonooka, Koji Sato, Toshihisa Osaki, Ryuji Kawano, and Shoji Takeuchi*
This
paper describes picoliter-sized lipid bilayer chambers and their theoretical model for the rapid detection of membrane transport. To prepare the chambers, semispherical aqueous droplets are patterned on a hydrophilic/hydrophobic substrate and then brought into contact with another aqueous droplet in lipid-dispersed organic solvent, resulting in the formation of the lipid bilayers on the semispherical droplets. The proposed method implements the lipid bilayer chambers with 25-fold higher ratio of lipid membrane area (S) to chamber volume (V) compared to the previous spherical droplet chambers. Using these chambers, we are able to trace the time-course of Ca2+ influx through α-hemolysin pores by a fluorescent indicator. Moreover, we confirm that the detection time of the substrate transport is inversely proportional to the S/V ratio of the developed chambers, which is consistent with the simulation results based on the developed model. Our chambers and model might be useful for rapid functional analyses of membrane transport phenomena.
1. Introduction Fluorescence detection of membrane transport by artificial lipid bilayers has been used for functional analyses of membrane proteins such as transporters that have low transport capacity or the function of weakly-charged molecule transport.[1–4] The detection system used for this purpose generally utilizes fluorescence molecules and a small chamber sealed with a lipid bilayer (lipid bilayer chamber), and traces the change in fluorescence intensity resulting from transportation through the bilayer. For rapid fluorescence detection targeting high-throughput analyses of membrane transport phenomena, the ratio of lipid bilayer area and
T. Tonooka, K. Sato, T. Osaki, S. Takeuchi Institute of Industrial Science The University of Tokyo 4–6–1 Komaba Meguro-ku, Tokyo 153–8505, Japan E-mail: [email protected] T. Osaki, R. Kawano, S. Takeuchi Kanagawa Academy of Science and Technology 3–2–1 Sakado Takatsu-ku, Kawasaki Kanagawa, 213–0012, Japan DOI: 10.1002/smll.201303332 small 2014, 10, No. 16, 3275–3282
chamber volume (S/V) should be as high as possible; it is considered that a larger bilayer area increases the number of transporters incorporated while a smaller chamber volume condenses the transported substrate.[5] Although liposomes are generally used as the lipid bilayer chambers,[2–4,6,7] it is difficult to control the chamber size of these systems, owing to which they are unsuitable for high-throughput quantitative analyses. Two systems have been proposed for use as volumecontrollable chambers: micromachined chambers sealed with lipid bilayers[5,8–11] and droplet chambers in which monodisperse microdroplets come into contact with other droplets in a lipid-dispersed organic solvent.[12,13] Micromachined chambers provide controlled volume reduction, resulting in a variable S/V ratio;[5,8–11] however, it is difficult to achieve reproducible sealing of the chamber.[13–15] On the other hand, droplet chambers provide high reproducibility of the sealing;[13] nevertheless, the S/V ratio is relatively low (S/V < 4 µm2/pL) owing to the large droplet volume (several nanoliters).[12,13] Although scaling down of the lipid bilayer chamber is a straightforward approach for increasing the S/V ratio, handling of floating microdroplets, especially for volumes on the picoliter scale, is difficult that in turn makes the formation of reproducible lipid bilayers difficult. Consequently, droplet chambers with a high S/V ratio have not been realized yet.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
3275
full papers
T. Tonooka et al.
Herein, we propose a picoliter droplet chamber array with improved S/V ratio of up to 100 µm2/pL. To fabricate the droplet chambers, we implement an array of picoliter aqueous droplets in lipid-dispersed organic solvent using a hydrophilic/hydrophobic patterned substrate. The droplets become lipid bilayer chambers by bringing another droplet into contact on top of them. Since the droplets are arrayed and stabilized on the substrate, this method allows reproducible formation of the lipid bilayer chambers on the droplets of picoliter volumes. Additionally, the semispherical shape of this droplet chamber provides a higher S/V ratio than that achievable using previously reported spherical droplet chambers of the same volume. In this paper, we first discuss a theoretical model for fluorescence detection using the lipid bilayer chamber to investigate how the detection time of membrane transport changes according to the S/V ratio of the chamber. We then conduct formation of the picoliter droplet chambers and evaluate the S/V ratio of the chambers. Finally, we examine fluorescence detection of membrane transport by the droplet chambers with different S/V ratios.
2. Results and Discussion 2.1. A theoretical Model for Fluorescence Detection at Various S/V ratios For fluorescence observation of the membrane transport phenomena, the area (S [m2]) of a lipid bilayer membrane is an important factor in addition to the volume (V [m3]) of the chamber that contains the transported substrates.[5] Thus, we evaluated the influence of the S/V ratio on the fluorescence detection of substrate transportation by the following model. The relative fluorescence intensity (ΔF/F0) in the chamber, i.e., the intensity increment (ΔF) relative to the intensity at time zero (F0), was used as the index for fluorescence detection. The intensity increase was due to the binding of fluorescence indicators in the chamber with transported substrates with a dissociation constant (Kd [mol/m3]). We assumed that the number of transporters in the lipid bilayer chamber sigmoidally increased with time as the incorporation of transporters would start after the bilayer formation at time zero and saturate at the maximum surface density.[16] Under this assumption, the number of transporters is given by N (t ) =
N max 1 + exp[ −Gn (t − t s )]
t
∫0 N (t )ST dt V
(2)
where T [mol/s] is the transport rate of a single membrane protein per unit time. Therefore, the time-course of the
3276 www.small-journal.com
( )
is the maximum relative fluorescence intensity where ⎛⎜⎝ ΔFF0 ⎞⎟⎠ max of a fluorescent indicator and Gi is the gain of the sigmoidal curve of relative fluorescence intensity. Figure 1a shows sigmoidal curves plotted based on Equation(1), expressing four different examples of transporter incorporation (from steep (no. 1) to gradual (no. 4)). Figure 1b shows the time courses of the fluorescence intensity normalized by (ΔF/F0)max as 100 [%] with different S/V ratio values. We defined the detection time td as the time that corresponds to the fluorescence detection threshold. Note that by fluorescence observation, substrate transportation will be detected only after a certain intensity threshold is exceeded. Although the threshold for the detection of fluorescence signals is defined based on the background noise, we here simply set the threshold to 15% owing to the difficulty in reliable calculation of the noise level. As shown in Figure 1b, the curve shifts toward the left with an increase in the S/V ratio. These results reconfirm that a larger S/V ratio will enable faster detection of substrate transportation. Figure 1c and 1d show the relationship between td and S/V ratio. The results show that td becomes shorter with increasing S/V ratio and the maximum slope of the curves fits in inverse manner. Note that there is a range where the slope is gradually changing, deviating from the inverse proportion. This range is due to the time-dependent incorporation of transporters. When td exists at the transition state of the incorporation in Figure 1a, increasing S/V ratio to shorten td is not effective because the total number of transporters incorporated drastically decreases with decreasing td; smaller number of transporters causes longer detection time even at the larger S/V ratio. Moreover, we found that the curves mainly shift to the right direction when NmaxT, i.e. transport capacity per unit area of the lipid bilayers, becomes smaller (from 10×NmaxT to 0.01×NmaxT shown in Figure 1c), while the curves mainly shift to the upper direction when the sigmoidal curve of incorporation becomes more gradual (Figure 1d). These characteristics may provide a guideline to design lipid bilayer chambers with various S/V ratios.
(1)
where t [s] is time, Nmax [proteins/m2] is the maximum number of membrane proteins per unit area of a lipid bilayer, ts [s] and Gn are parameters that determine the sigmoidal curve of the membrane protein incorporation per unit area of the lipid bilayer. Using N(t), the concentration of substrates (transported molecules) in a lipid bilayer chamber is given by C (t ) =
relative fluorescence intensity is expressed by a sigmoidal curve as ΔF F0 max ΔF (t ) = (3) F0 1 + exp[ −Gi ln[C (t )/K d ]]
2.2. Formation and Characterization of Lipid Bilayer Chambers on Semispherical Microdroplets Lipid bilayer chambers were formed on a hydrophilic/hydrophobic patterned glass substrate fabricated by standard photolithography (Figure 2a). The fabricated substrate was patterned such that the diameters of hydrophilic circles were 40 and 200 µm, and the spaces between the patterns were 40 and 200 µm, respectively (Figure 2b). First, a phospholipid dispersed in n-hexadecane was introduced on the hydrophilic/ hydrophobic patterned glass substrate (Figure 2c1). Second, a semispherical droplet array was formed on the circular hydrophilic areas by spotting a buffer solution (Figure 2c2).
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 16, 3275–3282
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport
(a)
(b) 100
N(t)/Nmax (%)
80 60 no. 1
(ts:50,Gn:0.12)
40
no. 2
(ts:100,Gn:0.06)
no. 3
20
(ts:200,Gn:0.03)
no. 4
(ΔF/F0)/(ΔF/F0)max (%)
100
(ts:400,Gn:0.015)
0 0
300 600 Time (s)
(c) 106
(d)
105
60 40
td (s)
103
S/V=4
600 1200 Time (s) td
1800
no. 1 no. 2 no. 3 no. 4
105 104 103 102
10
10 1
Detection threshold
106
102
1
S/V=10 (μm2/pL)
20
td (s)
104
S/V=30
0 0
900
0.01×NmaxT 0.1×NmaxT 1×NmaxT 10×NmaxT
S/V=100
80
10
102 103 104 105 106
S/V (μm2/pL)
1
1
10
102 103 104 105 106
S/V (μm2/pL)
Figure 1. Simulation results of fluorescence detection of molecular transport through transporters by using a lipid bilayer chamber. (a) Time courses of transporter incorporation into a lipid bilayer; the curves represent examples of steep (no.1) to gradual (no.4) incorporation. The parameters are shown in the legend. (b) Time course of the relative fluorescence intensities calculated based on the curve no.1 in Figure 1a with different S/V ratios. Detection threshold was set at 15%. (c) Detection time vs. S/V ratio calculated based on the curve no.1 in Figure 1a with different NmaxT values. (d) Detection time vs. S/V ratio with different types of transporter incorporation corresponding to no.1–4 in Figure 1a. The parameters used in the solid curves were NmaxT [mol/s·m2] = 5.0 × 10−8, Gi = 2.34, Kd [mol/m3] = 0.662. Gi and Kd were experimentally estimated (Supporting information 1).
Lipid monolayers autonomously covered the droplet surfaces because of the amphiphilic characteristics of lipid molecules. Third, an upper aqueous droplet was infused on and brought into contact with the droplet array on the substrate (Figure 2c3). Since a lipid monolayer covered the upper droplet, a thick lipid/hexadecane/lipid membrane was formed when the droplets came into contact. The thick membrane became a lipid bilayer within several minutes, forming a lipid bilayer chamber. The lipid bilayer formation started when a part of the two lipid monolayers[18–20] on the patterned and upper droplets came into contact with each other; the area of the lipid bilayer gradually increased and finally reached a steady state (Figure S2).[21] Figure 3a shows the lipid bilayer chamber array: Thirty-six lipid bilayer chambers were observed in the field of view. The boundary of a lipid bilayer on a semispherical droplet is visible in a closeup image (Figure 3b). It should be noted that the patterned semispherical droplets were always smaller than the hydrophilic areas on the substrate, probably because of the balance of interfacial tension between the patterned aqueous droplet, small 2014, 10, No. 16, 3275–3282
lipid solution, and substrate surface. For precise control of the droplet diameter, this balance of the tension should be adjusted by surface modification of the substrate or contents of the solutions. Lipid bilayer chambers were then characterized by the volume of the chamber (V), the area of the lipid bilayer (S), and the S/V ratio (Table 1). Figure 3c shows cross-sectional images of a lipid bilayer chamber formed on a substrate with 40-µm-diameter hydrophilic patterns (named as 40-µm chamber). Interestingly, the formed lipid bilayer membranes had slight curvature along the surface of the patterned semispherical droplets, as shown in the longitudinal image. We considered that the curvature was due to the difference between the internal pressure in the upper and patterned droplets. The average volume of the 40-µm chambers was hundred-fold smaller than that of the 200-µm chambers, while the membrane area of the 40-µm chambers became 35 times smaller than that of the 200-µm chambers. The variations in both chamber volume and area were relatively small, as described in Table 1. Accordingly, the S/V ratio obtained
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3277
full papers
T. Tonooka et al.
(a)
(b)
(i) Cytop spincoat, Al deposition, S1818 spincoat
Hydrophilic area
(iv) Cytop etching
Hydrophobic area
(ii) S1818 patterning (v) S1818, Al removal
(iii) Al etching Glass Cytop
(c1)
S1818 Al
(c2)
Capillary Semispherical Lipid solution aqueous (hexadecane) Hydrophobic droplet
Hydrophilic
Lipid
Lipid monolayer
Capillary
Upper aqueous droplet
Lipid bilayer
Semispherical microdroplet
Hexadecane
(c3)
Lipid bilayer chamber
Figure 2. Formation of lipid bilayer chambers on a hydrophilic/hydrophobic patterned substrate. (a) Fabrication process for the hydrophilic/ hydrophobic patterned substrate. (b) Microscopic image of the substrate. The scale bar represents 100 µm. (c) Formation procedure for lipid bilayer chambers: (c1) filling with lipid-dispersed organic solvent, (c2) patterning of an aqueous solution on the substrate provides a picolitersized semispherical droplet array, and (c3) contact of an upper aqueous droplet on the semispherical droplets, resulting in lipid bilayer formation on the semispherical droplets.
for the 40-µm chambers was 100 µm2/pL, and three times larger than that of the 200-µm chambers. Note that the droplet chambers commonly used in previous works showed lower S/V ratio of up to 4 µm2/pL; these droplet chambers consist of droplets with volumes as low as 10 nL and diameters of a few hundred micrometers.[12,13] Our droplet patterning method using a hydrophilic/hydrophobic pattern is simple, and the implementation of a reproducible methodology scales down the chamber volume to a few picoliters. Additionally, the easily observable bilayer-membrane area is another advantage of the semispherical chamber design, since it potentially allows microscopic observations of membrane proteins on the lipid bilayer membrane directly.[22–24] 2.3. Fluorescence Detection of Ca2+ Transport Through Membrane Proteins The Ca2+ influx through an α-hemolysin nanopore [25–27] was monitored by the developed lipid bilayer chambers to validate our method for fluorescence detection of membrane transport. The lipid bilayer chambers were formed under the following conditions. The upper aqueous droplet contained Ca2+ with/without α-hemolysin, while the patterned semispherical droplets contained a Ca2+ fluorescent indicator, Fluo-4. Figure 4a is representative time-lapse images of a lipid
3278 www.small-journal.com
bilayer chamber array showing the fluorescence intensity increase due to Ca2+ influx through α-hemolysin pores. In this experiment, 30 lipid bilayer chambers out of 36 spots (83%) remained after the upper aqueous droplet came into contact, while the rest were fused with the upper droplet via membrane rupture. The fluorescence intensities of all 30 chambers increased within 15 min. Figure 4b shows the time courses of the fluorescence intensities in the individual lipid bilayer chambers under the initial conditions of including/excluding α-hemolysin. We confirmed that the relative fluorescence intensities (ΔF/F0) increased in all lipid bilayer chambers only with the addition of α-hemolysin; the convex increase followed by plateau is consistent with the model derived in section 2.1. Without α-hemolysin, the change in the relative fluorescence intensity was negligible throughout the experiment. These results demonstrated that the developed lipid bilayer chambers were useful for introducing α-hemolysin nanopores and for observing ion transports through the pores.
2.4. Comparison Between 40- and 200-μm Chambers for the Fluorescence Detection We compared the detection time at two S/V ratios of the 40and 200-µm chambers using Ca2+ transport experiments with
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 16, 3275–3282
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport
to increase steeply with Ca2+ influx and reached the detection threshold within a shorter period than in the case of the 200-µm chambers (Figure 5b). As the S/V ratio becomes 3-fold higher from the 200-µm chamber to the 40-µm chamber, the detection time reduced by about 3-fold. This result also agreed with the simulation results that showed the detection time was inversely proportional to the S/V ratios (Figure 1c). We further performed Ca2+ imaging with a lower number of incorporated nanopores using the two chambers (Figure 5c). As a result, we confirmed that 59% of the 40-µm chambers exceeded the detection threshold within 15 min; on the other hand, no 200-µm chamber reached the detection threshold throughout the experiment. A higher S/V ratio was advantageous for detecting minimal membrane transport phenomena within a given period. These results confirmed that increasing the S/V ratio for a lipid bilayer chamber system not only allows rapid detection but also promotes high sensitivity detection of membrane transport activities.
3. Conclusion
Figure 3. Lipid bilayer chamber array formed on a substrate with the 40-µm-diameter pattern. (a) Array of lipid bilayer chambers. (b) Close-up image of a lipid bilayer chamber. (c) Confocal fluorescence image of (b). Green fluorescence stained by calcein shows a semispherical aqueous droplet; magenta fluorescence stained by DiI shows the lipid-dispersed organic solvent. (d) Longitudinal image at the white dotted line in (c). The white dotted line in (d) indicates the height of the transverse image in (c). The black and white scale bars are 100 and 10 µm, respectively.
α-hemolysin pores. A representative result of Ca2+ imaging for each chamber is presented in Figure 5a. The relative fluorescence intensity increased in a sigmoidal manner, which agrees with the model shown in Figure 1b. We determined the detection threshold at F0+10σF0; the average (F0) and standard deviation (σF0) of background intensity were estimated from the initial 30 s of recording in each chamber. The fluorescence intensities of the 40-µm chambers tended
In this study, we developed lipid bilayer chambers with increased S/V ratios of up to 100 µm2/pL by using picolitersized semispherical droplets. The developed chambers successfully detected Ca2+ influx through the α-hemolysin pores by a fluorescent indicator. The increase of the fluorescence in the chambers agreed with the theoretical model. By comparing two chambers having different S/V ratios, we confirmed that the larger S/V ratio provides rapid detections. The model we derived would allow us to understand how the S/V ratio affects the detection time with various properties of membrane proteins including transport capacity and incorporation manner of proteins. We believe that our picoliter-sized droplet chambers are useful for rapid screening of membrane transport such as drug resistance transporters on tumor cells.
4. Experimental Section Materials and Chemicals: The substrate for obtaining a lipid bilayer chamber array was fabricated from a glass slip (thickness of 150–200 µm; Matsunami Glass Ind., Osaka, Japan) and CytopTM (perfluoro polymer; CTL-809M; Asahi Glass, Tokyo, Japan). S1818 (positive photoresist; Shipley, MA, USA), aluminum etchant (Wako Pure Chemical Industries, Tokyo, Japan), and NMD (S1818 developer; Tokyo Ohka Kogyo, Kawasaki, Japan) were used in the fabrication process. Diphytanoylphosphatidylcholine (DPhPC; Avanti Polar Lipids, AL, USA), asolectin (Sigma-Aldrich, MO, USA), and n-hexadecane (Kanto Chemical, Tokyo, Japan) were used for pre-
Table 1. Properties of 40- and 200-µm lipid bilayer chambers. The S/V ratios were significantly different (p < 0.01, Mann-Whitney U test). n
Lipid bilayer area (S)
Chamber volume (V)
S/V ratio
40-µm chamber
9
136 ± 24 µm2
1.4 ± 0.1 pL
100 ± 10 µm2/pL
200-µm chamber
4
4760 ± 1257 µm2
158 ± 25 pL
30 ± 3 µm2/pL
small 2014, 10, No. 16, 3275–3282
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3279
full papers
T. Tonooka et al.
Device Fabrication: The hydrophilic/ hydrophobic pattern was fabricated on a 0 min 2 min glass substrate by a common photolithography process.[17] A cleaned glass substrate was used as the hydrophilic surface while the Cytop pattern was used as the hydrophobic part on the glass. The detailed process is as follows. First, a glass substrate was rinsed with acetone and 2-propanol. Cytop (9 wt%) was then spin-coated on the glass Lipid bilayer chamber substrate at 3000 rpm (MS-A100; Mikasa, Tokyo, Japan). Coated Cytop was then baked at 50 °C for 30 min and then at 180 °C for 1 h to evaporate the solvent. A thin Al layer was deposited on the substrate by using a high-vacuum evaporator (SVC-700 TM; Sanyu 3 min 5 min Electron, Tokyo, Japan). On the Al layer, a positive photoresist was spin-coated. An array of 40-µm or 200-µm circles was patterned on the photoresist by UV photolithography (PEM-6M; Union Optical, Tokyo, Japan). The pattern on the resist and Al was developed using NMD and Al etchant. Finally, Cytop was etched by applying O2 plasma (FA-1; Samco, Kyoto, Japan) and a circular pattern was obtained on the Cytop layer. The remaining S1818 and Al were removed using acetone and etchant. The substrate was heated at 150 °C for 1 h to reactivate the hydropho(b) 35 bicity of Cytop, and immersed in a Piranha α-hemolysin (+) solution at 60 °C for 1 min to enhance the 30 α-hemolysin (-) hydrophilicity of the glass surface. The 25 obtained hydrophilic/hydrophobic pattern is shown in Figure 2b. The thickness of Cytop 20 was 1.0 µm, measured by a stylus profilom15 eter (Dektak 6M; Veeco, NY, USA). Formation of Lipid Bilayer Chamber: The 10 formation process for lipid bilayer cham5 bers consists of the following three steps. In the first step (Figure 2c1), about 100 µL 0 of n-hexadecane was poured on the hydro-5 philic/hydrophobic patterned substrate. In 0 1 2 3 4 5 6 7 8 the second step (Figure 2c2), droplet patTime (min) terning was conducted using a glass capilFigure 4. Ca2+ imaging of Ca2+ influx through α-hemolysin pores using the 40-µm lipid bilayer lary (inner diameter ranged from 200 µm to chamber array. (a) Fluorescence time-lapse images of lipid bilayer chambers with 100 nM α-hemolysin. Scale bar is 100 µm. (b) The relative fluorescence intensity obtained from Ca2+ 1 mm; Narishige, Tokyo, Japan) and a microimaging in each lipid bilayer chamber over time. Each line corresponds to the conditions with pump (PicoPipet; Altair, Kanagawa, Japan); the capillary was filled with a buffer solution the presence (black lines, n = 30) or absence (gray lines, n = 27) of 100 nM α-hemolysin. (10 mM HEPES, pH 7.4 NaOH) and mounted on a micromanipulator (Eppendorf, NY, paring the lipid bilayer chambers. Calcein (Sigma-Aldrich) and USA). The buffer solution was ejected from the capillary by using 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlo- the pump while the capillary tip traced the pattern on the subrate (DiI; Invitrogen, CA, USA), α-hemolysin from Stapylococcus strate. This procedure automatically patterned the buffer on the aureus (Sigma-Aldrich), and Fluo-4 (Pentapotassium salt; Ca2+ flu- hydrophilic circle area of the substrate, and produced an array orescent indicator; Dojindo Laboratories, Kumamoto, Japan) were of semispherical microdroplets. In the third step (Figure 2c3), a used for the characterization of the lipid bilayer chamber array. buffer solution for the upper aqueous droplet (10 mM HEPES, pH Other chemicals were purchased from Kanto Chemical or Wako 7.4 NaOH) was ejected into n-hexadecane in the same manner, Pure Chemical Ind. (Tokyo, Japan). Milli-Q water (Millipore, MA, and brought into contact with the patterned semispherical droplets on the substrate by controlling the manipulator. Owing to USA) was used as deionized water (DI water).
ΔF/F0
(a)
3280 www.small-journal.com
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 16, 3275–3282
Lipid Bilayers on a Picoliter Microdroplet Array for Rapid Fluorescence Detection of Membrane Transport
(a)
(b) 25
8
40-μm chamber
20 Time (min)
6
ΔF/F0
15 10 Threshold (40-μm)
5 0
200-μm chamber
3
6 9 12 Time (min)
5 4 3 2 1
Threshold (200-μm)
0
*
7
15
0
S/V: 100 μm2/pL (40-μm chamber)
S/V: 30 μm2/pL (200-μm chamber)
(c) 1.8 1.5 40-μm chamber
ΔF/F0
1.2 0.9 0.6 0.3
Threshold (40-μm) Threshold (200-μm)
0
Supporting Information
200-μm chamber
0
used a 10 mM HEPES buffer solution (pH 7.4 NaOH) containing 100 µM Fluo-4 and 500 µM EDTA for patterned semispherical aqueous droplets, and the same 10 mM HEPES buffer solution (pH 7.4) containing 100 mM CaCl2 for an upper aqueous droplet. To make the osmotic pressure equal, 150 mM of KCl was added to the semispherical droplets. EDTA was used as a chelator to suppress background Ca2+ in the semispherical droplets. The fluorescence images were taken by an inverted microscope (IX71; Olympus, Tokyo, Japan). The time course of fluorescence intensity in the droplets was monitored using Aquacosmos software and an EM-CCD camera (C9100; Hamamatsu Photonics, Shizuoka, Japan). The average fluorescence intensities inside the droplets were evaluated using Image J. The fluorescence intensities (F) were normalized by the intensities right after lipid bilayer formation (F0) as (F – F0)/ F0 (denoted as ΔF/F0).
3
6 9 12 Time (min)
15
Supporting Information is available from the Wiley Online Library or from the author.
Figure 5. Comparison between 40- and 200-µm lipid bilayer chambers for fluorescence detection of Ca2+ influx through the α-hemolysin pores. (a) Representative time-courses of fluorescence intensity in the 40- and 200-µm chambers with 100 nM α-hemolysin. The detection thresholds, determined as F0+10σF0, are presented in the graph. (b) Detection times obtained in the 40-µm chamber (n = 30) and 200-µm chamber (n = 4). Error bars: standard deviation; *: p < 0.05 (Student’s t-test). (c) Representative time-courses of fluorescence intensity in the 40- and 200-µm lipid bilayer chambers with 10 nM of α-hemolysin.
the pressure from the upper aqueous droplet, the membrane at the interface between the droplets was transformed into a lipid bilayer (Figure 2c3).[19] Estimation of Volume and Membrane Area of Lipid Bilayer Chambers: The volume of the lipid bilayer chambers was determined as follows. A series of cross-sectional images of the lipid bilayer chambers was obtained using an inverted confocal microscope (LSM780, Carl Zeiss Microscopy, Jena, Germany) together with a bright field image. Here, the chambers and n-hexadecane were visualized with calcein (green) and DiI (magenta), respectively, as shown in Figure 3c. The transverse confocal images of the chambers were taken in increments of 0.5 µm step in perpendicular direction. The images of calcein were binarized by the particular threshold using an image processing software (Image J; NIH, USA), and areas with intensity higher than the threshold were integrated along the perpendicular direction, defining the volume of the lipid bilayer chamber. The lipid bilayer areas were measured from the bright filed images by using Image J. Ca2+Imaging of the Lipid Bilayer Chambers: Ca2+ imaging of the lipid bilayer chambers was conducted as follows. We
small 2014, 10, No. 16, 3275–3282
Acknowledgements
We thank Prof. H. Noji, Drs. K. Tabata and H. Onoe at the University of Tokyo for their helpful discussion. T. Tonooka was supported by Grant-in-Aid for JSPS Fellowship, JSPS, Japan. This work was partly supported by Platform for Dynamic Approaches to Living System, MEXT, Japan.
[1] R. Peters, E. Coutavas, J. P. Siebrasse, J. Struct. Biol. 2002, 140, 268–278. [2] K. Venema, R. Gibrat, J. P. Grouzis, C. Grignon, Biochim. Biophys. Acta 1993, 1146, 87–96. [3] K. Akashi, H. Miyata, H. Itoh, K. Kinosita, Biophys. J. 1996, 71, 3242–3250. [4] D. Basting, M. Lorch, I. Lehner, C. Glaubitz, Faseb J. 2008, 22, 365–373. [5] K. Sumitomo, A. McAllister, Y. Tamba, Y. Kashimura, A. Tanaka, Y. Shinozaki, K. Torimitsu, Biosens. Bioelectron. 2012, 31, 445–450. [6] T. Robinson, P. Kuhn, K. Eyer, P. S. Dittrich, Biomicrofluidics 2013, 7, 044105. [7] H. Sasaki, R. Kawano, T. Osaki, K. Kamiya, S. Takeuchi, Lab Chip 2012, 12, 702–704. [8] A. Kleefen, D. Pedone, C. Grunwald, R. Wei, M. Firnkes, G. Abstreiter, U. Rant, R. Tampe, Nano Lett. 2010, 10, 5080–5087. [9] S. Ota, H. Suzuki, S. Takeuchi, Lab Chip 2011, 11, 2485–2487.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3281
full papers
T. Tonooka et al.
[10] P. V. Ganesan, S. G. Boxer, Proc. Nati. Acad. Sci. USA 2009, 106, 5627–5632. [11] T. Osaki, Y. Watanabe, R. Kawano, H. Sasaki, S. Takeuchi, J. Microelectromech. Syst. 2011, 20, 797–799. [12] O. K. Castell, J. Berridge, M. I. Wallace, Angew. Chem. Int. Ed. 2012, 51, 3134–3138. [13] S. Leptihn, O. K. Castell, B. Cronin, E. H. Lee, L. C. M. Gross, D. P. Marshall, J. R. Thompson, M. Holden, M. I. Wallace, Nat. Protoc. 2013, 8, 1048–1057. [14] Y. L. Cheng, R. J. Bushby, S. D. Evans, P. F. Knowles, R. E. Miles, S. D. Ogier, Langmuir 2001, 17, 1240–1242. [15] H. Bayley, B. Cronin, A. Heron, M. A. Holden, W. L. Hwang, R. Syeda, J. Thompson, M. Wallace, Mol. Biosyst. 2008, 4, 1191–1208. [16] E. Gallucci, S. Micelli, G. Monticelli, Biophys. J. 1996, 71, 824–831. [17] S. Sakakihara, S. Araki, R. Iino, H. Noji, Lab Chip 2010, 10, 3355–3362. [18] K. Funakoshi, H. Suzuki, S. Takeuchi, Anal. Chem. 2006, 78, 8169–8174. [19] A. M. El-Arabi, C. S. Salazar, J. J. Schmidt, Lab Chip 2012, 12, 2409–2413.
3282 www.small-journal.com
[20] R. Kawano, Y. Tsuji, K. Sato, T. Osaki, K. Kamiya, M. Hirano, T. Ide, N. Miki, S. Takeuchi, Sci. Rep. 2013, 3, 1995. [21] H. Suzuki, K. V. Tabata, H. Noji, S. Takeuchi, Biosens. Bioelectron. 2007, 22, 1111–1115. [22] T. Ide, Y. Takeuchi, T. Yanagida, Single Mol. 2002, 3, 33–42. [23] T. Ide, Y. Takeuchi, H. Noji, K. V. Tabata, Langmuir 2010, 26, 8540–8543. [24] A. J. Heron, J. R. Thompson, B. Cronin, H. Bayley, M. I. Wallace, J. Am. Chem. Soc. 2009, 131, 1652–1653. [25] E. N. Ervin, R. J. White, H. S. White, Anal. Chem. 2009, 81, 533–537. [26] L. Z. Song, M. R. Hobaugh, C. Shustak, S. Cheley, H. Bayley, J. E. Gouaux, Science 1996, 274, 1859–1866. [27] D. Branton, D. W. Deamer, A. Marziali, H. Bayley, S. A. Benner, T. Butler, M. Di Ventra, S. Garaj, A. Hibbs, X. H. Huang, S. B. Jovanovich, P. S. Krstic, S. Lindsay, X. S. S. Ling, C. H. Mastrangelo, A. Meller, J. S. Oliver, Y. V. Pershin, J. M. Ramsey, R. Riehn, G. V. Soni, V. Tabard-Cossa, M. Wanunu, M. Wiggin, J. A. Schloss, Nat. Biotechnol. 2008, 26, 1146–1153.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Received: October 24, 2013 Published online: March 10, 2014
small 2014, 10, No. 16, 3275–3282
