sign in sign up
<<
Home , Septin

Langmuir

This document is confidential and is proprietary to the American Chemical Society and its authors. Do not copy or disclose without written permission. If you have received this item in error, notify the sender and delete all copies.

Septin Interferes with the Temperature-Dependent Domain Formation and Disappearance of Lipid Bilayer Membranes

Journal: Langmuir

Manuscript ID la-2016-03452g.R1

Manuscript Type: Article

Date Submitted by the Author: 03-Nov-2016

Complete List of Authors: Yamada, Shunsuke; Nagoya University, Biological Science Isogai, Takumi; Department of Materials, Physics and Energy Engineering, Graduate School of Engineering, Nagoya University, Nagoya 464-8602, Japan. Tero, Ryugo; Toyohashi University of Technology, Department of Environmental and Life Sciences Tanaka-Takiguchi, Yohko; Nagoya Daigaku, Structural Biology Research Center Ujihara, Toru; Nagoya University, Department of Materials Science and Engineering, Kinoshita, Makoto; Division of Biological Science, Graduate School of Science, Nagoya University Takiguchi, Kingo; Nagoya University, Biological Science

ACS Paragon Plus Environment Page 1 of 39 Langmuir

1 2 3 4 5 6 7 Interferes with the Temperature-Dependent 8 9 10 11 12 Domain Formation and Disappearance of Lipid 13 14 15 16 Bilayer Membranes 17 18 19 20 21 Shunsuke Yamada,† Takumi Isogai,‡ Ryugo Tero,§ Yohko Tanaka-Takiguchi, Toru Ujihara,‡ 22 23 24 Makoto Kinoshita,† and Kingo Takiguchi,†,,* 25 26 27 †: Division of Biological Science, Graduate School of Science, and ‡: Department of Materials, 28 29 30 Physics and Energy Engineering, Graduate School of Engineering, Nagoya University, Nagoya 31 32 464-8602, Japan. §: The Electronics-Inspired Interdisciplinary Research Institute, Toyohashi 33 34 35 University of Technology, Toyohashi 441-8580, Japan. : Structural Biology Research Center, 36 37 Nagoya University, Nagoya 464-8601, Japan. 38 39 40 41 42 43 44 KEYWORDS 45 46 47 Membrane domain formation, Partitioning of lipid membrane, Septin, Supported lipid bilayer. 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment 1

Langmuir Page 2 of 39

1 2 3 4 ABSTRACT 5 6 7 8 Domain formation or compartmentalization in a lipid bilayer membrane has been thought to take 9 10 place dynamically in membranes and play important roles in the spatiotemporal regulation of 11 12 13 their physiological functions. In addition, the membrane skeleton, which is a assembly 14 15 beneath the , also regulates the properties, as well as the morphology, of 16 17 18 membranes, due to its role as a diffusion barrier against constitutive molecules of the membrane 19 20 or as a scaffold for physiological reactions. Therefore, it is important to study the relationship 21 22 between lipid bilayer membranes and that form the membrane skeleton. Among 23 24 25 cytoskeletal systems, septin is unique because it forms arrays on liposomes that contain 26 27 phosphoinositides, and this property is thought to contribute to the formation of the annulus in 28 29 . In this study, a supported lipid bilayer (SLB) was used to investigate the effect 30 31 32 of septin on lipid bilayers because SLBs rather than liposomes are suitable for observation of the 33 34 membrane domains formed. We found that SLBs containing phosphatidylinositol (PI) reversibly 35 36 37 form domains by decreasing the temperature, and that septin affects both the formation and the 38 39 disappearance of the cooling-induced domain. Septin inhibits the growth of cooling-induced 40 41 domains during decreases in temperature, and inhibits the dispersion and the disappearance of 42 43 44 those domains during increases in temperature. These results indicate that septin complexes, i.e. 45 46 filaments or oligomers assembling on the surface of lipid bilayer membranes, can regulate the 47 48 dynamics of domain formation, via their behavior as an anchor for PI molecules. 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment 2

Page 3 of 39 Langmuir

1 2 3 1. Introduction 4 5 6 1.1. Membrane Domains 7 8 The basic structure of the cell membrane is a lipid bilayer, and a number of proteins 9 10 11 indispensable for the activities of cells are incorporated. In the that was 12 13 considered initially, lipids and proteins were thought to diffuse freely within the membrane. But 14 15 currently, the model wherein specific lipids and proteins assemble and form membrane domains, 16 17 18 where the compositions and characteristics are different from the surrounding areas, is accepted. 19 20 1–4 As the representative examples, lipid rafts and caveolae structures, micro-domains rich in 21 22 cholesterol, sphingomyelin or glycolipids, are membrane domains that have attracted much 23 24 25 attention. Caveolae are relatively stable structures formed by the interaction between a 26 27 membrane and caveolin, a membrane-lining protein. In addition, recent studies revealed the 28 29 30 physiological significance of membrane domains for the interactions between proteins and lipid 31 32 membranes, the formation of protein assemblies such as membrane skeletons, the localization of 33 34 proteins at membrane sites where the proteins to exhibit the function, and the membrane vesicle 35

36 5–7 37 traffic in cells. However, the mechanism of formation and maintenance of membrane domains 38 39 where proteins are involved is still unclear. 40 41 42 43 44 1.2. Supported Lipid Bilayers (SLBs) 45 46 The supported lipid bilayer (SLB) was used as a model of a lipid bilayer membrane in this 47 48 8–13 49 study. The basic structure of the SLB is identical to that of membrane vesicles that have been 50 51 used in many studies as the simplest model of the lipid bilayer membrane. The SLB exists at the 52 53 interface between a solid substrate and an aqueous solution, and has a flat shape and maintains 54 55 56 lateral diffusion properties, because about 1 nm thick water layer exists between the SLB and the 57 58 59 60 ACS Paragon Plus Environment 3

Langmuir Page 4 of 39

1 2 3 solid substrate. Therefore, SLBs have the ability to cause phase separation in the same way as 4 5 6 other lipid membranes, and that ability can be regulated by varying the conditions. SLBs have 7 8 recently been noted because they are suitable for analyzing the dynamics of target membrane- 9 10 11 constituting molecules and investigating the mechanisms of reactions that occur in the membrane. 12 13 Utilizing SLBs, the phase separation concerning membrane domains or rafts and the relationship 14 15 between rafts and membrane proteins have been studied. 14–18 To observe SLBs, fluorescence 16 17 18 microscopy, including confocal or total internal reflection microscopy, and atomic force 19 20 microscopy (AFM) are frequently used. 21 22 It is already known that domains are inducible in SLBs prepared from synthetic lipids, e.g., 23 24 25 dimyristoylphosphatidylcholine (DMPC) and dioleoylphosphatidylcholine (DOPC), by changing 26 27 the temperature. In those SLBs, the two phospholipids are completely mixed at 13 or higher 28 29 30 temperatures, whereas the domains emerge at around 12.5 and they grow in size with 31 32 8 33 decreasing temperature. In this study, we also found that SLBs made from phosphatidylcholine 34 35 (PC) and phosphatidylinositol (PI), phospholipids obtained from natural sources but not synthetic 36 37 ones, show the formation and disappearance of PI-free domains by changing the temperature. 38 39 40 41 42 1.3. Septin, A Possible Regulator of the Membrane Domain 43 44 can be purified from diverse types of eukaryotic cells, and they are regarded as the 45

46 19–21 47 fourth in addition to , and intermediate filaments. Septin 48 49 filaments polymerized can further assemble into diverse higher-order structures in vivo and in 50 51 52 vitro, such as linear/circular/spiral bundles or gauze-like arrays, and have been implicated in a 53 20, 22–29 54 variety of organization processes of biological membranes. However, the detailed 55 56 mechanism that septins exert the physiological roles remains unknown. 57 58 59 60 ACS Paragon Plus Environment 4

Page 5 of 39 Langmuir

1 2 3 Septin is unique compared to the other because it interacts directly with 4 5 30 6 negatively charged lipid membranes via a cluster of basic amino acid residues at its surface. 7 8 Septins forming arrays beneath the cell membrane are estimated to give rigidity to the 9 10 22 11 and to become a scaffold for membrane proteins, although the details are unclear. In addition, 12 13 our previous study revealed that, when septin is added to giant liposomes, 1) the 14 15 polymerization of septin filaments is stimulated, 2) the septin filaments spontaneously assemble 16 17 18 into two-dimensional arrays on the liposome surface (Figure 1), and 3) liposomes show robust 19 20 tubulation as a result of the array formation. 30 For these reactions to occur, an acidic 21 22 phospholipid, such as PI or PI-4,5-bisphosphate (PIP ), has to be contained in the membrane. 23 2 24 25 The two-dimensional arrays of septin formed on the membrane surface are also considered to 26 27 serve as a diffusion barrier. 28 29 30 The effects of the interaction with a lipid membrane on the nature of septin have been reported 31 32 by studies using PIP2-containing SLBs or lipid monolayers that were prepared on a flat substrate. 33 34 31, 32 On the other hand, the effect of the interaction with septin on the nature of a lipid membrane, 35 36 37 especially the partitioning of the membrane such as the domain formation, has not yet been 38 39 revealed, except that an in silico study suggested that septin filaments work as a membrane 40 41 diffusion barrier. 33 In in vitro experimental studies, mainly the of membranes has 42 43 30, 34 44 been studied regarding the impact of septin. In this study, in order to use SLBs placed at low 45 46 temperatures as a model for membrane domains, we prepared SLBs from PC and PI, and then 47 48 49 cooled them to cause the domain formation. By adding septin to these SLBs before and after the 50 51 membrane domains are formed, we investigated whether septin can regulate the membrane 52 53 domain formation and affect the fluidity of the lipid bilayer membrane. 54 55 56 57 58 59 60 ACS Paragon Plus Environment 5

Langmuir Page 6 of 39

1 2 3 2. Materials and Methods 4 5 6 2.1. Regents 7 8 Phosphatidylinositol (PI) isolated from bovine liver was purchased from Avanti Polar Lipids 9 10 11 (Alabaster, AL USA). Fluorescent-labeled annexin V and fluorescent-labeled lipid, N-(Texas 12 13 Red sulfonyl)-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (TR-DHPE), were 14 15 purchased from Takara Bio (Kusatsu, Shiga, Japan). A fluorescent-labeled lipid, 2-(4,4-difluoro- 16 17 18 5,7-diphenyl-4-bora-3a,4a-diaza-s-indacene-3-pentanoyl)-1-hexadecanoyl-sn-glycero-3- 19 20 phosphocholine (BODIPY-PC) was purchased from Invitrogen (Waltham, MA USA). 21 22 Phosphatidylcholine (PC) and phosphatidylglycerol (PG) isolated from egg yolk, 23 24 25 phosphatidylserine (PS) isolated from bovine brain, 1,2-dioleoyl-sn-glycero-3- 26 27 phosphatidylcholine (DOPC), and 1,2-dimyristoyl-sn-glycero-3-phosphatidylserine (DMPS) 28 29 30 were purchased from Sigma (St. Louis, MO USA). The protein inhibitor cocktail (complete and 31 32 EDTA-free tablet) used for preparing the septin fraction was purchased from Roche Diagnostics 33 34 (Tokyo, Japan). Other chemicals were of analytical grade, and were used without further 35 36 37 purification. 38 39 40 41 2.2. Supported Lipid Bilayers (SLBs) 42 43 44 In this study, SLBs were prepared by the method of vesicle fusion, because that method has the 45 46 advantage of producing uniform SLBs throughout the solid substrates, regardless of their three- 47 48 8–13, 35 49 dimensional shape and size. Each vesicle solution required to prepare SLBs was obtained 50 51 as follows. 36 Phospholipids were dissolved in chloroform/methanol solution (98:2, vol/vol) at 52 53 concentrations of 100 μM for fluorescent-labeled lipids (BODIPY-PC and TR-DHPE) and 10 54 55 56 mM for the other lipids. They were stored away from light at 4 until use. A total 300 μL 10 57 58 59 60 ACS Paragon Plus Environment 6

Page 7 of 39 Langmuir

1 2 3 mM lipids (PC and PI in this study usually) and 30 μL 100 μM BODIPY-PC or TR-DHPE were 4 5 6 mixed into pre-washed vial tubes (10 mL). The lipid composition of SLBs was PC, PI, and 7 8 BODIPY-PC (1:1:0.002, mol/mol), or PC, PI, and TR-DHPE (4:1:0.005, mol/mol). It is noted 9 10 11 here that, unless otherwise specified, SLBs used in these experiments were the first type. To 12 13 prepare a lipid film, the solvent was vaporized with nitrogen gas flow at room temperature, and 14 15 the film was put under vacuum for longer than 6 hr. To the vacuum-dried lipid film, 5 mL buffer 16 17 18 L (25 mM Tris-HCl, pH 8.0, 50 mM KCl, 1 mM DTT), that had been pre-warmed at 45, was 19 20 added and stirred for 1 hr with a vortex mixer at 45 after nitrogen gas was injected and the 21 22 23 tube was sealed. The solution was frozen using liquid nitrogen, and warmed again to 45. This 24 25 26 freeze-thaw procedure was repeated five times. The resulting solution was passed 10 times 27 28 through a polycarbonate filter (Cyclopore Polycarbonate Membranes, 0.1 μm pore size, 29 30 Whatman, GE Healthcare, Little Chalfont, UK). It was then diluted with the same volume of 31 32 33 buffer L and was sonicated at 28 kHz for 5 hr or more (58 hr). This long-time sonication is 34 35 required to prevent the formation of a stable layer of absorbing vesicles that may be due to the 36 37 38 failure to completely break the vesicles. Each vesicle suspension prepared was stored away from 39 40 light at 4 after the nitrogen gas injection until use. 41 42 43 Planar lipid bilayer membranes were formed on a mica substrate as follows. The vesicle 44 45 suspensions obtained as noted above were diluted with the same volume of buffer M10 (25 mM 46 47 48 Tris-HCl, pH 8.0, 50 mM KCl, and 10 mM CaCl2) and 400 μL of the diluted solution was 49 50 dropped on the surface of the mica substrate (natural mica, Nilaco, Tokyo, Japan) that had been 51 52 53 hydrophilized, and was incubated for 1 hr at 45 . After 10 min incubation at room temperature, 54 55 the substrates were washed 10 times with 400 μL buffer M5 (25 mM Tris-HCl, pH 8.0, 50 mM 56 57 KCl, and 5 mM CaCl ). Hereafter, all procedures were performed using ice-cold buffers. After 58 2 59 60 ACS Paragon Plus Environment 7

Langmuir Page 8 of 39

1 2 3 10 min incubation, each sample was washed 10 times with buffer M3 (25 mM Tris-HCl, pH 8.0, 4 5 6 50 mM KCl, and 3 mM CaCl2), and was then immediately washed 5 times with buffer M1 (25 7 8 mM Tris-HCl, pH 8.0, 50 mM KCl, and 1 mM CaCl2). After 10 min incubation, each sample was 9 10 11 washed 10 times with buffer M (25 mM Tris-HCl, pH 8.0, 50 mM KCl), then washed 3 times 12 13 with buffer ME (25 mM Tris-HCl, pH 8.0, 50 mM KCl, and 0.5 mM EDTA), and after 10 min 14 15 incubation, washed 10 times with buffer M. In order to obtain SLBs without defects or multi- 16 17 18 lamellar regions, a number of buffers were used serially for washing SLBs as described above. It 19

20 is noted that mica instead of SiO2/Si was used as the substrate for SLB in the above preparation 21 22 method, in order to obtain SLBs that will not form defects even under Ca2+-free conditions. 23 24 25 SLBs were observed using a fluorescence microscope (epi-fluorescence microscope BX51-FL, 26 27 Olympus, Tokyo, Japan) with fluorescent lighting devices (X-Cite exact, EXFO, Quebec, 28 29 30 Canada), a water immersion objective lens (LUMPlanFl, ×40, N.A. 0.80, Olympus) and a stage 31 32 equipped with a temperature control. When BODIPY-PC was observed, a mirror unit (U- 33 34 MWIG3, Olympus) was used. The excitation light intensity was adjusted by introducing 25 and 35 36 37 6% ND filters between the light source and the filter set. Images were obtained using a CCD 38 39 camera (Coolsnaps ES, Photometrics, Tucson, AZ USA) and were stored using software (Meta 40 41 Vue, Molecular Devices, Tokyo, Japan) as 16-bit grayscale TIFF files. The fluorescent images 42 43 44 obtained as above were analyzed using Image J computer software (http://imagej.nih.gov/ij/). 45 46 Prior to the experiments, SLBs obtained were confirmed for fluidity by Fluorescence Recovery 47 48 49 After Photobleaching (FRAP) analysis. 50 51 Atomic force microscopy (AFM) of SLBs was performed with PicoScan2500 (Keysight 52 53 Technologies, formerly Molecular Imaging, Tempe, AZ, USA) with a MAC lever Type VI 54 55 56 cantilever. The probe used was 100 μm in length, 18 μm in width, 0.2 N/m of elastic constant 57 58 59 60 ACS Paragon Plus Environment 8

Page 9 of 39 Langmuir

1 2 3 and 66 kHz of resonant frequency. AFM observation was carried out in a solution using 4 5 6 Magnetic-AC mode, a kind of tapping mode. 37, 38 The scanning speed was 0.11.0 line/sec, the 7 8 resolution was 256 dots/line and the scan range was performed in 20 μm × 20 μm. 9 10 11 Temperature control was carried out using an aluminum handmade stage with a Peltier element 12 13 (Netsu-denshi Kogyo, Tokyo, Japan) (Figure S1). The stage consists of two parts and holds the 14 15 16 Peltier element between the bottom and upper parts. Water flows through the inside of the 17 18 bottom part to cool the stage. Unless otherwise specified, the temperature of each SLB was 19 20 changed stepwise by every 5, and the SLBs were kept at each temperature for 15 min. The 21 22 23 reason to stand for 15 min is that, in the preliminary experiments, SLBs reached to a stable state 24 25 within 10 min at the latest after changing the temperature. 26 27 28 The septin fraction obtained as described below was centrifuged (15,000 ×g for 20 min, 4) 29 30 and was diluted two-fold with buffer M just prior to addition to the SLB, in order to remove the 31 32 33 much larger septin polymers and/or the debris of denatured proteins. Four hundred μL of the 34 35 septin fraction (protein concentration was 30 μg/mL) or the buffer solution resulting from the 36 37 38 final dialysis outer solution (as a control) was applied, after its temperature was adjusted and 39 40 buffer on the surface of each SLB was removed so as not to expose them to the air. 41 42 For each observation and numerical analysis, at least three independent experiments using 43 44 45 different sets of SLBs and protein sample were performed. 46 47 48 49 2.3. Giant Liposomes 50 51 30 52 Giant liposomes were prepared by natural swelling as detailed previously. In brief, 10 mM of 53 54 each phospholipid stock solution in chloroform/methanol (98:2, v:v) were mixed (total 200 nmol 55 56 57 of phospholipids). The lipid compositions of liposomes examined were PC alone (PC liposome), 58 59 60 ACS Paragon Plus Environment 9

Langmuir Page 10 of 39

1 2 3 PC:PI = 1:1 (PC+PI liposome), PC:PS = 1:1 (PC+PS liposome), PC:PG = 1:1 (PC+PG liposome) 4 5 6 or DOPC:DMPS = 1:1 (mol:mol). The solvent was evaporated under a flow of nitrogen gas and 7 8 the lipids were further dried in vacuo for at least 120 min. The dried lipid films were hydrated 9 10 11 with 250 μL 150 mM sucrose at 50°C for 120 min, and the liposome suspensions were diluted 12 13 20-fold with buffer L. 14 15 Giant liposomes were observed by dark-field microscopy (BHF, Olympus) with an objective 16 17 18 lens (HiApo40, Olympus) and a condenser (DC, Olympus). We manually tracked flowing 19 20 liposomes in focus. The serial images were recorded using a DVD recorder (DMR-XW 120, 21 22 Panasonic, Osaka, Japan). Fluorescent-labeled liposomes were observed using a fluorescence 23 24 25 microscope (BH2-RFCA, Olympus) attached to the dark-field microscope. 26 27 28 29 30 2.4. Septin Fraction 31 32 The brain extract was prepared as reported previously. 30 In brief, the forebrain and cerebellum 33 34 (about 90 gr) were dissected from a fresh porcine brain, washed with buffer A (25 mM Tris-HCl, 35 36 37 pH 8.0, 2 mM EGTA, 500 mM KCl, 10 mM MgCl2, 250 mM sucrose, 1 mM DTT, and two 38 39 protease inhibitor cocktail tablets), and then homogenized in 230 mL buffer A using a Waring 40 41 blender. The homogenate was cleared by consecutive centrifugations at 5,000 ×g for 1 hr, 10,000 42 43 44 ×g for 1 hr and 171,000 ×g for 2 hr. The final supernatant was dialyzed against buffer L. The 45 46 protein precipitate was salted out by adding ammonium sulfate to 40% (w/v) and incubation for 47 48 49 30 min, then collected by centrifugation at 10,000 ×g for 30 min and resuspended in buffer L (12 50 51 mL). The suspension was dialyzed against buffer L. After centrifugation at 171,000 ×g for 2 hr, 52 53 the supernatant (termed the brain extract) was used either for further purification or was quickly 54 55 56 57 58 59 60 ACS Paragon Plus Environment 10

Page 11 of 39 Langmuir

1 2 3 frozen with liquid nitrogen after addition of a final concentration of 1 mM GTP and stored at 4 5 6 −80. 7 8 The brain extract was fractionated first with a cation exchange column HiTrap SP FF (GE 9 10 11 Healthcare) attached to a FPLC liquid chromatography system (Pharmacia). The brain extract 12 13 was charged on the column equilibrated with buffer C (25 mM Tris-HCl, pH 8.0, 10 mM NaCl), 14 15 16 and bound proteins were eluted with a linear gradient of 10−500 mM NaCl in buffer C. The most 17 18 active fraction was applied to a hydroxyapatite column, Econo-Pac CHT-II (Bio-Rad), and was 19 20 eluted with a linear gradient of 10-500 mM potassium-phosphate buffer (pH 7.0). The major 21 22 23 fractions were dialyzed against buffer L. After the dialysis, the fractions were measured for 24 25 protein concentration using the BCA method with BSA as a standard, and were quickly frozen in 26 27 28 the presence of a final concentration of 1 mM GTP with liquid nitrogen and stored at −80 until 29 30 use. All procedures were conducted at 4 unless specifically noted. 31 32 33 In each step, fractions were applied to SDS-PAGE and assayed for liposome-transforming 34 35 activity. A 10 μL liposome solution containing 40 μM phospholipid was placed in a glass flow 36 37 38 cell at 25 and was observed using the dark-field microscope. Two μL of each sample solution, 39 40 which had been dialyzed against buffer L, was placed into the flow cell by capillary force and the 41 42 43 liposome suspensions were mixed by diffusion. 44 45 46 47 3. Results 48 49 50 3.1. Septin Arrays Assemble on the Surface of the Lipid Membrane 51 52 It is believed that septin is involved in the partition of biological membranes by restricting the 53 54 lateral movements of lipids and proteins constituting the membranes. The partition of 55 56 57 membranes is indispensable for the precise acceptance of complex signals by receptors and for 58 59 60 ACS Paragon Plus Environment 11

Langmuir Page 12 of 39

1 2 3 the exhibition of a variety of functions and the reinforcement of a unique morphology of 4 5 6 individual cells. 7 8 It has been known that septin interacts with lipid membranes containing acidic phospholipids, 9 10 30 11 and causes uniform bracing of the membrane as the result of the interaction. In addition, we 12 13 found in this study that septin can assemble into massive arrays of filaments on the surface of 14 15 lipid membrane vesicles, regardless of the tubulation (Figure 1). The arrays of filaments formed 16 17 18 on the membrane surface seemed to be involved in the compartment or domain formation of the 19 20 membrane. Thus, in this study, we verified that hypothesis using SLBs. 21 22 23 24 25 3.2. Formation and Disappearance of Cooling-Induced Domains in SLBs in the Absence 26 27 of Septin 28 29 30 Because SLBs are lipid membranes prepared on the surface of a flat substrate so that their 31 32 fluidity is observable both by optical fluorescence microscopy and by AFM, it is a favorable 33 34 experimental material to analyze the lateral movements of large numbers of molecules present in 35

36 39, 40 37 the membrane. In addition, it has been known that SLBs made from two synthetic 38 39 phospholipids, DMPC and DOPC, form domains of the gel phase at temperatures lower than 40 41 8 42 12 , because the Tc of DMPC and DOPC are different as mentioned above. 43 44 In this study, we used SLBs that were made from PC and PI to investigate the effect of septin 45 46 on lipid membranes other than the activity to induce membrane tubules, because PC and PI is the 47 48 30 49 favorable composition for interactions between membranes and septin. Incidentally, FRAP 50 51 analysis showed that the SLBs prepared retain their fluidity at room temperature, either in the 52 53 54 presence or absence of septin (Figure S2). 55 56 57 58 59 60 ACS Paragon Plus Environment 12

Page 13 of 39 Langmuir

1 2 3 4 When the temperature was lower than 15 , we found that SLBs form domains where the 5 6 regions appear darker than the surrounding areas in fluorescent images (Figure 2 and Figure S3). 7 8 9 In the process of decreasing the temperature, the domains that appeared around 15 increased in 10 11 their size with further decreases of temperature. Since both the number and the size of domains 12 13 14 did not change even when SLBs were placed at 5 for longer than 30 min, the domains formed 15 16 were rather stable unless the temperature changed. Since most domains formed had a fractal 17 18 19 shape, the domain formation might be from the diffusion-limited aggregation of lipid molecules. 20 21 Fluorescent-labeled annexin V is often used as a probe to detect PS on the outside of cell 22 23 membranes, annexin V being a protein that can bind to acidic phospholipids, PI and PG as well 24 25 26 as PS (Figure S4). If an SLB was made from PC and PI, the domains induced at a cold 27 28 temperature, which are dark by fluorescent microscopy using a fluorescent-labeled lipid, were 29 30 31 not recognized with the fluorescent annexin V (Figure S5 A C). Conversely, the surrounding 32 33 regions were recognized with that fluorescent annexin V. These observations suggest that the 34 35 cooling-induced domains are PI-free regions. It should be noted here that the domain formation 36 37 38 observed in this study is unique to the case of SLBs made from PC and PI obtained from natural 39 40 sources. When the lipid composition of SLBs was PC and PS, which is another acidic 41 42 43 phospholipid that is obtained from a natural source, the domain was not formed even if the SLB 44 45 was kept at 5 (Figure S5D). Moreover, even though the head group of the phospholipids that 46 47 48 constitute the SLBs is the same, the different configuration of the hydrocarbon chains can cause 49 50 a different domain formation. When the lipid composition of the SLBs was DOPC and DMPS, 51 52 the regions observed as dark by fluorescent microscopy using a fluorescent-labeled lipid 53 54 55 appeared in the cooled SLBs, while the regions were recognized with fluorescent-labeled 56 57 annexin V (Figure S5E). Since the conditions and mechanism for the formation of the cooling- 58 59 60 ACS Paragon Plus Environment 13

Langmuir Page 14 of 39

1 2 3 induced domain are complicated, the domains were used simply as a tool to examine the effect of 4 5 6 septin on the dynamics of lipid membranes in this study. 7 8 The cooling-induced domains disappeared reversibly by increasing the temperature (Figures 2 9 10 11 and S3). When the temperature was increased from 5 to 10, another region that possesses a 12 13 bright intensity in fluorescent images, penetrated into the domains from the surrounding areas, 14 15 16 resulting in the dispersion of the domains (Figure 2 and Figure S3, and see also Figure 4). Thus, 17 18 the number, and also the total area in some cases, of the domains increased once in the process of 19 20 increasing the temperature, even though the domain size, that is the average size of the individual 21 22 23 domains, decreased. When the temperature was higher than 15, the total area of domains 24 25 26 decreased with increasing temperature (Figure 3). When the temperature reached 25 , the 27 28 domains completely disappeared. 29 30 31 32 33 3.3. Effect of Septin on the Formation of the Domains 34 35 When septin was added to an SLB at 25, i.e., the solution covering an SLB was exchanged 36 37 38 with a solution containing septin prior to starting the process of decreasing the temperature of the 39 40 SLB, the size of the cooling-induced domains formed in the SLB was extremely restricted to a 41 42 43 much smaller size and inversely their number was increased, which resulted in a decrease of the 44 45 total area of the domains formed (Figures 2 and 3). When the solution on an SLB was exchanged 46 47 with a solution under the same conditions but without septin, these differences were never 48 49 50 induced. Therefore, the differences in results between cases in the presence or absence of septin 51 52 result from the effect of septin on the formation of the cooling-induced domain in SLBs. 53 54 The individual domains retained a fractal shape, regardless of the presence of septin (Figure 2). 55 56 57 The temperature when the domains appeared in the process of decreasing the temperature (about 58 59 60 ACS Paragon Plus Environment 14

Page 15 of 39 Langmuir

1 2 3 4 15 ) and the temperature when the domains completely disappeared in the process of increasing 5 6 the temperature (2025) was not affected by septin (Figure 3). Altogether, septin affects the 7 8 9 size and number, but not the basis of the formation, of the cooling-induced domain. Septin may 10 11 inhibit the movement of PC molecules toward the domains that had already appeared in the 12 13 14 SLBs and/or PI molecules from the surroundings of the domains via binding with PI molecules. 15 16 17 18 3.4. Effect of Septin on the Disappearance of the Domains 19 20 21 When septin was added to an SLB at 5, i.e., the solution covering an SLB was exchanged 22 23 with a solution containing septin after the formation of cooling-induced domains and before 24 25 26 starting the process of increasing the temperature, the cooling-induced domains that formed in 27 28 the SLB became very stable against the temperature increase and tended to remain even after the 29 30 31 SLB was placed at 25 for over than 60 min (Figures 4 and 5). Also, in this case, exchanging 32 33 the solution on the SLB with a solution with the same conditions but without septin, never 34 35 induced the difference. Therefore, the difference in results between cases in the presence or 36 37 38 absence of septin results from the effect of septin on the disappearance of the cooling-induced 39 40 domain that formed in SLBs. 41 42 43 Septin robustly prevented the shrinking and dispersion of the domains (Figure 4). Thus, during 44 45 the temperature increasing process, the shape of the individual domains was maintained (Figure 46 47 4) and changes in all parameters, i.e. the domain size, the number of domains, and the total area 48 49 50 of the domains, were extremely delayed by the addition of septin (Figure 5). Septin may inhibit 51 52 the movement of lipid molecules constituting the SLB via binding with PI molecules. 53 54 55 56 57 3.5. Septin on the Surface of SLBs 58 59 60 ACS Paragon Plus Environment 15

Langmuir Page 16 of 39

1 2 3 Because septin is thought to inhibit the movement of lipid molecules within an SLB as 4 5 6 described above, to determine the mechanism involved, next we tried to observe septin 7 8 interacting with SLBs using AFM (Figure 6). The SLBs prepared usually had a flat surface 9 10 11 (Figure 6A and B), except rarely showed some very small protrusions (Figure 6C) or defects 12 13 (Figure 6D). 14 15 When septin was added, although the fluorescent images did not show remarkable difference 16 17 18 (Figure 6E), some objects tended to be observed on the surface of SLBs by AFM (Figure 6F and 19 20 G). They were string-like structures or round-shape protrusions, which have a width or length 21 22 ranging from tens of nanometers to several micrometers and a height of about 5 nm. Septin 23 24 25 forms a hetero-hexamer, which consists of two hetero-trimers arranged in tandem, as a unit 26 27 complex for polymerization. The septin complex then polymerizes to filaments with a diameter 28 29 30 of about 5 nm, and subsequently those filaments associate with each other in a side-by-side 31 32 manner to form larger arrays. Therefore, the objects found on the surface of SLBs are assumed to 33 34 be assemblies of septin complexes. Membrane tubulation or the formation of long septin 35 36 37 filaments was not observed, which is reason that the amounts of septin added were kept low in 38 39 order to protect SLBs from decay caused by their deformation. The size of their lateral directions 40 41 and the shape of the assembly were not dependent on the amount of septin applied (Figure 6G). 42 43 44 Depending on the location on the SLBs rather than the amount of septin, various structures, i.e., 45 46 string-like structures or round-shape protrusions, are present. The reason may be due to the 47 48 49 formation of varied assemblies in each experiment because the nucleation that is a critical step 50 51 for the polymerization is a labile process. In addition, in the concentration ranges of septin 52 53 examined here, no remarkable difference was found in the septin-applied SLBs and the same 54 55 56 interference effects of septin on the SLBs were observed, even when the concentrations of septin 57 58 59 60 ACS Paragon Plus Environment 16

Page 17 of 39 Langmuir

1 2 3 added were changed. With increasing of the septin concentration, only the reproducibility of the 4 5 6 results was increased. 7 8 9 10 11 4. Discussion 12 13 4.1. Cooling-Induced Domains Formed in SLBs 14 15 Because both the BODIPY-PC and TR-DHPE used preferentially partition into a liquid crystal 16 17 18 phase in SLBs that causes phase separation, the bright region in fluorescent images that appears 19 20 at low temperatures is thought to be in the liquid crystal phase. 36 The remaining regions, which 21 22 are dark in fluorescent images and were confirmed not to be defects by AFM, that is, the 23 24 25 cooling-induced domains may be in the gel phase. In addition, as mentioned above, the 26 27 observations using fluorescent-labeled annexin V suggest that the domains are PI-free regions. 28 29 30 The domain formation observed in the cooled SLBs depends on the charge or size of the head 31 32 group and the length or degree of unsaturation of the hydrocarbon chains of the phospholipids 33 34 contained in the membrane and on the diversity (or uniformity) of the phospholipids constituting 35 36 37 the membrane, and is caused by a complicated involvement of these factors. The shape of the 38 39 domain suggests that the domain formation may result from the diffusion-limited aggregation of 40 41 phospholipids. 41 Since the conditions and mechanism for the formation of the cooling-induced 42 43 44 domain found in this study remain unclear, they should be further investigated. Moreover, the 45 46 lipid composition of the SLBs used in this study and the condition for the domain formation are 47 48 49 different from the naturally occurring. Therefore, studies using membrane domains that can be 50 51 induced under more physiological conditions should be required. 52 53 54 55 56 4.2. Effect of Septin on the Formation and Disappearance of the Domain 57 58 59 60 ACS Paragon Plus Environment 17

Langmuir Page 18 of 39

1 2 3 Septin may inhibit the diffusion-propelled long-range movements of lipids via the formation of 4 5 6 some complexes with PI molecules on the SLBs (Figure 7). If septin is added to an SLB before 7 8 the temperature decrease, i.e., prior to the domain formation, the inhibitory effect of septin would 9 10 11 stagnate the movement of PI molecules receding from the domain and thus restrict the number of 12 13 PC molecules gathering per domain. As a result, in comparison with the case without the 14 15 addition of septin, the domains that appear during the process of the temperature decrease remain 16 17 18 small without growth and increase their number without coalescence (Figure 2). In contrast, if 19 20 septin is added to an SLB after the domain formation caused by the temperature decrease, the 21 22 inhibitory effect of septin would stagnate the movement of PI molecules penetrating to the 23 24 25 domain region and thus obstruct the movement of PC molecules leaving the domain. As a result, 26 27 in comparison with the case without the addition of septin, the domains show a delayed 28 29 30 disappearance during the process of the temperature increase (Figure 4). 31 32 As described above, the domain is estimated as a PI-free region. Septin interacts with PI but 33 34 not with PC. Together, these results suggest that septin interacts with the domain-forming SLB 35 36 37 and is distributed outside the domain region. Therefore, the complexes consisting of septin 38 39 proteins on the surface of the SLBs and PI lipids in the SLBs, are thought to be formed outside 40 41 the domain regions (Figure 7). On the surface of the SLBs, septin proteins polymerize to form 42 43 44 filaments, or assemble to form oligomers, for example, nuclei that are the previous stage of 45 46 filament elongation. Against the diffusion movements of phospholipids in the SLBs, the 47 48 49 complexes can play the role of a fence in the former case (the upper model in Figure 7) and the 50 51 pickets in the latter case (the bottom model in Figure 7). Unfortunately, the AFM images 52 53 obtained show that there are both fibrous structures and round-shape protrusions on the septin- 54 55 56 treated SLBs, suggesting the possibility that both structures occur, and thus we could not reach a 57 58 59 60 ACS Paragon Plus Environment 18

Page 19 of 39 Langmuir

1 2 3 decision about this (Figure 6). In either case, electrostatic interactions between septin and acidic 4 5 6 lipids may be important for the formation of the diffusion barrier in the lipid membrane, as 7 8 reported previously by a simulation study. 33 That study also indicated that adequate blockage 9 10 11 against diffusion by a placed on the membrane surface is hard to provide, which 12 13 is consistent with the results of the FRAP analysis of SLBs at room temperature, which were not 14 15 affected by septin (Figure S2). Possibly, septins interacting with PI lipids can block the 16 17 18 movement of the populations of lipids, which are clustered by the interaction, but not the 19 20 diffusion of individual lipid molecules. 21 22 The domain formation investigated here is particular to the conditions of low temperature. The 23 24 25 mechanism whereby septin affects the formation and disappearance of the cooling-induced 26 27 domain remains unknown, whether the fence by filaments or the pickets by oligomers. In any 28 29 30 case, the results obtained in this study clearly demonstrate that septin, a membrane-interacting 31 32 cytoskeletal protein, can affect the size, number, and stability of the cooling-induced domain that 33 34 is induced in the membrane. 35 36 37 Septin has been estimated to not only be involved in the reinforcement of membrane structures, 38 39 but also to play a role as a diffusion barrier against membrane proteins in diverse types of cells 40 41 such as at the neck of sperm or the base of spinal nerve synapses. 42, 43 The effects of septin on the 42 43 44 membrane domain as directly observed in our in vitro study here strengthens the earlier 45 46 suggestion of the involvement of septin in the partitioning of cell membranes proposed by in vivo 47 48 49 and in silico studies. The ability of septin as demonstrated here will enable us to regulate the 50 51 domain formation of lipid bilayer membranes, although the detailed mechanism involved 52 53 remains unclear. In addition to septin, many other proteins have been shown to interact with 54 55 56 57 58 59 60 ACS Paragon Plus Environment 19

Langmuir Page 20 of 39

1 2 3 membranes and to assemble at their surface. Those proteins might also be useful for the control 4 5 6 of the properties of lipid bilayer membranes. 7 8 9 10 11 5. Conclusions 12 13 Free control of processes related to the partitioning of lipid membranes, including phase 14 15 separation and domain formation, is very important physiologically as well as physically and is 16 17 18 an engineering issue that allows the diversification of membrane function to evolve. In this study, 19 20 we investigated the effect of septin on lipid bilayer membranes using SLBs, and found that PI- 21 22 containing SLBs reversibly form domains when the temperature is decreased, and that septin 23 24 25 affects both the formation and the disappearance of the cooling-induced domain. Septin inhibits 26 27 the growth of those domains during the decrease of temperature, and inhibits the dispersion and 28 29 30 disappearance of the domains during the increases of temperature. The effects of septin on the 31 32 membrane domain directly observed in this study strengthen the hypothesis of the physiological 33 34 role of septin concerning the partitioning of cell membranes. These functions of septin should 35 36 37 enable the regulation of domain formation and the creation of unique regions in lipid bilayer 38 39 membranes. 40 41 42 43 44 FIGURE CAPTIONS 45 46 47 Figure 1. Giant liposome coiled with septin filaments. Electron micrographs show a PC+PI 48 49 liposome in the absence (left) or presence (center) of the septin fraction. Arrows indicate regions 50 51 52 that are expected to begin to tubulation. Septin filaments are wrapped around these short tubules. 53 54 The right panel shows an enlarged and contrast enhanced image of the region indicated by the 55 56 57 blue box in the center panel, in order to show bundles of septin filaments assembled on the 58 59 60 ACS Paragon Plus Environment 20

Page 21 of 39 Langmuir

1 2 3 surface of the liposome. Some of the bundles are indicated by blue brackets. Bars indicate 200 4 5 6 nm. 7 8 9 10 11 12 Figure 2. Effect of septin on the cooling-induced domain formation in SLBs made from PC, PI, 13 14 15 and fluorescence-labeled PC. A buffer solution without (upper) or with septin (bottom) that had 16 17 been adjusted to 25 was applied to an SLB. The temperature was then decreased from 25 to 18 19 20 5 by 5 every 15 min. After standing at 5 for 30 min, the temperature was increased to 21 22 25 by 5 every 15 min. Fluorescent images of SLBs at 25 and 15 during the temperature 23 24 25 decreasing process and at 5, 15, and 25 during the temperature increasing process are shown. 26 27 28 Bars indicate 10 μm. The same field of SLBs was continuously observed. More detail is shown 29 30 in Figure S3 regarding the case without septin (controls). 31 32 33 34 35 36 37 Figure 3. Size (top), number (middle), and total area (bottom) of the cooling-induced domains 38 39 formed in SLBs in the presence (red: ‘+ Septin’) or the absence of septin (blue: ‘Control’). 40 41 Conditions are the same as detailed for Figures 2 and S3; that is, a buffer solution without or 42 43 44 with septin, that had been adjusted to 25, was added to the SLBs before changing the 45 46 temperature. The size (μm2) of the domains formed is shown by boxplot. The top and bottom of 47 48 49 each whisker indicate the maximum and minimum values, the upper and lower ends of each box 50 51 indicate the first and third quarter points, and the white line in the box represents the median. 52 53 The number of domains formed and the ratio (%) of total area of the domains against the total 54 55 56 area of SLBs are shown by line graphs. Error bars indicate standard deviation (S.D.). When 57 58 59 60 ACS Paragon Plus Environment 21

Langmuir Page 22 of 39

1 2 3 results obtained in the presence of septin are compared with those in the absence of septin, the p 4 5 6 values of <0.0001, 0.001, and 0.01 are indicated by ***, **, and *, respectively. The Wilcoxon- 7 8 Mann-Whitney signed-rank test was used for statistical analysis. 9 10 11 12 13 14 15 Figure 4. Effect of septin on the cooling-induced domain formed in SLBs. SLBs were made 16 17 from PC, PI, and fluorescence-labeled PC. The temperature of the SLBs was decreased to 5 18 19 20 and the cooling-induced domain was formed. A buffer solution without (upper) or with septin 21 22 (bottom) that was at 0 was then applied to the SLBs. After standing at 5 for 90 min, the 23 24 25 temperature was increased to 25 by 5 every 15 min, then left for up to 120 min at 25. 26 27 28 Fluorescent images of SLBs at 5, 10, 15, and 25 during the temperature increase are shown in 29 30 the case of the absence of septin, and images at 5, 10, and 15 during the temperature increase 31 32 33 and after standing for 60 min at 25 are shown in the case of the presence of septin. Bars 34 35 indicate 10 μm. 36 37 38 39 40 41 42 Figure 5. Size (top), number (middle), and total area (bottom) of the cooling-induced domains 43 44 formed in SLBs in the presence (red: ‘+ Septin’) or absence of septin (blue: ‘Control’). 45 46 47 Conditions are the same as detailed for Figure 4; that is, after SLBs were cooled down to 5 and 48 49 the cooling-induced domain was formed, a buffer solution without or with septin that was at 0, 50 51 52 was added to the SLBs. The size (μm2) of domains formed is shown by boxplot. The number of 53 54 domains formed and the ratio (%) of total area of the domains against the total area of SLBs are 55 56 57 shown by line graphs. In the middle and bottom panels, when results obtained in the presence of 58 59 60 ACS Paragon Plus Environment 22

Page 23 of 39 Langmuir

1 2 3 septin are compared with those in the absence of septin, the p values of <0.0001, 0.001, and 0.01 4 5 6 are indicated by ***, **, and *, respectively. 7 8 9 10 11 12 Figure 6. Properties of the SLB surface analyzed by fluorescence microscopy and by AFM. 13 14 15 SLBs were made from PC, PI, and fluorescence-labeled PC. SLBs without (AD) or with septin 16 17 (EG) were observed by fluorescence microscopy (A and E) or by AFM (BD, F, and G). The 18 19 20 bar in each fluorescent image indicates 10 μm. The vertical and horizontal range shown in the 21 22 topographic image of AFM is 2 μm × 2 μm (BD) or 20 μm × 20 μm (F and G). The color scale 23 24 25 on the right side of each topographic image indicates the height of the SLB surface (nm). (B) 26 27 Results of an SLB, which has an almost flat surface. (C) Results for the case when some particles 28 29 30 were found on the SLB surface. (D) Results of an SLB, which has defects. The final amounts of 31 32 septin added were 6.8 (E), 12 (F), 0.85, 6.8, 10, and 12 μg (from left to right in G). In the case of 33 34 35 the AFM observation of SLB shown here, a septin solution stored at 0 was added to an SLB 36 37 incubated at 25. Observations were performed at 25. In (C), (D), and (F), the height of the 38 39 40 SLB surface (nm) obtained by the line scanning in the location indicated by the white line in the 41 42 topographic image is also shown. 43 44 45 46 47 48 49 Figure 7. Model for the effect of septin on the formation and disappearance of the cooling- 50 51 induced domain in SLBs; i) septin filament behaving as a fence suppressing the lateral diffusion 52 53 of phospholipids (upper two panels), and ii) septin facilitating the nucleation by binding to the 54 55 56 membrane and tethering PI molecules (lower two panels). In each model, the upper layer is the 57 58 59 60 ACS Paragon Plus Environment 23

Langmuir Page 24 of 39

1 2 3 4 case when septin is added to an SLB incubated at 25 and thus has not yet formed the domain, 5 6 and the bottom layer is the case when septin is added to an SLB incubated at 5 and thus has 7 8 9 already formed the domain. Phospholipids shown in orange and red are PI and PC, respectively. 10 11 We estimate that the cooling-induced domain, which has a lower fluorescent intensity, is a PI- 12 13 14 free region. Blue indicates septin filaments or oligomers (nucleus) that are assembling on the 15 16 membrane. Septin can interact with the lipid bilayer membrane via associating with PI, an acidic 17 18 phospholipid. The concentrations of septin added were restricted so as not to induce membrane 19 20 21 tubulation because membrane tubulation causes defects in SLBs. In such concentration ranges, 22 23 septin might not assemble into filaments. Septin did not alter the result of FRAP analysis that 24 25 was carried out at room temperature (Figure S2). Therefore, septin proteins might be in a state 26 27 28 that allows diffusion together with lipid molecules. Altogether, the model shown in the lower 29 30 half might be more accurate than that shown in the upper half. 31 32 33 34 35 36 37 ASSOCIATED CONTENT 38 39 40 Supporting Information 41 42 43 Photograph of the stage of the fluorescence microscope. Sequences of fluorescent images 44 45 showing typical FRAP results of SLBs prepared. Sequence of fluorescent images showing the 46 47 48 cooling-induced domain formed in SLBs made from PC, PI, and fluorescence-labeled PC. Dark- 49 50 field and fluorescent images of giant liposomes in the presence of fluorescent-labeled annexin V. 51 52 53 The distribution of the cooling-induced domain and the distribution of SLB-binding annexin V. 54 55 This information is available free of charge via the Internet at http://pubs.acs.org. 56 57 58 59 60 ACS Paragon Plus Environment 24

Page 25 of 39 Langmuir

1 2 3 4 5 AUTHOR INFORMATION 6 7 8 Corresponding Author 9 10 11 *Division of Biological Science, Graduate School of Science, Nagoya University, Furo-cho, 12 13 Chikusa-ku, Nagoya 464-8602, Japan. Telephone: +81 52 788 6248, FAX: +81 52 747 6471, E- 14 15 16 mail: [email protected] (K. T.). 17 18 19 20 21 22 Conflict of Interest 23 24 The authors declare no competing financial interest. 25 26 27 28 29 30 ACKNOWLEDGMENTS 31 32 33 34 We are grateful to Tamiki Umeda (Kobe University) and Yuichiro Maéda, Akihiro Narita, and 35 36 Michio Homma (Nagoya University) for their invaluable support, useful comments, and 37 38 encouragement. This work was supported by a Grant-in-Aid for Challenging Exploratory 39 40 41 Research (Project No. 24651134) of the Japan Society for the Promotion of Science, and on 42 43 Innovative Areas “Molecular Robotics” (Project No. 24104004) of the Ministry of Education, 44 45 Culture, Sports, Science, and Technology of Japan. 46 47 48 49 50 51 ABBREVIATIONS 52 53 SLB, supported lipid bilayer; AFM, atomic force microscopy; PI, phosphatidylinositol; TR- 54 55 DHPE, N-(Texas Red sulfonyl)-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine; 56 57 58 BODIPY-PC, 2-(4,4-difluoro-5,7-diphenyl-4-bora-3a,4a-diaza-s-indacene-3-pentanoyl)-1- 59 60 ACS Paragon Plus Environment 25

Langmuir Page 26 of 39

1 2 3 hexadecanoyl-sn-glycero-3-phosphocholine; PC, phosphatidylcholine; DOPC, 1,2-dioleoyl-sn- 4 5 6 glycero-3-phosphatidylcholine; PG, phosphatidylglycerol; PS, phosphatidylserine; DMPS, 1,2- 7 8 dimyristoyl-sn-glycero-3-phosphatidylserine. 9 10 11 12 13 REFERENCES 14 15 (1) Makushok, T.; Alves, P.; Huisman, S. M.; Kijowski, A. R.; Brunner, D. Sterol-Rich 16 17 18 Membrane Domains Define Fission Cell Polarity. Cell 2016, 165, 1182–1196. 19 20 21 (2) Salamone, M.; Pavia, F. C.; Ghersi, G. Proteolytic Enzymes Clustered in Specialized 22 23 Plasma-Membrane Domains Drive Endothelial Cells' Migration. PLoS ONE 2016, 11, e0154709. 24 25 26 27 (3) Ikenouchi, J.; Hirata, M.; Yonemura, S.; Umeda, M. Sphingomyelin Clustering is Essential 28 29 for the Formation of Microvilli. J. Cell Sci. 2013, 126, 3585–3592. 30 31 32 (4) Arita, Y.; Nishimura, S.; Ishitsuka, R.; Kishimoto, T.; Ikenouchi, J.; Ishii, K.; Umeda, M.; 33 34 35 Matsunaga, S.; Kobayashi, T.; Yoshida, M. Targeting Cholesterol in a Liquid-Disordered 36 37 Environment by Theonellamides Modulates Cell Membrane Order and Cell Shape. Chem. Biol. 38 39 2015, 22, 604–610. 40 41 42 (5) van den Bogaart, G.; Meyenberg, K.; Risselada, H. J.; Amin, H.; Willig, K. I.; Hubrich, B. 43 44 45 E.; Dier, M.; Hell, S. W.; Grubmuller, H.; Diederichsen, U.; Reinhard, J. Membrane Protein 46 47 Sequestering by Ionic Protein–Lipid Interactions. Nature 2011, 479, 552–555. 48 49 50 (6) Tomishige, M.; Kusumi, A. Compartmentalization of the Erythrocyte Membrane by the 51 52 53 Membrane Skeleton: Intercompartmental Hop Diffusion of Band 3. Mol. Biol. Cell 1999, 10, 54 55 2475–2479. 56 57 58 59 60 ACS Paragon Plus Environment 26

Page 27 of 39 Langmuir

1 2 3 (7) Nakada, C.; Ritchie, K.; Oba, Y.; Nakamura, M.; Hotta, Y.; Iino, R.; Kasai, R. S.; 4 5 6 Yamaguchi, K.; Fujiwara, T.; Kusumi, A. Accumulation of Anchored Proteins Forms Membrane 7 8 Diffusion Barriers During Neuronal Polarization. Nat. Cell Biol. 2003, 5, 626–632. 9 10 11 (8) Tero, R.; Sazaki, G.; Ujihara, T.; Urisu, T. Anomalous Diffusion in Supported Lipid 12 13 14 Bilayers Induced by Oxide Surface Nanostructures. Langmuir, 2011, 27, 9662–9665. 15 16 17 (9) Castellana, E. T.; Cremer, P. S. Solid Supported Lipid Bilayers: From Biophysical Studies 18 19 to Sensor Design. Surf. Sci. Rep. 2006, 61, 429–444. 20 21 22 23 (10) Keller, C. A.; Kasemo, B. Surface Specific Kinetics of Lipid Vesicle Adsorption 24 25 Measured with a Quartz Crystal Microbalance. Biophys. J. 1998, 75, 1397–1402. 26 27 28 (11) Reviakine, I. Brisson, A. Formation of Supported Phospholipid Bilayers from Unilamellar 29 30 31 Vesicles Investigated by Atomic Force Microscopy. Langmuir 2000, 16, 1806–1815. 32 33 34 (12) Rossetti, F. F.; Bally, M.; Michel, R.; Textor M.; Reviakine, I. Interactions between 35 36 Titanium Dioxide and Phosphatidyl Serine-Containing Liposomes: Formation and Patterning of 37 38 39 Supported Phospholipid Bilayers on the Surface of a Medically Relevant Material. Langmuir 40 41 2005, 21, 6443–6450. 42 43 44 45 (13) Tero, R.; Watanabe, H.; Urisu, T. Supported Phospholipid Bilayer Formation on 46 47 Hydrophilicity-Controlled Silicon Dioxide Surfaces. Phys. Chem. Chem. Phys. 2006, 8, 3885– 48 49 3894. 50 51 52 (14) Dietrich, C.; Bagatolli, L. A.; Volovyk, Z. N.; Thompson, N. L.; Levi, M.; Jacobson, K.; 53 54 55 Gratton, E. Lipid Rafts Reconstituted in Model Membranes. Biophys. J. 2001, 80, 1417–1428. 56 57 58 59 60 ACS Paragon Plus Environment 27

Langmuir Page 28 of 39

1 2 3 (15) Dietrich, C.; Volovyk, Z. N.; Levi, M.; Thompson, N. L.; Jacobson, K. Seeing is 4 5 6 Believing: Visualization of Rafts in Model Membranes, Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 7 8 10642–10647. 9 10 11 (16) Crane, J. M.; Tamm, L. K. Role of Cholesterol in the Formation and Nature of Lipid Rafts 12 13 14 in Planar and Spherical Model Membranes. Biophys. J. 2004, 86, 2965–2979. 15 16 17 (17) Riniaa, H. A.; Snelb, M. M. E.; van der Eerden, J. P. J. M.; de Kruijff, B. Visualizing 18 19 Detergent Resistant Domains in Model Membranes with Atomic Force Microscopy. FEBS Lett. 20 21 22 2001, 501, 92–96. 23 24 25 (18) Silvius, J. R. Partitioning of Membrane Molecules Between Raft and Non-Raft Domains: 26 27 Insights from Model-Membrane Studies. Biochem. Biophys. Acta 2005, 1746, 193–202. 28 29 30 31 (19) Hsu, S.-C.; Hazuka, C. D.; Roth, R.; Foletti, D. L.; Heuser, J.; Scheller, R. H. Subunit 32 33 Composition, Protein Interactions, and Structures of the Mammalian Brain Sec6/8 Complex and 34 35 Septin Filaments. Neuron 1998, 20, 1111–1122. 36 37 38 39 (20) Kinoshita, M.; Field, C. M.; Coughlin, M. L.; Straight, A. F.; Mitchison, T. J. Self- and 40 41 Actin-Templated Assembly of Mammalian Septins. Dev. Cell, 2002, 3, 791–802. 42 43 44 (21) Weirich, C. S.; Erzberger, J. P.; Barral, Y. The Septin Family of : Architecture 45 46 47 and Dynamics. Nat. Rev. Mol. Cell Biol. 2008, 9, 478–489. 48 49 50 (22) Kinoshita, M. Diversity of Septin Scaffolds. Curr. Opin. Cell Biol. 2006, 18, 54–60. 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment 28

Page 29 of 39 Langmuir

1 2 3 (23) Mavrakis, M.; Azou-Gros, Y.; Tsai, F.-C.; Alvarado, J.; Bertin, A.; Iv, F.; Kress, A.; 4 5 6 Brasselet, S.; Koenderink, G. H.; Lecuit, T. Septins Promote F-Actin Ring Formation by 7 8 Crosslinking Actin Filaments into Curved Bundles. Nat. Cell Biol. 2014, 16, 322–334. 9 10 11 (24) Surka, M. C.; Tsang, C. W.; Trimble, W. S. The Mammalian Septin MSF Localizes with 12 13 14 Microtubules and is Required for Completion of . Mol. Biol. Cell 2002, 13, 3532– 15 16 3545. 17 18 19 (25) Nagata, K.; Kawajiri, A.; Matsui, S.; Takagishi, M.; Shiromizu, T.; Saitoh, N.; Izawa, I.; 20 21 22 Kiyono, T.; Itoh, T. J.; Hotani, H.; Inagaki, M. Filament Formation of MSF-A, a Mammalian 23 24 Septin, in Human Mammary Epithelial Cells Depends on Interactions with Microtubules. J. Biol. 25 26 27 Chem. 2003, 278, 18538–18543. 28 29 30 (26) Joo, E.; Surka, M. C.; Trimble, W. S. Mammalian SEPT2 is Required for Scaffolding 31 32 Nonmuscle Myosin II and its Kinases. Dev. Cell 2007, 13, 677–690. 33 34 35 (27) Dent, J.; Kato, K.; Peng, X.-R.; Martinez, C.; Cattaneo, M.; Poujol, C.; Nurden, P.; 36 37 38 Nurden, A.; Trimble, W. S.; Ware, J. A Prototypic Platelet Septin and its Participation in 39 40 Secretion. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 3064–3069. 41 42 43 (28) Ihara, M.; Kinoshita, A.: Yamada, S.; Tanaka, H.; Tanigaki, A.; Kitano, A.; Goto, M.; 44 45 46 Okubo, K.; Nishiyama, H.; Ogawa, O.; Takahashi, C.; Itohara, S.; Nishimune, Y.; Noda, M.; 47 48 Kinoshita, M. Cortical Organization by the Septin Cytoskeleton is Essential for Structural and 49 50 Mechanical Integrity of Mammalian Spermatozoa. Dev. Cell 2005, 8, 343–352. 51 52 53 54 (29) Ihara, M.; Yamasaki, N.; Hagiwara, A.; Tanigaki, A.; Kitano, A.; Hikawa, R.; Tomimoto, 55 56 H.; Noda, M.; Takanashi, M.; Mori, H.; Hattori, N.; Miyakawa, T. Kinoshita, M. Sept4, a 57 58 59 60 ACS Paragon Plus Environment 29

Langmuir Page 30 of 39

1 2 3 Component Of Presynaptic Scaffold and Lewy Bodies, is Required for the Suppression of Alpha- 4 5 6 Synuclein Neurotoxicity. Neuron 2007, 53, 519–533. 7 8 9 (30) Tanaka-Takiguchi, Y.; Kinoshita, M.; Takiguchi, K. Septin-Mediated Uniform Bracing of 10 11 Phospholipid Membranes. Curr. Biol. 2009, 19, 140–145. 12 13 14 15 (31) Bridges, A. A.; Zhang, H.; Mehta, S. B.; Occhipinti, P.; Tani, T.; Gladfelter, A. S. Septin 16 17 Assemblies form by Diffusion-Driven Annealing on Membranes. Proc. Natl. Acad. Sci. U.S.A. 18 19 2014, 111, 2146–2151. 20 21 22 23 (32) Bertin, A.; McMurray, M. A.; Thai, L.; Garcia III, G.; Votin, V.; Grob, P.; Allyn, T.; 24 25 Thorner, T.; Nogales, E. Phosphatidylinositol-4,5-bisphosphate Promotes Budding Yeast Septin 26 27 Filament Assembly and Organization. J. Mol. Biol. 2010, 404, 711–731. 28 29 30 31 (33) Lee, K. I.; Im, W.; Pastor, R. W. Langevin Dynamics Simulations of Charged Model 32 33 Phosphatidylinositol Lipids in the Presence of Diffusion Barriers: Toward an Atomic Level 34 35 Understanding of Corralling of PIP by Protein Fences in Biological Membranes. BMC 36 2 37 38 Biophysics 2014, 7, 13–22. 39 40 41 (34) Bridges, A. A.; Jentzsch, M. S.; Oakes, P. W.; Occhipinti, P.; Gladfelter, A. S. Micron- 42 43 Scale Plasma Membrane Curvature is Recognized by the Septin Cytoskeleton. J. Cell Biol. 2016, 44 45 46 213, 23-32. 47 48 49 (35) Kiessling, V.; Yang, S.-T.; Tamm, L. K. Supported Lipid Bilayers as Models for Studying 50 51 Membrane Domains. Curr. Top. Memb. 2015, 75, 1–23. 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment 30

Page 31 of 39 Langmuir

1 2 3 (36) Ujihara, T.; Suzuki, S.; Yamauchi, Y.; Tero, R.; Takeda, Y. Local Concentration of Gel 4 5 6 Phase Domains in Supported Lipid Bilayers under Light Irradiation in Binary Mixture of 7 8 Phospholipids Doped with Dyes for Photoinduced Activation. Langmuir 2008, 24, 10974–10980. 9 10 11 (37) Han, W.H.; Lindsay, S.M. Probing Molecular Ordering at a Liquid-Solid Interface with a 12 13 14 Magnetically Oscillated Atomic Force Microscope. Appl. Phys. Lett. 1998, 72, 1656–1658. 15 16 17 (38) Guanglu, G.; Dong, H.; Danying, L.; Weiguo, C.; Yunxu, S.; Lei, J.; Wanyun, M.; Chen, 18 19 W. MAC Mode Atomic Force Microscopy Studies of Living Samples, Ranging from Cells to 20 21 22 Fresh Tissue. Ultramicroscopy 2007, 107, 299–307. 23 24 25 (39) Isogai, T.; Piednoir, A.; Akada, E.; Akahoshi, Y.; Tero, R.; Harada, S.; Ujihara, T.; 26 27 Tagawa, M. Forming Two-Dimensional Structure of DNA-Functionalized Au Nanoparticles via 28 29 30 Lipid Diffusion in Supported Lipid Bilayers. J. Cryst. Growth 2014, 401, 494–498. 31 32 33 (40) Gumi-Audenis, B.; Sanz, F.; Giannotti, M. I. Impact of Galactosylceramides on the 34 35 Nanomechanical Properties of Lipid Bilayer Models: An AFM-Force Spectroscopy Study. Soft 36 37 38 Matter 2015, 11, 5447–5454. 39 40 41 (41) Witten, T. A.; Sander, L. M. Diffusion-Limited Aggregation, a Kinetic Critical 42 43 Phenomenon. Phys. Rev. Lett. 1981, 47, 1400–1403. 44 45 46 47 (42) Ewers, H.; Tada, T.; Petersen, J. D.; Racz, B.; Sheng, M.; Choquet, D. A Septin- 48 49 Dependent Diffusion Barrier at Dendritic Spine Necks. PLoS ONE 2014, 9, e113916. 50 51 52 (43) Caudron, F.; Barral, Y. Septins and the Lateral Compartmentalization of Eukaryotic 53 54 55 Membranes. Dev. Cell 2009, 16, 493–506. 56 57 58 59 60 ACS Paragon Plus Environment 31

Langmuir Page 32 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Septin is one of the proteins that form a cytoskeletal network in biological membranes. A 22 23 membrane domain formation is inducible in supported lipid bilayers (SLBs) consisting of 24 25 26 phosphatidylcholine and phosphatidylinositol by cooling. In the presence of septin, the formation 27 28 and disappearance of the cooling-induced domain is affected. This finding demonstrates that it is 29 30 possible to regulate membrane domain formation by utilizing the membrane beneath the 31 32 33 cytoskeletal protein. 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment 32

Page 33 of 39 Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Figure 1. Giant liposome coiled with septin filaments. Electron micrographs show a PC+PI liposome in the 23 absence (left) or presence (center) of the septin fraction. Arrows indicate regions that are expected to begin 24 to tubulation. Septin filaments are wrapped around these short tubules. The right panel shows an enlarged and contrast enhanced image of the region indicated by the blue box in the center panel, in order to show 25 bundles of septin filaments assembled on the surface of the liposome. Some of the bundles are indicated by 26 blue brackets. Bars indicate 200 nm. 27 28 74x32mm (600 x 600 DPI) 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment Langmuir Page 34 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Figure 2. Effect of septin on the cooling-induced domain formation in SLBs made from PC, PI, and 26 fluorescence-labeled PC. A buffer solution without (upper) or with septin (bottom) that had been adjusted to 27 25℃ was applied to an SLB. The temperature was then decreased from 25 to 5℃ by 5℃ every 15 min. After 28 standing at 5℃ for 30 min, the temperature was increased to 25℃ by 5℃ every 15 min. Fluorescent images of SLBs at 25 and 15 during the temperature decreasing process and at 5, 15, and 25 during the 29 ℃ ℃ temperature increasing process are shown. Bars indicate 10 µm. The same field of SLBs was continuously 30 observed. More detail is shown in Figure S3 regarding the case without septin (controls). 31 32 90x47mm (300 x 300 DPI) 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment Page 35 of 39 Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 Figure 3. Size (top), number (middle), and total area (bottom) of the cooling-induced domains formed in SLBs in the presence (red: ‘+ Septin’) or the absence of septin (blue: ‘Control’). Conditions are the same as 48 detailed for Figures 2 and S3; that is, a buffer solution without or with septin, that had been adjusted to 49 25℃, was added to the SLBs before changing the temperature. The size (µm2) of the domains formed is 50 shown by boxplot. The top and bottom of each whisker indicate the maximum and minimum values, the 51 upper and lower ends of each box indicate the first and third quarter points, and the white line in the box 52 represents the median. The number of domains formed and the ratio (%) of total area of the domains 53 against the total area of SLBs are shown by line graphs. Error bars indicate standard deviation (S.D.). When 54 results obtained in the presence of septin are compared with those in the absence of septin, the p values of 55 <0.0001, 0.001, and 0.01 are indicated by ***, **, and *, respectively. The Wilcoxon-Mann-Whitney signed-rank test was used for statistical analysis. 56 57 149x264mm (300 x 300 DPI) 58 59 60 ACS Paragon Plus Environment Langmuir Page 36 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment Page 37 of 39 Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Figure 4. Effect of septin on the cooling-induced domain formed in SLBs. SLBs were made from PC, PI, and 27 fluorescence-labeled PC. The temperature of the SLBs was decreased to 5℃ and the cooling-induced domain 28 was formed. A buffer solution without (upper) or with septin (bottom) that was at 0℃ was then applied to 29 the SLBs. After standing at 5℃ for 90 min, the temperature was increased to 25℃ by 5℃ every 15 min, then 30 left for up to 120 min at 25℃. Fluorescent images of SLBs at 5, 10, 15, and 25℃ during the temperature 31 increase are shown in the case of the absence of septin, and images at 5, 10, and 15℃ during the temperature increase and after standing for 60 min at 25℃ are shown in the case of the presence of septin. 32 Bars indicate 10 µm. 33 34 90x50mm (300 x 300 DPI) 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment Langmuir Page 38 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 Figure 5. Size (top), number (middle), and total area (bottom) of the cooling-induced domains formed in SLBs in the presence (red: ‘+ Septin’) or absence of septin (blue: ‘Control’). Conditions are the same as 48 detailed for Figure 4; that is, after SLBs were cooled down to 5℃ and the cooling-induced domain was 49 formed, a buffer solution without or with septin that was at 0℃, was added to the SLBs. The size (µm2) of 50 domains formed is shown by boxplot. The number of domains formed and the ratio (%) of total area of the 51 domains against the total area of SLBs are shown by line graphs. In the middle and bottom panels, when 52 results obtained in the presence of septin are compared with those in the absence of septin, the p values of 53 <0.0001, 0.001, and 0.01 are indicated by ***, **, and *, respectively. 54 150x264mm (300 x 300 DPI) 55 56 57 58 59 60 ACS Paragon Plus Environment Page 39 of 39 Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 Figure 6. Properties of the SLB surface analyzed by fluorescence microscopy and by AFM. SLBs were made 42 from PC, PI, and fluorescence-labeled PC. SLBs without (A−D) or with septin (E−G) were observed by 43 fluorescence microscopy (A and E) or by AFM (B−D, F, and G). The bar in each fluorescent image indicates 44 10 µm. The vertical and horizontal range shown in the topographic image of AFM is 2 µm × 2 µm (B−D) or 45 20 µm × 20 µm (F and G). The color scale on the right side of each topographic image indicates the height of the SLB surface (nm). (B) Results of an SLB, which has an almost flat surface. (C) Results for the case 46 when some particles were found on the SLB surface. (D) Results of an SLB, which has defects. The final 47 amounts of septin added were 6.8 (E), 12 (F), 0.85, 6.8, 10, and 12 µg (from left to right in G). In the case 48 of the AFM observation of SLB shown here, a septin solution stored at 0℃ was added to an SLB incubated at 49 25℃. Observations were performed at 25℃. In (C), (D), and (F), the height of the SLB surface (nm) 50 obtained by the line scanning in the location indicated by the white line in the topographic image is also 51 shown. 52 53 180x185mm (300 x 300 DPI) 54 55 56 57 58 59 60 ACS Paragon Plus Environment Langmuir Page 40 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 Figure 7. Model for the effect of septin on the formation and disappearance of the cooling-induced domain in 45 SLBs; i) septin filament behaving as a fence suppressing the lateral diffusion of phospholipids (upper two 46 panels), and ii) septin facilitating the nucleation by binding to the membrane and tethering PI molecules 47 (lower two panels). In each model, the upper layer is the case when septin is added to an SLB incubated at 48 25℃ and thus has not yet formed the domain, and the bottom layer is the case when septin is added to an SLB incubated at 5℃ and thus has already formed the domain. Phospholipids shown in orange and red are PI 49 and PC, respectively. We estimate that the cooling-induced domain, which has a lower fluorescent intensity, 50 is a PI-free region. Blue indicates septin filaments or oligomers (nucleus) that are assembling on the 51 membrane. Septin can interact with the lipid bilayer membrane via associating with PI, an acidic 52 phospholipid. The concentrations of septin added were restricted so as not to induce membrane tubulation 53 because membrane tubulation causes defects in SLBs. In such concentration ranges, septin might not 54 assemble into filaments. Septin did not alter the result of FRAP analysis that was carried out at room 55 temperature (Figure S2). Therefore, septin proteins might be in a state that allows diffusion together with 56 lipid molecules. Altogether, the model shown in the lower half might be more accurate than that shown in the upper half. 57 58 59 60 ACS Paragon Plus Environment Page 41 of 39 Langmuir

1 2 3 94x106mm (300 x 300 DPI) 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment