Current Biology, Vol. 14, 1002–1006, June 8, 2004, 2004 Elsevier Ltd. All rights reserved. DOI 10.1016/j .cub.2004.05.044

A Barrier to Lateral Diffusionin the Cleavage Furrowof Dividing Mammalian Cells

provides an experimentally tractable way of measuringdiffusion in live cell membranes [16, 17]. Diffusion con-stants determined by FRAP agree well with those de-rived from other techniques applied over similar timescales and distances [5, 8, 10]. As a first step, diffusion

Katja Schmidt and Benjamin J. Nichols*MRC Laboratory of Molecular BiologyHills RoadCambridge, CB2 2QHUnited Kingdom

was measured in interphase HeLa and NRK cells. FRAPof a 4 �m circle allowed measurement of the rate oflateral diffusion (D) and mobile fraction (MF, degree ofSummaryrecovery as determined by fitting data as a single expo-nential. Figure 1C and Table 1) [17]. Rates of diffusionBarriers to diffusion of proteins and lipids play an im-varied from 0.7 �m2/s to 0.15 �m2/s in the order DiI�Lyn-portant role in generating functionally specialized re-GFP�YFP-Gl-GPI�LYFPGT46. Diffusion was slightlygions of the plasma membrane. Such barriers havefaster in NRK than HeLa cells, but the overall relationshipbeen reported at the base of axons, at the bud neckbetween rates of diffusion for the different markers wasin Saccharomyces cerevisiae, as well as at the tightrecapitulated in both cell types. These values of D arejunctions of epithelia [1–3]. How diffusion barriers arecomparable to those obtained previously for similar pro-formed and how they effect behavior of both inner andteins [17, 18]. In order to confirm that observed valuesouter leaflets of the bilayer are not fully understood.of D are not influenced by photobleaching conditions,Here, we provide evidence for a cortical barrier toD for Lyn-GFP and YFP-Gl-GPI was calculated both bydiffusion within the cleavage furrow of mammalianusing minimal photobleaching light and by using an or-cells. Photobleaching-based assays were used toder of magnitude more light (10 or 100 laser scans; seemeasure diffusion of three membrane proteins withExperimental Procedures). This made no difference todiffering topologies and putative lipid raft association,the observed value of D (not shown). Additionally, D foras well as the lipid analog dialkylindocarbocyanine (DiIa version of YFP-Gl-GPI where YFP was replaced byC18, [4]), across the cleavage furrow [5, 6]. There wasphotoactivatable GFP (PAFP-Gl-GPI) [19] was calcu-a block in diffusion of proteins with a cytosolic domain,lated by activation within a 4 �m circle in HeLa cellsbut not of proteins anchored in the outer leaflet of the(Figure S1). D for PAFP-Gl-GPI was 0.18 � 0.02, thePM or of DiI. Diffusion of lipid raft proteins in the innersame as that for YFP-Gl-GPI. As photoactivation re-and outer leaflets of the membrane was not directlyquires approximately 10-fold less light than photo-coupled. The distribution of Septin proteins, as op-bleaching, we can conclude that variation of illuminationposed to cortical actin, was consistent with a func-during photoactivation, and photobleaching by 100-foldtional role in limiting diffusion [2, 7].has no effect on the observed D.

Latrunculin B was used to depolymerise actin andResults and Discussion thereby determine the role of cortical actin filaments

in modulating diffusion in both HeLa and NRK cells.Models in which lateral diffusion of plasma membrane Concentrations of Latrunculin that were sufficient to(PM) constituents is restricted or compartmentalized by cause loss of visible actin bundles and drastic changesinteraction of the membrane with the cortical cytoskele- in cell shape had little or no effect on lateral diffusionton predict that in situations where the cortical cytoskel- (Table 1). Thus it is unlikely that cortical actin filamentseton is especially dense and organized, a corresponding play a major role in restricting diffusion in interphasereduction in lateral diffusion will be observed [1, 8–10]. cells [8, 9]. We reasoned that effects of the corticalWe reasoned that the cleavage furrow of dividing cells cytoskeleton on diffusion should be more apparentpotentially constitutes one such situation [11]. In order when many cytoskeletal components are concentratedto test this hypothesis, we measured diffusion of three together in the cleavage furrow of dividing cells.different types of fluorescent protein, as well as an exog- Photobleaching was used to assay lateral diffusion ofenous fluorescent lipid analog, dialkylindocarbocyanine YFP-Gl-GPI, LYFPGT46, Lyn-GFP, and DiI across the(DiI C18 [4]) (Figure 1A). YFP-Gl-GPI is embedded in the cleavage furrow in HeLa cells. One half of the dividingouter leaflet of the PM via a glycosylphosphatidylinositol cell was bleached, and the loss of fluorescence fromanchor, Lyn-GFP is embedded in the inner leaflet by the unbleached half was quantified over time (Figuresmyristyl and palmityl moieties [12], and LYFPGT46 is a 2A and 2B). This approach avoided problems associatedtransmembrane protein [13–15]. YFP-Gl-GPI and Lyn- with quantifying recovery in the photobleached region,GFP are both found in detergent-resistant membrane particularly diffusion of out-of-focus fluorescence. Pho-fractions (biochemically defined lipid rafts [6]), so this tobleaching did not induce any detectable delay or per-combination of markers allowed us to investigate the turbation in cytokinesis. In order to demonstrate thatcontribution of both raft microdomains and membrane cytokinesis was not complete in any of the cells analyzedorientation in modulating diffusion. in this way, soluble, cytosolic GFP was also bleached,

Fluorescence recovery after photobleaching (FRAP) and diffusion through the cleavage furrow was mea-sured (Figures 2A and 2B). No diffusion of LYFPGT46and Lyn-GFP across the cleavage furrow was detected.*Correspondance: ben@mrc-lmb.cam.ac.uk

A Diffusion Barrier in the Cleavage Furrow1003

was readily detected and occurred at a similar rate todiffusion of this protein during interphase (Figures 2Band 2C). Similarly, diffusion of DiI across the cleavagefurrow was readily detected (Figure 2C). Thus diffusionacross the cleavage furrow of PM constituents that havea cytosolic domain appears to be much less rapid thandiffusion at other regions of the plasma membrane,whereas diffusion in the outer leaflet of the plasma mem-brane is apparently unaffected.

We sought direct confirmation of the existence of adiffusion barrier for LYFPGT46 and Lyn-GFP, but notYFP-Gl-GPI, in the cleavage furrow. The morphologyand highly dynamic nature of the plasma membraneduring early stages of furrow ingression made it difficultto visualize the relative distributions of the three pro-teins. Later during cytokinesis, however, the dividingcells remain attached and relatively stationary over sev-eral minutes. During this time an optically dense struc-ture, the midbody, is observed between the cells [11].Photobleaching and recovery of cytosolic GFP showedthat the cytosol is still continuous at this stage (datanot shown). High-resolution images of thin, confocalsections revealed that although there was marked accu-mulation of LYFPGT46 in intracellular membranes oneither side of the midbody, the region of the PM incontact with the midbody was completely devoid ofLYFPGT46 (Figure 3A). Therefore LYFPGT46 does notdiffuse into this area of the PM. Cytosolic GFP wasclearly visible in the midbody, confirming that the cytosolFigure 1. Lateral Mobility of Plasma Membrane Proteinsis still continuous at this stage (Figure 3B). Moreover,(A) Distribution of Lyn-GFP (associated with the inner leaflet of theYFP-Gl-GPI was present at constant density all the wayplasma membrane), YFP-Gl-GPI (attached to the outer leaflet), the

transmembrane protein LYFPGT46, and the fluorescent lipid analog across the midbody and clearly delineated the regionDiI in NRK cells. Images are single confocal sections of the base of of the membrane devoid of LYFPGT46 (Figure 3C). DiIthe cell. Similar distributions were seen in Hela cells. An asterisk had a similar distribution to YFP-Gl-GPI (not shown). Asindicates position of G/YFP in the construct. Bar � 25 �m.

in the case of LYFPGT46, Lyn-GFP fluorescence was(B) Representative photobleach experiment for Lyn-GFP: a 4.1 �mnot detected in the PM at the midbody (Figure 3D). Thuscircle was bleached and recovery of fluorescence monitored overLYFPGT46 and Lyn-GFP cannot diffuse into or acrosstime.the midbody region. This agrees well with the photo-(C) Recovery curves of the different constructs. The ratio between

mean fluorescence intensity of the bleached circle and the mean bleaching experiments described above, though it isfluorescence of the whole cell was normalized to the prebleach ratio important to emphasize that the photobleaching dataand expressed as a function of time (Experimental Procedures). The imply that restriction of diffusion of LYFPGT46 and Lyn-data were fitted to a single exponential curve (solid lines) to calculate

GFP begins prior to assembly of the midbody. Interest-diffusion coefficients (D) and mobile fractions (MF, Table 1) n � 8 �ingly, expression of a YFP-Septin2 fusion protein re-SD. Curve fits were to complete data sets for 150 s recovery.vealed that the region of the intercellular bridge devoidof LCFPGT46 fluorescence contains Septin2, a member

In contrast, despite the fact YFP-Gl-GPI diffuses slower of the Septin family of proteins thought to restrict diffu-than Lyn-GFP during interphase (Figure 1C and Table sion during cytokinesis in budding yeast [2, 3] (Figure

3E). Antibody staining confirmed the presence of endog-1), diffusion of YFP-Gl-GPI across the cleavage furrow

Table 1. Lateral Mobility of Plasma Membrane Proteins Is Not Affected by the Actin Cytoskeleton

Control 50 nM Latrunculin 150 nM Latrunculin

D (�m2/s)a MF (%)b D (�m2/s) MF (%) D (�m2/s) MF (%)

Lyn-GFP NRK 0.40 � 0.08 85 � 5.9 0.38 � 0.08 85 � 6.6 0.36 � 0.04 87 � 3.7HeLa 0.29 � 0.05 76 � 6.5 0.30 � 0.10 79 � 7.1 0.29 � 0.06 78 � 6.9

YFP-Gl-GPI NRK 0.23 � 0.04 81 � 5.0 0.22 � 0.03 81 � 5.3 0.24 � 0.05 78 � 9.7HeLa 0.19 � 0.06 74 � 8.3 0.23 � 0.05 75 � 6.6 0.25 � 0.05 77 � 8.9

LYFPGT46 NRK 0.16 � 0.02 84 � 5.8 0.18 � 0.02 81 � 5.0 0.17 � 0.02 73 � 8.8HeLa 0.16 � 0.05 74 � 8.3 0.14 � 0.01 78 � 5.6 0.13 � 0.04 68 � 8.8

DiI NRK 0.75 � 0.04 80 � 5.0 0.69 � 0.07 79 � 5.3 0.72 � 0.06 73 � 9.7HeLa 0.58 � 0.05 82 � 7.1 0.55 � 0.09 77 � 6.6 0.54 � 0.06 77 � 5.9

a Diffusion coefficient expressed as mean � SD.b Mobile fraction (mean � SD).

Current Biology1004

Figure 2. A Diffusion Barrier in the Cleavage Furrow of Dividing Cells

(A) Photobleach experiment. One daughter cell of dividing Hela cells cotransfected with soluble GFP and LYFPGT46 was bleached by using20 full intensity scans from the 488 and 514 nm lines of a 30 mW Argon laser. Fluorescence intensities were monitored over time. GFP andYFP emission were unmixed by using Zeiss META. Note soluble GFP is evenly distributed between the daughter cells after 500 s postbleach,whereas the fluorescence intensity of LYFPGT46 is unchanged in the unbleached daughter. Recovery of LYFPGT46 fluorescence in theunbleached daughter is due to incomplete bleaching. Images show one focal plane of a Z stack. Bar � 10 �m.(B) Quantification of the loss of fluorescence in unbleached daughter cell. The ratio between mean fluorescence intensities of unbleacheddaughter (ROI1 in [A]) and mean fluorescence intensities of both daughters (ROI2) was normalized to the prebleach ratio and expressed asa percentage of the first postbleach ratio (Experimental Procedures). No loss of LYFPGT46 fluorescence in the unbleached daughter of dividingcells indicates a block of lateral mobility across the cleavage furrow (black solid line). As a control, one half of an interphase cell was bleached.The loss of fluorescence in the unbleached half over time shows “free” diffusion of LYFPGT46 (black dashed line). Gray lines show theunrestricted diffusion of soluble GFP from the unbleached daughter/half to the bleached region. Data are from one typical experiment.(C) Lateral mobility of membrane proteins with a cytoplasmic domain is impaired across the cleavage furrow. The data were analyzed as in(B). Mean fluorescence intensities in the unbleached daughter of dividing cells after different time points were compared with those in theunbleached half of interphase cells. The small apparent decrease observed for Lyn-GFP in dividing cells was not statistically significant(p � 0.05 for comparison of 150 s and 1200 s time points in two-tailed T test).

enous Septin2, but not actin, in the optically dense mid- sue-culture cells reveal that localized diffusion over dis-tances not resolved by FRAP (�0.5 �m) is likely to bebody (Figure 3F). This indicates that Septin2 may play

a direct role in the formation of the midbody diffusion faster than is measured by longer time and distance-scale FRAP experiments [8–10, 13, 16]. Compartmental-barrier.

The combination of FRAP and direct observation em- ization of the membrane by cortical barriers, stearic orcrowding effects due to high protein density, and theployed in this study allows us to make three firm conclu-

sions: that there is a barrier to free diffusion of membrane presence of membrane microdomains have all been in-voked as nonexclusive explanations for this phenome-components in the cleavage furrow in dividing mamma-

lian cells, that this barrier is effective in restricting diffu- non [5, 8, 10]. The lack of a strong effect on diffusionupon Latrunculin treatment indicates that the polymer-sion of membrane proteins that have a cytosolic domain

but has little or no effect on diffusion of markers for the ization state of cortical actin does not have a directeffect on lateral diffusion. It may be that other corticalouter leaflet of the PM, and that partitioning of Lyn-GFP

and YFP-Gl-GPI into the same biochemically defined proteins are also important, or that residual actin poly-mers that are not completely disassembled under themembrane microdomains, or lipid rafts [14, 15], does

not lead to direct and stable coupling of the diffusion conditions we have used can still significantly restrictdiffusion [8]. There has been some controversy over theof these proteins in the inner and outer leaflets of the

membrane, respectively. The finding that soluble, cyto- extent to which diffusion of GPI-linked proteins, lipidanalogs, and transmembrane proteins across the axon-solic GFP diffuses freely through the cleavage furrow

even at late stages in cytokinesis means that the most soma junction of neurons is blocked [1, 20, 21]. Ourfindings emphasize that a block in diffusion in the innerlikely explanation for these observations is that diffusion

of the cytosolic domain proteins is impeded by the high leaflet of the membrane does not directly lead to equiva-lent effects in the outer leaflet.density of specific cortical proteins in the cleavage

furrow. In S. cerevisiae, Septin proteins are thought to berequired for block in diffusion of some proteins acrossAlthough it is necessary to be cautious in extrapolat-

ing from the specific case of diffusion in the cleavage the bud neck [2, 3]. We have shown that the localizationof the ubiquitous mammalian Septin, Septin2 [7, 22], isfurrow, our data are pertinent to other situations where

diffusion is restricted or anomalous. Studies applying fully consistent with these proteins having a similar rolein mammalian cells. Various approaches to perturbingsingle-molecule tracking approaches in interphase tis-

A Diffusion Barrier in the Cleavage Furrow1005

play an important role in generation of asymmetry duringcell division [23], and further functional studies will beneeded to test this hypothesis. Moreover, it is possiblethat equivalent or analogous machinery is used to gener-ate restricted diffusion in other situations, for example,during establishment of polarity within the plasma mem-brane [23, 24].

Experimental Procedures

Constructs and AntibodiesYFP-Gl-GPI and LYFPGT46 constructs were from P. Keller [13, 15],and Lyn-GFP was from T. Meyer. Plasmids pEG/YFP-N1 (ClontechLaboratories) were used to express soluble G/YFP. The LCFPGT46-expressing construct was produced as described [25]. YFP in YFP-Gl-GPI was replaced with photoactivateable GFP [19] by using AgeIand BsrGI sites to produce PAFP-Gl-GPI. Anti-Septin2 polyclonalantibody was a gift from M. Kinoshita.

Transfection, Drug Treatment, Synchronization,Immunofluorescence, and DiI LabelingTransfections were carried out with Fugene6 (Roche). HeLa cellswere synchronized by a thymidine-nocodazole block 20 hr aftertransfection. To disrupt the actin cytoskeleton, NRK cells were incu-bated in imaging medium (DMEM without phenol red, 10% FCS,50 mM HEPES [pH 7.2]) containing 50 nM and 150 nM Latrunculin B(Molecular Probes), respectively. DiI was introduced into the plasmamembrane by using Vybrant DiI labeling kit (Molecular Probes) ac-cording to the manufacturer’s apostrophe symbol instructions. La-beling was for 10 min at room temperature, cells were examinedwithin 20 min to minimalize internalization of the lipid probe.

Microscopy and PhotobleachingAll imaging experiments employed a Zeiss LSM510 META confocalmicroscope equipped with a 40 mW Argon laser (514 nm and 488 nmlines), a 5 mW HeNe laser (546 nm line), and an 80 mW Kryptonlaser (413 nm line). Images of fixed cells were taken with a 63� 1.4NA objective with the confocal pinhole set to one Airy unit. Live-cell microscopy was carried out at 35�C in imaging medium. Photo-bleaching experiments were performed with the confocal zoom setto 3 and a 40� 1.3 NA objective lens with the confocal pinhole setto 1–3 Airy units. To assay the lateral mobility of YFP-Gl-GPI, Lyn-

Figure 3. Localization of Plasma Membrane Proteins in the Cleav- GFP, and LYFPGT46 in the plasma membrane of NRK cells, a circleage Furrow of interest (r � 4.1 �m) was photobleached using 25 scans with the

488 nm laser line for GFP and the 514 nm laser line for YFP at full(A) Distribution of LYFPGT46 in diving cells. LYFPGT46 is excludedlaser power. DiI was bleached by using 40 scans with 514 nm andfrom the DIC-dense midbody structure (arrow). Image is a projection546 nm laser lines. Activation of photoactivateable GFP was carriedof 25 thin, confocal slices.out by using one scan with the 413 nm line [19]. Pre- and postbleach(B) In contrast to LYFPGT46 (red), soluble GFP (green) is detectableimages were monitored at low laser-intensity scanning for at leastthroughout the cleavage furrow.150 s (pixel time; 0.96 �s, 8 bit). Fluorescence recovery in the(C) Localization of LCFPGT46 (red) and YFP-Gl-GPI (green) is dis-bleached region, the diameter of the bleach region, and the overalltinct. Unlike LCFPGT46 (red), YFP-Gl-GPI (green) evenly localizesfluorescence in the whole cell during the time series were quantifiedto the whole membrane in the cleavage furrow (arrow). A singleby using the Zeiss LSM software. After subtracting the backgroundconfocal section is shown.(� mean fluorescence intensity in the bleached region after bleach)(D) Lyn-GFP (red) is also excluded from the plasma membrane adja-the ratio between mean fluorescence intensity of the bleached circlecent to the midbody (arrow). A single confocal section is shown.and the mean fluorescence of the whole cell was expressed as a(E) YFP-Septin2 (green) is present in the midbody region (arrow)percentage of the prebleach ratio of these values. These normalizeddevoid of LCFPGT46 (red).data were fitted to a single exponential curve by using the PRISM(F) Septin2 (red) is detectable by indirect immunofluorescence insoftware (GraphPad Software Inc., San Diego) to derive mobile frac-the midbody (arrow). Actin (green) was stained with phallacidin. Ation MF (� amount of recovery) and characteristic diffusion timesingle confocal section is shown. Bars � 2 �m.tD, which indicates the time at which half of the fluorescence hasrecovered. Diffusion coefficients were then calculated from tD byusing D � 0.224 � r2/tD [26].

Septin function, including siRNA and use of truncation To monitor the lateral mobility of membrane proteins across thecleavage furrow of dividing cells, Hela cells were cotransfected withor point mutants, cause drastic changes to other com-soluble G/YFP and a G/YFP plasma membrane construct (pEGFP-ponents of the cortical cytoskeleton including actin bun-N1 YFP-Gl-GPI, pEGFP-N1 LYFPGT46, or pEYFP-N1 and Lyn-dles ([7], our unpublished data), so further experimentsGFP). Free diffusion of soluble cytoplasmic G/YFP between thewill be required to rigorously confirm the identity of mo-daughter cells was used to check whether dividing cells were still

lecular components responsible for mediating restricted connected. One daughter cell of dividing cells was photobleacheddiffusion across the cleavage furrow. The presence of by using 20 scans with the 488 and 514 laser line at full laser power.

Z stacks of pre- and postbleach images (four to five slices/time point)a diffusion barrier during cytokinesis could potentially

Current Biology1006

were taken with the 488 laser line at low laser-intensity scanning a of live cells as revealed by single molecule techniques (Review).Mol. Membr. Biol. 20, 13–18.128 � 128 pixel frame every 4.2 s (pixel time: 12.8 �s, 12 bit). YFP

and GFP emission spectra were acquired and linearly unmixed by 10. Saxton, M.J., and Jacobson, K. (1997). Single-particle tracking:applications to membrane dynamics. Annu. Rev. Biophys. Bio-using Zeiss META detector and software. Mean fluorescence inten-

sities in the unbleached region and the overall fluorescence in the mol. Struct. 26, 373–399.11. Glotzer, M. (2001). Animal cell cytokinesis. Annu. Rev. Cell Dev.whole cell (� both daughters together) during the time series were

quantified by using the Zeiss LSM software. Overall prebleach fluo- Biol. 17, 351–386.12. Kovarova, M., Tolar, P., Arudchandran, R., Draberova, L., Rivera,rescence intensities of all images of a Z stack were added and only

cells expressing the same amount of membrane protein and soluble J., and Draber, P. (2001). Structure-function analysis of Lynkinase association with lipid rafts and initiation of early signalingprotein (GFP/YFP ratio � 0.9–1.1) were analyzed. In order to confirm

that intracellular fluorescence makes a negligible contribution, in events after Fcepsilon receptor I aggregation. Mol. Cell. Biol.21, 8318–8328.some cases selected small regions of PM fluorescence were mea-

sured instead of all of the cell/unbleached half of the cell—this made 13. Pralle, A., Keller, P., Florin, E.L., Simons, K., and Horber, J.K.(2000). Sphingolipid-cholesterol rafts diffuse as small entitiesno difference to the results. Because of the depth of dividing cells,

bleaching was incomplete, and diffusion of out-of-focus fluores- in the plasma membrane of mammalian cells. J. Cell Biol. 148,997–1008.cence contributed to recovery in the bleached daughter cell. There-

fore fluorescence recovery was not an appropriate measure to 14. Pyenta, P.S., Holowka, D., and Baird, B. (2001). Cross-correla-tion analysis of inner-leaflet-anchored green fluorescent proteinanalyze diffusion from one daughter to the other. The loss of fluores-

cence in the unbleached daughter was calculated instead. The ratio co-redistributed with IgE receptors and outer leaflet lipid raftcomponents. Biophys. J. 80, 2120–2132.between mean fluorescence intensities of unbleached daughter

(ROI1) and mean fluorescence intensities of both daughters (ROI2) 15. Keller, P., Toomre, D., Diaz, E., White, J., and Simons, K. (2001).Multicolour imaging of post-Golgi sorting and trafficking in livewas normalized by subtracting to set the prebleach ratio to zero,

rescaling so that first postbleach ratio was one, and then adding cells. Nat. Cell Biol. 3, 140–149.16. Feder, T.J., Brust-Mascher, I., Slattery, J.P., Baird, B., andone. The normalized loss of fluorescence in the unbleached region

was plotted as a function of time. Webb, W.W. (1996). Constrained diffusion or immobile fractionon cell surfaces: a new interpretation. Biophys. J. 70, 2767–2773.Supplemental Data

17. Lippincott-Schwartz, J., Snapp, E., and Kenworthy, A. (2001).A supplemental figure is available at http://www.current-biology.com/Studying protein dynamics in living cells. Nat. Rev. Mol. Cellcgi/content/full/14/11/1002/DC1/.Biol. 2, 444–456.

18. Niv, H., Gutman, O., Kloog, Y., and Henis, Y.I. (2002). ActivatedAcknowledgments K-Ras and H-Ras display different interactions with saturable

nonraft sites at the surface of live cells. J. Cell Biol. 157, 865–872.Many thanks to H. Pelham and S. Munro for critical comments 19. Patterson, G.H., and Lippincott-Schwartz, J. (2002). A photoac-on the manuscript and to M. Kinoshita for generously providing tivatable GFP for selective photolabeling of proteins and cells.antibodies. Science 297, 1873–1877.

20. Howard, M.R., Spiller, D., Reed, J.E., McNamee, C., White, M.,and Moss, D.J. (2003). No barrier to diffusion between cell somaReceived: February 11, 2004

Revised: March 18, 2004 and neurite membranes in sympathetic neurons for a GPI-anchored glycoprotein. Mol. Cell. Neurosci. 24, 296–306.Accepted: March 29, 2004

Published: June 8, 2004 21. Winckler, B., Forscher, P., and Mellman, I. (1999). A diffusionbarrier maintains distribution of membrane proteins in polarizedneurons. Nature 397, 698–701.References

22. Xie, H., Surka, M., Howard, J., and Trimble, W.S. (1999). Charac-terization of the mammalian septin H5: distinct patterns of cy-1. Nakada, C., Ritchie, K., Oba, Y., Nakamura, M., Hotta, Y., Iino,toskeletal and membrane association from other septin pro-R., Kasai, R.S., Yamaguchi, K., Fujiwara, T., and Kusumi, A.teins. Cell Motil. Cytoskeleton 43, 52–62.(2003). Accumulation of anchored proteins forms membrane

23. Knoblich, J.A. (2001). Asymmetric cell division during animaldiffusion barriers during neuronal polarization. Nat. Cell Biol. 5,development. Nat. Rev. Mol. Cell Biol. 2, 11–20.626–632.

24. Lecuit, T., and Wieschaus, E. (2000). Polarized insertion of new2. Takizawa, P.A., DeRisi, J.L., Wilhelm, J.E., and Vale, R.D. (2000).membrane from a cytoplasmic reservoir during cleavage of thePlasma membrane compartmentalization in yeast by messen-Drosophila embryo. J. Cell Biol. 150, 849–860.ger RNA transport and a septin diffusion barrier. Science 290,

25. Glebov, O.O., and Nichols, B.J. (2004). Lipid raft proteins have341–344.a random distribution during localized activation of T-cell recep-3. Barral, Y., Mermall, V., Mooseker, M.S., and Snyder, M. (2000).tor. Nat. Cell Biol. 6, 238–243.Compartmentalization of the cell cortex by septins is required

26. Adams, C.L., Chen, Y.T., Smith, S.J., and Nelson, W.J. (1998).for maintenance of cell polarity in yeast. Mol. Cell 5, 841–851.Mechanisms of epithelial cell-cell adhesion and cell compaction4. Mukherjee, S., Soe, T.T., and Maxfield, F.R. (1999). Endocyticrevealed by high-resolution tracking of E-cadherin-green fluo-sorting of lipid analogues differing solely in the chemistry ofrescent protein. J. Cell Biol. 142, 1105–1119.their hydrophobic tails. J. Cell Biol. 144, 1271–1284.

5. Dietrich, C., Yang, B., Fujiwara, T., Kusumi, A., and Jacobson,K. (2002). Relationship of lipid rafts to transient confinementzones detected by single particle tracking. Biophys. J. 82,274–284.

6. Simons, K., and Ikonen, E. (1997). Functional rafts in cell mem-branes. Nature 387, 569–572.

7. Kinoshita, M., Field, C.M., Coughlin, M.L., Straight, A.F., andMitchison, T.J. (2002). Self- and actin-templated assembly ofMammalian septins. Dev. Cell 3, 791–802.

8. Fujiwara, T., Ritchie, K., Murakoshi, H., Jacobson, K., and Ku-sumi, A. (2002). Phospholipids undergo hop diffusion in com-partmentalized cell membrane. J. Cell Biol. 157, 1071–1081.

9. Ritchie, K., Iino, R., Fujiwara, T., Murase, K., and Kusumi, A.(2003). The fence and picket structure of the plasma membrane