• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Oct 30, 2007; 104(44): 17370–17375.
Published online Oct 24, 2007. doi:  10.1073/pnas.0708066104
PMCID: PMC2077263
Applied Physical Sciences, Biophysics

Dynamic clustered distribution of hemagglutinin resolved at 40 nm in living cell membranes discriminates between raft theories

Abstract

Organization in biological membranes spans many orders of magnitude in length scale, but limited resolution in far-field light microscopy has impeded distinction between numerous biomembrane models. One canonical example of a heterogeneously distributed membrane protein is hemagglutinin (HA) from influenza virus, which is associated with controversial cholesterol-rich lipid rafts. Using fluorescence photoactivation localization microscopy, we are able to image distributions of tens of thousands of HA molecules with subdiffraction resolution (≈40 nm) in live and fixed fibroblasts. HA molecules form irregular clusters on length scales from ≈40 nm up to many micrometers, consistent with results from electron microscopy. In live cells, the dynamics of HA molecules within clusters is observed and quantified to determine an effective diffusion coefficient. The results are interpreted in terms of several established models of biological membranes.

Keywords: hemagglutinin, microdomains, fluorescence photoactivation localization microscopy, photoactivation, rafts

Because lateral organization in biological membranes is necessary to orchestrate cell function (1), it is crucial to determine the molecular basis of this organization, which is exploited by viruses for their entry into host cells (2). One example of a clustered membrane protein is influenza hemagglutinin (HA), which mediates membrane fusion and entry of influenza virus (3): its lateral distribution into microscopic clusters is crucial for that function (3, 4). However, the visualization of such membrane clusters in live cells has been limited by the resolution of light microscopy (5). Recent developments in subdiffraction microscopy now permit imaging of intracellular protein distributions at nanometer length scales (68). Surprisingly, fluorescence photoactivation localization microscopy (FPALM) (7) can be used in living cells to image the lateral distribution and dynamics of HA, with a demonstrated resolution of ≈40 nm and molecular position information every ≈150 ms. HA molecules are mostly mobile, indicating fluidity, not gel phase; yet the borders of their boundaries are irregular, inconsistent with expectations for liquid-ordered domains. Thus, many models of membrane lateral organization are inconsistent with the results, leaving only models that feature mobile protein molecules in fluid membranes excluded from certain regions; e.g., by the cytoskeleton or other structures. Moreover, our results suggest a means to decipher the mechanism of many biological processes with time-resolved data at nanometer resolution in living cells. In principle, these methods can be applied to image any biological structure (i) that does not remodel on time scales faster than the acquisition time, (ii) that can be labeled with a photoactivatable fluorophore, and (iii) whose structures of interest can be found partially or entirely within the focal plane of the microscope in use.

HA enables viral entry by opening a fusion pore in the host endosomal membrane, leading to the release of viral RNA into the cytoplasm (3, 9). The clustered lateral distribution of HA trimers in cell plasma membranes (10, 11) depends on cholesterol-rich raft membrane microdomains (1, 2) for budding (12) and fusion (4). However, consensus on the existence, structure, size, and dynamics of rafts continues to be elusive because of disparate results obtained by a variety of methods (13). Instead, rafts are defined principally by biochemical means (14, 15), and numerous biophysical models including lipid shells (5), temporary lateral confinement (16, 17), preferential partitioning (18), and nanodomains (15, 19) cannot be distinguished easily because they predict crucial structural details on length scales much smaller than the wavelength of visible light. Further distinction between these membrane models would provide insight into a large number of biological processes that depend on rafts (5, 15, 20).

We describe the use of FPALM (7) in living cells to image the nanoscale distribution and dynamics of HA tagged with photoactivatable green fluorescent protein (PA-GFP) (21). The range of length scales accessible to FPALM (≈40 nm to >100 μm) is crucial for direct visualization of rafts, which in live cells are proposed to be smaller than the diffraction-limited resolution of visible light (5).

Results and Discussion

FPALM can be used to visualize protein distributions in living cells at length scales much less than the diffraction-limited resolution, r0, as shown in images of PA-GFP-tagged influenza HA (PA-GFP-HA) expressed in fibroblasts and compared with confocal and widefield fluorescence microscope images (Figs. 1 and and2).2). By FPALM and confocal microscopy, EGFP-HA was both membrane-bound and intracellular (Fig. 1). During FPALM measurements, focusing the objective on the coverslip-proximal portion of the plasma membrane revealed numerous green-fluorescent molecules. Molecules were observed to flicker and blink, as well as to apparently photobleach irreversibly. Focusing further into the cell typically yielded fewer visible molecules within the focal plane and limited the size of membrane patches that could be imaged in a given focal plane. Out-of-focus molecules were sometimes intermittently visible but did not produce a bright enough signal to be localized and rather contributed to the spatially dependent fluorescence background level. A typical fixed-cell widefield image, FPALM image, and zoom-in with high density of visible PA-GFP molecules are shown in Fig. 1 A–H. A comparison between widefield fluorescence and FPALM imaging of the upper portion of the same cell is shown in Fig. 1 C–H, demonstrating significant resolution enhancement. The position of each molecule is plotted as a Gaussian spot with amplitude proportional to the observed intensity of the molecule and width of ≈40 nm (close to the estimated localization precision), consistent with the FWHM of ≈41 ± 2 nm of one of the cell processes in Fig. 1I [see supporting information (SI) Text and SI Fig. 11 for more details on rendering methods]. Note that the labeled HA distribution is clustered (Figs. 1 and and2),2), and that HA sometimes forms elongated structures (Fig. 2). Both compact and elongated clusters were observed in en face membrane sections labeled with α-HA primary antibody and 10-nm gold colloid-labeled secondary antibody, critical-point dried, and imaged by transmission EM (Fig. 2B).

Fig. 1.
Nanoscale visualization of intracellular proteins by FPALM. HA from influenza was tagged with PA-GFP, expressed in HAb2 fibroblasts, and imaged by widefield fluorescence microscopy (A) and FPALM (B), illustrating the agreement between the two methods ...
Fig. 2.
Elongated HA clusters imaged by FPALM in live cells and by EM of fixed cell membrane sheets. (A) FPALM image of a subregion of the coverslip-proximal plasma membrane of a live fibroblast, with molecular positions plotted in gray. For comparison with EM, ...

Because cellular membrane organization is expected to be highly dynamic (16, 19), we have quantified the motion of >105 single molecules of fluorescently tagged HA and analyzed that motion to test various membrane raft models. Because the diffusion coefficient of HA is very small (D ~ 0.09 μm2/s) (22, 23), FPALM can be used successfully to localize single HA molecules in live cells; the mean-squared displacement expected from two-dimensional diffusion of an HA is 0.054 μm2 in 0.15 s, compared with r02 ~ 0.078 μm2, where r0 ~ 0.28 μm is the estimated 1/e2 radius of the point-spread function (PSF) (7).

To consider shorter length and time scales (e.g., intracluster dynamics), positions of individual localized PA-GFP-HA molecules were plotted as a function of time, along with the overall distribution of molecules ultimately obtained (Fig. 3; also see SI Movies 1 and 2). Molecules are frequently visible for two to four successive frames, during which time (≈150–600 ms) they move on ≈100-nm length scales. However, these motions were found to be spatially constrained and did not span the entire lateral space available within the membrane (Fig. 3C). Instead, motions mapped out regions with elongated shapes and irregular boundaries (Fig. 3; also see SI Text and SI Figs. 12 and 13), suggesting that HAs may move along one-dimensional paths within the two-dimensional membrane. Alternatively, because some membrane regions appear to be inaccessible to HA, mobile HA molecules may undergo tethered diffusion. Such results are consistent with observations by confocal microscopy and EM that the clustered distribution of HA leaves many areas with significantly lower than average density (10).

Fig. 3.
Time dependence of positions of localized HA molecules within an HA cluster in a live fibroblast at room temperature. (A) FPALM image of a whole cell. (B) Zoomed view of area in A enclosed in dashed magenta box. (C) Successive frames from the 0.4 μm ...

Localized HA positions were analyzed for displacements from one frame to the next, and from the ith frame to the (i + k)th frame, for live and fixed cells (see SI Text for further details). The value of k can be converted into a time difference τ = τF × k, where τF is the time between successive frames. Because it is difficult to decisively identify that a particular molecule in an earlier frame is the same molecule in a later frame, all distances between each localized HA and all others in the (i + k)th frame are tabulated, and the distribution of distances squared (including displacements in both lateral directions) is plotted for very short length scales. Striking differences in distance distributions are apparent when comparing live and fixed cells (Fig. 4 A and B). The distribution in fixed cells is significantly narrower, with a peak at ≈(40 nm)2 = 1,600 nm2, whereas the peak for live cells shifts from ≈(60 nm)2 at k = 1 to (260 nm)2 at k = 4. The location of the peak does not change significantly in fixed cells, although the amplitude of the peak decreases exponentially with k, because of photobleaching of molecules after a few frames (SI Figs. 5–8). The decay of the distribution as k increases is also different when comparing the live and fixed cells, and the portion of the distribution that does decay is confined to shorter length scales in fixed cells, again indicating that little motion is occurring from frame to frame.

Fig. 4.
Distribution of radial HA–HA distances vs. time in live and fixed HAb2 fibroblasts expressing PA-GFP-HA. (A and B) Striking differences in distribution of distances squared between HA molecules in a given frame and HA molecules in the frame k ...

The significant differences between live and fixed cell distance distributions reflect the motion of HA molecules in live cells and can be used to quantify the dynamics of such molecules within the membrane. Diffusion of molecules can be described in two dimensions by a normalized Gaussian probability density function:

equation image

which predicts that the probability of finding the same molecule at a distance r from its position at time t = 0 decays exponentially with distance squared. For diffusion alone, σ2 = 2Dt, where D is the diffusion coefficient and σ is the standard deviation in position. The histogram of all distances was calculated as a function of displacement squared for live cells and fixed cells (corrected for differences in time between frames) and is shown in Fig. 4 C and D. Eq. 1 was then used to fit the distribution of displacements for r2 > 0.01 μm2 (live cells) and r2 > 4 × 10−4 μm2 (fixed cells), using σ as a free parameter (lines in Fig. 4 C and D). The fits yield σL2 = 0.0338 ± 0.002 μm2 for live cells, whereas the significantly smaller σF2 = 0.0025 ± 0.0003 μm2 for fixed cells reflects the finite localization precision. The distribution of distances between HAs in successive frames (Fig. 4B; k = 1) is peaked at a value of 0.0016 μm2 = (40 nm)2 in fixed cells, which is consistent with the localization uncertainty σx ~ 41 nm calculated due to background and limited signal photons. However, in living cells the distance distribution is significantly broadened and includes displacements much larger than σx. We interpret these results as a measure of diffusion of molecules within the time between frames (τF ~ 10−1 s). Using the variance in live cells, subtracting σF2 to correct for localization accuracy, and taking t = τF yields σL2 − σF2 ~ 2DeffτF. The form 2Dτ is used rather than 4Dτ to account for the apparently one-dimensional structures in which HAs were observed to move. The resulting effective diffusion coefficient Deff = 0.086 ± 0.018 μm2/s is within uncertainty of the values measured by fluorescence recovery after photobleaching (FRAP) in HAb2 cells (D = 0.097 ± 0.025 μm2/s) (22) and in CV-1 cells at room temperature (23). Analysis of distance histograms by using an analytical diffusion model yielded similar results (see SI Text and SI Figs. 9 and 10).

The time dependence of the distance histograms shown in Fig. 4 A and B was quantified by using Eq. 1 plus a time-independent constant offset to fit the total number of observed HA–HA distances between 0.1 μm and 0.56 μm, as a function of time delay between frames (Fig. 4E). As expected, little time dependence was observed in fixed cells. For the live cells, using the diffusion coefficient Deff = 0.086 μm2/s (from above) and r = 0.338 μm (equal to the average value of r within the range of distances that were binned) described the time dependence of the histograms, except for time differences <0.3 s. Thus, the distribution of HA–HA distances is fairly well described by a diffusion-like process (Eq. 1) on short length scales (<0.4 μm), except that the molecules cannot access certain membrane regions (Figs. 113). The deviation in Fig. 4E between the histogram of HA–HA distances and Eq. 1 in live cells indicates that more HAs were close to another HA in the next frame than would be predicted if all HAs were diffusing with Deff. However, it is known from FRAP studies of HA in cell membranes that a fraction of HA molecules (15–25%) are immobile (18, 23). Such an immobile fraction would lead to a small (≈2.3–6.3%) increase in the number of observed HA–HA distances that are shorter than predicted by diffusion. The deviation between observed and expected at τ = 0.19 s in Fig. 4E is equal to 6,270 out of 109,790, or ≈5.7%. Although this deviation is consistent with the immobile fraction determined by FRAP, we cannot be certain that it is the immobile fraction.

Ripley's K test was used to quantify the degree of HA clustering as a function of length scale. Results (Fig. 4F) show clustering of HA on all length scales tested, consistent with previous results from EM (10) extended to longer length scales in both living and fixed cells (see SI Figs. 12 and 13 for additional images). Clustering on many length scales implies that definition of a single cluster size is not possible, consistent with the finding that rafts observed by many methods have many different (seemingly contradictory) sizes (5).

Our results for HA are inconsistent with several established membrane domain models. In most cases, clusters do not have perimeter-minimized boundaries and often show elongated shapes and narrow extensions on submicrometer length scales (Figs. 113). In contrast, rounded boundaries would be expected for fluid phase domains in coexistence with another fluid phase (24), and if domains were ideally mixed, the distribution of HA would be expected to be uniform on length scales much less than the domain size (10). Rather, we observe nonrandomness on all length scales tested, which is inconsistent with partitioning into ideally mixed fluid domains with any well defined size measurable by FPALM or EM. Models that predict a tightly packed, static collection of HAs into a solid or gel phase are also inconsistent with our observations. One would expect HA inside a solid phase to be immobile, but our FPALM results show lateral motion of HA on length scales of 200 nm and smaller, and FRAP results show similar motion on longer length scales, except for a small immobile fraction (18, 23). However, FRAP cannot access details on subdiffraction length scales and does not detect differences in the behavior of individual molecules. Although normal microscopy methods (such as confocal and FRAP) can image dynamics, they have limited resolution, and although EM has excellent resolution, it is difficult to use to image dynamics. The ability to both image the clusters and the motions of individual molecules comprising those clusters on 40-nm length scales has led to the insights reported here. The observed HA clusters in live cells are much larger than just a few molecules or a few nanometers, making these results for HA inconsistent with a purely nanoscopic domain membrane model (19). However, nanometer-sized domains that are themselves part of larger clusters, or clusters with a range of sizes extending down to sizes of a few nanometers in width but not length (for example in disperse or elongated clusters where domain perimeter and area are not simply related), could describe these observations. The motion of clusters, range of cluster sizes, and diffusion-like motion of molecules within clusters suggest that the apparent diffusion rate of membrane objects will depend on the length scale being tested, explaining earlier published data (18, 25).

The observed elongated clusters and irregular domain boundaries suggest that line tension (and thus lipid fluid phase behavior) plays a limited role in domain shape (24), but interactions between the cytoskeleton and membrane proteins could certainly produce such a distribution of HA clusters within which lipids and proteins are confined but diffuse locally, such as has been suggested by Edidin, Kusumi, Vale, and others (16, 17, 26, 27). Interdependence of raft protein trafficking, lipid second messengers, and the cytoskeleton merit further investigation (2729). Finally, the demonstrated ability to visualize nanoscale intracellular dynamics suggests an even broader range of applications accessible to PALM and FPALM, in contrast to previous predictions that PALM and FPALM would not be possible in live cells (6).

Materials and Methods

Generation of pHA-PAGFP Construct.

A modified pEGFP-N1 plasmid (Clontech) containing a PA-GFP in place of EGFP (pPA-GFP) was digested with XmaI and PstI to linearize the plasmid. A pEGFP-N1 plasmid containing an N-terminal HA (X:31 strain) tag (pHA-EGFP) that had been created and described previously (18) was also digested with XmaI and PstI to generate a fragment containing the HA tag. Both digested pPA-GFP vector and HA insert fragments were gel-purified and ligated together. Transformed clones were screened by XmaI and PstI restriction digestion. Clones containing correct insert and vector fragments were sequenced at the cloning junctions to ensure correct integration of the HA tag into the GFP protein sequence. The resulting N-terminal HA-tagged photoactivatable GFP (pHA-PAGFP) construct was purified by using Maxiprep (Eppendorf) and used in subsequent experiments.

Cell Culture.

HAb2 fibroblasts were grown to ≈80% confluence on eight-well chambers with #1.5 coverslip bottom, transfected by using Lipofectamine 2000 (Invitrogen) with PA-GFP-HA, grown for an additional 24–48 h, rinsed three times with PBS plus glucose, and imaged at room temperature. Please see SI Text for further details. For fixation, cells were removed from the incubator, rinsed three times in PBS (Sigma–Aldrich), and bathed for 20–30 min in 4% paraformaldehyde (Sigma–Aldrich) in PBS at room temperature.

Lipid Labeling and Confocal Microscopy.

HAb2 and NIH 3T3 fibroblasts (the parent cell line from which the HAb2 cell was derived) (22) were transfected with pEGFP-HA (X31). Approximately 24–48 h posttransfection, cells were removed from the incubator, washed three times with PBS plus 10 mM glucose, and incubated with a warmed solution of ≈3 μM lissamine-rhodamine-B-dioleoyl-phosphatidyl-ethanolamine (Rhod-DOPE) (810150; Avanti Polar Lipids) plus 0.2% (by volume) dimethyl sulfoxide (Sigma–Aldrich).

Confocal Imaging.

A Zeiss LSM 510 inverted microscope was used at room temperature with a ×40 1.2 N.A. water objective and excitation at 488 and 568 nm. Kalman averages of 3–4 linescans were recorded. Linear image brightness and contrast were adjusted independently for the red and green channels.

FPALM Imaging.

Live or fixed cells at room temperature in an eight-well NUNC chamber with #1.5 coverslip bottom were then illuminated by two lasers: (i) the readout beam, 6–10 mW of continuous-wave Argon ion laser power at 496 nm, spread over an area at the sample of ≈1,000 μm2 to yield ≈600–1,000 W/cm2; and (ii) the activation beam, a 405-nm laser (CrystaLaser) with ≈0.5–1 mW of power spread over ≈125–250 μm2 to yield 400 W/cm2. The activation beam was aligned to illuminate the same (central) region of the field as the readout beam. The readout beam continuously illuminated the sample during data acquisition, while the activation beam was pulsed for ≈1–10 s to photoactivate PA-GFP molecules whenever the density of visible molecules within the sample declined to fewer than ≈10. Activation of PA-GFPs by the readout beam (496-nm illumination only) occurred at a slow rate but reduced the frequency at which 405-nm activation pulses were needed. Please see SI Text and SI Table 1 for further details. The lower (coverslip-proximal) surface of the cell was imaged, rather than the top surface of the cell, to maximize the area within which the plasma membrane was in focus. Photoactivated molecules were visualized by imaging the fluorescence onto a charge-coupled device camera (QuantiFire; Optronics) or an electron-multiplication charge-coupled device camera (Cascade 512B; Photometrics). Frames were acquired (i) with the QuantiFire camera using 2 × 2 binning, 0.1–0.2 s acquisition time per frame (time between frames of ≈0.25 s), and gain 6–8, and (ii) with the Cascade 512 camera using 1 × 1 binning, 0.15 s acquisition time per frame (time between frames of 0.19 s), on-chip multiplication gain 1,500–3,500, and conversion gain 6 e/ADU. For live-cell imaging, overall acquisition time was kept as short as possible (in most cases <300 s), or analysis was restricted to the first 500 frames (≈100 s) to minimize gross motion of cellular structures or the cells themselves.

Postacquisition image analysis determined molecular positions by screening images for objects above an intensity threshold (typically a few hundred photons) and below a second threshold for objects too bright or large to be single molecules. The positions of objects which were within the thresholds were then determined. First, the centroid coordinates were determined and used as the initial guess in a least-squares fit of the background-subtracted image of the object using a Gaussian profile I(x, y) = I0e−2[(xx0)2+(yy0)2]/r02, where I0 is the peak pixel value, x0 and y0 are the coordinates of the center of the fluorescent object, and r0 is the 1/e2 radius of the point-spread function (PSF). Best-fit results yielded the x–y coordinates (x0, y0) and intensity I0 (proportional to the number of detected photons) for each bright object. Either a constant background was subtracted from each image before analysis or a nonuniform background was subtracted as follows. A widefield sum was generated for a given time-series acquisition by summing (over time) all widefield images analyzed. The widefield sum image was then divided by its average intensity to generate a normalized intensity profile (an image with an average pixel value of 1). For each individual image of the acquisition, the average intensity was also calculated. The nonuniform background profile was then calculated as either 90% or 95% of the product of the intensity profile and this average intensity, and was subtracted from the original image before analysis.

Image Rendering.

Two-dimensional maps of intensity-weighted molecular positions (FPALM images) were generated from the coordinates and intensities determined by image analysis. Each molecule is represented as a spot with amplitude proportional to the number of detected photons and Gaussian radially symmetric profile of width (unless otherwise noted) 40 nm, which is approximately equal to the estimated lateral resolution. Alternatively, molecules were rendered with amplitude proportional to the number of detected photons and spot size equal to the calculated localization precision (30), using the measured background noise and number of detected photons.

Widefield imaging was carried out simultaneously in the FPALM setup (the same images are saved and analyzed to localize molecules for FPALM, or summed to generate a widefield image). The excitation wavelength λx = 496 nm and detection wavelength range λd = 510–560 nm, objective (1.2 N.A. ×60 water), and camera(s) used were therefore identical to those used for FPALM.

K-Test Analysis.

Ripley's K test (31) was performed by methods published previously (10). Raw K-test amplitudes were normalized (divided) by the 99% confidence interval for presentation and comparison; therefore, a K-test amplitude of 1 indicates a 99% probability of clustering and a 1% chance of a random distribution.

Histogram Analysis.

The histogram H(r, k) of all distances r between each localized HA and all others localized k frames later was determined for values of k = 0–10. This histogram was found to contain contributions from individual molecular motions, from the overall pattern of clustering, and from photobleaching. In live cells, the time dependence (k dependence) of H(r, k) was described well by Eq. 1 plus a constant baseline. Data taken using different frame rates were combined after the mean-squared displacement values were scaled linearly by the time between frames, which was either 190 or 270 ms for all live-cell data.

Membrane Retrieval and Preparation for Electron Microscopy.

The following methods were described previously (10). Briefly, HAb2 fibroblasts, which express the Japan strain of HA0 (here referred to as HA), were grown as a monolayer on glass coverslips (22 × 22 mm) to ≈80% confluence. Membranes were extracted by the procedure of Sanan and Anderson (32), labeled with primary antibody against HA, and fixed with 2% paraformaldehyde and 0.05% glutaraldehyde in 0.1 M Hendry's phosphate buffer for 20 min at room temperature. Alternatively, cells were fixed before labeling, and then the membranes were extracted. Secondary antibody with 10-nm colloidal gold was used to label the primary antibody and provide image contrast. Samples were critical-point dried before imaging with a transmission electron microscope (100 CX; JEOL) at ×50,000 magnification and 80 kV. Please see SI Text for further details.

Supplementary Material

Supporting Information:

Acknowledgments

We thank George Patterson for the PA-GFP construct and purified protein, Julie Gosse for help with cell culture, C. T. Hess and Dean Astumian for useful discussions, Thomas Tripp for machining, The Jackson Laboratory for an apparatus grant, and Stephen Smith and Paul Blank for laser expertise. S.T.H. was supported by National Institutes of Health Career Development Award K25AI65459 and University of Maine startup funds. This work was supported in part by the intramural program of the National Institute of Child Health and Human Development, National Institutes of Health.

Abbreviations

FPALM
fluorescence photoactivation localization microscopy
FRAP
fluorescence recovery after photobleaching
PA-GFP
photoactivatable green fluorescent protein.

Footnotes

The authors declare no conflict of interest.

This article contains supporting information online at www.pnas.org/cgi/content/full/0708066104/DC1.

References

1. Simons K, Ikonen E. Nature. 1997;387:569–572. [PubMed]
2. Manes S, del Real G, Martinez C. Nat Rev Immunol. 2003;3:557–568. [PubMed]
3. White J, Helenius A, Gething MJ. Nature. 1982;300:658–659. [PubMed]
4. Takeda M, Leser GP, Russell CJ, Lamb RA. Proc Natl Acad Sci USA. 2003;100:14610–14617. [PMC free article] [PubMed]
5. Anderson RGW, Jacobson K. Science. 2002;296:1821–1825. [PubMed]
6. Betzig E, Patterson GH, Sougrat R, Lindwasser OW, Olenych S, Bonifacino JS, Davidson MW, Lippincott-Schwartz J, Hess HF. Science. 2006;313:1642–1645. [PubMed]
7. Hess ST, Girirajan TPK, Mason MD. Biophys J. 2006;91:4258–4272. [PMC free article] [PubMed]
8. Klar TA, Engel E, Hell SW. Phys Rev E. 2001;64 066613. [PubMed]
9. Wiley DC, Skehel JJ. Annu Rev Biochem. 1987;56:365–394. [PubMed]
10. Hess ST, Kumar M, Verma A, Farrington J, Kenworthy A, Zimmerberg J. J Cell Biol. 2005;169:965–976. [PMC free article] [PubMed]
11. Ruigrok RW, Krijgsman PC, de Ronde-Verloop FM, de Jong JC. Virus Res. 1985;3:69–76. [PubMed]
12. Scheiffele P, Rietveld A, Wilk T, Simons K. J Biol Chem. 1999;274:2038–2044. [PubMed]
13. Mahaffey DT, Moore MS, Brodsky FM, Anderson RGW. J Cell Biol. 1989;108:1615–1624. [PMC free article] [PubMed]
14. Brown DA, London E. J Biol Chem. 2000;275:17221–17224. [PubMed]
15. Brown DA. Physiology (Bethesda) 2006;21:430–439. [PubMed]
16. Fujiwara T, Ritchie K, Murakoshi H, Jacobson K, Kusumi A. J Cell Biol. 2002;157:1071–1081. [PMC free article] [PubMed]
17. Nakada C, Ritchie K, Oba Y, Nakamura M, Hotta Y, Iino R, Kasai RS, Yamaguchi K, Fujiwara T, Kusumi A. Nat Cell Biol. 2003;5:626–632. [PubMed]
18. Kenworthy AK, Nichols BJ, Remmert CL, Hendrix GM, Kumar M, Zimmerberg J, Lippincott-Schwartz J. J Cell Biol. 2004;165:735–746. [PMC free article] [PubMed]
19. Sharma P, Varma R, Sarasij RC, Ira, Gousset K, Krishnamoorthy G, Rao M, Mayor S. Cell. 2004;116:577–589. [PubMed]
20. Edidin M. Nat Rev Mol Cell Biol. 2003;4:414–418. [PubMed]
21. Patterson GH, Lippincott-Schwartz J. Science. 2002;297:1873–1877. [PubMed]
22. Ellens H, Bentz J, Mason D, Zhang F, White JM. Biochemistry. 1990;29:9697–9707. [PubMed]
23. Shvartsman DE, Kotler M, Tall RD, Roth MG, Henis YI. J Cell Biol. 2003;163:879–999. [PMC free article] [PubMed]
24. Baumgart T, Hess ST, Webb WW. Nature. 2003;425:821–824. [PubMed]
25. Lenne PF, Wawrezinieck L, Conchonaud F, Wurtz O, Boned A, Guo XJ, Rigneault H, He HT, Marguet D. EMBO J. 2006;25:3245–3256. [PMC free article] [PubMed]
26. Douglass AD, Vale RD. Cell. 2005;121:937–950. [PMC free article] [PubMed]
27. Kwik J, Boyle S, Fooksman D, Margolis L, Sheetz MP, Edidin M. Proc Natl Acad Sci USA. 2003;100:13964–13969. [PMC free article] [PubMed]
28. del Pozo MA, Alderson NB, Kiosses WB, Chiang HH, Anderson RG, Schwartz MA. Science. 2004;303:839–842. [PubMed]
29. Heo WD, Inoue T, Park WS, Kim ML, Park BO, Wandless TJ, Meyer T. Science. 2006;314:1458–1461. [PMC free article] [PubMed]
30. Thompson RE, Larson DR, Webb WW. Biophys J. 2002;82:2775–2783. [PMC free article] [PubMed]
31. Ripley BD. J R Stat Soc B. 1979;41:368–374.
32. Sanan DA, Anderson RGW. J Histochem Cytochem. 1991;39:1017–1024. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...