The intracellular mobility of a viral membrane glycoprotein measured by confocal microscope fluorescence recovery after photobleaching

The structure of all biological membranes is fundamentally similar, consisting of a lipid bilayer in which proteins are embedded and able to diffuse. Diffusion properties of membrane proteins have so far been characterized in detail only for the plasma membrane, by a series of physical techniques including fluorescence recovery after photobleaching (FRAP; for reviews see Hennis, 1989; Jacobson et al., 1987). From these studies, a series of general principles have emerged leading to the conclusion that in vivo plasma membrane proteins are indeed mobile. However, their diffusion rates are typically much lower (10- to 30-fold) than expected for free diffusion of a protein in a lipid bilayer and, in fact, a significant fraction of most plasma membrane proteins is immobile (Jacobson et al., 1987). The reduced mobility of plasma membrane proteins is thought to be due to different combinations of interactions of membrane proteins with each other and with components of the cytoskeleton and the extracellular matrix (Edidin, 1993; Lee et al., 1993; Zhang et al., 1991). The degree to which protein-protein and other possible interactions, such as inclusion in glycolipid rafts (Simons and Wandinger-Ness, 1990), may affect the dynamic properties of proteins in intracellular membranes in vivo is unknown because of lack of reagents and limited accessiblity. To date, techniques such as FRAP, which have been applied to the characterization of membrane proteins at the cell surface, have not been successfully used for studying dynamic properties of intracellular membrane proteins.

Here, we demonstrate the validity of a FRAP approach for measuring the mobility of the temperature-sensitive mutant of the vesicular stomatitis virus glycoprotein (ts-O45-G), a model intracellular transmembrane protein. This protein is well characterized and has been a powerful tool in dissecting steps along the secretory pathway. At nonpermissive temperature (39.5°C), ts-O45-G is blocked in the endoplasmic reticulum (ER) and at 15°C or 20°C accumulates in the intermediate compartment (Schweizer et al., 1990) or trans-Golgi network (TGN; Griffiths and Simons, 1986), respectively. Shifting infected cells to the permissive temperature (31°C) releases the block and induces transport of ts-O45-G to the cell surface. Transport of ts-O45-G from the ER to the Golgi apparatus also depends on coat proteins (COPs) associated with nonclathrin-coated transport vesicles. Microinjected antibodies against β-COP, a protein homologous to the clathrin adaptor protein β-adaptin (Duden et al., 1991), block transport of the viral glycoproteins at the interface between the ER and the Golgi apparatus (Pepperkok et al., 1993).

Most importantly for the present study, we have established previously that the monoclonal antibody P5D4 directed against the cytoplasmic domain of ts-O45-G binds with high affinity to the viral glycoprotein in vivo and that its Fab fragments have no effect on the transport of the glycoprotein through the secretory pathway (Kreis, 1986). In the present experiments, microinjected rhodamine-conjugated P5D4 Fab fragments (rh-P5D4-Fabs) have been used to visualize the mobility of ts-O45-G in the TGN and while in transit from the ER to the TGN. The primary outcome of this work has been to demon-strate the applicability of FRAP to the study of intracellular membrane proteins. We show by FRAP that ts-O45-G is highly mobile in intracellular membranes, even at lower incubation temperatures when transport to the surface is inhibited. However, a significant fraction of the model viral glycoprotein is immobilized under conditions where its transport to the cell surface is blocked by injected antibodies to β-COP, yet the diffusion coefficient of mobile membrane protein remains unchanged. The possibility of generalizing this approach to other resident or transient transmembrane proteins through the use of epitope-tagged proteins is discussed.

Monolayer Vero cells (African green monkey kidney cells, ATCC CCL81) were cultured as described previously (Kreis, 1986). For experimental use, cells were plated in 10 cm plastic tissue culture dishes containing 22 mm circular glass coverslips and maintained in culture for 2 days. Cultures were about 80% confluent. Coverslips were washed in methanol and, for microinjection experiments, inscribed with a centrally located ‘X’ prior to autoclaving.

A temperature-sensitive VSV mutant viral stock (ts-O45 VSV, Indiana serotype, gift from Dr Kai Simons, EMBL, Heidelberg) was prepared by propagation of the virus at an initial concentration of 0.5 pfu/Vero cell. Three days post-infection the culture supernatant was harvested and clarified by centrifugation at 600 g_avfor 10 minutes followed by filtration through a 0.22 μm filter. The resulting ts-O45 VSV stock was divided into aliquots, quick frozen in liquid N₂and stored at a −70°C. For experimental use, stock viruses were diluted into three volumes of Dulbecco’s phosphate-buffered saline (PBS) containing 1% fetal bovine serum, adsorbed to coverslip cultured Vero cells for 1 hour at room temperature, and cells were further incubated in regular culture medium at the temperatures and for the times indicated.

P5D4, an affinity-purified murine monoclonal antibody (mAb) against peptide P-15 comprising the carboxyl-terminal 15 amino acids of the cytoplasmic tail of VSV-G (Kreis, 1986); anti-110-12, anti-EAGE and anti-A1 affinity-purified rabbit antibodies against peptides comprising amino acids 391-411, 496-513 and 1161-1175, respectively, of rat β-COP (Pepperkok et al., 1993); and 1A2, a murine mAb against tubulin (Kreis, 1987), have all been described previously. Rabbit anti-VSV-G antibody was a gift from Dr Kai Simons, EMBL, Heidelberg. Secondary antibodies were luorescein- or rhodamine-labeled goat antibodies against mouse or rabbit IgG (Kreis, 1986). Affinity-purified Fab fragments of anti-EAGE were prepared as described (Pepperkok et al., 1993).

Rhodamine-conjugated affinity-purified P5D4 and its Fab fragments (rh-P5D4-Fabs) were prepared by modifications of previously published procedures (Kreis, 1986). Briefly, P5D4 antibody was affinity-purified from ascites fluid on P-15 coupled to Sephadex, the peak eluant fractions containing purified antibody pooled, dialyzed against PBS, and coupled with rhodamine. Rhodamine-conjugated antibody was separated from free dye over a Sephadex G-50 (Pharmacia-LKB) column and concentrated by a DE52 (Whatman) column. The conjugate had an A₂₈₀/A₅₇₅ratio of 2.4, equivalent to ∼1.5 moles rhodamine per mole IgG. Following digestion with papain for 2 hours at 37°C, small digestion products were removed from larger fragments by Sephadex G-50 chromatography in 30 mM NaCl, 10 mM sodium phosphate, pH 8.0, and the rhodamine-conjugated Fab fraction subsequently collected as the flow-through of a DE52 column equilibrated in the same buffer. Aliquots of the rh-P5D4-Fabs were quick frozen with liquid N₂and stored at −70°C. The concentration of the rh-P5D4-Fabs was 0.6 mg/ml.

Cultured Vero cells grown on glass coverslips were injected either with rh-P5D4-Fabs or with rh-P5D4-Fabs and antibodies against β-COP using the Zeiss automated injection system (AIS, Carl Zeiss, Oberkochen, Germany). For each experimental condition, 50-60 cells near the intersection point of the coverslip mark were injected. The computer-controlled dwell time of the injection capillary in the cells was 0.1 second. ts-O45 VSV-infected cells were injected either at 39.5°C, 2-2.5 hours post-infection, or at 20°C following a 2 hour shift to 20°C during which time the ts-O45-G accumulates in the TGN (Matlin and Simons, 1983). For injections at 39.5°C, the microscope stage was fitted with a water-jacketed tissue culture dish holder. Following a 30 minute antibody binding period at the injection temperature, cells were either processed immediately for fluorescence microscopy/confocal laser photobleaching, or shifted from nonper-missive to permissive temperature (39.5°C to 31°C) in complete culture medium. Stock solutions of rh-P5D4-Fabs were brought to 0.1 M KCl before microinjection. In double-antibody microinjection experiments, rh-P5D4-Fabs were mixed: one part rh-P5D4-Fab stock solution plus two parts anti-β-COP antibody. The starting concentrations of anti-β-COP antibodies were as described by Pepperkok et al. (1993). All antibody solutions were clarified by centrifugation prior to microinjection.

For direct P5D4 labeling, ts-O45 VSV-infected Vero cells were fixed with methanol followed by acetone at −20°C (Kreis, 1986) and then incubated with various dilutions of rh-P5D4-Fabs. In double-label experiments, virally infected cells were first injected with rh-P5D4-Fabs at 39.5°C or after a 2 hour chase at 20°C, then fixed either immediately or after a brief chase at 31°C and finally labeled with rabbit anti-VSV-G antibody as described (Pepperkok et al., 1993). In some experiments, cells were treated with nocodazole to depolymerize microtubules (Ho et al., 1990), fixed with methanol-acetone as described above, and labeled with 1A2 followed by rhodamine-labeled second antibody. These preparations were mounted in Hanks’ balanced saline solution and the coverslips mounted on dried nailpolish posts. Labeled cells were viewed by either conventional or confocal fluorescence microscopy.

The Spectra Physics 2016 multiline argon laser (Mt View, CA, USA) of the EMBL compact confocal microscope (CCM; Stelzer et al., 1991) was used with an ×100 Zeiss infinity corrected, fluorite objective (NA, 1.3; Carl Zeiss, Oberkochen, Germany). Rhodamine was excited with the 529 nm laser line (total laser power was 1-1.2 watts), the dichroic mirror in the CCM had a cutoff of 530 nm and the emission filter in front of the detector was a 570 nm long-pass filter. Fluorescein was excited with the 476 nm laser line (the total laser power was 0.01-0.1 watt), the dichoic mirror had a cutoff of 500 nm, and the emission filter was a 25 nm wide band-pass filter centered at 530 nm. For samples labeled with rh-P5D4-Fabs, a focal plane was selected that passed through the central distribution of fluorescence. Double-labeled samples were first illuminated in the fluorescein channel and then, in the same focal plane, in the rhodamine channel. The laser line and power, dichroic mirror and emission filter, and detector photomultiplier gain were selected as a set each time the respective dye was illuminated. 512×512 pixels, 256 gray level images were collected in the XYor XZplane at a raster scan rate of 900 lines per second (pixel dwell time, 720 nanoseconds). Images were averaged during collection and stored on floppy discs and/or photographed using Polaroid freeze-frame equipment (Polaroid, Cambridge, MA, USA). Digitally stored images were printed to negatives using an Agfa SlideWriter at 4000 lines resolution.

Coverslips were mounted in Hanks’ balanced saline solution supplemented with 1% fetal calf serum and non-essential amino acids. Cells on coverslips were mounted on dried nail-polish posts and sealed with nail-polish. Observation temperatures were 20-25°C (ambient temperature).

Rhodamine photobleaching was performed with the EMBL CCM. A 30 μm×30 μm field was imaged. This corresponded to the approximate area of a single cell. The bleached area was selected manually using the computer mouse to define on a pre-bleach XYconfocal image of the cell the path of laser movement. Bleaching was achieved by exposing the designated cellular path of diffraction limited width to the laser for 30 milliseconds at each of 10 evenly spaced points along a linear path. Typically, the bleached path was ∼2 μm long. The XYcoordinates of the bleached path were automatically reported by the microscope system. In most experiments, the three confocal microscope control functions: timeseries, bleachpath and saveseries, were automatically linked; typically, two images of the cell were collected before the bleachpath procedure and 10 images were collected afterwards. The first post-bleach image was collected ∼1 second following the bleachpath procedure. Because of the time required for scanning the entire 30 μm×30 μm field, data could be collected no more rapidly than once every 2 seconds. In some experiments, pre- and post-bleach images were collected individually under manual control.

Fluorescence images were further processed and quantified using the software program NIH Image (version 1.44, Wayne Rasband, author, National Institute of Mental Health, NIH, Bethesda, MD or [email protected],Internet or Bitnet) on an Apple Macintosh computer. The width of the bleach path was determined by manual inspection of images displayed on a computer monitor at a 2-4 times zoom factor. Typically, fluorescence intensity was quantified for a 0.4 μm wide block of pixels along the length of the bleach path. Average pixel brightness in the bleach area was corrected for any photobleaching effect during repeated observations of the field and for variations in laser illumination intensity by dividing the average intensity observed in the bleach area by the average pixel intensity measured over two nonbleached areas. Image series exhibiting any variation in focal plane, i.e. change in nuclear diameter or shape of the rh-P5D4-Fab-stained area within the cell in an animated play-back of the image set, were not scored. For each experimental condition, half-times for fluorescence recovery were calculated individually from an exponential curve fit for each image series and then averaged. Lateral diffusion coefficients (D) were calculated as previously described by Axelrod et al. (1976)with g factor corrections for bleach intensity. The immobile fraction was determined from the observed and/or extrapolated fluorescence recovery at 110 seconds according to the formula:

F_(im)=(FI_prebleach−FI₁₁₀)/(FI_prebleach−FI_postbleach)×100,

where F_(im)is the immobile fraction and FIis fluorescence intensity before (pre), after 110 seconds (110) and immediately after bleaching (post). Periods of recording were limited to 110 seconds to avoid photobleaching artefacts and any shift in microscope stage height relative to the objective.

Our experimental goal was to characterize in vivo the mobility of a newly synthesized transmembrane protein en route to the plasma membrane. We have used ts-O45-G as a model protein in Vero cells and temperature blocks or microinjected antibodies against β-COP to modulate its transport. ts-O45-G was visualized in living cells by microinjected rh-P5D4-Fabs and its mobility was assessed by FRAP using the confocal microscope. We present first experiments validating the FRAP approach and then describe the effect of microinjected anti-β-COP antibodies on the mobility of ts-O45-G in cytoplasmic membranes at the interface of the ER and the cisside of the Golgi complex.

Initial experiments were aimed at characterizing the system. Rhodamine (rh-P5D4-Fabs) was excited with the argon laser of the confocal microscope. To photobleach, the laser dwell time was increased ∼45,000-fold (from 720 nanoseconds to 30 milliseconds) along a line of 1-3 μm consisting of 10 evenly spaced spots. Bleaching a line rather than a single spot was chosen as the most effective and reproducible method of determining the mobility of a membrane protein within a complex organelle in the cytoplasm of a living cell.

To determine the geometry of the bleached line, an evenly distributed cytoplasmic marker was generated by solubilizing tubulin by treating cells with nocodazole. Tubulin was labeled with rhodamine by indirect immunofluorescence and subsequently a line was photobleached in these fixed cells (Fig. 1). The extent of fluorescence photobleaching was 40-70%. This resulted in a continuous, narrow bleach line with an average width of 0.52±0.02 μm (s.e.m., n=16) and an average depth in the Zdimension of 4.01±0.31 μm (s.e.m., n=11). Such a bleach-line was sufficient to create a bisecting photobleach plane through the Golgi apparatus. The width of the path bleached by the laser is consistent with the optical properties of the microscope. For the standard microscope objective (×100; NA, 1.3), the smallest light spot in the plane of focus is 0.46 μm (Stelzer et al., 1991).

The use of the confocal microscope for intracellular FRAP experiments also requires the ability to collect time series of high-resolution images rapidly with little loss of fluorescence and to distinguish the possible contribution of soluble fluorophore from membrane-associated fluorophore. In fixed cells, with rhodamine-stained depolymerized tubulin, little loss in fluorescence was apparent if seven 512×512 images were collected with limited image averaging (2×: 5-10%, 4×: 15-20%). These values of bleaching during image acquisition were in fact lower in the living uninfected cells microinjected with rh-P5D4-Fabs (10 images 4x averaged: ∼5%; Fig. 2). Further, averaging was not necessary for image analysis and increased bleaching significantly. Therefore, all series of images in the subsequent FRAP experiments were collected with 2- or 4-fold averaging. Acquisition of a 4-fold averaged 512×512 image to disk required 2.3 seconds, which is sufficient to determine the rate and extent of membrane-associated fluorescence recovery within the bleached line for rh-P5D4-Fabs (see below). It should be noted that higher apparent photostability of soluble rh-P5D4-Fabs may reflect in part their high mobility. Under our line bleach conditions, the 30 millisecond laser dwell time produced no detectable photobleaching of microinjected rh-P5D4-Fabs in noninfected cells. Diffusion of soluble rh-P5D4-Fabs must be so rapid that full diffusion recovery has already occurred during the first ∼1 second (averaging of two frames) required by the confocal microscope to collect the first post-bleach image.

Rhodamine-labeled Fab fragments of a monoclonal antibody reacting with the cytoplasmic tail of VSV-G (rh-P5D4-Fabs) were microinjected into cells to analyze the dynamic properties of the viral glycoprotein in vivo. Approximately 80% of the rh-P5D4-Fabs bound to beads conjugated with the peptide corresponding to the cytoplasmic tail of VSV-G (P-15; Kreis, 1986). As judged by SDS-polyacrylamide gel electrophoresis, the rh-P5D4-Fab preparation was depleted of intact divalent antibody and contained no free rhodamine (Fig. 3). When injected into infected cells, the rh-P5D4-Fabs labeled all the compartments containing the viral glycoprotein. An ER-like pattern was visualized when cells were injected and kept at 39.5°C, and when subsequently shifted for 15 minutes to 31°C in the presence of cycloheximide (100 μg/ml), membranes of the Golgi complex and intermediate compartment were labeled (Fig. 4, see also Kreis, 1986). These patterns are identical to those obtained by immunofluorescence labeling with a rabbit polyclonal antibody against VSV-G (Fig. 4). When injected, virus-infected cells were shifted from nonpermissive to permissive temperature, ts-O45-G was transported with normal kinetics to the cell surface (data not shown, see also Kreis, 1986). We conclude that rh-P5D4-Fabs can be used as fluorescent tracers for ts-O45-G in living Vero cells.

For the initial measurements of ts-O45-G mobility in vivo, infected cells were injected with rh-P5D4-Fabs at 20°C with ts-O45-G accumulated in the TGN in the presence of cyclo-heximide. These initial conditions were chosen for reasons of convenience with respect to both the stable concentration of ts-O45-G in a defined subcellular compartment (Fig. 5) and the ease of visualization of the fluorescence at 20-25°C (i.e. ambient temperature). Under these conditions, FRAP of rh-P5D4-Fabs bound to ts-O45-G revealed visually considerable recovery within 16 seconds, while organelle morphology appeared constant (Fig. 5). Initial fluorescence was measured within the first two seconds after bleaching. Quantification of the fluorescence recovery in this specific cell revealed 75% mobility of the viral glycoprotein within 21 seconds (Fig. 6).

The mobility of ts-O45-G within membranes of the Golgi complex was further determined quantitatively under two different conditions. In the first, the mobility of ts-O45-G accumulated in the TGN at 20°C was determined and quantified in eight cells as described above (Fig. 7, Table 1). In the second, cells in which ts-O45-G was accumulated for 2.5 hours in the ER were shifted in the presence of cycloheximide for at least 10 minutes to 31°C to allow its transport to the intermediate compartment and the Golgi complex. Injection with rh-P5D4-Fabs was at 39.5°C. Quantification was for 9 cells (Fig. 7, Table 1). TGN-associated ts-O45-G was essentially fully mobile over a 110 second observation period at room temperature (immobile fraction 3-4%, 20°C data extrapolated to 110 seconds). No significant immobilization of the viral glycoprotein accumulated in the intermediate compartment and the Golgi complex was detectable using this methodology. Due to technical limitations of the confocal microscopy set-up used for these experiments, measurements of FRAP at 39.5°C (non-permissive temperature, with ts-O45-G blocked in the ER) were impossible.

Several factors may have contributed to fluorescence recovery in these experiments. The first and most likely is lateral diffusion of ts-O45-G/rh-P5D4-Fab conjugates within the membranes of the Golgi complex. Nonetheless, this contribution may be confounded by other mechanisms such as diffusion of soluble rh-P5D4-Fabs into the bleach area, delivery of labeled protein into the organelle, or replacement of bleached Fabs by Fabs from the nonbleached soluble pool. Contribution of soluble rh-P5D4-Fab diffusion to the measured fluorescence recovery in the Golgi membranes is insignificant. This system is incapable of measuring such fast diffusion, about one second or less (see above, Fig. 2). The fluorescence recovery observed is over tens of seconds. It is unlikely that exchange of Fabs or delivery of labeled protein to the Golgi complex contributed to fluorescence recovery. rh-P5D4-Fabs remain bound to the antigen during direct immunofluorescence where cells are repeatedly washed with a large excess of PBS (see also, Fig. 4). FRAP experiments were performed in the presence of cycloheximide, an inhibitor of protein synthesis; no increase in fluorescence intensity in nonbleached regions of the Golgi complex was detected (data not shown). Furthermore, no recovery occurs in photobleached isolated Golgi elements detached from the body of the Golgi complex (data not shown), in agreement with previous work with fluorescent lipid-labeled Golgi membranes (Cooper et al.,1990). We conclude that membrane continuity with the main body of the organellar complex is essential for fluorescence recovery and that this method of FRAP using the confocal fluorescence microscope allows the quantitative measurement of the mobility of a transmembrane protein within intracellular organelles.

A series of anti-peptide antibodies against rat β-COP have previously been generated and characterized (Duden et al., 1991; Pepperkok et al., 1993). Two (anti-EAGE and anti-110-12) bind to β-COP associated with cytoplasmic membranes in microinjected Vero cells; another (anti-A1) appears not to react with membrane associated β-COP (Pepperkok et al., 1993). Anti-EAGE and its Fab fragments block transport of ts-O45-G at the interface of the ER and the Golgi apparatus; in contrast, anti-110-12, has no effect on ts-O45-G transport (Pepperkok et al., 1993).

We have characterized the effect of anti-EAGE (and its Fab fragments), anti-110-12 and anti-A1 (control) on the mobility of ts-O45-G following its release from the ER. In all experiments, affinity-purified antibodies were injected together with the rh-P5D4-Fabs at 39.5°C, following a 2.5 hour accumulation of ts-O45-G protein in the ER. Cells were then shifted, in the presence of cycloheximide, to 31°C for 10 minutes. As shown in Fig. 8, the distribution of the viral glycoprotein was affected by the microinjected anti-EAGE, but not by microinjected anti-110-12 or anti-A1 antibodies. In cells injected with anti-EAGE, ts-O45-G accumulated in tubular membrane structures at the interface of the ER and Golgi apparatus (arrows in Fig. 8F; see also Pepperkok et al., 1993). In agreement with previous experiments (Pepperkok et al., 1993), anti-EAGE (co-injected with rh-P5D4-Fabs) blocked transport of ts-O45-G from the ER to the cell surface (data not shown).

FRAP associated with mobility of labeled ts-O45-G accumulated in the region of the Golgi complex (see Fig. 8, arrrow-heads in F point to typical areas for photobleaching in the case of EAGE-injected cells) was incomplete when infected cells were co-injected with rh-P5D4-Fabs and intact antibodies capable of binding β-COP in vivo (Fig. 9, Table 1). Co-injection of anti-EAGE or anti-110-12 revealed an immobile fraction after 110 seconds of 30% and 27%, respectively. This is in contrast to cells co-injected with anti-A1 or anti-EAGE-Fabs, where the immobile fraction was 6% or 12%, respectively. Virtually no immobile fraction was observed when infected cells were injected with rh-P5D4-Fabs alone (4%). We conclude that anti-EAGE and anti-110-12 cause the immobilization of a significant fraction of the viral glycoprotein. With respect to the mobile fraction of ts-O45-G, for which no interaction with β-COP antibodies is expected, neither the anti-EAGE antibody and its Fab fragments, anti-110-12 or anti-A1, had any significant effect on the lateral diffusion coefficient of ts-O45-G (Table 1).

We have probed, by a confocal microscope FRAP approach, the mobility of a newly synthesized intracellular model membrane protein (ts-O45-G) en route to the plasma membrane. Using microinjected rh-P5D4-Fabs directed against the cytoplasmic tail of ts-O45-G to visualize its dynamic properties, we found that the viral glycoprotein displays similar high mobility following release from the ER and after accumulation in the TGN at 20°C. Only when COPs were affected by co-injected antibodies binding to β-COP in vivo did we observe a significant effect on the mobility of ts-O45-G: the pool of mobile ts-O45-G was reduced by ∼30%. These results indicate that membrane proteins in transit to the cell surface are significantly more mobile in intracellular membranes than at the plasma membrane (Zhang et al., 1991; Fire et al., 1991; Lee et al., 1993). This may be a general trait of cell surface transmembrane proteins.

Technically, our experimental approach is novel. We visualized the intracellular dynamic properties of a transmembrane protein through the behavior of microinjected fluorescent Fab fragments. So far FRAP experiments have probed the mobility of membrane proteins at the cell surface, where excess antibodies can be washed away. In this work, the high affinity of rh-P5D4-Fabs for the cytoplasmic tail of ts-O45-G was sufficient; free intracellular Fabs made no contribution to the observed fluorescence recovery of the viral glycoprotein. Most FRAP experiments employ either spot photobleaching (e.g. see Axelrod et al., 1976) or fringe pattern photobleaching (e.g. see Davoust et al., 1982). Here, we used a line photobleaching procedure in which a series of partially overlapping, sequential spots are bleached to give a continuous line bisecting the organelle. Because the Golgi apparatus and associated perinuclear organelles have a complex structure with an irregular shape and often consist of tubular elements, this was, in our opinion, the optimal bleaching procedure; it allows fitting the bleaching path to the geometry of the organelle and averaging fluorescence recovery over several tubular structures. Finally, out of focal plane fluorescence contributions were eliminated with the confocal fluorescence microscopy. This is an important consideration for intracellular organelles that are a few μm in thickness.

Using this approach, we observed an average Dof ∼10× 10⁻¹⁰cm²/s for ts-O45-G irrespective of blockade of its transport at 20°C within the TGN or the presence of antibodies that block membrane traffic from the ER to the Golgi complex. This value is about 3-fold higher than those reported for the viral glycoproteins at the cell surface. The Dvalues of transfected VSV-G protein and influenza HA expressed at the cell surface are 3.8×10⁻¹⁰cm²/s (22°C, Zhang et al., 1991) and 3.3×10⁻¹⁰cm²/s (22°C, Fire et al., 1991), respectively. We also found that, in the absence of perturbations, little intracellular ts-O45-G protein was immobile. In striking contrast, ∼50% of transfected cell surface VSV-G is immobile (Zhang et al., 1991), probably due to the interaction of cell surface viral glycoprotein with extracellular matrix components (Zhang et al., 1991; Lee et al., 1993).

Interestingly, microinjected antibodies capable of binding to membrane-associated β-COP in vivo (anti-EAGE; anti-110-12; Pepperkok et al., 1993) immobilized ∼30% of the intracellular ts-O45-G. This effect is specific; Fab fragments or β-COP antibodies that do not bind to the antigen in vivo produced insignificant immobilization of ts-O45-G. Most likely, anti-EAGE and anti-110-12 lead to crosslinking (but not precipitation) of membrane associated β-COP. This result suggests that the cytoplasmic domain of the viral glycoprotein may interact directly or indirectly with the COP complex and is consistent with previous evidence that the cytoplasmic domain of VSV-G plays a role in its transport to the cell surface. Microinjected Fab fragments of polyclonal antibodies directed against the cytoplasmic tail of ts-O45-G block its transport to the cell surface (Kreis, 1986). It should also be noted that, in a possibly analogous case, perturbation of clathrin-coated pits results in the cell surface immobilization of a mutant of influenza hemag-glutinin containing tyrosine in its cytoplasmic domain (Fire et al., 1991).

The observation that injection of both anti-EAGE and anti-110-12 leads to significant immobilization of ts-O45-G yet only one of them (anti-EAGE) interferes with membrane traffic, appears paradoxical. The following possibility offers, in our opinion, a simple plausible explanation for this finding. Microinjected antibodies binding to β-COP in vivo may reduce the dynamic properties of the coatomer and putative associated proteins (e.g. by crosslinking β-COP subunits) and may lead to the formation of a more rigid matrix at the exit sites at the ER to Golgi interface. This leads to transient (<2 minutes, the time period of FRAP) entrapment and immobilization of a fraction (∼30%) of cargo membrane protein, including ts-O45-G. The observation that binding of an antibody against the cytoplasmic tail (αP4; Kreis, 1986) also interferes with transport of ts-O45-G is consistent with the possibility that an interaction of cargo with components of a membrane-associated coat structure is essential for transport. Injected anti-110-12 does not further interfere with transport, since the budding machinery remains unaffected and membrane traffic proceeds apparently undisturbed (over a time period of >30 minutes; Pepperkok et al., 1993). In contrast, anti-EAGE, which inhibits membrane traffic at the boundary of the intermediate compartment and the Golgi, in addition interferes with the formation of the obligatory transport intermediates delivering ts-O45-G to the cis-Golgi cisternae. Anti-EAGE may prevent binding of additional factors to the coatomer and thus prevent formation of a functional COP complex, which is essential for transport (Pepperkok et al., 1993; Peter et al., 1993).

In conclusion, we describe a novel approach for the analysis of the intracellular mobility of organellar proteins. With this approach we could show that, in contrast to the plasma membrane, ts-O45-G is highly mobile in intracellular membranes. In addition, antibodies that bind to β-COP in vivo immobilize ∼30% of ts-O45-G, raising the possibility of an interaction between nonclathrin-coated vesicle cargo membrane proteins and components of the COP complex. We have applied this approach to the specific case of ts-O45-G; it may, however, easily be generalized by (transient or stable) expression of various epitope-tagged proteins in cells. Several high-affinity antibodies against defined epitopes are available and may thus be used to characterize the intracellular mobility of a wide range of organellar transmembrane proteins.

We express our appreciation to Brigitte Joggerst-Thomalla and Pekka Hänninen for their excellent technical assistance. We thank W. Ansorge for providing the microinjection facilities. We appreciate the helpful comments of Ariel Blocker, Jean Gruenberg, Bernard Hoflack and Kai Simons on the manuscript. B.S. was supported by a Fogarty Senior International Fellowship from the U. S. National Institutes of Health while at EMBL, and R.P. in part by an EMBO long-term postdoctoral fellowship. This work was completed with support by grants from the US National Science Foundation (DCB-9022817; to B.S.) and the Swiss National Science Foundation (31-33546.92; to T.E.K.).