The length of cargo-protein transmembrane segments drives secretory transport by facilitating cargo concentration in export domains

The membranes of secretory organelles maintain their distinct composition despite ongoing and extensive fluxes of membrane exchange. This dynamic homeostasis is mediated by specific interactions involving targeting sequences within the cargo protein and the coat complexes of the secretory machinery. The organelle membranes are the platform for all of these transport steps. Integral cargo proteins are selected and concentrated within differentiated export membrane domains (Kuge et al., 1994). These domains then mature and bud to form vesicular carriers that translocate and fuse with the membranes of target organelles (Presley et al., 1997; Scales et al., 1997). The transmembrane domain (TMD) of the secretory proteins also has transport and targeting functions. It is hypothesized that a gradual change in the composition and thickness of the secretory organelle membranes constitutes the platform for the targeting function of the TMD. Mismatch interactions between TMDs and the hydrophobic tails of membrane lipids are key to delineating these processes (Mitra et al., 2004). The targeting of Golgi-resident proteins by a distinct length of TMD was the first example of these interactions (Munro, 1995). More recently, endoplasmic reticulum (ER) export of tail-anchored proteins was demonstrated to be facilitated by a sufficiently long TMD (Ronchi et al., 2008). The thermoreversible mutant model cargo protein, vesicular stomatitis virus G tsO45 (VSVG) requires both an intact TMD (Cole et al., 1998; Guan et al., 1988) and a COPII-interacting acidic motif (Miller et al., 2003; Nishimura and Balch, 1997) at its cytosolic tail to exit the ER and arrive at the plasma membrane (PM).

This study was designed to gain in-depth information on the contribution of TMD to the transport and sorting processes. Live-cell analysis was applied to delineate the secretory transport and sorting dynamics of YFP-tagged VSVG with a TMD that was elongated (by two or four amino acids) or shortened in a stepwise fashion (two amino acids per step, from one to nine amino acids). Secretory transport was analyzed by quantifying the Golgi fluorescence after shift to the permissive temperature (32°C). Golgi export rate was specifically calculated using a 20°C temperature block to compare the control VSVG, an elongated TMD mutant (In2) and the shortest TMD mutant that still arrives at the PM (ΔTM7). Quantitative live-cell analysis was also applied to quantify the concentrations of the mutant proteins in the ER exit sites (ERESs). Results showed that the speed of secretory transport is associated with the length of the TMD. ER-to-ERES sorting analysis demonstrated that the TMD determines the extent of the cargo's concentration in export domains. These data are discussed in terms of how the TMD and targeting signals orchestrate transport processes.

The relationship between TMD length and sorting and transport dynamics was analyzed using the thermoreversible cargo protein YFP-VSVG transiently expressed in living COS-7 cells. Site-directed mutagenesis was applied to generate a series of deletions or additions within the predicted TMD, and to replace the acidic amino acids within the cytosolic DXE motif with alanines, as detailed in Fig. 1A. The amino acids selected for deletion were Val, Ile or Leu, known to be associated with intra-membrane α-helix formation. As demonstrated in Fig. 1B with cells incubated overnight at the non-permissive temperature (39.5°C) and imaged 30 minutes after shift to the permissive temperature (32°C), the DXE-deficient mutant (YFP-VSVGtsO45-AIA) was retained in the ER, whereas YFP-VSVG localized to the Golgi complex or to scattered mini-Golgi stacks in cells with depolymerized microtubules. Initially, we wanted to verify the arrival of the various mutants at the PM by prolonged incubation at the permissive temperature. The mutants In2 and In4 with extended TMDs localized exclusively to the PM after 24 hours incubation at the permissive temperature of 32°C (data not shown). Fig. 1C shows images of typical cells incubated at the permissive temperature for 24 (top panel) or 48 (lower panel) hours after transfection with plasmids coding for the various YFP-VSVG deletion mutants. Two distinct intracellular distributions were observed for the deletion mutants (Fig. 1C): they either localized to the PM or were retained within ER membranes. The ER retention of ΔTM8 and ΔTM9 did not differ between 24 and 48 hours in the presence or absence (inserts in lower panel) of 100 μg/ml cycloheximide, a protein-synthesis inhibitor. To confirm that ΔTM8 and ΔTM9 were still transmembrane proteins and had not changed their topology to become ER lumina-soluble proteins, we imaged cells expressing ΔTM8, ΔTM9 or CFP with hen egg lysozyme signal sequence (ssCFP) before and after mild permeabilization with 0.1% (v/v) saponin. Since the saponin-treated membranes still contained ΔTM8 or ΔTM9 and not ssCFP (data not shown), we concluded that ΔTM8 and ΔTM9 are transmembrane proteins. These data confirmed that VSVG with a predicted TMD longer than 15 amino acids is ultimately targeted to the PM. The mutant ΔTM1a has a single deletion of Leu478, which is the eighth amino acid deleted from ΔTM7 that confers its ER retention at all temperatures. The ΔTM1a mutant was further used in this study to demonstrate that ER retention of ΔTM8 (TMD with deletion of eight amino acids) results entirely from a cumulative effect of TMD shortening, rather than from the absence of that particular amino acid. Based on these results, we carried out all of our analyses at the permissive temperature of 32°C, thereby ensuring consistent arrival of all ER-export-competent deletion mutants (up to seven amino acids deleted) at the PM.

Next, we carried out experiments to follow the secretory transport of the TMD deletion or addition mutants. Cells transfected with the various YFP-VSVG mutants were incubated overnight at the non-permissive temperature (39.5°C). Time-lapse digital images were collected as previously described (Vasserman et al., 2006) for approximately 3 hours at 30-second intervals. Fig. 2A shows images of representative cells at different times after the shift to the permissive temperature, each cell expressing one of the deletion or addition mutants (see also Movies 1 and 2 in supplementary material). The images in Fig. 2A demonstrate that shortening the TMD of VSVG significantly slows down its transport: the shorter the TMD, the longer it took for the ER to be depleted of its VSVG content. In addition, Golgi-to-PM transport also slowed, as Golgi fluorescence became brighter at 180 minutes after the shift with shorter TMDs. To perform a semi-quantitative analysis of the transport dynamics of the TMD deletion or addition mutants, we focused on transport through the Golgi complex – the intermediate organelle between the ER and the PM. Accordingly, at any given time during secretory transport of VSVG, the fluorescence intensity of the Golgi is a consequence of the flux of incoming cargo from the ER and the efflux of cargo being exported to the PM. Golgi fluorescence intensity was analyzed using regions of interest (ROIs) surrounding the Golgi complex, as indicated by the colored circles in Fig. 2A. The plots in Fig. 2B demonstrate averaged normalized profiles of the time-based changes in fluorescence intensity of a ROI surrounding the Golgi in cells expressing the insertion and deletion mutations In4 and In2 (top), VSVG control, ΔTM1 and ΔTM1a (middle), and ΔTM3 ΔTM5 and ΔTM7 mutants (bottom). The graph in Fig. 2C shows the average values without error bars to demonstrate the relationship between TMD length and transport through the Golgi complex. Shortening the TMD has two effects on the time-dependent change in Golgi fluorescence intensity: (1) peak fluorescence in the Golgi complex occurs at later times, after the shift to the permissive temperature; (2) the slopes of the increase and decrease in fluorescence intensity become flatter. Extending the TMD by two or four amino acids affected the profiles of Golgi fluorescence intensity as these changed in a steeper manner compared to control VSVG (Fig. 2B,C). In addition, visualization of time-lapse movies of In2 seems to show a faster arrival at the PM compared with controls (data not shown). These data clearly demonstrate an association between secretory transport rate and the length of the TMD of VSVG.

We next determined the effect of changing TMD length on the Golgi exit rate in isolation. We accumulated YFP-VSVG, YFP-In2 or YFP-ΔTM7 in the Golgi by shifting cells expressing each one of the three mutants from 39.5°C to 20°C for 3 hours (Griffiths et al., 1989). Cells were then shifted back to 32°C and time-lapse images were captured (Fig. 3A). Although the Golgi effluxes of YFP-VSVG and In2 were comparable, the Golgi export of ΔTM7 was significantly slower. The images in Fig. 3A demonstrate that in addition to the slow emptying of ΔTM7 from the Golgi, the post-Golgi carriers that bud and translocate to the cell periphery fail to fuse with the PM and therefore accumulate. This is clearly seen in Movie 3 in supplementary material.

The semi-log plot in Fig. 3B demonstrates the difference in export rates of YFP-In2 and YFP-VSVG versus YFP-ΔTM7. The experimental data (filled circles) of Golgi export after a temperature block at 20°C were used to obtain rate coefficients for the Golgi-to-PM transport of YFP-VSVG, YFP-In2 and YFP-ΔTM7 using SAAM II modeling and simulation software. The model consisted of two compartments (Golgi and PM) and a single rate coefficient to represent Golgi-to-PM transport (Fig. 3B). Data of both Golgi and PM fluorescence intensities were used for the fitting process. The average exit rate constants obtained from the simulations were 2.01±0.2% per minute (n=11), 2.3±0.3% per minute (n=14) and 0.90±0.1% per minute (n=13) for YFP-VSVG, YFP-In2 and YFP-ΔTM7, respectively. These data demonstrate that shortening the TMD of VSVG slows cargo efflux from the Golgi complex.

The scheme in Fig. 4C demonstrates a plot of four arbitrary lines generated with Eqn 1 that differ only in the value of the parameter a which corresponds to the maximum variance value representing the concentrated cargo state in the ERES. The sorting and concentration dynamics of the various TMD deletion mutants and the In2 addition mutant were compared by analyzing the time-dependent increase in the fluorescence-intensity variance within a ROI surrounding a single cell. Fig. 4D shows averaged normalized variance values for nocodazole-treated cells expressing the control YFP-VSVG, In2 and the deletion mutations, including the ER-retained ΔTM8. The continuous lines represent the fit with Eqn 1. The curve-fitting parameters (R² values) for In2, YFP-VSVG and ΔTM1-7 were between 0.99 and 0.98. For the In2 mutant, curve-fitting was applied only to the data obtained during the initial 22 minutes at the permissive temperature. This is because transport from the ER through ERESs and mini-Golgies to the PM for this mutant was faster on average than for the control. Thus, there was no sigmoid asymptote in the variance values plot as found for the other deletion mutants. This asymptote represents the concentrated cargo during transport between the superimposed ERESs and mini-Golgi stacks. During this time period, the high variance values appear constant. This observation demonstrates that extending the TMD facilitates cargo concentration and to some extent, secretory transport as well. The ER-retained mutant ΔTM8 could not be fitted to Eqn 1 since its variance values did not change during the experiment. The values of the a coefficient in the fitted sigmoid equation were 2.79±0.16, 2.23±0.03, 2.50±0.03, 2.07±0.02, 1.72±0.08 and 1.77±0.05 for In2, VSVG, ΔTM1, ΔTM3, ΔTM5 and ΔTM7, respectively. Deletions within the TMDs were also associated with the increase of the values of b and c coefficients (see Table S1 in supplementary material). The inflection point of the sigmoid curve changed from 17.30 minutes for In2 to 27.44 minutes and 35.58 minutes for VSVG and ΔTM7, respectively. The value of the c coefficient changed from 2.65 for In2 to 3.68 and 6.29 for VSVG and ΔTM7, respectively. These data demonstrate that the capacity of the transport machinery to concentrate cargo in the ERES depends on the length of the TMD of the cargo proteins. Overall, the data are consistent with the hypothesis that the role of the TMD in secretory transport is to thermodynamically promote the concentration of cargo in export domains, thereby determining the rate of transport.

To further substantiate these results, we analyzed cargo concentration during secretory transport of the Golgi-resident protein GalT-CFP (Dukhovny et al., 2008). A GalT-CFP variant with three amino acids deleted from its TMD (GalT-Δ3) was generated (Fig. 5A, left panel). To compare the sorting and concentration in the ERESs of these molecules, cells expressing GalT-CFP or GalT-Δ3 were transferred to ice for 20 minutes and nocodazole was added to the tissue-culture medium. Images of the cells were then captured at 30-second intervals for 1 hour at 37°C. During this period, the depolymerization of microtubules induced redistribution of Golgi membranes to the ER followed by their emergence in ERESs (Fig. 5A, right). This process is demonstrated in Fig. 5B by the images of representative cells expressing GalT-CFP or GalT-Δ3 at various times in the presence of nocodazole. The central Golgi fluorescence decreased while peripheral punctate structures appeared. The decrease in fluorescence intensity of the central Golgi (yellow ROI in Fig. 5B) indicates the rate of retrograde transport of GalT-CFP and GalT-Δ3 (Fig. 5C, left graph). However, the emergence of peripheral punctate structures represents ER-to-ERES transport which involves the sorting and concentration processes. Fig. 5C shows plots of both retrograde transport and analysis of concentration in ERESs by utilizing the changes in fluorescence intensity variance. Although the retrograde transport of GalT-CFP and GalT-Δ3 was comparable, the concentration of GalT-Δ3 in the ERESs was significantly decreased. Fitting the normalized averaged values with Eqn 1 gave a values of 4.3±0.11 and 3.6±0.09 for GalT-CFP and GalT-Δ3, respectively. These data also support our finding with YFP-VSVG that the TMD facilitates cargo-protein concentration during export.

Next, we analyzed the relationship between TMD length and COPII. Although it is accepted that COPII-mediated recognition of specific targeting sequences is a prerequisite for entry of cargo proteins into ERESs, the relationship between COPII and the cargo TMD is unclear. To elucidate this, we treated cells with both nocodazole and brefeldin A (BFA) to depolymerize microtubules and to deplete the membranes of ARF1 and COPI complex, respectively. As demonstrated previously, selective sorting of VSVG to the dilated ERESs proceeds via the DXE motif-COPII interaction (Dukhovny et al., 2008), a process that occurs in the absence of ARF1 and COPI. However, absence of COPI results in the transformation of tubular vesicular ERESs to spherical membranes. These dilated ERESs are unstable as they undergo cycles of cargo accumulation and collapse back into the ER. In addition, as they have an approximate diameter of 1 μm, the dilated ERES membranes can be clearly resolved from those of the ER using standard confocal microscopy. Fig. 6 shows images of cells coexpressing YFP-ΔTM8 and CFP-ERGIC53 (Fig. 6A) or expressing YFP-VSVG or YFP-ΔTM7 (Fig. 6B), subjected to nocodazole and BFA during incubation on ice prior to the shift to the permissive temperature. After 1 hour at the permissive temperature, YFP-ΔTM8 was retained within the ER membranes and was completely absent from CFP-ERGIC53-labeled dilated ERESs (Fig. 6A). At the permissive temperature in the presence of nocodazole and BFA, both YFP-VSVG and ΔTM7 localized to the ER membrane and the dilated ERESs (Fig. 6B). Fold concentration within the dilated ERES membranes was analyzed by dividing the fluorescence intensity of the dilated ERESs by that of the surrounding ER. No significant differences were observed between YFP-VSVG and ΔTM7, with apparent fold concentration values within the dilated ERESs of 3.29±0.7 and 3.14±0.7, respectively. These results suggest that the TMD-mediated facilitation of cargo concentration requires membrane modulation by ARF1 and COPI (Manneville et al., 2008).

In summary, we found a direct correlation between the length of the cargo TMD and the speed of its secretory transport and extent of its concentration in ERESs. This demonstrates that the extent of cargo concentration in ERESs, a key determinant of transport rate, is limited by TMD length.

In this study, we focused our analysis on the relationship between TMD length and secretory transport, as well as its relationship with a defined activity of the transport machinery, i.e. cargo concentration in export domains. We demonstrate that the TMD lengths of cargo proteins determine their rate of secretory transport at least in part by facilitating their concentration in export domains during coat-complex-mediated sorting. These data are in line with our hypothesis that interactions between TMD and membrane lipids stabilize the concentrated state of the cargo proteins within export-domain membranes. Furthermore, the association of decreased TMD length with decreased transport rate, as well as with a reduction in the extent of cargo concentration, supports the following model: hydrophobic mismatch interactions arising from differences between the hydrophobic lengths of the protein TMD and the hydrophobic width of the membrane lipids are a principal driving force for the thermodynamically unfavorable processes of cargo-protein concentration (Jensen and Mouritsen, 2004).

These data strongly rely on the prospect that membrane thickness increases in the anterograde direction and that export-site membranes have a composition that is more reminiscent of the membrane of their target organelle. This hypothesis is strongly supported by the report that cholesterol, a molecule associated with thicker liquid-ordered domains, is essential for proper cargo loading to ERESs (Runz et al., 2006).

A possible comment on these results is that the effect of the TMD is rather minimal, based on the finding that a deletion of up to seven amino acids has no effect on the final intracellular distribution of VSVG. Moreover, one might argue that the effect of shortening the TMD on cargo concentration is rather small. The response to these arguments is that a role for TMD length in facilitating cargo concentration in fact predicts an effect on transport dynamics, rather than on the final intracellular distribution. The effect of TMD deletions on reducing the capacity to concentrate cargo in ERESs might not be fully represented by the data obtained using the variance analysis: masking can occur by a variety of uncontrollable differences among individual cells, such as differences in expression levels. Nevertheless, given that the fold concentration of a given cargo protein in an export domain directly affects the magnitude of the efflux, any reduction in cargo concentration significantly affects transport rate.

The data presented in this manuscript lay the foundation for an inclusive model that introduces complementary and successive roles of the targeting sequences and TMDs in secretory transport. The targeting-sequence–COPII complex interaction is a specific protein-protein recognition process that involves the selection step in which proteins are specifically excluded from, or selected for export. Concomitantly, the TMD-membrane interactions provide thermodynamic drive for the cargo-protein concentration process, via the resultant movement from the ER to the export domains to alleviate the mismatch between the TMD and its surrounding membranes. In the case of VSVG, the cytosolic acidic DXE targeting sequence is obligatory for ER export. In its absence, VSVG is retained in the ER as it is apparently excluded from ERESs. Shortening the TMD has a clear and gradual effect on the sorting and transport dynamics up to deletion of the eighth amino acids that confers retention in the ER. As in the case of the VSVG with a mutated DXE motif, retention of the ΔTM8 mutation is also associated with its apparent exclusion from ERESs. Interestingly, for the VSVG deletion mutants involving one to seven amino acids, there is little or no difference in their intracellular distribution at long (24-48 hours) incubations at the permissive temperature. Based on previous observations (Munro, 1995), we expected enhanced localization within the Golgi membranes of the intermediate deletion mutants. However, we found only a single shift in intracellular distribution, associated with the failure to enter ERESs (Fig. 6A). These findings suggest that Golgi retention or residence requires other as-yet undiscovered motifs, in addition to the targeting signals present in VSVG. Thus, for VSVG, and possibly for other PM-targeted membrane cargo proteins as well, crossing of the ER-ERES boundary with mediation by COPII is the critical step in determining their intracellular localization.

Shortening the TMD affects the rate of efflux of VSVG from the Golgi. Unlike the early stage of the ER-to-ERES transport step where cargo sorting is associated with processes such as protein folding and quality control, VSVG in the Golgi complex is in its folded state. Hence, the data in Figs 2 and 3 argue against the possibility that the observed effect on the export rate of shortening the TMDs of VSVG and GalT is exclusively due to folding as the limiting step associated with the quality-control boundary in the ER. Moreover, it demonstrates that the TMD facilitates the cargo-concentration step at different stages of secretory transport.

Figs 2 and 3 demonstrate the association between TMD length and the speed of secretory transport. Moreover, from the experiments on the time-based changes in Golgi fluorescence (Fig. 2A) in ΔTM3, ΔTM5 and ΔTM7 mutants, shortening the TMD also affects the life span of post-Golgi carriers. This is also evident for ΔTM7 (Fig. 3A). Movie 3 in supplementary material demonstrates that the post-Golgi carriers remain for extended periods of time before their fusion with the PM. These data reveal yet another transport process that is potentially affected by the length of the cargo TMD. The mechanism governing the effect of shortened TMDs on fusion efficiency is not clear. One possibility is that the lipid content of the post-Golgi carriers is limited to lipids with shorter acyl tails to match the shortened TMD of ΔTM7. Lipids with shorter acyl chains are expected to be characterized by a less negative spontaneous curvature compared with common phospholipids. Therefore, they should have a lower propensity to form the stalk intermediates of membrane fusion and, hence, the whole fusion process should be hindered (Michael Kozlov, personal communication). This effect of TMD deletion on inter-organelle carrier functions emphasizes the robustness of the secretory machinery by showing that a series of major perturbations affecting transport dynamics and protein cargo-membrane lipid interactions do not culminate in changing the cellular destination of the protein.

The correlation between ER retention and exclusion from ERES membranes was confirmed by the use of dilated ERESs obtained by treating cells with nocodazole and BFA (Dukhovny et al., 2008). COPII was shown to mediate its selective sorting functions by interacting with the DXE export signal of VSVG. However, it is unclear which of the COP complexes is associated with the function of TMD. A plausible hypothesis based on the data presented here is that the membrane composition of the ERES should be tightly regulated for it to allow facilitation of cargo protein concentration via the TMD. More specifically, the ERES membranes should be kept on average thicker than the ER to allow proper concentration of cargo. COPI and ARF1 are associated with stabilization of ERES filled with VSVG cargo, as well as with maintaining its typical highly curved tubular vesicular structure (Dukhovny et al., 2008). ΔTM7 concentration is impaired within ERESs of nocodazole-treated cells (Fig. 4D), but is indistinguishable from VSVG in BFA- and nocodazole-treated cells (Fig. 6). Based on these observations, it is tempting to speculate that COPII is associated with cargo sorting, based on specific recognition of cargo-targeting signals. By contrast, COPI and ARF1 are associated with housekeeping of the ERES membrane composition to render it a cargo concentration-competent and transport-competent domain.

Shortening the TMD may lead to various effects, provoked mainly by altering mismatch interactions. Extreme alterations of such interactions might result in affecting the intracellular distribution of a protein (Fig. 1) (Ronchi et al., 2008). In general, protein-lipid mismatch interactions may lead to various outcomes, such as protein clustering to reduce the surface of mismatched contact zones and tilting of the transmembrane helices (Schmidt et al., 2008). VSVG is known to exist as a trimer. It is possible that VSVG partitioning between monomer and trimer states is driven in part by positive mismatching with local membrane lipids, as well as by an intrinsic propensity for oligomerization. A possible effect of shortening the TMD might be via an effect on diffusion within the ER membranes. However, in several cases, such diffusion has been found to be much faster than cargo loading on ERESs (Dukhovny et al., 2008; Runz et al., 2006). In a study using coarse-grained membrane simulations, mismatching was found to have little effect on diffusion (Schmidt et al., 2008).

In summary, the ERES is a committed membrane domain that specifically selects and concentrates correctly folded secretory cargo proteins. The ERES also excludes resident, misfolded and transport-incompetent proteins. These functions are mediated by proteins and protein complexes, some of which interact directly with the cargo proteins (COPII) whereas others are involved in the cargo-attracting functions of ERES by establishing and stabilizing its unique membrane content and structure (COPI). As the ERES membranes accumulate cargo, additional tasks become necessary, such as stabilization of the cargo-loaded membranes and maturation of the membrane to a transport carrier. Now that data on the sorting and transport dynamics can be obtained, the effect of genetic and pharmaceutical perturbations on the dynamics of ERES function should be utilized to understand the detailed mechanistic features of export domains.

All reagents were purchased from Sigma Chemical Co. (St Louis, MO) unless otherwise specified. YFP-VSVGtsO45 was prepared as described elsewhere (Ward et al., 2001). CFP-ERGIC53 (Ward et al., 2001) was a kind gift from the laboratory of Theresa H. Ward (London School of Hygiene & Tropical Medicine, London, UK). YFP-VSVGtsO45-AIA (DXE-deficient mutant), addition mutants In2-4, TMD deletion mutants ΔTM1-9 and GalT-Δ3 were prepared using the Quick-Change kit from Stratagene (La Jolla, CA). The primers used for the PCR were: 5′-GACTATTCTTGGTTGTCGTCCTCCGAGTTGG-3′ for In2; 5′-GACTATTCTTGGTTGTCGTCGTCGTCCTCCGAGTTGG-3′ for In4; 5′-GACAGATTTATACAGCCATAGCGATGAACCGAC-3′ for AIA; 5′-GGGTTAATCATTGGACTATTCTTGGTTCGAGTTGG-3′ for ΔTM1; 5′-GGGTTAATCATTGGATTCTTGGTTCTCCGAG-3′ for ΔTM1a; 5′-GGGTTAATCATTGGACTATTCCGAGTTGG-3′ for ΔTM3; 5′-CATAGGGTTAGGACTATTCCGAGTTGG-3′ for ΔTM5; 5′-GCCTCTTTTTTCTTTGGGTTAGGACTATTCCGAGTTGG-3′ for ΔTM7; 5′-GCCTCTTTTTTCTTTGGGTTAGGATTCCGAGTTGG-3′ for ΔTM8; 5′-GCCTCTTTTTTCTTTGGGGGATTCCGAG-3′ for ΔTM9; 5′-CTGCTCGTGTGCCTGCACCTTG-3′ for GalT-Δ3.

COS-7 (African green monkey) cells were grown at 37°C in a humidified atmosphere with 5% CO₂. Cell cultures were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum and penicillin and streptomycin (Biological Industries, Bet-Haemek, Israel). For microscopy, cells were subcultured in glass coverslip chambers (Nalgene Nunc, Rochester, NY), grown to 20-30% confluence and then transiently transfected with 1.5 g DNA/chamber using FuGENE-6 reagent (Roche Applied Science, Mannheim, Germany).

Cells were imaged in DMEM without Phenol Red but with supplements, including 20 mM HEPES, pH 7.4. Transfection and imaging were carried out in Labtek chambers (Nunc). A Zeiss LSM PASCAL (Carl Zeiss MicroImaging, Jena, Germany) was equipped with an Axiovert 200 inverted microscope, and Ar 458 nm, 488 nm and 514 nm laser lines for ECFP, EGFP and EYFP, respectively. The confocal and time-lapse images were captured using a Plan-Apochromat 63 NA 1.4 objective (Carl Zeiss MicroImaging). Temperature on the microscope stage was maintained during time-lapse sessions using an electronic temperature-controlled airstream incubator. Images and movies were generated and analyzed using the Zeiss LSM software, NIH Image and ImageJ software (W. Rasband, NIH, Bethesda, MD).

Over 1-hour long time-lapse image sequences were captured using the autofocusing function integrated into the `advanced time series' macro set (Carl Zeiss MicroImaging). Data were then fitted to the relevant equations using Kalaidagraph (Synergy software, Essex-Junction, VT). Pixel fluorescence intensity variance (PFIVar) value calculation and fluorescence intensity analysis were carried out using ImageJ (Wane Rasband, NIH, Bethesda, MD). Values were normalized by dividing all variance values by the initial value. Experiments on Golgi export after the 20°C block were simulated using SAAM II software (Univ. of Washington, Seattle, WA). The experiment was simulated using a two-compartment model with one flux delineating YFP-VSVG movement from the Golgi to the PM ROI. Experimental data (fluorescence intensities) of both Golgi and PM ROIs were used to obtain the values of the Golgi exit rate coefficients.