A positive signal prevents secretory membrane cargo from recycling between the Golgi and the ER

Abstract

The Golgi complex and ER are dynamically connected by anterograde and retrograde trafficking pathways. To what extent and by what mechanism outward-bound cargo proteins escape retrograde trafficking has been poorly investigated. Here, we analysed the behaviour of several membrane proteins at the ER/Golgi interface in live cells. When Golgi-to-plasma membrane transport was blocked, vesicular stomatitis virus glycoprotein (VSVG), which bears an ER export signal, accumulated in the Golgi, whereas an export signal-deleted version of VSVG attained a steady state determined by the balance of retrograde and anterograde traffic. A similar behaviour was displayed by EGF receptor and by a model tail-anchored protein, whose retrograde traffic was slowed by addition of VSVG's export signal. Retrograde trafficking was energy- and Rab6-dependent, and Rab6 inhibition accelerated signal-deleted VSVG's transport to the cell surface. Our results extend the dynamic bi-directional relationship between the Golgi and ER to include surface-directed proteins, uncover an unanticipated role for export signals at the Golgi complex, and identify recycling as a novel factor that regulates cargo transport out of the early secretory pathway.

Keywords: Rab6; VSV glycoprotein; live cell imaging; retrograde transport; secretory pathway.

Figure 1. Characterization of ER-to-Golgi anterograde transport of FP-22 at 20°C

1. NRK cells, co-microinjected with mEGFP-22 and GalNAc-mCherry (upper row), or co-transfected with VSVG-EGFP and GalNAc-mCherry at 39.3°C (lower row), were incubated for 2 h at 20°C before fixation. Images are maximum intensity projections of z-stacks. Scale bars, 10 μm.

2. Representative FRAP experiment of Golgi regions (arrowheads) or of a portion of the ER (asterisk) in co-microinjected cells allowed to express mCerulean-22 and GalNAc-mCherry, incubated for 60 min at 20°C before imaging. Lower row: magnifications of the boxed area (rotated clockwise by 90°) in the upper row, showing a Golgi apparatus and a portion of the ER that were bleached (red contours; see Supplementary Movie S2). Scale bars, 20 (upper and middle rows) and 10 (lower row) μm.

3. Recovery curves of mCerulean-22 in the Golgi apparatus and in the ER in the presence or absence of energy together with the curves fitted to one-phase (ER, green and violet traces) and two-phase (Golgi, red and blue traces) exponential functions. Data points represent means ± SEM, and the 95% confidence limits of the fitted curves are indicated by the surrounding lightly coloured areas (n = 17, 7, 8 and 10 cells, for Golgi, ER, Golgi minus ATP, and ER–ATP, respectively; number of experiments ≥ 5).

4. Representative line scans (yellow) crossing the Golgi apparatus and portions of surrounding ER in FRAP experiments performed as described in (B), under control conditions (upper panel) and in the absence of energy (lower panel), before bleaching (red), immediately after bleaching (blue), and after 60 min of recovery (green) (related to Supplementary Movie S3). Scale bars, 10 μm.

5. FRAP of a portion of the ER (yellow contour) was performed on energy-depleted NRK cells, microinjected and incubated as described in (B). Scale bar, 15 μm.

Figure 2. FP-22 is included in a Golgi-to-ER retrograde transport pathway at 20 and 37°C

1. Loss of Golgi fluorescence after bleaching of the ER in NRK cells co-microinjected with mCerulean-22 and mRFP-KDEL. Cells were incubated for 1 h at 20°C and imaged alive at 20°C for 30 min, before ER bleaching (red and yellow contours). Enlargements (lower row) of the boxed area (upper row) in the mCerulean-22 channel show Golgi emptying in more detailed temporal resolution. Red arrowhead indicates a thin FP-22-positive tubule emanating from the Golgi apparatus (see Supplementary Movie S4). Scale bars, 15 (upper and middle rows) and 3 (lower row) μm.

2. Inhibition of loss of Golgi fluorescence by energy depletion. Cells were microinjected and incubated as in (A). The mCerulean channel is shown. Bleached ROIs are indicated by the yellow contours. Scale bars, 20 μm.

3. Quantitative analysis of fluorescence loss from the Golgi after ER bleaching in the presence or absence of ATP, and recovery of the ER in the same conditions (n = 12 cells, number of experiments ≥ 5). Data points represent means ± SEM. The coloured lines show the fitted two-phase exponential decay and association curves.

4. Bleaching of the ER, as described in (A), was performed at 37°C on microinjected cells. The red contour (upper panel) indicates the bleached ER ROI; the red arrowhead in the enlargement (lower row) of the boxed area highlights an FP-22-positive tubule originating from the Golgi. Scale bars, 10 (upper two rows) and 5 (lower row) μm.

5. Time course of fluorescence loss from the Golgi after ER bleaching at 37°C (n = 8 cells). Data points represent means ± SEM.

Figure 3. VSVG is poorly recruited into Golgi-to-ER retrograde trafficking pathways

A NRK cells, co-transfected with VSVG-EGFP and mRFP-KDEL at 39.3°C and then incubated 15 min at 32°C, were imaged alive at 20°C before or after bleaching the ER, as indicated by the red contours. The GFP channel is shown. Scale bar, 10 μm. B Quantitative analysis of VSVG fluorescence loss from the Golgi after ER bleaching. Data points represent means ± SEM, and the 95% confidence limits of the fitted curves are indicated by the surrounding lightly coloured areas. The green trace shows the fitted two-phase exponential decay curve (n = 16). The corresponding curve of FP-22 (see Fig2C) is displayed for comparison. C–F Effect of H89 treatment on the distribution of FP-22 and VSVG. ERGIC-53 (p58) distribution in NRK cells treated (right) or not (left) with H89 for 60 min at 20°C (C). NRK cells, transfected at 39.3°C with VSVG-EGFP and GalNAc-mCherry, were incubated for 15 min at 32°C and imaged alive for 60 min at 20°C in the presence of H89 (D). The same protocol was carried out on NRK cells microinjected with mCerulean-22 and GalNAc-mCherry (E). The inset shows an enlargement of the boxed area. Time course of Golgi fluorescence loss (F) in the presence of H89 of FP-22 and VSVG-EGFP (FP-22, n = 9; VSVG-EGFP, n = 12). Data points represent means ± SEM. Scale bars: 20 mm (C), 10 mm (D), 10 and 2 mm (E, inset).

Figure 4. FP-22 retrograde transport is not affected by co-expression of VSVG

1. Intracellular distribution of VSVG-EGFP and tdTomato-22 in co-microinjected NRK cells incubated for 90 min at 20°C before fixation. Scale bar, 10 μm.

2. NRK cells microinjected with Cerulean-22 and VSVG-Venus cDNAs were incubated 75 min at 39.3°C followed by 60 min at 20°C before live cell imaging at the same temperature. The ER was bleached in the Cerulean channel (yellow contour), and the loss of Golgi fluorescence was recorded for 60 min after bleaching. Scale bar, 10 μm.

3. Quantitative analysis of the iFRAP illustrated in (B). Data points represent means ± SEM. The fitted two-phase exponential decay curve is also shown (n = 10 cells).

Figure 5. Distribution of FP-22 and VSVG-EGFP within the Golgi apparatus

1. NRK cells were co-microinjected with mCerulean-22 and GalNAc-mCherry and then incubated for 90 min at 20°C before fixation and immunostaining with anti-TGN38 mAb. A maximum intensity projection of a representative z-stack is shown. The boxed area is enlarged in the images of the lower row.

2. NRK cells were co-transfected with VSVG-EGFP and GalNAc-mCherry plasmids at 39.3°C and then incubated and processed as described in (A).

Data information: In (A) and (B), the graphs on the right report fluorescence intensities for each channel along the yellow arrow drawn in the enlargements. Scale bars: 15 μm and 5 μm in upper and lower row, respectively.

Figure 6. Signal-deleted VSVG is recruited into a Golgi-to-ER retrograde transport pathway

1. Localization of VSVGAxA-EGFP in NRK cells co-transfected with GalNAc-mCherry at 39.3°C and incubated for 2 h at 20°C before fixation. Scale bar, 15 μm.

2. Loss of VSVGAxA-EGFP Golgi fluorescence after bleaching of the ER. NRK cells, transfected as in (A), were incubated for 60 min at 20°C and then imaged alive at 20°C. The red contour indicates the bleached area (see Supplementary Movie S5). Scale bar, 15 μm.

3. Time course of fluorescence loss in the Golgi and recovery in the ER after ER bleaching (n = 13). The data were fitted to two-phase exponential equations.

4. NRK cells, transfected as in (B) and incubated for 60 min at 32°C, were imaged at 20°C in the presence of H89. Scale bar, 10 μm.

5. Time course of VSVGAxA-EGFP Golgi fluorescence loss in cells treated with H89 (n = 9). Data points represent means ± SEM. The data points were fitted to a mono-exponential decay function. The trace for VSVG-EGFP, obtained by fitting the data points of Fig4D to a monoexponential function, is also reported.

6. iFRAP of NRK cells microinjected with Cerulean-22 YTDIE and GalNAc-mCherry. Cells were incubated for 1 h at 20°C and then imaged at the same temperature for 30 min before and 60 min after ER bleaching (red contour). The unbleached region corresponding to the Golgi apparatus was chosen based on co-localization with the Golgi marker, GalNAc-mCherry (inset). Scale bar, 15 μm.

7. Quantitative analysis of Golgi fluorescence intensities before and after bleaching as in (F). Data points represent means ± SEM. The green line represents the fitted two-phase exponential decay equation (n = 11 cells). Golgi emptying of the parent construct FP-22 is also shown for comparison (red trace from Fig2C).

Figure 7. EGFR, but not Syb2, is recruited into a retrograde transport pathway from the Golgi to the ER

1. Steady-state distribution of EGFR-EGFP and Syb2 after transfection at 37°C. Scale bars, 10 μm.

2. Intracellular localization of EGFR-EGFP and Syb2 75 min after microinjection of each of the two plasmids together with GalNAc-mCherry. Scale bars, 15 μm.

3. NRK cells, co-microinjected with EGFR-EGFP and GalNAc-mCherry, were incubated for 90 min at 20°C and then fixed. Scale bar, 10 μm.

4. Loss of EGFR-EGFP fluorescence from the Golgi region after bleaching of the ER. Cells co-microinjected with EGFR-EGFP and GalNAc-mCherry were incubated for 60 min at 20°C and then imaged alive at the same temperature. The red contour indicates the bleached ER region. Scale bar, 15 μm.

5. Quantitative analysis of Golgi emptying by EGFR-EGFP after ER bleaching (n = 9 cells). Data points represent means ± SEM, and the 95% confidence limits of the fitted curves are indicated by the surrounding lightly coloured areas. Values were fitted to a two-phase exponential decay function.

Figure 8. The signal-deficient form of VSVG is recycled from the Golgi to the ER through the Rab6-dependent pathway

1. Intracellular localization of VSVGAxA-EGFP and endogenous ERGIC-53 in HeLa cells co-transfected at 39.3°C with Arf1 (WT or Q71L) and incubated for 3 h at 32°C. Asterisks indicate transfected cells. Cells transfected with the GTP-locked Arf1 show a redistribution of ERGIC-53. Scale bars, 15 μm.

2. Golgi/ER mean fluorescence ratio of VSVGAxA-EGFP in HeLa cells transfected and incubated as described in (A) (n = 29, 25 and 23 cells transfected with empty vector, WT or Q71L Arf1, respectively). **P = 0.002 determined by one-way ANOVA followed by Dunn's post-test.

3. HeLa cells were co-transfected with VSVGAxA-EGFP and myc-Rab6 (WT or T27N), exposed to Cy3-STxB at 4°C for 20 min, followed by washing and incubation for 5 h at 32°C to allow toxin internalization and, concomitantly, VSVGAxA export from the ER. Yellow arrowheads indicate two transfected cells where Rab6 T27N inhibits the arrival of STxB to the ER and causes an increased concentration of VSVGAxA in the Golgi. Maximum intensity projections of z-stacks are shown. Scale bars, 15 μm.

4. Golgi/ER fluorescence ratio of Cy3-STxB and VSVGAxA-EGFP in HeLa cells transfected and incubated as described in (C). ***P < 0.0001 determined by one-way ANOVA followed by Dunn's post-test (STxB quantitation (left): n = 46, 44 and 28 cells transfected with empty vector, myc-Rab6 WT and myc-Rab6 T27N, respectively; VSVGAxA quantitation (right): n = 51, 48 and 49 cells transfected with empty vector, myc-Rab6 WT and myc-Rab6 T27N, respectively).

5. Analysis of Golgi emptying by VSVGAxA in HeLa cells co-transfected with VSVGAxA-EGFP and either GalNAc-mCherry (CTRL) or Rab6 T27N and incubated as in (C) before live cell imaging at 20°C. The Rab6 T27N-transfected cells were exposed to Cy3-STxB during the 4°C incubation, to check the efficacy of the dominant-negative mutant (n = 9 and 7 for CTRL and Rab6 T27N, respectively). Data points represent means ± SEM, and the 95% confidence limits of the fitted curves are indicated by the surrounding lightly coloured areas.

Figure 9. Inhibition of Rab6 function accelerates the transport of VSVGAxA-EGFP, but not of VSVG-EGFP, to the cell surface

1. NRK cells were co-transfected with VSVG-EGFP (upper panel) or VSVGAxA-EGFP (lower panel) and either WT Rab6 or T27N Rab6 at 39°C; 24 h after transfection, the cells were incubated at 32°C for 30 min or 2 h, in the case of WT or signal-deficient VSVG, respectively, before surface labelling of non-permeabilized cells. Images are maximum intensity projections of z-stacks. Scale bars, 15 μm.

2. Box blots of cell surface expression of VSVG-EGFP or VSVGAxA-EGFP expressed as ratio of antibody-revealed surface fluorescence to total EGFP-VSVG fluorescence (n = 20 images, for a total number of cells of ˜150). ***P < 0.0001 and *P = 0.024 determined by Student's t-test.

3. Working model of transport of signal-bearing or signal-deficient cargoes along the secretory pathway. See text for explanation.