GRASP55 and GRASP65 play complementary and essential roles in Golgi cisternal stacking

Abstract

In vitro studies have suggested that Golgi stack formation involves two homologous peripheral Golgi proteins, GRASP65 and GRASP55, which localize to the cis and medial-trans cisternae, respectively. However, no mechanism has been provided on how these two GRASP proteins work together to stack Golgi cisternae. Here, we show that depletion of either GRASP55 or GRASP65 by siRNA reduces the number of cisternae per Golgi stack, whereas simultaneous knockdown of both GRASP proteins leads to disassembly of the entire stack. GRASP55 stacks Golgi membranes by forming oligomers through its N-terminal GRASP domain. This process is regulated by phosphorylation within the C-terminal serine/proline-rich domain. Expression of nonphosphorylatable GRASP55 mutants enhances Golgi stacking in interphase cells and inhibits Golgi disassembly during mitosis. These results demonstrate that GRASP55 and GRASP65 stack mammalian Golgi cisternae via a common mechanism.

Figure 1.

Knockdown of GRASP55 reduces the number of cisternae per stack. (A and B) Confocal fluorescence images of GRASP55 knockdown cells. HeLa cells transfected with indicated siRNA were fixed after 96 h and immunostained for GRASP55 and GM130. (C and D) Fluorescence images of HeLa cells in which endogenous GRASP55 was replaced by exogenous GRASP55. HeLa cells expressing GFP or rat GRASP55-GFP using an inducible retroviral expression system were transfected with control siRNA or siRNA specific for human GRASP55. Doxycycline was added 48 h after transfection. Cells were fixed and stained for GM130. (E–H) Representative EM micrographs of cells described in A–D. Arrowheads indicate Golgi stacks. Note that the number of cisternae in the stacks was reduced in GRASP55 knockdown cells (F), which was restored by the expression of rat GRASP55 (H), but not by GFP (G). (I) Immunoblots of cells described in C and D. Duplicate samples were loaded; asterisk indicates a nonspecific band. “End. p55”, endogenous GRASP55. (J) Quantitation of the EM images in E and F from three sets of independent experiments. Results are expressed as the mean ± SEM. (K) The numbers of cisternae per stack from one representative experiment were presented in a histogram format. Note that most stacks contained 5–6 cisternae in control RNAi cells, which was reduced to 3–4 in GRASP55 RNAi cells. (L) Quantitation of G–H. Note that GRASP55 knockdown significantly reduced the number of cisternae per stack, whereas expression of rat GRASP55 restored it. ***, P < 0.001.

Figure 2.

Effect of GRASP55 depletion on Golgi ribbon linking. (A–C) Enlarged confocal fluorescence images of the Golgi with classified morphology in GRASP55 knockdown cells stained for GM130. Bar, 10 µm. mild frag., mildly fragmented. (D) Quantitation (mean ± SEM) of classified Golgi morphology in HeLa cells treated with control or GRASP55 siRNA from three independent experiments. GM130, GRASP65, and Gos28 were used as Golgi markers and 300 cells were counted. (E and F) FRAP analysis of ManII-GFP–expressing HeLa cells transfected with indicated siRNAs. The indicated regions (arrows) were photobleached with laser pulses and fluorescence recovery was recorded. Representative images at indicated times are shown. Bar, 5 µm. (G) Quantitation of FRAP results. Fluorescence recovery was represented by the ratio of GFP fluorescence intensity in the bleached area to that of the entire Golgi. Normalization was set between the values before bleaching and the first time point after bleaching. Results represent mean ± SEM from two independent experiments, with more than 17 cells quantified in each case.

Figure 3.

Depletion of both GRASP55 and GRASP65 leads to disassembly of the entire Golgi stack. (A and B) Confocal fluorescence images of GRASP55 and GRASP65 double-knockdown cells. HeLa cells were transfected with a mixture of GRASP55 and GRASP65 siRNA (B) or with control siRNA (A). Cells were fixed and immunostained for both GRASP55 and GRASP65 (anti-GRASP55/65), and for Gos28. (C) Immunoblots of HeLa cells transfected with indicated siRNA. (D and E) Representative EM micrographs of cells transfected with indicated siRNAs. Arrowheads in D indicate single cisternae and Golgi fragments. Arrows in E indicate normal Golgi stacks. (F) Quantitation (mean ± SEM) of GRASP55 and GRASP65 knockdown efficiency and effects on Golgi stacking from three independent experiments. Statistical significance was assessed by comparison with the control siRNA cells. *, double knockdown cells were not quantified due to the complete disassembly of the Golgi stacks.

Figure 4.

GRASP55 is phosphorylated during mitosis. (A) GRASP55 schematic with phosphorylation sites indicated. Predicted phosphorylation sites (T222, T225, S245, T249, and T435) are indicated by asterisks. myr: myristoylated N-terminal glycine. (B) Detection of GRASP55 phosphorylation using a band-shift assay. Golgi membranes were either treated with buffer (lane 1) or mitotic cytosol (MC, lanes 2–4). Membranes were reisolated and further treated with calf intestine alkaline phosphatase (CIP) in the absence (lane 3) or presence (lane 4) of a general phosphatase inhibitor, β-glycerophosphate (β-GP). Note the increase in molecular weight of GRASP55 under mitotic conditions was reversed by CIP. (C) GRASP55 phosphorylation by ERK2. Golgi membranes were incubated with indicated proteins followed by Western blotting. Note that ERK2/MEK1 phosphorylated GRASP55 to a similar extent as MC. (D) The mitotic phosphorylation sites of GRASP55 are in the SPR domain. GST-tagged GRASP55 constructs were incubated with buffer (lane 1), or interphase (IC, lane 2) or mitotic (MC, lane 3) cytosol and analyzed by immunoblotting. Note that the SPR domain (213–454), but not the GRASP domain (1–212), exhibited a band shift after MC treatment, whereas deletion of aa 213–231 largely abolished the phosphorylation. (E) Mapping the phosphorylation sites of GRASP55. Potential sites (asterisks in A) in His-GRASP55 were mutated to alanines or glutamates and the purified proteins were analyzed as in D.

Figure 5.

Cell cycle–regulated GRASP55 oligomerization is sufficient to link adjacent surfaces. (A) Co-purification of MBP-GRASP55 and His-GRASP55. Differently tagged proteins were separately purified, mixed, and incubated in the presence of buffer (control) or mitotic cytosol (mitotic). The protein complex was isolated using either amylose (lanes 1–3) or nickel beads (lanes 4–6). Equal proportions of input (I), unbound (U), or bound (B) fractions were analyzed by immunoblotting for GRASP55. Note the copurification of the two proteins under control but not mitotic conditions. (B) Analysis of GRASP55 oligomerization by nondenaturing gels. His-GRASP55 was incubated in the absence (−, lane 1) or presence (+, lane 2) of ERK2/MEK1 kinases followed by nondenaturing electrophoresis and Western blotting. Molecular weight standards are indicated on the left. Note that the higher molecular weight bands in lane 1 (arrowheads) diminished after kinase treatment. (C) Aggregation of GRASP55-coated beads. Purified His-GRASP55 or BSA was covalently coupled to the surface of magnetic Dynal beads and incubated with BSA, interphase cytosol (IC), or mitotic cytosol (MC). After incubation the beads were placed on glass slides and random fields were photographed. A representative image of each condition is shown. Note that GRASP55-coated beads aggregated slightly in the presence of BSA; this was enhanced by IC, but inhibited by MC. (D) As in C, except that GRASP55 beads were first aggregated using IC and then treated with either MC (IC→MC) or with purified ERK2/MEK1 kinases (IC→K). Note that aggregates were reversibly disassembled by MC and kinases. Bar, 100 µm. (E) Quantitation (mean ± SEM) of C and D; n = 3. (F) Treatment of Golgi membranes with purified kinases leads to cisternal membrane unstacking. Purified Golgi stacks were treated with either buffer or indicated kinases, and analyzed by EM. Shown are the quantitation results from a representative experiment. Statistical significance was assessed by comparison of kinase treatment with buffer treatment. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Figure 6.

Mapping the domain structure of GRASP55 for oligomerization and mitotic regulation. (A) Purified His-tagged GRASP domain (1–212) was coupled to Dynal beads and sequentially incubated with interphase cytosol (IC) and mitotic cytosol (IC→MC). Note that aggregates of beads formed by IC treatment were not dispersed by MC. (B) As in A, using the T222A–T225A–S245A mutant. Beads aggregated under both conditions. (C) Beads coated with the T222E–T225E mutant were incubated with BSA or IC. Note that beads did not aggregate after IC treatment. (D) Wild-type GRASP55-coupled beads were incubated with IC in the presence of soluble full-length (FL) GRASP55, the GRASP domain (1–212), or the SPR domain (213–454) recombinants. Bar, 100 µm. Note that the aggregation was suppressed by the soluble full-length protein and the GRASP domain, but not by the SPR domain. (E) Quantitation of A and B from three independent experiments (mean ± SEM). (F) Quantitation of C. Aggregation of the beads coupled with the T222E–T225E mutant was significantly reduced compared with those coated with wild-type GRASP55 when treated with IC. (G) Quantitation of D. **, P ≤ 0.01; ***, P ≤ 0.001.

Figure 7.

Overexpression of nonregulatable GRASP55 mutants enhances Golgi stack formation in interphase cells. (A–F) Representative EM images of interphase cells expressing indicated GRASP55 constructs. Bar, 0.5 µm. Note that the Golgi structures in cells expressing the GRASP domain (C) and the T222A–T225A–S245A mutant (D) are better organized than those in the GFP cell line (A). (G) Quantitation (mean ± SEM) of 20 cells from conditions indicated in A–F. Statistical significance was assessed by comparison to the GFP cell line. ***, P < 0.001. (H) Cells in A–F were lysed in SDS buffer and analyzed by Western blotting.

Figure 8.

Rescue of Golgi structure by expression of exogenous GRASP55 and GRASP65 in cells with both GRASPs depleted. (A–E) EM images of HeLa cell lines transfected with a mixture of GRASP55 and GRASP65 siRNAs and induced to express indicated protein. Arrowheads indicate unstacked cisternae. (F) Quantitation of 20 cells in B–E. Statistical significance was assessed by comparison to the GFP cell line treated with control siRNA. ***, P < 0.001. (G) Western blots of cells in A–E. “End. p55”, endogenous p55; “End. p65”, endogenous GRASP65. Asterisk indicates a nonspecific band.

Figure 9.

Expression of nonphosphorylatable GRASP55 mutants inhibits mitotic Golgi disassembly. (A–E) Confocal images of mitotic cells of indicated HeLa cell lines stained for DNA and GRASP65. Note that cells expressing the GRASP domain (C) and the T222A–T225A–S245A mutant (D) had more mitotic clusters, whereas Golgi fragmentation in cells expressing the T222E–T225E mutant (E) was more complete compared with that in GFP-expressing cells (A). Bar, 20 µm. (F–I) Representative EM images of mitotic cells expressing indicated GRASP55 constructs. Mitotic cells were collected by shake-off after release from a double-thymidine block and processed for EM. Bar, 0.5 µm. Note the remaining Golgi cisternal membranes or stacks (arrows) and vesicle clusters (arrowheads) in the GRASP domain–expressing (H) and the T222A–T225A–S245A mutant-–expressing (I) cells. Asterisks indicate condensed chromosomes. (J) Quantitation of the number of Golgi clusters in mitotic cells expressing indicated proteins. 20 cells were examined in each case. Results are expressed as the mean ± SEM. Statistical significance was assessed by comparison to the GFP cell line. ***, P < 0.001.