Setting up an approach to force entry into mitosis with an intact Golgi ribbon

To investigate the role of GC unlinking in mitotic entry, we have developed a two-step strategy to first accumulate cells in G2 with an intact Golgi ribbon and then force them to enter mitosis. To block GC unlinking and investigate the downstream effects, we employed a well-characterized strategy involving the pharmacological inhibition of the phosphorylation of GRASP65, which induces a potent GC-dependent G2 arrest, coupled with artificial disassembly of the GC as a control to identify the direct consequences of the block of the GC unlinking (29).

To this purpose, we employed normal rat kidney (NRK) cells as they are noncancerous cells and generally used as a mitosis model because they are easy to synchronize. The cells were arrested at the beginning of the S-phase using the aphidicolin-block protocol (38) (Figs 1A and S1A and B). Then, aphidicolin was washed out, and during G2, the cells were treated for 2 h with DMSO as a control or with SP600125 (SP), which inhibits the JNK2-mediated phosphorylation of GRASP65, resulting in the formation of a continuous Golgi ribbon and G2 arrest (14, 29). 2 h after the addition of SP, part of the cells was also treated with MK1775 (MK), an inhibitor of Wee1, which is an inhibitor of the kinase CDK1 (35). This last treatment induces the activation of CDK1 and a rapid and synchronous entry into mitosis (35). At the end of the incubation with MK, the cells were fixed and stained for immunofluorescence with an anti-GRASP65 antibody and Hoechst to label the GC and the DNA, respectively. Confocal microscopy analysis of the cells in G2 showed that the GC of control cells appeared in the usual form of distinct groups of membranes localized mainly around the nucleus, whereas SP inhibited GC unlinking, as previously shown (Fig 1B) (29). The quantification of Golgi objects shows that the addition of MK had no effects on the Golgi structure of the cells in G2, whereas SP reduced its number by about 40%, in line with the effect of ribbon reformation. Furthermore, MK added to SP did not counteract the Golgi compaction effect induced by SP (Fig 1B and C). Next, we examined if MK treatment was able to force SP-treated cells to enter mitosis. Quantification of the mitotic index showed that SP induced about 75% inhibition of entry into mitosis compared with control cells, as previously shown (29). As expected, MK addition caused a 50% increase of the mitotic index in control cells and, more importantly, also fully rescued the SP-induced G2 arrest (Fig 1B and D). These results indicate that MK can override the Golgi checkpoint in cells with an intact GC ribbon in G2, indicating that the assay is suitable for testing whether the Golgi status in G2 (i.e., in the form of disassembled stacks as in controls cells or intact, in the case of SP-treated cells) affects the subsequent mitotic processes.

Golgi unlinking during G2 is required for proper spindle formation

Next, we investigated the functional consequences of the forced entry into mitosis in cells with an intact GC in G2, focusing first on the spindle structure, whose formation is completed during metaphase. The major task of the spindle is to separate sister chromatids between daughter cells. Thus, this approach offers the opportunity to clearly identify potential defects (7). To visualize the spindle aberrations, the cells were forced to enter mitosis, as indicated above. Then, after fixation, they were labeled with antibodies against α-tubulin and GRASP65 to identify the MTs and the GC, respectively, and Hoechst to visualize the DNA (Fig 2A and B). We found that 60% of the cells treated with SP and MK showed severe spindle alterations, whereas only a minor fraction of control cells, about 10%, showed alterations (Fig 2B and C), in line with the presence of transient defects normally detected also during unperturbed mitosis (39). The prevalent defect was multipolarity, with a minor fraction represented by disorganization of the spindle fibers (Fig S2) (39). Treatment with MK or SP alone did not significantly increase the frequency of spindle defects compared with the control (Fig 2C). Therefore, to test whether the observed abnormalities were a direct consequence of the inhibition of GC unlinking, NRK cells were treated as above but with the difference that 1 h before addition of SP, part of the samples were also treated with brefeldin A (BFA) (Fig 2A) to induce artificial disruption of the GC (29, 40). BFA does not interfere with cell division and induces redistribution of Golgi enzymes into the endoplasmic reticulum (ER), whereas Golgi matrix proteins aggregate in the form of dispersed tubular vesicular clusters (41). Notably, the BFA-induced Golgi disassembly induced a 60% rescue of bipolar spindle formation events in cells treated with SP and MK (Fig 2C), indicating that a significant fraction of spindle defects directly resulted from an intact GC ribbon in G2 and not from untimely FA disassembly potentially caused by MK (5).

A previous study based on blocking stack disassembly through the formation of large diaminobenzidine polymers in the CG lumen has shown that inhibition of GC stacks disassembly in mitosis induces the formation of monopolar spindles, spindle assembly checkpoint activation, and metaphase arrest as a consequence of a steric hindrance of the intact Golgi stacks (23). Thus, a possible explanation of our results is that forcing entry into mitosis of G2 cells with an intact ribbon could affect spindle formation because of the persistence of large Golgi clusters and that BFA alleviated the defects by dissolving these clusters. To examine this hypothesis, we evaluated the extent of mitotic Golgi fragmentation by measuring the size of mitotic clusters in control cells or cells treated with SP and MK in the absence or presence of BFA. In control cells, the Golgi underwent extensive fragmentation, as expected. Similarly, in cells treated with SP and MK, the Golgi was also highly fragmented, albeit to a slightly lesser extent, as shown by the about 50% increase in the average area of the GC clusters (Fig 2D and E). The latter result is not surprising because even if SP inhibits GC unlinking, the activation of CDK1 after the mitosis onset stimulates Golgi disassembly pathways distinct from the JNK2-mediated GRASP65 phosphorylation (42). Importantly, the addition of BFA to cells treated with SP and MK did not alter the size distribution of the disassembled GC clusters compared with cells treated only with SP and MK (Fig 2D and E), excluding the possibility that BFA recovers spindle formation because of the reduction of the size of the clusters, suggesting that the spindle defects observed in metaphase are the direct consequence of the lack of Golgi unlinking during the G2 phase.

To test this conclusion by an independent approach, we induced constitutive and irreversible unlinking of GC into isolated stacks through the depletion of GRASP55 and examined whether this condition rescued spindle formation. Here, we employed TERT-RPE1 (RPE-1) cells as they are noncancerous cells that can be efficiently synchronized and also transfected. The cells were transfected with control non-targeting or GRASP55-specific siRNAs and synchronized using the double-thymidine protocol (29) (Fig S3). Constitutive GC unlinking is a condition that overrides the effect of SP on Golgi compaction and G2 arrest (29). After thymidine washout, the cells were incubated in the absence or presence of SP and MK, fixed at the mitotic peak, and stained with antibodies against GRASP55 and α-tubulin (Fig S4). The quantification of Golgi objects shows that the addition of SP to control cells, even in the presence of MK, induces a reduction in Golgi objects, about 40%, as already shown for NRK cells. Instead, in cells depleted of GRASP55, the number of Golgi objects is increased by about three times, showing that the depletion of GRASP55 induces an irreversible unlinking of the Golgi structure. In these cells, the addition of SP and MK can still induce a minor reduction of Golgi objects, whose number remains about 40% higher than in control cells (Fig 3A). Importantly, the investigation of spindle morphology showed that in RPE-1 cells, the treatment with SP or MK alone induced an increase of aberrant spindles. However, the combination of SP and MK had a potent additive effect in cells treated with non-targeting siRNA (about 55% of cells with defective spindles, more than double with respect to SP or MK alone) but had a significantly lower impact in cells with the GC irreversibly unlinked by depletion of GRASP55 (GR-55 KD; about 30% of cells with defective spindles; Fig 3B). Thus, our data demonstrated that a significant fraction of the aberrant spindles observed in cells treated with SP and MK is the direct consequence of the lack of Golgi unlinking during G2.

Golgi unlinking controls spindle formation through Aurora-A

Next, we addressed the question of why Golgi unlinking in G2 is necessary for spindle formation. The main organizers of the spindle are the CEs, which are duplicated during S-phase (43) and then separated during G2, in concomitance with ribbon unlinking (38, 44). The separated CEs reach the definitive localization at metaphase when they are located at the opposite poles of the cells. Starting from the G2 phase, the pericentriolar matrix (PCM) surrounding the CEs undergoes a profound modification. The PCM is a matrix composed of multiprotein complexes that control the polymerization and stability of three types of spindle fibers: the kinetochore MTs connecting the PMC to the kinetochores, the polar MTs that interconnect at the spindle midzone to push the spindle poles apart, and the astral MTs anchoring the PCM to the plasma membrane (43). This complex machine generates a combination of pulling and pushing forces on the chromosomes and the CEs to drive the formation of a symmetric and bioriented spindle, which is necessary for chromosome alignment at the metaphase plate (45).

The CEs are also physically connected to the GC through the polymerization of radial MTs, which form tracks along which Golgi membranes are transported to the cell center (46, 47). Therefore, it is possible that an intact Golgi ribbon could exert a drag force on the separating CEs during the G2/M transition, altering their position and thus resulting in aberrant spindle formation. To evaluate this possibility, we examined whether a bulky unseparated GC could perturb the correct separation and positioning of the CEs during G2 and prophase. To this purpose, HeLa cells were synchronized for G2/M transition by double-thymidine block and treated with vehicle or SP and MK, as shown in Fig 1A. The cells were then fixed and labeled with antibodies against pericentrin and GRASP65 to identify the CEs and the GC, respectively, and with Hoechst to monitor the level of chromosomes condensation (Fig 4A). To measure potential perturbations of CEs separation, we assessed their “centralization” by measuring the differential distance of the two CEs from the nucleus center (∆ = delta centralization) (Fig 4B) and CE “separation” by measuring the distance between the two centrosomes (d = distance) (Fig 4C). Our results demonstrated that the block of Golgi unlinking did not significantly alter the positioning or the separation of the CEs during both G2 and prophase (Fig 4B and C), excluding the possibility that a bulky unlinked GC could exert an important drag force on CEs repositioning. Our data are also supported by the finding that the connection between the CEs and the GC is temporarily loosened during G2 (48).

Having excluded that an intact ribbon could hamper CE positioning in G2, we focused on the major Golgi-dependent spindle defect (i.e., multipolarity). Using confocal microscopy, we observed that this alteration is the consequence of the formation of multiple MT nucleation foci (Fig 5A) that, in turn, are the result of the fragmentation of either CEs or PCM (45). To this end, the cells treated with control buffer or with SP and MK were fixed and labeled with antibodies against ninein (49), which is a core CE marker that localizes primarily to the postmitotic mother centriole and pericentrin, a PCM component (50). As shown in Fig 5B, control cells showed two pericentrin foci that colocalized with two ninein-labeled structures. Conversely, the cells treated with SP and MK showed multiple pericentrin-labeled foci but only two ninein-marked CEs. Thus, these results demonstrated that lack of ribbon unlinking in G2 causes PCM fragmentation in metaphase, resulting in the aberrant spindle formation (45).

The fragmentation of PCM results from imbalanced forces exerted by the MT fibers on the PCMs. A major regulator of spindle fibers formation and function is the kinase Aurora-A, which during G2 becomes associated with the CEs to prepare the PCM to control MT nucleation and stabilization. The PCM comprises many scaffold proteins forming a large multimeric structure that anchors the newly nucleated MTs (51). Significantly, reduced Aurora-A activity results in altered MT nucleation and stabilization, leading to unbalanced MT-mediated forces during prometaphase and PCM fragmentation (51). Of note, ribbon unlinking acts as a specific signal to stimulate the activation of Src at the GC, resulting in the phosphorylation of Aurora-A at Tyr148, a residue that is distinct from the canonical Thr288 activation site. The Src-mediated phosphorylation stimulates the kinase activity and also the recruitment at the CEs (14). Thus, we tested the hypothesis that forcing the cells to enter mitosis with an intact ribbon could result in reduced Aurora-A recruitment, which is necessary for the CE-based functions of this kinase (16). To this purpose, we used HeLa cells as we have previously addressed the role of Aurora-A and the experimental conditions to precisely control its expression in this cell model (14, 38, 44). The cells were synchronized for G2/M transition and treated with SP and MK in the absence or presence of BFA. The cells were fixed and stained for immunofluorescence with antibodies against pericentrin and Aurora-A (Fig 6A) to measure Aurora-A fluorescence levels at the CEs during G2 and prophase (38). Aurora-A activity can be stimulated by multiple mechanisms, including phosphorylation of sites distinct from Thr288 and binding to allosteric regulators, which operate only after its recruitment (52). Therefore, the level of recruitment at the CEs is a reliable proxy of Aurora-A activity (38, 52). Quantification of Aurora-A fluorescence intensity at the CEs during G2 and prophase showed that the treatment with SP and MK caused about a 40% reduction of its levels (Fig 6B). As a control, the fluorescence intensity of pericentrin during G2 and prophase did not change after treatment with SP and MK (Fig S5), indicating that the reduced Aurora-A recruitment was not an indirect effect of reduced availability of PMC. Aurora-A functional activity is not directly proportional to its local concentration but requires the reaching of a specific threshold to be effective (53), indicating that the observed reduction is physiologically relevant (53). Importantly, the reduction of Aurora-A recruitment was a direct consequence of the lack of Golgi unlinking as it was fully recovered by BFA-induced GC disassembly (Fig 6B).

To examine whether the reduced level of Aurora-A at the CEs is a major cause of the spindle defects, we restored its local physiological concentration by transfecting Aurora-A-GFP. The precise control of Aurora-A expression is crucial because a high expression level can alter many mitosis-related events (16). Therefore, the experimental conditions were fine-tuned to induce expression levels of exogenous Aurora-A similar to the endogenous protein, as previously described (38). As shown in Fig 6C and D, treatment of HeLa cells with SP and MK induced aberrant spindle in about 74% of the cells, in comparison to about 25% in untreated cells. Importantly, transfection of Aurora-A in HeLa cells synchronized for mitotic entry, induced a substantial recovery, about 50%, of correct spindle formation. These results indicate that a major cause of the GC-dependent spindle alterations is an insufficient level of Aurora-A activation at the CEs during G2/prophase, likely resulting in defective control of MT polymerization or stabilization the at the PCM (51, 54).

Golgi unlinking is also required for cytokinesis

Defective spindles can result in cell death or abnormalities during mitotic exit (55). Therefore, we also examined the effects on postmitotic events after entering into mitosis with an intact GC in G2. To this end, NRK cells were synchronized for G2/M enrichment and treated to force entry into mitosis with an intact GC, as described in Fig 2A, with the difference that the cells were fixed 16 h after the mitotic peak as this time is sufficient to allow the duplication of most of the cells. The fixed cells were labeled for immunofluorescence with antibodies against GRASP65 and α-tubulin to detect the GC and the MTs, respectively, and Hoechst to visualize the DNA. Strikingly, confocal imaging showed that after forcing mitotic entry with an intact GC (SP + MK), more than 60% of the cells became binucleated (Fig 7A and B). MK alone did not induce significant alterations of cell duplication. At the same time, the treatment with SP alone resulted in about 25% of binucleated cells (Fig 7B), in line with previous observations (56). Importantly, BFA-induced GC disassembly rescued cell division to levels similar to SP-treated cells, indicating that the block of Golgi unlinking is the direct cause of more than 50% of the observed binucleation events (Fig 7B).

We also tested this conclusion with an independent approach. RPE-1 cells were depleted of GRASP55 to induce irreversible Golgi unlinking (27, 57) and synchronized for entry into mitosis by the double-thymidine block. The cells were forced to enter mitosis with an intact GC, as previously described, and fixed 16 h after the mitotic peak. The cells were stained with antibodies against GRASP55 and α-tubulin to label the GC and the MTs, respectively, and Hoechst to label the DNA. MK alone did not alter the binucleation index, whereas SP caused about 22% of binucleation. The combined treatment with SP and MK resulted in more than 55% of binucleated cells (Fig 7C). Importantly, in cells with constitutively unlinked GC (GR55-KD), the normal duplication was rescued to levels analogous to the treatment with SP alone (Fig 7C), further confirming that a significant fraction of the binucleation events is the direct result of the lack of Golgi unlinking.

On the other hand, binucleation can be the consequence of either failure of midbody cleavage or endomitosis, the latter consisting of the formation of two nuclei without cell division (37, 58). Thus, to distinguish between these two possibilities, the cells were synchronized, treated to force entry into mitosis in the absence or presence of BFA, and fixed 90 min after the mitotic peak. This is a suitable timing for measuring metaphase and cytokinesis events, the last-mentioned defined by the formation of the cleavage furrow. After fixation, the cells were stained with α-tubulin to visualize the cleavage furrow, GRASP65 to observe the GC, and Hoechst to label the DNA (Fig S6A). The quantification of the results showed that, compared with control cells, BFA did not alter the fraction of cells in metaphase or cytokinesis (Fig S6B). Treatment with SP reduced both populations, in line with the inhibitory effect on mitotic entry. At the same time, MK increased the fraction of cells in cytokinesis of about two times, a likely consequence of the stimulation of mitotic entry. Finally, the treatment with the combination of SP and MK induced an increase in the number of cells in metaphase, about the double compared with control cells, indicating a delay of the metaphase/anaphase transition, with a consequential and partial reduction of cells in cytokinesis. Importantly, the addition of BFA to cells treated with SP and MK, a treatment that rescues normal cell division, did not change the numbers of cells in cytokinesis compared with cells treated with SP and MK alone (Fig S6B). Therefore, the BFA-induced rescue of normal cell division in cells treated with SP and MK does not correlate with an increased number of cytokinesis events, indicating that endomitosis is not the mediator of the Golgi-dependent binucleation, which likely could result from failed midbody cleavage. In addition, the BFA-mediated recovery of correct cell duplication does not cause cell death, as evidenced by the counting at the microscope of the number of nuclei per field of view (n/f.o.v.) 24 h after aphidicolin washout (Fig S6C), indicating that BFA does not cause an apparent recovery of cell duplication by death-mediated elimination of the failed mitotic events.

Altogether, these data indicate that the substantial increase of binucleation observed in cells treated with SP and MK could be the direct consequence of midbody cleavage failure in cells with an intact GC ribbon in G2. Binucleation can have critical pathological consequences as the associated tetraploidization, even if transient, can lead to chromosomal instability, aneuploidy, and tumorigenesis (59).

Cell cultures

NRK cells (Cat# 86032002, RRID:CVCL_3758; ECACC) were from the American Type Culture Collection (Manassas, Virginia) and HeLa (Cat# 93021013, RRID:CVCL_0030; ECACC) cells were from the European Collection of Authenticated Cell Cultures (Manassas, Virginia) and were cultured in DMEM and MEM, respectively (Invitrogen), supplemented with 10% FCS (Biochrom), 100 μM minimal essential medium nonessential amino-acid solution, 2 mM L-glutamine, 1 U/ml penicillin, and 50 μg/ml streptomycin (all Invitrogen). hTERT‐RPE1 (kindly provided by B. Franco, TIGEM) were grown in DMEM/F12 with 10% FBS (Gibco, BRL), 2 mM l‐glutamine, 1 U/ml penicillin, and 50 μg/ml streptomycin (Invitrogen). All the cell lines were grown under a controlled atmosphere in the presence of 5% CO₂ at 37°C.

Synchronization protocol

HeLa and RPE-1 cells are synchronized using double-thymidine cell cycle arrest: the cells were maintained in growth medium plus 2 mM thymidine for 16 h and then rinsed and maintained in growth medium for 8 h. The cells were then maintained in thymidine for additional 16 h before the final drug washout. For the cell cycle synchronization of NRK cells, the population was incubated with aphidicolin O/N, and the day after, cells were released from the aphidicolin block. The G2/M phase of NRK and HeLa cells was reached after 8 h from the washout.

Mitotic index assay

For mitotic index analysis of synchronized cells, the cells were plated at 60% confluence on 15-mm coverslips. The mitotic index was estimated by measuring the number of cells, stained with Hoechst 33342 and showing clear mitotic (condensed chromosomes) and interphase (diffuse nuclear staining) features, using a confocal laser microscope (Zeiss LSM700 confocal microscope system; Carl Zeiss) equipped with ×63 1.4 NA oil objective.

Flow cytometry

NRK cells were synchronized with aphidicolin and RPE-1 cells with the double-thymidine cell cycle arrest protocol. Trypsinized cells were pelleted, washed in cold PBS, and resuspended in ice-cold ethanol while vortexing. The cells were incubated O/N at 4°C. The next day, the ethanol was removed by centrifugation and the cells were washed in cold PBS and incubated with 50 μg/ml propidium iodide (Invitrogen) for 30 min in the presence of RNAse (Sigma-Aldrich). The cells were then analyzed using the Becton Dickinson FACSCantoA instrument (BD). The data were plotted with Diva software, with 20,000 events analyzed for each sample.

Antibodies and reagents

Aphidicolin (2.5 μg/ml), thymidine (2 mM), brefeldin A (200 ng/ml), SP600125 (25 μM), and fibronectin (10 μg/ml) were from Sigma‐Aldrich. MK1775 (0.3 μM) was from Selleckchem. DMSO was from Carlo Erba. Hoechst 33342 was from Invitrogen. Mowiol 4‐88 was from Sigma‐Aldrich. The antibodies were from the following sources: anti‐GRASP65 (Cat# ab48533, RRID:AB_880296; Abcam) (1:3,000) and anti-pericentrin (Cat# ab4448, RRID:AB_304461; Abcam) were from Abcam; anti‐α‐tubulin (Clone DM1‐A) (RRID:AB_3067992) (1:5,000) was from Sigma-Aldrich; anti-GRASP55 was from Novus Biologicals (Cat# NBP1-00862, RRID:AB_1503318; Novus) (1:500); anti-Aurora-A was from BD Transduction Laboratories (1:200); anti-Ninein was from BioLegend (1:200) (Cat# 602802, RRID:AB_2251471; BioLegend). Alexa 488– and Alexa 568–conjugated secondary antibodies (Cat# 710369, RRID:AB_2532697, and Cat# A-11004, RRID:AB_2534072, respectively) (1:400) were from Thermo Fisher Scientific.

Plasmids, siRNAs, and transfection

The GRASP55 vector was targeted using a smart pool from Dharmacon directed against the sequences: #1 5′-GGAGUGAGCAUUCGUUUCU-3′; #2,5′-GUAAACCAGUCCCUCACUU-3′; #3,5′-GACCACACAGUGAUUAUAU-3′; #4,5′-UGUCGAGAAGUGAUUAUUA-3′. The siRNA duplexes were transfected using Lipofectamine 2000 (Invitrogen), according to the manufacturer instructions. As a non-targeting (control), we used the siRNA duplex 5′‐UUCUCCGAACGUGUCACGUdTdT‐3′ (Sigma-Aldrich). The cells were further processed according to the experimental design, and depletion was assessed by immunofluorescence. For transfection of Aurora-A-GFP, HeLa cells were arrested in S-phase with double-thymidine block and transfected in 24 well with 0.25 μg of vector for 24 h with the TransIT-LT1 Transfection Reagent (Mirus), according to the manufacturer’s instructions. The cDNA of full-length Aurora-A fused to GFP was a kind gift from Dr. Stefano Ferrari (Institute of Molecular Cancer Research, University of Zurich, Zurich, Switzerland).

Immunofluorescence microscopy

The cells were grown on 10 μg/ml fibronectin‐coated glass coverslips (Sigma‐Aldrich) and treated as described above. They were then fixed with 4% PFA (Electron Microscopy Sciences, Hatfield, Pennsylvania) for 10 min at RT. The blocking buffer (0.5% BSA, 0.1% saponin, 50 mM NH₄Cl in PBS supplemented with sodium azide) was then added to the cells for 20 min. For the labeling of α‐tubulin, the cells were fixed with methanol at −20°C for 7 min and blocked with 1% BSA in PBS. In any case, the samples were washed in PBS and incubated for 1 h at RT or overnight at 4°C with the primary antibodies in the blocking buffer. The secondary antibodies plus 2 μg/ml Hoechst were incubated for 40 min at RT. The coverslips were then mounted on glass microscope slides with Mowiol 4-88 (Sigma‐Aldrich). The cells were fixed and stained with Hoechst 33342 for labeling of the nuclei. Immunofluorescence samples were examined using a confocal laser microscope (Zeiss LSM700 confocal microscope system; Carl Zeiss) equipped with ×63 1.4 NA oil objective. Optical confocal sections were taken at 1 Airy unit, with a resolution of 512 × 512 pixels or 1,024 × 1,024 pixels, and exported as .JPEG or .TIF files. The images were cropped with Adobe Photoshop CS6 and composed using Adobe Illustrator CS6. High‐resolution images were acquired at the Zeiss LSM880 using an Airyscan detector and further processed with deconvolution algorithms of Zeiss Zen Black 2.1 software.

Quantitative analysis of GC area fragmentation

The analysis was performed using ImageJ. Images were subjected to thresholding, and the number of particles (“Golgi objects”) was calculated with the Analyze Particles function. All the images were acquired at maximal resolution, under fixed-imaging conditions. For qualitative analysis of the phenotype of the GC, the GRASP65 staining was processed with the freeware ImageJ software package using the “Invert” function.

Statistical analysis

Statistical significance was determined by an unpaired, two‐tailed, t test.