Role of Cytoskeletal Elements in the Recruitment of Cx43-GFP and Cx26-YFP into Gap Junctions

Cytoskeletal elements may be important in connexin transport to the cell surface, cell surface gap junction plaque formation and/or gap junction internalization. In this study, fluorescence recovery after photobleaching was used to examine the role of microfilaments and microtubules in the recruitment and coalescence of green fluorescent protein-tagged Cx43 (Cx43-GFP) or yellow fluorescent tagged-Cx26 (Cx26-YFP) into gap junctions in NRK cells. In untreated cells, both Cx26-YFP and Cx43-GFP were recruited into gap junctions within photobleached areas of cell-cell contact within 2 hrs. However, disruption of microfilaments with cytochalasin B inhibited the recruitment and assembly of both Cx26-YFP and Cx43-GFP into gap junctions within photobleached areas. Surprisingly, disruption of microtubules with nocodazole inhibited the recruitment of Cx43-GFP into gap junctions but had limited effect on the transport and clustering of Cx26-YFP into gapjunctions within the photobleached regions of cell-cell contact. These results suggest that the recruitment of Cx43-GFP and Cx26-YFP to the cell surface or their lateral clustering into gap junctions plaques is dependent in part on the presence of intact actin microfilaments while Cx43-GFP was more dependent on intact microtubules than Cx26-YFP.     

Role of cytoskeletal elements in the recruitment of Cx43-GFP and Cx26-YFP into gap junctions

Cytoskeletal elements may be important in connexin transport to the cell surface, cell surface gap junction plaque formation and/or gap junction internalization. In this study, fluorescence recovery after photobleaching was used to examine the role of microfilaments and microtubules in the recruitment and coalescence of green fluorescent protein-tagged Cx43 (Cx43-GFP) or yellow fluorescent tagged-Cx26 (Cx26-YFP) into gap junctions in NRK cells. In untreated cells, both Cx26-YFP and Cx43-GFP were recruited into gap junctions within photobleached areas of cell-cell contact within 2 hrs. However, disruption of microfilaments with cytochalasin B inhibited the recruitment and assembly of both Cx26-YFP and Cx43-GFP into gap junctions within photobleached areas. Surprisingly, disruption of microtubules with nocodazole inhibited the recruitment of Cx43-GFP into gap junctions but had limited effect on the transport and clustering of Cx26-YFP into gap junctions within the photobleached regions of cell-cell contact. These results suggest that the recruitment of Cx43-GFP and Cx26-YFP to the cell surface or their lateral clustering into gap junctions plaques is dependent in part on the presence of intact actin microfilaments while Cx43-GFP was more dependent on intact microtubules than Cx26-YFP.     

Role of Cytoskeletal Elements in the Recruitment of Cx43-GFP and Cx26-YFP into Gap Junctions

Cytoskeletal elements may be important in connexin transport to the cell surface, cell surface gap junction plaque formation and/or gap junction internalization. In this study, fluorescence recovery after photobleaching was used to examine the role of microfilaments and microtubules in the recruitment and coalescence of green fluorescent protein-tagged Cx43 (Cx43-GFP) or yellow fluorescent tagged-Cx26 (Cx26-YFP) into gap junctions in NRK cells. In untreated cells, both Cx26-YFP and Cx43-GFP were recruited into gap junctions within photobleached areas of cell-cell contact within 2 hrs. However, disruption of microfilaments with cytochalasin B inhibited the recruitment and assembly of both Cx26-YFP and Cx43-GFP into gap junctions within photobleached areas. Surprisingly, disruption of microtubules with nocodazole inhibited the recruitment of Cx43-GFP into gap junctions but had limited effect on the transport and clustering of Cx26-YFP into gapjunctions within the photobleached regions of cell-cell contact. These results suggest that the recruitment of Cx43-GFP and Cx26-YFP to the cell surface or their lateral clustering into gap junctions plaques is dependent in part on the presence of intact actin microfilaments while Cx43-GFP was more dependent on intact microtubules than Cx26-YFP.     

Actin microfilaments facilitate the retrograde transport from the Golgi complex to the endoplasmic reticulum in mammalian cells

The morphology and subcellular positioning of the Golgi complex depend on both microtubule and actin cytoskeletons. In contrast to microtubules, the role of actin cytoskeleton in the secretory pathway in mammalian cells has not been clearly established. Using cytochalasin D, we have previously shown that microfilaments are not involved in the endoplasmic reticulum–Golgi membrane dynamics. However, it has been reported that, unlike botulinum C2 toxin and latrunculins, cytochalasin D does not produce net depolymerization of actin filaments. Therefore, we have reassessed the functional role of actin microfilaments in the early steps of the biosynthetic pathway using C2 toxin and latrunculin B. The anterograde endoplasmic reticulum-to-Golgi transport monitored with the vesicular stomatitis virus-G protein remained unaltered in cells treated with cytochalasin D, latrunculin B or C2 toxin. Conversely, the brefeldin A-induced Golgi membrane fusion into the endoplasmic reticulum, the Golgi-to-endoplasmic reticulum transport of a Shiga toxin mutant form, and the subcellular distribution of the KDEL receptor were all impaired when actin microfilaments were depolymerized by latrunculin B or C2 toxin. These findings, together with the fact that COPI-coated and uncoated vesicles contain β/γ-actin isoforms, indicate that actin microfilaments are involved in the endoplasmic reticulum/Golgi interface, facilitating the retrograde Golgi-to-endoplasmic reticulum membrane transport, which could be mediated by the orchestrated movement of transport intermediates along microtubule and microfilament tracks.     

Control of Intracellular Movement of Connexins by E-Cadherin in Murine Skin Papilloma Cells

The gap junctional intercellular communication-deficient mouse skin papilloma cell line P3/22 expresses Cx43 but not E-cadherin. The E-cadherin gene-transfected cells (P3E1) communicate in a calcium-dependent manner and they were used to study how E-cadherin restores the function of connexins. At low calcium, Cx43 molecules remain in the cytoplasm of P3E1 cells and appear at cell-cell contact areas only in high-calcium medium. While Cx43 is unphosphorylated in P3E1 cells in low-calcium medium, two phosphorylated bands appeared at high calcium. However, when Cx26, which has no C-terminal tail that can undergo phosphorylation, was expressed in P3E1 cells, this connexin also moved to the plasma membrane after the calcium shift and partly colocalized with Cx43, suggesting that C-terminal phosphorylation is not essential for E-cadherin-mediated intracellular transport of connexins. In low calcium, both Cx26 and Cx43 remained and colocalized in the endoplasmic reticulum. As early as 30 min after the shift to high-calcium medium, both Cx43 and Cx26 began to accumulate in the Golgi apparatus. Intracellular movement of connexins to the cytoplasmic membrane at high calcium was effectively blocked by cytochalasin D and brefeldin A. These results suggest that E-cadherin junction formation at high calcium leads to formation of actin cables, which directly or indirectly transport connexins from the cytoplasm to the cell-cell contact membranes via the Golgi apparatus.     

Trafficking, assembly, and function of a connexin43-green fluorescent protein chimera in live mammalian cells.

To examine the trafficking, assembly, and turnover of connexin43 (Cx43) in living cells, we used an enhanced red-shifted mutant of green fluorescent protein (GFP) to construct a Cx43-GFP chimera. When cDNA encoding Cx43-GFP was transfected into communication-competent normal rat kidney cells, Cx43-negative Madin-Darby canine kidney (MDCK) cells, or communication-deficient Neuro2A or HeLa cells, the fusion protein of predicted length was expressed, transported, and assembled into gap junctions that exhibited the classical pentalaminar profile. Dye transfer studies showed that Cx43-GFP formed functional gap junction channels when transfected into otherwise communication-deficient HeLa or Neuro2A cells. Live imaging of Cx43-GFP in MDCK cells revealed that many gap junction plaques remained relatively immobile, whereas others coalesced laterally within the plasma membrane. Time-lapse imaging of live MDCK cells also revealed that Cx43-GFP was transported via highly mobile transport intermediates that could be divided into two size classes of <0.5 microm and 0.5-1.5 microm. In some cases, the larger intracellular Cx43-GFP transport intermediates were observed to form from the internalization of gap junctions, whereas the smaller transport intermediates may represent other routes of trafficking to or from the plasma membrane. The localization of Cx43-GFP in two transport compartments suggests that the dynamic formation and turnover of connexins may involve at least two distinct pathways.     

Glycosylation Regulates Pannexin Intermixing and Cellular Localization

The pannexin family of mammalian proteins, composed of Panx1, Panx2, and Panx3, has been postulated to be a new class of single-membrane channels with functional similarities to connexin gap junction proteins. In this study, immunolabeling and coimmunoprecipitation assays revealed that Panx1 can interact with Panx2 and to a lesser extent, with Panx3 in a glycosylation-dependent manner. Panx2 strongly interacts with the core and high-mannose species of Panx1 but not with Panx3. Biotinylation and dye uptake assays indicated that all three pannexins, as well as the N-glycosylation-defective mutants of Panx1 and Panx3, can traffic to the cell surface and form functional single-membrane channels. Interestingly, Panx2, which is also a glycoprotein and seems to only be glycosylated to a high-mannose form, is more abundant in intracellular compartments, except when coexpressed with Panx1, when its cell surface distribution increases by twofold. Functional assays indicated that the combination of Panx1 and Panx2 results in compromised channel function, whereas coexpressing Panx1 and Panx3 does not affect the incidence of dye uptake in 293T cells. Collectively, these results reveal that the functional state and cellular distribution of mouse pannexins are regulated by their glycosylation status and interactions among pannexin family members.     

Pannexin 2 Is Expressed by Postnatal Hippocampal Neural Progenitors and Modulates Neuronal Commitment

The pannexins (Panx1, -2, and -3) are a mammalian family of putative single membrane channels discovered through homology to invertebrate gap junction-forming proteins, the innexins. Because connexin gap junction proteins are known regulators of neural stem and progenitor cell proliferation, migration, and specification, we asked whether pannexins, specifically Panx2, play a similar role in the postnatal hippocampus. We show that Panx2 protein is differentially expressed by multipotential progenitor cells and mature neurons. Both in vivo and in vitro, Type I and IIa stem-like neural progenitor cells express an S-palmitoylated Panx2 species localizing to Golgi and endoplasmic reticulum membranes. Protein expression is down-regulated during neurogenesis in neuronally committed Type IIb and III progenitor cells and immature neurons. Panx2 is re-expressed by neurons following maturation. Protein expressed by mature neurons is not palmitoylated and localizes to the plasma membrane. To assess the impact of Panx2 on neuronal differentiation, we used short hairpin RNA to suppress Panx2 expression in Neuro2a cells. Knockdown significantly accelerated the rate of neuronal differentiation. Neuritic extension and the expression of antigenic markers of mature neurons occurred earlier in stable lines expressing Panx2 short hairpin RNA than in controls. Together, these findings describe an endogenous post-translational regulation of Panx2, specific to early neural progenitor cells, and demonstrate that this expression plays a role in modulating the timing of their commitment to a neuronal lineage.     

The Golgi apparatus in honeybee photoreceptor cells: structural organization and spatial relationship to microtubules and actin filaments

The architecture of the Golgi complex in honeybee photoreceptors has been analyzed by electron-microscopic techniques. The Golgi apparatus consists of several hundred individual stacks of cisternae dispersed throughout the soma of the photoreceptor cell. Two distinct subpopulations of Golgi stacks are distinguishable by their topographic features: (1) a dense row of Golgi stacks is aligned along the palisade-like cisternae of smooth endoplasmic reticulum backing the photoreceptive microvilli; (2) other Golgi stacks are scattered in the remainder of the cell body. The spatial relationship of Golgi stacks to microtubules and actin filaments has also been determined. Electron-microscopic examination of high-pressure-frozen freeze-substituted retinae reveals that Golgi stacks backing the submicrovillar endoplasmic reticulum reside in a cell area without microtubules, whereas the second subpopulation of Golgi stacks is embedded amidst microtubules. Labeling studies with several actin-specific probes, viz., rhodamine phalloidin, monoclonal anti-actin antibodies, and myosin fragments, provide evidence for a juxtaposition of the submicrovillar Golgi stacks to actin filaments. The Golgi membranes are thus ideally positioned to facilitate the transport of Golgi-derived material toward the microvilli along actin filaments.     

Microinjected anti-actin antibodies decrease gap junctional intercellular commmunication in cultured astrocytes

The purpose of the present study was to investigate the influence of the actin cytoskeleton on gap junctional intercellular communication (GJIC) in cultured astrocytes. This report describes an decrease of GJIC following microinjection of anti-actin antibodies or cytochalasin D treatment. Dye transfer of microinjected neurobiotin was used to assay gap junctional permeability in cultured astrocytes. Besides this translocation of connexins to the plasma membrane we investigated subsequent anti-actin antibody injection. While control cultures exhibited intensive dye spreading of microinjected neurobiotin, GJIC was impaired by microinjection of anti-actin antibodies. Additionally, impaired GJIC was observed after cytochalasin D treatment for 15 min. After the drug had been washed out, a recovery of GJIC was achieved. Cultured astrocytes exhibited a prominent actin cytoskeleton, with strong staining of actin filaments at the plasma membrane. Confocal laser scanning microscopy revealed an impaired translocation of Cx 43 from the Golgi apparatus to the cell membrane of cell processes following anti-actin antibody injection. These results suggest that the morphological integrity of microfilaments seems to be fundamental for GJIC, probably by means of associations among actin filaments, actin binding proteins, and Cx 43 at the plasma membrane or indirectly through the transport of connexins from the cytoplasm to the cell membrane.     

Loss of Gap Junction Plaques and Inhibition of Intercellular Communication in Ilimaquinone-treated BICR-M1Rk and NRK Cells

To examine the mechanism(s) and pathways of gap junction formation and removal a novel and reversible inhibitor of protein secretion, ilimaquinone (IQ), was employed. IQ has been reported to cause the vesiculation of Golgi membranes, block protein transport at the cis-Golgi and depolymerize cytoplasmic microtubules. Connexin43 (Cx43) immunolabeling and dye microinjection experiments revealed that gap junction plaques were lost and intercellular communication was inhibited following IQ treatment for 1 hr in BICR-M1R_k rat mammary tumor cells and for 2 hr in normal rat kidney (NRK) cells. Gap junction plaques and intercellular communication recovered within 2 hr when IQ was removed. IQ, however, did not affect the distribution of zonula occludens-1, a protein associated with tight junctions. Western blot analysis revealed that the IQ-induced loss of gap junction plaques was accompanied by a limited reduction in the highly phosphorylated form of Cx43, previously shown to be correlated with gap junction plaques. The presence of IQ inhibited the formation of new gap junction plaques in BICR-M1R_k cells under conditions where preexisting gap junctions were downregulated by brefeldin A treatment. Treatment of BICR-M1R_k and NRK cells with other microtubule depolymerization agents did not inhibit plaque formation or promote rapid gap junction removal. These findings suggest that IQ disrupts intercellular communication by inhibiting the events that are involved in plaque formation and/or retention at the cell surface independent of its effects on microtubules. Our results also suggest that additional factors other than phosphorylation are necessary for Cx43 assembly into gap junction plaques. Received: 16 January 1996/Revised: 20 September 1996     

Pannexin 1 and Pannexin 3 Channels Regulate Skeletal Muscle Myoblast Proliferation and Differentiation

Pannexins constitute a family of three glycoproteins (Panx1, -2, and -3) forming single membrane channels. Recent work demonstrated that Panx1 is expressed in skeletal muscle and involved in the potentiation of contraction. However, Panxs functions in skeletal muscle cell differentiation, and proliferation had yet to be assessed. We show here that Panx1 and Panx3, but not Panx2, are present in human and rodent skeletal muscle, and their various species are differentially expressed in fetal versus adult human skeletal muscle tissue. Panx1 levels were very low in undifferentiated human primary skeletal muscle cells and myoblasts (HSMM) but increased drastically during differentiation and became the main Panx expressed in differentiated cells. Using HSMM, we found that Panx1 expression promotes this process, whereas it was impaired in the presence of probenecid or carbenoxolone. As for Panx3, its lower molecular weight species were prominent in adult skeletal muscle but very low in the fetal tissue and in undifferentiated skeletal muscle cells and myoblasts. Its overexpression (∼43-kDa species) induced HSMM differentiation and also inhibited their proliferation. On the other hand, a ∼70-kDa immunoreactive species of Panx3, likely glycosylated, sialylated, and phosphorylated, was highly expressed in proliferative myoblasts but strikingly down-regulated during their differentiation. Reduction of its endogenous expression using two Panx3 shRNAs significantly inhibited HSMM proliferation without triggering their differentiation. In summary, our results demonstrate that Panx1 and Panx3 are co-expressed in human skeletal muscle myoblasts and play a pivotal role in dictating the proliferation and differentiation status of these cells.     

Limiting transport steps and novel interactions of Connexin-43 along the secretory pathway

Connexins are four-transmembrane-domain proteins expressed in all vertebrates which form permeable gap junction channels that connect cells. Here, we analysed Connexin-43 (Cx43) transport to the plasma membrane and studied the effects of small GTPases acting along the secretory pathway. We show that both GTP- and GDP-restricted Sar1 prevents exit of Cx43 from the endoplasmic reticulum (ER), but only GTP-restricted Sar1 arrests Cx43 in COP II-coated ER exit sites and accumulates 14-3-3 proteins in the ER fraction. FRET-FLIM data confirm that already in ER exit sites Cx43 exists in oligomeric form, suggesting an in vivo role for 14-3-3 in Cx43 oligomerization. Exit of Cx43 from the ER can be blocked by other factors—such as expression of the β subunit of the COP I coat or p50/dynamitin that acts on the microtubule-based dynein motor complex. GTP-restricted Arf1 blocks Cx43 in the Golgi. Lastly, we show that GTP-restricted Arf6 removes Cx43 gap junction plaques from the cell–cell interface and targets them to degradation. These data provide a molecular explanation of how small GTPases act to regulate Cx43 transport through the secretory pathway, facilitating or abolishing cell–cell communication through gap junctions.     

Characterization and dynamics of cytoplasmic F-actin in higher plant endosperm cells during interphase, mitosis, and cytokinesis

We have identified an F-actin cytoskeletal network that remains throughout interphase, mitosis, and cytokinesis of higher plant endosperm cells. Fluorescent labeling was obtained using actin monoclonal antibodies and/or rhodamine-phalloidin. Video-enhanced microscopy and ultrastructural observations of immunogold-labeled preparations illustrated microfilament-microtubule co-distribution and interactions. Actin was also identified in cell crude extract with Western blotting. During interphase, microfilament and microtubule arrays formed two distinct networks that intermingled. At the onset of mitosis, when microtubules rearranged into the mitotic spindle, microfilaments were redistributed to the cell cortex, while few microfilaments remained in the spindle. During mitosis, the cortical actin network remained as an elastic cage around the mitotic apparatus and was stretched parallel to the spindle axis during poleward movement of chromosomes. This suggested the presence of dynamic cross-links that rearrange when they are submitted to slow and regular mitotic forces. At the poles, the regular network is maintained. After midanaphase, new, short microfilaments invaded the equator when interzonal vesicles were transported along the phragmoplast microtubules. Colchicine did not affect actin distribution, and cytochalasin B or D did not inhibit chromosome transport. Our data on endosperm cells suggested that plant cytoplasmic actin has an important role in the cell cortex integrity and in the structural dynamics of the poorly understood cytoplasm-mitotic spindle interface. F-actin may contribute to the regulatory mechanisms of microtubule-dependent or guided transport of vesicles during mitosis and cytokinesis in higher plant cells.     

Cx43/beta-gal inhibits Cx43 transport in the Golgi apparatus

A connexin construct consisting of bacterial beta-galactosidase fused to the C-terminus of connexin43 (Cx43/beta-gal) was used to examine Cx43 assembly in NIH 3T3 cells. Cx43/beta-gal is retained in a perinuclear compartment and inhibits Cx43 transport to the cell surface. The intracellular connexin pool trapped by Cx43/beta-gal was retained in a compartment that co-localized with a medial Golgi apparatus marker by immunofluorescence microscopy and that was readily disassembled by treatment with brefeldin A. Further analysis by sucrose gradient fractionation showed that Cx43 and Cx43/beta-gal were assembled into a sub-hexameric complex, and that Cx43/beta-gal expression also inhibited Cx43 assembly into hemichannels. While this is consistent with Cx43 hemichannel assembly in the trans Golgi network (TGN), these data also suggest that the dominant negative effect of Cx43/beta-gal on Cx43 trafficking may reflect a putative sub-hexameric assembly intermediate formed in the Golgi apparatus.     

Cx43/β-Gal Inhibits Cx43 Transport in the Golgi Apparatus

A connexin construct consisting of bacterial β-galactosidase fused to the C-terminus of connexin43 (Cx43/β-gal) was used to examine Cx43 assembly in NIH 3T3 cells. Cx43/β-gal is retained in a perinuclear compartment and inhibits Cx43 transport to the cell surface. The intracellular connexin pool trapped by Cx43/β-gal was retained in a compartment that co-localized with a medial Golgi apparatus marker by immunofluorescence microscopy and that was readily disassembled by treatment with brefeldin A. Further analysis by sucrose gradient fractionation showed that Cx43 and Cx43/β-gal were assembled into a sub-hexameric complex, and that Cx43/β-gal expression also inhibited Cx43 assembly into hemichannels. While this is consistent with Cx43 hemichannel assembly in the trans Golgi network (TGN), these data also suggest that the dominant negative effect of Cx43/β-gal on Cx43 trafficking may reflect a putative sub-hexameric assembly intermediate formed in the Golgi apparatus.     

Pannexin 3 Regulates Intracellular ATP/cAMP Levels and Promotes Chondrocyte Differentiation

Pannexin 3 (Panx3) is a new member of the gap junction pannexin family, but its expression profiles and physiological function are not yet clear. We demonstrate in this study that Panx3 is expressed in cartilage and regulates chondrocyte proliferation and differentiation. Panx3 mRNA was expressed in the prehypertrophic zone in the developing growth plate and was induced during the differentiation of chondrogenic ATDC5 and N1511 cells. Panx3-transfected ATDC5 and N1511 cells promoted chondrogenic differentiation, but the suppression of endogenous Panx3 inhibited differentiation of ATDC5 cells and primary chondrocytes. Panx3-transfected ATDC5 cells reduced parathyroid hormone-induced cell proliferation and promoted the release of ATP into the extracellular space, possibly by action of Panx3 as a hemichannel. Panx3 expression in ATDC5 cells reduced intracellular cAMP levels and the activation of cAMP-response element-binding, a protein kinase A downstream effector. These Panx3 activities were blocked by anti-Panx3 antibody. Our results suggest that Panx3 functions to switch the chondrocyte cell fate from proliferation to differentiation by regulating the intracellular ATP/cAMP levels.     

Trafficking pathways of Cx49-GFP in living mammalian cells

In the present study we examined the trafficking pathways of connexin49 (Cx49) fused to green fluorescent protein (GFP) in polar and non-polar cell lines. The Cx49 gene was isolated from ovine lens by RT-PCR. Cx49 cDNA was fused to GFP and the hybrid cDNA was transfected into several cell lines. After transfection of Cx49-GFP cDNA into HeLa cells, it was shown using the double whole-cell patch-clamp technique that the expressed fusion protein was still able to form conducting gap junction channels. Synthesis, assembly, and turnover of the Cx49-GFP hybrid protein were investigated using a pulse-chase protocol. A major 78-kDa protein band corresponding to Cx49-GFP could be detected with a turnover of 16-20 h and a half-life time of 10 h. The trafficking pathways of Cx49-GFP were monitored by confocal laser microscopy. Fusion proteins were localized in subcellular compartments, including the endoplasmic reticulum (ER), the ER-Golgi intermediate compartment, the Golgi apparatus, and the trans-Golgi network, as well as vesicles traveling towards the plasma membrane. Time-dependent sequential localization of Cx49-GFP in the ER and then the Golgi apparatus supports the notion of a slow turnover of Cx49-GFP compared to other connexins analyzed so far. Gap junction plaques resembling the usual punctuate distribution pattern could be demonstrated for COS-1 and MDCK cells. Basolateral distribution of Cx49-GFP was observed in polar MDCK cells, indicating specific sorting behavior of Cx49 in polarized cells. Together, this report describes the first characterization of biosynthesis and trafficking of lens Cx49.     

Coalescence of Golgi fragments in microtubule-deprived living cells

The process of stack coalescence, an important mechanism of Golgi recovery from mitosis, was examined using novel experimental paradigms. In living cells with disrupted (by nocodazole) microtubules, galactosyl transferase-GFP-labelled Golgi fragments constantly appeared, grew, sometimes moved with a speed of 1-2 microns/min, coalesced or gradually diminished and disappeared. The rate of Golgi fragment turnover and coalescence was highly balanced to maintain a constant number of Golgi units per cell. Moreover some Golgi islands appear and some received new GalTase-GFP after photobleaching of cell cytoplasm. Short tubules extending from the rims of scattered Golgi fragments frequently formed bridges between ministacks, inducing their coalescence. The frequency of coalescence could also be inhibited by disruption of actin microfilaments. After the Golgi redistribution into endoplasmic reticulum induced by brefeldin A, either the growth of small Golgi fragments or their coalescence leads to compartmentalized stack formation without the participation of microtubules. These results demonstrate that this coalescence between isolated Golgi stacks is microtubule-independent and could thus be mediated by membranous tubules.