Cytoskeletal Membrane Dynamics - Mark A. McNiven

Actin and Arf1-dependent Recruitment of a Cortactin–dynamin Complex to the Golgi Regulates Post-Golgi Transport

Nature Cell Biology Volume 7 | Number 5 | May 2005 pp483-492

Hong Cao, Shaun Weller, James D. Orth, Jing Chen, Bing Huang, Ji-Long Chen, Mark Stamnes and Mark A. McNiven

Cortactin is an actin-binding protein that has recently been implicated in endocytosis. It binds directly to dynamin-2 (Dyn2), a sm GTPase that mediates the formation of vesicles from the plasma membrane and the Golgi. Here we show that cortactin associates with the Golgi to regulate the actin- and Dyn2-dependent transport of cargo. Cortactin antibodies stain the Golgi apparatus, labelling peripheral buds and vesicles that are associated with the cisternae. Notably, in vitro or intactcell experiments show that activation of Arf1 mediates the recruitment of actin, cortactin and Dyn2 to Golgi membranes. Furthermore, selective disruption of the cortactin–Dyn2 interaction significantly reduces the levels of Dyn2 at the Golgi and blocks the transit of nascent proteins from the trans-Golgi network, resulting in swollen and distended cisternae. These findings support the idea of an Arf1-activated recruitment of an actin, cortactin and Dyn2 complex that is essential for Golgi function.

enlarge

Figure 1 Immunofluorescence staining of cultured epithelial cells reveals a localization of cortactin to the Golgi. (a) The diagram depicts the different domains of cortactin, and the asterisks mark the three different epitope regions of cortactin that are used for the generation of the specific antibodies utilized in this study. (b–c) Immunofluorescence staining of Clone 9 cells and acinar cells with antibodies specific to: the third F-actin-binding site (AB3) of cortactin (b), C-terminal tyrosines (C-Tyr) that are phosphorylated by Src kinases. Cortactin antibody staining is also found in the ruffle and at endocytic pits (arrowhead), which are both known sites of cortactin function. As an alternative to antibody staining, cortactin B–RFP was expressed in Clone 9 cells (c) then fixed and co-stained with an antibody to TGN38 (c´). A nearly exact colocalization between the expressed tagged cortactin protein and the TGN38 antigen can be seen. Scale bars, 10 μm.

enlarge

Figure 2 Cortactin is in intimate contact with Golgi-associated buds and vesicles. (a–e) Electron micrographs of Clone 9 cells that have been incubated with primary antibodies to two distinct cortactin domains, C-Tyr (a–c) and AB3 (d, e), and labelled with a gold-conjugated secondary antibody. Arrowheads indicate cortactin labelling of the peripheral buds of Golgi stacks, which is seen frequently, whereas arrows indicate labelling of peripheral Golgi buds and vesicles at both the cis and trans sides. Little gold labelling is observed on the Golgi stacks proper (GS). (f) A total of 20 distinct Golgi fields were examined to determine the number of gold particles labelling either Golgi or mitochondrial membranes (M). The average number of gold particles along each micrometre of membrane within the two distinct structures was calculated. There was, on average, fivefold more particles for each micrometre of Golgi membrane than of mitochondria membrane. Error bars denote s.d. Scale bars, 100 nm.

enlarge

Figure 3 Binding of cortactin and dynamin to isolated Golgi membranes is dependent upon Arf1-activated actin recruitment. (a) Western blot analysis of an in vitro Golgi-binding assay was used to determine what factors are required for recruitment of cortactin and dynamin proteins to Golgi membranes. With GTP-γS in the assay, there was a marked increase in Golgi membrane-bound cortactin and dynamin. However, inclusion of either BFA, to inhibit Arf1 activity, or the actin-filament-disrupting drug latrunculin A (Lat A; 3 μM) resulted in a marked inhibition of actin, cortactin and dynamin recruitment to these membranes. The recovery of membranes after the assay period was assessed by blotting for the Golgi resident protein MannII. (b) To test further the role of Arf1 in the GTPdependent recruitment, the same assay was performed with, or without, added Arf1 protein ‘preloaded’ with GTP and in the presence or absence of BFA. As in a, there was a substantial recruitment of the complex to Golgi membranes, and the inclusion of BFA prevented this recruitment. Whereas the inclusion of preloaded Arf1-GTP (b; lane 3) to the assay had the same positive effect on complex recruitment as the addition of GTP-γS (a; lane 2), the action of preloading the Arf1 before assay addition negated the inhibitory effects of BFA on recruitment (compare a, lane 3 with b, lanes 3 and 4). For cellular studies, rat fibroblasts were treated with 0.5 μM of cytochalasin D for 30 min, fixed and double stained for cortactin (c–e) and TGN38 (c´–e´). Whereas cytochalasin treatment induced some modest disorganization of the Golgi apparatus in cells, cortactin localization to the Golgi (arrows) was altered and appeared as dispersed, peripheral puncta (d, e). Scale bar, 10 μm.

enlarge

Figure 4 Disruption of cortactin function reduces Dyn2 recruitment to the Golgi apparatus. (a, b) Immunofluorescence images of Clone 9 cells expressing CortΔSH3 (asterisks) resulted in a significant loss of Golgi-localized dynamin, noted by Dyn2 (a´) and Hudy-1 staining (b´). (c–e) Microinjection of either a 24-amino-acid peptide derived from the cortactin SH3 domain (c, c´), or purified polyclonal antibodies (d, d´, e, e´) to distinct regions of cortactin (C-Tyr and AB3) also induced a marked reduction in Dyn2 associated with the Golgi. Injected cells are marked with an asterisk and Golgi regions are marked with arrowheads. Scale bar, 10 μm.

enlarge

Figure 5 VSV-G-ts–GFP that is transported from the ER accumulates in the Golgi of cells expressing truncated cortactin proteins. (a–h) Fluorescence micrographs of cultured BHK-21 cells co-transfected with plasmids encoding the secretory marker protein VSV-G-ts–GFP and wild-type (wt) cortactin or truncated CortΔSH3, CortΔY or Dyn2ΔPRD. Co-transfected BHK-21 cells expressing either wild-type (a–d) or mutant (e–h) cortactin following a 16- h recovery period at 40 ºC. The VSV-G-ts–GFP protein is distributed in a diffuse pattern throughout the cytoplasm, consistent with retention in the ER. Following a 15-min incubation at the permissive temperature (32 ºC), both wild-type (b) and mutant cells (f) have transported nascent VSV-G-ts–GFP to a Golgi-like perinuclear compartment (arrows). (c, d) With increased incubation times at the permissive temperature (60 or 120 min), cells expressing wild-type cortactin transported most, if not all, of the nascent VSV-G-ts–GFP out of the perinuclear region to the cell surface. (g, h) During this same time period, cells expressing mutant cortactin retained the viral protein in a perinuclear region, even after 120 min (h). Circles denote standardized areas of quantification in which fluorescence intensities were measured for each time point (i). Scale bar, 10 μm. A similar retention of VSV-G-ts–GFP protein was seen at the 60- and 120-min time points in cells expressing CortΔSH3, CortΔY or Dyn2ΔPRD. (i) The histogram shows the average intensity for at least 40 cells within a given time point after the permissive temperature switch, from two separate experiments. (j) BHK-21 cells co-transfected with VSV-G-ts–GFP and wild-type cortactin or truncated CortΔY were temperature-shifted as in a–i. VSV-G protein was then processed using endo-H glycosidase to assess maturation from an endo-H-sensitive (endo-HS) ER-localized form to an endo-H resistant (endo- HR) form, consistent with transport to the cis-/medial-Golgi compartments. Densitometric analysis of the two molecular weight forms of VSV-G protein indicates that the VSV-G maturation was nearly identical for cells expressing either form of cortactin protein, indicating no effect on ER-to-Golgi transport. Error bars denote the standard error of the mean.