Src kinase regulates the integrity and function of the Golgi apparatus via activation of dynamin 2

Abstract

The size and integrity of the Golgi apparatus is maintained via a tightly controlled regulation of membrane traffic using a variety of different signaling and cytoskeletal proteins. We have recently observed that activation of c-Src has profound effects on Golgi structure, leading to dramatically vesiculated cisternae in a variety of cell types. As the large GTPase dynamin (Dyn2) has been implicated in Golgi vesiculation during secretion, we tested whether inhibiting Dyn2 activity by expression of a Dyn2K44A mutant or siRNA knockdown could attenuate active Src-induced Golgi fragmentation. Indeed, these perturbations attenuated fragmentation, and expression of a Dyn2Y(231/597)F mutant protein that cannot be phosphorylated by Src kinase had a similar effect . Finally, we find that Dyn2 is markedly phosphorylated during the transit of VSV-G protein through the TGN whereas expression of the Dyn2Y(231/597)F mutant significantly reduces exit of the nascent protein from this compartment. These findings demonstrate that activation of Dyn2 by Src kinase regulates Golgi integrity and vesiculation during the secretory process.

In mammalian cells, the Golgi apparatus is arranged as an organized array of membranous stacks that are supported by an associated cytoskeletal matrix (1). Because the Golgi apparatus is a central site for processing and packaging secreted proteins, the organelle is in constant flux. These dynamics are crucial for mediating the secretory function of the Golgi and also play a key role in mitotic progression at the G2/M phase of the cell cycle (2). The regulation of Golgi dynamics, morphology, and function has important implications for vesicular trafficking and cell proliferation.

When testing the effects of active c-Src on the endocytic process, we observed a marked and near complete fragmentation of the Golgi apparatus in cultured cells that can be prevented by pretreatment with the Src inhibitors PP2 or SU6656. In addition, a recent report (3) has shown that Src kinase is activated upon transit of nascent cargo from the ER through the Golgi stacks. Together, these findings suggest that Src may play an important role in defining Golgi structure and regulating its dynamics. Of several Golgi-associated Src substrates that could participate in this regulatory process, the ubiquitous form of dynamin 2 (Dyn2) is an attractive candidate for multiple reasons as it is a well-established mechanoenzyme with a confirmed role in membrane fission at several endocytic and exocytic traffic steps (4, 5), including exit from the Golgi (6–9). Interestingly, others have observed a selective requirement for dynamin in the formation of post-Golgi carriers in nonpolarized cell models. Furthermore, there appears to be a selective dynamin dependence for apical vs. basolateral membrane targeted cargo in polarized cell models, where other proteins including CtBP1/BARS have been reported to play a role (10–12). Moreover, Dyn2 is a substrate of Src and has been shown to be phosphorylated by this kinase during the activation and subsequent endocytic internalization of specific membrane receptors (13–15). In this study, we provide evidence that Src kinase, through regulation of Dyn2, acts to control Golgi dynamics in both normal and neoplastic cells as well as vesiculation of the TGN during the secretory process.

Modulation of Src Activity Affects Golgi Compartment Integrity in Normal and Neoplastic Cells

Transient expression of a constitutively active form of c-Src (SrcY530F) in a variety of cultured cells [rat fibroblasts (RF); HeLa fibroblast-like cells; and rat hepatocytes, Clone 9] resulted in a dramatic fragmentation and vesiculation of the Golgi complex (Fig. 1 A–D). Visual scoring of this phenotype indicated that more than 60% of cells expressing SrcY530F exhibited a fragmented (FRAGM) Golgi phenotype (Fig. 1E). An additional 10–15% of cells showed some marked changes in Golgi integrity with distended cisternae and tubules, which was categorized as an altered (ALT) phenotype. These morphological changes induced by the expression of constitutively active Src could be attenuated and even reversed by overnight treatment of cells with the Src inhibitor PP2, which led to intact and tightly compacted Golgi structures that had few, if any, extending tubules (Fig. 1 F and G). We have used the inhibitory effect of the PP2 drug in combination with expression of active Src to dramatically reduce the time needed to observe Golgi fragmentation. This was achieved by transfecting cells with active Src kinase, allowing a brief recovery period, then incubating these cells with PP2 overnight to prevent fragmentation. Upon rinsing the cells in drug-free media the next day, we could observe a significant vesiculation/fragmentation of the Golgi in just 1–2 h.

Fig. 1.

Active Src kinase induces marked Golgi fragmentation in multiple cell types. (A–D) Golgi (GM130) staining of rat fibroblasts (RFs), HeLa and Clone 9 cells expressing constitutively active Src (SrcY530F). Asterisks denote transfected cells. Arrows highlight the fragmented Golgi phenotype present in transfected cells; arrowheads indicate the normal reticular Golgi morphology present in untransfected cells. (A’) Increased magnification of boxed region in A. (E) Scoring of Golgi morphologies in RFs expressing constitutively active Src. Golgi phenotypes in transfected cells were categorized as normal (NML), having an altered, distended, or compacted appearance (ALT), or displaying fragmentation/vesiculation of the entire Golgi compartment (FRAGM). Data represent the average from three independent experiments; error bars denote SD. (F) Golgi staining (GM130) of RFs after treatment with the Src-inhibitory drug PP2 (50 μM). (Bars, 10 μm.) (G) Quantitation of the average Golgi region major axis lengths in RFs treated with DMSO vehicle or PP2 drug. Data represent average length of the major axis of the Golgi area (pixels) in 50 cells over two experiments. Error bars denote SD.

A variety of tumor cell types exhibit Golgi complexes that are fragmented and dispersed by an unknown mechanism (16). Cultured cells from different human pancreatic epithelial tumor cell lines (HPAF-II, BxPC-3) were stained for a Golgi marker that was detected by immunofluorescence. These cells exhibited markedly fragmented Golgi cisternae that appeared to be evenly dispersed throughout the cytoplasm (Fig. 2 A–B). Western blot analysis of lysates from these pancreatic tumor cells with an antibody that recognizes the phospho-active form of Src revealed markedly increased levels of active Src, considerably (3- to 9-fold) higher than the RFs (Fig. 2 G and H).

Fig. 2.

Tumor-derived cells have a fragmented Golgi phenotype that can be reversed via inhibition of Src activity. (A and B) Golgi staining (GM130) of cultured cells from two different cell lines derived from human pancreatic tumors, HPAF-II (A) and BxPC-3 (B). (C and D) Golgi staining of HPAF and BxPC3 cells following a 12-h or 16-h treatment with the Src-inhibitory compound SU6656 (1.5 μM / 25 μM), respectively. (E) Quantitation of Golgi element length in HPAF-II cells after treatment with either DMSO vehicle or the Src inhibitor SU6656 (25μM) over 16 h. Data represent measurements of Golgi element lengths in a total of 30 cells over three experiments. (F) Average number of Golgi elements measured in HPAF cells treated with 1.5 μM SU6656 or DMSO vehicle for 12 h. Data are average measurements of at least 150 cells in each treatment group. (G) Western blot of whole-cell lysates from RFs, HPAF-II, and BxPC3 cells. Active Src was detected with the anti–phospho-Y416/418 Src antibody (P-Src), and total Src protein was detected with a pan-Src kinase polyclonal antibody (Src). (H) Levels of active Src expressed as relative densitometry units obtained from ratio of P-Src to total Src protein.

To test whether inhibition of Src activity in these tumor cells might lead to a more normal, contiguous Golgi phenotype, cells were treated with the Src inhibitory drug, SU6656 (1.5–25 μM), for 12–16 h and then stained for the Golgi marker protein GM130. Remarkably, drug-treated cells showed considerable reassembly of the previously dispersed Golgi fragments into long, peri-nuclear, stacked elements, similar to normal epithelial cells (Fig. 2 C and D). Quantitative measurement of Golgi element lengths showed a greater than 2-fold increase in the drug-treated cells compared with DMSO-treated control cells (Fig. 2E). Accordingly, there was a decrease in the average number of observable Golgi elements per cell (Fig. 2F). These findings support the idea that Src activity in cells can be sufficiently high to spontaneously alter Golgi morphology.

Active Dynamin Is Required for Src-Induced Golgi Fragmentation.

The findings described above suggest that Src profoundly affects Golgi organelle integrity, possibly by stimulating the vesiculation of Golgi membranes. The large GTPase dynamin 2 (Dyn2) is a Src substrate that has been implicated in the vesiculation events necessary for the exit of nascent cargo from the TGN (6, 8, 17). To test whether inhibition of Dyn2 function prevents Src-induced fragmentation of the Golgi, RFs expressing SrcY530F were cotransfected with the GTPase-defective, dominant-negative mutant Dyn2K44A. Alternatively, Dyn2 levels were reduced by siRNA-mediated knockdown and subsequent re-expression of this dominant-negative dynamin mutant. For both approaches, Golgi morphology was analyzed by fluorescence microscopy after staining for the Golgi marker GM130. As shown in Fig. 3A, RFs expressing wild-type Dyn2 (WTDyn2) and SrcY530F exhibited a fragmented Golgi phenotype, similar to RFs expressing SrcY530F alone (Fig. 1 A and B). In contrast, cells expressing the Dyn2K44A mutant in combination with active SrcY530F displayed Golgi with fully intact reticular stacks (Fig. 3 B and C). Similarly, Golgi integrity appeared to be stabilized in RFs that were treated for 48 h with siRNA oligos targeting Dyn2 (>90% reduction in endogenous levels of Dyn2; Fig. 3D) before expression of SrcY530F and DN-dynamin (Fig. 3 E and F). siRNA treatment alone prevented Golgi fragmentation to a lesser extent, but did not result in the markedly extended Golgi reticulum observed in the Dyn2K44A-expressing cells.

Fig. 3.

Dyn2 function is required for active Src-induced Golgi fragmentation. (A) Golgi staining (GM130) of RFs coexpressing wild-type Dyn2 (WTDyn2) and constitutively active Src kinase (SrcY530F). (B and C) Golgi staining of RFs coexpressing dominant-negative Dyn2 (Dyn2K44A) and constitutively active Src kinase (SrcY530F). (D) Western blot analysis of lysates from RFs treated with mock reagent (M) or with siRNA oligos targeting Dyn2 (D2-KD). Antibodies were against Dyn2 and actin (loading control). (E and F) Golgi staining of RFs depleted of Dyn2 by siRNA treatment and subsequently transfected to express active Src kinase along with a dominant-negative mutant of Dyn2 (D2-KD/D2K44A/Src530). Transfected cell nuclei are denoted with an asterisk. (G) Quantitation of Golgi morphologies observed in RFs following expression of active Src alone (SrcY530F; black bars) or together with either wild-type Dyn2 (SrcY530F/WTDyn2; white bars) or a dominant-negative mutant of Dyn2 (SrcY530F/Dyn2K44A; gray bars). Golgi phenotypes were categorized as described in the legend to Fig. 1. Data represent the average from three experiments. (H) Average frequency of RF cells exhibiting a fragmented Golgi phenotype. Cells were treated with either mock reagents (Mock), expression of constitutively active Src kinase alone (Src530), or active Src kinase expression subsequent to siRNA-mediated knockdown of Dyn2 (Src530/D2-KD) with or without re-expression of a dominant-negative mutant of Dyn2 (Src530/D2-KD/D2K44A). Error bars denote SD. (Bars, 10 μm.)

Quantitation of Golgi phenotypes indicated that Golgi stacks in RFs, expressing SrcY530F alone or in combination with WTDyn2, were fragmented ∼60–70% of the time (FRAGM), with an additional 15% of Golgi exhibiting a partially fragmented morphology (ALT). However, expression of Dyn2K44A with constitutively active Src reduced the percentage of RFs displaying a fragmented Golgi morphology to 35–40%, with a coincident increase in cells containing Golgi of normal (NML) or only partially fragmented morphology (Fig. 3G). The frequency of cells displaying a fragmented Golgi phenotype was reduced by nearly 20% following the knockdown of endogenous dynamin protein (Fig. 3H). Golgi fragmentation associated with active Src expression was substantially reduced (>50%) following the combination of Dyn2 knockdown and subsequent dominant negative re-expression, further implicating the Dyn2 dependence of this Src-driven process.

Stimulation of Secretory Process Leads to Src-Dependent Vesiculation of TGN.

As cisternae of the TGN are used during the formation of nascent secretory vesicles for subsequent transport to the cell surface, it is likely that the structural integrity of this compartment is compromised upon a regulated secretory stimulus (18, 19). Indeed, we have observed in RFs that a timed-pulsed delivery of the temperature-sensitive mutant of the vesicular stomatitis virus G glycoprotein (VSV-G) from the restrictive to the permissive temperature (20) induces a marked fragmentation or complete disappearance of TGN marker stainable compartments of the cell as the VSV-G is transported from the ER, through the Golgi and from the TGN. Surprisingly, TGN38 staining was often observed in a discrete population of cytoplasmic vesicles (Fig. S1 A and A″) or had disappeared completely (Fig. S1 B and B″). Consistent with these findings in fibroblasts, we have previously observed a similar process in bona fide regulated secretory cells. Primary pancreatic acinar cells, in which the Golgi has been labeled with BODIPY-ceramide (Fig. S1C), exhibit dispersal of this TGN lipid marker upon stimulation with 2 × 10⁻⁹ M concentrations of the secretory agonist CCK (21) (Fig. S1 D and E). Thus, it is attractive to predict that activation of a Golgi-associated Src kinase (22, 23), when activated during the transit of nascent secretory proteins from the ER to the Golgi (3), could induce TGN vesiculation.

To confirm that Src activation within the Golgi occurs in response to a traffic pulse of secretory cargo through the Golgi in our cell system, we conducted a series of experiments in BHK-21 fibroblasts. Cells expressing VSV-Gts045-GFP were treated with Src inhibitors or DMSO vehicle while cells were maintained at the restrictive temperature and proteins remained in the ER (schematic in Fig. S2). VSV-G cargo exited the ER and reached the Golgi within 15 min of the permissive temperature shift in both DMSO- and PP2-treated cells (Fig. S2 A and B). Extended incubation of control cells at 32 °C led to the generation of VSV-G cargo–bearing carriers (Fig. S2A′), which were eventually delivered to the cell surface as indicated by diffuse plasma membrane labeling 2 h after the temperature shift (Fig. S2A″). However, PP2-treated cells accumulated and retained the bolus of VSV-G cargo in the Golgi compartment at 2 h (Fig. S2B″).

VSV-G-GFP was immunoprecipitated following cell surface biotinylation, and the amount of protein trafficked was measured by densitometric analysis. As shown in Fig. S2 C and D, control (DMSO-treated) cells had transported nascent VSV-G-GFP to the cell surface by 1 h and 2 h. Even after 2 h at the permissive temperature, the amount of VSV-G-GFP at the cell surface in PP2-treated cells was just 10–20% of the level in control cells.

Dyn2 Phosphorylation by Src Regulates Golgi Integrity.

To define the Src substrates that are phosphorylated during passage of nascent cargo proteins through the Golgi, phospho-protein profiles of homogenates from quiescent cells (i.e., cells in which traffic of VSV-G was arrested in the ER at 40 °C) were compared with those of actively trafficking cells (cells released from the 40 °C temperature block). By Western blot analysis of these lysates with an antibody to p-Tyr, we observed two major groups of phosphoproteins between 100 kDa and 130 kDa and between 70 kDa and 83 kDa (Fig. 4A). As Dyn2 has been shown by several groups to play an important role in Golgi function (7, 9, 10, 24), is a Src substrate, and has a MW of 100 kDa, we tested for Dyn2 activation in trafficking BHK-21 cells. Dyn2 was immunoprecipitated from lysates of VSV-G−trafficking cells and subjected to Western blotting with an anti–p-Tyr antibody. Importantly, a 2- to 3-fold increase in Dyn2 tyrosine phosphorylation was observed in actively trafficking cells (Fig. 4B), and this increase could be attenuated by treatment of cells with the Src kinase inhibitor SU6656. Furthermore, HeLa cells constitutively expressing VSVG cargo over 2 days had a higher level of dynamin tyrosine phosphorylation (Fig. 4C), indicating that the Src-based phosphorylation of dynamin seen with the pulse delivery of Golgi cargo was not an artifact of experimental manipulations involving temperature shifts.

Fig. 4.

Transit of nascent VSV-G protein through the Golgi stimulates tyrosine phosphorylation of Dyn2. (A) Western blot analysis of total p-Tyr (PY-99) in BHK cell lysates incubated at 40 °C for 3 h (40 °C) and in lysates 30 min after the temperature block release of VSV-G cargo (32 °C). (B) Western blot analysis of tyrosine phosphorylated Dyn2 immunoprecipitated from BHK cells expressing wild-type Dyn2-GFP. Cells were either maintained at the 40 °C temperature block or released from the block (32 °C) for 30 min in the presence or absence of SU6656 (5 μM). Graph displaying the densitometric quantitation of Dyn2 tyrosine phosphorylation from two experiments. (C) HeLa cells cotransfected with Dyn2-FLAG and VSVG-GFP or empty vector were incubated at 37 °C for 48 h then lysed and Dyn2 was immunoprecipitated with an anti-FLAG antibody and probed for p-Tyr and Dyn2 by Western blotting. Graph represents average level of Dyn2 phosphorylation relative to total of immunoprecipitated Dyn2, with or without constitutive transport of VSVG-GFP cargo. Data are average obtained from two experiments; error bars denote SD. Dyn2 phosphorylation by Src is required for complete Golgi fragmentation. (D) Western blot analysis of Dyn2 immunoprecipitated from RFs expressing wild-type Dyn2 (WTDyn2) alone (−) or in combination with constitutively active Src kinase (SrcY530F; +). In addition, Dyn2 was immunoprecipitated from RFs expressing active Src kinase in combination with either of two Dyn2 tyrosine phospho-mutants, Dyn2Y597F or Dyn2Y231/597F, and analyzed by Western blot. (E–G) Golgi (GM130) staining of RFs expressing active Src kinase and either wild-type Dyn2 (SrcY530F/WTDyn2; E) or the Dyn2Y231/597F tyrosine phospho-mutant (SrcY530F/Dyn2YF; F and G). (H) Average frequency of RF cells exhibiting a fragmented Golgi phenotype following expression of constitutively active Src kinase (Alone), active Src kinase along with the Dyn2Y231/597F tyrosine phospho-mutant (D2YF), or active Src kinase expression subsequent to siRNA-mediated knockdown of Dyn2 (D2-KD) with or without re-expression of the Dyn2Y231/597F tyrosine phospho-mutant mutant of Dyn2 (D2-KD/D2YF). Data represent average frequency of cells with fragmented Golgi phenotype from at least three experiments. (I) Average Golgi element length measurements from 20 cells expressing constitutively active Src kinase alone (Src530) or in combination with wild-type Dyn2 (Src530/WTD2) or with the Dyn2Y231/597F tyrosine phospho-mutant (Src530/Dyn2YF). Error bars denote SD. (Bars, 10 μm.)

As Dyn2 was identified to be one of the Golgi-associated, tyrosine phosphorylated substrates during VSV-G transit, we predicted that Src-mediated phosphorylation of Golgi-associated Dyn2 could be responsible for the dramatic Golgi fragmentation observed in cells over-expressing this kinase. It has been shown that Src-based phosphorylation of Dyn1 at Y231 and Y597 regulates the assembly and GTPase activity of this enzyme to markedly enhance endocytosis (13, 15). First, we confirmed that Dyn2 is tyrosine phosphorylated in RFs expressing active Src kinase and that Y231 and Y597, conserved between Dyn1 and Dyn2, were the relevant sites of phosphorylation. Indeed, the levels of tyrosine-phosphorylated Dyn2 were increased substantially in cells expressing constitutively active Src compared with mock-transfected cells (Fig. 4D). Importantly, Dyn2 complexes immunoprecipitated from RFs expressing either of the two Dyn2 tyrosine mutants showed only modest staining with the anti–phospho-tyrosine antibody.

The Dyn2Y231/597F tyrosine phospho-mutant was coexpressed with constitutively active Src in RFs, and the effects on Golgi morphology were observed in fixed cells. In comparison with cells expressing wild-type Dyn2 and SrcY530F (Fig. 4E), a more reticular Golgi phenotype of continuous elements was observed in cells expressing Dyn2Y231/597F (Dyn2YF) in combination with active Src (Fig. 4 F and G). When the Dyn2 tyrosine phospho-mutant was expressed with constitutively active Src, the number of RFs that exhibited fragmented Golgi was reduced by ∼50% compared with the expression of constitutively active Src alone or in combination with wild-type Dyn2 (Fig. 3G and Fig. 4H). A similar inhibition of Golgi fragmentation was seen following Dyn2 knockdown and re-expression of this Dyn2 tyrosine phospho-mutant. Furthermore, the average length of Golgi elements was increased ∼2- to 3-fold in cells expressing the Dyn2Y231/597F tyrosine phospho-mutant as detected by fluorescence microscopy (Fig. 4I). In addition, correlative light and electron microscopy confirmed the altered Golgi membrane phenotype induced by expression of constitutively active Src (Fig. S3 A–D). Coexpression of the tyrosine phospho-mutant of Dyn2 resulted in a retention of normal Golgi phenotype that correlated with increases in both average cisternal length and number of cisternae (Fig. S3E). Together, these findings strongly suggest that Src-mediated phosphorylation and activation of Dyn2 plays an important role in Golgi integrity and vesiculation.

Carrier Formation from Golgi Requires Src-Activated Phosphorylation of Dyn2.

As Dyn2 appears to be a Src substrate that participates in Golgi fragmentation (Fig. 3 and Fig. 4), it was important to test whether this signaling event was required for the exit of nascent cargo from the TGN. To this end, we again used the Dyn2Y231/597F(Dyn2YF) mutant in the context of VSV-G trafficking from the TGN in BHK-21 cells. We observed that VSV-G-GFP cargo exited the ER and reached the Golgi within 15 min of the permissive temperature shift in both vector control and Dyn2YF-expressing cells (Fig. 5 A and B). Extended incubation of control cells at 32 °C led to the generation of VSV-G-GFP cargo-bearing carriers by 1 h (Vector CT, Fig. 5A’) that was eventually delivered to the cell surface as indicated by the diffuse plasma membrane labeling at 2 h after the temperature shift (Vector CT, Fig. 5A”). Importantly, Dyn2YF-expressing cells accumulated and retained the VSV-G-GFP cargo in pleiomorphic structures of variable size that were connected with the Golgi even after 2 h of transport (Fig. 5b”). A similar retention of cargo was observed previously in cells that were transfected with a GTPase-deficient Dyn2 mutant (Dyn2K44A) (24) and presumably represent post-Golgi carriers that are unable to undergo fission from the TGN. Together, the two lines of evidence revealed by Src inhibition and the use of a phospho-depleted Dyn2 mutant strongly support the hypothesis that phosphorylation of Dyn2 by Src is essential for the fission of post-Golgi carriers and for cargo export from the Golgi.

Fig. 5.

Tyrosine phosphorylation of Dyn2 is required for normal Golgi-to-plasma membrane transport. (A and B) VSV-G-GFP transport over 2-h period following shift to permissive temperature in BHK-21 cells coexpressing either control vector (A and A”) or phospho-mutant form of dynamin (D2YF) (B and B”). Marked accumulation of nascent VSV-G protein (arrows) is observed at Golgi of mutant Dyn2-expressing cells. (C) Biochemical analysis of VSV-G-GFP trafficking in BHK-21 cells that were coexpressing VSV-G-GFP and either vector alone, Dyn2Y231/597F tyrosine phospho-mutant (Dyn2YF), or GTPase-deficient form of Dyn2 (Dyn2KA). (D) Densitometric quantitation of the experiment shown in C, which were repeated four times for each condition. Error bars denote SEM. (Bars, 10 μm.)

A biochemical method to assess VSV-G-GFP trafficking was used to determine the effects of Src-mediated phosphorylation of Dyn2 on secretion. BHK-21 cells were transfected with either empty vector (control), constructs encoding an inactive GTPase-deficient dynamin (Dyn2K44A), or a construct encoding the Dyn2 Y231/597F phospho-mutant (Dyn2YF). Compared with control cells, Dyn2 mutant–expressing cells showed some reduction in the transport of VSV-G-GFP to the cell surface at the 1-h time point; however, that was markedly increased at 2 h (Fig. 5 C and D). Indeed, we observed a substantial reduction (40–50%) in Golgi-to-plasma membrane transport of VSV-G-GFP in cells expressing the Dyn2YF phospho-mutant.

Finally, an immunofluorescence-based approach was used to detect the relative level of VSVG cargo arriving at the surface of cells following Dyn2 protein knockdown and re-expression (Fig. S4). In mock-treated cells, a substantial amount of surface antigen (LD-VSVG) was present after 90 min of transport following a permissive temperature shift. Within the same permissive temperature shift timeframe, a retention of VSVG-GFP cargo was seen in a juxta-nuclear region (Fig. S4, arrows) in the Dyn2 knock-down cells. Correlating with this cargo retention was a decreased level of VSVG surface antigen. This accumulation was reduced upon re-expression of a WT Dyn2 in these knockdown cells but was maintained in the cells re-expressing the Dyn2 phospho-mutant (Fig. S4, D2YF). As a control, cells that were depleted of endogenous Dyn2 through siRNA mediated knock-down were examined for defects in retrograde transport from the Golgi to the ER through the use of a temperature-sensitive VSVG-KDELR fusion protein. Following incubation of transfected cells at 32 °C, the cargo was found exclusively in the Golgi compartment (Fig. S5). Upon shifting the cells to 40 °C, a condition that favors the retrograde transport to the ER via the KDELR retrieval process, cargo was found within the ER regardless of the Dyn2 protein level. Quantitation of this retrograde transport process showed no significant measurable difference of Golgi to ER transport in control or Dyn2 knock-down cells (Fig. S5C).

To expand upon these findings, additional experiments were performed with a Dyn2 GTPase inhibitory drug (Dynasore). Following treatment with Dynasore, cells exhibit a dramatic retention of VSV-GFP cargo within the TGN, even after 2-h shift to the permissive temperature, whereas vehicle control treated cells exhibit a significant increase in the amount of cargo arriving to the cell surface (Fig. S6A). Quantitation of the Golgi localized cargo at 2 h post–permissive temperature shift revealed a near 3-fold increase in the relative level of retained cargo (Fig. S6B). Interestingly, the dynasore compound mimicked the results of the dynamin knock-down experiments, exhibiting no observable effect on retrograde transport of the VSVG-KDELR chimera (Fig. S6 C and D).

Discussion

This study provides insights into mechanisms of Golgi function, structure, and dynamics. First, in support of the recent study by Pulvirenti et al. (3), we observed that the transit of nascent VSV-G protein through the Golgi cisternae led to substantial activation of the associated Src kinase. Second, traffic-induced activation of Src resulted in marked tyrosine phosphorylation of Dyn2. This Src-mediated Dyn2 phosphorylation is required for membrane vesiculation at the TGN and for exit of protein cargo. Third, prolonged activation of Src kinase stimulated exaggerated, Dyn2-dependent vesiculation that resulted in fragmentation and dispersal of the Golgi stacks. Finally, the abnormally high endogenous levels of Src kinase activity observed in some human tumor cell lines appeared to be responsible for the fragmented Golgi phenotype. These findings could lead to an increased understanding of how aberrant Src activation might affect Golgi morphology and function, protein transport, and cell cycle progression.

While tyrosine phosphorylation has been observed in response to various ligands (13–15, 25–28), this event is thought to play a role in the regulation of ligand-induced uptake of various receptors. However, the mechanisms involved in regulating Dyn2 function during secretion and, more specifically, whether Src kinase-mediated phosphorylation plays a role in this process have remained largely unexplored. Although differences certainly exist, there are marked similarities in the machineries used for vesicle formation at the plasma membrane and at the Golgi apparatus (5, 29). Thus, it will be interesting to compare and contrast how tyrosine phosphorylation of Dyn2 regulates interactions with membranes and proteins at the plasma membrane vs. the Golgi and the effects on endocytosis and secretion, respectively.

A central contribution of the present study is the observation that activation of Src kinase and Dyn2 at the Golgi, either by release of a secretory protein bolus from the ER or by expression of a constitutively active Src protein, leads to a structural imbalance that favors vesiculation of the TGN. This model is supported by the compact Golgi morphology that we observed in cells treated with PP2 and that others have observed in cells lacking Src kinase family members, such as SYF^−/− cells (30). Thus, in normal cells, Src activity is expected to vesiculate the Golgi during a secretory stimulus.

It is important to note that although the expression of active c-Src kinase induces fragmentation of the Golgi, and Src inhibitors exert the opposite affect, we cannot say with certainly that c-Src is the specific kinase in this family responsible for the Golgi dynamics. A yet unidentified nonreceptor tyrosine kinase could provide the final phosphorylation event.

Although our observations implicate Dyn2 as an important substrate of Src kinase that mediates Golgi vesiculation, other Src targets likely contribute to this process. One could easily envision Src activation of multiple Golgi-associated microtubules or actin-based molecular motors (31) or of the many structural components that link Golgi stacks and vesicles (1, 32). However, the fact that inhibition of Dyn2 function alone (by expression of Dyn2K44A or Dyn2Y231/597F proteins or by siRNA knockdown) resulted in maintenance of Golgi integrity in constitutively active Src-expressing cells suggests that this GTPase plays a central role in Golgi vesiculation.

The involvement of Src kinase in Golgi integrity raises the exciting possibility that Src activity is required for Golgi fragmentation and for regulation of the Golgi G2/M mitotic checkpoint. Dispersal of the Golgi apparatus is required for progression into mitosis (2, 33). Furthermore, Src activity plays a role in multiple phases of mitosis (34) and is required for progression beyond G2 (35). It is particularly interesting that the pancreatic tumor cells exhibit fragmented Golgi structures spontaneously. Whether this chronic dispersal actually contributes to the perpetual growth of these cells will need to be defined. It is also interesting that these neoplastic cells secrete proteins despite a completely dispersed Golgi and a severe loss of polarity. How this is achieved is unclear, but it is likely that the normal trafficking and insertion of polarized plasma membrane proteins is compromised in these cells, as reviewed elsewhere (36). Additional studies are needed to define the role of Src-based regulation of Dyn2 in Golgi dynamics during mitotic progression and regulated secretion.

Methods

Cell Culture and Reagents.

Cell lines were obtained from American Type Culture Collection and were maintained as previously described (8).

The polyclonal antibody against Dyn2 was as described previously (37). The Golgi apparatus was detected with antibodies against either p115 or GM130 (BD Biosciences. The anti–pan-Src (SC-18) and anti–c-Src (H-12) antibodies were from Santa Cruz. The anti–phospho-Y416/Y418 Src antibody was from Transduction Labs, and the antibody recognizing phospho-tyrosine residues (4G10) was from Upstate. Secondary antibodies for immunofluorescence and Western blot analysis were described previously (8). The Src inhibitory compounds PP2 and SU6656 were from Calbiochem.

Constructs, siRNA, and Transfection.

The constructs encoding wild-type Dyn2, the dominant-negative Dyn2K44A mutant were described previously (6, 24). Wild-type c-Src was amplified from rat liver cDNA (GenBank accession no. AF157016). The construct encoding temperature-sensitive VSV-G was a gift from Jennifer Lippincott-Schwartz (National Institutes of Health) and was reconstructed as a GFP fusion protein (8).

RNA oligonucleotides targeting Dyn2 were purchased from Dharmacon as described (38).

Fluorescence Microscopy.

Cells were fixed in formaldehyde and processed for immunofluorescence as described previously (39). Images were taken with a Hamamatsu Orca II camera (Hamamatsu Photonics) and analyzed with IPLab software (Scanalytics).

Correlative Light-Electron Microscopy.

Confocal images were obtained using a Zeiss LSM510 META confocal system (Carl Zeiss), were immediately prepared for electron microscopy, and were analyzed under a Philips Tecnai-12 electron microscope (FEI/Philips Electron Optics), as described previously (40).

VSV-G-GFP Trafficking.

Morphological and biochemical analyses of VSV-G-GFP trafficking were performed essentially as described previously (8, 24).