Kinase Signaling Initiates Coat Complex II (COPII) Recruitment and Export from the Mammalian Endoplasmic Reticulum*

The events regulating coat complex II (COPII) vesicle formation involved in the export of cargo from the endoplasmic reticulum (ER) are unknown. COPII recruitment to membranes is initiated by the activation of the small GTPase Sar1. We have utilized purified COPII components in both membrane recruitment and cargo export assays to analyze the possible role of kinase regulation in ER export. We now demonstrate that Sar1 recruitment to membranes requires ATP. We find that the serine/threonine kinase inhibitor H89 abolishes membrane recruitment of Sar1, thereby preventing COPII polymerization by interfering with the recruitment of the cytosolic Sec23/24 COPII coat complex. Inhibition of COPII recruitment prevents export of cargo from the ER. These results demonstrate that ER export and initiation of COPII vesicle formation in mammalian cells is under kinase regulation.

The events regulating coat complex II (COPII) vesicle formation involved in the export of cargo from the endoplasmic reticulum (ER) are unknown. COPII recruitment to membranes is initiated by the activation of the small GTPase Sar1. We have utilized purified COPII components in both membrane recruitment and cargo export assays to analyze the possible role of kinase regulation in ER export. We now demonstrate that Sar1 recruitment to membranes requires ATP. We find that the serine/threonine kinase inhibitor H89 abolishes membrane recruitment of Sar1, thereby preventing COPII polymerization by interfering with the recruitment of the cytosolic Sec23/24 COPII coat complex. Inhibition of COPII recruitment prevents export of cargo from the ER. These results demonstrate that ER export and initiation of COPII vesicle formation in mammalian cells is under kinase regulation.
Components comprising the cytosolic coat complex II (COPII) 1 are now recognized to be involved in cargo selection and export from the endoplasmic reticulum (ER) (1,2). Export from the ER is initiated by the activation of the small GTPase Sar1 through exchange of GDP for GTP by the membraneassociated Sec12 guanine nucleotide exchange factor (GEF). This activation step leads to the recruitment of the COPII subunits Sec23/24 from the cytosol to the membrane to form a tertiary complex that interacts with cargo and cargo receptors, initiating selection prior to export (3,4). Subsequent recruitment of the Sec13/31 complex allows the selected cargo to be exported from the ER by budding vesicles. Following COPII-mediated sorting from the ER, cargo-containing COPII vesicles are believed to fuse to form pre-Golgi intermediates containing tubular elements (5). Pre-Golgi intermediates are the first step in the exocytic pathway involved in the retrieval of recycling components to the ER using the coat complex I (COPI) components (6), thus separating forward moving cargo from the membrane-bound components of the COPII budding and fusion machinery (2,(7)(8)(9). The integration of these two sorting steps enables the forward moving cargo to be selectively delivered to the Golgi complex for transport to the cell surface (7).
The mechanisms by which different sorting steps in the early secretory pathway are regulated remain unknown. We (10,11) and others (12) have previously implicated kinases and phosphatases in regulating transport between the ER and Golgi compartments. More recently, a qualitative morphological study using indirect immunoflourescence suggested that H89, an isoquinolinesulfonamide that is frequently used as a selective serine/threonine and protein kinase A inhibitor, was involved in a late step of COPII vesicle coat assembly following Sar1 activation (13). We have now examined quantitatively the step in ER export directly regulated by H89 sensitive kinase(s) utilizing purified COPII components to follow the ordered recruitment of COPII to membranes, COPII-mediated vesicle formation, and cargo export. We show that Sar1 recruitment to ER membranes is sensitive to kinase activation, demonstrating that the first step in COPII vesicle formation, Sar1 activation, is under kinase regulation.
Coat Recruitment Assays-Sec23 recruitment (two-stage recruitment assay with salt wash) was performed as described previously (7). For Sar1 binding, microsome membranes, prepared as described previously (8) (20 -40 g), were incubated in the presence of wild type Sar1 (0.1 g) and Sec23/24 (0.7 g) or rat liver cytosol (200 g) in a final volume of 60 l in a reaction mix containing 36 mM Hepes pH 7.2, 70 mM KOAc, 2.5 mM MgOAc, 250 mM sorbitol, 1.8 mM CaCl 2ϩ , 5 mM EGTA, and 100 M GDP in the presence or absence of an ATP-regenerating system as described previously (11). Following incubation, the reactions were layered on a 15% sucrose cushion (180 l) containing 75 mM KOAc and 2 mM MgOAc. The reactions were centrifuged for 15 min at 16,000 ϫ g at 4°C. The membrane-containing pellet was solubilized with Laemmli SDS-sample buffer (15) and analyzed by SDS-polyacrylamide gel electrophoresis and quantitative immunoblotting (8). All experiments presented were performed at least twice with identical results.
In Vitro Vesicle Formation Assay-In vitro vesicle formation assay was performed with purified COPII components utilizing wild type Sar1 (2 g), Sec23/24 (1 g), and Sec13/31 (12 g) as described previously (3). All experiments presented were performed at least twice with identical results.

RESULTS AND DISCUSSION
The formation of COPII vesicles requires the sequential recruitment of the Sar1 GTPase and the Sec23/24 coat complex followed by the Sec13/31 complex (16). We have previously demonstrated biochemically that the recruitment of the Sec23/24 coat complex to ER membranes requires Sar1 activation and ATP (3,7,17). Consistent with these results, and as expected, ATP is also required for the recruitment of Sec13/31 (13). One possible explanation of the need for ATP was that ER membranes need to be primed with ATP prior to Sar1 activation and COPII recruitment. Alternatively, Sar1 activation may be initiated prior to the ATP-dependent step, which is then required for Sec23/24 recruitment and coat assembly. To test these two possibilities, ER membranes were incubated in a "stage 1" reaction in the presence of crude cytosol with or without ATP and/or the nonhydrolyzable form of GTP, GTP␥S, for 10 min at 32°C as described previously (7) (Fig. 1A). Following incubation, the membranes were either transferred to ice or collected by centrifugation, the latter being resuspended for further incubation in "stage 2" in the presence or absence of ATP or GTP␥S for an additional 10 min at 32°C prior to transfer to ice. At the end of the incubation, the membranes were then collected by centrifugation, solubilized, and analyzed for Sec23/24 recruitment using immunoblotting (7). As shown in Fig. 1A, preincubation of ER membranes in the presence of cytosol led to COPII recruitment (as analyzed by Sec23 binding); this occurred only in the presence of both ATP and GTP␥S, the latter being necessary to activate the endogenous cytosolic Sar1 (7). Moreover, a similar pattern was observed in the stage 2 incubation, as Sec23/24 binding to the membranes was not detected when membranes were primed with ATP alone nor when endogenous cytosolic Sar1 was first activated by the addition of GTP␥S in the stage 1 incubation. Therefore, Sar1 activation cannot be separated from the ATP-dependent step.
The above results suggested that ATP may be required for Sar1 recruitment and activation. To analyze this possibility directly, we measured the recruitment of both Sar1 and Sec23/24 to membranes by using purified Sar1 and Sec23/24 COPII subunits. To assure membrane-dependent Sar1 activation, we utilized recombinant wild type Sar1 and supplemented all incubations with GDP (100 M) in the presence or absence of GTP␥S (100 M). We have previously demonstrated that the mammalian Sar1 has a 10-fold higher affinity for GDP over GTP, assuring that only GEF-driven activation of Sar1 is measured in our assay (18). Membranes were incubated in the presence or absence of ATP, with or without GTP␥S, and in the presence of crude cytosol or wild type Sar1 and Sec23/24 subunits for 15 min at 32°C. At the end of the incubation, the reactions were layered on top of a 15% sucrose cushion and centrifuged to remove the unbound COPII components from the membrane-bound forms. The recruitment of endogenous cytosolic Sar1 was dependent on the addition of GTP␥S and ATP (Fig. 1B). Similarly, when purified COPII components were utilized, efficient membrane binding of Sar1 required both the addition of ATP and GTP␥S. Therefore, the Sar1 recruitment and activation step involving GTP, which initiates COPII assembly, requires ATP.
The dependence of Sar1 on ATP suggests the possible involvement of a protein kinase activity in the initiation of ER export. Two recent reports have suggested that the kinase inhibitor H89 could block general ER-to-Golgi traffic, whereas a variety of other kinase inhibitors had no detectable affect (12,13). To determine whether the H89-sensitive kinase affected the first step, Sar1 binding to the membranes, we incubated microsomal membranes in the presence of ATP, GTP␥S, purified Sar1, Sec23/24, and increasing concentrations of the kinase inhibitor H89. H89 abolished both Sar1 and the subsequent Sec23/24 interaction with the membranes at a K i of 25 M (Fig. 2A). The inhibition of Sar1 recruitment led to a parallel inhibition of Sec23/24 recruitment, albeit at a slightly higher H89 concentration (Fig. 2B). As Sec23/24 recruitment is absolutely dependent on Sar1 activation (3,7,8), the increased concentration of H89 required to block Sec23/24 recruitment may reflect the minimal levels of Sar1 activity sufficient to support maximal Sec23/24 recruitment. A similar inhibition of Sec23/24 and Sar1 recruitment was observed when whole cytosol was utilized (not shown).
We next analyzed the role of the H89-sensitive kinase in regulating cargo export from the ER. We utilized an in vitro budding assay containing ER microsomes incubated with the purified COPII coat components Sar1, Sec23/24, and Sec13/31 to focus directly on the ER export step involving the kinase (3,8), thereby eliminating the possibility that inhibition reflects COPI-dependent recycling steps from the Golgi. The assay follows the mobilization of a reporter cargo molecule, VSV-G, which is released from dense ER membranes into a more slowly sedimenting COPII-containing vesicle fraction, using differential centrifugation. Microsomes were incubated with the purified COPII components in the absence or presence of increasing concentrations of H89. As we have previously reported, the addition of purified COPII led to efficient mobilization of VSV-G to the vesicle fraction (ϳ15-30%) (Fig. 3A, lane 2, compare medium speed (M) with high speed (H)), similar to the amount of VSV-G that is mobilized to vesicles when microsomes are incubated with whole cytosol (not shown) (3). Incubation of microsomes with increasing concentrations of H89 led to the complete inhibition of cargo export from the ER at a concentration consistent with its ability to block Sar1 recruitment (Fig. 2).
Given that H89 is a specific inhibitor of PKA at nanomolar concentration, whereas the concentration of H89 required to inhibit COPII recruitment was in the M range, it is unlikely that protein kinase A was the target for H89 inhibition. Indeed, related inhibitors of PKA such as H8 or KT5720 do not inhibit ER-to-Golgi transport (13). Moreover, we have analyzed the effects of a specific protein kinase A peptide inhibitor, PKI (K i ,

FIG. 1. ATP is required for Sar1 recruitment to ER membranes.
A, membranes were incubated in the presence or absence of an ATP regenerating system (ATPr) and/or 100 M GTP␥S, as indicated, for 10 min at 32°C (Stage 1) as described under "Experimental Procedures." Following incubation, membranes were either transferred to ice or collected by a brief centrifugation and resuspended in the presence or absence of ATPr or GTP␥S, as indicated, for an additional 10 min at 32°C (Stage 2). At the end of the stage 2 incubation, membranes were collected and analyzed for Sec23/24 binding by immunoblotting as described under "Experimental Procedures." B, membranes were incubated either with rat liver cytosol (cyt) or purified Sar1 and Sec23/24 COPII coat components, in the presence of GDP, ATPr, or GTP␥S as indicated. Binding of Sar1 and Sec23/24 was determined as described under "Experimental Procedures." 36 nM). Concentrations of up to 250 nM PKI failed to affect Sar1 and Sec23/24 recruitment (Fig. 2C) or COPII vesicle formation (Fig. 3B, see below). These results are consistent with the inability of other PKA inhibitors to block ER-to-Golgi transport (12,13). H89 was also utilized previously as an inhibitor of serine/threonine kinases of the PKC family. However, related inhibitors such as H7 or chelerythrine do not affect ER-to-Golgi transport (13). Moreover, recent morphological observations raised the possibility that the isozyme of protein kinase C (PKD) involved Golgi disassembly may also be involved in the transport of VSV-G between the ER and the Golgi (12). Like the PKI peptide substrate, using a specific PKD peptide substrate to antagonize PKD function, we failed to interfere with COPII vesicle budding compared with that observed for the control, scrambled peptide (Fig. 3B, lanes 3-5). PKD peptides also failed to inhibit Sar1 and Sec23/24 recruitment to membranes (Fig. 2C). Thus, although the identity of the H89-sensitive kinase(s) involved in ER export remains unknown, it does not appear to be either PKA or the isoform of PKD that mediates Golgi disassembly (12) nor a host of other known kinases (13). It is also unlikely to involve phosphatidylinositol 3 (PI3)-lipid kinases as we have previously shown that wortmannin, a PI3specific inhibitor, has no effect on the export of either soluble or transmembrane cargo from the ER (19). Other soluble lipid kinases, including PI4-kinases, are unlikely to be involved given that we were unable to prime membranes by preincubating them with ATP. Furthermore, the addition of purified COPII components, in the absence of cytosol, supports efficient export. However, further experiments are necessary to address the potential role of lipid kinases, as a PI4-kinase of unknown function has been demonstrated to be enriched on ER membranes (20).
Our results demonstrate that an H89-sensitive kinase initiates Sar1 recruitment, the first step in COPII vesicle formation. Our conclusions differ significantly from a recent report in which H89 was proposed to affect a late, but not early, step in COPII vesicle formation (13); that report based its conclusion on experimental results in which H89 was observed to inhibit GTP␥S stabilization of the Sec13/31 complex to membranes. It was assumed that the inclusion of GTP␥S in the assay was sufficient for passive Sar1 activation. In contrast, we now demonstrate that the recruitment and activation of Sar1 by GTP␥S is dependent on ATP and inhibited by H89. These results are consistent with the requirement for physiological temperatures for recruitment of COPII to membranes in vitro and with the relatively low affinity of Sar1 for guanine triphosphate nucleotides in the absence of a Sar1 specific exchange activity (3,7,8,17). Thus, Sar1 binding of GTP␥S is not passive and requires the Sec12 GEF. Consistent with this observation, in the present experiments we have observed that in the absence of Sar1 activation by Sec12, the recruitment of the Sec23/24 complex, and as a consequence, the subsequent recruitment of the Sec13/31 complex, cannot occur. Indeed, the inability morphologically to detect ER export sites using indirect immunofluorescence based on the detection of recruited Sec13/31 complex (13), is entirely consistent with our results that demonstrate that the primary step affected by the H89-sensitive kinase is the first step, Sar1 recruitment and activation.
Control of small GTPase activation by kinases is not unprec- edented. Activation of Ras through the recruitment of the Grb2⅐SOS GEF complex is initiated by protein tyrosine kinases (21). By analogy to Ras, recruitment and activation of Sar1 through Sec12, a transmembrane GEF, may be regulated by a membrane-associated kinase. This conclusion is consistent with our observations that recruitment of Sar1 is the limiting component in our budding reaction (3). In preliminary in vitro experiments we have failed to detect direct phosphorylation of either Sar1 or Sec12 during the formation of COPII vesicles. However, further studies are required to establish or preclude a role for Sar1 or Sec12 phosphorylation during these potentially transient events involved in coat assembly. Alternatively, our previous studies have suggested a role for cargo in modulating the COPII export machinery (17). Although it remains to be seen whether cargo availability per se can regulate kinase and Sar1 activation, ER chaperones, which interact with cargo during the folding cascade, may mediate these events, thereby linking cargo availability to ER export. This interpretation is consistent with the role of the ER resident chaperone Ig heavy chain-binding protein (BIP) in activation of ER stress receptor kinases in response to misfolded cargo in the ER (22), an event that leads to a global up-regulation of the exocytic pathway (23). Additional targets for kinase activity may therefore include proteins of the ER folding and selection machinery that can interact with cargo prior to its association with the recruited COPII coat (24 -26).
Placing ER budding under kinase control would allow the export machinery to respond to extracellular signaling pathways, thus integrating the secretory pathway with cellular physiology. We have previously demonstrated that activation of IgE receptors in mast cell lines enhances ER export (27) and that ER export is regulated by a calphostin C-sensitive, diacylglycerol-binding protein (28). Moreover, it has been shown that modulation of kinase activity plays a major role during mitosis, leading to cessation of ER export and dispersal of ER export sites (29). We propose that the ability of the ER export machinery to integrate with kinase signaling cascades may represent an important first step in the general function of cargo selection by COPII machinery in the exocytic pathway in mammalian cells (26).