The Src-family tyrosine kinase Lyn has a role in signal transduction at the cytoplasmic face of the plasma membrane upon extracellular ligand stimulation. After synthesis in the cytoplasm, Lyn accumulates on the Golgi and is subsequently transported to the plasma membrane. However, the mechanism of Lyn trafficking remains elusive. We show here that the C-lobe of the Lyn kinase domain is associated with long-chain acyl-CoA synthetase 3 (ACSL3) on the Golgi in a manner that is dependent on Lyn conformation but is independent of its kinase activity. Formation of a closed conformation by CSK prevents Lyn from associating with ACSL3, resulting in blockade of Lyn export from the Golgi. Overexpression and knockdown of ACSL3 accelerates and blocks Golgi export of Lyn, respectively. The post-Golgi route of Lyn, triggered by ACSL3, is distinct from that of vesicular stomatitis virus glycoprotein (VSV-G) and of caveolin. Moreover, an ACSL3 mutant lacking the LR2 domain, which is required for the catalytic activity, retains the ability to associate with Lyn and accelerate Golgi export of Lyn. These results suggest that initiation of Golgi export of Lyn involves association of ACSL3 with the Lyn C-lobe, which is exposed to the molecular surface in an open conformation.
Src-family tyrosine kinases, which are non-receptor-type tyrosine kinases, consist of proto-oncogene products and structurally related proteins and include at least eight highly homologous proteins: Src, Lyn, Yes, Fyn, Fgr, Hck, Lck and Blk (Brown and Cooper, 1996; Thomas and Brugge, 1997). Src-family tyrosine kinases have crucial roles in the regulation of cell proliferation, differentiation, migration and cell-shape changes (Thomas and Brugge, 1997). Src, Lyn, Yes and Fyn are widely expressed in a variety of cell types, whereas Blk, Hck, Fgr and Lck are found primarily in hematopoietic cells (Bolen and Brugge, 1997; Thomas and Brugge, 1997).
Src-family tyrosine kinases are composed of: (1) an N-terminal Src homology (SH) 4 domain, (2) a poorly conserved ‘unique’ domain, (3) an SH3 domain, which can bind to specific proline-rich sequences, (4) an SH2 domain, which can bind to specific sites of tyrosine phosphorylation, (5) an SH1 tyrosine kinase catalytic domain and (6) a C-terminal negative regulatory tail for autoinhibition of the kinase activity (Brown and Cooper, 1996). The tyrosine kinase activity of Src-family kinases is repressed through CSK-induced ‘closed conformation’ as a result of the intramolecular binding of the SH2 domain to the tyrosine-phosphorylated tail catalyzed by CSK-family kinases and the SH3 domain to the SH2-kinase linker (Brown and Cooper, 1996; Sicheri et al., 1997; Xu et al., 1997). Src-family tyrosine kinases can interact with a large number of upstream regulators and downstream substrates via SH2- and SH3-domain-mediated protein-protein interactions (Pawson, 1995; Thomas and Brugge, 1997; Blume-Jensen and Hunter, 2001).
Src-family tyrosine kinases, which are classified as cytosolic enzymes, localize at the cytoplasmic face of the plasma membrane through post-translational lipid modification (Resh, 1994), but an appreciable fraction is found at intracellular compartments, such as late endosomes, lysosomes, the Golgi, secretory granules and the nucleus (Kaplan et al., 1992; Ley et al., 1994; Möhn et al., 1995; Bijlmakers et al., 1997; Yamaguchi et al., 2001; Kasahara et al., 2004; Kasahara et al., 2007a; Chu et al., 2007; Grimmler et al., 2007; Ikeda et al., 2008; Sato et al., 2009; Takahashi et al., 2009). Distinctive localizations of individual members of Src-family kinases have been implicated in their specific functions.
Lyn, a member of Src-family tyrosine kinases, is widely expressed in a variety of organs, tissues and cell types, including epithelial, hematopoietic and neuronal cells, and has an important role in signal transduction at the cytoplasmic face of the plasma membrane (Tsukita et al., 1991; Hibbs et al., 1995; Nishizumi et al., 1995; Hirao et al., 1998; Hayashi et al., 1999; Sheets et al., 1999; Tada et al., 1999; Rivera and Olivera, 2007; Kuga et al., 2007; Kuga et al., 2008; Sohn et al., 2008). We recently showed that newly synthesized Lyn in the cytoplasm accumulates on Golgi membranes and is subsequently transported to the plasma membrane along the secretory pathway in a manner that is dependent on the Lyn kinase domain (Kasahara et al., 2004; Sato et al., 2009). The Lyn kinase domain is also required for targeting of Lyn to caveolin-positive Golgi membranes (Ikeda et al., 2009), which suggests that this domain has a crucial role in Lyn trafficking in addition to its role in catalysis of substrate tyrosine phosphorylation. We further showed that Lyn tyrosine-phosphorylates annexin II on Golgi membranes, leading to the translocation of annexin II from the Golgi complex to the endoplasmic reticulum (ER) (Matsuda et al., 2006), suggesting that Lyn functions not only at the plasma membrane but also at the Golgi complex. Although the localization of Lyn is crucial for its functions, the mechanism of the intracellular trafficking of Lyn remains elusive.
In this study, we investigate the mechanism of the trafficking of Lyn from the Golgi to the plasma membrane. We show that the C-lobe of the Lyn kinase domain associates with ACSL3 when Lyn has an open conformation, thereby initiating the export of Lyn from the Golgi. Our results provide the first demonstration of a direct role of the Lyn kinase C-lobe in Golgi export of Lyn.
Association of the Lyn kinase C-lobe with ACSL3
We previously showed that the trafficking of newly synthesized Lyn from the Golgi to the plasma membrane involves the four charged amino acid residues (Asp346, Glu353, Asp498 and Asp499) in the C-lobe of the Lyn kinase domain (Kasahara et al., 2004). We thus hypothesized that the Lyn C-lobe has a role in Lyn trafficking through a protein-protein interaction. To search for a protein that associates with the Lyn C-lobe, we constructed a glutathione-S-transferase (GST)-Lyn C-lobe fusion protein (GST-C-lobe), which encompasses amino acids 326-506 in Lyn (Fig. 1A). We found that a 70 kDa protein (p70) was pulled down with GST-C-lobe but not with GST alone from Triton-X-100-treated HeLa cell lysates (Fig. 1B). We then constructed GST-C-lobe-mt, an alanine-substitution mutant (Fig. 1A). p70 could be pulled down with GST-C-lobe-mt but to a lesser extent than seen with GST-C-lobe (Fig. 1B), suggesting that p70 preferentially associates with the intact C-lobe. Consequently, the p70 band was excised and subjected to proteolytic cleavage followed by MALDI-TOF-MS. The peptide derived from p70 (VLSEAAISASLEK) exhibited 100% identity with the amino acid sequence of human long-chain acyl-CoA synthetase-3 (ACSL3; EC 184.108.40.206).
To verify the association of the Lyn C-lobe with ACSL3 in vitro, we expressed N-terminally Myc-tagged ACSL3 (Myc-ACSL3; Fig. 4A) in COS-1 cells. Upon 5-minute incubation with Myc-ACSL3-containing Triton X-100 cell lysates at 4°C Myc-ACSL3 was pulled down preferentially with GST-C-lobe rather than GST-C-lobe-mt, although upon incubation at 4°C for 2 hours, Myc-ACSL3 could be pulled down with GST-C-lobe-mt to a similar extent as that with GST-C-lobe (Fig. 1C). Next, we tested whether Lyn was associated with ACSL3 in vivo. When Lyn-HA and Myc-ACSL3 were coexpressed in COS-1 cells, Lyn-HA was coimmunoprecipitated with Myc-ACSL3 and vice versa (Fig. 1D). Similarly to Lyn-HA, kinase-inactive Lyn(K275A)-HA was coimmunoprecipitated with Myc-ACSL3 (Fig. 1E). These results suggest that Lyn is associated with ACSL3 in a manner that is independent of the Lyn kinase activity.
We hypothesized that the charged residues in the C-lobe are exposed to the molecular surface of Lyn when it is in an open conformation (Kasahara et al., 2004). Since a closed, repressed conformation of Src-family kinases is created by CSK-mediated phosphorylation of their C-terminal tyrosine residue, we examined whether overexpression of CSK could inhibit the association of Lyn with ACSL3. COS-1 cells were triply cotransfected with wild-type Lyn (Lyn-wt), Myc-ACSL3 and CSK. Western blot analysis confirmed that the kinase activity of Lyn-wt was decreased because of the induction of a closed conformation of Lyn-wt by overexpression of CSK. Intriguingly, the association of Lyn-wt with Myc-ACSL3 was affected by CSK overexpression (Fig. 1F), suggesting that a closed conformation of Lyn inhibits its association with ACSL3.
Using anti-ACSL3 antibody that recognises both Myc-ACSL3 and endogenous ACSL3 in COS-1 and HeLa cells (Fig. 1G; supplementary material Fig. S1), we showed that two alternative spliced isoforms of endogenous Lyn were associated with endogenous ACSL3 (Fig. 1H). Although a fraction of ACSL3 appeared in the immunoprecipitate with control antibody, actin did not appear in the immunoprecipitates, suggesting that ACSL3 might nonspecifically interact with immunoglobulin and/or protein-G-bound beads. Furthermore, ACSL3 was seen with the Golgi caveolin and β-1,4-galactosyltransferase (GalT) where Lyn accumulated, despite colocalization of Myc-ACSL3 largely with ER-resident calnexin (supplementary material Fig. S2). We then transfected COS-1 cells with Lyn-wt to examine the localization of Lyn and ACSL3, because endogenous Lyn could not be visualized owing to its low expression in COS-1 cells. At 18 hours after transfection, Lyn-wt was localized to the Golgi and the plasma membrane. Endogenous ACSL3 was seen mainly at the perinuclear areas but not at the plasma membrane, consistent with previous studies (Fujimoto et al., 2004), and Lyn-wt was colocalized with endogenous ACSL3 at the Golgi, but not the plasma membrane (Fig. 1I, upper panels). In addition, Lyn-wt was colocalized with Myc-ACSL3 at the Golgi but not the plasma membrane in COS-1 cells cotransfected with Lyn-wt and Myc-ACSL3 (Fig. 1I, lower panels). Localization of Myc-ACSL3 was comparable with that of endogenous ACSL3 (Fig. 1G). These results suggest that Lyn is associated with ACSL3 at the Golgi.
Requirement of Lyn open conformation for its association with ACSL3
To examine the role of ACSL3 in Lyn localization, we overexpressed ACSL3 in COS-1 cells and found that coexpression of Myc-ACSL3 permitted Lyn-HA and Lyn-wt to localize to the plasma membrane at 12 hours and 18 hours after transfection, respectively (Fig. 2A,B). The trafficking of Lyn from the Golgi to the plasma membrane was accelerated by the lack of the C-terminal negative-regulatory tail in Lyn (compare Fig. 2A with 2B). Similar results were obtained in HeLa cells (supplementary material Fig. S3). By contrast, overexpression of ACSL3 did not affect the localization of Src, the Golgi pool of caveolin or Golgi-resident GM130 and GalT (supplementary material Fig. S4), suggesting that overexpression of ACSL3 specifically accelerates the trafficking of Lyn from the Golgi. Note that Myc-ACSL3 consistently remained in the perinuclear areas (insets in Fig. 2A,B; Fig. 3B). In human monocytic THP-1 cells, endogenous Lyn can be visualized at the Golgi and the plasma membrane (Ikeda et al., 2008; Sato et al., 2009). We found that expression of Myc-ACSL3 in THP-1 cells significantly decreased the levels of endogenous Lyn at the Golgi and increased its levels at the plasma membrane (Fig. 2C). Moreover, excessive overexpression of CSK greatly decreased the levels of the association of Lyn-wt with endogenous ACSL3 in COS-1 cells, leading to blockade of Golgi export of Lyn-wt even at 24 hours after transfection, whereas excessive overexpression of CSK did not inhibit Lyn-HA association with ACSL3 nor Golgi export of Lyn-HA (Fig. 2D-F). To further examine the relationship between the localization of Lyn and its association with ACSL3, we compared the association of Lyn with ACSL3 between 12 hours and 24 hours after transfection. The amount of ACSL3 that associated with Lyn-wt 24 hours after transfection was drastically decreased compared with the amount associated with CSK-treated Lyn-wt (Fig. 2G), suggesting that Lyn present at the plasma membrane was not colocalized with ACSL3 (see also Fig. 1I; Fig. 2A-C, Fig. 3B, Fig. 4C, Fig. 7C; supplementary material Fig. S3). These results suggest that Golgi association of Lyn in its open conformation with ACSL3 initiates and accelerates Golgi export of Lyn toward the plasma membrane.
Role of Lyn C-lobe charged residues in ACSL3 association in vivo
Next, we compared the effect of ACSL3 overexpression on the trafficking between Lyn-HA and Tetra-mt-HA, which lacks the four charged residues in the C-lobe (Fig. 1A). The levels of Tetra-mt-HA coimmunoprecipitated with Myc-ACSL3 were lower than those of Lyn-HA coimmunoprecipitated with Myc-ACSL3 (Fig. 3A). Tetra-mt-HA was predominantly seen at the Golgi in COS-1 cells transfected with Tetra-mt-HA alone at 11 hours and 12 hours after transfection (Fig. 3B, upper panels), consistent with our previous study (Kasahara et al., 2004). We further showed that in COS-1 cells coexpressing Tetra-mt-HA plus Myc-ACSL3 or Lyn-HA plus Myc-ACSL3, expression of Myc-ACSL3 did not accelerate Golgi export of Tetra-mt-HA but did increase export of Lyn-HA (Fig. 3B-D). These results suggest that the lack of the charged residues in the C-lobe leads to Golgi accumulation of Lyn through an insufficient association of Lyn with ACSL3.
Role of the ACSL3-LR1 domain in Golgi export of Lyn
ACSL3, a membrane-bound protein with a C-terminal large cytoplasmic portion, has a short N-terminal portion, a transmembrane domain (TM), a luciferase-like region (LR) 1 domain, which contains the phosphate-binding P-loop and the fatty acid Gate-domain, and an LR2 domain, which contains the adenine-binding motif and the long chain acyl-CoA synthetase signature motif (Fujino et al., 1997; Black et al., 1997; Soupene and Kuypers, 2008). To examine a region in ACSL3 responsible for Lyn trafficking, we constructed two Myc-ACSL3 mutants lacking the LR2 (Myc-ΔLR2) or LR1 domain (Myc-ΔLR1) (Fig. 4A). When COS-1 cells were cotransfected with Lyn-HA and each Myc-ACSL3 mutant, we showed that Lyn-HA was indeed coimmunoprecipitated with Myc-ΔLR2 but not Myc-ΔLR1 (Fig. 4B). Next, we cotransfected COS-1 cells with Lyn-HA and each Myc-ACSL3 mutant and examined their localizations 12 hours after transfection. Intriguingly, Myc-ΔLR1 did not accelerate Golgi export of Lyn-HA, whereas Myc-ΔLR2 significantly accelerated Golgi export of Lyn-HA (Fig. 4C). However, it is of interest to note that the localizations of Myc-ΔLR2 and Myc-ΔLR1 were similar to that of Myc-ACSL3 (Fig. 4C). Given that the LR1 and LR2 domains are both required for the acyl-CoA synthetase activity (Iijima et al., 1996), these results suggest that the ACSL3 LR1 domain has a crucial role in Golgi export of Lyn through its ACSL3 association in a manner that is independent of the acyl-CoA synthetase activity.
Requirement of ACSL3 for Golgi export of Lyn
To examine whether ACSL3 was required for Golgi export of Lyn, we knocked down endogenous ACSL3 expression using short hairpin RNAs (shRNAs). To visualize shRNA-transfected cells, the mCherry expression cassette was introduced into the shRNA expression vector. Western blotting analysis showed that the protein levels of endogenous ACSL3 in HeLa cells were reduced by 56-94% at 20 hours after transfection with ACSL3-A1, ACSL3-A2 or ACSL3-A3 shRNA (Fig. 5A). When Lyn-HA was seen predominantly at the plasma membrane at 20 hours after transfection, we showed that coexpression of Lyn-HA with ACSL3-A1, ACSL3-A2 or ACSL3-A3 shRNA significantly increased the levels of Golgi localization of Lyn-HA (Fig. 5B,C). However, knockdown of ACSL3 did not affect Golgi localization of GalT and GM130 (supplementary material Fig. S5A,B). Then, to examine the effect of ACSL3 knockdown on the trafficking of endogenous Lyn, we used human megakaryocytic Dami cells in which endogenous Lyn was detectable by immunostaining (Sato et al., 2009) and found that transfection of Dami cells with ACSL3-A2 or ACSL3-A3 shRNA resulted in a ~45% reduction of endogenous ACSL3 levels (Fig. 5D). In control Dami cells, endogenous Lyn was mainly found at the plasma membrane, and a very small fraction of Lyn was seen at the Golgi (Fig. 5E, upper panels), which is consistent with our recent results (Sato et al., 2009). Intriguingly, transfection with ACSL3-A2 or ACSL3-A3 shRNA increased the levels of Golgi accumulation of endogenous Lyn, whereas transfection with control shRNA did not affect the localization of endogenous Lyn (Fig. 5E). In addition, transfection with ACSL3-A3 shRNA did not inhibit the cell-surface localization of the transmembrane protein CD43 (supplementary material Fig. S5C). These results suggest that ACSL3 is crucial for export of Lyn from the Golgi toward the plasma membrane.
Comparison of the localization of Src, Lyn and Yes upon knockdown of ACSL3
Similarly to Lyn, Yes, which is another member of the Src-family kinases, is also transported to the plasma membrane through the Golgi (Sato et al., 2009). However, Src is rapidly exchanged between the plasma membrane and late endosomes (Kasahara et al., 2007a; Kasahara et al., 2008). We thus examined the localization of Src, Lyn and Yes upon knockdown of ACSL3 in HeLa cells. We found that knockdown of ACSL3 blocked Golgi export of Yes as well as Lyn, but the knockdown did not affect the localization of Src (Fig. 6A), supported by ACSL3 association with Lyn and Yes but not Src (Fig. 6B). Note that ACSL3 knockdown inhibited plasma-membrane localization of Lyn and Yes but not Src. These results suggest that ACSL3 association is required for the trafficking of the Src-family members, such as Lyn and Yes, which are biosynthetically transported to the plasma membrane through the Golgi.
Golgi export of Lyn and VSV-G by different post-Golgi carriers
To examine whether ACSL3 affected the conventional export route from the Golgi to the plasma membrane, we tested the effect of ACSL3 knockdown on Golgi export of vesicular stomatitis virus glycoprotein (VSV-G) in COS-1 cells. The levels of endogenous ACSL3 in COS-1 cells were reduced by 45-47% upon transfection with ACSL3-A3 shRNA, resulting in Golgi accumulation of Lyn-HA (supplementary material Fig. S6). Cells cotransfected with green fluorescent protein (GFP)-tagged VSV-G (VSV-G-GFP) plus mCherry (control) or ACSL3-A3 shRNA-mCherry were incubated at 40°C to accumulate VSV-G-GFP in the ER and then shifted to 32°C for 30 minutes or 6 hours to transport VSV-G-GFP through the Golgi to the plasma membrane. Knockdown of ACSL3 did not affect Golgi export of VSV-G-GFP (Fig. 7A), suggesting that Golgi export of VSV-G-GFP is independent of ACSL3.
To further examine whether Lyn and VSV-G could be sorted into different post-Golgi transport carriers (PGCs), COS-1 cells cotransfected with Lyn-wt and VSV-G-GFP were treated with tannic acid, which inhibits fusion of PGCs with the plasma membrane to accumulate PGCs (Polishchuk et al., 2004). Tannic acid treatment enabled us to visualize PGCs containing Lyn-wt at the cell periphery (Fig. 7B). Intriguingly, most PGCs containing Lyn-wt were different to those carrying VSV-G-GFP and were also different to caveolin-positive PGCs. Consistent with our observations that Lyn is colocalized with caveolin at the Golgi but not at the cell periphery (Kasahara et al., 2004; Ikeda et al., 2009), these results suggest that Lyn is exported from the Golgi by unconventional PGCs.
Next, to examine how ACSL3 is involved in Golgi export of Lyn, COS-1 cells cotransfected with Lyn-wt, VSV-G-GFP and Myc-ACSL3 were treated with tannic acid. We found that ACSL3 overexpression significantly increased the number of PGCs containing Lyn-wt but not that of PGCs carrying VSV-G-GFP (Fig. 7C). However, tannic acid treatment did not affect the localization of Myc-ACSL3, and at the cell periphery Myc-ACSL3 did not colocalize with PGCs containing Lyn-wt. These results suggest that ACSL3 has a key role in triggering Golgi export of Lyn on Golgi membranes through an increase in the number of a specific subset of PGCs that contain Lyn.
In the present study, we provide evidence that a novel protein-protein interaction mediated by the Lyn kinase domain has a crucial role in the trafficking of newly synthesized Lyn. First, the Lyn kinase C-lobe is associated with ACSL3 and the association is inhibited by the lack of the four charged amino acid residues in the Lyn kinase C-lobe. Second, the association of Lyn with ACSL3 is inhibited by CSK-induced closed conformation, where the four charged amino acid residues are masked. Third, overexpression of CSK induces Golgi accumulation of Lyn, whereas overexpression of ACSL3 accelerates Golgi export of Lyn toward the plasma membrane. Fourth, the LR1 domain of ACSL3 is sufficient for association with Lyn and acceleration of Golgi export of Lyn. Fifth, Golgi export of Lyn but not VSV-G is blocked by ACSL3 knockdown and mediated by PGCs distinct from those carrying VSV-G. We show for the first time the significance of the Lyn kinase domain in an open conformation for Golgi export of newly synthesized Lyn through association with ACSL3.
Thus, we present a model for Golgi export of Lyn initiated by the association of the ACSL3 LR1 domain with the Lyn C-lobe in when Lyn is in an open conformation (Fig. 8). Newly synthesized Lyn in the cytoplasm is accumulated on Golgi membranes. In a closed conformation, mediated by CSK-catalyzed phosphorylation of the C-terminal tyrosine residue, Lyn is unable to associate with ACSL3 and remains on Golgi membranes. Once Lyn forms an open conformation by dephosphorylation of the tyrosine-phosphorylated tail, Lyn is associated with ACSL3 via the kinase C-lobe and is then released from Golgi membranes toward the plasma membrane while undergoing dissociation from ACSL3.
Src-family tyrosine kinases have modular domains for protein-protein interactions, such as the SH3 and SH2 domains (Pawson, 1995). The SH3 and SH2 domains of Src-family kinases have an important role in their localization, such as focal adhesions, lysosomes, late endosomes and the Golgi (Kaplan et al., 1994; Kasahara et al., 2007a; Kasahara et al., 2008; Li et al., 2008; Ikeda et al., 2009). The kinase domain of Src-family tyrosine kinases is believed to solely exert the catalytic activity. Nonetheless, we have shown that the Lyn kinase C-lobe, but not kinase activity, is required for the proper trafficking of Lyn from the Golgi (Kasahara et al., 2004) and the targeting of Lyn to the Golgi pool of caveolin (Ikeda et al., 2009). In fact, our results shown in Figs 1, 2 and 3 substantiate a protein-protein interaction through the C-lobe of the Lyn kinase domain by identifying a binding partner of the C-lobe. This unusual kinase-domain-mediated protein-protein interaction is emphasized by the findings that the association of Lyn with ACSL3 is inhibited by CSK-mediated phosphorylation of the C-terminal tyrosine residue of Lyn (Fig. 1F; Fig. 2D,G) and the kinase activity of Lyn is dispensable for the association (Fig. 1E). Given that four negative-charged amino acid residues on the αE and αI helices in the C-lobe intramolecularly interact with three positively charged amino acid residues on the αA helix in the SH2 domain in a closed conformation (Sicheri et al., 1997; Xu et al., 1997; Kasahara et al., 2004), an open conformation of Lyn unmasks the αE and αI helices in the C-lobe for an association with ACSL3.
The mammalian ACSL family has five members with different localizations: ACSL1, in the ER and the cytosol; ACSL3, in the ER and the Golgi (this study); ACSL4 and ACSL5, in mitochondria; and ACSL6, in the plasma membrane (Lewin et al., 2001; Soupene and Kuypers, 2008). Long chain acyl-CoA, which is generated by the catalytic activity of ACSL, is required for budding of transport vesicles from Golgi membranes (Glick and Rothman, 1987; Pfanner et al., 1989; Barr and Shorter, 2000). The ACSL3 activity is also required for very low-density-lipoprotein-mediated secretion of hepatitis C virus particles in hepatoma cells (Yao and Ye, 2008). However, given that the ACSL-LR1 domain per se does not exert the catalytic activity (Iijima et al., 1996; Black et al., 1997; Soupene and Kuypers, 2008), ACSL3 is likely to act as a component of a hypothetical initiator complex for Golgi export of Lyn and might function as an adaptor protein on Golgi membranes.
Lck, another member of the Src family, travels from the perinuclear region to the plasma membrane all the way with specific transport carriers containing MAL in T cells (Antón et al., 2008). In contrast to MAL, ACSL3 has a unique character to physically associate with Lyn on the Golgi and remain on the Golgi without formation of ACSL3-positive PGCs (Figs 1,2,7; supplementary material Fig. S2). These results suggest that unlike MAL in Lck trafficking, ACSL3 association with newly synthesized Lyn takes place on the Golgi, and ACSL3 is released from Lyn when Lyn leaves the Golgi. Presumably, ACSL3 can trigger and stimulate Golgi export of Lyn but not deliver Lyn from the Golgi to the plasma membrane.
It is of interest to note that formation of PGCs containing Lyn but not those carrying VSV-G is stimulated by ACSL3 (Fig. 7C), and that Lyn and caveolin are sorted into different PGCs despite their colocalization at the Golgi (Fig. 7B). Given that VSV-G and caveolin are sorted into separate PGCs and destined for different plasma membrane domains (Presley et al., 1997; Tagawa et al., 2005; Hayer et al., 2010), the targeting of Lyn toward a specific domain(s) of the plasma membrane could be directed by ACSL3-dependent transport.
Recently, we showed that Lyn and Yes, which are monopalmitoylated Src-family kinases, are transported to the plasma membrane through the Golgi (Sato et al., 2009), whereas Src, which is nonpalmitoylated Src-family kinase, is rapidly exchanged between the plasma membrane and late endosomes or lysosomes (Kasahara et al., 2007a). Intriguingly, Yes but not Src, is indeed associated with ACSL3, and Golgi export of Yes is blocked by ACSL3 knockdown, but the localization of Src is not affected (Fig. 6), leading to the hypothesis that ACSL3 association is required for the trafficking of the monopalmitoylated Src-family member that are biosynthetically transported to the plasma membrane via the Golgi. Given that Lyn and Yes have a crucial role in signal transduction mostly at the cytoplasmic face of the plasma membrane upon extracellular ligand stimulation (Thomas and Brugge, 1997), ACSL3 is indispensable for plasma membrane localization of Lyn and Yes. Lyn is known to participate in the B-cell-antigen receptor (BCR) and FcεRI signaling at the plasma membrane in B-cells and mast cells, respectively (Nishizumi et al., 1995; Bolen and Brugge, 1997; Thomas and Brugge, 1997; Sheets et al., 1999; Rivera and Olivera, 2007; Sohn et al., 2008). Presumably, the association of Lyn with ACSL3 has a crucial role in proper Lyn localization to the plasma membrane, which leads to the appropriate BCR and FcεRI signaling. Moreover, it is shown that Golgi membranes can serve as a platform of Src-family kinases for signal transduction: Golgi-localized Lyn tyrosine-phosphorylates annexin II on Golgi membranes under oxidative stress (Matsuda et al., 2006) and Golgi-localized Lck is needed for the activation of Raf-1 under weak stimulation through the T-cell receptor (Li et al., 2008). Thus, ACSL3 might be involved in regulation of the balance of localization of Lyn and Yes between the plasma membrane and Golgi membranes.
In conclusion, we show that Golgi export of Lyn requires the association of the Lyn kinase C-lobe with ACSL3 when Lyn is in its open conformation. Despite the high homology of Src-family kinases, the localization of Lyn and Yes but not Src is affected by the lack of ACSL3, suggesting the importance of spatio-temporal localization of the Src-family kinases. It would be interesting to look for a possible protein complex, including Lyn and ACSL3, on Golgi membranes to determine which factors are responsible for regulating initiation of Golgi export of Lyn and to determine whether ACSL3 is involved in the trafficking of other proteins from the Golgi to the plasma membrane.
Materials and Methods
cDNAs encoding human wild-type Lyn (Lyn-wt) (1-512; with 1 designating the initiator methionine) and human wild-type Yes (1-543) were provided by Tadashi Yamamoto (The University of Tokyo, Tokyo, Japan) (Yamanashi et al., 1987; Sukegawa et al., 1987). cDNA encoding human wild-type Src (1-536) (Biorge et al., 1995) (provided by Donald J. Fujita, University of Calgary, Calgary, Canada) was previously described (Kasahara et al., 2007b). Three HA-tagged Lyn constructs lacking the C-terminal negative-regulatory tail, Lyn-HA (1-506; kinase-active), Lyn(K275A)-HA(1-506; kinase-dead) and Tetra-mt-HA (1-506; Asp346→Ala, Glu353→Ala, Asp498→Ala, Asp499→Ala), were described previously (Kasahara et al., 2004). For protein expression, all constructs were subcloned into the pcDNA4/TO vector (Invitrogen). Human ACSL3 cDNA subcloned into the pBluescript vector (Stratagene) was provided by Takahiro Fujino (Ehime University, Matsuyama, Japan) (Minekura et al., 1997). To construct N-terminally Myc-tagged ACSL3 (Myc-ACSL3), the ScaI-NotI fragment containing full-length ACSL3 was ligated into the SacII (blunted) and NotI sites of pcDNA3-Myc-cyclinA-GFP (Clute and Pines, 1999; provided by Jonathon Pines, Gurdon Institute, Cambridge, UK). The resulting Myc-ACSL3 cDNA was subcloned into pcDNA4/TO, and removal of the Zeocin-resistant gene generated pcDNA4/TO-Myc-ACSL3(ΔZeor). The spacer sequence MVGNSAPGPATREAGSALLALQQTALQEDQENINPEKAAPVQQPL was inserted between the Myc tag and ACSL3. Myc-ACSL3ΔLR2 (1-444) was generated from pcDNA4/TO-Myc-ACSL3(ΔZeor) by SspI and NotI digestion and blunting. To construct Myc-ACSL3ΔLR1 (1-94 and 438-720), the NdeI (blunted) and BsstBI (blunted) fragment of pcDNA4/TO-Myc-ACSL3(ΔZeor) was replaced with the SalI (blunted)-XhoI (blunted) fragment of pBluescript as a spacer encompassing three amino acid residues (RPR). Rat Csk cDNA subcloned into the pME18S vector was provided by Masato Okada and Shigeyuki Nada (Osaka University, Osaka, Japan) (Nada et al., 1991). The plasmid for vesicular stomatitis virus glycoprotein fused to green fluorescent protein (VSV-G-GFP) was provided by Jennifer Lippincott-Schwartz (National Institutes of Health, Bethesda, MD) (Presley et al., 1997) through Mitsuo Tagaya (Tokyo University of Pharmacy and Life Sciences, Tokyo, Japan) (Hirose et al., 2004). A glutathione-S-transferase-fusion protein containing the Lyn C-lobe (326-506) (GST-C-lobe) was constructed by subcloning the Lyn C-lobe between the BamHI (blunted) and SmaI sites of the pGEX-5X2 vector (Amersham Biosciences). GST-C-lobe-mt (326-506; Asp346→Ala, Glu353→Ala, Asp498→Ala, Asp499→Ala) was constructed by replacing the C-lobe with the C-lobe-mt of Tetra-mt-HA.
The following antibodies were used: HA epitope (F-7 and Y-11; Santa Cruz Biotechnology), Myc epitope (PL14; Medical & Biological Laboratory, Nagoya, and 9E10 and A-14; Santa Cruz Biotechnology), Lyn (Lyn44 and H-6; Santa Cruz Biotechnology, and Lyn9; Wako Pure Chemicals, Osaka, Japan), CSK (#52; BD Transduction Laboratories), actin (MAB1501; Chemicon International), Src (#327; Oncogene Research, GD11; Millipore), Src phosphorylated on Y416 (Src-P-family; Cell Signaling Technology), Yes (#1; BD Biosciences), GM130 (#35, BD Biosciences), GalT (Yamaguchi and Fukuda, 1995) (provided by Michiko N. Fukuda, The Burnham Institute for Medical Research, La Jolla, CA), caveolin (Transduction Laboratories), calnexin (Stressgen Bioreagents), and CD43 (MEM59, Sanbio B.V.). MOPC21 (Sigma) was used for mouse control IgG. Rabbit anti-ACSL3 antiserum was generated against a GST-fusion protein containing the N-terminal amino acid residues (70-259) of human ACSL3 (Fujimoto et al., 2004) (provided by Y. Fujimoto and T. Takano, Teikyo University, Tokyo, Japan) and extensively absorbed with GST before use. Horseradish peroxidase (HRP)-F(ab′)2 of anti-mouse IgG, anti-rabbit Ig and anti-rat IgG secondary antibodies were from Amersham Biosciences. FITC-F(ab′)2 of anti-rabbit IgG or of anti-mouse IgG, TRITC-anti-rabbit IgG, TRITC-anti-mouse IgG (Fc specific) and Alexa Fluor 647-anti-rabbit IgG secondary antibodies were from BioSource International, Sigma and Invitrogen.
Cells and transfection
HeLa (Japanese Collection of Research Bioresource, Osaka, Japan) and COS-1 cells were cultured in Iscove's modified DME containing 5% fetal bovine serum at 37°C. THP-1 cells (provided by Atsushi Iwama, Chiba University, Chiba, Japan) were grown in suspension in Iscove's modified DME containing 5% fetal bovine serum and 50 μM 2-mercaptoethanol, and Dami cells were maintained in suspension culture in Iscove's modified DME supplemented with 7.5% horse serum, as described (Greenberg et al., 1988; Hirao et al., 1998; Sato et al., 2009), and attached to culture dishes during ~2-3 days of culture in Iscove's modified DME supplemented with 2.5% fetal bovine serum and 2.5% horse serum. Transient transfection was performed using TransIT transfection reagent (Mirus), LipofectAMINE™ 2000 reagent (Invitrogen) or linear polyethylenimine (25 kDa; Polysciences) (Matsuda et al., 2006; Kasahara et al., 2007a; Fukumoto et al., 2010).
Western blotting and immunoprecipitation
Cell lysates were prepared in SDS-PAGE sample buffer or Triton X-100 lysis buffer (50 mM HEPES, pH 7.4, 10% glycerol, 1% Triton X-100, 4 mM EDTA, 100 mM NaF, 50 μg/ml aprotinin, 100 μM leupeptin, 25 μM pepstatin A, and 2 mM PMSF), and subjected to SDS-PAGE and electrotransferred onto polyvinylidene difluoride membranes (Millipore). Immunodetection was performed by enhanced chemiluminescence (Amersham Biosciences), as described (Yamaguchi et al., 2001; Kasahara et al., 2004; Matsuda et al., 2006; Kasahara et al., 2007b; Ikeda et al., 2008; Kuga et al., 2008; Sato et al., 2009). Sequential reprobing of membranes with a variety of antibodies was performed after the complete removal of primary and secondary antibodies from membranes in stripping buffer or inactivation of HRP by 0.1% NaN3, according to the manufacturer's instructions. Results were analyzed using an image analyzer LAS-1000plus equipped with Science Lab software (Fujifilm, Tokyo, Japan) or ChemiDoc XRSPlus (Bio-Rad). Immunoprecipitation was performed using antibody-precoated protein-G beads, as described (Mera et al., 1999; Yamaguchi et al., 2001; Kasahara et al., 2004; Kasahara et al., 2007b; Ikeda et al., 2008; Sato et al., 2009). Intensity of chemiluminescence was measured using the ImageJ software (NIH, USA).
GST-fusion proteins were expressed in the E. coli AD202 strain (Nakano et al., 1993) (provided by Takashi Saito, RIKEN, Yokohama, Japan) upon incubation with 0.1 mM IPTG at 30°C for 3 hours. After bacteria were lysed at 4°C in phosphate-buffered saline (PBS) containing 1% Triton X-100, the Triton X-100-insoluble fraction was lysed at 4°C with Empigen lysis buffer (50 mM HEPES, pH 7.4, 10% glycerol, 3% Empigen BB, 4 mM EDTA, 100 mM NaF, 50 μg/ml aprotinin, 100 μM leupeptin, 25 μM pepstatin A, and 2 mM PMSF), as described previously (Yamaguchi et al., 2001). GST-fusion proteins were collected on glutathione-Sepharose beads from Empigen lysates and washed four times with Empigen lysis buffer. To identify proteins associated with the C-lobe, a pulldown assay was performed at 4°C for 2 hours in Triton X-100 lysates prepared from HeLa cells. After extensively washing with Triton X-100 lysis buffer, the bead pellets were analyzed by SDS-PAGE and Coomassie brilliant blue (CBB) staining.
Identification of p70 by peptide mapping
Proteins pulled down with GST-C-lobe were resolved by SDS-PAGE using 7% acrylamide and 2.7% bisacrylamide to increase separation of p70 and stained with CBB. The protein band corresponding to p70 was cut out and digested with Achromobacter protease I (Lys-C) in digestion buffer (100 mM Tris-HCl, pH 9.0, 1 mM EDTA, and 0.1% SDS). After the digestion, molecular mass analysis of Lys-C fragments was performed by matrix-assisted laser desorption-ionization time-of-flight mass spectrometry (MALDI-TOF-MS). Identification of protein was carried out by comparison between the molecular weights determined by MALDI-TOF-MS and theoretical masses from the proteins registered in NCBInr.
Immunofluorescence staining was performed as described (Yamaguchi and Fukuda, 1995; Tada et al., 1999; Nakayama and Yamaguchi, 2005; Kasahara et al., 2007c; Ikeda et al., 2008; Sato et al., 2009). In brief, cells were washed in warmed PBS and fixed in 4% paraformaldehyde for 20 minutes. Fixed cells were permeabilized and blocked in PBS containing 0.1% saponin and 3% bovine serum albumin for 30 minutes, and then incubated with a primary and a secondary antibody for 1 hour each. After washing with PBS containing 0.1% saponin, cells were mounted with ProLong antifade reagent (Molecular Probes). Confocal images were obtained using a Fluoview FV500 laser-scanning microscope with a ×40 1.00 NA oil-immersion objective (Olympus, Tokyo). 200-400 cells were scored for each assay. For immunofluorescence of THP-1 cells, cells in suspension were directly fixed with 4% paraformaldehyde and then attached onto coverslips by cytocentrifugation. Composite figures were prepared using Photoshop 11.0 and Illustrator 14.0 software (Adobe). Intensity of fluorescence was measured using the ImageJ software.
Tannic acid treatment
COS-1 cells cotransfected with Lyn-wt and VSV-G-GFP were incubated overnight at 40°C and then shifted to 19°C for 2 hours to accumulate both proteins in the Golgi. Subsequently, cells were treated with 0.5% tannic acid for 30 or 60 minutes at 32°C and fixed for confocal microscopy, as described (Polishchuk et al., 2004). After background subtraction using the ImageJ software, the number of PGCs containing Lyn-wt or VSV-G-GFP was counted from 15-25 cells.
ACSL3 gene silencing with shRNAs
Three different target-specific short hairpin RNA (shRNA) sequences were selected for silencing ACSL3 with help of web-based algorithms (http://www.clontech.com/) (ACSL3-A1, GCTGAAACGCAAAGAGCTT; ACSL3-A2, GACCAACATCGCCATCTTC; ACSL3-A3, CTCTTACAAACAAAGTTGA). EGFP-targeted shRNA was used as a control shRNA (GGCAAGCTGACCCTGAAGTTCA). The oligonucleotides for shRNA were annealed and subcloned into the XbaI and BglII sites of the pENTR4-H1 vector (provided by Hiroyuki Miyoshi, RIKEN BRC, Tsukuba, Japan). To visualize cells that were transfected with shRNA encoding pENTR4-H1, an mCherry-coexpressing shRNA vector was constructed. In brief, the mCherry vector was constructed from the pEGFP-C1 vector (Clontech) by replacing EGFP with the monomeric red fluorescent protein mCherry of the pRSET-B-mCherry vector (Shaner et al., 2004) (provided by Roger Y. Tsien, University of California, San Diego, CA), and the H1 promoter-driven shRNA cassette of pENTR4-H1 was subsequently inserted into the mCherry vector. The resulting shRNA-mCherry vector encodes both shRNA and mCherry. When shRNA-mCherry vectors were transiently transfected, the transfection efficiencies of shRNA were 47-71% in HeLa and COS-1 cells and 5-10% in Dami cells when we counted the number of cells expressing mCherry under a confocal laser-scanning microscope.
VSV-G-GFP transport assay
COS-1 cells cotransfected with VSV-G-GFP in conjunction with mCherry (control) or ACSL3-A3 shRNA-mCherry vector were incubated at 40°C for 14 hours and then shifted to 32°C for 30 minutes or 6 hours in the presence of 200 μg/ml cycloheximide for the last 3 hours, as described (Hirschberg et al., 1998; Hirose et al., 2004; Sato et al., 2009).
We thank Tadashi Yamamoto, Takahiro Fujino, Masato Okada, Shigeyuki Nada, Hiroyuki Miyoshi, Jennifer Lippincott-Schwartz, Mitsuo Tagaya, Roger Y. Tsien, Jonathon Pines, Takashi Saito, Atsushi Iwama, Tatsuya Takano and Yasuyuki Fujimoto for plasmids, cells and antibody. This work was supported in part by grants-in-aid for Scientific Research, Global COE Program (Global Center for Education and Research in Immune Regulation and Treatment) and Special Funds for Education and Research (Development of SPECT probes for Pharmaceutical Innovation) from the Japanese Ministry of Education, Culture, Sports, Science and Technology, and research grants from the Suzuken Memorial Foundation and the Mochida Memorial Foundation for Medical and Pharmaceutical Research. Y.O. is a Global COE Research Assistant.
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.066266/-/DC1
- Accepted May 4, 2010.
- © 2010.