The Rho GTPase Rac1 is required for recycling endosome-mediated secretion of TNF in macrophages
Rho GTPases are required for many cellular events such as adhesion, motility, and membrane trafficking. Here we show that in macrophages, the Rho GTPases Rac1 and Cdc42 are involved in lamellipodia and filopodia formation, respectively, and that both of these Rho GTPases are essential for the efficient surface delivery of tumor necrosis factor (TNF) to the plasma membrane following TLR4 stimulation. We have previously demonstrated intracellular trafficking of TNF via recycling endosomes in lipopolysaccharide (LPS)-activated macrophages. Here, we further define a specific role for Rac1 in intracellular TNF trafficking, demonstrating impairment in TNF release following TLR4 stimulation in the presence of a Rac inhibitor, in cells expressing a dominant negative (DN) form of Rac1, and following small interfering RNA (siRNA) knockdown of Rac1.
Rac1 activity was required for TNF trafficking but not for TLR4 signaling following LPS stimulation. Reduced TNF secretion was due to a defect in Rac1 activity, but not of the closely related Rho GTPase Rac2, demonstrated by the additional use of macrophages derived from Rac2-deficient mice. Labeling recycling endosomes by the uptake of fluorescent transferrin enabled us to show that Rac1 was required for the final stages of TNF trafficking and delivery from recycling endosomes to the plasma membrane. Thus, actin remodeling by the Rho GTPase Rac1 is required for TNF cell surface delivery and release from macrophages.
Keywords: cytokine; TNF; macrophage; Rac1; Rho GTPase; trafficking
Macrophages are crucial for establishing an inflammatory response to pathogenic invasion and injury1,2 and are also involved in chronic wound healing and tissue repair.3 Macrophages are stimulated by bacterial products, such as lipopolysaccharide (LPS), to release major pro-inflammatory cytokines, including tumor necrosis factor (TNF) and interleukin-6.4,5 The secretion of TNF and other cytokines is indispensable for immunity; however, dysregulation of this process contributes to the pathogenesis of chronic inflammatory disorders, such as inflammatory bowel disease and rheumatoid arthritis.6–8 TNF is a critical target for current medications such as anti-TNF antibodies and soluble analogs of the TNF receptor, aimed at treating inflammatory disorders.9 Thus, understanding more about the secretory pathway for TNF in macrophages is an area of significant scientific and clinical interest.
We have previously shown that newly synthesized TNF is trafficked in macrophages via the Golgi complex and recycling endosomes to specialized regions of the plasma membrane, such as phagocytic cups and filopodia, for secretion.10,11 Notably, TNF is delivered via
recycling endosomes to these sites on the plasma membrane that are cholesterol-rich12 and F-actin-rich.10 Therefore, F-actin-modifying proteins potentially have direct roles in mediating cytokine secretion. Rho GTPases are essential for actin cytoskeleton remodelling, which is a fundamental process in many cellular events such as cell adhesion, motility and membrane trafficking.13,14 With regulation by guanine nucleotide exchange factors (GEFs), GTPase-activating proteins and guanine nucleotide dissociation inhibitors, Rho proteins cycle between GDP-bound inactive states and GTP-bound active states in which they interact with effector molecules.15 Among the Rho GTPases characterized in mammalian systems are Rac1 and Cdc42, ubiquitously expressed and implicated in the formation of cell surface protrusions, cell signaling and protein trafficking.16,17 Rac1 has a closely related isoform, Rac2, whose expression is largely confined to cells of the immune system, and where each isoform has both redundant and distinct roles.14,15,18,19 Rac1 induces the formation of lamellipodia, actin-rich protrusions on the leading edge of mobile cells, while Cdc42 promotes the formation of filopodia,actin-rich finger-like projections for sensing and phagocytosis.14,15 In macrophages, Rac1 and Cdc42 are also important for Fc receptor- mediated phagocytosis, cell spreading and chemotaxis.
Specific Rho GTPases are also involved in protein trafficking and exocytosis in various cell types. Rac1 and Cdc42 are important for polarized, surface delivery of E-cadherin in kidney epithelial cells, for exocytosis in developing axons and dendrites in neuronal cells, and for regulated exocytosis of secretory granules from mast cells.23–26 Rac2 is uniquely required for exocytosis of azurophilic granules in neutrophils.27 Cdc42 is also essential for Golgi exit, and Rac1 has an auxiliary role in Golgi exit by binding to protein complexes that coordinate Rac1 signaling during clathrin-adaptor protein 1-coated vesicle biogenesis at the trans-Golgi network.
Thus, while these findings show that actin remodeling regulated by Rho GTPases is involved in protein trafficking and secretion in many cells, a role for these proteins in cytokine secretion from macrophages has not been explored. In this study, we hypothesized that Rho GTPase-mediated actin dynamics are required for cytokine trafficking from the Golgi complex to the cell membrane via recycling endo- somes in the F-actin-rich periphery of activated macrophages.
RESULTS
Rac1 GTPase and Cdc42 mutants alter membrane ruffling, lamellipodia and filopodia formation in macrophages F-actin-modifying proteins potentially have direct roles in mediating post-Golgi transport and surface release of cytokines from macro- phages. To explore Rac1 GTPase function in this context, GFP-Rac1 and its constitutively active (CA) (L61) and dominant negative (DN) (N17) mutants (GFP-Rac1 CA and GFP-Rac1 DN, respectively) were each expressed in RAW264.7 cells. As shown in Figure 1a, GFP-Rac1 wild type (WT) was found predominantly in the cytoplasm with some localization to the plasma membrane and intracellular mem- branes, with little or no effect on overall cell morphology. However, GFP-Rac1 CA-transfected cells exhibited dramatic alterations in their morphology, evident in extensive membrane ruffling and the forma- tion of numerous lamellipodia that were enriched in F-actin bundles. Interestingly, these lamellipodia were in constant motion in live cells (data not shown). In contrast, cells transfected with GFP-Rac1 DN were rounded in shape, with occasional filopodia. These results thus confirm previous reports of Rac1 localization and activity in macrophages.
RAW264.7 cells were transfected with another Rho GTPase, Cdc42, fused to eGFP (GFP-Cdc42) and its CA (V12) and DN (N17) mutants (GFP-Cdc42 CA and GFP-Cdc42 DN, respectively). Similar to GFP-Rac1, GFP-Cdc42 WT localized to the cytoplasm with some expression in intracellular sites and at the plasma membrane (Figure 1b). In contrast, however, GFP-Cdc42 CA induced a marked degree of filopodia formation with intensely spiky F-actin-rich projections. Cells expressing GFP-Cdc42 DN demonstrated fewer filopodia, but retained lamellipodia formation and demonstrated enhanced F-actin patches in pseudopods. These findings also confirm earlier reports on the behavior of Cdc42 in macrophages.
Effect of Rac1 GTPase mutants on TNF trafficking
We have previously established an assay for the measurement of TNF surface delivery as the penultimate step in its trafficking and secretion.31 Incubation of cells with TAPI to block TACE cleavage of TNF at the cell surface, followed by immunostaining of TNF on unpermeabilized cells, demonstrates surface TNF labeling in LPS- activated macrophages (Figure 2a). Expression of WT GFP-Rac1 did not interfere with the surface delivery of TNF, and abundant surface staining of TNF was evident in these cells, with enrichment in filopodia (Figures 2b and c). When complementary cells were permeabilized and stained for intracellular TNF, we confirmed that TNF was actively produced as denoted by staining in the perinuclear Golgi complex (Figure 2d).
Figure 1 Lamellipodia formation by GFP-Rac1 CA expression and filopodia formation by GFP-Cdc42 CA expression. Epifluorescence images of RAW264.7 cells that were transfected with GFP-Rac1 WT, CA or DN constructs (a) or GFP-Cdc42 WT, CA or DN constructs (b) for 5-6 h prior to fixation and staining with Alexa 594 phalloidin and DAPI. Results are representative of three independent experiments. Bar represents 10 mm.
In cells expressing GFP-Rac1 CA, there was also substantial surface staining of TNF, although this time, the TNF decorated Rac1-induced membrane ruffles rather than filopodia, and TNF was prominent on enhanced lamellipodia (Figure 2b). However, GFP-Rac1 DN inhibited the appearance of surface TNF in LPS-stimulated macrophages (Figures 2b and c). Quantification revealed a 450% decrease in surface delivery of TNF in cells expressing Rac1 DN (Figures 2b and c). TNF synthesis was stimulated in cells expressing GFP-Rac1 CA and GFP-Rac1 DN, as evidenced by the presence of TNF in the peri- nuclear Golgi complex (Figure 2d) and notably in peripheral punctate structures, presumably representing recycling endosomes as the penultimate destination for secretory TNF, prior to delivery to the plasma membrane. Interestingly, accumulation of TNF in these punctate structures was increased in cells expressing Rac1 DN (Figure 2d), concurrent with the reduction in TNF surface delivery. These results suggest that GTP-bound Rac1 is not needed to mediate signaling from LPS nor to initiate the synthesis of TNF, but it is required for a distal step in its trafficking and delivery to the cell surface.
Figure 2 TNF colocalizes with lamellipodia induced by GFP-Rac1 CA. (a) Assay for measurement of cell surface delivery of TNF. RAW264.7 cells were stimulated with LPS for 2 h to induce TNF secretion in the presence of TAPI. Cells were labeled for cell surface TNF (red) prior to permeabilisation and labeling for intracellular TNF (green). (b–d) RAW264.7 macrophages transiently transfected with WT, CA or DN mutants of GFP-tagged Rac1 were LPS- stimulated for 2 h in the presence of TAPI, then fixed and labeled for surface or intracellular TNF. (b) GFP-Rac1 WT and CA-expressing cells displayed prominent surface TNF staining, whereas surface TNF labeling was absent from cells expressing GFP-Rac1 DN (arrow). (c) Quantification of surface TNF in GFP-Rac1 mutant-transfected cells. Untransfected cells were quantified from a parallel well of RAW264.7 cells. AU indicates arbitrary units. (d) GFP-Rac1 WT, CA and DN-expressing cells showed intracellular staining of TNF in typical perinuclear and peripheral punctate localizations to indicate synthesis of TNF. (e) RAW264.7 macrophages were co-transfected with mCherry-TNF together with GFP-tagged constructs of Rac1 WT, CA and DN. Inset shows colocalized mCherry-TNF with GFP-Rac1 CA. Results are representative of at least four independent experiments. Bar indicates 10 mm.
TNF trafficking and secretion can also be examined in cells expressing fluorescently tagged constructs of TNF.31 In cells co-transfected with mCherry-TNF and GFP-Rac1 CA, the labeled TNF was clearly colocalized with Rac1 on lamellipodia, in Golgi and in peripheral compartments (inset, Figure 2e). This further demonstrates the trafficking of TNF through Rac1-enriched compartments and delivery of cytokines to Rac1-induced lamellipodia. Furthermore, TNF surface delivery was inhibited in cells co-transfected with mCherry- TNF and GFP-Rac1 DN, further suggesting that functional Rac1 is required for TNF trafficking to the plasma membrane.
Effect of Cdc42 mutants on TNF trafficking
Similar to Rac1 mutants, expression of GFP-Cdc42 DN inhibited surface delivery of TNF, while GFP-Cdc42 WT and CA did not alter this activity (Figure 3a). Although surface TNF was prominent on the enhanced lamellipodia produced by CA Rac1, there was less surface TNF associated with the exaggerated filopodia in cells expressing CA Cdc42. The DN form of Cdc42 prevented surface delivery of TNF, suggesting that cycling of this GTPase may also be required for TNF trafficking. The presence of intracellular TNF labeling could be demonstrated in these cells (Figure 3b), showing that functional TNF was made but not trafficked to the cell surface in cells expressing GFP-Cdc42 DN. TNF was present in fewer peripheral punctate structures in cells expressing GFP-Cdc42 DN than cells expressing GFP Rac1 DN, suggesting that Cdc42 may be required for initial stages of TNF trafficking for release from the Golgi complex. Thus Cdc42, as well as Rac1, contributes to TNF trafficking in macrophages.
Figure 3 GFP-Cdc42 mutants affect TNF trafficking. RAW264.7 macrophages transiently transfected with WT, CA or DN mutants of GFP-tagged Cdc42 were LPS-stimulated for 2 h in the presence of TAPI, then fixed and labeled for surface or intracellular TNF. (a) GFP-Cdc42 WT and CA-expressing cells displayed prominent surface TNF staining, whereas surface TNF labeling was absent from cells expressing GFP-Cdc42 DN (arrow). (b) GFP-Cdc42 WT, CA and DN-expressing cells showed intracellular staining of TNF in typical perinuclear and peripheral punctate localizations. Results are representative of at least four independent experiments. Bar indicates 10 mm.
Rac1 inhibition suppresses TNF release
We focused hereafter on the role of Rac1 in the trafficking and release of TNF, given the interesting punctate localization of TNF in cells expressing GFP-Rac1 DN. The transfected DN mutant of Rac1 may have numerous inhibitory effects due to its ability to sequester GEFs that bind to multiple Rho GTPases.14 Rac1 —/ — mice are embryonic lethal;32 therefore, we also tested the effects of the Rac inhibitor, EHT1864, on TNF secretion in RAW264.7 cells and in primary macrophages. This inhibitor binds to target amino acid residues in Rac1 and Rac2, and prevents association with various GEFs during receptor activation.33,34 EHT1864 treatment inhibited TNF release over a time course following LPS stimulation of primary bone marrow macrophages (BMMs) (Figure 4a) and RAW264.7 cells (Supplementary Figure 1). To investigate whether this effect was due to a trafficking defect or a role for Rac1 in TLR4 signaling following LPS stimulation, we first treated BMMs with LPS for 2 h to initiate TLR4 signaling, followed by subsequent treatment with EHT1864 (or DMSO as control) for a further 2 h. TNF secretion was reduced by B50% in cells treated with EHT1864 in this delayed manner compared with control treated cells (Figure 4b), suggesting that the reduced TNF secretion was due to a post-Golgi trafficking defect. Labeling BMMs for surface and intracellular TNF demonstrated that whilst EHT1864 treatment had no effect on TNF synthesis it led to reduced levels of surface TNF compared with control cells (Figure 4c). Importantly, this effect could be titrated by reducing the amount of EHT1864. BMMs treated with EHT1864 also exhibited reduced TNF-positive punctate structures in the cell periphery, indicating that TNF post-Golgi trafficking was perturbed by Rac1 inhibition. These data confirm that Rac1 inhibition reduces surface delivery and secretion of TNF, but not its synthesis.
Figure 4 The Rac inhibitor EHT1864 blocks TNF secretion and TNF surface delivery. BMMs were differentiated from the bone marrow of C57BL/6 mice and incubated with LPS in the presence or absence of 15 mM of the Rac inhibitor, EHT1864 or DMSO as a control. (a) Media samples were harvested every 2 h for a total of 6 h to measure TNF secretion by ELISA. Data show reduced TNF secretion from EHT1864-treated cells and are presented as mean±s.e.m. of three independent experiments, analyzed by one-way ANOVA with Bonferroni post-test applied (**Po0.01). (b) WT BMMs were incubated with LPS for 2 h, prior to incubation with 15 mM of the Rac inhibitor, EHT1864 or DMSO for 2 h. Media samples were analysed for TNF secretion by ELISA. Data show reduced TNF secretion following delayed treatment with EHT1864 and are presented as mean±s.e.m. of three independent experiments, analyzed by Student’s t-test (*Po0.05). (c) BMMs were incubated with LPS and TAPI in the presence or absence of 5 or 15 mM of the Rac inhibitor, EHT1864 for 2 h, prior to labeling for intracellular and cell surface TNF. Epifluorescence microscopy revealed normal trafficking of TNF to the cell surface in control cells. Cells treated with EHT1864 showed a concentration-dependent reduction of surface TNF. Line scan graphs of representative cells to the right of each panel represent the fluorescence intensities of surface TNF (red) and intracellular TNF (green) across the distance of the yellow line. Peak fluorescence intensities at the plasma membrane (surface TNF, red) and at the Golgi apparatus (intracellular TNF, green) are demonstrated by colored arrows. These images are representative of three independent experiments. Bar indicates 10 mm.
Rac2 is not required for TNF trafficking
Macrophages also express the Rac2 isoform,35 which has the potential for functional redundancy with Rac1. Overlapping or compensatory actions of Rac2 could compromise the effects of Rac inhibitor experiments, since EHT1864 is also known to bind to Rac2.33 We were able to test for a role for Rac2 in TNF secretion through the use of BMMs differentiated from Rac2 —/ — mice that express normal levels of Rac1 but no Rac2.36 Activation of WT and Rac2 —/ — BMMs with LPS resulted in an almost identical TNF secretion profile over a 6 h time course (Figure 5a), indicating that the lack of Rac2 does not compromise the amount or pattern of TNF secretion, and that Rac2 alone is not required for this function. Rac2 —/ — BMMs were then treated with the Rac inhibitor EHT1864, aimed at additionally inhibiting Rac1 in these cells. Our findings showed that EHT1864 significantly inhibited TNF secretion over the 6 h time course in Rac2 —/ — BMMs (Figure 5b), in a manner similar to that seen in WT BMMs (Figure 4a). Furthermore, TNF trafficking and surface delivery were markedly decreased in Rac2 —/ — BMMs upon treatment with EHT1864, whilst TNF synthesis and labeling in the Golgi complex was unaffected (Figure 5c). This additionally confirms that Rac1 inhibition reduces TNF secretion in both WT and Rac2 —/ — BMMs, and shows that there is no additive effect and therefore no discernible role for Rac2 in TNF secretion.
Figure 5 Rac2 gene deletion has no effect on TNF secretion. (a) BMMs were differentiated from bone marrow of WT and Rac2 —/ — mice and incubated with LPS for 6 h. Media samples were harvested every 2 h to measure TNF by ELISA, and showed no difference in TNF secretion. Data are presented as mean±s.e.m. of three independent experiments for a total of 3 pairs of WT and Rac2 —/ — mice. (b) Rac2 —/ — BMMs were treated with LPS and EHT1864 or DMSO as a control, and media samples were collected every 2 h for a total of 6 h and analyzed for TNF secretion by ELISA. Rac2 —/ — BMM treated with EHT1864 showed a significant decrease in TNF secretion following LPS stimulation. Data are presented as mean±s.e.m. of three independent experiments, analyzed by one-way ANOVA with Bonferroni post-test applied (***Po0.001). (c) Rac2 —/ — BMMs were incubated with LPS and TAPI in the presence or absence of 5 or 15 mM of the Rac inhibitor, EHT1864 for 2 h, prior to labeling for intracellular and cell surface TNF. DMSO-treated cells, as well as those treated with 5 mM EHT1864, displayed typical strong accumulation of surface TNF, while cells treated with 15 mM EHT1864 exhibited reduced surface TNF. All cells were positive for intracellular TNF, indicating that TNF synthesis was not affected under any of the conditions. Line scan graphs of representative cells to the right of each panel indicate the fluorescent intensities of surface TNF (red) and intracellular TNF (green) across the distance of the yellow line. Peak fluorescence intensities at the plasma membrane (surface TNF, red) and at the Golgi apparatus (intracellular TNF, green) are demonstrated by colored arrows. Images are representative of a total of three mice analyzed. Bar indicates 10 mm.
Effect of siRNA knockdown of Rac1
To explore whether Rac1 itself is important in TNF trafficking and secretion, we used small interfering RNA (siRNA) to knockdown Rac1 expression in RAW264.7 cells and to avoid nonspecific effects on other Rho GTPase effector pathways. Three pooled siRNAs that targeted mouse Rac1 were transfected into RAW264.7 macrophages. Western blotting on cell extracts confirmed an almost complete suppression of Rac1 expression in siRNA knockdown cells (Figure 6a). The complementary negative control (high GC content) RNA sequence did not reduce Rac1 expression. Next, TNF secretion into the medium was measured following LPS stimulation, and showed a significant decrease (34%) in TNF release after Rac1 knockdown compared with control cells (Figure 6b).
The intracellular localization of TNF was next investigated in siRNA knockdown cells. We observed an accumulation of TNF in punctate structures in LPS-stimulated siRNA knockdown cells (Figure 6c). The cells were also incubated with fluorescent transferrin, a known cargo for recycling endosomes11 (Supplementary Figure 2). TNF accumulation coincided with transferrin-positive recycling endosomes. These findings suggest that Rac1 is required for traffick- ing of TNF from the recycling endosome to the cell surface, and that inhibition of Rac1 leads to an accumulation of TNF in recycling endosomes. Taken together, these results show that Rac1 itself is required in the TNF secretory pathway in macrophages.
Rac1 is required to mediate the final stages of TNF delivery from recycling endosomes to plasma membrane
To further investigate the mechanism of reduced TNF delivery to the plasma membrane in cells deficient in Rac1 activity we next examined the intracellular localization of TNF in RAW264.7 cells and BMMs treated with EHT1864. In control-treated cells double-labeled for intracellular and cell surface TNF, there was significant TNF labeling on the plasma membrane, concentrated in ruffles and lamellipodia (Figures 7a and b). Intracellular TNF was present in the Golgi complex and in small punctate recycling endosomes scattered throughout the cell periphery. Surface delivery of TNF was completely abrogated in cells treated with EHT1864, whilst significant intracel- lular pools of TNF could be observed in the perinuclear region in RAW264.7 cells, colocalized with transferrin in recycling endosomes (Figure 7a). In BMMs treated with EHT1864, intracellular TNF continued to be concentrated in perinuclear regions and in large peripheral structures that co-localized with transferrin (Figure 7b,i and ii). Therefore, TNF trafficking to recycling endosomes at peripheral and perinuclear localizations was not affected in either RAW264.7 cells or BMMs treated with EHT1864. However, subse- quent trafficking of TNF to the plasma membrane was severely impaired when Rac1 was inhibited, coinciding with additional accumulation of TNF in recycling endosomes. These findings demonstrate that Rac1 is required for the final stages of TNF delivery to the cell surface.
We then examined TNF localization in RAW264.7 cells expressing
GFP-Rac1 DN compared with control cells expressing GFP only. Transferrin-loaded recycling endosomes were present in all GFP-Rac1- transfected cells, indicating that the integrity of recycling endosomes is not affected by the activation status of Rac1. TNF colocalized with transferrin more frequently in the cell periphery in cells expressing GFP Rac1-DN compared with cells expressing GFP only (Figure 7c), suggesting that recycling endosome to plasma membrane trafficking of TNF is interrupted in cells that express a DN mutation in Rac1.
Figure 6 siRNA knockdown of Rac1 reduces TNF secretion and causes accumulation of TNF in recycling endosomes. (a) RAW264.7 cells were incubated for 48 h with two treatments of Rac1 pooled siRNA every 24 h at three doses (in pmol). At the end of the incubation, cells were lysed and 20 mg per lane protein was loaded for immunoblot analysis to detect Rac1. Negative control was high GC content at the same doses. (b) LPS- induced TNF secretion from macrophages was inhibited by treatment with siRNA to Rac1 compared with negative control RNA. Data are presented as mean±s.e.m. of three independent experiments, analyzed by one-way ANOVA with Bonferroni post-test applied (*Po0.05, ***Po0.001). (c) Intracellular TNF immunoreactivity strongly colocalizes with mouse transferrinþ recycling endosomes in siRNA-treated macrophages. Grey indicates F-actin staining from phalloidin conjugated to Alexa 647. Bar indicates 5 mm.
Figure 7 Rac1 is required for the final stages of TNF trafficking from recycling endosomes to plasma membrane. (a) RAW264.7 cells were treated with DMSO or 30 mM EHT1864 and stimulated with LPS for 2 h to induce TNF secretion in the presence of TAPI. Cells were incubated in the presence of transferrin-A647 to label recycling endosomes (blue). Cells were labeled for cell surface TNF (green) prior to permeabilization and labeling for intracellular TNF (red). (b) BMMs were differentiated from the bone marrow of WT mice and incubated with LPS for 2 h, prior to incubation with 15 mM of the Rac inhibitor, EHT1864 or DMSO as a control, and TAPI for 2 h. BMMs were incubated in the presence of transferrin-A647 to label recycling endosomes, then labeled for intracellular and cell surface TNF. Epifluorescence microscopy revealed normal trafficking of TNF to the cell surface in control cells. Cells treated with EHT1864 showed reduced surface TNF with concurrent accumulation and colocalization of intracellular TNF with transferrin in perinuclear and peripheral locations. (c, d) RAW264.7 macrophages were transfected with GFP only (c) or GFP-tagged Rac1 DN ((d) shown in white), and stimulated with LPS for 2 h to induce TNF secretion. Cells were also incubated with transferrin-A647 to label recycling endosomes (green). Cells were permeabilised and labeled for TNF (red). Transfection with GFP-Rac1 DN leads to increased accumulation of TNF in recycling endosomes. Pearson’s R coefficient is shown in the corner of each image to denote the degree of colocalization between TNF and transferrin in these images. These images are representative of three independent experiments.
DISCUSSION
Rho GTPases are ubiquitously expressed and contribute to cell homeostasis through regulation of many cell functions, including cell division, motility, morphological changes, and signal transduc- tion.13,14 They have also been implicated in exocytic trafficking of protein cargo in mast cells,25 neutrophils27 and eosinophils,37 but to date, there has been no detailed investigation of a role for Rho GTPases in cytokine trafficking in macrophages. Previously, Rac1 has been implicated in TNF synthesis from LPS-activated macrophages by the use of Rac1 DN, which suppressed extracellular release of TNF.38 Rac1 is also implicated in pro interleukin-1 cytokine expression in response to macrophage scavenger receptor stimulation.39 However, no studies have determined the precise contribution of Rac1 or Cdc42 to cytokine trafficking in macrophages. In this study, we have given evidence that Rac1, and potentially Cdc42, are essential for regulating the post-Golgi trafficking and secretion of TNF. Here, we used three complementary approaches to reduce active Rac1 in both the RAW264.7 macrophage cell line and in primary BMMs, expression of a DN mutant of Rac1, siRNA knockdown of Rac1 and a pharmacologic inhibitor of Rac activity.
Transient expression of functional mutants of Rac1 and Cdc42 in macrophages produced dramatic changes in cell morphology. Macro- phages expressing CA Rac1 exhibited an abundance of highly mobile lamellipodia. Conversely, expression of active, GTP-bound Cdc42 produced exaggerated filopodia in cells. These finding are consistent with previous studies in macrophages and fibroblasts, which have shown that activation of Rac1 and Cdc42 results in the formation of F-actin-rich lamellipodia and filopodia, respectively.30,40–43
Although the CA mutants produced dramatic changes in cell shape, including exaggerated filopodia that are normally associated with TNF release,10 and lamellipodia shown to accept recycling membrane delivery,44 neither of these mutants visibly altered (increased) the surface delivery of TNF. Thus, cell surface protrusions alone are neither sufficient nor necessary to enhance trafficking or fusion of TNF exocytic carriers with the plasma membrane. However, the expression of DN mutants of GFP-tagged Rac1 and Cdc42 in macrophages substantially reduced both surface protrusions and the amount of TNF delivered to the cell surface, although the cytokine was still actively synthesized and reached the perinuclear Golgi complex. These DN mutations are not simply latent GDP-bound forms and can lead to binding of GEFs upstream of Racs and Rhos, since GEFs are promiscuous. Therefore, the Rac1 DN mutant may block activation of other GTPases that are dependent on GEF activation. To rule out the contribution of non-specific effects of Rac1DN on other GTPases we also used a Rac inhibitor and siRNA to Rac1.
The mode of action of the Rac inhibitor EHT1864 is to prevent the association of GTP-bound Rac1 with its effectors (specifically, p21- activated kinase or PAK) via nucleotide displacement.33 EHT1864 very effectively reduced the pool of TACE-dependent TNF at the cell surface in staining experiments and reduced TNF secretion by B40%. To further address the question of where Rac1 might be acting in the TNF secretory pathway we delayed drug treatment with EHT1864 until 2 h following LPS stimulation and demonstrated that reduced
TNF secretion following Rac1 inactivation is due to impaired TNF trafficking rather than reduced TLR4 signaling. This was concurrent with reduced surface labeling of TNF and increased accumulation of TNF in recycling endosomes labeled with transferrin. In summary, our observations require a more intensive pharmacological follow-up to determine whether EHT1864 may be a potential candidate for therapeutic control of excessive TNF secretion in disease.
We also considered a redundant or complementary role for the immune-specific Rac2 isoform in macrophages. Despite sharing 490% homology, Rac1 and Rac2 have both overlapping and separate functions, and bind to distinct proteins in T cells and neutrophils.15 For example, Rac2-deficient but not Rac1-deficient BMMs exhibit decreased levels of F-actin and numbers of actin-rich podosomes, while Rac1-deficient but not Rac2-deficient BMMs are impaired in their ability to migrate through 3D extracellular matrices.45 Using Rac2 —/ — BMMs, we found that LPS activation of these cells resulted in TNF synthesis and secretion in a pattern and at levels fully comparable with those of WT BMMs, implying that Rac2 is not required for cytokine secretion in macrophages. Thus, TNF secretion is a macrophage function where Rac1 and Rac2 have differential or exclusive roles, as also seen in NADPH oxidase regulation, chemotaxis, phagocytosis and spreading.45–47 In an attempt to eliminate the function of both Rac1 and Rac2, Rac2 —/ — BMMs were additionally treated with EHT1864. We found that EHT1864 also inhibited TNF secretion in Rac2 —/ — BMMs. These findings suggest that Rac1 is the predominant isoform required for TNF secretion, and that there is no additive effect of knocking out Rac2 activity in BMMs.
Finally, in vitro siRNA knockdown was also used to study the effects of Rac1 in macrophage TNF trafficking. A near-complete knockdown of Rac1 expression led to B34% inhibition of TNF secretion, similar to the level of inhibition achieved with the Rac1 inhibitor EHT1864. Analysis of the intracellular sites of TNF in siRNA-treated macrophages revealed that TNF was harbored in transferrin-containing recycling endosomes, suggesting that TNF was trapped in recycling endosomes and prevented from reaching the cell surface, following LPS stimulation. BMMs and RAW264.7 cells treated with EHT1864 or expressing the DN form of Rac1 also displayed increased accumulation of TNF in transferrin-loaded recycling endosomes, concurrent with reduced cell surface labeling of TNF. These observations are consistent with TNF being transported via recycling endosomes to the cell surface by a Rac1-dependent pathway or mechanism.
The actin cytoskeleton regulates the localization of organelles.48,49 Actin filament tracks also serve a direct role in trafficking as ‘railways’ for cargo-loaded vesicle or endosome movement.50,51 Actin anchors recycling endosomes at the cell periphery,52 and Rho GTPases have been implicated in regulating actin dynamics for the motility and positioning of endosomes.53–55 It is possible that inhibition of Rac1 prevents the recruitment of effectors needed for Rac1-mediated, actin- dependent recycling endosome positioning or movement. Rac isoforms may mediate TNF trafficking by activating actin nucleators (Arp2/3 complex) via the recruitment of WAVE proteins to create transport tracks to the recycling endosomes or the cell surface. Our observation that intracellular TNF in EHT1864-treated cells, cells expressing a DN form of Rac1 and Rac1 siRNA-treated cells was concentrated in the Golgi complex and recycling endosomes (confirmed from co-localization experiments with transferrin) may further narrow down the site of Rac1 action to transport steps between recycling endosomes to the cell surface, which are myosin and actin-dependent steps.56
In summary, we have provided substantial, novel evidence that associates Rac1 with the TNF secretory pathway in macrophages. Inhibition of both Rac1 and Cdc42 via expression of DN mutants disrupted surface delivery of TNF. Likewise, pharmacologic inhibition of Rac1 resulted in decreased delivery of TNF to the cell surface. In vivo ablation of Rac2 in Rac2 —/ — macrophages is definitive, and in our experiments we used primary cells to rule out a role for Rac2 in TNF trafficking and secretion. Finally, the accumulation of TNF in recycling endosomes after Rac1 inactivation suggests that Rac1 exerts its effect on trafficking at a post-Golgi step and functions to orchestrate the final stages of delivery from recycling endosomes to the cell surface. Rac proteins quite possibly regulate actin-dependent positioning and distribution of recycling endosomes, and control the formation of actin filament tracks, thus promoting the delivery of TNF-loaded vesicles to the cell surface.
METHODS
Antibodies, plasmids and inhibitors
Antibodies to Rac1 (ARC03 mouse monoclonal, Cytoskeleton Inc., Denver, CO, USA), pan-actin (MAB1501 mouse monoclonal, Millipore, Kilsyth, Victoria, Australia) and mouse TNF (rabbit polyclonal, Merck, Darmstadt, Germany and rat monoclonal, BD Biosciences, San Jose, CA, USA) were used for Western blotting and immunofluorescence. Secondary antibodies for immunofluorescence were Alexa Fluor 488- or Cy3-conjugated anti-mouse or -rabbit IgGs (from Jackson ImmunoResearch Laboratories, West Grove, PA, USA). The following Rho GTPase cDNA constructs were used in this project: eGFP-tagged WT mouse Rac1 and Cdc42; eGFP-tagged DN Rac1 (N17) and Cdc42 (N17),30,41,57,58 and cGFP-tagged CA Rac1 (L61)59 and Cdc42 (V12).41
These constructs were inserted into the pEGFP-C and -N versions of commercially available GFP plasmids (Clontech, Mountain View, CA, USA) and were obtained from Dr Gary Eitzen, Department of Cell Biology, University of Alberta, Edmonton, Canada. The full-length TNF sequence was N-terminally tagged with mCherry (TNF-mCherry). Mouse transferrin was conjugated to Alexa Fluor 647. siRNA sequences to mouse Rac1 were obtained from Invitrogen (Mulgrave, VIC, Australia), and TNF protease inhibitor (TAPI) was purchased from Merck. The Rac1 inhibitor, EHT1864 was purchased from Tocris Bioscience (Bristol, UK) and used at final concentrations of 5–30 mM.
Animals
Rac2 —/ — mice bred against the C57BL/6 background (Jackson Lab strain: B6.129S6-Rac2tm1Mddw/J) were originally shipped from the laboratory of Dr Mary C Dinauer, Washington University School of Medicine, St Louis, MO, USA. These mice were generated by targeted disruption of exon 1 of the Rac2 gene via homologous recombination in embryonic stem cells from C57BL/6 mice.36 Rac2 —/ — mice are viable and fertile, and have a normal lifespan, but exhibit immunosuppressive defects.36,60,61 Mice used in these experiments were 8–20 weeks of age. Age-matched C57BL/6 mice were used as WT controls, and were housed and treated under the same conditions. All animal experiments were approved by the University of Queensland Animal Ethics Committee.
Cell culture and transfection
The mouse RAW264.7 macrophage cell line was cultured in RPMI-1640 (BioWhittaker, Lonza, Mt Waverley, Victoria, Australia), supplemented with 10% heat-inactivated fetal bovine serum and 1% L-glutamine, and maintained in a humidified 5% CO2 atmosphere at 37 1C, as described previously.31 BMMs were also used. BM was harvested from mouse femurs and BMMs were differentiated in vitro from WT and Rac2 —/ — mice and used on day 7.62 All cells were cultured at 37 1C in 5% CO2 in RPMI-1640 complete media (Lonza) containing 10% fetal bovine serum (Thermo Trace, Scoresby, Victoria, Australia) and 1% L-glutamine (Invitrogen), supplemented with 20 Uml—1 penicillin (Sigma-Aldrich, Castle Hill, New South Wales, Australia), 20 mg ml —1 streptomycin (Invitrogen) and 100 ng ml—1 recombinant macrophage colony- stimulating factor 1 (R&D Systems, Minneapolis, MN, USA).63
For transient expression of cDNA, cells at 50% confluence were transfected using Lipofectamine 2000 (Invitrogen), according to the manufacturer’s instructions. Cells were typically used for experiments 6 h after transfection. For siRNA knockdown of Rac1, cells were transfected with pooled specific stealth siRNAs (Invitrogen) using Lipofectamine 2000 (Invitrogen), cultured for 24 h, retransfected under the same conditions, and cultured for another 24 h before LPS activation. Controls were cells treated with GC-matched negative control sequences (Invitrogen). Western blot analysis was used to verify protein expression knockdown following siRNA transfection.
Cell lysate preparation and Western blot analysis
To prepare lysates, RAW264.7 cells were washed with ice-cold PBS and lysed with 200 ml cold lysis buffer (10 mM Tris, pH 7.4, and 0.5% Triton X-100) containing protease inhibitor cocktail (Roche Applied Science, Brisbane, Queensland, Australia), followed by passaging through a 27.5-gauge needle and centrifugation at 14 000 g to pellet cellular debris and nuclei. Supernatants were assayed for protein content (Bradford assay, Bio-Rad Laboratories, Gladesville, New South Wales, Australia). Proteins were separated on SDS– PAGE with 20 mg of total protein from each sample loaded per well.31
For Western blot analysis, proteins were transferred by semi-dry transfer to PVDF membranes (Millipore). Membranes were blocked and probed with primary anti-Rac1 antibody (Cytoskeleton, Inc.). Anti-Rac1 was detected by horseradish peroxidase-conjugated secondary antibody, followed by substrate addition (SuperSignal West Pico Chemiluminescent Substrate Kit, Pierce, Rockford, IL, USA). Chemiluminescence was captured using film exposure and development.
TNF secretion assay
To induce synthesis and secretion of TNF, cells were treated with 100 ng ml—1 LPS (Salmonella enterica serotype Minnesota Re 595; Sigma-Aldrich) at 37 1C for varying times. To assess cell surface-delivered TNF, secreted TNF cleavage was blocked by addition of a TACE inhibitor, TAPI (TNF protease inhibitor, Merck, Dallas, TX, USA) added to cells at 2 mM during LPS stimulation.10–12,31 Levels of secreted TNF in supernatants were determined by BD OpTEIA Mouse TNF ELISA Set II kit (BD Biosciences) using the manufacturer’s protocol.
Immunofluorescence and imaging
Immunofluorescence staining was performed as described previously64 on coverslip-adherent macrophages fixed for 30 min in 4% PFA. Surface TNF was labeled on cells prior to permeabilisation with a rabbit polyclonal antibody (Merck). Cells were then permeabilized with 0.1% Triton X-100 for 5 min and then labeled for intracellular TNF with a rat monoclonal antibody (BD Biosciences), and nuclei were counterstained with DAPI. To label recycling endosomes, cells were serum-starved for 30 min and incubated with 10 mg ml —1 mouse transferrin (Sigma-Aldrich) conjugated to A647 (Invitrogen) for 20 min, followed by a 10-min chase prior to fixation and labelling. Coverslips were mounted in ProLong Gold reagent (Invitrogen) prior to imaging.
Epifluorescence images were captured on an Olympus BX-51 microscope with a 60 × /NA 1.35 oil immersion objective using an Olympus DP-70 camera. Confocal images were acquired on a Carl Zeiss LSM510 or 710 META inverted microscope with a 60 × /NA 1.4 oil immersion objective. Deconvo- luted images were obtained from a Deltavision live cell deconvolution microscope with a 60 × /NA 1.4 oil immersion objective. Images were analyzed using associated software: Olympus DP Controller v. 2.1 (epifluorescence), LSM510 META v. 3.2 analysis software (confocal) or Imaris 7.0 image analysis software (Bitplane, Inc., Saint Paul, MN, USA) for deconvoluted images from the Deltavision system. Images were further analyzed as necessary using Adobe Photoshop v. 9 and ImageJ v. 1.42q (NIH).
Image analysis
Fluorescence intensity of surface TNF labeling in Figure 2 was quantified by selection of region of interests from GFP þ cells in at least 5 high-powered fields per condition (total of 20–30 cells per bar). Line scan graphs of representative cells in Figures 4 and 5 indicate the fluorescent intensities of surface TNF (red) and intracellular TNF (green) across the distance of the line in highlighted cells. Red arrows are used to demonstrate peak surface TNF labeling at the plasma membrane, whilst green arrows indicate peak intracellular TNF labeling at the Golgi complex.
Statistical analyses
Statistical analysis on experimental data was done with GraphPad Prism versions 4.0/5.0. Data are presented as mean±s.e.m., except in cases where data is representative. Depending on the number of variables, data were analyzed using one-way or two-way ANOVA with Bonferroni post-hoc analysis applied,EHT 1864 and P values o0.05 were considered significant.