Mechanobiology June 26th - June 2nd 2016

Mechanobiology: June 26th  - June 2nd 2016

Summary

Increases in reactive oxygen species (ROS) have been implicated in age-related diseases, including cancer. The serine/threonine kinase protein kinase D1 (PKD1) is a stress-responsive kinase and sensor for reactive oxygen species, which can initiate cell survival through NF-κB signaling. We have previously shown that in response to ROS, PKD1 is activated at the mitochondria. However, the initial signaling events leading to localization of PKD1 to the mitochondria are unknown. Here, we show that formation of mitochondrial diacylglycerol (DAG) and its binding to PKD1 is the means by which PKD1 is localized to the mitochondria in response to ROS. Interestingly, DAG to which PKD1 is recruited in this pathway is formed downstream of phospholipase D1 (PLD1) and a lipase-inactive PLD1 or inhibition of PLD1 by pharmacological inhibitors blocked PKD1 activation under oxidative stress. To date it has been viewed that monosaturated and saturated DAG formed via PLD1 have no signaling function. However, our data describe a role for PLD1-induced DAG as a competent second messenger at the mitochondria that relays ROS to PKD1-mediated mitochondria-to-nucleus signaling.

Introduction

Mitochondrial oxidative stress caused by reactive oxygen species (ROS) is frequently implicated in a wide range of age-related disease processes, including various cancers, atherosclerosis, diabetes, neurodegenerative diseases and, in general, processes that regulate cellular and organismal aging (Beal, 2003; Finkel, 2005; Lustbader et al., 2004; Schieke and Finkel, 2006). In tumor cells, ROS homeostasis affects a wide range of cellular processes and, depending on the level, source and species of radical produced, can elicit responses that include proliferation and survival or senescence and apoptosis (Leslie et al., 2003; Matsuzawa et al., 2002; Storz, 2005; Wang et al., 2000). Therefore, the signaling pathways initiated by ROS are of great interest as their modulation could be an efficient therapeutic approach. In this context, the serine/threonine kinase protein kinase D1 (PKD1) is gaining much interest as a potential therapy target, because it is an oxidative stress sensor that induces mitochondria-to-nucleus signaling. However, further insight into the mechanism by which PKD1 is activated in response to oxidative stress is necessary to develop strategies to modulate PKD1-mediated signaling events.

Mitochondria are the major source of ROS within the cell, as well as being a major target for ROS-mediated damage (Balaban et al., 2005). Mitochondrial reactive oxygen species (mROS) are formed as by-products of the electron transport chain during the generation of ATP, or by an imbalance of cellular oxidant/antioxidant systems. Mitochondrial ROS activate PKD1, which is subsequently involved in mitochondria-to-nucleus signaling (Storz et al., 2005b). PKD enzymes are serine/threonine kinases, which were classified as members of the Ca2+/calmodulin-dependent kinases (CaM-K) family of kinases (Manning et al., 2002). The PKD group includes the isoenzymes PKD1 (Johannes et al., 1994), PKD2 (Sturany et al., 2001) and PKD3 (Matthews et al., 2003), which share a unique molecular structure that comprises a catalytic (kinase) domain and a regulatory region, including two cysteine-rich zinc-finger domains C1a and C1b (CRD).

PKD1 can be activated by a variety of stimuli and its functions within cells range from Golgi complex organization and plasma membrane-directed transport to stress responses and cellular survival (for reviews, see Rozengurt et al., 2005; Rykx et al., 2003; Wang, 2006). In response to H2O2 and mROS, PKD1 activates the transcription factor NF-κB through its canonical activation pathway (Storz et al., 2004), leading to the induction of genes promoting cellular detoxification and survival (Storz et al., 2005a; Storz et al., 2005b). The kinase activity of PKD1 is tightly regulated by phosphorylation and lipid binding. Depending on the stimulus and cellular compartment, different mechanisms are used to regulate PKD1 localization and activation. The most common form of PKD1 activation occurs at the plasma membrane (Rey et al., 2001; Van Lint et al., 1998; Wood et al., 2005). In various receptor-activated pathways, activation of phospholipase Cβ or γ (PLCβ/γ) leads to the production of DAG in the membrane. The increase in DAG causes PKD1 and novel PKC isoforms (nPKCs) to localize from the cytosol to the inner leaflet of the plasma membrane, where both proteins bind to DAG via their respective C1 domains (Matthews et al., 1999). Localization to the membrane releases autoinhibition of PKD1 and brings it into proximity with nPKCs, which transphosphorylate the PKD1 activation loop residues S738/S742. Phosphorylation of these sites correlates with PKD1 activity and has been shown to be sufficient for the full activation of PKD1 (Baron and Malhotra, 2002; Diaz Anel and Malhotra, 2005; Iglesias et al., 1998b).

In response to oxidative stress, PKD1 translocates to the mitochondria, where activation occurs via additional sequential phosphorylation steps at three distinct sites. These are coordinated signaling events initiated by Src kinase. First, Src kinase activates Abl, which in turn directly phosphorylates PKD1 at Y463 in the PH domain (Storz and Toker, 2003). This releases autoinhibition of PKD1, exposing the catalytic domain, most probably via an increase in negative charge that results in a conformational change (Storz and Toker, 2003). Src further phosphorylates PKD1 directly at tyrosine residue 95 (Doppler and Storz, 2007). Phosphorylation at this site creates a motif matching the consensus sequence for a binding site for the PKCδ C2 domain. Binding of PKCδ facilitates transphosphorylation of the PKD1 activation loop residues S738/S742 in the exposed catalytic kinase domain and fully activates the kinase (Storz et al., 2003). At this point, no role was known for DAG in this activation mechanism.

Translocation of proteins with C1 domains occurs due to rapid, transient increases of DAG levels in the target membrane that are produced by phosholipase C and phospholipase D (PLD) enzymes. PLC hydrolyses phosphatidylinositol-4,5-bisphosphate [PtdIns(4,5)P2; PIP2] to form DAG and inositol-1,4,5-triphosphate [Ins(1,4,5)P3; IP3]. PLD catalyses hydrolysis of phosphatidylcholine (PC) to phosphatidic acid (PA) and free choline. PA can then be converted into DAG and lysophosphatidic acid (LPA) via phosphatidic acid phosphohydrolases. PKD1 has been shown to be activated in response to both DAG generated by PLCγ and PLD1 (Baron and Malhotra, 2002; Kam and Exton, 2004; Wood et al., 2005; Zugaza et al., 1997). Previously published data indicate that PLD1 and PLCγ can both be activated in response to oxidative stress and are, therefore, candidates for DAG production under such conditions, although the consequences of this activation are still ill-defined (Banno and Nozawa, 2003; Hagele et al., 2007; Min et al., 1998; Varadharaj et al., 2006; Wang et al., 2001).

We here analyzed the mechanism by which PKD1 activation is initiated by oxidative stress. We show that formation of mitochondrial diacylglycerol (DAG) and its binding to PKD1 are the means by which PKD1 is localized to the mitochondria in response to ROS. Interestingly, the DAG to which PKD1 is recruited in this pathway is formed downstream of phospholipase D1. It is currently thought that monosaturated and saturated DAG formed via PLD1 have no signaling function (Pettitt et al., 1997). However, our data describe a role for PLD1-induced DAG as a competent second messenger at the mitochondria that relays ROS to PKD1 signaling.

Results

Activation of PKD1 by mitochondrial oxidative stress

Protein kinase D1 is a sensor for oxidative stress that is activated by reactive oxygen species (Storz and Toker, 2003; Waldron and Rozengurt, 2000; Zhang et al., 2005). Oxidative stress leads to the phosphorylation of PKD1 at the activation loop serines S738 and S742, an event that directly correlates with increased PKD1 activity (Waldron et al., 2001). The stimulation of HeLa cells with Rotenone, a mitochondrial complex I inhibitor, which is an inducer of endogenous mitochondria-generated oxidative stress (mROS), increased both the translocation of PKD1 to the mitochondria (Fig. 1A) as well as the phosphorylation of PKD1 at its activation loop (Fig. 1B). The treatment of cells with hydrogen peroxide served as a positive control. H2O2 is itself a type of reactive oxygen species and has also been shown to induce further release of ROS at the mitochondria, as well as to induce the translocation of PKD1 to mitochondria (Storz, 2007; Storz et al., 2005b).

Fig. 1.

Translocation and activation of PKD1 by mitochondrial oxidative stress. (A) HeLa cells were transfected with pDsRed2-Mito and HA-tagged PKD1 and stimulated with Rotenone (20 μM, 1 hour) or left untreated (control). Localization of PKD1 and mitochondria was analyzed by confocal microscopy. (B) HeLa cells were transfected with wild-type PKD1 and stimulated with H2O2 (10 mM, 10 minutes) or Rotenone (20 μM, 1 hour). PKD1 was immunoprecipitated, samples were separated on SDS-PAGE and transferred to nitrocellulose. Samples were analyzed by western blotting with α-pS738/S742 (phosphorylated activation loop). Total PKD1 was detected by stripping and re-probing for PKD1. All experiments were performed three times with similar results.

Oxidative stress induces DAG formation at the mitochondria

PKD1 localizes to the mitochondria in response to mitochondrial oxidative stress (Storz et al., 2005b); however, the underlying mechanisms have not yet been determined. PKD1 has two C1 domains, which both bind to diacylglycerol, serving as membrane anchors. Binding of DAG is a required step in the PKD1 activation mechanisms at cellular compartments such as the Golgi complex or plasma membrane (Baron and Malhotra, 2002; Rey et al., 2001). To evaluate the possibility of DAG binding as the means of PKD1 localization to the mitochondria, we first determined whether ROS lead to mitochondrial DAG formation. Therefore, we used a previously described FRET-based mitochondria-localized reporter for DAG (Sato et al., 2006). Stimulation of cells with oxidative stress showed a rapid, but sustained, formation of DAG at the mitochondria, as indicated by a decrease in CFP intensity and subsequent increase in YFP intensity of the reporter (Fig. 2A). Mitochondrial DAG formation over time is indicated by the CFP/YFP intensity ratio, with decreased ratio value when DAG is formed (Fig. 2B). We were next interested in defining both the role of mitochondrial DAG in the activation of PKD1 and the source of DAG formation.

Fig. 2.

Oxidative stress induces mitochondrial DAG formation. (A,B) HeLa cells were transfected with a mitochondrial DAG reporter (Daglas-mit1). Cells were stimulated over time with 10 mM H2O2. H2O2-induced mitochondrial DAG formation was analyzed by measuring the CFP/YFP intensity ratio using a confocal microscope. (A) Real-time imaging of DAG formation at the mitochondria. Shown are CFP and YFP mitochondrial signals at start time (T0) and end time (T30=870 seconds). The decrease of CFP signal and increase of YFP signal indicates FRET caused by mitochondrial DAG formation. (B) Quantification of the experiment by depicting the CFP/YFP intensity ratio. The arrow indicates the start time for H2O2 stimulation (T6=150 seconds). All experiments were performed several times and with similar results.

Functional C1 domains of PKD1 are required for oxidative-stress-mediated activation

In order to determine the role of DAG binding in ROS-mediated activation of PKD1, we analyzed a subset of PKD1 deletion and point mutants (Fig. 3A) for their capability to be activated by oxidative stress. A PKD1 mutant with a deletion of the first 321 amino acids was not activated by oxidative stress, as judged by its activation loop phosphorylations in response to ROS (Fig. 3B). Amino acids 1-321 of PKD1 contain a CRD (cysteine-rich domain) domain that comprises two C1 domains. In order to define the role of the DAG-binding domains C1a and C1b for PKD1 activation in oxidative stress signaling, we then compared a PKD1 mutant with a CRD deletion (PKD1.ΔCRD) with wild-type PKD1 for its capability to be activated by oxidative stress. We found that the deletion of both C1 domains leads to the loss of PKD1 activation in response to ROS (Fig. 3C). To specifically demonstrate a role of DAG-binding capacity in ROS-mediated PKD1 activation, we then blocked PKD1-mediated DAG binding by using point mutations in C1a (PKD1.P157G mutant), C1b (PKD1.P281G mutant) or both (PKD1.P157G.P281G mutant). Proline to glycine mutations at these sites result in the complete loss of DAG and phorbol ester binding (Iglesias et al., 1998a), and therefore should prevent activation of PKD1 if DAG binding is necessary for this process. To investigate whether DAG binding is a vital event upstream of PKD1 activation in oxidative stress signaling, activation loop phosphorylation of wild-type PKD1, PKD1.P157G, PKD1.P281G and PKD1.P157G.P281G was examined in response to H2O2 (Fig. 3D). All mutants, particularly PKD1.P157G.P281G, showed decreased activation loop phosphorylation compared with wild-type PKD1 in response to stimulation with H2O2, suggesting that DAG binding by both C1 domains is important for PKD1 activation by oxidative stress.

Fig. 3.

Functional N-terminal and C1 domains in PKD1 are required for its activation by oxidative stress. (A) Overview PKD1 structure and mutants. PKD1 consists of two C1 domains, C1a and C1b (CRD domain), a pleckstrin homology domain (PH) and a kinase domain (KD). PKD1 deletion and point mutants were generated to examine the importance of DAG binding on PKD1 activation in response to oxidative stress. PKD1.Δ1-321: PKD1 mutant with deletion of the first 321 amino acids of the N-terminal domain. PKD1.ΔCRD: PKD1 mutant with deletion of the region comprising C1a and C1b (CRD domain). PKD1.P157G: PKD1 point mutant with a P to G mutation that blocks DAG binding by the C1a domain, but leaves the tertiary structure of PKD1 intact. PKD1.P281G: PKD1 point mutant with a P to G mutation that blocks DAG binding by the C1b domain. (B) HeLa cells were transfected with vector control, wild-type PKD1 or PKD1.Δ1-321. Cells were stimulated with H2O2 (10 mM, 10 minutes) as indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blotting was performed with α-pS738/S742. Total PKD1 was detected by stripping and re-probing for PKD1. (C) HeLa cells were transfected with vector control, wild-type PKD1 or PKD1.ΔCRD. Cells were stimulated with H2O2 (10 mM, 10 minutes) as indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blotting was performed with α-pS738/S742. Total PKD1 was detected by stripping and re-probing for PKD1. (D) HeLa cells were transfected with vector control, wild-type PKD1, PKD1.P157G, PKD1.P281G or PKD1.P157G.P281G. Cells were stimulated with H2O2 (10 mM, 10 minutes) as indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blotting was performed with α-pS738/S742. Total PKD1 was detected by stripping and re-probing for PKD1. All experiments were performed at least three times and obtained similar results.

DAG-binding capacity of PKD1 is required for mROS-mediated PKD1 activation and mitochondrial localization

We then used Rotenone as a specific inducer of mitochondrial reactive oxygen species to determine the role of DAG binding for the mitochondrial localization of PKD1. By comparing wild-type PKD1 with a PKD1.P157G.P281G mutant, we found that the loss of DAG binding capacity led to a loss of mitochondrial translocation of PKD1 in response to mROS generation (Fig. 4A-D). Similar to the results obtained for hydrogen peroxide in Fig. 4A, this also directly translated to a loss of Rotenone-mediated activation of PKD1 (Fig. 4E). Our data suggest both that mitochondrial DAG serves an anchor for PKD1 and that DAG binding is required for PKD1 activation by mitochondrial oxidative stress.

Fig. 4.

Mitochondrial localization and activation of PKD1 is dependent on DAG. (A-D) HeLa cells were transfected with GFP-tagged wild-type PKD1 or GFP-PKD1.P157G.P281G. Cells were stimulated with Rotenone (20 μM, 60 minutes) where indicated. GFP-PKD1 localization was detected using confocal microscopy. (E) Cells were transfected with vector control, HA-tagged wild-type PKD1 or PKD1.P157G.P281G. Cells were stimulated with Rotenone (20 μM, 60 minutes) where indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blots were performed against active PKD1 by staining for activation loop phosphorylations (α-pS738/742). Total PKD1 was detected by stripping and re-probing for PKD1. All experiments were performed three times and similar results were obtained.

Mutation of DAG-binding sites blocks pY95-mediated activation of PKD1

When PKD1 is activated in response to oxidative stress, it undergoes several Src-mediated phosphorylation events (Storz and Toker, 2003). The direct phosphorylation of PKD1 at Y95 by Src is crucial in creating a consensus binding motif matching that of a binding site for the C2 domain of PKCδ (Doppler and Storz, 2007). Phosphorylation of Y95 is necessary for allowing PKCδ binding and subsequent PKCδ-mediated phosphorylation of the PKD1 activation loop S738/742 to fully activate PKD1 (Doppler and Storz, 2007). To determine whether DAG binding of PKD1 is necessary for tyrosine phosphorylation to occur at Y95, we examined whether this phosphorylation event was abolished by a PKD1.P157G.P281G mutant. In HeLa cells, the PKD1.P157G.P281G mutant showed decreased Y95 phosphorylation in response to stimulation with hydrogen peroxide when compared with wild-type PKD1 (Fig. 5A). This correlates with the decrease seen in activation loop phosphorylation (Fig. 3D) and also indicates that DAG binding regulates early events in H2O2-mediated PKD1 activation. In a similar experiment, PKD1 activation was induced by overexpression of constitutively active Src, the upstream kinase that phosphorylates Y95. Overexpression of Src induced Y95 and activation loop phosphorylation in wild-type PKD1, which were both attenuated in the PKD1.P157G.P281G mutant (Fig. 5B). These results demonstrate that PKD1 binding to DAG occurs ahead of Src-mediated phosphorylation events.

Fig. 5.

Mutation of DAG-binding sites blocks tyrosine phosphorylation of PKD1. (A) Cells were transfected with vector control, wild-type PKD1 or PKD1.P157G.P281G, and were stimulated with H2O2 (10 mM, 10 minutes) where indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blots were performed against PKD1 phosphorylated at Y95 (α-pY95). Total PKD1 was detected by stripping and re-probing for PKD1. (B) Cells were co-transfected with vector control, PKD1 or PKD1.P157G.P281G and either empty vector or constitutively active Src (Src.Y527F). PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blots were performed with α-pY95, re-probed with α-pS738/S742 and α-PKD1. Immunoblotting was performed against Src (α-Src) to control overexpression. All experiments were performed three times and similar results were obtained.

Oxidative-stress-mediated activation of PKD1 is mediated via PLD1

We next identified the source for DAG in oxidative-stress-mediated PKD1 activation. Several lipases within cells have been implicated in the activation of PKD1, most prominent PLCγ and PLD (Baron and Malhotra, 2002; Kam and Exton, 2004; Wood et al., 2005; Zugaza et al., 1997). HeLa cells express PLCγ as well as PLD1, but not PLD2 (Fig. 6A). Both PLCγ and PLD1, when overexpressed can lead to PKD1 activation (Fig. 6B). However, PKD1 showed a marked increase in activation loop phosphorylation in response to overexpression of PLD1, whereas the effect of PLCγ was marginal (Fig. 6B). This points to PLD1 as being the likely source of DAG production that leads to PKD1 activation.

PLC hydrolyses PtdIns(4,5)P2 to DAG and Ins(1,4,5)P3, whereas PLD catalyses the hydrolysis of PC to PA and free choline. PA can then be converted into DAG, and lysophosphatidic acid (LPA) via PA-phosphohydrolases (Fig. 6C). To confirm the source of DAG, which participates in the activation of PKD1 by oxidative stress, specific inhibitors of PLCγ and PLD activity were used. Pre-incubation of cells with U73122, a specific PLC inhibitor (Smith et al., 1990), had no effect on PKD1 activation loop phosphorylation induced after stimulation with H2O2 (Fig. 6D). This indicates PLC is not involved in the pathway leading to PKD1 activation in response to ROS. In order to determine whether PKD1 is activated via PLD1 in response to oxidative stress, we used two inhibitors of PLD-mediated DAG formation (1-butanol and propranolol). 1-butanol is an established inhibitor of PLD-mediated phosphatidic acid (PA) production and can be used to examine the requirement of PLD for signaling processes (Baron and Malhotra, 2002; Kam and Exton, 2004). Increased generation of PA then leads to subsequent DAG formation by phosphatidic acid phosphohydrolases (Fig. 6C), a process that can be inhibited by propranolol (Kam and Exton, 2004). We found that both 1-butanol and propranolol (Fig. 6E,F) inhibited oxidative-stress-mediated activation of PKD1, suggesting a role for PLD1 in this process.

Fig. 6.

PLD activity is required for oxidative-stress-mediated activation of PKD1. (A) HeLa cells express PLD1 and PLCγ. Cells were lysed 24 hours after plating and proteins were resolved by SDS-PAGE. Immunoblotting was performed against PLD1, PLD2 and PLCγ. Equal loading was controlled by immunoblotting against actin (α-Actin). (B) PLCγ and PLD1 activate PKD1. Cells were co-transfected with PKD1 and control vector, PLCγ or PLD1. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blotting was performed with α-pS738/S742, and the blot re-probed with α-PKD1 to control expression. (C) Inhibition of pathways of DAG production. Abbreviations: DAG, diacylglycerol; DGK, diacylgylcerol kinase; IP3, inositol 1,4,5-triphosphate; LPA, lysophosphatidic acid; LPAAT, lysophosphatidic acid acyltransferase; PA, phosphatidic acid; PAP, phosphatidic acid phosphohydrolase; PI 4,5P2, phosphatidylinositol 4,5-bisphosphate; PLA, phospholipase A; PLC, phospholipase C; PLD, phospholipase D. (D) Inhibition of PLC by U73122 has no effect on PKD1 activation in response to oxidative stress. Cells were transfected with PKD1, and either incubated with U73122 (5 μM, 30 minutes) or vehicle control. Cells were stimulated with H2O2 (10 mM, 10 minutes) where indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blotting was performed with α-pS738/S742, and the blot re-probed with α-PKD1 to control expression. (E,F) Inhibition of PLD blocks PKD1 activation in response to oxidative stress. Cells were transfected with PKD1, and either incubated with 1-butanol (0.2%, 10 minutes; E), propranolol (250 μM, 30 minutes; F) or left untreated. Cells were stimulated with H2O2 (10 mM, 10 minutes) where indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blots were performed with α-pS738/742, and the blot re-probed with α-PKD1 to control expression. Experiments were performed three times and similar results were obtained each time.

A catalytically inactive PLD1 mutant blocks PKD1 activation

Next, we compared PKD1 activity in response to Rotenone or to hydrogen peroxide after its co-expression with PLD1 or the lipase-inactive mutant (Sung et al., 1997) (Fig. 7A,B). Activation loop phosphorylation of PKD1 was greatly increased in response to mROS induced by Rotenone or to H2O2 in cells transfected with wild-type PLD1. This effect was attenuated by the expression of a catalytically inactive PLD1.K898R mutant. This indicates that PKD1 is activated via PLD1 in response to oxidative stress and that this process involves enzymatic activity of PLD1. We further tested whether PLD1 is necessary for the activation of other PKD isoenzymes, such as PKD2 and PKD3/PKCν. We have previously found that only PKD1 and PKD2 are responsive to oxidative stress (Doppler and Storz, 2007). The co-expression of PKD2 and PKD3 with either wild-type PLD1 or catalytically inactive PLD1, followed by induction of oxidative stress with hydrogen peroxide revealed that PKD1 and PKD2 are activated via PLD1. PKD3 is unresponsive to oxidative stress and PLD1 (Fig. 7B,C).

PLD1 activity is required for PKD1 localization to the mitochondria

Next, we determined whether DAG formation by PLD1 is necessary for localization of PKD1 to the mitochondria. Therefore, we treated cells with inhibitors of PLD-mediated DAG formation, induced mROS with Rotenone and then analyzed the localization of PKD1. 1-Butanol and propranolol both inhibited Rotenone-mediated mitochondrial localization, whereas the inhibition of PLC with U73122 had no effect (Fig. 8A-P). We also compared PKD1 localization at the mitochondria after expression of wild-type or lipase-inactive PLD1 and induction of mROS with Rotenone (Fig. 8A-L). The overexpression of wild-type PLD1 already increased localization of PKD1 to the mitochondria in untreated cells (Fig. 9E), whereas a lipase-inactive PLD1 completely blocked mitochondrial localization in response to mROS (Fig. 9K). This illustrates DAG formation via PLD1 is necessary for the localization of PKD1 to the mitochondria in response to ROS.

Fig. 7.

Lipase-deficient PLD1 blocks PKD1 activation in response to ROS. (A,B) Cells were co-transfected with PKD1 and vector control, wild-type PLD1 or PLD1.K898R. Cells were stimulated with Rotenone (20 μM, 1 hour; A) or H2O2 (10 mM, 10 minutes; B) where indicated. PKD1 was immunoprecipitated (α-HA) and resolved by SDS-PAGE. Western blots were performed with α-pS738/S742. Total PKD1 was detected by stripping and re-probing for PKD1. (C) Cells were co-transfected FLAG-tagged PKD2 or GST-tagged PKD3 with vector control, wild-type PLD1 or PLD1.K898R. Cells were stimulated with H2O2 (10 mM, 10 minutes) where indicated. PKD2 (α-FLAG) or PKD3 (α-GST) was immunoprecipitated and resolved by SDS-PAGE. Western blots were performed with α-pS738/S742. Total PKD1 was detected by stripping and re-probing for PKD1. All experiments were performed three times and similar results were obtained.

Fig. 8.

Inhibition of PLD1-mediated DAG formation regulates the localization of PKD1 to the mitochondria. (A-P) Cells were seeded on eight-well μ-slides and transfected with GFP-tagged PKD1 and pDsRED2-Mito (mitochondrial marker). Cells were either treated with solvent or treated with U73122 (5 μM, 10 minutes), 1-butanol (0.2%, 10 minutes) or with propranolol (250 μM, 10 minutes) and then stimulated with Rotenone (20 μM, 1 hour). After stimulation, the cells were fixed and analyzed. The experiment was performed three times and similar results were obtained.

Inhibition of DAG formation or PLD1 blocks the mitochondrial localization of a PKD1.Y463E mutant

To obtain a clearer picture of whether DAG formation by PLD1 is necessary for localization of PKD1 to the mitochondria in oxidative stress signaling, we investigated whether propranolol could block localization of a PKD1 mutant (PKD1.Y463E). This mutation simulates a phosphorylation, leading to an open conformation of the kinase, rendering the mutant constitutively activated via the oxidative stress signaling pathway (Storz et al., 2005b). PKD1.Y463E strongly colocalizes with the mitochondria in HeLa cells (Fig. 10G-I). However, this co-localization is markedly reduced by treatment of cells with propranolol (Fig. 10J-L) as well as co-expression of a lipase-dead PLD1 (Fig. 10M-P). This shows that both the phosphorylation of PKD1 at Y463, as well as mitochondrial DAG formation via PLD1 is necessary for the localization of PKD1 to the mitochondria.

Src-mediated phosphorylation at Y463 regulates PKD1 localization to the mitochondria

We have shown before that Src induces the phosphorylation of PKD1 at Y463, and that this is an initial step in ROS-mediated PKD1 activation as it leads to an open conformation of the kinase (Storz et al., 2005b). Therefore, we performed additional experiments to determine whether Src regulates mitochondrial localization of PKD. We first expressed constitutively active Src (Src.Y527F) and found that active Src induced the mitochondrial localization of PKD1 (Fig. 11A). Additionally, the inhibition of Src with the SFK family inhibitor PP2 blocked the Rotenone-induced localization of PKD1 to the mitochondria (Fig. 11B), further showing that Src is involved in the mitochondrial localization of PKD1.

Fig. 9.

PLD1 regulates the localization of PKD1 to the mitochondria. (A-L) Cells were seeded on eight-well μ-slides and transfected with GFP-tagged PKD1, pDsRED2-Mito (mitochondrial marker) and control vector, wild-type PLD1 or PLD1.KR, and then stimulated with Rotenone (20 μM, 1 hour). After stimulation, the cells were fixed and analyzed. The experiments were performed three times and similar results were obtained.

Discussion

Protein kinase D1 is increasingly gaining interest as a potential therapy target in oxidative-stress-mediated diseases, owing to its function as a ROS sensor that relays oxidative stress to downstream signaling (Storz, 2007). In this capacity, it has been shown to activate a mitochondrion-to-nucleus signaling pathway, leading to the activation of NF-κB, and expression of genes promoting cellular detoxification and survival (Storz et al., 2005a; Storz et al., 2005b). The intracellular location of PKD1 has been shown to be highly important in determining its function in many cell types and processes (Marklund et al., 2003; Mullin et al., 2006). To perform as a sensor for mROS, PKD1 must be located at the mitochondria, as mitochondria-generated reactive oxygen species are quickly broken down (Storz, 2007). In immunofluorescence studies, wild-type PKD1 localized to the mitochondria in response to Rotenone, an inducer of mitochondrial ROS (Fig. 1A). This is consistent with PKD1 translocation to the mitochondria in response to H2O2, as previously reported (Storz et al., 2005b). Here, we show that PKD1 localizes to the mitochondria in response to oxidative stress by binding to mitochondrial DAG, and that this is an essential step necessary to activate PKD1 in order for it to participate in oxidative stress signaling. This is the first report describing a function for mitochondria-generated DAG. With phospholipase D1, we further determined the cellular source of mitochondrial DAG formation.

Previously described are the phosphorylation-dependent activation mechanisms of PKD1 in response to inducers of oxidative stress (Storz et al., 2005b). Activation requires phosphorylation at Y463 to relieve autoinhibition (Storz and Toker, 2003), phosphorylation at Y95 by Src to create a PKCδ binding site (Doppler and Storz, 2007) and subsequent phosphorylation of the activation loop serines (S738/S742 in human PKD1) by PKCδ for full activation (Storz et al., 2004). It is not known whether DAG binding was involved in PKD1-mediated oxidative stress signaling. We used a mitochondria-localized FRET-reporter for DAG to show that DAG can be formed at the mitochondria as a result of oxidative stress (Fig. 2). This was the first time that direct DAG formation at the mitochondria was measured in response to increased oxidative stress. Furthermore, we found that DAG contributes to the localization of PKD1 to the mitochondria in response to oxidative stress, as an essential step in activation. PKD1.P157G.P281G, a PKD1 point mutant that renders it incapable of DAG binding (Baron and Malhotra, 2002), showed significantly decreased tyrosine phosphorylation at Y95 and activation loop phosphorylation in response to inducers of ROS (Fig. 3D, Fig. 4E, Fig. 5A). The hypothesis that DAG binding is necessary for PKD1 activation was also supported by studies with Src, the kinase responsible for phosphorylating Y95 in response to oxidative stress. Active Src increased phosphorylation of Y95, which promotes subsequent phosphorylation of the PKD1 activation loop. Both phosphorylation events were attenuated in the PKD1.P157G.P281G mutant (Fig. 5B). This clearly indicates that DAG binding is necessary and occurs prior to both key phosphorylation events (pY95 and pS738/742) that mediate PKD1 activation in response to ROS. The importance of functional C1 domains of PKD1 is also demonstrated in Fig. 4A-D, where we show that point mutations in the DAG-binding sites of PKD1 block Rotenone-mediated translocation of PKD1 to the mitochondria.

Fig. 10.

PLD1 regulates the localization of Y463-phosphorylated PKD1 to the mitochondria. (A-L) Cells were transfected with HA-tagged PKD1 or PKD1.Y463E mutant and pDsRED2-Mito (mitochondrial marker). Twenty-four hours after transfection, the cells were seeded on glass coverslips. Cells were either left untreated (A-C,G-I) or stimulated with propranolol (250 μM, 30 minutes) to block DAG formation from PA (D-F,J-L). Immunofluorescence samples were stained and processed as described in the Materials and Methods. Samples were analyzed by confocal microscopy. Green, PKD1 stained with α-P8 (α-PKD1 antibody); red, pDsRED2-Mito (mitochondria); blue, DAPI (nuclei). (M-P) Cells were seeded on eight-well μ-slides and transfected with GFP-tagged PKD1.Y463E, pDsRED2-Mito (mitochondrial marker) and control vector or PLD1.KR. Twenty-four hours after transfection, the cells were fixed and analyzed. The experiments were performed three times and similar results were obtained.

Fig. 11.

Src regulates the localization of PKD1 to the mitochondria. (A) Cells were seeded on eight-well μ-slides and transfected with GFP-tagged PKD1, pDsRED2-Mito (mitochondrial marker) and vector control or Src.Y527F. Twenty-four hours after transfection, the cells were fixed and analyzed by immunofluorescence. (B1-B8) Cells were seeded on eight-well μ-slides and transfected with GFP-tagged PKD1 and pDsRED2-Mito (mitochondrial marker). Twenty-four hours after transfection, the cells were pre-treated with PP2 (10 μM, 1 hour) and then stimulated with Rotenone (20 μM, 1 hour) as indicated. Cells were then fixed and analyzed by immunofluorescence. All experiments were performed three times and similar results were obtained.

As these results clearly suggest, DAG binding is the method used by PKD1 to localize to the mitochondria, our next aim was to uncover the source of increased DAG production under these conditions. DAG is present only at low levels in most membranes; however, in response to various stimuli, it rapidly accumulates owing to the action of PLC and PLD enzymes (Wakelam, 1998). PLD1 and PLCγ were both present in HeLa cells as potential sources of DAG production (Fig. 6A). However, when PLD1 and PLCγ were overexpressed, PLCγ had only a marginal effect on PKD1 activation in comparison with PLD1, which strongly induced PKD1 activation loop phosphorylation (Fig. 6B). Of the PLC enzymes, only PLCγ was examined, as PLCγ subtypes alone are activated in response to oxidative stress where it has a role in cell survival (as has PKD1) (Blake et al., 1993; Schieven et al., 1993). For example, PLCγ-deficient mouse embryonic fibroblasts showed greatly increased sensitivity to H2O2-induced cell death (Wang et al., 2001). PLCγ activation in response to H2O2 involves growth factor receptor activation (Banan et al., 2001; Wang et al., 2001). In HeLa cells, in which PKD1 activation by oxidative stress was established, and in which we have executed our studies, PKD1 is not activated by various growth factors (unpublished results). This is consistent with our results that PLCγ is not involved in PKD1 activation. Further investigation demonstrated that the use of a specific PLC inhibitor (U73122) had no effect on the activation of PKD1 in response to H2O2, effectively eliminating PLC as the source of DAG in this pathway (Fig. 6D). Similar to PLCγ, PLD1 is activated by oxidative stress in various cell lines (Hagele et al., 2007; Varadharaj et al., 2006). PLD is a ubiquitously expressed enzyme that hydrolyzes phosphatidylcholine and increases cellular phosphatidic acid levels (Exton, 2002). ROS-induced PLD1 activation involves PKC and tyrosine kinases in various mammalian cell systems (Banno and Nozawa, 2003; Min et al., 1998). Our results with the PLD inhibitors 1-butanol and propranolol clearly indicate that PLD1 mediates the ROS-induced increase in DAG, which facilitates PKD1 localization to the mitochondria and its activation (Fig. 6E; Fig. 6F; Fig. 8).

Fig. 12.

Mechanisms of sequential PKD1 activation in response to oxidative stress. Oxidative-stress-mediated PLD1 activation leads to an increase in PA and subsequent DAG formation. The initial phosphorylation of PKD1 at Y463 via Src (1) allows the recruitment of PKD1 to the mitochondria via DAG binding (2). Binding leads to a change in conformation, allowing further Src-mediated phosphorylations. Src directly phosphorylates Y95, creating a C2 domain-binding motif (3), to which PKCδ binds and phosphorylates S738/742 to fully activate PKD1 (4). PKD1 can then activate downstream targets to facilitate mitochondria-to-nucleus signaling.

PLD1 was described to be located at most cellular organelles, including the perinuclear region, mitochondria and the Golgi apparatus (Freyberg et al., 2001; Jin et al., 2006; Liscovitch et al., 1999). Because an additional mitochondrial protein (mitoPLD) with PLD enzyme activity has been recently described (Choi et al., 2006). We also compared mitoPLD and PLD1 in cells stimulated with Rotenone and found that mitoPLD1 had no effect on PKD1 activity (suppplementary material Fig. S1), suggesting that PLD1 is indeed responsible for oxidative-stress-mediated DAG formation at the mitochondria. Further support of the role of PLD1 in PKD1 activation and mitochondrial localization came from studies using a catalytically inactive PLD1 (PLD1.K898R). This mutant blocked the mitochondrial localization (Fig. 9) and the activation of PKD1 in response to oxidative stress, as indicated by the decreased phosphorylation of the activation loop serines S738 and S742 (Fig. 7A,B). Therefore, it is reasonable to conclude that PLD1 enzymatic activity is necessary for the mitochondrial localization and activation of PKD1 under oxidative stress.

The PLD1 product phosphatidic acid is an important second messenger implicated in signaling through various pathways, including activation of PI4P 5-kinase (Jenkins et al., 1994) and mTOR (Fang et al., 2001), and binding to Raf-1 kinase (Rizzo et al., 1999). Until now, it has been viewed that monosaturated and saturated diacylglycerols formed from PA have no signaling function (Pettitt et al., 1997). However, our results strongly indicate this is not the case in oxidative stress signaling. This is a very interesting prospect as it raises issues about the relative contributions of DAG and PA signaling in different cell systems and in response to specific stimuli. Future studies will elucidate whether PLD1 in this pathway generates PA at the mitochondria or whether its product PA diffuses to the mitochondria and is transformed to DAG at the mitochondria by a mitochondrial enzyme.

Mitochondria are the major source of ROS within the cell, as well as being a major target for ROS damage (Balaban et al., 2005). Therefore, PKD1 is gaining much interest as a potential therapy target, owing to its role as an oxidative stress sensor. Not only does our current work provide further insight into the sequential events necessary for ROS-mediated PKD1 localization and activation at the mitochondria (Fig. 12), it also for the first time demonstrates a signaling function for mitochondrially generated DAG as a competent second messenger at the mitochondria.

Materials and Methods

Cell culture, antibodies and reagents

The HeLa cell line was purchased from American Type Culture Collection and maintained in high-glucose Dulbecco's modified Eagle's medium (MediaTech, Herndon, VA) supplemented with 10% fetal bovine serum (Biosource, Camarillo, CA). All transient transfections were carried out 1 day after plating using TransIT HeLa Monster reagent (Mirus, Madison, WI) or Superfect (QIAGEN, Valencia, CA) according to manufacturer's instructions. The anti-PKD1 antibody was from Santa Cruz (Santa Cruz, CA). The anti-PLD1 and -PLD2 antibodies were from Sigma-Aldrich (St Louis, MO). The anti-PLCγ antibody was a kind gift from Dr Alex Toker (Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA). The anti-pS738/S742 antibody recognizing the phosphorylated PKD1 activation loop was purchased from Biosource. The rabbit polyclonal pY95-PKD1 antibody has been described elsewhere (Doppler and Storz, 2007). The secondary HRP-linked anti-mouse or anti-rabbit antibodies were from Roche (Indianapolis, IN). The anti-Src antibody was from Upstate (Charlottesville, VA). Anti-HA was from Boehringer (Ridgefield, CT). The P8 antibody (anti-PKD1) was a kind gift from Dr Klaus Pfizenmaier (University of Stuttgart, Germany). H2O2 was purchased from Fisher Scientific (Suanee, GA) and Rotenone was from Calbiochem (La Jolla, CA). U73122 was bought from Biosource and propranolol was purchased from Calbiochem.

Expression plasmids

The cloning of an expression vector for N-terminal HA-tagged human PKD1 has been described previously (Storz et al., 2003; Storz and Toker, 2003). Expression plasmids for PKD1 mutants are based on this plasmid. Mutagenesis was carried out by PCR using QuikChange (Stratagene, La Jolla, CA) according to the manufacturer's instructions. Primers used for PKD1.P157G were 5′-GTTCATTCATACAGAGCTGGAGCTTTCTGTGATCACTGT-3′ and 5′-ACAGTGATCACAGAAAGCTCCAGCTCTGTATGAATGAAC-3′, and for PKD1.P281G were 5′-CACTCCTACACCCGGGGCACAGTGTGCCAGTAC-3′ and 5′-GTACTGGCACACTGTGCCCCGGGTGTAGGAGTG-3′. PLD1 expression plasmids were based on an N-terminal FLAG-tagged PLD1 derived from full-length non-tagged human PLD1 (obtained from Dr M. Liscovitch) using the following primer pairs: 5′-GCGGCGGCCGCATGGACTATAAGGACGATGATGACAAATCACTGAAAAACGAGCCACGGGTA-3′ and 5′-CGCTCTAGATTAAGTCCAAACCTCCATGGGCAC-3′, and cloned into pcDNA4/TO/myc-His via NotI and XbaI. Mutagenesis with the primer pairs 5′-GAGCTTATCTATGTCCACAGCAGGTTGTTAATTGCTGATG-3′ and 5′-CATCAGCAATTAACAACCTGCTGTGGACATAGATAAGCTC-3′ was carried out by PCR using QuikChange to generate the PLD1.K898R mutant. All constructs were verified by DNA sequencing. The mitochondria marker pDsRed2-Mito was from Clontech, Mountain View, CA. All other plasmids have been described previously (Storz et al., 2003; Storz et al., 2004).

Immunoblotting and immunoprecipitation

Cells were plated in 60 mm dishes at 6.5×105 cells/dish. Cells were transfected using TransIT HeLa Monster reagent. Twenty-four hours after transfection, cells were stimulated as indicated, then washed twice with ice-cold phosphate-buffered saline [PBS; 140 mM NaCl, 2.7 mM KCl, 8 mM Na2HPO4, 1.5 mM KH2PO4 (pH 7.2)] and lysed with lysis buffer [50 mM Tris-HCl pH 7.4, 1% Triton X-100, 150 mM NaCl, 5 mM EDTA (pH 7.4)] plus Protease Inhibitor Cocktail (Sigma-Aldrich, St Louis, MO). Lysates were vortexed, incubated on ice for 30 minutes and centrifuged (10,000 g, 15 minutes, 4°C). Lysates were used for immunoblot analysis by adding an equal volume of 2× Laemmli buffer, or proteins of interest were immunoprecipitated by a 1 hour incubation with a specific antibody (2 μg) followed by a 30 minute incubation with protein G-Sepharose (Amersham Biosciences). Immune complexes were washed three times with TBS (50 mM Tris-HCl pH 7.4, 150 mM NaCl). The last wash was aspirated and 20 μl each of TBS and Laemmli buffer was added. Immunoprecipitates or lysates were resolved by SDS-PAGE. After SDS-PAGE, proteins were transferred to nitrocellulose membranes and visualized by immunostaining and enhanced chemiluminescence using Supersignal reagents (Pierce, Rockford, IL).

Analysis of DAG generation

Daglas-mit1, the reporter for mitochondrial DAG, has been described previously (Sato et al., 2006). The confirmed mitochondrial localization of Daglas-mit1 allows the spatial analysis of DAG dynamics at the mitochondria. Cells were transfected with Daglas-mit1 and after 24 hours reseeded on 3 cm glass bottom dishes coated with Collagen I (22.5 mg/ml 1 hour) in L15 media with 10% FCS and glutamate without Phenol Red. Measurement of mitochondrial DAG-induced FRET occurred in a live-imaging P-chamber at 37°C. Images were taken every 30 seconds and the stimulation with 10 mM H2O2 was started after 150 seconds (T6). Nine random control and 9 H2O2-treated cells each experiment were analyzed. Samples were examined using a LSM 510META confocal laser scanning microscope (Zeiss, Jena, Germany) with a Plan-Apochromat 63× oil immersion objective. The LSM settings were: single track CFP (CH2)/YFP (CH3); CFP BP 470-500/YFP LP560 Ex; and 458 5% (all PH MAX of 1000). Other settings were: gain/offset linear range (range indicator); scan speed 8/line average 4 (best signal/noise ratio); ROIs for single cells: ratio calculation CFP (CH2)/YFP(CH3) for time series using 8 bit coding and intensity range settings.

Immunofluorescence

Cells were seeded and transfected in μ-slides (Integrated Biodiagnostics, Martinsried, Germany) and 24 hours after transfection were stimulated, fixed and analyzed. For confocal microscopy, cells were transfected using TransIT HeLa Monster reagent and 24 hours after transfection plated on glass coverslips at a density of 3.5×105 cells/well in a 24-well plate. The next day, cells were transfected with 1 μg plasmid of interest using TransIT HeLa Monster reagent. Twenty-four hours after transfection, cells were plated on coverslips at 70,000 cells/well; cells were stimulated as indicated. Samples were washed twice with ice-cold PBS and fixed in 3.5% paraformaldehyde (15 minutes, 37°C). After permeabilization (0.1% Triton X-100, 2-10 minutes), cells were blocked with PBS containing 3% bovine serum albumin (BSA) and 0.05% Tween-20 for 30 minutes at room temperature. The coverslips were incubated with primary antibody solution diluted 1:2000 in PBS-BSA (3% BSA in PBS) overnight at 4°C. Cells were washed five times in PBS, then incubated with the secondary antibody diluted 1:5000 in PBS-BSA (donkey anti-rat IgG Cy3-conjugated from Invitrogen, Carlsbad, CA) also containing DAPI (1:2500) for 2 hours. After five washes in PBS, coverslips were mounted in Gel Mount from Biomeda (Foster City, CA) and examined. Samples were examined using an Olympus IX71 fluorescence microscope or a LSM 510META confocal laser scanning microscope (Zeiss, Jena, Germany) with a Plan-Apochromat 63×/1.4 DIC oil immersion objective. Images were processed using NIH ImageJ.

Footnotes

  • We thank Tim Eiseler for help with the FRET analysis. We also thank Mordechai Liscovitch (Weizmann Institute, Israel) for a PLD1 expression plasmid, Michael A. Frohman (SUNY Stony Brook, NY) for a mitoPLD expression plasmid and Alex Toker (Harvard Medical School, MA) for the PLCγ antibody. This work was in part sponsored by funds from the Mayo Foundation and the Mayo Comprehensive Cancer Center, as well as by a R21 from the NCI (CA135102) to P.S. and a DFG grant (DFG HA3557/2-1) to A.H. The authors declare that they have no competing financial interests. Deposited in PMC for release after 12 months.

  • Supplementary material available online at http://jcs.biologists.org/cgi/content/full/122/7/919/DC1

  • Accepted November 21, 2008.

References

View Abstract