|Home | About | Journals | Submit | Contact Us | Français|
Phospholipid-enriched membranes such as the plasma membrane can serve as direct regulators of kinase signaling. Pak1 is involved in growth factor signaling at the plasma membrane and its dysregulation is implicated in cancer. Pak1 adopts an autoinhibited conformation that is relieved upon binding to membrane-bound Rho GTPases Rac1 or Cdc42, but whether lipids also regulate Pak1 in vivo is unknown. We show here that phosphoinositides, particularly PIP2, potentiate Rho-GTPase-mediated Pak1 activity. A positively charged region of Pak1 binds to phosphoinositide-containing membranes, and this interaction is essential for membrane recruitment and activation of Pak1 in response to extracellular signals. Our results highlight an active role for lipids as allosteric regulators of Pak1 and suggest that Pak1 is a “coincidence detector” whose activation depends on GTPases present in phosphoinositide-rich membranes. These findings expand the role of phosphoinositides in kinase signaling and suggest how altered phosphoinositide metabolism may upregulate Pak1 activity in cancer cells.
Phosphoinositides are important spatial and temporal regulators of cellular signaling. Reversible phosphorylation of phosphatidylinositol generates seven distinct species with differing functions and subcellular localizations (Di Paolo and De Camilli, 2006). Phosphatidylinositol 4,5-bisphosphate (PIP2) and phosphatidylinositol 3,4,5-trisphosphate (PIP3) are localized predominantly to the plasma membrane where they mediate recruitment of Rho family GTPases including Rac and Cdc42 (Heo et al., 2006). PIP3 also directly regulates the recruitment and activity of several protein kinases in mammalian cells (Newton, 2009), while PIP2 has only been reported to promote membrane association of the G protein-coupled receptor kinases to facilitate substrate interactions (Pitcher et al., 1998).
Accumulating evidence suggests that phosphoinositides do not merely serve as passive membrane platforms for signaling but can play active roles as allosteric protein activators. PIP2, for example, binds to a basic region in the scaffolding protein N-WASP (neural Wiskott-Aldrich syndrome protein), triggering conformational changes that alter its biochemical activity (Papayannopoulos et al., 2005). Interestingly, basic regions are found at similar positions in a subset of Rho GTPase effectors including Pak1, suggesting that Pak1 might also be regulated by phosphoinositides.
Pak1 is a member of an evolutionarily conserved family of serine/threonine kinases that plays fundamental roles in cell morphology and proliferation (Arias-Romero and Chernoff, 2008). Group I Paks (Paks 1–3) are regulated by autoinhibition which is relieved upon binding to active Cdc42 or Rac1. Once activated, Paks undergo autophosphorylation at specific sites and then phosphorylate a variety of substrate proteins (Bokoch, 2003).
While Rho GTPases are well-characterized activators of Pak1, the role of other factors in activating Pak1 is less clear. A variety of lipid micelles have been shown to activate mammalian Pak1 in vitro (Bokoch et al., 1998) but the relevance of these findings to Pak1 regulation in vivo is unknown. Here we show that a basic region of Pak1 binds PIP2 and stimulates Pak1 activity in concert with Rac1. We further show that PIP2-binding is essential both for Pak1 activation and plasma membrane recruitment in response to extracellular signals. These results reveal a regulatory mechanism critical for Pak1 activation that may be relevant to other Rho GTPase effectors. Our findings thus expand the repertoire of mammalian kinases that are activated by phosphoinositides and establish Pak1 as a bona fide lipid-regulated kinase.
As an initial test to determine whether phosphoinositides can promote Pak1 activation, full-length recombinant Pak1 was incubated with liposomes of diverse compositions and Pak1 activation was assessed by substrate phosphorylation (Figure 1A) or by gel mobility shift of the kinase (Figure S1A) due to autophosphorylation (Manser et al., 1997). Pak1 incubated with active Rac1 served as a positive control (Figures 1A and S1A, last lanes). Incubation with most liposomes had little effect on Pak1 activity but phosphatidylcholine (PC):phosphatidylinositol (PI):PIP2 (molar ratio 48:48:4; hereafter “PIP2 liposomes”) stimulated Pak1 autophosphorylation and substrate phosphorylation comparable to that observed with Rac1. PIP2 appeared to be the most critical liposome component since PC:PI liposomes lacking PIP2 activated Pak1 only weakly. Other phosphoinositides tested showed little activity and, interestingly, PIP2 present in other lipid contexts (PC:phosphatidylethanolamine (PE); 4:48:48) was mostly inactive. We then investigated whether Pak1 activity is potentiated in the presence of various concentrations of both PIP2 liposomes and Rac1. At concentrations at which each activator was weakly active on its own (80 µM PIP2 liposomes, 0.1 µM Rac1), their combination enhanced both substrate phosphorylation and Pak1 autophosphorylation (Figures 1B and S1B). Since active Rac1 is normally tethered to phosphoinositide-rich membranes by a prenyl moiety, the degree of potentiation observed here with unmodified, soluble Rac1 likely underestimates the true potentiation in vivo. Thus, PIP2 promotes Pak1 activation in a lipid context-dependent manner that is cooperative with Rac1.
The Pak1 autoregulatory domain encompasses a region enriched in basic residues followed by the canonical GTPase-binding domain. This arrangement is conserved in Pak1 orthologues in metazoans (Figure 1C) and yeast (Takahashi and Pryciak, 2007). To determine whether Pak1 activation by PIP2 liposomes is mediated by this basic region, we generated mutants of Pak1 in which either four (4T Pak1) or eight (8T Pak1) lysine residues in the basic region were mutated to the polar but uncharged residue threonine (Figure 1C). The wild-type and mutant proteins migrated similarly by SDS-PAGE and in gel filtration chromatography, (Figure S1C), demonstrating that the mutations do not disrupt the ability of Pak1 to form homodimers. In an in vitro kinase assay, activation of wild-type Pak1 was dramatically enhanced by Rac1 and PIP2 liposomes as shown by the molecular weight increase of Pak1 on a silver-stained gel (Figure 1D, upper panel) as well as by western blotting using phospho-specific antibodies against Pak1 autophosphorylation sites Ser144 and Ser199 (Figure 1D, lower panels). This shift was indeed due to phosphorylation as phosphatase treatment of the Rac1+PIP2 reaction collapsed the upper bands (lane 5). In contrast, 4T Pak1 and 8T Pak1 showed progressively weaker responses to PIP2 liposomes (compare lane 4 with 9 & 13), although activation by Rac1 remained intact (lanes 7 & 11). Titrations of wild-type (WT) and 8T Pak1 with Rac1 showed quantitatively similar stimulation of Pak1 kinase activity, indicating that the 8T mutation does not affect activation by Rac1 (Figure 1E). Therefore, we conclude that the basic region of Pak1 is necessary only for activation by PIP2.
To determine whether the basic region is sufficient to mediate PIP2 binding, we purified the N-terminal domain of WT Pak1, 4T Pak1, and 8T Pak1 as GST-fusion proteins and monitored their ability to co-sediment with PIP2 liposomes (Figure 1F). As a positive control, the pleckstrin homology domain of phospholipase Cδ (PH-PLCδ), bound robustly to PIP2 liposomes and was partially found in the pellet fraction, as was WT Pak1 and, to a lesser extent, the 4T Pak1 mutant. 8T Pak1 and GST alone did not co-sediment with PIP2 liposomes. This demonstrates that the N-terminus of Pak1 is sufficient for PIP2 binding and that this interaction requires the Pak1 basic region.
Whereas several Pak-related kinases in yeast bind phosphoinositides via pleckstrin homology domains (Sells et al., 1998; Wild et al., 2004), Saccharomyces cerevisiae Ste20p and mammalian N-WASP bind to membranes via electrostatic interactions that depend on the net charge but not the precise sequence of amino acids (Papayannopoulos et al., 2005; Takahashi and Pryciak, 2007). We therefore tested whether PIP2 binding could be restored to 8T Pak1 by the insertion of a series of eight lysine residues within the Pak1 autoregulatory region (8T+8K Pak1; Figure 1C). This insertion restored the ability of 8T Pak1 to bind PIP2 liposomes (Figure 1F), implying that Pak1 binding to PIP2 is mediated by electrostatic binding to the basic region.
We next tested if the ability of phosphoinositides to activate Pak1 correlated with their ability to bind the Pak1 basic region by liposome co-sedimentation. As expected, PIP2 liposomes (lane 9), and to a lesser extent PC:PI liposomes (lane 7), bound Pak1 (Figure 1G), consistent with their respective abilities to activate full-length Pak1 (Figure 1A). In addition, we observed weaker binding to PC:PE:PIP2 (lane 4) and PC:PS liposomes (lane 6), consistent with a charge-dependent interaction with acidic phospholipids. Interestingly, however, no binding was observed to PI(3)P- and PI(3,4,5)P3-containing liposomes, nor to liposomes lacking acidic phospholipids (PC:PE). Thus, liposome binding generally correlates with Pak1 activation. We conclude that the evolutionarily conserved basic region in Pak1 is necessary and sufficient for binding to PIP2 via a sequence-independent, electrostatic interaction.
To confirm the physiological relevance of phosphoinositide binding by Pak1 observed in vitro, we investigated Pak1 activation by PIP2 liposomes in cytoplasmic extracts of Xenopus laevis eggs. Stimulation of these extracts with PIP2 results in the activation of endogenous Rac1 and Cdc42 as well as the effector N-WASP (Pelish et al., 2006; Peterson et al., 2004). We used an established in-gel kinase assay (Ding et al., 1996) to monitor Pak activation under these conditions. Two Xenopus kinases migrating with the expected molecular weights of XPak1 and XPak2 (58 and 57 kDa, respectively) were activated in a time-dependent and liposome dose-dependent manner in extracts stimulated with PIP2 (Figure 2A). We used three approaches to confirm the identity of these kinases as XPaks (Figure S2A–C).
To determine whether PIP2-stimulated activation of Pak1 in Xenopus egg extracts requires Rho GTPases or whether PIP2 activates XPak1/2 directly, we used three distinct inhibitors of Rac1/Cdc42 signaling. Extracts were pre-incubated with either recombinant Rho guanine nucleotide dissociation inhibitor (RhoGDI, a negative regulator of both Rac1 and Cdc42; (Hoffman et al., 2000)), dominant negative Rac1 (RacN17), or a small-molecule inhibitor of Rac (NSC23766; (Gao et al., 2004)). The extracts were then stimulated with PIP2 liposomes and analyzed by in-gel kinase assay (Figure 2B) and western blotting with anti-phospho-Ser144 Pak1 antibody (Figure 2C). Strikingly, all three inhibitors blocked XPak activation but importantly, did not affect Rac-independent actin polymerization in these extracts (Figure S2D). These results indicate that Rho GTPases play an essential role in XPak activation in this system.
Although PIP2 acts upstream of Rac1/Cdc42 activation in Xenopus egg extract, we next assessed whether it might also play a second role in Pak1 activation by binding directly to Pak1. To test this, extracts were supplemented with WT Pak1, 4T Pak1, 8T Pak1, or 8T+8K Pak1 and then were either mock stimulated, treated with inert liposomes (PC:PE), or stimulated with PIP2 liposomes. Activation of endogenous XPak and the recombinant Pak1 proteins were monitored by anti-phospho-Ser144 and anti-phospho-Ser199 western blotting (Figure 2D, upper panels). Total amounts of exogenous Pak1 were detected using antibodies against mammalian Pak1 (lower panel). Whereas endogenous XPak and exogenous WT Pak1 became activated (Ser144-and Ser199-phosphorylated) in response to PIP2, 4T Pak1 was weakly activated and 8T Pak1 remained inactive. Remarkably, activation of 8T+8K Pak1 was restored relative to 8T Pak1. These results demonstrate an essential role of the Pak1 basic region in PIP2-mediated Pak1 activation. Taken together with the previous experiment, these findings indicate that activation of Pak1 requires binding to both a GTPase and PIP2.
Pak1 has been implicated in the formation of actin-rich plasma membrane ruffles in cells stimulated with phorbol myristate acetate (PMA) (Viaud and Peterson, 2009). We therefore investigated whether phosphoinositides play a role in Pak1 regulation during PMA-stimulated membrane ruffling. Confocal immunofluorescence microscopy showed that Pak1 is recruited to actin-rich ruffles in PMA-stimulated BSC-1 cells (Figure 3A). These ruffles are also enriched in PIP2, as revealed by the PIP2 marker GFP-PH-PLCδ (Figure 3B), suggesting that PIP2 might play a role in recruiting Pak1. Importantly, endogenous Pak1 localized to these PIP2-enriched ruffles as well (Figure S3A), suggesting that localization is not due to overexpression of the protein. We then transfected WT Pak1, 8T Pak1, and 8T+8K Pak1 and monitored their recruitment to ruffles. WT Pak1 was enriched in PMA-induced membrane ruffles whereas 8T Pak1 was not (Figure 3C, see also Figure S3B,C). Consistent with the functional rescue in vitro, 8T+8K Pak1 was enriched in ruffles, as was GFP fused to the N-terminal domain of Pak1 (GFP-Pak1-1-84; Figure 3D). These findings demonstrate that, in the context of live cells, phosphoinositide binding is essential for Pak1 recruitment to the plasma membrane.
Stimulation of cells with platelet-derived growth factor (PDGF) activates Rac1 and results in the formation of membrane ruffles that contain type I phosphatidylinositol phosphate kinase α (PIPKIα), an enzyme responsible for PIP2 synthesis (Doughman et al., 2003). PIPKIα catalytic activity is required for these actin-rich ruffles, indicating that PIP2 plays a role in their formation (Doughman et al., 2003). Pak1 is also activated by PDGF and contributes to Erk activation in NIH3T3 cells (Beeser et al., 2005), and Pak1 is recruited to PDGF-stimulated membrane ruffles and promotes their formation (Dharmawardhane et al., 1997; Zhou et al., 2006). Therefore, to assess the functional importance of the basic region of Pak1 in PDGF-stimulated Pak1 activation, we analyzed the kinase activity of wild-type and basic-region mutants of Pak1 in PDGF-stimulated NIH3T3 cells.
Cells transfected with plasmids encoding wild-type or basic-region mutants of Pak1 were stimulated with PDGF. Transfected Pak1 was immunoprecipitated from cell lysates and analyzed by in vitro kinase assay (Figure 4A). Anti-phospho-Erk and anti-total-Erk western blots of total cell lysates confirmed that PDGF stimulation was effective in all cases (Figure S4). Anti-myc western blots (myc-Pak1) and silver-stained SDS-PAGE of the immunoprecipitates demonstrated recovery of similar amounts of wild-type and mutant Pak1 (Figure 4A, top panels). Coomassie blue-staining of the lower portion of the gel documented that similar amounts of substrate (GST-Paktide) were used. As expected, wild-type Pak1 was robustly activated by PDGF, as evidenced by autophosphorylation of Pak1 (next to last panel), phosphorylation of the substrate GST-Paktide (last panel), and by reduced Pak1 mobility (Figure S4). In contrast, 8T Pak1 was not responsive to PDGF. Strikingly, the 8T+8K Pak1 mutant restored responsiveness to PDGF to levels approaching wild-type Pak1 (Figure 4B). This finding demonstrates that phosphoinositide-binding plays an essential role in Pak1 activation in response to growth factor stimulation of mammalian cells.
Pak1 and PIP2 both play roles in the regulation of the actin cytoskeleton, maintenance of focal contacts, and cytokinesis (van den Bout and Divecha, 2009), and our results suggest that the regulation of Pak1 by PIP2 may provide an underlying mechanism for these observations. While we cannot exclude that other phosphoinositides might also regulate Pak1 under physiological conditions, we find that PIP2, the most abundant phosphoinositide at the plasma membrane, is also the most potent phosphoinositide activator of Pak1 under the conditions tested.
The potentiation of Pak1 activation by phosphoinositides and Rho GTPases in vitro begs the question of the relative importance of these two inputs for Pak1 activation in vivo. Overexpression of constitutively active forms of Rac1 or Cdc42 alone in cultured cells is sufficient to activate Pak1 (Knaus et al., 1998). Our findings suggest the possibility that this common experimental protocol may overestimate the importance of the Rho GTPase input on Pak1 activation by producing non-physiological concentrations of active Rac1/Cdc42. Rac1/Cdc42 overexpression might, in fact, promote PIP2 synthesis as Rac1 interacts directly with phosphatidylinositol 4-phosphate 5-kinases (PIP5K), and both Rac1 and Cdc42 enhance PIP5K activity (van den Bout and Divecha, 2009). Similarly, local synthesis of PIP2 has been shown to promote Rho GTPase recruitment via interactions with a C-terminal basic region (Heo et al., 2006). These interactions may therefore provide a positive feedback loop to promote synergistic enrichment of both active Rho GTPase and PIP2, which would ultimately promote Pak1 activation.
Our data indicate a critical role for PIP2-Pak1 binding in vitro, in cell extracts, and in cultured mammalian cells suggesting that Pak1 activation may be phosphoinositide-dependent in many, if not all, physiological contexts. Our findings also provide an explanation for previous observations that Pak1 lysine residues Lys66-68 (mutated in both 4T and 8T Pak1) play a role in Pak1 activation independent of GTPase binding (Knaus et al., 1998). Furthermore, our demonstration of Pak1 activation solely by PIP2 liposomes raises the possibility that pathologically elevated levels of phosphoinositides on their own might lead to Pak1 activation. Elevated Pak1 kinase activity is associated with cancers of the breast and other tissues, leading to significant interest in the therapeutic potential of Pak1 inhibitors (Deacon et al., 2008; Molli et al., 2009). However, unlike the situation with Ras-induced neoplasia, activating mutations in the upstream GTPases Rac1 and Cdc42 are not commonly found in human tumors. Thus, perturbations in phosphoinositide metabolism could account for the Pak1 hyperactivity associated with various cancers.
In electronic circuits, coincidence detectors reduce stochastic noise by requiring cotemporaneous signals from multiple inputs to produce a positive signal. We postulate that the cooperative regulation of Pak1 by PIP2 and Rho GTPases may be functioning in an analogous manner in the cell to ensure Pak1 activation only at sites where PIP2 and activated Rho GTPase overlap. The evolutionary conservation of basic regions in the related Paks 2 and 3 suggests that phosphoinositide-dependent activation may be a hallmark of all three Group I Pak isoforms. Interestingly, a number of Rac1/Cdc42 effectors also have basic amino acid-enriched sequences adjacent to their GTPase binding domains including myotonic dystrophy kinase-related Cdc42-binding kinase, BORG1/2, and the mixed lineage kinases, suggesting that phophoinositide binding may be a common regulatory mechanism. In striking contrast, other Rac/Cdc42 effectors, such as Ack, lack identifiable basic regions in this position, implying that the combinatorial activation by Rac1/Cdc42 and phosphoinositides may provide a mechanism for selectively activating a subset of effectors downstream of Rho GTPases.
Lipids were from Avanti Polar Lipids, Echelon Biosciences, Calbiochem, or Cayman Chemicals. PDGF-BB and PMA were from Sigma. NSC23766 was from NCI. Anti-phospho-Ser144 Pak1 antibody was from Invitrogen. Anti-phospho-Ser199 Pak1 antibody was from Cell Signaling Technology. Anti-total Pak (C19) and anti-myc (9E10) antibodies were from Santa Cruz Biotechnology.
Human Pak1 in pJ3H (Sells and Chernoff, 1995) was mutagenized using the Quikchange Kit (Agilent). All constructs were fully sequenced. For expression of myc-tagged Pak1, the Pak1 open reading frame was subcloned using BamHI/EcoRI from pJ3H-Pak1 into pCMV6 (OriGene). The Pak1 autoregulatory domain (amino acids 1–84) was amplified by PCR as a BamHI/SalI or BglII/EcoRI fragment and subcloned into pGEX-KG-KAN (Novagen) or pEGFP-C1 (Clontech), respectively. For cloning of GST-Paktide, oligonucleotides encoding peptide sequence GRRRRRSWYWDG were annealed and ligated into pGEX-6P-1 (GE Healthcare) using BamHI/EcoRI.
Pak1 basic region mutants were subcloned using BamHI/EcoRI from pJ3H-Pak1 into pFastBac HTb (Invitrogen). Baculovirus stocks were prepared and proteins were expressed and purified from Sf9 cells as described (Rennefahrt et al., 2007). GST fusion proteins were expressed in and purified from E. coli using standard protocols (GE Healthcare).
Liposomes were prepared by mixing chloroform stocks, drying, and resuspending with sonication in 10 mM HEPES, pH 7.7, 100 mM NaCl, 50 mM sucrose, 5 mM EGTA. Liposomes were stored at 4°C and used within one week of preparation.
Liposome sedimentation assays were conducted as in (Takahashi and Pryciak, 2007) except 100 µM liposomes and 20 µg of GST fusion proteins were used. Supernatant and pellet fractions were analyzed by SDS-PAGE and Coomassie blue staining.
Pak1 was incubated with 800 µM ATP/[32P]-γ-ATP and GST-Paktide (10 µg) or with 1 mM ATP without substrate plus the indicated concentrations of Rac1 and/or liposomes at 22 or 30°C for ~30 min. Reactions were stopped by the addition of 95°C sample buffer and were subjected to SDS-PAGE and Coomassie blue staining, silver staining, or immunoblotting as indicated.
A high speed supernatant of Xenopus egg cytoplasmic extract (Peterson et al., 2001) supplemented with 1 mM ATP and 0.1 mM GTPγS was stimulated with the indicated dose of PIP2 liposomes for the indicated times at 22°C. Reactions were stopped by 95°C SDS-PAGE sample buffer and were analyzed by western blotting or in-gel kinase assay as described in (Beeser et al., 2005) except using peptides GGRRRRRSWASPGGK (0.1 mg/ml; Fig. 2A) (Rennefahrt et al., 2007) or R698 (RRRRRSWYWDG-CONH2; 0.1 mg/ml; Fig. S2) as substrates. Inhibitors were added to extracts prior to PIP2 stimulation. For Figure 2D, Xenopus egg extracts were supplemented with 66 nM human WT or mutant Pak1 proteins and were stimulated with 80 µM PIP2 liposomes for 30 min at 30°C.
Sf9 cells were maintained in SFM-900 with Pluronic F-68 and antibiotic/antimycotic (Gibco). BSC-1 and NIH3T3 cells were grown in DMEM with 10% fetal bovine serum, L-glutamine, and penicillin/streptomycin. Lipofectamine 2000 (Invitrogen) was used for transfection.
BSC-1 cells seeded on glass coverslips were transfected for 48 hours. Cells were stimulated with PMA (250 ng/ml) for 15 min and were then fixed in 4% formaldehyde. Following permeabilization, coverslips were stained using anti-HA and Alexa-488-anti-mouse antibodies and co-stained with Texas Red X-phalloidin (Invitrogen). Cells were imaged using a Leica TCS SP5 spectral confocal system with a 63x PlanApo objective. The 488- and 594-nm laser lines were used for sequential excitation. Images were captured with a 4-channel internal spectrometric detector.
NIH3T3 cells transfected with plasmids encoding myc-tagged WT, 8T, or 8T+8K Pak1 or mock transfected were serum-starved overnight. Cells were stimulated with 50 ng/ml PDGF-BB for 30 min. Cells were lysed in 50 mM Tris pH 8.0, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 5 mM NaVO4, 10 mM β-glycerol phosphate, 50 mM NaF, 5 mM MgCl2, 5 mM Na4P2O7, 10% glycerol containing protease inhibitor (Roche). Lysates were centrifuged at 20,000 × g and supernatants were pre-cleared by incubation with protein A/G agarose (Thermo). Pak1 was immunoprecipitated with anti-myc antibodies. Beads were washed three times with lysis buffer, three times with PBS, and twice with kinase buffer (50 mM HEPES pH 7.6, 13.5 mM NaCl, 0.65 mM MgCl2). Kinase reactions were started by the addition of a mix containing GST-Paktide (10 µg) and 30 µM ATP/[32P]-γ-ATP. After 45 min, reactions were stopped by the addition of sample buffer and analyzed by SDS-PAGE and autoradiography. Quantification was performed as described in (Viaud and Peterson, 2009). Statistical significance was calculated with a Student’s t-test.
We thank N. Morin, A. O’Reilly, and M. Lemmon for equipment and reagents; W. Kruger, M. Murphy, and J. Chernoff for comments on the manuscript; and J. Chernoff for plasmids and discussions. This work was supported by a W.W. Smith Foundation Award and an American Cancer Society Scholar Award to J.R.P. and by National Institutes of Health (NIH) awards RO1 GM083025 to J.R.P., T32 CA009035 to T.I.S., and P30 CA006927 to Fox Chase Cancer Center.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.