The physiology of membrane transport and endomembrane-based signalling

Abstract

Some of the important open questions concerning the physiology of the secretory pathway relate to its homeostasis. Secretion involves a number of separate compartments for which their transport activities should be precisely cross-coordinated to avoid gross imbalances in the trafficking system. Moreover, the membrane fluxes across these compartments should be able to adapt to environmental ‘requests' and to respond to extracellular signals. How is this regulation effected? Here, we consider evidence that endomembrane-based signalling cascades that are similar in organization to those used at the plasma membrane coordinate membrane traffic. If this is the case, this would also represent a model for a more general inter-organelle signalling network for functionally interconnecting different intracellular activities, a necessity for the maintenance of cellular homeostasis and to express harmonic global cellular responses.

In vivo physiology of the secretory pathway

The genetics and biochemistry of intracellular membrane trafficking have been progressively elucidated over the last few decades, and some of the molecular machineries underlying elementary transport events, such as membrane bending, budding, fusion and fission, are now understood in great detail (Lee et al, 2004). In contrast, there has been much less progress towards a true understanding of supramolecular, or ‘global', aspects of the trafficking system that underlie the physiology of membrane transport in vivo. Among these are the issues concerning the micro-anatomy and micro-dynamics of the trafficking compartments and the organizational principles of transport in vivo (Trucco et al, 2004; Luini et al, 2005), some aspects of which have seen significant advances only recently.

The reasons for this knowledge gap are multiple, and although these include historical/cultural aspects, the main cause has been the lack of suitable technologies. An understanding of the physiology of trafficking requires the visualization of traffic events in vivo with high resolution in time and space—a possibility that has only become (partially) available in the last few years owing to the advent of increasingly powerful video-microscopy and electron tomography approaches, as well as of effective ancillary techniques (e.g. traffic synchronization and correlative electron microscopy) (Polishchuk et al, 2000). These advances have spurred a series of studies that have led to new models of the organization of transport in vivo—models that, incidentally, have not always been easy to reconcile with previous schemes that were based mostly on genetic and in vitro biochemical data (Mellman and Warren, 2000; Pelham and Rothman, 2000; Lee et al, 2004; Behnia and Munro, 2005). Moreover, as a natural consequence of these studies, new fundamental issues about the physiology of traffic have come to the fore. One in particular is the issue of the identity of the mechanism underlying the internal coordination and homeostasis of these compartments as well as the regulation of transport organelles by exogenous cues. Here, we will focus on the signalling circuits that might support the coordination of the compartments of the secretory pathway.

Homeostasis of the secretory compartments

Intracellular trafficking pathways comprise a number of anatomically separate compartments that are constantly exchanging their membranes and cargo proteins in an organized sequence of trafficking segments (e.g. endoplasmic reticulum (ER) to Golgi complex, Golgi complex to plasma membrane (PM), etc.). The membrane fluxes along each of these segments should be precisely coordinated under physiological conditions to avoid gross imbalances in the system; moreover, the overall trafficking fluxes (including those regulated and constitutive) should be able to adapt to environmental ‘requests' and to react to extracellular signals as part of the global cellular response.

How is this regulation effected? Several mechanisms are in principle conceivable. For instance, it is possible to explain the ability of the secretory system to control traffic fluxes across different compartments and to maintain the size and composition of these compartments simply on the basis of a few parameters, such as the properties and amounts of the coat and SNARE proteins in the system (Bonifacino and Glick, 2004; Heinrich and Rapoport, 2005). Moreover, it is possible that simple constitutive phosphorylation and dephosphorylation events, such as those proposed to participate in the clathrin-dependent endocytic cycle, might also have roles in complex trafficking responses (Flett et al, 2005). Other considerations, however, suggest that more sophisticated control mechanisms might exist (see below), superimposed onto these basic ‘automatic' processes, acting perhaps in the same way in which computerized electronic circuits regulate the performance of mechanical parts of modern cars. Here, we will consider one specific hypothesis; namely, that these regulatory circuits involve endomembrane-based signalling cascades that are similar in their organization and components to those used by the cell at the PM. Moreover, these signalling cascades should be initiated by traffic itself and affect the activity of other organelles (this type of signalling organization could therefore be defined as ‘inter-organelle signalling'). This hypothesis is supported by two main arguments. The first is that a variety of ‘classical' signalling molecules, including G proteins, kinases and phospholipases, are physically associated with secretory endomembranes, and in particular with the Golgi complex (Stow and de Almeida, 1993; Denker et al, 1996; Donaldson and Lippincott-Schwartz, 2000; Bard et al, 2002; Nagahama et al, 2002; Bivona and Philips, 2003; Wylie et al, 2003; Larocca et al, 2004; Preisinger et al, 2004; Diaz Anel and Malhotra, 2005; Ghanekar and Lowe, 2005; Le Niculescu et al, 2005). They are therefore well placed to participate in traffic-initiated signalling (Helms et al, 1998; Cabrera et al, 2003). The second, more speculative argument is that given the sophisticated spatio-temporal control that has evolved for PM-initiated signal transduction pathways (Pierce et al, 2002), it would be uneconomical for cells not to use these mechanisms to generate signals operating across intracellular organelles to mediate inter-compartment coordination of membrane trafficking.

In the following, we will review some of the available evidence pertaining to the presence of signalling molecules on endomembranes and to the role of these molecules in the regulation/coordination of the trafficking pathway. We do not plan to be exhaustive in this survey. Potentially relevant aspects of the relationship between signalling and traffic have been reviewed recently and will not be treated here. These include the well-established links between signalling and endocytosis, whereby signalling complexes generated at the PM can modulate endocytic activity and can themselves be modulated while they reside on endocytic organelles (Di Fiore and De Camilli, 2001; Sorkin and Von Zastrow, 2002; Conner and Schmid, 2003; Miaczynska et al, 2004). They also include the links between signalling and autophagy (Codogno and Meijer, 2005). Neither will we discuss the significance and the mechanisms of regulation of overall secretory traffic rates by surface receptors, aspects that have been previously reviewed by us and others (Luini and De Matteis, 1993; Bannykh et al, 1995). Rather, we will focus mainly on the secretory pathway and on a selected group of ‘traditional' signalling protein families and pathways that have been more extensively characterized in the context of secretory traffic and that provide relevant examples of the issues we are dealing with here. We will also summarize the literature data on a larger number of molecules for which only limited information is at present available, but which might turn out to be relevant to the issue being examined (Table I).

Table 1.

Signalling proteins and their functions on intracellular organelles

Signalling protein	Organelle localization	Molecular function	Role on endomembranes	References
AKAP350	Centrosome, ERGIC	A-kinase-anchoring protein	Unknown	Shanks et al (2002)
ATF6	ER	Transcription factor	Transduces effects of UPR	Rutkowski and Kaufman (2004)
BIG2	Golgi	Nucleotide exchange factor and AKAP	Unknown	Li et al (2003)
C3G	Golgi	Rap1 exchange factor	MAP kinase pathway	Radha et al (2004)
Cbl	Golgi	E3 ubiquitin ligase	Unknown	Bard et al (2002)
Cdc42	Golgi	Small GTP-binding protein	Involved in ER-to-Golgi, TGN-to-PM and Golgi-to-ER transport	Luna et al (2002), Matas et al (2004), Stamnes (2002)
ERK	Golgi	Serine/threonine kinase	Involved in Golgi disassembly during mitosis	Cha and Shapiro (2001), Acharya et al (1998)
Fps	Golgi	Tyrosine kinase	Unknown	Zirngibl et al (2001)
GIV	COPI vesicles	Regulator of G-protein signalling	Unknown	Le Niculescu et al (2005)
GRP1	Golgi	Ras exchange factor	Necessary for Ras activation on the Golgi	Bivona et al (2003)
Gαi	Golgi	Heterotrimeric GTP-binding protein	Regulates TGN-to-PM transport	Pimplikar and Simons (1993)
Gαi3	Cis-Golgi	Heterotrimeric GTP-binding protein	Regulates transport of constitutively secreted proteins	Wilson et al (1994), Stow and de Almeida (1993)
Gαq	Golgi, mitochondria	Heterotrimeric GTP-binding protein	Unknown	Wilson et al (1994), Denker et al (1996)
Gαs	TGN	Heterotrimeric GTP-binding protein	TGN-to-PM transport	Pimplikar and Simons (1993)
Gαz	Golgi	Heterotrimeric GTP-binding protein	Maintenance of Golgi structure	Nagahama et al (2002) Yamaguchi et al (2000)
IRE1	ER	Serine/threonine kinase	Transduces effects of UPR	Rutkowski and Kaufman (2004)
JAK-2	Transitional ER	Tyrosine kinase	Coordinates ERES assembly	Lavoie et al (2000)
KDELr	ER, Golgi	Receptor protein	Transduces effects of UPR	Yamamoto et al (2003)
Lck	Golgi	Tyrosine kinase	Unknown	Bijlmakers et al (1997)
LimK	Golgi	Serine/threonine kinase	Kinase-dead mutant produces tubular processes emerging from the Golgi stack	Rosso et al (2004)
MEK-1	Golgi	Serine/threonine kinase	Involved in Golgi disassembly during mitosis	Colanzi et al (2003)
MINK	Golgi	Serine/threonine kinase	Unknown	Hu et al (2004)
MST4	Golgi	Serine/threonine kinase	Depletion by siRNA causes Golgi dispersion	Preisinger et al (2004)
MTG	Cis-Golgi	Myeloid translocation gene	Unknown	Asirvatham et al (2004)
Myomegalin	Golgi, centrosome	Scaffold protein	Unknown	Verde et al (2001)
PCTAIRE	ER	Cycin-dependent kinase	Regulator of ER-to-Golgi transport	Palmer et al (2005)
PDE4D3	Golgi, centrosome	Phosphodiesterase	Unknown	Jin et al (1998)
PDE7A	Cis-Golgi	Phosphodiesterase	Unknown	Asirvatham et al (2004)
PERK	ER	Serine/threonine kinase	Transduces effects of UPR	Rutkowski and Kaufman (2004)
PI3K	Golgi	Lipid kinase	Involved in Golgi disassembly during mitosis	Domin et al (2000)
PKA	Golgi	Serine/threonine kinase	Regulates endosome-to-TGN, Golgi-to-ER, and TGN-to-PM transport	Birkeli et al (2003), Muniz et al (1997), Cabrera et al (2003)
PKC	Golgi	Serine/threonine kinase	Regulates TGN-to-PM transport and apoptosis	Buccione et al (1996) Kajimoto et al (2004)
PKD	Golgi	Serine/threonine kinase	TGN-to-PM transport	Diaz Anel and Malhotra (2005), Ghanekar and Lowe (2005)
PLK3	Golgi	Serine/threonine kinase	Involved in Golgi disassembly during mitosis	Xie et al (2004)
PSKH1	Cis-Golgi, centrosome, nuclei	Serine/threonine kinase	Mutant kinase causes Golgi disassembly	Brede et al (2003)
PTPH1	ERES	Protein tyrosine phosphatase	Coordinates ERES assembly	Lavoie et al (2000)
Rac	TGN	Small GTP-binding protein	TGN sorting	Faucherre et al (2003)
Rap1	Golgi	Small GTP-binding protein	Unknown	Wienecke et al (1996)
Ras	Golgi, ER	Small GTP-binding protein	Supports cell transformation via MAPK	Bivona and Philips (2003)
RGS2	Golgi	Regulator of G-protein signalling	Unknown	Sullivan et al (2000)
RGS4	Golgi	Regulator of G-protein signalling	Regulates Golgi-to-PM transport via interaction with β-COP	Sullivan et al (2000)
RGS-GAIP	TGN	Regulator of G-protein signalling	Unknown	De Vries et al (1998)
RGSZ	Golgi	Regulator of G-protein signalling	Regulates ER-to-PM transport and Golgi disruption	Chatterjee and Fisher (2000), Nagahama et al (2002)
Sef	Golgi	Scaffold protein	Spatial regulator of ERK signalling	Torii et al (2004)
sGαi2	Golgi	Heterotrimeric GTP-binding protein	Maintenance of Golgi structure	Montmayeur and Borrelli (1994)
Src	Golgi	Tyrosine kinase	Regulates retrograde transport of KDELr	Bard et al (2003), Bard et al (2002)
TrkA	Golgi	Tyrosine kinase receptor	Neuronal survival	Rajagopal et al (2004)
XLGαs	Golgi	Heterotrimeric GTP-binding protein	Unknown	Pasolli et al (2000) Ugur and Jones (2000)
YSK1	Golgi	Serine/threonine kinase	Kinase-dead and siRNA causes Golgi dispersion	Preisinger et al (2004)
βγ	Unknown	Heterotrimeric GTP-binding protein	TGN-to-PM transport and regulation of Golgi complex organization	Jamora et al (1999), Diaz Anel and Malhotra (2005)
AKAP, protein kinase A-anchoring scaffold protein; GTP, guanine 5′-triphosphate; MAP, mitogen-activated protein; PM, plasma membrane; ER, endoplasmic reticulum; ERGIC, ER–Golgi intermediate compartment; TGN, trans-Golgi network; ERES, ER exit sites; UPR, unfolded protein response; KDELr, KDEL receptor.	The membranes of the secretory system host several signalling proteins, ranging from the heterotrimeric G-proteins to small G-proteins, scaffold proteins, kinases, phosphodiesterases, GTPase-activating proteins and guanine nucleotide exchange factors. For many of these, the only information available relates to the effects of their perturbation on Golgi morphology and function, with their physiological roles remaining, as yet, unknown.

While this evidence will be discussed in the light of an inter-organelle signalling hypothesis for traffic coordination (Figure 1A), it is also necessary to consider other schemes that can explain the presence of signalling molecules on endomembranes. An alternative possibility, for instance, is that Golgi-based transduction proteins might serve to mediate inputs that initiate at the PM but regulate Golgi function (Figure 1C), as shown previously by us and others (De Matteis et al, 1993; Luton et al, 1999) (a similar regulatory mechanism has been shown to exist also in the case of endocytosis (Sorkin and Von Zastrow, 2002; Miaczynska et al, 2004)). Furthermore, it is possible that transduction molecules present on the Golgi complex act as relay devices in signalling networks initiated at the PM that control cell growth and other functions (Figure 1D). This has been shown to be the case for Golgi- and endosome-based signalling proteins (Di Fiore and De Camilli, 2001; Sorkin and Von Zastrow, 2002; Bivona et al, 2003; Miaczynska et al, 2004). Finally, signalling proteins might simply be in transit through the Golgi to the PM following their synthesis (Michaelson et al, 2002) (Figure 1E). It should also to be borne in mind that these models (schematized in Figure 1) are not mutually exclusive.

Figure 1.

Generic models for the roles of signalling proteins at the Golgi complex. (A, B). Direct activation of Golgi signalling by endogenous events (e.g. incoming traffic)—self-regulation of the trafficking pathway (A) and/or regulation of other cellular functions (B); (C) activation of PM signals—signal translation—activation of Golgi signaling—regulation of trafficking; (D) activation of PM signals—signal translation—activation of Golgi signaling—regulation of other cellular functions (e.g. cell proliferation, motility); (E) passive residence of signalling molecules in transit to the PM.

The cAMP-PKA pathway

The centrepiece of the cAMP-PKA pathway is protein kinase A (PKA), a tetrameric serine/threonine kinase that consists of two regulatory and two catalytic subunits. The kinase becomes active when the catalytic and regulatory subunits dissociate in the presence of cAMP. In addition to PKA, the pathway includes the enzyme adenylyl cylase (AC, which synthesizes cAMP) and AC activators (such as Gαs), pathway inactivating phosphodiesterases (PDEs, which hydrolyze cAMP) and the PKA-anchoring scaffold proteins(s) (AKAPs, see below). Remarkably, all of these components have been shown to be located on endomembranes, and particularly on the Golgi complex (Cheng and Farquhar, 1976a, 1976b; Maier et al, 1995; Denker et al, 1996; Pooley et al, 1997; Martin et al, 1999; Birkeli et al, 2003; Li et al, 2003; Asirvatham et al, 2004; Larocca et al, 2004); our unpublished observations).

The cAMP-PKA pathway can exert potent effects on many trafficking steps. Transport from the ER to the Golgi complex can be accelerated through stimulation of cAMP synthesis and impaired by the PKA inhibitor H89 (Muniz et al, 1996). However, this transport step is not affected by other PKA blockers (i.e. KT5720 and PKI) (Aridor and Balch, 2000; Lee and Linstedt, 2000), suggesting that H89 acts through other kinase(s) and indicating that while the cAMP-PKA cascade can accelerate this step, it is not required for its execution at steady state. PKA also regulates intra-Golgi transport and transport from the trans Golgi network (TGN) to the PM (Muniz et al, 1996). Velasco and co-workers showed that PKA catalytic activity is required for the budding of constitutive transport vesicles from the TGN in vitro (Muniz et al, 1997) and that the stimulation of cAMP synthesis in vivo accelerates transport from the Golgi to the PM. An increase in cAMP also modifies the structure of the Golgi complex, with induction of the formation of numerous tubules interconnecting nonequivalent Golgi cisternae (Muniz et al, 1996). Finally, the activity of PKA appears also to be required for retrograde transport from the endosomal compartment to the Golgi complex and from the Golgi to the ER (Birkeli et al, 2003) (Cabrera et al, 2003).

The mechanisms of action through which PKA exerts these effects have been identified only in part. One mechanism relates to the ability of PKA to recruit ADP ribosylation factor-1 (ARF1) to the Golgi complex (Martin et al, 2000). ARF1 belongs to the Ras superfamily of small GTP-binding proteins and is the master regulator of many fundamental steps along the secretory pathway, including the assembly of coat proteins, such as COPI and clathrin (Donaldson et al, 2005). The recruitment of ARF1 from the cytosol to Golgi membranes has been shown to be increased by the catalytic portion of PKA and by cAMP (Martin et al, 2000). Conversely, ARF1 recruitment is impaired by PKI (a selective PKA inhibitor) and by PKA depletion (Martin et al, 2000). This has been confirmed in vivo by direct stimulation of cAMP formation using forskolin, which led to an increase in the recruitment of ARF1 to the Golgi complex (Martin et al, 2000).

Another target of PKA that might contribute to the regulation of membrane transport is the KDEL receptor (KDELr). The KDELr is a seven-transmembrane-domain protein that cycles between the ER and the Golgi complex (Lewis and Pelham, 1992). In doing so, it returns to the ER the ER-resident chaperones that have leaked through to post-ER compartments during membrane transport. PKA phosphorylates a COPI-binding motif on the C-tail of the KDELr, which is necessary for the retrograde transport of the KDELr itself (Cabrera et al, 2003). PKA also phosphorylates certain SNAREs (soluble N-ethyl-maleimide-sensitive fusion protein attachment protein receptor) and regulates their ability to support membrane fusion in endocytic and exocytic steps in mammals (Hong, 2005), as well as in ER-to-Golgi and intra-Golgi transport in yeast (Hong, 2005; Weinberger et al, 2005). Furthermore, PKA is involved in the ER quality control process for a class of proteins that bear the RXR transient-ER-retention motif through which they bind to COPI (Michelsen et al, 2005). These include G-protein-coupled receptors (GPCRs) and multimeric ion channels, the transport of which is delayed in the ER through a COPI-mediated retrieval mechanism. When these protein complexes are properly assembled, their RXR motif is masked by an interaction with the partner subunit(s) or by PKA or protein kinase C (PKC)-mediated phosphorylation of a nearby serine or threonine, which itself generates a binding site for the adaptor protein 14-3-3 that prevents the RXR motif from interacting with COPI (Kuwana et al, 1998; Michelsen et al, 2005).

The presence of the components of the cAMP-PKA pathway (including those responsible for the formation of cAMP) on the Golgi and their regulation of traffic through this organelle is consistent with a scheme (Figure 1A) in which PKA-based Golgi-initiated signalling coordinates the activities of different Golgi subcompartments. However, the complete circuit in which this pathway might be involved, and the circumstances in which this circuit might be activated, remains to be defined, and alternative explanations for the presence of the pathway components on the Golgi cannot be excluded at this time (e.g. schemes in Figure 1B and C).

The DAG-PKC pathway

The protein kinase C (PKC) are a family of at least 10 serine/threonine kinases that can be structurally and functionally subclassified into three main groups: those Ca²⁺- and diacylglycerol (DAG)-dependent, those calcium-independent and those atypical, calcium- and DAG-independent. Most of these can localize to the Golgi complex via their C1 lipid-binding domains (Schultz et al, 2004), which can bind to DAG and/or to ceramides, retinoic acid and archidonic acid. This binding can have important functional consequences. In neuroblastoma and HeLa cells treated with ceramides, PKCδ and PKCɛ show a C1-domain-dependent translocation to the Golgi complex, and this translocation is necessary for the apoptotic effects promoted by ceramides (Kajimoto et al, 2001; Schultz et al, 2003; Kajimoto et al, 2004). Similar effects are seen with IFN-γ treatment: this acts via ceramide production and the consequent translocation of PKCδ to the Golgi complex, which results in cell apoptosis. PKCδ is activated only after its Golgi translocation, via direct phosphorylation on tyrosines 311 and 332 by a Golgi-located Src, and this phosphorylation is necessary for the apoptotic effect promoted by PKCδ (Kajimoto et al, 2004). Another Golgi targeting mechanism is via the RACK (receptor for activated PKC)-anchoring proteins. PKCɛ is targeted to the Golgi complex via a direct interaction with the β′ subunit of the coatomer β′-COP, which acts as a RACK through its characteristic WD-40 repeat that is found in other RACKs (Csukai et al, 1997).

The recruitment of the PKCs to endomembranes also has regulatory effects on traffic. The intermediate compartment that operates between the ER and the Golgi is an important sorting station where anterograde-directed proteins are segregated from recycling proteins. Here, Rab2, a small GTPase and a key player in the intermediate compartment function, recruits PKCι/λ (Tisdale and Artalejo, 2006), which promotes the binding of COPI to intermediate compartment membranes (Tisdale, 2000, 2003), and phosphorylates glyceraldehyde 3-phosphate dehydrogenase (GAPDH). GAPDH then ultimately influences microtubule dynamics and transport in the early secretory pathway (Tisdale, 2002). ER-to-Golgi transport is impaired by the specific PKC inhibitor calphostin C, whereas it is activated by DAG analogues (e.g. phorbol 12-myristate 13-acetate (PMA)) (Fabbri et al, 1994). However, several other PKC inhibitors that act as ATP competitors (e.g. H7, H8, staurosporine) and the downregulation of PKC have no effects on ER-to-Golgi transport, thus casting doubts as to the role of PKC here, and suggesting instead the involvement of another (as yet unidentified) protein that contains a DAG-binding domain (Fabbri et al, 1994).

Constitutive intra-Golgi and TGN-to-PM transport can be modulated by PKC (De Matteis et al, 1993; Fabbri et al, 1994). De Matteis et al (1993) have demonstrated that the transport of glucosaminoglycans (GAGs) and the temperature-sensitive variant of the G protein of vesicular stomatitis virus (VSVG) from the Golgi to the PM requires PKC activity (Buccione et al, 1996), and that the direct activation of PKC by PMA or via activation of the IgE receptor stimulates GAG secretion. In this context, the action of PKC is likely to involve more than one mechanism, including the regulation of the GTP-dependent binding of ARF1 to Golgi membranes (De Matteis et al, 1993). As the IgE receptor has been shown to stimulate traffic from the ER to the Golgi in addition to Golgi-to-PM traffic (Bannykh et al, 1995; Buccione et al, 1996), these data identify a case of stimulation of the entire constitutive secretory pathway by an extracellular ligand. The significance of these observations has been discussed previously (Luini and De Matteis, 1993). In an analogous study, Fabbri et al (1994) showed that the PKC inhibitor calphostin C blocks the transport of VSVG from the TGN to the PM. Using an in vitro system to generate post-Golgi vesicles, Sabatini and co-workers showed that PKC is involved in membrane carrier fission from the TGN (however, they proposed that in this case PKC does not act via its kinase activity) (Simon et al, 1996). Finally, in a series of studies that is further discussed below, the Malhotra laboratory (see above) showed that PKCη acts as a kinase that activates protein kinase D (PKD) and that this is required for the fission of TGN carriers (Diaz Anel and Malhotra, 2005).

Therefore, PKC on the Golgi complex appears to be involved in at least two main functions: apoptosis-related signalling and the regulation of traffic. One of the PKC-dependent pathways that controls traffic initiates at the cell surface with the activation of IgE receptors, and so this fits well with scheme C in Figure 1. Whether PKC is also activated on the Golgi by traffic itself (scheme A in Figure 1) is unknown, but it is possible that traffic can induce changes in the lipid composition (DAG, ceramides) in specific Golgi domains, which could then result in the regulation of PKC activity (Baron and Malhotra, 2002; Diaz Anel and Malhotra, 2005).

Heterotrimeric G proteins

Heterotrimeric G proteins are a family of GTPases that are primarily involved in the transduction of signals initiated at the PM by seven-transmembrane-domain receptors. However, they have also been proposed to have direct roles in membrane transport based on the evidence that a number of different Gα subunits are located on internal membranes, including the Golgi complex (Denker et al, 1996; Helms et al, 1998; Qian et al, 2002), and that manipulation of the levels or activities of these G proteins markedly affects the functioning of transport pathways.

The distribution of the different Gα subunits on endomembranes has been the target of many studies in different laboratories (Stow et al, 1991b; Wilson et al, 1994). Stow and co-workers demonstrated that Gα_i3 is present on Golgi membranes in epithelial cells (Stow et al, 1991a; Stow and de Almeida, 1993) and Farquhar and co-workers (Wilson et al, 1994) showed that Gα_i3 is mostly on the cis side of the Golgi in pituitary cells. A partial Golgi distribution was also described for Gα_q (Denker et al, 1996). Both a short and a long form of Gα_s have been identified in the Golgi in liver and adrenal medulla cells (Maier et al, 1995), and Huttner and co-workers showed that Gα_s colocalizes with the trans-Golgi marker protein TGN38 in PC12 cells (Leyte et al, 1992). Gα_z has also been shown to localize to the Golgi complex in BHK cells, where it overlaps with the Golgi marker mannosidase II and follows the redistribution of Golgi proteins after treatment with brefeldin A (BFA, a fungal toxin that disassembles and redistributes the Golgi into the ER) (Nagahama et al, 2002).

From a functional standpoint, several secretory transport steps have been proposed to be regulated by heterotrimeric G proteins on the basis of a variety of lines of evidence. Transport from the ER to the Golgi complex was suggested to be controlled by these G proteins, since mastoparan (an activator of Gα_i) causes a block in the exit of VSVG from the ER (Schwaninger et al, 1992). A similar result was also seen in permeabilized cells treated with purified βγ subunits (which deplete the Gα pool by associating with it), suggesting that both activatory and inhibitory heterotrimeric G proteins are involved in the export of proteins from the ER.

Heterotrimeric G proteins have also been implicated in intra-Golgi and post-Golgi trafficking. Manipulation (overexpression or treatment with pertussis toxin) of Gα_i3 (which is located in the Golgi; see above) indicated that this protein has an overall inhibitory role on anterograde secretory traffic (Stow et al, 1991a). COPI (a protein complex crucially involved in intra-Golgi traffic (Rothman, 1994)) has been shown to interact with the hetrotrimeric G protein GTPase-activating protein RGS4 (regulator of G protein signalling-4) (Sullivan et al, 2000). This interaction impairs COPI binding to Golgi membranes (although ARF1 localization and Golgi integrity are not affected), and apparently results in inhibition of the transport of acquaporin 1 to the PM and the secretion of placental alkaline phosphatase. Thus, it was hypothesized that RGS4 regulates transport through the sequestration of β-COP from the cytosol, with this sequestration inhibiting β-COP recruitment to Golgi membranes. Potentially linked to these observations, Velasco and co-workers showed that the in vitro interaction of PKA with Golgi membranes is sensitive to modulators of heterotrimeric G proteins, and that myristoylated Gα_i3 can stimulate this interaction, suggesting a role for G proteins in the recruitment of PKA to Golgi membranes (Martin et al, 1999). PKA participates in COPI-dependent retrograde transport of the KDELr by the phosphorylation of the C-tail of the KDELr, which is mandatory for its interaction with β-COP and its inclusion in COPI-coated vesicles (see also below). Collectively, the above data suggest that G-protein-dependent signalling is involved in the regulation of the COPI trafficking machinery, although the precise mechanism of this involvement remains unclear.

Heterotrimeric G proteins also appear to be involved in the regulation of TGN-to-PM transport. In an early study, Pimplikar and Simons (1993) showed that treatment with reagents that influence Gα_s specifically impair apical, and not basolateral, transport. On the other hand, a selectivity towards basolateral transport was seen with treatments influencing Gα_i, leading to the conclusion that Gα_s and Gα_i could be involved in the regulation of polarized TGN-to-PM transport. In addition to Gα subunits, the βγ subunits appear to have a role in transport from the TGN. In what is the most complete series of studies in this area to date, Malhotra and co-workers showed that PKD, a key regulator of membrane fission at the TGN, is activated by a signalling pathway that involves βγ subunits (β₁γ₂/β₃γ₂), and that this activation passes through PKCη, in this way regulating transport to the cell surface and organization of the Golgi complex (Diaz Anel and Malhotra, 2005) (Liljedahl et al, 2001; Yeaman et al, 2004). RGS proteins are also present on the TGN, and it has been shown that RGS-Gα-interacting protein (RGS-GAIP) is localized to clathrin-coated buds and vesicles in the Golgi region. However, at this stage, the function of RGS-GAIP remains to be defined (De Vries et al, 1998; Wylie et al, 1999, 2003; Chatterjee and Fisher, 2000; Gleeson et al, 2004).

Gα_z and Gα_i2 also appear to have key roles in the maintenance of the overall structure of the Golgi complex (Yamaguchi et al, 2000). Their overexpression results in inhibition of nordihydroguaiaretic acid (NDGA)-induced disassembly of the Golgi complex. Conversely, overexpression of the regulator (repressor) of Gα_z signalling (RGS_z) stimulated disruption of the Golgi complex, and delayed ER-to-PM transport of VSVG (Nagahama et al, 2002).

Thus, there is abundant (albeit heterogeneous) evidence for a role for heterotrimeric G proteins and of some of their accessory proteins in secretory transport, as well as in the control of the structure of the Golgi. The fact that they are both present and able to exert effects on this organelle suggests that they might operate within local regulatory circuits, which would be consistent with scheme A in Figure 1. However, a coherent picture of the pathway(s) in which they are involved remains to be defined.

The Ras signalling pathway

Ras is the prototype of a large family of small GTPases that are involved in multiple cellular functions, including the transduction of the effects of many extracellular ligands, such as growth factors, via PM receptors. Recently, Ras has been shown to be present on the Golgi and to be activated there as a consequence of PM-based signalling. Different Ras family proteins localize to distinct subdomains of the PM (Prior et al, 2001) and/or to various intracellular compartments, including the Golgi, in a palmitoylation-dependent fashion (Choy et al, 1999; Michaelson et al, 2001; Rocks et al, 2005). Moreover, downstream elements of the Ras pathway have also been demonstrated to be located on the Golgi complex (Philips, 2004). Over the last few years, this has prompted a number of studies to address the question of whether these proteins can actually signal from endomembranes. In particular, Philips and co-workers (Chiu et al, 2002) investigated the pool of Ras localized on endomembranes using an in vivo probe for activated Ras. They concluded that stimulation with mitogens results in the activation of Ras both on the PM and on the Golgi complex, but with different kinetics and through different mechanisms (some of these findings have, however, been disputed by others (Augsten et al, 2006). Philips and colleagues have also shown that activation of H-Ras on the Golgi complex, but not on the PM, is dependent on Src-family kinases (Chiu et al, 2002). Thus, following Src-dependent activation of phospholipase Cγ (PLCγ), Ras guanine nucleotide releasing protein 1 (RasGRP1) translocates to the Golgi complex, where it activates Ras (Bivona and Philips, 2003). This pathway differs from that at the PM, indicating that distinct modes of activation and deactivation can regulate Ras in different subcellular compartments. This strategy might allow the cell to differentially regulate the activation of Ras in different cellular locations (Bivona and Philips, 2003).

As noted, not only Ras but also other components of the Ras pathway appear to be located on the Golgi complex. Recently, Nishida and co-workers (Torii et al, 2004) demonstrated the ‘spatial' modulatory activity of a novel scaffold protein, Sef, which is localized to the Golgi. Sef had been previously identified as a negative regulator of FGF signalling, and here it was shown to bind the active MEK/ERK complex, anchoring it to the Golgi complex (Sorkin, 2005). When ERK is trapped within this heterotrimeric complex, it can still phosphorylate one of its cytosolic substrates (RSK2), although it cannot dissociate and translocate to the nucleus to activate the array of nuclear ERK effectors (e.g. Elk-1). Reinforcing the concept that Sef works as a spatial modulator, it was shown that Golgi-entrapped ERK can still phosphorylate the nuclear target Elk-1 if the latter is made to move into the cytosol (Torii et al, 2004). What is still unknown is which proteins control the function of Sef and when they promote the switch from nuclear to cytosolic signalling of the ERK pathway. To complete the picture, Sef activation switches off the nuclear ERK pathway while activating the TAK1/JNK apoptotic pathway (Yang et al, 2004).

Finally, it has been shown that the nerve growth factor (NGF) receptor (TrkA) can localize to the Golgi complex, where it can be functionally activated, and thus impinge on the Ras pathway (Sorkin, 2005). Stimulation of the G protein-coupled receptors for adenosine or pituitary adenylate cyclase-activating polypeptide (PACAP) in neuronal cells has been shown to support cell survival via trans-activation of the pool of Golgi-based TrkA receptor. TrkA transactivation requires new protein synthesis and engages a population of immature receptors that are localized exclusively at the Golgi complex (Rajagopal et al, 2004). A BFA treatment, which results in Golgi disassembly and redistribution into the ER, completely prevented TrkA transactivation, while not affecting its stimulation via NGF from the PM, indicating the requirement of a preserved Golgi structure for this signalling pathway. Moreover, trans-activation of TrkA on the Golgi complex depends on Src and calcium ions (Rajagopal et al, 2004).

To date, these observations have been interpreted as indicating a role of the Ras signalling pool on the Golgi complex as a relay device for signalling cascades that are initiated at the PM and are involved in growth and proliferation (Figure 1D). There is no evidence so far for a link between Ras and the regulation of traffic. If this is so, the case of Ras might be different from that of the signalling pathways discussed above, which appeared to be involved in traffic regulation.

Physiological significance of endomembrane-based signalling and concluding remarks

In addition to the signalling pathways discussed above, a myriad of other signalling molecules of various kinds have been found to reside on secretory compartments, and in particular in the Golgi complex (Table 1). Most of these are only partially (or not yet) characterized in terms of their function at the Golgi, or of their precise subcompartment locations. This collective body of knowledge strongly suggests the existence of a hugely complex and rich endomembrane-based signalling network, the function and connectivity of which are only beginning to be glimpsed at.

To begin to bring order to this complex body of information, we propose the five generic model pathways illustrated in Figure 1. Evidence is so far available for the sequence in Figure 1C (activation of PM signals—signal translation— activation of Golgi signaling—regulation of traffic (De Matteis et al, 1993; Fabbri et al, 1994; Buccione et al, 1996) and Figure 1D (activation of PM signals—signal translation—activation of Golgi signaling—regulation of other functions (e.g. cell proliferation and motility) (Bivona and Philips, 2003). Moreover, for some G proteins, evidence has been produced that their Golgi location reflects their transit or cycling through the secretory pathway (Michaelson et al, 2002; Takida and Wedegaertner, 2004; Rocks et al, 2005) (Figure 1E). The formal demonstration of direct activation of a Golgi-based signalling pathway by a traffic event (Figure 1A and B) is instead still lacking, despite a large number of tantalizing observations that have indicated that such pathways exist. However, an example of initiation of signalling cascades on a secretory organelle by endogenous events is provided by the case of the ER stress responses (Patil and Walter, 2001; Zhang and Kaufman, 2004), whereby ER stresses can activate signalling pathways on the cytosolic surface of the ER, which then regulate survival-related and other cellular functions.

The overall picture discussed above defines almost by itself the tasks being called for. Much is known about the endomembrane localization and the effects on traffic of signalling molecules. What is largely lacking, however, is an understanding of how information flows through this maze of signalling molecules that would allow us to define complete specific pathways, along with their physiological significance and context. In particular, the question of the activation mechanisms of signalling complexes at trafficking endomembranes remains open. Let us take, for instance, the case of the heterotrimeric G proteins. How might they be activated at their intracellular sites? GPCR-like molecules that can bind G proteins on their cytosolic tail have been reported to reside on internal endomembranes (Boivin et al, 2003; Konger et al, 2005). Moreover, more than one type of GPCR-unrelated molecule has been reported that might in principle participate in the activation of G proteins at the Golgi complex (Blumer and Lanier, 2003). However, no information is available so far on possible functional links between any of these molecules and the Golgi heterotrimeric G proteins, nor is it known whether the initial activating stimuli are driven by extracellular signals or by endogenous (e.g. traffic initiated) events. Also, the identity of the G protein effectors on endomembranes remains unclear, although there is some evidence for the existence of traffic regulation by ‘classical' G-protein-effector pathways (as discussed above, and see Figure 2). Similar questions apply to other signalling pathways. The discovery of endogenous activation mechanisms that can initiate inter-organelle signalling would greatly advance our understanding of the coordination of the many intracellular activities involved in the maintenance of harmonic global cellular behaviour, a critical, yet largely unexplored, aspect of the physiology of traffic and of other cell functions.

Figure 2.

Representative Golgi-based signalling pathways. Several signalling complexes have been shown to be present on intracellular organelles. Here we depict the main elements of four representative pathways: the IgE receptor (IgEr) on the PM activates PKC, which promotes the recruitment of ARF1 and COPI to the Golgi complex via an unknown mediator. This pathway regulates the transport of cargo toward the PM. A hypothetical receptor (R?) with a potential intracellular location activates Gα_s that, in turn, activates adenylyl cyclase, to generate the cAMP that is necessary for activation of PKA at the Golgi complex. This active PKA phosphorylates the KDELr, uncovering the COPI-binding motif that is necessary for its retrograde transport. A hypothetical receptor (R?) activates a heterotrimeric G protein to release its βγ subunit. This generates diacylglycerol (DAG) through an unknown mechanism, which in turn activates PKCη. The subsequent phosphorylation and activation of PKD is required for membrane fission at the TGN. Stimulation of the EGF receptor (EGFr) produces via PLCγ the Ca²⁺ and DAG that are necessary for translocation of the Ras exchange factor Ras-GRP1 to the Golgi complex. Ras-GRP1 activates Ras on the Golgi, and the resulting activation of the MAP kinase (MEK/ERK) pathway supports basic cellular functions, such as cell growth and differentiation.