- Email: [email protected]
TGN38 and its orthologues: roles in post-TGN vesicle formation and maintenance of TGN morphology
Biochimica et Biophysica Acta 1355 Ž1997. 209–217
Review
TGN38 and its orthologues: roles in post-TGN vesicle formation and maintenance of TGN morphology G. Banting a , S. Ponnambalam a
b
b
Department of Biochemistry and BBSRC Funded Molecular Recognition Centre, UniÕersity of Bristol, UniÕersity Walk, Bristol, BS8 1TD, UK The Brunskill Laboratory, Department of Biochemistry, Medical Sciences Institute, UniÕersity of Dundee, Dundee DD1 4HN, UK Received 24 July 1996; revised 1 October 1996; accepted 9 October 1996
Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
210
2. Effects of Brefeldin A on the TGN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
210
3. Cycling of TGN38 between the TGN and the cell surface . . . . . . . . . . . . . . . . . . . . .
210
4. Internalisation of TGN38 from the cell surface . . . . . . . . . . . . . . . . . . . . . . . . . . .
212
5. Transmembrane localisation motifs in TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
213
6. Cytosolic proteins which interact with TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . .
213
7. Phosphorylation of the cytosolic domain of TGN38 . . . . . . . . . . . . . . . . . . . . . . . .
214
8. Orthologues of TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
214
9. Over-espression of TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
215
10. Model of the roleŽs. played by TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
216
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
216
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
216
0167-4889r97r$17.00 Copyright q 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 7 - 4 8 8 9 Ž 9 6 . 0 0 1 4 6 - 2
210
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
1. Introduction TGN38 is a type I integral membrane protein which is predominantly localised to the trans-Golgi network ŽTGN. of rodent cells. It was originally identified as a result of a screening strategy designed to isolate cDNA clones encoding organelle-specific integral membrane proteins w1x. Sequence analysis predicted a protein with a molecular mass of 38 kDa ŽFig. 1., but immunoblot analysis of whole cell lysate using a rabbit polyclonal antiserum raised against a plasmid-encoded b-galactosidase-TGN38 fusion protein identified a heterogenously glycosylated protein of 85–95 kDa w1x. This protein was later shown to be identical to the previously described protein GIMPt Žw2,3x, Reaves and Banting unpublished observations. and can thus be described as being heavily sialylated, both N- and O-glycosylated and with a half-life of approximately 9 h. Immunofluorescence analysis using the same antiserum generated a juxtanuclear pattern of staining characteristic of the stacked cisternae of the Golgi apparatus, whilst immuno-gold labelling and electron microscopic analysis of thin frozen sections demonstrated that TGN38 is localised to a tubulo-vesicular network at the trans side of the Golgi apparatus w1x. It had previously been shown
that, in cells infected with Vesicular Stomatitis Virus ŽVSV., the G protein of the virus accumulates in the TGN of cells incubated at 208C w4x; TGN38 co-localised with VSV-G in virally infected cells which had been incubated at 208C w1x. Thus, TGN38 provided the first endogenous marker for the TGN.
2. Effects of Brefeldin A on the TGN However, even with current advances in light and laser microscopy, it remains impossible to discriminate between membranes of the Golgi stack and those of the TGN Že.g. using antibodies to mannosidase II to decorate the medial cisternae of the Golgi stack and antibodies to TGN38 to decorate the TGN. by immunofluorescence microscopy alone; the resolution afforded by immuno-gold electron microscopy is required. This limitation of immunofluorescence microscopy can be circumvented by incubation of cells in the fungal metabolite brefeldin A Ž BFA. . It was observed by several groups that, in NRK cells, the membranes of the Golgi stack and those of the TGN respond differently to BFA w5–7x. The membrane and contents of the Golgi stack redistribute into the endoplasmic reticulum ŽER., whilst those of the TGN collapse upon the microtubule organising centre ŽMTOC.ŽFig. 2. w5–7x. Thus, even at the level of immunofluorescence microscopy it is possible to discriminate between membranes of the Golgi stack and those of the TGN.
3. Cycling of TGN38 between the TGN and the cell surface
Fig. 1. General features of TGN38 and its orthologues. See w39x for a more detailed sequence comparison.
Until the recent observation that newly synthesised transferrin receptors can pass through an endocytic compartment on their way to the plasma membrane w8x, the TGN might have been described as the final organelle through which newly-synthesised secretory proteins traffic. As it is, the TGN remains a major sorting station on the secretory pathway. It is from this organelle that proteins are delivered to lysosomes, constitutive secretory vesicles and regulated secretory vesicles. Is there any evidence that TGN38 may play a role in these processes? Yes, but nothing conclusive as yet.
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
211
Fig. 2. Dual label immunofluorescence analysis of methanol fixed NRK cells. Panels A–D represent cells incubated with a rabbit polyclonal antiserum vs. ratTGN38 and a mouse monoclonal antibody vs. mannosidase II as primary antibodies and a TRITC-conjugated swine anti-rabbit Ig Žpanels A and C. and a FITC-conjugated sheep anti-mouse Ig Žpanels B and D. as secondary antibodies. Panels E and F represent cells incubated with a rabbit polyclonal antiserum vs. ratTGN38 and a rat monoclonal antibody vs. tubulin as primary antibodies and a TRITC-conjugated swine anti-rabbit Ig Žpanel E. and a FITC-conjugated goat anti-mouse Ig Žpanel F. as secondary antibodies. Cells in panels A and B were incubated in the absence of BFA prior to fixation, cells in panels C-F were incubated in 5 m grml BFA for 2 h at 378C prior to fixation. Experimental procedures were as previously described w6x.
Although primarily localised to the TGN, TGN38 constitutively cycles between the TGN and the cell surface w7,9,10x with individual TGN38 molecules cycling rapidly. Gruenberg and colleagues showed that bafilomycin A1 ŽBafA1. Ža specific inhibitor of vacuolar ATPase. blocks the formation of endocytic carrier vesicles ŽECVs. ; structures normally generated as post early endosome compartments along the endocytic pathway w11x. It was subsequently shown that incubation of cells in the presence of BafA1 or
chloroquine Ž another perturbant of endosomal lumenal pH. led to an accumulation of TGN38 in swollen early endosomal structures w12,13x. The total population of TGN38 molecules was observed to be redistributed from the TGN to the swollen early endosomes within 45–60 minutes w12x, indicating that all TGN38 molecules complete the TGN to cell surface to TGN circuit at least once every hour. Thus, immuno-localisation studies provide an accurate representation of the steady-state distribution of TGN38,
212
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
but fail to indicate the highly dynamic nature of the protein in terms of its cyclic traffic between the TGN and the cell surface. Under control conditions in interphase cells there is very little TGN38 at the cell surface w6,7,9,10,14,15x; in fact, recent experiments suggest that only ; 1% of the total population of TGN38 molecules is at the cell surface at any one time ŽGirotti and Banting, unpublished observations.. However, the level of TGN38 at the cell surface increases approximately ten-fold in mitotic cells Ž Horn and Banting, unpublished observations. and in cells incubated in the presence of okadaic acid w15x. This is presumably because endocytosis is inhibited, but transport to the cell surface still occurs Žalbeit at a reduced rate., under these conditions w48x.
4. Internalisation of TGN38 from the cell surface The rapid cycling of TGN38, coupled with its low level of expression at the cell surface imply that it is efficiently internalised from the cell surface. Thus it was hardly surprising that it was detected in clathrin coated structures at the cell surface w7x and that sequences within its cytosolic domain are required for its TGN localisation w1,16–18x. The cytosolic domain of TGN38 was originally shown to be important for the correct localisation of the protein to the TGN, since in monkey ŽCOS. cells expressing wild type ratTGN38 the protein was localised to the TGN whilst in COS cells expressing ratTGN38 lacking its cytosolic domain the protein was expressed at the cell surface w1x. This observation not only showed the importance of the cytosolic domain of TGN38 in intracellular targeting, but also demonstrated that the machinery which recognises that cytosolic domain is conserved between rodents and primates. In addition, this observation led to a series of experiments in different labs in which different regions of the cytosolic domain of TGN38 were appended to the lumenal and transmembrane domains of various reporter proteins normally expressed at the cell surface. These hybrid proteins were expressed in transiently Žor, in some cases, stably. transfected heterologous cells and led to the identification of the hexapeptide ‘SDYQRL’ as being the motif required for internalisation from the cell surface and efficient return to the TGN w16–18x Žsee Fig. 1.; the tyrosine ŽY. and
leucine ŽL. residues being most critical w16–18x, with the serine ŽS. also playing a role w17x. Thus the internalisation motif in the cytosolic domain of TGN38 fell into the growing class of tyrosine-containing signals required for rapid internalisation from the cell surface via a clathrin mediated mechanism. Such motifs have also been suggested to play a role in the delivery of integral membrane proteins to the basolateral domain of polarised epithelial cells. It was therefore not surprising that when some of the hybrid constructs used to identify the ‘SDYQRL’ motif were expressed in polarised MDCK cells they were found to cycle between the TGN and the basolateral plasma membrane w19x. Interestingly, whilst structural analysis of synthetic peptides corresponding to some tyrosine-containing internalisation motifs has been interpreted as showing that the critical tyrosine residue lies at the end of a tight turn, data from similar studies performed on a synthetic peptide encompassing the ‘SDYQRL’ motif in TGN38 have been interpreted as showing that the critical tyrosine residue lies within a nascent helix w20x. Determining whether these are genuine structural differences or simply differences in experimental procedure and data interpretation awaits further study. However, Ohno et al. w21x used a yeast genetic screen and found that the ‘SDYQRL’ motif interacts directly with the m 2 sub-unit of the AP2 Žplasma membrane. adaptor complex. Mutational analysis showed that the tyrosine residue is necessary for this interaction to occur, and there is a preference for a bulky hydrophobic residue in the position occupied by the leucine. Mutations of the other residues in the motif had no significant effect on the interaction with m 2 w21x. During clathrin-mediated endocytosis from the cell surface, the clathrin cage is linked to the cytosolic domains of specific integral membrane proteins by a specific hetero-tetrameric complex designated AP2 w22x; one component of this complex is a 50 kDa protein Ž m 2.. The tyrosine residue in the ‘SDYQRL’ motif is needed for interaction with m 2 w21x providing a functional reason for its requirement in the internalisation of TGN38 from the cell surface. There is another adaptor complex ŽAP1. which links the clathrin cage to the cytosolic domains of integral membrane proteins in membranes destined to form vesicles which will be delivered to late endosomes andror lysosomes w22,23x. The equivalent of m 2 in
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
the AP1 hetero-tetramer is m1. It is of note that the ‘SDYQRL’ motif is also capable of binding to m1 w21x, albeit with an approximately ten-fold lower affinity than for m 2. The data from Ohno et al. are from in vitro experiments, but, if they represent the in vivo situation, would imply that TGN38 has the capacity to enter vesicles Žat the TGN. destined to be delivered to late endosomes andror lysosomes. This is in addition to the previously defined routing of TGN38 to the cell surface w9,10x. The potential for direct delivery to lysosomes coupled with a potentially less than 100% efficient trafficking of internalised TGN38 to the TGN Žas opposed to lysosomal delivery via late endosomes. would help to account for its relatively short Ž9 h. reported half-life w3x. However, potential interactions between TGN38 and the AP-1 Ž TGN. adaptor complex must be viewed with some caution since relatively weak, in vitro interactions may not accurately reflect the in vivo situation. The precise route followed by internalised TGN38 on its way back to the TGN has not been completely defined, but recent experimental data would imply that it is delivered to the TGN directly from early endosomes rather than from a later endocytic compartment w24x. Thus, whilst the PI-3-kinase inhibitor wortmannin blocks both the recycling of the transferrin receptor w25x and the delivery of the lysosomal integral membrane protein lgp120 from the cell surface to lysosomes w24x, it has no effect on the internalisation of TGN38 from the cell surface and its subsequent delivery to the TGN w24x.
5. Transmembrane localisation motifs in TGN38 Despite its undoubted importance, localisation of ratTGN38 to the TGN is not solely dependent on the ‘SDYQRL’ motif. Further experiments, in which the transmembrane domain of either CD4 or CD8 Žproteins normally expressed at the cell surface. was replaced with that of ratTGN38 and the chimeric protein transiently expressed in human ŽHeLa. cells, demonstrated that the transmembrane domain of ratTGN38 is sufficient to localise these chimeras to the TGN w26x. These data were supported by the observation that in stably transfected COS cells expressing mutant ratTGN38 lacking a cytosolic do-
213
main, the recombinant protein could still be detected in the TGN w27x. Thus, both retrieval Žthe ‘SDYQRL’ internalisation motif. and retention Žthe transmembrane domain. signals are involved in maintaining the steady state distribution of TGN38 w28x.
6. Cytosolic proteins which interact with TGN38 The mechanisms by which the previously described regions of the protein affect its localisation and intracellular itinerary remain to be elucidated. However, several cytosolic proteins have been shown to interact with the cytosolic domain of TGN38 w21,29x. The direct interaction with m 2 and m1 w21x has already been discussed. Immunoprecipitation of TGN38 in the presence of a zwitterionic detergent ŽCHAPS. indicated that the cytosolic domain of TGN38 interacts with a complex of proteins which includes small GTP-binding proteins Žone of which is rab6. and a 62 kDa cytosolic protein designated p62 w29x. This complex appears to possess PI-3-kinase activity, whilst p62 is reported to have sequence homology to the regulatory sub-unit of PI-3-kinase w30x. Data from an in vitro budding assay using a sub-cellular fraction enriched ; 20 fold for stacked Golgi Žmembranes and lumenal contents. suggests that TGN38 and cytosolic factors are required for post-TGN vesicle formation w29x. Phosphorylation of p62 regulates its association with the cytosolic domain of TGN38 w29,31x and specific residues within the cytosolic domain of TGN38 have been identified as being important for interaction with p62 w32x. In addition, electroporation of cells with antibodies to the cytosolic domain of TGN38 leads to a significant decrease in protein secretion w33,34x. Thus, formation of the p62rrab6rTGN38 cytosolic domain complex appears to be a pre-requisite for vesicle formation at exit sites from the TGN and underlies the functional importance of TGN38 in this part of the cell. It is also worthy of note that TGN38 is probably dimeric at this point in its recycling pathway w29x. Narula and Stow w35x have shown that another cytosolic protein Žp200. w36x can be localised to TGN membranes, and is associated with the cytosolic face of vesicles budding from TGN membranes under conditions which activate GTP-binding proteins w35x. These authors also showed, by immuno-gold la-
214
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
belling and electron microscopy on thin frozen cryosections, that there is an overlapping distribution of TGN38 and p200 on TGN membranes; thus raising the possibility that p200 and the cytosolic domain of TGN38 play a concerted role in the formation of a particular class of vesicles at the TGN. As is the case with antibodies to the cytosolic domain of TGN38, electroporation of cells with antibodies to p200 leads to a significant decrease in protein secretion w33,34x. These data have led to the suggestion that a complex involving p200, p62, a small GTP-binding protein, and a dimer of TGN38 is involved in the formation of exocytic transport vesicles at the TGN w34x. A growing number of coats and adaptor complexes w37x is being shown to be associated with TGN membranes; the minimal amount of data available so far would suggest that the cytosolic domain of TGN38 is a component of structures which assemble on the cytosolic face of TGN membranes as a prerequisite to the formation of particular types of post-TGN vesicles.
context that incubation of cells in the presence of okadaic acid Žan inhibitor of protein phosphatases 1 and 2A. leads to a ten-fold elevation of expression of TGN38 at the cell surface w15x. It is unclear whether this is by virtue of a direct or indirect effect upon TGN38, however data from recent experiments designed to address the role played by the serine residue which lies within the ‘SDYQRL’ internalisation motif in the cytosolic domain of TGN38 have shown that its mutation leads to an elevation in cell surface expression of TGN38 ŽRoquemore and Banting, unpublished observations.. These findings suggest that subtle phosphorylationrdephosphorylation events may act to regulate the movement of TGN38 to and from different intracellular compartments. Importantly, such mechanisms may also be a means to recruit andror disassemble the cytosolic TGN38-associated complexes which are required for vesicle biogenesis at the TGN.
8. Orthologues of TGN38 7. Phosphorylation of the cytosolic domain of TGN38 There are, as yet, no published data which show that TGN38 is phosphorylated in vivo. In fact, under its guise as GIMPt, it was shown not to be phosphorylated w3x. However, in vitro studies have shown that both serine residues and the single threonine residue in the cytosolic domain of TGN38 are substrates for phosphorylation by protein kinase A andror protein kinase C w31x. Phosphorylation of the membrane proximal serine residue regulates binding of p62 in vitro w31x. Thus reversible phosphorylation of specific residues in the cytosolic domain of TGN38 may regulate its intracellular trafficking; this would be analogous to the situation observed with furin Žanother type 1 integral membrane protein which is predominantly localised to the TGN, but cycles between the TGN and the cell surface. where reversible phosphorylation of its cytosolic domain by casein kinase II modulates its intracellular trafficking w47x. The cytosolic domain of TGN38 is, however, not a substrate for casein kinase II w31x; thus regulation of its intracellular trafficking is not via the same mechanism as that imposed upon furin. It should be noted in this
An isoform of TGN38 Ždesignated TGN41. was identified several years ago w38x, it behaves as TGN38 and is identical in sequence to TGN38 apart from the last three amino acids of TGN38 which are replaced by a stretch of 26 amino acids in TGN41 w38x. Only recently have cDNA clones encoding species homologues Žorthologues. of ratTGN38 been isolated w39,40x. No orthologue of TGN41 has been identified. Alignment of the conceptual translations of rat w1x, mouse Žtwo isoforms differing in the number of lumenal domain repeats-see below. w40x, rabbit, hamster, monkey and human w39x cDNA orthologues demonstrates that the sequences of transmembrane and cytosolic domains of the protein are highly conserved Žsee Fig. 3 for alignment of these regions., with absolute conservation of both the cytosolic ‘SDYQRL’ motif and the upstream serine Žwhich is supposedly involved in regulation of p62 binding.. This might have been predicted since these are the regions of the rat protein which have Ži. been shown to be required for the appropriate localisation of the protein and Žii. work as localisation signals in diverse species. However, the high degree of sequence conservation throughout these regions would suggest that molecules other than those already identified might
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
215
Fig. 3. Alignment of the predicted transmembrane domain and cytosolic domain sequences of ratTGN38 and its orthologues. Differences in amino acid sequence are indicated. The ‘stop transfer’ sequence Ždelineating the transmembrane domainrcytosolic domain boundary. is underlined, the conserved ‘SDYQRL’ internalisation motif is in bold. cDNA clones encompassing the entire coding sequences of human, macaque and mouse Žtwo isoforms. orthologues of ratTGN38 have been isolated. The different orthologues have been designated humTGN46, macTGN47 and MmTGN38 respectively w39x. The rabbit and hamster sequences are derived from PCR products generated using primers based upon the rat sequence w39x.
play roles in regulating the intracellular trafficking of TGN38 by virtue of sequencerstructure-dependent interactions. The more surprising observation which came out of comparison of the orthologous sequences was that there is very little homology between the bulk of the lumenal Žextracellular. domains of the proteins, in fact they show only 20–26% identity w39x. Both rodent and primate lumenal domains include direct repeat sequences; in rodents there are either six Ž rat and mouse isoform A. w1,40x or seven Ž mouse isoform B. w40x copies of an octapeptide repeat, whilst in primates there are fourteen or fifteen copies of a tetradecapeptide repeat w39x. There is little sequence similarity between the primate and rodent repeats, but both contain proline, serine and threonine residues. These features, coupled with the fact that both primate and rodent orthologues of TGN38 are heavily glycosylated w1,38,39x led to the suggestion that TGN38 and its orthologues may define a novel family of mucin-like proteins w39x. Features such as extensive N- and O-glycosylation, direct tandem repeats containing proline, serine and threonine residues, and divergence in the sequence of repeats between orthologues are also exhibited by the mucin gene products w41,42x. Intriguingly, another protein which has been described as being mucin-like Žfor the same reasons as those listed above. is the ligand for human P-selectin ŽPSGL-1. w43x. PSGL-1 is a type I integral membrane glycoprotein which contains fifteen direct copies of a decapeptide repeat; both the P-selectin repeat and the repeats in the primate orthologues of TGN38 contain the tetrapeptide sequence ‘EAQT’. The functional significance of this observation is not yet clear, but it may indicate a role for the lumenal domain of TGN38 and
its orthologues in protein:protein andr or protein:carbohydrate interactions. Perhaps some previously published data should be revisited in light of the observation that the lumenal domains of the different TGN38 orthologues are so different. Much of the work which led to the identification of the localisation motifs in ratTGN38 was performed using reporter constructs expressed in heterologous cell systems in which various regions of the transmembrane domain andror cytosolic domain of ratTGN38 were appended to the lumenal andror transmembrane domains of proteins normally resident at the cell surface w16–19,26x. These experiments failed to take any account of the possible role which the lumenal domain of TGN38 may play in the intracellular trafficking of the protein. Thus a hybrid protein whose sole TGN38 component is derived from the ratTGN38 cytosolic domain is primarily localised to the TGN of MDCK cell and cycles via the basolateral cell surface w19x, but this may not necessarily be the case for the endogenous canine orthologue of TGN38 in the same cells. It is conceivable that the lumenal domain of TGN38 may possess apical targeting information to complement the previously characterised basolateral targeting information in the cytosolic domain.
9. Over-expression of TGN38 It is also of note that studies on the localisation of ratTGN38 demonstrated that its elevated expression in monkey ŽCOS. or human ŽHeLa. cells led to a fragmentation of the TGN without any increase in cell surface expression w27x. Thus, the endocytic machinery has the capacity to accommodate a signifi-
216
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
cant increase in the level of expression of ratTGN38 in a heterologous cell system, but the TGN itself appears to be incapable of remaining intact under such circumstances. Intriguingly, when recombinant ratTGN38 is expressed at elevated levels in stably transfected rat ŽNRK. cells Žas a Green Fluorescent Protein-tagged construct. no fragmentation of the TGN is observed w44x. These observations have led to the suggestion that interactions between the lumenal domains of TGN38 molecules are required for maintenance of the morphology of the TGN w44x. Elevated levels of expression of the endogenous form of TGN38 do not lead to fragmentation of the TGN Žpresumably because compatible sets of repeat sequence are being expressed. , but elevated levels of expression of TGN38 orthologues does lead to fragmentation Žpresumably because non-compatible sets of lumenal domain repeats are being expressed in the same TGN..
10. Model of the role(s) played by TGN38 The precise roleŽs. played by TGN38 and its orthologues remains unclear. However the data published to date are consistent with a model in which the cytosolic domain of the protein is involved in the reversible recruitment to the membrane of cytosolic proteins involved in vesicle formation, whilst the lumenal domain is involved in maintaining the morphology of the TGN. The mucin-like nature of the lumenal domain of TGN38 would imply that it is highly hydrated, a feature which would facilitate concentration of the lumenal contents of the TGN. This might aid sorting of the contents of the TGN, with some glycoproteins being concentrated by association with the lumenal domain of TGN38 Žby virtue of a P-selectin:PSGL type interaction. and others being excluded from interaction with TGN38 and concentrated in the remaining aqueous milieu. It is of note that increased concentration is part of the mechanism which is used to sort proteins that enter the regulated secretory pathway w45x. Binding of ‘cargo’ to the lumenal domain of TGN38 Ž as proposed by Stanley and Howell, w46x. might then trigger recruitment of the p62rrab6 complex to the cytosolic domain of the protein, leading to the formation of transport vesicles bearing a specific sub-set of the
total population of proteins present in the TGN lumen. Such a model would require that ‘cargo’ binds to the lumenal domain of TGN38 in the TGN and is released after either vesicle formation or vesicle fusion with the cell surface. This reversible association could be accommodated by the different pH of the different environments. An extension of this argument is that, since the cytosolic domain of TGN38 acts as a basolateral targeting motif in polarised cells w19x, TGN38 may play a role in the targeted delivery of TGN lumenal contents to the basolateral surface of such cells. It is clear that, once TGN38 reaches the cell surface, the cytosolic domain of the protein is critical for its rapid internalisation and subsequent return to the TGN. Acknowledgements The authors would like to thank the CRC, AFRC, Wellcome Trust ŽGB. and MRC Ž SP and GB., for financial support. We are also grateful to Dr. K.E. Howell, Dr. B. Reaves, Dr. K. Stanley, Dr. E.P. Roquemore and Dr. D. Stephens for their critical contributions. References w1x Luzio, J.P., Brake, B., Banting, G., Howell, K.E., Braghetta, P. and Stanley, K.K. Ž1990. Biochem. J. 270, 97–102. w2x Sandoval, I.V. and Bakke, O. Ž1994. Trends Cell Biol. 4, 292–297. w3x Yuan, L., Barriocanal, J., Bonifacino, J. and Sandoval, I. Ž1987. J. Cell Biol. 105, 215–227. w4x Griffiths, G. and Simons, K. Ž1986. Science 234, 438–443. w5x Lippincott-Schwartz, J., Yuan, L., Tipper, C., Amherdt, M., Orci, L. and Klausner, R.D. Ž1991. Cell 67, 601–616. w6x Reaves, B. and Banting, G. Ž1992. J. Cell Biol. 116, 85–94. w7x Ladinsky, M.S. and Howell, K.E. Ž1992. EJCB 59, 92–105. w8x Futter, C.E., Connolly, C.N., Cutler, D.F. and Hopkins, C.R. Ž1995. J. Biol. Chem. 270, 10999–11003. w9x Reaves, B., Horn, M. and Banting, G. Ž1993. Mol. Biol. Cell 4, 93–105. w10x Ladinsky, M.S. and Howell, K.E. Ž1992. Mol. Biol. Cell 3, A 309–A 309. w11x Clague, M.J., Urbe, S., Aniento, F. and Gruenberg, J. Ž1994. J. Biol. Chem. 269, 21–24. w12x Reaves, B. and Banting, G. Ž1994. FEBS Letters 345, 61–66. w13x Chapman, R.E. and Munro, S. Ž1994. EMBO J. 13, 2305– 2312.
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217 w14x Ladinsky, M.S. and Howell, K.E. Ž1993. J. Cell Sci. S17, 41–47. w15x Horn, M. and Banting, G. Ž1994. Biochem. J. 301, 69–73. w16x Bos, K., Wraight, C. and Stanley, K.K. Ž1993. EMBO J. 12, 2219–2228. w17x Wong, S.H. and Hong, W.J. Ž1993. J. Biol. Chem. 268, 22853–22862. w18x Humphrey, J.S., Peters, P.J., Yuan, L.C. and Bonifacino, J.S. Ž1993. J. Cell Biol. 120, 1123–1135. w19x Rajasekaran, A.K., Humphrey, J.S., Wagner, M., Miesenbock, G., Lebivic, A., Bonifacino, J.S. and RodriguezBoulan, E. Ž1994. Mol. Cell Biol. 5, 1093–1103. w20x Wilde, A., Dempsey, C. and Banting, G. Ž1994. J. Biol. Chem. 269, 7131–7136. w21x Ohno, H., Stewart, J., Fournier, M.C., Bosshart, H., Rhee, I., Miyatake, S., Saito, T., Gallusser, A., Kirchhausen, T. and Bonifacino, J.S. Ž1995. Science 269, 1872–1875. w22x Pearse, B.M.F. and Robinson, M.S. Ž1990. Ann. Rev. Cell Biol. 6, 151–171. w23x Kornfeld, S. Ž1992. Annu. Rev. Biochem. 61, 307–330. w24x Reaves, B.J., Bright, N.A., Mullock, B.M. and Luzio, J.P. Ž1996. J. Cell Sci. 109, 749–762. w25x Shepherd, P.R., Soos, M.A. and Siddle, K. Ž1995. Biochem. Biophys. Res. Comun. 211, 535–539. w26x Ponnambalam, S., Rabouille, C., Luzio, J.P., Nilsson, T. and Warren, G. Ž1994. J. Cell Biol. 125, 253–268. w27x Reaves, B. and Banting, G. Ž1994. FEBS Letters 351, 448–456. w28x Luzio, J.P. and Banting, G. Ž1993. Trends Biochem. Sci. 18, 395–398. w29x Jones, S.M., Crosby, J.R., Salamero, J. and Howell, K.E. Ž1993. J. Cell Biol. 122, 775–788. w30x Jones, S.M. and Howell, K.E. Ž1995. Mol. Biol. Cell 6, 689. w31x Zehavifeferman, R., Burgess, J.W. and Stanley, K.K. Ž1995. FEBS Lett. 368, 122–124. w32x Stanley, K.K. Ž1996. Mol. Mem. Biol. 13, 19–27.
217
w33x Wang, J., Brown, J.L. and Howell, K.E. Ž1995. Mol. Biol. Cell 6, 1700. w34x Wang, J., Ladinsky, M.S. and Howell, K.E. Ž1995. Cold Spring Harbor Symposia on Quant. Biol. LX, 139–146. w35x Narula, N. and Stow, J.L. Ž1995. Proc. Natl. Acad. Sci. 92, 2874–2878. w36x Narula, N., McMorrow, I., Plopper, G., Doherty, J., Matlin, K.S., Burke, B. and Stow, J.L. Ž1992. J. Cell Biol. 117, 27–38. w37x Simpson, F., Bright, N., West, M., Newman, L.S., Darnell, R.B. and Robinson, M.S. Ž1996. J. Cell Biol. 133, 749–760. w38x Reaves, B., Wilde, A. and Banting, G. Ž1992. Biochem. J. 283, 313–316. w39x Ponnambalam, S., Girotti, M., Yaspo, M.-L., Owen, C.W., Perry, A.C.F., Suganuma, T., Nilsson, T., Fried, M., Banting, G. and Warren, G. Ž1996. J. Cell Sci. 109, 675–685. w40x Kasai, K., Takahashi, S., Murakami, K. and Nakayama, K. Ž1995. J. Biol. Chem. 270, 14471–14476. w41x Burchell, J. and Taylor-Papdimitriou, J. Ž1993. in Guidebook to the Extracellular Matrix and Adhesion Proteins ŽKreis, T. and Vale, R., eds.., pp. 73–74, Oxford University Press. w42x Devine, P.L. and McKenzie, F.C. Ž1992. BioEssays 14, 619–625. w43x Sako, D., Chang, X.J., Barone, K.M., Vachino, G., White, H.M., Shaw, G., Bean, K.M., Ahern, T.J., Furie, B., Cumming, D.A. and Larsen, G.R. Ž1993. Cell 75, 1179–1186. w44x Girotti, M. and Banting, G. Ž1996. J. Cell Sci., in press. w45x Bauerfeind, R. and Huttner, W.B. Ž1993. Curr. Op. Cell Biol. 5, 628–635. w46x Stanely, K.K. and Howell, K.E. Ž1993. Trends Cell Biol. 3, 252–255. w47x Jones, B.G., Thomas, J., Molloy, S.S., Thulin, C.D., Fry, M.D., Walsh, K.A. and Thomas, G. Ž1995. EMBO. J. 14, 5869–5883. w48x Kreiner, T. and Moore, H.P. Ž1990. Cell Regul. 1, 415–424.
Review
TGN38 and its orthologues: roles in post-TGN vesicle formation and maintenance of TGN morphology G. Banting a , S. Ponnambalam a
b
b
Department of Biochemistry and BBSRC Funded Molecular Recognition Centre, UniÕersity of Bristol, UniÕersity Walk, Bristol, BS8 1TD, UK The Brunskill Laboratory, Department of Biochemistry, Medical Sciences Institute, UniÕersity of Dundee, Dundee DD1 4HN, UK Received 24 July 1996; revised 1 October 1996; accepted 9 October 1996
Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
210
2. Effects of Brefeldin A on the TGN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
210
3. Cycling of TGN38 between the TGN and the cell surface . . . . . . . . . . . . . . . . . . . . .
210
4. Internalisation of TGN38 from the cell surface . . . . . . . . . . . . . . . . . . . . . . . . . . .
212
5. Transmembrane localisation motifs in TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
213
6. Cytosolic proteins which interact with TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . .
213
7. Phosphorylation of the cytosolic domain of TGN38 . . . . . . . . . . . . . . . . . . . . . . . .
214
8. Orthologues of TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
214
9. Over-espression of TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
215
10. Model of the roleŽs. played by TGN38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
216
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
216
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
216
0167-4889r97r$17.00 Copyright q 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 7 - 4 8 8 9 Ž 9 6 . 0 0 1 4 6 - 2
210
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
1. Introduction TGN38 is a type I integral membrane protein which is predominantly localised to the trans-Golgi network ŽTGN. of rodent cells. It was originally identified as a result of a screening strategy designed to isolate cDNA clones encoding organelle-specific integral membrane proteins w1x. Sequence analysis predicted a protein with a molecular mass of 38 kDa ŽFig. 1., but immunoblot analysis of whole cell lysate using a rabbit polyclonal antiserum raised against a plasmid-encoded b-galactosidase-TGN38 fusion protein identified a heterogenously glycosylated protein of 85–95 kDa w1x. This protein was later shown to be identical to the previously described protein GIMPt Žw2,3x, Reaves and Banting unpublished observations. and can thus be described as being heavily sialylated, both N- and O-glycosylated and with a half-life of approximately 9 h. Immunofluorescence analysis using the same antiserum generated a juxtanuclear pattern of staining characteristic of the stacked cisternae of the Golgi apparatus, whilst immuno-gold labelling and electron microscopic analysis of thin frozen sections demonstrated that TGN38 is localised to a tubulo-vesicular network at the trans side of the Golgi apparatus w1x. It had previously been shown
that, in cells infected with Vesicular Stomatitis Virus ŽVSV., the G protein of the virus accumulates in the TGN of cells incubated at 208C w4x; TGN38 co-localised with VSV-G in virally infected cells which had been incubated at 208C w1x. Thus, TGN38 provided the first endogenous marker for the TGN.
2. Effects of Brefeldin A on the TGN However, even with current advances in light and laser microscopy, it remains impossible to discriminate between membranes of the Golgi stack and those of the TGN Že.g. using antibodies to mannosidase II to decorate the medial cisternae of the Golgi stack and antibodies to TGN38 to decorate the TGN. by immunofluorescence microscopy alone; the resolution afforded by immuno-gold electron microscopy is required. This limitation of immunofluorescence microscopy can be circumvented by incubation of cells in the fungal metabolite brefeldin A Ž BFA. . It was observed by several groups that, in NRK cells, the membranes of the Golgi stack and those of the TGN respond differently to BFA w5–7x. The membrane and contents of the Golgi stack redistribute into the endoplasmic reticulum ŽER., whilst those of the TGN collapse upon the microtubule organising centre ŽMTOC.ŽFig. 2. w5–7x. Thus, even at the level of immunofluorescence microscopy it is possible to discriminate between membranes of the Golgi stack and those of the TGN.
3. Cycling of TGN38 between the TGN and the cell surface
Fig. 1. General features of TGN38 and its orthologues. See w39x for a more detailed sequence comparison.
Until the recent observation that newly synthesised transferrin receptors can pass through an endocytic compartment on their way to the plasma membrane w8x, the TGN might have been described as the final organelle through which newly-synthesised secretory proteins traffic. As it is, the TGN remains a major sorting station on the secretory pathway. It is from this organelle that proteins are delivered to lysosomes, constitutive secretory vesicles and regulated secretory vesicles. Is there any evidence that TGN38 may play a role in these processes? Yes, but nothing conclusive as yet.
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
211
Fig. 2. Dual label immunofluorescence analysis of methanol fixed NRK cells. Panels A–D represent cells incubated with a rabbit polyclonal antiserum vs. ratTGN38 and a mouse monoclonal antibody vs. mannosidase II as primary antibodies and a TRITC-conjugated swine anti-rabbit Ig Žpanels A and C. and a FITC-conjugated sheep anti-mouse Ig Žpanels B and D. as secondary antibodies. Panels E and F represent cells incubated with a rabbit polyclonal antiserum vs. ratTGN38 and a rat monoclonal antibody vs. tubulin as primary antibodies and a TRITC-conjugated swine anti-rabbit Ig Žpanel E. and a FITC-conjugated goat anti-mouse Ig Žpanel F. as secondary antibodies. Cells in panels A and B were incubated in the absence of BFA prior to fixation, cells in panels C-F were incubated in 5 m grml BFA for 2 h at 378C prior to fixation. Experimental procedures were as previously described w6x.
Although primarily localised to the TGN, TGN38 constitutively cycles between the TGN and the cell surface w7,9,10x with individual TGN38 molecules cycling rapidly. Gruenberg and colleagues showed that bafilomycin A1 ŽBafA1. Ža specific inhibitor of vacuolar ATPase. blocks the formation of endocytic carrier vesicles ŽECVs. ; structures normally generated as post early endosome compartments along the endocytic pathway w11x. It was subsequently shown that incubation of cells in the presence of BafA1 or
chloroquine Ž another perturbant of endosomal lumenal pH. led to an accumulation of TGN38 in swollen early endosomal structures w12,13x. The total population of TGN38 molecules was observed to be redistributed from the TGN to the swollen early endosomes within 45–60 minutes w12x, indicating that all TGN38 molecules complete the TGN to cell surface to TGN circuit at least once every hour. Thus, immuno-localisation studies provide an accurate representation of the steady-state distribution of TGN38,
212
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
but fail to indicate the highly dynamic nature of the protein in terms of its cyclic traffic between the TGN and the cell surface. Under control conditions in interphase cells there is very little TGN38 at the cell surface w6,7,9,10,14,15x; in fact, recent experiments suggest that only ; 1% of the total population of TGN38 molecules is at the cell surface at any one time ŽGirotti and Banting, unpublished observations.. However, the level of TGN38 at the cell surface increases approximately ten-fold in mitotic cells Ž Horn and Banting, unpublished observations. and in cells incubated in the presence of okadaic acid w15x. This is presumably because endocytosis is inhibited, but transport to the cell surface still occurs Žalbeit at a reduced rate., under these conditions w48x.
4. Internalisation of TGN38 from the cell surface The rapid cycling of TGN38, coupled with its low level of expression at the cell surface imply that it is efficiently internalised from the cell surface. Thus it was hardly surprising that it was detected in clathrin coated structures at the cell surface w7x and that sequences within its cytosolic domain are required for its TGN localisation w1,16–18x. The cytosolic domain of TGN38 was originally shown to be important for the correct localisation of the protein to the TGN, since in monkey ŽCOS. cells expressing wild type ratTGN38 the protein was localised to the TGN whilst in COS cells expressing ratTGN38 lacking its cytosolic domain the protein was expressed at the cell surface w1x. This observation not only showed the importance of the cytosolic domain of TGN38 in intracellular targeting, but also demonstrated that the machinery which recognises that cytosolic domain is conserved between rodents and primates. In addition, this observation led to a series of experiments in different labs in which different regions of the cytosolic domain of TGN38 were appended to the lumenal and transmembrane domains of various reporter proteins normally expressed at the cell surface. These hybrid proteins were expressed in transiently Žor, in some cases, stably. transfected heterologous cells and led to the identification of the hexapeptide ‘SDYQRL’ as being the motif required for internalisation from the cell surface and efficient return to the TGN w16–18x Žsee Fig. 1.; the tyrosine ŽY. and
leucine ŽL. residues being most critical w16–18x, with the serine ŽS. also playing a role w17x. Thus the internalisation motif in the cytosolic domain of TGN38 fell into the growing class of tyrosine-containing signals required for rapid internalisation from the cell surface via a clathrin mediated mechanism. Such motifs have also been suggested to play a role in the delivery of integral membrane proteins to the basolateral domain of polarised epithelial cells. It was therefore not surprising that when some of the hybrid constructs used to identify the ‘SDYQRL’ motif were expressed in polarised MDCK cells they were found to cycle between the TGN and the basolateral plasma membrane w19x. Interestingly, whilst structural analysis of synthetic peptides corresponding to some tyrosine-containing internalisation motifs has been interpreted as showing that the critical tyrosine residue lies at the end of a tight turn, data from similar studies performed on a synthetic peptide encompassing the ‘SDYQRL’ motif in TGN38 have been interpreted as showing that the critical tyrosine residue lies within a nascent helix w20x. Determining whether these are genuine structural differences or simply differences in experimental procedure and data interpretation awaits further study. However, Ohno et al. w21x used a yeast genetic screen and found that the ‘SDYQRL’ motif interacts directly with the m 2 sub-unit of the AP2 Žplasma membrane. adaptor complex. Mutational analysis showed that the tyrosine residue is necessary for this interaction to occur, and there is a preference for a bulky hydrophobic residue in the position occupied by the leucine. Mutations of the other residues in the motif had no significant effect on the interaction with m 2 w21x. During clathrin-mediated endocytosis from the cell surface, the clathrin cage is linked to the cytosolic domains of specific integral membrane proteins by a specific hetero-tetrameric complex designated AP2 w22x; one component of this complex is a 50 kDa protein Ž m 2.. The tyrosine residue in the ‘SDYQRL’ motif is needed for interaction with m 2 w21x providing a functional reason for its requirement in the internalisation of TGN38 from the cell surface. There is another adaptor complex ŽAP1. which links the clathrin cage to the cytosolic domains of integral membrane proteins in membranes destined to form vesicles which will be delivered to late endosomes andror lysosomes w22,23x. The equivalent of m 2 in
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
the AP1 hetero-tetramer is m1. It is of note that the ‘SDYQRL’ motif is also capable of binding to m1 w21x, albeit with an approximately ten-fold lower affinity than for m 2. The data from Ohno et al. are from in vitro experiments, but, if they represent the in vivo situation, would imply that TGN38 has the capacity to enter vesicles Žat the TGN. destined to be delivered to late endosomes andror lysosomes. This is in addition to the previously defined routing of TGN38 to the cell surface w9,10x. The potential for direct delivery to lysosomes coupled with a potentially less than 100% efficient trafficking of internalised TGN38 to the TGN Žas opposed to lysosomal delivery via late endosomes. would help to account for its relatively short Ž9 h. reported half-life w3x. However, potential interactions between TGN38 and the AP-1 Ž TGN. adaptor complex must be viewed with some caution since relatively weak, in vitro interactions may not accurately reflect the in vivo situation. The precise route followed by internalised TGN38 on its way back to the TGN has not been completely defined, but recent experimental data would imply that it is delivered to the TGN directly from early endosomes rather than from a later endocytic compartment w24x. Thus, whilst the PI-3-kinase inhibitor wortmannin blocks both the recycling of the transferrin receptor w25x and the delivery of the lysosomal integral membrane protein lgp120 from the cell surface to lysosomes w24x, it has no effect on the internalisation of TGN38 from the cell surface and its subsequent delivery to the TGN w24x.
5. Transmembrane localisation motifs in TGN38 Despite its undoubted importance, localisation of ratTGN38 to the TGN is not solely dependent on the ‘SDYQRL’ motif. Further experiments, in which the transmembrane domain of either CD4 or CD8 Žproteins normally expressed at the cell surface. was replaced with that of ratTGN38 and the chimeric protein transiently expressed in human ŽHeLa. cells, demonstrated that the transmembrane domain of ratTGN38 is sufficient to localise these chimeras to the TGN w26x. These data were supported by the observation that in stably transfected COS cells expressing mutant ratTGN38 lacking a cytosolic do-
213
main, the recombinant protein could still be detected in the TGN w27x. Thus, both retrieval Žthe ‘SDYQRL’ internalisation motif. and retention Žthe transmembrane domain. signals are involved in maintaining the steady state distribution of TGN38 w28x.
6. Cytosolic proteins which interact with TGN38 The mechanisms by which the previously described regions of the protein affect its localisation and intracellular itinerary remain to be elucidated. However, several cytosolic proteins have been shown to interact with the cytosolic domain of TGN38 w21,29x. The direct interaction with m 2 and m1 w21x has already been discussed. Immunoprecipitation of TGN38 in the presence of a zwitterionic detergent ŽCHAPS. indicated that the cytosolic domain of TGN38 interacts with a complex of proteins which includes small GTP-binding proteins Žone of which is rab6. and a 62 kDa cytosolic protein designated p62 w29x. This complex appears to possess PI-3-kinase activity, whilst p62 is reported to have sequence homology to the regulatory sub-unit of PI-3-kinase w30x. Data from an in vitro budding assay using a sub-cellular fraction enriched ; 20 fold for stacked Golgi Žmembranes and lumenal contents. suggests that TGN38 and cytosolic factors are required for post-TGN vesicle formation w29x. Phosphorylation of p62 regulates its association with the cytosolic domain of TGN38 w29,31x and specific residues within the cytosolic domain of TGN38 have been identified as being important for interaction with p62 w32x. In addition, electroporation of cells with antibodies to the cytosolic domain of TGN38 leads to a significant decrease in protein secretion w33,34x. Thus, formation of the p62rrab6rTGN38 cytosolic domain complex appears to be a pre-requisite for vesicle formation at exit sites from the TGN and underlies the functional importance of TGN38 in this part of the cell. It is also worthy of note that TGN38 is probably dimeric at this point in its recycling pathway w29x. Narula and Stow w35x have shown that another cytosolic protein Žp200. w36x can be localised to TGN membranes, and is associated with the cytosolic face of vesicles budding from TGN membranes under conditions which activate GTP-binding proteins w35x. These authors also showed, by immuno-gold la-
214
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
belling and electron microscopy on thin frozen cryosections, that there is an overlapping distribution of TGN38 and p200 on TGN membranes; thus raising the possibility that p200 and the cytosolic domain of TGN38 play a concerted role in the formation of a particular class of vesicles at the TGN. As is the case with antibodies to the cytosolic domain of TGN38, electroporation of cells with antibodies to p200 leads to a significant decrease in protein secretion w33,34x. These data have led to the suggestion that a complex involving p200, p62, a small GTP-binding protein, and a dimer of TGN38 is involved in the formation of exocytic transport vesicles at the TGN w34x. A growing number of coats and adaptor complexes w37x is being shown to be associated with TGN membranes; the minimal amount of data available so far would suggest that the cytosolic domain of TGN38 is a component of structures which assemble on the cytosolic face of TGN membranes as a prerequisite to the formation of particular types of post-TGN vesicles.
context that incubation of cells in the presence of okadaic acid Žan inhibitor of protein phosphatases 1 and 2A. leads to a ten-fold elevation of expression of TGN38 at the cell surface w15x. It is unclear whether this is by virtue of a direct or indirect effect upon TGN38, however data from recent experiments designed to address the role played by the serine residue which lies within the ‘SDYQRL’ internalisation motif in the cytosolic domain of TGN38 have shown that its mutation leads to an elevation in cell surface expression of TGN38 ŽRoquemore and Banting, unpublished observations.. These findings suggest that subtle phosphorylationrdephosphorylation events may act to regulate the movement of TGN38 to and from different intracellular compartments. Importantly, such mechanisms may also be a means to recruit andror disassemble the cytosolic TGN38-associated complexes which are required for vesicle biogenesis at the TGN.
8. Orthologues of TGN38 7. Phosphorylation of the cytosolic domain of TGN38 There are, as yet, no published data which show that TGN38 is phosphorylated in vivo. In fact, under its guise as GIMPt, it was shown not to be phosphorylated w3x. However, in vitro studies have shown that both serine residues and the single threonine residue in the cytosolic domain of TGN38 are substrates for phosphorylation by protein kinase A andror protein kinase C w31x. Phosphorylation of the membrane proximal serine residue regulates binding of p62 in vitro w31x. Thus reversible phosphorylation of specific residues in the cytosolic domain of TGN38 may regulate its intracellular trafficking; this would be analogous to the situation observed with furin Žanother type 1 integral membrane protein which is predominantly localised to the TGN, but cycles between the TGN and the cell surface. where reversible phosphorylation of its cytosolic domain by casein kinase II modulates its intracellular trafficking w47x. The cytosolic domain of TGN38 is, however, not a substrate for casein kinase II w31x; thus regulation of its intracellular trafficking is not via the same mechanism as that imposed upon furin. It should be noted in this
An isoform of TGN38 Ždesignated TGN41. was identified several years ago w38x, it behaves as TGN38 and is identical in sequence to TGN38 apart from the last three amino acids of TGN38 which are replaced by a stretch of 26 amino acids in TGN41 w38x. Only recently have cDNA clones encoding species homologues Žorthologues. of ratTGN38 been isolated w39,40x. No orthologue of TGN41 has been identified. Alignment of the conceptual translations of rat w1x, mouse Žtwo isoforms differing in the number of lumenal domain repeats-see below. w40x, rabbit, hamster, monkey and human w39x cDNA orthologues demonstrates that the sequences of transmembrane and cytosolic domains of the protein are highly conserved Žsee Fig. 3 for alignment of these regions., with absolute conservation of both the cytosolic ‘SDYQRL’ motif and the upstream serine Žwhich is supposedly involved in regulation of p62 binding.. This might have been predicted since these are the regions of the rat protein which have Ži. been shown to be required for the appropriate localisation of the protein and Žii. work as localisation signals in diverse species. However, the high degree of sequence conservation throughout these regions would suggest that molecules other than those already identified might
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
215
Fig. 3. Alignment of the predicted transmembrane domain and cytosolic domain sequences of ratTGN38 and its orthologues. Differences in amino acid sequence are indicated. The ‘stop transfer’ sequence Ždelineating the transmembrane domainrcytosolic domain boundary. is underlined, the conserved ‘SDYQRL’ internalisation motif is in bold. cDNA clones encompassing the entire coding sequences of human, macaque and mouse Žtwo isoforms. orthologues of ratTGN38 have been isolated. The different orthologues have been designated humTGN46, macTGN47 and MmTGN38 respectively w39x. The rabbit and hamster sequences are derived from PCR products generated using primers based upon the rat sequence w39x.
play roles in regulating the intracellular trafficking of TGN38 by virtue of sequencerstructure-dependent interactions. The more surprising observation which came out of comparison of the orthologous sequences was that there is very little homology between the bulk of the lumenal Žextracellular. domains of the proteins, in fact they show only 20–26% identity w39x. Both rodent and primate lumenal domains include direct repeat sequences; in rodents there are either six Ž rat and mouse isoform A. w1,40x or seven Ž mouse isoform B. w40x copies of an octapeptide repeat, whilst in primates there are fourteen or fifteen copies of a tetradecapeptide repeat w39x. There is little sequence similarity between the primate and rodent repeats, but both contain proline, serine and threonine residues. These features, coupled with the fact that both primate and rodent orthologues of TGN38 are heavily glycosylated w1,38,39x led to the suggestion that TGN38 and its orthologues may define a novel family of mucin-like proteins w39x. Features such as extensive N- and O-glycosylation, direct tandem repeats containing proline, serine and threonine residues, and divergence in the sequence of repeats between orthologues are also exhibited by the mucin gene products w41,42x. Intriguingly, another protein which has been described as being mucin-like Žfor the same reasons as those listed above. is the ligand for human P-selectin ŽPSGL-1. w43x. PSGL-1 is a type I integral membrane glycoprotein which contains fifteen direct copies of a decapeptide repeat; both the P-selectin repeat and the repeats in the primate orthologues of TGN38 contain the tetrapeptide sequence ‘EAQT’. The functional significance of this observation is not yet clear, but it may indicate a role for the lumenal domain of TGN38 and
its orthologues in protein:protein andr or protein:carbohydrate interactions. Perhaps some previously published data should be revisited in light of the observation that the lumenal domains of the different TGN38 orthologues are so different. Much of the work which led to the identification of the localisation motifs in ratTGN38 was performed using reporter constructs expressed in heterologous cell systems in which various regions of the transmembrane domain andror cytosolic domain of ratTGN38 were appended to the lumenal andror transmembrane domains of proteins normally resident at the cell surface w16–19,26x. These experiments failed to take any account of the possible role which the lumenal domain of TGN38 may play in the intracellular trafficking of the protein. Thus a hybrid protein whose sole TGN38 component is derived from the ratTGN38 cytosolic domain is primarily localised to the TGN of MDCK cell and cycles via the basolateral cell surface w19x, but this may not necessarily be the case for the endogenous canine orthologue of TGN38 in the same cells. It is conceivable that the lumenal domain of TGN38 may possess apical targeting information to complement the previously characterised basolateral targeting information in the cytosolic domain.
9. Over-expression of TGN38 It is also of note that studies on the localisation of ratTGN38 demonstrated that its elevated expression in monkey ŽCOS. or human ŽHeLa. cells led to a fragmentation of the TGN without any increase in cell surface expression w27x. Thus, the endocytic machinery has the capacity to accommodate a signifi-
216
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217
cant increase in the level of expression of ratTGN38 in a heterologous cell system, but the TGN itself appears to be incapable of remaining intact under such circumstances. Intriguingly, when recombinant ratTGN38 is expressed at elevated levels in stably transfected rat ŽNRK. cells Žas a Green Fluorescent Protein-tagged construct. no fragmentation of the TGN is observed w44x. These observations have led to the suggestion that interactions between the lumenal domains of TGN38 molecules are required for maintenance of the morphology of the TGN w44x. Elevated levels of expression of the endogenous form of TGN38 do not lead to fragmentation of the TGN Žpresumably because compatible sets of repeat sequence are being expressed. , but elevated levels of expression of TGN38 orthologues does lead to fragmentation Žpresumably because non-compatible sets of lumenal domain repeats are being expressed in the same TGN..
10. Model of the role(s) played by TGN38 The precise roleŽs. played by TGN38 and its orthologues remains unclear. However the data published to date are consistent with a model in which the cytosolic domain of the protein is involved in the reversible recruitment to the membrane of cytosolic proteins involved in vesicle formation, whilst the lumenal domain is involved in maintaining the morphology of the TGN. The mucin-like nature of the lumenal domain of TGN38 would imply that it is highly hydrated, a feature which would facilitate concentration of the lumenal contents of the TGN. This might aid sorting of the contents of the TGN, with some glycoproteins being concentrated by association with the lumenal domain of TGN38 Žby virtue of a P-selectin:PSGL type interaction. and others being excluded from interaction with TGN38 and concentrated in the remaining aqueous milieu. It is of note that increased concentration is part of the mechanism which is used to sort proteins that enter the regulated secretory pathway w45x. Binding of ‘cargo’ to the lumenal domain of TGN38 Ž as proposed by Stanley and Howell, w46x. might then trigger recruitment of the p62rrab6 complex to the cytosolic domain of the protein, leading to the formation of transport vesicles bearing a specific sub-set of the
total population of proteins present in the TGN lumen. Such a model would require that ‘cargo’ binds to the lumenal domain of TGN38 in the TGN and is released after either vesicle formation or vesicle fusion with the cell surface. This reversible association could be accommodated by the different pH of the different environments. An extension of this argument is that, since the cytosolic domain of TGN38 acts as a basolateral targeting motif in polarised cells w19x, TGN38 may play a role in the targeted delivery of TGN lumenal contents to the basolateral surface of such cells. It is clear that, once TGN38 reaches the cell surface, the cytosolic domain of the protein is critical for its rapid internalisation and subsequent return to the TGN. Acknowledgements The authors would like to thank the CRC, AFRC, Wellcome Trust ŽGB. and MRC Ž SP and GB., for financial support. We are also grateful to Dr. K.E. Howell, Dr. B. Reaves, Dr. K. Stanley, Dr. E.P. Roquemore and Dr. D. Stephens for their critical contributions. References w1x Luzio, J.P., Brake, B., Banting, G., Howell, K.E., Braghetta, P. and Stanley, K.K. Ž1990. Biochem. J. 270, 97–102. w2x Sandoval, I.V. and Bakke, O. Ž1994. Trends Cell Biol. 4, 292–297. w3x Yuan, L., Barriocanal, J., Bonifacino, J. and Sandoval, I. Ž1987. J. Cell Biol. 105, 215–227. w4x Griffiths, G. and Simons, K. Ž1986. Science 234, 438–443. w5x Lippincott-Schwartz, J., Yuan, L., Tipper, C., Amherdt, M., Orci, L. and Klausner, R.D. Ž1991. Cell 67, 601–616. w6x Reaves, B. and Banting, G. Ž1992. J. Cell Biol. 116, 85–94. w7x Ladinsky, M.S. and Howell, K.E. Ž1992. EJCB 59, 92–105. w8x Futter, C.E., Connolly, C.N., Cutler, D.F. and Hopkins, C.R. Ž1995. J. Biol. Chem. 270, 10999–11003. w9x Reaves, B., Horn, M. and Banting, G. Ž1993. Mol. Biol. Cell 4, 93–105. w10x Ladinsky, M.S. and Howell, K.E. Ž1992. Mol. Biol. Cell 3, A 309–A 309. w11x Clague, M.J., Urbe, S., Aniento, F. and Gruenberg, J. Ž1994. J. Biol. Chem. 269, 21–24. w12x Reaves, B. and Banting, G. Ž1994. FEBS Letters 345, 61–66. w13x Chapman, R.E. and Munro, S. Ž1994. EMBO J. 13, 2305– 2312.
G. Banting, S. Ponnambalamr Biochimica et Biophysica Acta 1355 (1997) 209–217 w14x Ladinsky, M.S. and Howell, K.E. Ž1993. J. Cell Sci. S17, 41–47. w15x Horn, M. and Banting, G. Ž1994. Biochem. J. 301, 69–73. w16x Bos, K., Wraight, C. and Stanley, K.K. Ž1993. EMBO J. 12, 2219–2228. w17x Wong, S.H. and Hong, W.J. Ž1993. J. Biol. Chem. 268, 22853–22862. w18x Humphrey, J.S., Peters, P.J., Yuan, L.C. and Bonifacino, J.S. Ž1993. J. Cell Biol. 120, 1123–1135. w19x Rajasekaran, A.K., Humphrey, J.S., Wagner, M., Miesenbock, G., Lebivic, A., Bonifacino, J.S. and RodriguezBoulan, E. Ž1994. Mol. Cell Biol. 5, 1093–1103. w20x Wilde, A., Dempsey, C. and Banting, G. Ž1994. J. Biol. Chem. 269, 7131–7136. w21x Ohno, H., Stewart, J., Fournier, M.C., Bosshart, H., Rhee, I., Miyatake, S., Saito, T., Gallusser, A., Kirchhausen, T. and Bonifacino, J.S. Ž1995. Science 269, 1872–1875. w22x Pearse, B.M.F. and Robinson, M.S. Ž1990. Ann. Rev. Cell Biol. 6, 151–171. w23x Kornfeld, S. Ž1992. Annu. Rev. Biochem. 61, 307–330. w24x Reaves, B.J., Bright, N.A., Mullock, B.M. and Luzio, J.P. Ž1996. J. Cell Sci. 109, 749–762. w25x Shepherd, P.R., Soos, M.A. and Siddle, K. Ž1995. Biochem. Biophys. Res. Comun. 211, 535–539. w26x Ponnambalam, S., Rabouille, C., Luzio, J.P., Nilsson, T. and Warren, G. Ž1994. J. Cell Biol. 125, 253–268. w27x Reaves, B. and Banting, G. Ž1994. FEBS Letters 351, 448–456. w28x Luzio, J.P. and Banting, G. Ž1993. Trends Biochem. Sci. 18, 395–398. w29x Jones, S.M., Crosby, J.R., Salamero, J. and Howell, K.E. Ž1993. J. Cell Biol. 122, 775–788. w30x Jones, S.M. and Howell, K.E. Ž1995. Mol. Biol. Cell 6, 689. w31x Zehavifeferman, R., Burgess, J.W. and Stanley, K.K. Ž1995. FEBS Lett. 368, 122–124. w32x Stanley, K.K. Ž1996. Mol. Mem. Biol. 13, 19–27.
217
w33x Wang, J., Brown, J.L. and Howell, K.E. Ž1995. Mol. Biol. Cell 6, 1700. w34x Wang, J., Ladinsky, M.S. and Howell, K.E. Ž1995. Cold Spring Harbor Symposia on Quant. Biol. LX, 139–146. w35x Narula, N. and Stow, J.L. Ž1995. Proc. Natl. Acad. Sci. 92, 2874–2878. w36x Narula, N., McMorrow, I., Plopper, G., Doherty, J., Matlin, K.S., Burke, B. and Stow, J.L. Ž1992. J. Cell Biol. 117, 27–38. w37x Simpson, F., Bright, N., West, M., Newman, L.S., Darnell, R.B. and Robinson, M.S. Ž1996. J. Cell Biol. 133, 749–760. w38x Reaves, B., Wilde, A. and Banting, G. Ž1992. Biochem. J. 283, 313–316. w39x Ponnambalam, S., Girotti, M., Yaspo, M.-L., Owen, C.W., Perry, A.C.F., Suganuma, T., Nilsson, T., Fried, M., Banting, G. and Warren, G. Ž1996. J. Cell Sci. 109, 675–685. w40x Kasai, K., Takahashi, S., Murakami, K. and Nakayama, K. Ž1995. J. Biol. Chem. 270, 14471–14476. w41x Burchell, J. and Taylor-Papdimitriou, J. Ž1993. in Guidebook to the Extracellular Matrix and Adhesion Proteins ŽKreis, T. and Vale, R., eds.., pp. 73–74, Oxford University Press. w42x Devine, P.L. and McKenzie, F.C. Ž1992. BioEssays 14, 619–625. w43x Sako, D., Chang, X.J., Barone, K.M., Vachino, G., White, H.M., Shaw, G., Bean, K.M., Ahern, T.J., Furie, B., Cumming, D.A. and Larsen, G.R. Ž1993. Cell 75, 1179–1186. w44x Girotti, M. and Banting, G. Ž1996. J. Cell Sci., in press. w45x Bauerfeind, R. and Huttner, W.B. Ž1993. Curr. Op. Cell Biol. 5, 628–635. w46x Stanely, K.K. and Howell, K.E. Ž1993. Trends Cell Biol. 3, 252–255. w47x Jones, B.G., Thomas, J., Molloy, S.S., Thulin, C.D., Fry, M.D., Walsh, K.A. and Thomas, G. Ž1995. EMBO. J. 14, 5869–5883. w48x Kreiner, T. and Moore, H.P. Ž1990. Cell Regul. 1, 415–424.