R496 Dispatch

Protein : Puzzling receptors Christoph Thiele, Hans-Hermann Gerdes and Wieland B. Huttner

All known sorting receptors for soluble cargo in the sorting at these two levels appears to vary between cell secretory pathway are transmembrane proteins. For types. In the most abundant cell type endowed with regu- sorting to the regulated pathway, however, a lated protein secretion, the neuron, membrane traffic to subpopulation of secretory proteins, associated with the either the somatodendritic or the axonal cell surface membrane but not membrane-spanning, appears to link diverges in the trans-Golgi network [7], which is confined cargo and membrane in storage granule biogenesis. to the neuronal cell body. In neurons, therefore, the trans- Golgi network is likely to be a major site of protein sorting Address: Department of Neurobiology, University of Heidelberg, Im Neuenheimer Feld 364, D-69120 Heidelberg, Germany. during the biogenesis of neuropeptide-containing secre- tory granules, which typically are targeted to the axon, and Current Biology 1997, 7:R496–R500 vesicles targeted to the dendrites. (Some protein sorting may continue as neuronal secretory granules undergo mat- http://biomednet.com/elecref/09609822007R0496 uration during axonal transport.) © Current Biology Ltd ISSN 0960-9822 In exocrine cells, the rate of vesicle traffic to the basolat- Secretory granules are specialized organelles that mediate eral surface for cell maintenance, relative to that of apical regulated protein secretion — the exocytotic release of secretory granules to fulfill the specialized function of the proteins in response to a stimulus. They contain a selected cell, is presumably much less than in a peptidergic neuron set of the soluble and membrane-bound proteins that, fol- with an extensive dendritic tree, such as a cerebellar Purk- lowing synthesis in the rough , are inje cell. Hence, in exocrine cells, traffic from the trans- transported to, and through, the Golgi complex. This Golgi network into secretory vesicles other than immature selection is the result of protein sorting that occurs during secretory granules may be insignificant. The same appears assembly of the secretory granule matrix and membrane to hold true for certain endocrine cells, such as pancreatic [1,2]. Protein sorting can lead either to enrichment of a islet β-cells. In such endocrine and exocrine cells, most given protein in secretory granules relative to constitutive protein sorting during the course of secretory granule bio- secretory vesicles, as is the case for the regulated secretory genesis is thought to take place in the immature secretory proteins, or to a failure of the protein to appear in secre- granule [6]. tory granules, as is the case for certain membrane proteins destined for the plasma membrane [3] (Figure 1). Neuroendocrine cell lines, which have been frequently used as model systems for studying secretory granule bio- One hypothesis for how secretory proteins are sorted to genesis, resemble neurons in that secretory granule secretory granules postulates the existence of membrane protein sorting occurs at the level of the trans-Golgi receptors that bind regulated secretory proteins at the network [2], but they may in addition exhibit protein level of the trans-Golgi network and direct them to secre- sorting at the level of the immature secretory granule as tory granules [1]. Using pro-opiomelanocortin (POMC) as seen in certain exocrine and endocrine cells [3,6]. Studies their model regulated secretory protein, Loh and col- with pheochromocytoma (PC12) cells have shown [8] that leagues [4] have recently suggested that the membrane- distinct constitutive and regulated secretory markers are associated form of carboxypeptidase E (CPE) — an quantitatively segregated from each other upon exit from found in secretory granules of neuroendocrine the trans-Golgi network. Thus, markers such as free gly- cells that is involved in the generation of biologically cosaminoglycan chains [2] and α1-antitrypsin (A. Krömer, active peptides from larger precursors [5] — functions as and H.-H. G., unpublished observations) are largely pack- such a sorting receptor. Before discussing the evidence for aged into constitutive secretory vesicles, and markers such and against this proposition, it is useful to review certain as chromogranin B and secretogranin II are largely pack- aspects of protein sorting during secretory granule biogen- aged into immature secretory granules. esis that have a bearing on the case. In neuroendocrine cells where there is significant mem- Two levels of sorting to the regulated pathway of protein brane traffic into both constitutive secretory vesicles and secretion can be distinguished: sorting in the trans-Golgi immature secretory granules, passive flow — traffic by network during the formation of immature secretory gran- ‘default’ — of secretory cargo from the trans-Golgi network ules (vesicular intermediates in secretory granule biogene- appears to occur largely into constitutive secretory vesicles sis), and sorting in the immature secretory granules during rather than immature secretory granules. Thus, most sul- their maturation [1–3,6] (Figure 1). The relative extent of fophilin — an artificial secretory protein consisting of Dispatch R497

Figure 1

(a) Neurons and neuroendocrine cell lines (b) Endocrine and exocrine cells

TGN TGN

1 2 3 132

ISG ISG

4 5 45

CSV IDV MSG CSV IDV MSG

Constitutive Constitutive-like Regulated Constitutive Constitutive-like Regulated secretion secretion secretion secretion secretion secretion

1 Constitutive secretory proteins (red circles) exit from the trans-Golgi network (TGN) by passive flow in constitutive secretory vesicles (CSVs) 2 Aggregates of regulated secretory proteins (light green symbols) exit from the TGN in immature secretory granules (ISGs), via interaction with their membrane-associated counterparts (dark green symbols); constitutive secretory proteins are excluded from these aggregates 3 Lysosomal (blue circles) exit from the TGN in ISGs bound to mannose-6-phosphate receptors (black forks), which interact with a patchy clathrin coat 4 Receptor-bound lysosomal enzymes are removed from the ISG in clathrin-coated, immature secretory granule-derived vesicles (IDVs); IDVs also remove some soluble processing products of the regulated secretory proteins (green triangles), which are excluded from the condensing regulated secretory proteins and enter the IDVs by passive flow 5 During ISG maturation, regulated secretory proteins condense; they remain with the mature secretory granule (MSG) © 1997 Current Biology

Secretory proteins can be sorted at either of two levels during arrow), sorting occurs predominantly at the ISGs. In these cells, some secretory granule biogenesis — at the trans-Golgi network (TGN) or of the constitutive secretory proteins, though excluded from the the immature secretory granule (ISG). In neurons and neuroendocrine aggregates of regulated secretory proteins, exit from the TGN in ISGs. cell lines (left), in which there is signficant membrane traffic into both These constitutive secretory proteins are removed from the ISGs by constitutive secretory vesicles (CSVs) and ISGs — indicated by the immature secretory granule-derived vesicles (IDVs). Note that equal thickness of the arrows — sorting occurs primarily at the TGN. In lysosomal enzymes may also exit from the TGN in distinct clathrin- certain endocrine and exocrine cells (right), in which there is minimal coated vesicles, which are not shown for clarity. membrane traffic into CSVs (thin arrow) relative to the ISGs (thick twelve tandem repeats of the heptapeptide tyrosine sulfa- B [10]. This implies that the immunoglobulin-bearing chro- tion site of the regulated secretory protein chromogranin B mogranin B molecules interacted with other molecules of [9] and so unlikely to contain a sorting signal for packaging the nascent immature secretory granule, and hence the exit into constitutive secretory vesicles — nonetheless exits of chromogranin B from the trans-Golgi network in imma- from the trans-Golgi network in constitutive secretory vesi- ture secretory granules is not by default (Figure 1). cles (A. Krömer and H.-H.G., unpublished observations). Furthermore, in PC12 cells, an immunoglobulin — nor- Three observations provide insight into the intermolecular mally a constitutive secretory protein — was found to be interactions of regulated secretory proteins during sorting diverted to secretory granules when bound to chromogranin to immature secretory granules. One is that regulated R498 Current Biology, Vol 7 No 8

secretory proteins are present at high concentrations in the that of chromogranin B [16], is known to mediate its dimer- lumen of the secretory pathway and exhibit a tendency to ization (C.T. and W.B.H., unpublished observations). self-associate in the lumenal milieu [2]. This leads to their Therefore, the loop-mediated interaction of soluble chro- aggregation in the trans-Golgi network, which underlies mogranin molecules with each other and, presumably, with the electron-dense matrix seen by electron microscopy in their membrane-associated counterparts, is likely to be the trans-Golgi network of cells forming secretory granules required for their sorting to immature secretory granules. [2]. Like the condensation of regulated secretory proteins that occurs in the immature secretory granule — such as POMC also contains an amino-terminal disulfide-bonded that of the processed peptides in neuroendocrine cells [11] loop. From their previous work, Loh and colleagues con- — aggregation is a means of segregating regulated from cluded that this loop is required for the traffic of POMC to constitutive secretory proteins [2] (Figure 1). It is secretory granules [17,18], although the compartment in important to note, however, that aggregation differs from the secretory pathway — the trans-Golgi network or condensation, in that it is a mode of protein–protein inter- immature secretory granules — at which the POMC loop action sufficiently unrestrained to allow post-translational is thought to promote sorting remains to be established. modification of regulated secretory proteins, in particular, These data — which disagree with earlier observations by the proteolytic processing of peptide precursors in neu- others [19] — were taken as a basis for studying roendocrine cells [2]. membrane binding of the POMC loop and, in particular, for searching for a binding site or ‘sorting receptor’. Loh The second observation is that many regulated secretory and colleagues [4] have now reported that the POMC loop proteins associate tightly with the luminal leaflet of the binds to the membrane-associated form of CPE. As such, membrane. In neuroendocrine cells, membrane-associated this finding is well within the conceptual framework forms have been described for all types of regulated secre- outlined above [3,12] (Figure 1), and is an example of a tory protein: peptides, such as insulin or adrenocorti- heterophilic interaction between soluble and membrane- cotropic (ACTH) [12], processing enzymes, such associated regulated secretory proteins. as CPE [13] and prohormone convertases [14], and other secretory granule constituents of unknown function, such There are, however, puzzling aspects to the conclusion as the granins [12]. Membrane association of these pro- that CPE is essential for sorting of POMC to secretory teins can been detected at the level of the trans-Golgi granules. This conclusion is based on results obtained network and persists through secretory granule biogenesis with two experimental systems: clones derived from the and , so that cell-surface-associated immuno- neuroendocrine cell line Neuro-2a after transfection with reactivity for these proteins can be detected when regu- CPE antisense DNA [4], and pituitary neurointermediate lated secretion is induced [12]. lobes and dissociated anterior pituitary cells from CPEfat mutant mice [4,20], which have drastically reduced CPE The various regulated secretory proteins differ in the pro- enzyme levels [21]. From a conceptual point of view, an portion in which they exist in membrane-associated form, essential role for CPE revealed in these systems would and the nature of their membrane anchorage is understood imply either that CPE is the only membrane binding site in only a few cases (for example, [13]). Nonetheless, given that can mediate POMC sorting to secretory granules, or the ability and tendency of regulated secretory proteins to that the lack of CPE-catalyzed processing has some engage in homophilic and heterophilic interactions, the secondary effect that inhibits sorting. Unfortunately, the membrane-associated forms of these proteins can be data obtained by Loh and colleagues [4,20] in these two regarded as ‘sorting receptors’, in that their existence pro- systems are inconclusive. vides a simple mechanism for the membrane enveloping of aggregated regulated secretory proteins, and hence for First, it is unclear whether the decreased CPE levels in the budding of an immature secretory granule containing a the Neuro-2a clones isolated after transfection reflect a selected set of secretory proteins [3,12] (Figure 1). specific reduction in just one neuroendocrine gene product (CPE) or a generalized downregulation of the The third observation is that specific structures essential for neuroendocrine phenotype [4]. Second, the elevated level sorting to immature secretory granules have been of ACTH immunoreactivity — reflecting POMC and/or identified. The disulfide-bonded amino-terminal loop of its ACTH-containing fragments — recovered in the chromogranin B has been shown to be such an essential medium of these cells without stimulation is not necessar- structure. Reduction of this disulfide bond in vivo [15], or ily due to constitutive secretion — it may equally well deletion of the disulfide-bonded loop (A. Krömer and reflect an increased basal release from secretory granules. H.-H. G., unpublished observations), causes chromogranin Third, the regulated secretion of ACTH immunoreactiv- B to leave the trans-Golgi network in constitutive secretory ity from these cells was determined after an unusually vesicles rather than immature secretory granules. The disul- long release period (3 hours), considering that depolariza- fide-bonded loop of chromogranin A, highly homologous to tion was used as stimulus [4]. The increase in ACTH Dispatch R499

immunoreactivity in the wild-type Neuro-2a cells, but not seen as the underlying reason for the presence of clathrin the transfected Neuro-2a clones, may therefore also reflect on nascent dense-cored buds of the trans-Golgi network a differential response of the POMC biosynthetic machin- and on immature secretory granules [2]. Meanwhile, we ery to the elevated extracellular potassium ion concentra- have learned that this clathrin, the membrane binding of tion. Similar criticism holds true for the analysis of which is mediated by AP-1 adaptors [22], can be secretion of pulse-labeled growth hormone from anterior accounted for by the presence of furin (S. Tooze, personal pituitary cells of wild-type and CPEfat mice [20]. communication), an endopeptidase related to the prohor- mone convertases, and mannose-6-phosphate receptors A fourth problem is that the kinetics of secretion from [11,16,23]. The latter remove lysosomal enzymes, which pituitary neurointermediate lobe of pulse-labeled POMC may exit from the trans-Golgi network in immature secre- and its ACTH-containing fragments are virtually identical tory granules, from these organelles during their matura- for control and CPEfat mice [4], showing that POMC and tion [11,23] (Figure 1). its products were secreted by the same pathway in the two types of mice. In cells from wild-type animals, POMC is There are, however recent reports indicating that mature known to be efficiently sorted to the regulated secretory secretory granules contain transmembrane proteins — pathway, so it follows that the same is true for CPEfat such as ICA512 [24] and phogrin [25] — that are homolo- mice. In fact, the observation by Loh and colleagues [4] gous to receptor tyrosine phosphatases. As with any mem- that the release of labeled POMC and its ACTH-contain- brane protein of secretory granules, their relatively low ing fragments approaches a plateau at a higher level in abundance would allow binding of only a fraction of the CPEfat mice (~40%) than normal mice (~20%) points to regulated secretory proteins. Whether this makes a case this secretion having occurred via the constitutive-like that these transmembrane proteins may be receptors of pathway: incomplete proteolytic processing because of the highly multimeric ligands — either aggregated or con- lack of CPE might impair the condensation of ACTH in densed regulated secretory proteins — or points to other the immature secretory granule, resulting in a greater pro- roles in the regulated secretory pathway, remains to be portion being removed during secretory granule matura- seen. To bind and what to bind, that is still the question. tion [6] (Figure 1). Acknowledgements There thus appears to be little evidence to substantiate We thank P. Arvan, D. Castle, B. Eipper, L. Fricker, R. Mains, D. Steiner and S. Tooze for comments on the manuscript. the claim [4,20] that CPE has an essential role in the sorting of regulated secretory proteins, and that it acts at References the level of the trans-Golgi network. Disappointing as this 1. Burgess TL, Kelly RB: Constitutive and regulated secretion of proteins. Annu Rev Cell Biol 1987, 3:243-293. may be, it makes it easier to reconcile the observations 2. Tooze SA, Chanat E, Tooze J, Huttner WB: Secretory granule that anterior pituitary cells of CPEfat mice contain almost formation. In Mechanisms of Intracellular Trafficking and Processing as much 4.5 kDa ACTH as normal mice [20], and that islet of Proproteins. Edited by Loh YP. Boca Raton: CRC Press; 1993:157-177. fat β-cells of CPE mice neither contain lower numbers of 3. Bauerfeind R, Huttner WB: Biogenesis of constitutive secretory secretory granules [21] nor show significant constitutive vesicles, secretory granules and synaptic vesicles. Curr Opin Cell Biol 1993, 5:628-635. secretion of proinsulin (D. Steiner, personal communica- 4. Cool DR, Normant E, Shen F-S, Chen H-C, Pannell L, Zhang Y, Loh tion). Additional evidence against CPE having an obliga- YP: Carboxypeptidase E is a regulated secretory pathway sorting tory role in regulated secretory protein sorting comes from receptor: genetic obliteration leads to endocrine disorders in Cpefat mice. Cell 1997, 88:73-83. the observation that, in transgenic mice, chromogranin B 5. Fricker LD: Carboxypeptidase E. Annu Rev Physiol 1988, 50:309-321. expressed in pancreatic acinar cells, which lack CPE [5], is 6. Arvan P, Castle D: Protein sorting and secretion granule formation in regulated secretory cells. Trends Cell Biol 1992, 2:327-331. sorted to zymogen granules (S. Natori and W.B.H., unpub- 7. Simons K, Dupree P, Fiedler K, Huber LA, Kobayashi T, Kurzchalia T, lished observations). What we are left with, it seems, is Olkkonen V, Pimplikar S, Parton R, Dotti C: Biogenesis of cell- increasing evidence that regulated secretory proteins, surface polarity in epithelial cells and neurons. Cold Spring Harbor Symp Quant Biol 1992, 57:611-619. which have been known to exist in soluble as well as 8. Tooze SA, Huttner WB: Cell-free protein sorting to the regulated membrane-associated forms, are capable of interacting and constitutive secretory pathways. Cell 1990, 60:837-847. 9. Niehrs C, Huttner WB, Rüther U: In vivo expression and with each other in many different ways, providing a great stoichiometric sulfation of the artificial protein sulfophilin, a deal of flexibility for protein–protein interactions in the polymer of tyrosine sulfation sites. J Biol Chem 1992, biogenesis of the secretory granule (Figure 1). 267:15938-15942. 10. Rosa P, Weiss U, Pepperkok R, Ansorge W, Niehrs C, Stelzer EHK, Huttner WB: An antibody against secretogranin I (chromogranin B) What remains an open issue is whether regulated secre- is packaged into secretory granules. J Cell Biol 1989, 109:17-34. tory proteins bind to transmembrane proteins of secretory 11. Kuliawat R, Arvan P: Distinct molecular mechanisms for protein sorting within immature secretory granules of pancreatic β-cells. J granules and whether this binding is essential for sorting. Cell Biol 1994, 126:77-86. In early ideas about secretory granule biogenesis [1], this 12. Pimplikar SW, Huttner WB: Chromogranin B (secretogranin I), a secretory protein of the regulated pathway, is also present in a seemed the most obvious way that regulated secretory tightly membrane-associated form in PC12 cells. J Biol Chem proteins could be sorted to secretory granules, and was 1992, 267:4110-4118. R500 Current Biology, Vol 7 No 8

13. Fricker LD, Das B, Angeletti RH: Identification of the pH-dependent membrane anchor of carboxypeptidase E (EC 3.4.17.10). J Biol Chem 1990, 265:2476-2482. 14. Shennan KIJ, Taylor NA, Docherty K: Calcium- and pH-dependent If you found this dispatch interesting, you might also want aggregation and membrane association of the precursor of the to read the August 1997 issue of prohormone convertase PC2. J Biol Chem 1994, 269:18646-18650. Current Opinion in 15. Chanat E, Weiss U, Huttner WB, Tooze SA: Reduction of the disulfide bond of chromogranin B (secretogranin I) in the trans- Cell Biology Golgi network causes its missorting to the constitutive secretory pathway. EMBO J 1993, 12:2159-2168. 16. Huttner WB, Ohashi M, Kehlenbach RH, Barr FA, Bauerfeind R, which includes the following reviews, edited Bräunling O, Corbeil D, Hannah M, Pasolli A, Schmidt A, et al.: by Scott D. Emr and Vivek Malhotra, on Biogenesis of neurosecretory vesicles. Cold Spring Harbor Symp Quant Biol 1995, 60:315-327. Membranes and sorting: 17. Tam WWH, Andreasson KI, Loh YP: The amino-terminal sequence of pro-opiomelanocortin directs intracellular targeting to the COPII and secretory cargo capture into transport regulated secretory pathway. Eur J Cell Biol 1993, vesicles 62:294-306. M.J. Kuehn and R. Schekman 18. Cool DR, Fenger M, Snell CR, Loh YP: Identification of the sorting signal motif within pro-opiomelanocortin for the regulated Coatomer (COPI)-coated vesicles: role in intracellular secretory pathway. J Biol Chem 1995, 270:8723-8729. transport and protein sorting 19. Roy P, Chevrier D, Fournier H, Racine C, Zollinger M, Crine P, Boileau G: Investigation of a possible role of the amino-terminal pro- P. Cosson and F. Letourneur region of proopiomelanocortin in its processing and targeting to Linking cargo to vesicle formation: receptor tail secretory granules. Mol Cell Endocrinol 1991, 82:237-250. interactions with coat proteins 20. Shen F-S, Loh YP: Intracellular misrouting and abnormal secretion of adrenocorticotropin and growth hormone in Cpefat mice T. Kirchhausen, J.S. Bonifacino and H. Riezman associated with a carboxypeptidase E mutation. Proc Natl Acad The diversity of Rab proteins in vesicle transport Sci USA 1997, 94:5314-5319. 21. Naggert JK, Fricker LD, Varlamov O, Nishina PM, Rouille Y, Steiner P. Novick and M. Zerial DF, Carroll R, Paigen BJ, Leiter EH: Hyperproinsulinaemia in obese SNAREs and NSF in targeted membrane fusion fat/fat mice associated with a carboxypeptidase E mutation which reduces enzyme activity. Nature Genet 1995, 10:135-142. J.C. Hay and R.H. Scheller 22. Dittié AS, Hajibagheri N, Tooze SA: The AP-1 adaptor complex Calcium regulation of release: reliably binds to immature secretory granules from PC12 cells, and is unreliable? regulated by ADP-ribosylation factor. J Cell Biol 1996, 132:523-536. Y. Goda and T.C. Südhof 23. Kuliawat R, Klumpermann J, Ludwig T, Arvan P: Differential sorting of The role of lipid signaling in constitutive membrane lysosomal enzymes out of the regulated secretory pathway in pancreatic β-cells. J Cell Biol 1997, 137:595-608. traffic 24. Solimena M, Dirkx Jr. R, Hermel J-M, Pleasic-Williams S, Shapiro JA, M.G. Roth and P.C. Sternweis Caron L, Rabin DU: ICA 512, an autoantigen of type I diabetes, is an intrinsic membrane protein of neurosecretory granules. EMBO The trans-Golgi network: a late secretory sorting station J 1996, 15:2102-2114. L.M. Traub and S. Kornfeld 25. Wasmeier C, Hutton JC: Molecular cloning of phogrin, a protein tyrosine phosphatase homologue localized to insulin secretory Caveolae, DIGs, and the dynamics of granule membranes. J Biol Chem 1996, 271:18161-18170. sphingoloid–cholesterol microdomains T. Harder and K. Simons

the same issue includes the following reviews, edited by Matthias A. Hediger, on Membrane permeability:

The cystic fibrosis transmembrane conductance regulator and ATP S. Devidas and W.B. Guggino The ABCs of ATP-sensitive potassium channels: more pieces of the puzzle J. Bryan and L. Aguilar-Bryan Regulation of aquaporin-2 water channel trafficking by vasopressin M.A. Knepper and T. Inoue Family of neutral and acidic transporters: molecular biology, physiology and medical implications Y. Kanai Molecular biology of iron and zinc uptake in eukaryotes D. Eide