stores. Not suprisingly, therefore, there has been considerable interest in the identification and characterisation of the intracellular calcium stores in non-muscle cells.
Calcium Storage by the Sarcoplasmic Reticulum The sarcoplasmic reticulum has been widely used as a model for calcium storage in non-muscle systems since it is general1 accepted as the prototype calcium storage organelle('. *). It is, therefore, worth considering the properties of the SR which render it suitable for calcium storage. First, there is the ability to pump calcium out of the myoplasm into the luminal space. This is achieved by the presence of high concentrations of a calcium ATPase. Second, there is the presence of a high concentration of calsequestrin, a major low affinity high capacity calcium binding protein within the lumen which permits high concentrations of calcium to be accumulated within the SR. In addition, the SR has also been shown to contain another calcium binding protein (HACBP/calreticulin) which expresses both high and low affinity calcium binding sites. Although calsequestrin and calreticulin are soluble when isolated, they appear to be part of an insoluble matrix within the ER lumed3). This matrix may be important in localising the calcium storage capacity to specific parts of the SR(4). However, it must also be considered that the sequestration of calcium into the matrix is important for the performance of the organelle in calcium signalling. Thus the storage of calcium in the SR as a dynamically unstable complex might be crucial to the signalling process and this feature might also be important in analogous signalling systems in non-muscle cells. Finally, there is the issue of the mechanism by which calcium passes from the store to the cytoplasm of the cell. In the SR this appears to be mediated b a channel created by the so-called ryanodine receptor($. Thus, an SR-type calcium storage organelle would need to possess most of the above-mentioned characteristics.
Y
Summary Calcium storage is one of the functions commonly attributed to the endoplasmic reticulum (ER) in nonmuscle cells. Several recent studies have added support to this concept. Analysis of reticuloplasm, the luminal ER content, has shown that it contains several proteins (reticuloplasmins) which are prospective calcium storage proteins. One of these, calreticulin, is also present in the sarcoplasmic reticulum (SR). In sea urchin eggs, a calsequestrin-like protein has been clearly localised to the ER. The recent demonstration that the IP3 receptor, which has similarities with the calcium release channel in the SR is also localised in the ER membrane suggests that calcium stored in the ER is important for intracellular signalling. The alternative view, that the physiologically important calcium store is a specialised organelle, the calciosome, is not supported by these observations. Recent evidence also suggests that ER calcium might be important in ER structure and in the retention of the luminal ER proteins. Introduction In view of the vital role played by calcium ions in a wide variety of cellular processes, the ability to store calcium within intracellular membranous organelles must be considered as one of the major developments in cellular evolution. Cells which lack such systems are solely dependent on plasma membrane fluxes to maintain their intracellular calcium at acceptable levels and to generate changes in intracellular calcium for signalling. With the evolution of intracellular storage systems, excess cytoplasmic calcium could be taken up by the stores, thereby buffering cells against excessive changes in cellular calcium. Coincident with this increased buffering capacity, cells would acquire the ability to use increased levels of intracellular calcium as a general signalling mechanism. Although such signalling need not be absolutely dependent on the presence of an intracellular calcium store, it would be intrinsically less stable, and thus unreliable, if plasma membrane fluxes were the only mechanism for altering cytoplasmic calcium. An additional sophistication made possible by the development of the intracellular stores is the ability to generate local gradients of calcium ions which could serve as a signalling device within cells. Such gradients could arise through the pattern of distribution of the
Calcium Storage by the Endoplasmic Reticulum The idea that the E R performs a calcium storage function is based on a number of experimental observations including the accumulation of calcium in the ER of live cells and isolated microsomes as well as the direct measurement of cellular distributions by electron microprobe analysid6). However, it is recognised that none of these observations are without reservation and further evidence is required to formally establish the calcium storage function of the ER. One approach to this is to examine the extent to which the ER exhibits the properties associated with the SR calcium storage function. In this context, it must be remembered that muscle cells are highly specialised and thus particular properties are grossly hypertrophied in such cells. For example, the levels and organisation of myosin are much greater compared with non-muscle cells, although the basic motile processes appear to be
same, Thus, it is probable that an SR analogue in nonmuscle cells will not necessarily possess the precise properties of the SR but will nevertheless share the major common features. Calcium-binding Proteins of the Endoplasmic Reticulum The presence of high concentrations of the luminal SR proteins such as calsequestrin is considered a crucial requirement for the calcium storage function associated with this organelle. Thus, if the E R were to perform a storage function in non-muscle cells, the expectation is that analogous or even homologous proteins should exist in the ER. Such a possibility appeared unlikely until relatively recently, since the general view was that the major luminal constituents of the E R were the newly-synthesized 'secretory' proteins. These were likely to vary between cell types and it was difficult to visualise how a common property such as calcium storage could be attributed to them. Fortunately, it turns out that this is not an accurate general picture of the E R lumen. Rather it has emerged that the luminal material, or reticuloplasm, is predominantly composed of a family of proteins, collectively called reticuloplasmind'), which are the permanent residents of this space. Two independent studie&-") of the reticuloplasm of plasmacytoma cells, which have a well-developed ER, showed that there are at least five major proteins associated with the lumen of the ER. However, there is still disagreement about whether one of the proteins, endoplasmin (ERP99), is luminal or transmembrane. Our own studies show no evidence for the latter possibility("). Three of the other proteins have also been well characterised and are generally referred to as BiP (immunoglobulin heavy chain binding protein)(12), protein disulphide isomerase(13) and ~alreticulin('~~ "). One of the proteins, RP60, is still poorly defined. A significant feature of the reticuloplasmins is their concentration. Using the material from plasmacytoma cells we estimate that the proteins could be present at an average concentration of about 50mgml-'. This is equivalent to almost 1mM protein which is lower than, but of the order of, the concentration of the protein in the sarcoplasmic reticulum. A second significant feature of these proteins is their relative acidity due to the presence of large amounts of aspartic and glutamic acid residues in their sequence. A third factor which might be relevant is the fact that several of the proteins are overexpressed when cells are treated with calciumperturbing agents('6317). One of the characteristic features of calsequestrin is its staining with the carbocyanine dye, Stains-all. Analyses of isolated reticuloplasm with this reagent, after gel electrophoresis, showed that two of the proteins, endoplasmin and calreticulin, are stained blue, with the latter exhibitin a staining intensity comparable with calsequestrin . Use of the radioactive calcium overlay method for detecting calcium
(18
binding proteins under conditions which would detect low-affinity binding proteins confirmed that endoplasmin and calreticulin can bind significant amounts of calcium in the millimolar affinity range('.''). Small but significant calcium binding was also observed with BiP and protein disulphide isomerase. Calcium-binding was also detected with isolated reticuloplasm, endoplasmin and calreticulin using equilibrium dialysis("). It was estimated that endoplasmin and calreticulin could bind about 10 and 20 molecules of calcium per molecule, respectively, in the millimolar range. One concern about these studies is that they are carried out under conditions which might not reflect the true ionic conditions in vivo. However, the fact that parallel analyses of calsequestrin gave reasonable results suggests that the conditions used bear some physiological significance. Calreticulin an ER/SR Calcium Binding Protein One of the obvious questions arising in this context concerns the presence of calsequestrin itself in the ER of non-muscle cells. Our systematic analyses of the reticuloplasm from non-muscle cells have failed to demonstrate the presence of a homologue of calsequestrin. However, one of the reticuloplasmins, now referred to as calreticulin, was of particular interest since it did possess many of the general properties of calsequestrin, i. e. -55 kDa Stains-all binding, multiple low affinity calcium binding sites, etc. In fact, at least some of the claims of a calsequestrin in non-muscle cells can easily be attributed to calreticulin. In order to determine whether there was any structural homology between the two proteins we determined its rimary structure by cDNA cloning and sequencing('). The results clearly showed that it was a different protein although there was a short and quite characteristic sequence which is homologous with calsequestrin. The sequence also confirmed the results of earlier biochemical analyses which had shown that the low affinity calcium binding sites are located in a hyperacidic stretch close to the C-terminus of the protein, and the fact that the protein contained the C-terminal KDEL sequence characteristic of a reticuloplasmin. Although calreticulin is clearly different from calsequestrin, its properties showed significant similarity with the other major calcium binding protein in the SR called HACBP (high-affinity calcium binding protein). Immunofluoresence studies confirmed its presence in the SR of skeletal muscle. Subsequently the complete sequence of HACBP was also determined(15) and confirmed that the two proteins are indeed the same. The fact that one of the major calcium binding E R proteins is also present in the SR adds considerable circumstantial support to the view that the E R is also a calcium storage organelle, since calcium storage and release is the major known function of the SR. One puzzling issue arises from the presence in calreticulin from either muscle or non-muscle cells of the C-terminal KDEL sequence. This sequence has hitherto only been detected in resident E R proteins and is
thought to be part of the mechanism for precluding such proteins from the rest of the secretory pathway. If that is also the case in developing myocytes, it suggests that the SR originates directly from the E R without involvement of the Golgi apparatus. This possibility is supported by the observations that the carbohydrate moiety of calsequestrin is of the high mannose type and is fully Endo H sensitive(”). The suggestion is that during the development of the SR, calsequestrin and calreticulin accumulate within the E R and then partition into specialised vesicles which generate the SR itself. Since a special set of coated vesicles have been implicated in this process(20), it will be interesting to determine whether they contain calreticulin. Although calsequestrin is not detectable in nonmuscle cells, there is evidence of a calsequestrin-like protein in sea urchin eggs. This protein appears to cross-react immunologically with rabbit muscle calsequestrin and possesses most of the general properties of the latter. Significantly, immunolocalisation studies clearly show that the protein is in the general endoplasmic reticulum(2’) adding strong support to the view that the E R is an important calcium storage organelle. The Role of ER Calcium in lntracellular Signalling One of the major sources of the intense interest in the calcium storage capacity of the E R is the proposed role of this organelle in intracellular signalling. It is envisaged that events at the cell surface result in the release of IP3 (inositol 1,4,5-triphosphate), which interacts with receptors in the intracellular stores, releasing calcium ions which act as a second messenger to produce the observed effect(s)(22). Prevous studies pointed to the E R as the site of action of IP3 since the effects could be simulated with isolated microsomes. However, such studies are always beset with uncertainty since microsomes are never a pure preparation of E R vesicles. Recent studies on the IP3 receptor itself provide much clearer evidence about the location of these molecules in ~ e l l s ( ~ ~This - ~ ~protein ). is highly enriched in the cerebellar Purkinje cells of the rat and can be localised by immunocytochemical methods using monospecific antibodies to the receptor. These studies clearly demonstrate the presence of the protein in the E R of these cells. One uncertainty about the IP3 receptor concerns its apparent presence in other membranes. However, this has been resolved and the protein ap ears to be localised exclusively in the E R membraneP26).The fact that the IP3 receptor actually shows homology with the ryanodine receptor in the SR(23) is also very significant since the ryanodine receptor is generally considered to be the calcium release channel in the SR(27).Furthermore, reconstitution experiments in which the purified protein was incorporated into artificial vesicles showed an IP3inducible calcium release(28). These studies provide strong evidence that the IP3 sensitive calcium store, at
least in the cerebellar Purkinje cells, is located in the endoplasmic reticulum. Whither calciosomes? It is impossible within the context of an article of this scope to avoid an examination of the status of the structure referred to as the ‘calcios~tne’(~~). It was proposed that cells contained special vesicular elements which represent the physiologically relevant calcium store. The characteristics of this store are its distinction from the ER, the presence of calsequestrin as evidenced by the presence of a protein which appears to react with antibodies to rabbit muscle calsequestrin and the presence of an IP3-sensitive calcium release mechanism. However, the only proteins which appear to be capable of serving in calcium storage in nonmuscle cells are all characteristic reticuloplasmins. Our analyses of HL60 cells have shown that even in these cells the only calsequestrin-like protein is calreticulin. As mentioned above, the available evidence shows that the IP3 receptor is not associated with specialised structures but is predominantly an ER protein. Thus, two of the major bases for the proposal of a specialised structure appear insecure. If the calsequestrin-like protein in HL60 cells is calreticulin, is there an alternative explanation for the pattern of localisation observed in these cells? One possibility suggested by our own studies is that it is a problem of fixation. We have found that calreticulin, in contrast to endoplasmin, is very variably fixed in different cells and even in the same cell line under different conditions. The other possible explanation is that the distribution of calreticulin, and for that matter any of the reticuloplasmins, is not necessarily uniform within the lumen of the ER. This does not appear to be the case in plasmacytoma cells(’’) but could be true for others. Regions which contain high concentrations of the reticuloplasmins could arise by a variety of mechanisms the simplest of which would be the localised condensation of the proteins into immobile aggregates (see below). Even if this is true it does not appear necessary to invoke a novel organelle to account for calcium storage in nonmuscle cells and the E R remains the main candidate for a storage organelle. Organisation of Reticuloplasmins in Cells One of the significant features of the sarcoplasmic reticulum is the fact that luminal proteins such as calsequestrin do not exist as freely diffusing species but rather as part of an immobile network(3).For this and other reasons it is relevant to consider whether the reticuloplasmins are freely-diffusing species within the lumen of the E R or whether they might be part of some supramolecular complex. The difficulty with such a question arises from the intrinsic difficulty of measuring the diffusion of proteins within intact cells. However, it is possible to examine other properties which are themselves dependent on the physical state of the
Fig. 1. Endoplasmin distribution in the ER of a murine fibroblast. The staining of the peripheral elements (small arrowheads) is significantly lower than that of the central elements (large arrowheads). The pattern is consistently seen in cells with a dispersed reticular E R network.
protein within the cell. One such property is fixation with cross-linking agents. If a protein is diffusing freely, it will not be efficiently cross-linked to other proteins and will not be 'fixed' when cells are treated with crosslinking agents such as formaldehyde which only crosslink proteins inefficiently. Thus it was something of a surprise to find that the reticuloplasmins were generally very efficiently fixed in plasma cells treated for even short periods with formaldehyde. The degree of fixation of endoplasmin has since been quantitated and found to be >90% of the total in the ER. Such a high fixation efficiency is difficult to reconcile with the notion that the proteins are freely-diffusing species within the lumen of the E R and support the idea of their organisation into some higher order structure. There are other reasons for believing that a luminal structure might prevail within the E R . These include the indication that the distribution of the reticuloplasmins might not be uniform throughout the reticulum. For example the presence of the reticuloplasmins in the rough E R is clear, but their presence in the smooth E R is not yet established and at least some studies suggest that they are excluded or at least decreased in the latter(3"). Studies on the distribution of endoplasmin in
plasmacytoma cells revealed the presence of a distinct region in which the concentration of endoplasmin appeared to be higher, although it was possible that this represented a region in which the E R was packed differently(31). More recently, we have observed that the staining of the peripheral elements of the E R in fibroblastoid cells is significantly lower than that of the central ones and that this does not appear to reflect differences in the size or shape of the elements (Fig. 1). These observations are consistent with the suggestion that the reticuloplasmins can assume non-random distributions within the E R through the formation of higher order structures. We have suggested(") that such structures may be stabilised by calcium ions in a manner analogous to that observed in the SR, and that they may play an important role in the retention of the reticuloplasmins through a condensation sorting mechanism similar to that proposed for the sorting of secretory proteins into the secretory granule(33). Yet another aspect of this is that the E R itself might be stabilised by such a luminal network in a manner similar to that performed by the structural proteins associated with the plasma membrane. This suggestion is based on the observation that the E R exhibits different stability
in different cells and that stability appears to be increased in cells expressing higher levels of the reticuloplasmins. The significant point is that it is important to determine the extent to which the reticuloplasmins are freely diffusing or otherwise since a number of issues relevant to E R structure and function could depend upon this factor. In the context of this article, the role of calcium ions in the formation of such intra-luminal structures is particularly relevant because of the possible analogy with the organisation of the luminal calcium storage proteins in the sarcoplasmic reticulum and the suggestion that such an arrangement might be functionally important (see above). From this perspective, a luminal matrix that can undergo reversible sol-gel transformation may also provide an explanation for the oscillations in cytoplasmic calcium observed in cells undergoing continuous stimulation. Future Prospects Although progress in the field of E R calcium storage has been significant over the last decade many important aspects need clarification. Perhaps the most relevant concerns the accurate measurement of the calcium (free and bound) in the E R of live cells. It will also be important to determine the general ionic status of the lumen of the E R since the calcium-binding properties of proteins, especially the low affinity species, are very sensitive to the concentrations of other ions. Although some use has already been made of cellfree systems, they need to be improved to simulate the natural systems more closely. The demonstration that the IP3 receptor can be incorporated into vesicles suggests the possibility of loading such vesicles with the reticuloplasmins to determine whether this makes a difference to calcium storage and release. In the case of calreticulin, it would even be appropriate to exploit SR vesicle preparations, used previously to examine calsequestrin function, for this purpose. Since the E R proteins have been cloned and sequenced, preparation and analysis of mutant forms is clearly feasible. Finally, the impending availability of homologous recombination techniques for the replacement of normal genes with suitable variants could provide a powerful method for examination of the function of those proteins in calcium storage in vivo References 1 CARAFOLI. E. (1987). Intracellular calcium homeostasis. Annu. Rev. Biochem. 56. 395-433. 2 MACLENNAN, D. H . A N D HOI.LAND. P. C . (lY75). Calcium transport in the sarcoplasmic reticulum. Anmr. Rei,. Biophy.~.Bioeng. 4, 377-40.1. 3 MEISSNER.G. (1975). Isolation and characterisation of two types of sarcoplasmic reticulum vesicles. Biochem. Biophys. Acfa 389. 51-61. 4 JORCENSEN. A. O . , SHEN, A. C. Y. A N D CAMPBELL.K. P. (1985). Ultrastructural localisation of calsequestrin in adult rat atrial and vcntricular muscle cells. J . Cell Biol. 101. 257-268. 5 LAI.F. A,. ERICKSON. H. P.. ROUSSLALI. E.. Liu. Q. Y. A N D MEISSNER. G. (1988). Purification and reconstitution of the calcium release channel from skeletal muscle. Narirw 331. 315-319. 6 SOMLYO. A. P. (1984). Cellular sites of calcium regulation. Nnrrrre 309. 516-517. 7 KOCH.G. L. E. (1987). Rcticuloplasniins: a novel group of protcins in the endoplasmic reticulum. J . CeN Sci. 87. 491-492.
8 LEWIS. M. .I.. MAZZAREI.LA. R. A. m r ) G R ~ NM.. (1YSS). Structurc and assembly of the endoplasmic reticulum. J . Biol. C%em.260. 3(lSI)-3057. 9 KOCH.G. L. E.. SMITH. M. J.. MACER. D. R . J . . W L B ~ I L P. K .A N D MDK~,\R,\. R. A. (1086). Endoplasmic reticulum contains a common abundant calcium binding glycoprotein, endoplasmin. J. Cell Sci. 86. 217-232. 10 MACER.D. R. J. A N D KOCH.G. L. E. (19x8). Identification of ;I set o f calcium binding proteins in reticuloplasm. the lumnial content of the cndoplasmic reticulum. J . Cell Sci. 92. 61-70. 11 KOCH. G . L. E.. MACFA.D. R . J . ,\wr) W ~ O D I N G F.. . B. 1'. (IYSX). Endoplasmin IS a reticuloplasmin. J . Cell Sci. YO. 485-491. 12 BOLE. D. G.. HENDERSHOT. L. M. A N D KLARNLY. J . F. (lY86). Poqtranslational association of immunoglobulin hcavy chain binding protein M itli nascent heavy chains in non-secreting hybridoma\. J . Cell B i d . 102. 1.558-1556. 13 FREEDMAN. R. B . (19X4). Nativc disulphide bond formation in protcin biosynthesis: evidence for the role of protein disulphide isomerase. 7i.end\ Biochrm. Sci. 9. 438-441. 14 SMITH.M . J. A N D KOCH.G. L. E. (1989). Multiple zones i n the sequence of calreticulin (CRP55. calregulin, HACBP). a major calcium binding ER/SR protein. E M B O J . 8, 3581-3586. 15 FLEIGEL, L., BURNS.K., MACLENNAN. D. H.. REIrHMLlLH. R . A , F. ANI) MICHAI A K . M. (1989). Molecular cloning of the high affinity calciuni binding protcin (calrcticulin) of skeletal muscle sarcoplasmic reticulum, J . Brol. Clicm. 264. 21 522-21 528. 16 LEE. A. S. (1987). Co-ordinated regulation of a set of genes by gluco\e and calcium ionophores in mammalian cells. Trends Biockem. k i . 12. 20-23. 17 RLSLNDEZ. E.. ATCNELLO. J . W.. GHAI-SKY. A , . CHANG. C. S. A N D I . t t . A. S. (1985). Calcium ionophore A23187 induce\ crpression of glucose rcgulatcd genes and thcir heterologous fusion gene. Mol. Cell Biol. 5. 1212-121'). 18 KOCH.G. L. E . , MACER. D. R. J . A N D WOODING, F. B. P. (1989). i n Reid. R . E.. Cook. G. M. W. and Lurio. J. P. (eds) Biochemical approachc\ to cellular calcium. methodological surveys i n biochemistry and an;ilysi\. R o y [ Socirry Clrrm.. London 19, 159-165. 19 THOMAS. K.. NAVARKO, J.. BLNSON.R. J . J . , CAMPBELL. K. P.. R o i v m o . K. L. A N D F I N L R . E. (1989). Newly synthcsiscd calseyuettrin destined for the urcoplasmic reticulum is contained in early/intermediate golgi-tlcrivcd clathrin-coated vesicles. J . Biol. Chem. 264. 3140-314.5. 20 ROTUNDO.R. L.. THOMAS.K.. PORTER-JORDAN. K.. BtNSON. R. .I. J.. FERNANDEZ-VALLE. C. AND FINE.R . E . (IYSY). lntracellular transport. torting and turnover of acctylcholinesterase. J . Biol. Chem. 264. 3146-3152. 21 HENSON.J . H.. BECG,D. A , . B t A C l . l t l . S. M.. FISHKIND. D. J.. BONDER. E. M.. TERASAKI. M.. LEBECHE. D. AND KAMINER. B. (1989). A cal\cquc\lrinlike protein in the E R of the sea-urchin: localisation and dynamics i n tlic cgg and first cell cycle embryos. J. Cell Biol. 109. 149-161. 22 B ~ K R I U M. G ~ J. . (1987). lnositol triphosphate and diacylglycerol: Two interacting sccond messcngers. Aiimr. Rev. Biocliem. 56. 159-193. 23 MIGNFRY. G . A , , SUDHOF. T. c.. T A K E I . K. .\ND DE CI\MII.LO. P. (lY89). Putative receptor for inositol 1.4.5 triphosphatc \imilar to ryanodinc rcccptor. Nfrrrrre 342. 192-195. 24 Ross. C. A , . MELDOLESI. J.. MILNEK. T. A , . SATOH.T.. S L P . \ ~ ~ A P OS.N F . . A N D SNYDER. S . H. (1989). Inositol 1.4.5 triphosphate receptor localiscd to endoplasmic reticulum in cerebellar Purkinje neurons. Ncrrurc 342. J6X-4711. 25 FURICCHI. T., YOSHIKAWA. S., M I Y A W A K AI, ,. WAD,%.K.. MAEDA.N . ,\xi) MIKOSHIBA. K. (1989). Primary structure and functional expression nf the inositol 1.4.5 triphosphate binding protein P4(KI. Nrrrrrw 342. 32-38. 26 DE CAMII.I I , P.. TAKEI, T.. MICNERY. G . A. A N D SCDHOF. T. C. (l9YO). InsP3 receptor turnaround. Nurim 344. 495. 27 LAI.F. A. A N D MEISSNLR, Ci. (1989). Thc muscle ryanodinc receptor and its intrinsic Ca2+ channel activity. J . Bioeng. Bionwmh. 21. 227-246. 28 FERRIS. C. D., HUCANIR. R.L.. SUPATTAPONL. s. A N D S N Y D E R . s. H. (1'189). Purified inwitol 1.4.5 triphosphatc receptor mediates calcium flux i n reconstituted lipid vesicles. Nature 342. 87-89. 29 VOLPL. P.. KREUSE.K., HASHIMOTO. S . , ZORZATO.F.. POZZAN.T.. J . AND LEW.D. P. (1988). Calciosomc. a cytoplasmic organcllc: the MEI.DOLESI. IP3 sensitive Ca'+ store in non-muscle cells? Proc. Mztl Accd. Sci. USA 85. in9 I- 109s. 30 BOLE, D. G.. DOWIN.R.. DORIAL);. A . A N D JAMIESON. J . D. (IYX9). lmmunocytochemical localisation of BiP to the rough endoplasmic reticulum: Evidence for protein sorting by selective retention. J . HI~rochcni.~ ' y ~ i d w m . 37. 1817-1823. 31 KOCH.G. L. E., SMITH.M. J.. MACER.D. R. J.. BOOTH.C. ,\ND W O O D I N G . F. B. P. (1989). Structure and assembly of the cndoplasmic reticulum. Biocliein. Soc. Trcin\. 17. 328-331. 32 BOOTH.C . A N D KOCH.G. L. E . (1989). Perturbation of cellular calcium induces secretion of luminal E R proteins. Cell 59. 729-737. 33 BURGESS.T. L. A N D KELLY,R. B. (1987). Constitutive and regulated secretion of proteins. Annrr. Rev. Cell Biol. 3. 243-293.
Gordon L. E. Koch is at the Laboratory of Molecular Biology, Hills Rd., Cambridge CBI 2QH, UK.